diff options
Diffstat (limited to 'drivers/gpu')
101 files changed, 10117 insertions, 3573 deletions
diff --git a/drivers/gpu/drm/Kconfig b/drivers/gpu/drm/Kconfig index e4d971c..96eddd1 100644 --- a/drivers/gpu/drm/Kconfig +++ b/drivers/gpu/drm/Kconfig @@ -92,6 +92,7 @@ config DRM_I830 config DRM_I915 tristate "i915 driver" depends on AGP_INTEL + select SHMEM select DRM_KMS_HELPER select FB_CFB_FILLRECT select FB_CFB_COPYAREA @@ -102,6 +103,7 @@ config DRM_I915 select BACKLIGHT_CLASS_DEVICE if ACPI select INPUT if ACPI select ACPI_VIDEO if ACPI + select ACPI_BUTTON if ACPI help Choose this option if you have a system that has Intel 830M, 845G, 852GM, 855GM 865G or 915G integrated graphics. If M is selected, the diff --git a/drivers/gpu/drm/drm_crtc.c b/drivers/gpu/drm/drm_crtc.c index ba728ad..5cae0b3 100644 --- a/drivers/gpu/drm/drm_crtc.c +++ b/drivers/gpu/drm/drm_crtc.c @@ -482,6 +482,7 @@ void drm_connector_cleanup(struct drm_connector *connector) list_for_each_entry_safe(mode, t, &connector->user_modes, head) drm_mode_remove(connector, mode); + kfree(connector->fb_helper_private); mutex_lock(&dev->mode_config.mutex); drm_mode_object_put(dev, &connector->base); list_del(&connector->head); @@ -1555,8 +1556,6 @@ int drm_mode_cursor_ioctl(struct drm_device *dev, struct drm_crtc *crtc; int ret = 0; - DRM_DEBUG_KMS("\n"); - if (!req->flags) { DRM_ERROR("no operation set\n"); return -EINVAL; diff --git a/drivers/gpu/drm/drm_crtc_helper.c b/drivers/gpu/drm/drm_crtc_helper.c index fe86974..bbfd110 100644 --- a/drivers/gpu/drm/drm_crtc_helper.c +++ b/drivers/gpu/drm/drm_crtc_helper.c @@ -32,6 +32,7 @@ #include "drmP.h" #include "drm_crtc.h" #include "drm_crtc_helper.h" +#include "drm_fb_helper.h" static void drm_mode_validate_flag(struct drm_connector *connector, int flags) @@ -90,7 +91,15 @@ int drm_helper_probe_single_connector_modes(struct drm_connector *connector, list_for_each_entry_safe(mode, t, &connector->modes, head) mode->status = MODE_UNVERIFIED; - connector->status = connector->funcs->detect(connector); + if (connector->force) { + if (connector->force == DRM_FORCE_ON) + connector->status = connector_status_connected; + else + connector->status = connector_status_disconnected; + if (connector->funcs->force) + connector->funcs->force(connector); + } else + connector->status = connector->funcs->detect(connector); if (connector->status == connector_status_disconnected) { DRM_DEBUG_KMS("%s is disconnected\n", @@ -267,6 +276,66 @@ static struct drm_display_mode *drm_has_preferred_mode(struct drm_connector *con return NULL; } +static bool drm_has_cmdline_mode(struct drm_connector *connector) +{ + struct drm_fb_helper_connector *fb_help_conn = connector->fb_helper_private; + struct drm_fb_helper_cmdline_mode *cmdline_mode; + + if (!fb_help_conn) + return false; + + cmdline_mode = &fb_help_conn->cmdline_mode; + return cmdline_mode->specified; +} + +static struct drm_display_mode *drm_pick_cmdline_mode(struct drm_connector *connector, int width, int height) +{ + struct drm_fb_helper_connector *fb_help_conn = connector->fb_helper_private; + struct drm_fb_helper_cmdline_mode *cmdline_mode; + struct drm_display_mode *mode = NULL; + + if (!fb_help_conn) + return mode; + + cmdline_mode = &fb_help_conn->cmdline_mode; + if (cmdline_mode->specified == false) + return mode; + + /* attempt to find a matching mode in the list of modes + * we have gotten so far, if not add a CVT mode that conforms + */ + if (cmdline_mode->rb || cmdline_mode->margins) + goto create_mode; + + list_for_each_entry(mode, &connector->modes, head) { + /* check width/height */ + if (mode->hdisplay != cmdline_mode->xres || + mode->vdisplay != cmdline_mode->yres) + continue; + + if (cmdline_mode->refresh_specified) { + if (mode->vrefresh != cmdline_mode->refresh) + continue; + } + + if (cmdline_mode->interlace) { + if (!(mode->flags & DRM_MODE_FLAG_INTERLACE)) + continue; + } + return mode; + } + +create_mode: + mode = drm_cvt_mode(connector->dev, cmdline_mode->xres, + cmdline_mode->yres, + cmdline_mode->refresh_specified ? cmdline_mode->refresh : 60, + cmdline_mode->rb, cmdline_mode->interlace, + cmdline_mode->margins); + drm_mode_set_crtcinfo(mode, CRTC_INTERLACE_HALVE_V); + list_add(&mode->head, &connector->modes); + return mode; +} + static bool drm_connector_enabled(struct drm_connector *connector, bool strict) { bool enable; @@ -317,10 +386,16 @@ static bool drm_target_preferred(struct drm_device *dev, continue; } - DRM_DEBUG_KMS("looking for preferred mode on connector %d\n", - connector->base.id); + DRM_DEBUG_KMS("looking for cmdline mode on connector %d\n", + connector->base.id); - modes[i] = drm_has_preferred_mode(connector, width, height); + /* got for command line mode first */ + modes[i] = drm_pick_cmdline_mode(connector, width, height); + if (!modes[i]) { + DRM_DEBUG_KMS("looking for preferred mode on connector %d\n", + connector->base.id); + modes[i] = drm_has_preferred_mode(connector, width, height); + } /* No preferred modes, pick one off the list */ if (!modes[i] && !list_empty(&connector->modes)) { list_for_each_entry(modes[i], &connector->modes, head) @@ -369,6 +444,8 @@ static int drm_pick_crtcs(struct drm_device *dev, my_score = 1; if (connector->status == connector_status_connected) my_score++; + if (drm_has_cmdline_mode(connector)) + my_score++; if (drm_has_preferred_mode(connector, width, height)) my_score++; @@ -943,6 +1020,8 @@ bool drm_helper_initial_config(struct drm_device *dev) { int count = 0; + drm_fb_helper_parse_command_line(dev); + count = drm_helper_probe_connector_modes(dev, dev->mode_config.max_width, dev->mode_config.max_height); @@ -950,7 +1029,7 @@ bool drm_helper_initial_config(struct drm_device *dev) /* * we shouldn't end up with no modes here. */ - WARN(!count, "Connected connector with 0 modes\n"); + WARN(!count, "No connectors reported connected with modes\n"); drm_setup_crtcs(dev); diff --git a/drivers/gpu/drm/drm_edid.c b/drivers/gpu/drm/drm_edid.c index 90d76ba..b54ba63 100644 --- a/drivers/gpu/drm/drm_edid.c +++ b/drivers/gpu/drm/drm_edid.c @@ -109,7 +109,9 @@ static struct edid_quirk { /* Valid EDID header has these bytes */ -static u8 edid_header[] = { 0x00, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0x00 }; +static const u8 edid_header[] = { + 0x00, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0x00 +}; /** * edid_is_valid - sanity check EDID data @@ -500,6 +502,19 @@ static struct drm_display_mode *drm_find_dmt(struct drm_device *dev, } return mode; } + +/* + * 0 is reserved. The spec says 0x01 fill for unused timings. Some old + * monitors fill with ascii space (0x20) instead. + */ +static int +bad_std_timing(u8 a, u8 b) +{ + return (a == 0x00 && b == 0x00) || + (a == 0x01 && b == 0x01) || + (a == 0x20 && b == 0x20); +} + /** * drm_mode_std - convert standard mode info (width, height, refresh) into mode * @t: standard timing params @@ -513,6 +528,7 @@ static struct drm_display_mode *drm_find_dmt(struct drm_device *dev, */ struct drm_display_mode *drm_mode_std(struct drm_device *dev, struct std_timing *t, + int revision, int timing_level) { struct drm_display_mode *mode; @@ -523,14 +539,20 @@ struct drm_display_mode *drm_mode_std(struct drm_device *dev, unsigned vfreq = (t->vfreq_aspect & EDID_TIMING_VFREQ_MASK) >> EDID_TIMING_VFREQ_SHIFT; + if (bad_std_timing(t->hsize, t->vfreq_aspect)) + return NULL; + /* According to the EDID spec, the hdisplay = hsize * 8 + 248 */ hsize = t->hsize * 8 + 248; /* vrefresh_rate = vfreq + 60 */ vrefresh_rate = vfreq + 60; /* the vdisplay is calculated based on the aspect ratio */ - if (aspect_ratio == 0) - vsize = (hsize * 10) / 16; - else if (aspect_ratio == 1) + if (aspect_ratio == 0) { + if (revision < 3) + vsize = hsize; + else + vsize = (hsize * 10) / 16; + } else if (aspect_ratio == 1) vsize = (hsize * 3) / 4; else if (aspect_ratio == 2) vsize = (hsize * 4) / 5; @@ -538,7 +560,8 @@ struct drm_display_mode *drm_mode_std(struct drm_device *dev, vsize = (hsize * 9) / 16; /* HDTV hack */ if (hsize == 1360 && vsize == 765 && vrefresh_rate == 60) { - mode = drm_cvt_mode(dev, hsize, vsize, vrefresh_rate, 0, 0); + mode = drm_cvt_mode(dev, hsize, vsize, vrefresh_rate, 0, 0, + false); mode->hdisplay = 1366; mode->vsync_start = mode->vsync_start - 1; mode->vsync_end = mode->vsync_end - 1; @@ -557,7 +580,8 @@ struct drm_display_mode *drm_mode_std(struct drm_device *dev, mode = drm_gtf_mode(dev, hsize, vsize, vrefresh_rate, 0, 0); break; case LEVEL_CVT: - mode = drm_cvt_mode(dev, hsize, vsize, vrefresh_rate, 0, 0); + mode = drm_cvt_mode(dev, hsize, vsize, vrefresh_rate, 0, 0, + false); break; } return mode; @@ -602,6 +626,12 @@ static struct drm_display_mode *drm_mode_detailed(struct drm_device *dev, return NULL; } + /* it is incorrect if hsync/vsync width is zero */ + if (!hsync_pulse_width || !vsync_pulse_width) { + DRM_DEBUG_KMS("Incorrect Detailed timing. " + "Wrong Hsync/Vsync pulse width\n"); + return NULL; + } mode = drm_mode_create(dev); if (!mode) return NULL; @@ -623,6 +653,21 @@ static struct drm_display_mode *drm_mode_detailed(struct drm_device *dev, mode->vsync_end = mode->vsync_start + vsync_pulse_width; mode->vtotal = mode->vdisplay + vblank; + /* perform the basic check for the detailed timing */ + if (mode->hsync_end > mode->htotal || + mode->vsync_end > mode->vtotal) { + drm_mode_destroy(dev, mode); + DRM_DEBUG_KMS("Incorrect detailed timing. " + "Sync is beyond the blank.\n"); + return NULL; + } + + /* Some EDIDs have bogus h/vtotal values */ + if (mode->hsync_end > mode->htotal) + mode->htotal = mode->hsync_end + 1; + if (mode->vsync_end > mode->vtotal) + mode->vtotal = mode->vsync_end + 1; + drm_mode_set_name(mode); if (pt->misc & DRM_EDID_PT_INTERLACED) @@ -779,7 +824,7 @@ static int add_standard_modes(struct drm_connector *connector, struct edid *edid continue; newmode = drm_mode_std(dev, &edid->standard_timings[i], - timing_level); + edid->revision, timing_level); if (newmode) { drm_mode_probed_add(connector, newmode); modes++; @@ -829,13 +874,13 @@ static int add_detailed_info(struct drm_connector *connector, case EDID_DETAIL_MONITOR_CPDATA: break; case EDID_DETAIL_STD_MODES: - /* Five modes per detailed section */ - for (j = 0; j < 5; i++) { + for (j = 0; j < 6; i++) { struct std_timing *std; struct drm_display_mode *newmode; std = &data->data.timings[j]; newmode = drm_mode_std(dev, std, + edid->revision, timing_level); if (newmode) { drm_mode_probed_add(connector, newmode); @@ -964,7 +1009,9 @@ static int add_detailed_info_eedid(struct drm_connector *connector, struct drm_display_mode *newmode; std = &data->data.timings[j]; - newmode = drm_mode_std(dev, std, timing_level); + newmode = drm_mode_std(dev, std, + edid->revision, + timing_level); if (newmode) { drm_mode_probed_add(connector, newmode); modes++; diff --git a/drivers/gpu/drm/drm_fb_helper.c b/drivers/gpu/drm/drm_fb_helper.c index 2c46713..65ef011 100644 --- a/drivers/gpu/drm/drm_fb_helper.c +++ b/drivers/gpu/drm/drm_fb_helper.c @@ -40,6 +40,199 @@ MODULE_LICENSE("GPL and additional rights"); static LIST_HEAD(kernel_fb_helper_list); +int drm_fb_helper_add_connector(struct drm_connector *connector) +{ + connector->fb_helper_private = kzalloc(sizeof(struct drm_fb_helper_connector), GFP_KERNEL); + if (!connector->fb_helper_private) + return -ENOMEM; + + return 0; +} +EXPORT_SYMBOL(drm_fb_helper_add_connector); + +static int my_atoi(const char *name) +{ + int val = 0; + + for (;; name++) { + switch (*name) { + case '0' ... '9': + val = 10*val+(*name-'0'); + break; + default: + return val; + } + } +} + +/** + * drm_fb_helper_connector_parse_command_line - parse command line for connector + * @connector - connector to parse line for + * @mode_option - per connector mode option + * + * This parses the connector specific then generic command lines for + * modes and options to configure the connector. + * + * This uses the same parameters as the fb modedb.c, except for extra + * <xres>x<yres>[M][R][-<bpp>][@<refresh>][i][m][eDd] + * + * enable/enable Digital/disable bit at the end + */ +static bool drm_fb_helper_connector_parse_command_line(struct drm_connector *connector, + const char *mode_option) +{ + const char *name; + unsigned int namelen; + int res_specified = 0, bpp_specified = 0, refresh_specified = 0; + unsigned int xres = 0, yres = 0, bpp = 32, refresh = 0; + int yres_specified = 0, cvt = 0, rb = 0, interlace = 0, margins = 0; + int i; + enum drm_connector_force force = DRM_FORCE_UNSPECIFIED; + struct drm_fb_helper_connector *fb_help_conn = connector->fb_helper_private; + struct drm_fb_helper_cmdline_mode *cmdline_mode; + + if (!fb_help_conn) + return false; + + cmdline_mode = &fb_help_conn->cmdline_mode; + if (!mode_option) + mode_option = fb_mode_option; + + if (!mode_option) { + cmdline_mode->specified = false; + return false; + } + + name = mode_option; + namelen = strlen(name); + for (i = namelen-1; i >= 0; i--) { + switch (name[i]) { + case '@': + namelen = i; + if (!refresh_specified && !bpp_specified && + !yres_specified) { + refresh = my_atoi(&name[i+1]); + refresh_specified = 1; + if (cvt || rb) + cvt = 0; + } else + goto done; + break; + case '-': + namelen = i; + if (!bpp_specified && !yres_specified) { + bpp = my_atoi(&name[i+1]); + bpp_specified = 1; + if (cvt || rb) + cvt = 0; + } else + goto done; + break; + case 'x': + if (!yres_specified) { + yres = my_atoi(&name[i+1]); + yres_specified = 1; + } else + goto done; + case '0' ... '9': + break; + case 'M': + if (!yres_specified) + cvt = 1; + break; + case 'R': + if (!cvt) + rb = 1; + break; + case 'm': + if (!cvt) + margins = 1; + break; + case 'i': + if (!cvt) + interlace = 1; + break; + case 'e': + force = DRM_FORCE_ON; + break; + case 'D': + if ((connector->connector_type != DRM_MODE_CONNECTOR_DVII) || + (connector->connector_type != DRM_MODE_CONNECTOR_HDMIB)) + force = DRM_FORCE_ON; + else + force = DRM_FORCE_ON_DIGITAL; + break; + case 'd': + force = DRM_FORCE_OFF; + break; + default: + goto done; + } + } + if (i < 0 && yres_specified) { + xres = my_atoi(name); + res_specified = 1; + } +done: + + DRM_DEBUG_KMS("cmdline mode for connector %s %dx%d@%dHz%s%s%s\n", + drm_get_connector_name(connector), xres, yres, + (refresh) ? refresh : 60, (rb) ? " reduced blanking" : + "", (margins) ? " with margins" : "", (interlace) ? + " interlaced" : ""); + + if (force) { + const char *s; + switch (force) { + case DRM_FORCE_OFF: s = "OFF"; break; + case DRM_FORCE_ON_DIGITAL: s = "ON - dig"; break; + default: + case DRM_FORCE_ON: s = "ON"; break; + } + + DRM_INFO("forcing %s connector %s\n", + drm_get_connector_name(connector), s); + connector->force = force; + } + + if (res_specified) { + cmdline_mode->specified = true; + cmdline_mode->xres = xres; + cmdline_mode->yres = yres; + } + + if (refresh_specified) { + cmdline_mode->refresh_specified = true; + cmdline_mode->refresh = refresh; + } + + if (bpp_specified) { + cmdline_mode->bpp_specified = true; + cmdline_mode->bpp = bpp; + } + cmdline_mode->rb = rb ? true : false; + cmdline_mode->cvt = cvt ? true : false; + cmdline_mode->interlace = interlace ? true : false; + + return true; +} + +int drm_fb_helper_parse_command_line(struct drm_device *dev) +{ + struct drm_connector *connector; + + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + char *option = NULL; + + /* do something on return - turn off connector maybe */ + if (fb_get_options(drm_get_connector_name(connector), &option)) + continue; + + drm_fb_helper_connector_parse_command_line(connector, option); + } + return 0; +} + bool drm_fb_helper_force_kernel_mode(void) { int i = 0; @@ -87,6 +280,7 @@ void drm_fb_helper_restore(void) } EXPORT_SYMBOL(drm_fb_helper_restore); +#ifdef CONFIG_MAGIC_SYSRQ static void drm_fb_helper_restore_work_fn(struct work_struct *ignored) { drm_fb_helper_restore(); @@ -103,6 +297,7 @@ static struct sysrq_key_op sysrq_drm_fb_helper_restore_op = { .help_msg = "force-fb(V)", .action_msg = "Restore framebuffer console", }; +#endif static void drm_fb_helper_on(struct fb_info *info) { @@ -259,6 +454,109 @@ out_free: } EXPORT_SYMBOL(drm_fb_helper_init_crtc_count); +static int setcolreg(struct drm_crtc *crtc, u16 red, u16 green, + u16 blue, u16 regno, struct fb_info *info) +{ + struct drm_fb_helper *fb_helper = info->par; + struct drm_framebuffer *fb = fb_helper->fb; + int pindex; + + if (info->fix.visual == FB_VISUAL_TRUECOLOR) { + u32 *palette; + u32 value; + /* place color in psuedopalette */ + if (regno > 16) + return -EINVAL; + palette = (u32 *)info->pseudo_palette; + red >>= (16 - info->var.red.length); + green >>= (16 - info->var.green.length); + blue >>= (16 - info->var.blue.length); + value = (red << info->var.red.offset) | + (green << info->var.green.offset) | + (blue << info->var.blue.offset); + palette[regno] = value; + return 0; + } + + pindex = regno; + + if (fb->bits_per_pixel == 16) { + pindex = regno << 3; + + if (fb->depth == 16 && regno > 63) + return -EINVAL; + if (fb->depth == 15 && regno > 31) + return -EINVAL; + + if (fb->depth == 16) { + u16 r, g, b; + int i; + if (regno < 32) { + for (i = 0; i < 8; i++) + fb_helper->funcs->gamma_set(crtc, red, + green, blue, pindex + i); + } + + fb_helper->funcs->gamma_get(crtc, &r, + &g, &b, + pindex >> 1); + + for (i = 0; i < 4; i++) + fb_helper->funcs->gamma_set(crtc, r, + green, b, + (pindex >> 1) + i); + } + } + + if (fb->depth != 16) + fb_helper->funcs->gamma_set(crtc, red, green, blue, pindex); + return 0; +} + +int drm_fb_helper_setcmap(struct fb_cmap *cmap, struct fb_info *info) +{ + struct drm_fb_helper *fb_helper = info->par; + struct drm_device *dev = fb_helper->dev; + u16 *red, *green, *blue, *transp; + struct drm_crtc *crtc; + int i, rc = 0; + int start; + + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { + struct drm_crtc_helper_funcs *crtc_funcs = crtc->helper_private; + for (i = 0; i < fb_helper->crtc_count; i++) { + if (crtc->base.id == fb_helper->crtc_info[i].crtc_id) + break; + } + if (i == fb_helper->crtc_count) + continue; + + red = cmap->red; + green = cmap->green; + blue = cmap->blue; + transp = cmap->transp; + start = cmap->start; + + for (i = 0; i < cmap->len; i++) { + u16 hred, hgreen, hblue, htransp = 0xffff; + + hred = *red++; + hgreen = *green++; + hblue = *blue++; + + if (transp) + htransp = *transp++; + + rc = setcolreg(crtc, hred, hgreen, hblue, start++, info); + if (rc) + return rc; + } + crtc_funcs->load_lut(crtc); + } + return rc; +} +EXPORT_SYMBOL(drm_fb_helper_setcmap); + int drm_fb_helper_setcolreg(unsigned regno, unsigned red, unsigned green, @@ -270,10 +568,13 @@ int drm_fb_helper_setcolreg(unsigned regno, struct drm_device *dev = fb_helper->dev; struct drm_crtc *crtc; int i; + int ret; - list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { - struct drm_framebuffer *fb = fb_helper->fb; + if (regno > 255) + return 1; + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { + struct drm_crtc_helper_funcs *crtc_funcs = crtc->helper_private; for (i = 0; i < fb_helper->crtc_count; i++) { if (crtc->base.id == fb_helper->crtc_info[i].crtc_id) break; @@ -281,35 +582,11 @@ int drm_fb_helper_setcolreg(unsigned regno, if (i == fb_helper->crtc_count) continue; - if (regno > 255) - return 1; - - if (fb->depth == 8) { - fb_helper->funcs->gamma_set(crtc, red, green, blue, regno); - return 0; - } + ret = setcolreg(crtc, red, green, blue, regno, info); + if (ret) + return ret; - if (regno < 16) { - switch (fb->depth) { - case 15: - fb->pseudo_palette[regno] = ((red & 0xf800) >> 1) | - ((green & 0xf800) >> 6) | - ((blue & 0xf800) >> 11); - break; - case 16: - fb->pseudo_palette[regno] = (red & 0xf800) | - ((green & 0xfc00) >> 5) | - ((blue & 0xf800) >> 11); - break; - case 24: - case 32: - fb->pseudo_palette[regno] = - (((red >> 8) & 0xff) << info->var.red.offset) | - (((green >> 8) & 0xff) << info->var.green.offset) | - (((blue >> 8) & 0xff) << info->var.blue.offset); - break; - } - } + crtc_funcs->load_lut(crtc); } return 0; } @@ -322,7 +599,7 @@ int drm_fb_helper_check_var(struct fb_var_screeninfo *var, struct drm_framebuffer *fb = fb_helper->fb; int depth; - if (var->pixclock == -1 || !var->pixclock) + if (var->pixclock != 0) return -EINVAL; /* Need to resize the fb object !!! */ @@ -414,7 +691,7 @@ int drm_fb_helper_set_par(struct fb_info *info) int ret; int i; - if (var->pixclock != -1) { + if (var->pixclock != 0) { DRM_ERROR("PIXEL CLCOK SET\n"); return -EINVAL; } @@ -430,7 +707,7 @@ int drm_fb_helper_set_par(struct fb_info *info) if (crtc->fb == fb_helper->crtc_info[i].mode_set.fb) { mutex_lock(&dev->mode_config.mutex); - ret = crtc->funcs->set_config(&fb_helper->crtc_info->mode_set); + ret = crtc->funcs->set_config(&fb_helper->crtc_info[i].mode_set); mutex_unlock(&dev->mode_config.mutex); if (ret) return ret; @@ -479,11 +756,14 @@ int drm_fb_helper_pan_display(struct fb_var_screeninfo *var, EXPORT_SYMBOL(drm_fb_helper_pan_display); int drm_fb_helper_single_fb_probe(struct drm_device *dev, + int preferred_bpp, int (*fb_create)(struct drm_device *dev, uint32_t fb_width, uint32_t fb_height, uint32_t surface_width, uint32_t surface_height, + uint32_t surface_depth, + uint32_t surface_bpp, struct drm_framebuffer **fb_ptr)) { struct drm_crtc *crtc; @@ -497,8 +777,48 @@ int drm_fb_helper_single_fb_probe(struct drm_device *dev, struct drm_framebuffer *fb; struct drm_mode_set *modeset = NULL; struct drm_fb_helper *fb_helper; + uint32_t surface_depth = 24, surface_bpp = 32; + /* if driver picks 8 or 16 by default use that + for both depth/bpp */ + if (preferred_bpp != surface_bpp) { + surface_depth = surface_bpp = preferred_bpp; + } /* first up get a count of crtcs now in use and new min/maxes width/heights */ + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + struct drm_fb_helper_connector *fb_help_conn = connector->fb_helper_private; + + struct drm_fb_helper_cmdline_mode *cmdline_mode; + + if (!fb_help_conn) + continue; + + cmdline_mode = &fb_help_conn->cmdline_mode; + + if (cmdline_mode->bpp_specified) { + switch (cmdline_mode->bpp) { + case 8: + surface_depth = surface_bpp = 8; + break; + case 15: + surface_depth = 15; + surface_bpp = 16; + break; + case 16: + surface_depth = surface_bpp = 16; + break; + case 24: + surface_depth = surface_bpp = 24; + break; + case 32: + surface_depth = 24; + surface_bpp = 32; + break; + } + break; + } + } + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { if (drm_helper_crtc_in_use(crtc)) { if (crtc->desired_mode) { @@ -527,7 +847,8 @@ int drm_fb_helper_single_fb_probe(struct drm_device *dev, /* do we have an fb already? */ if (list_empty(&dev->mode_config.fb_kernel_list)) { ret = (*fb_create)(dev, fb_width, fb_height, surface_width, - surface_height, &fb); + surface_height, surface_depth, surface_bpp, + &fb); if (ret) return -EINVAL; new_fb = 1; @@ -583,7 +904,7 @@ int drm_fb_helper_single_fb_probe(struct drm_device *dev, fb_helper->fb = fb; if (new_fb) { - info->var.pixclock = -1; + info->var.pixclock = 0; if (register_framebuffer(info) < 0) return -EINVAL; } else { @@ -618,10 +939,12 @@ void drm_fb_helper_free(struct drm_fb_helper *helper) } EXPORT_SYMBOL(drm_fb_helper_free); -void drm_fb_helper_fill_fix(struct fb_info *info, uint32_t pitch) +void drm_fb_helper_fill_fix(struct fb_info *info, uint32_t pitch, + uint32_t depth) { info->fix.type = FB_TYPE_PACKED_PIXELS; - info->fix.visual = FB_VISUAL_TRUECOLOR; + info->fix.visual = depth == 8 ? FB_VISUAL_PSEUDOCOLOR : + FB_VISUAL_TRUECOLOR; info->fix.type_aux = 0; info->fix.xpanstep = 1; /* doing it in hw */ info->fix.ypanstep = 1; /* doing it in hw */ diff --git a/drivers/gpu/drm/drm_gem.c b/drivers/gpu/drm/drm_gem.c index 230c9ff..e9dbb48 100644 --- a/drivers/gpu/drm/drm_gem.c +++ b/drivers/gpu/drm/drm_gem.c @@ -142,6 +142,19 @@ drm_gem_object_alloc(struct drm_device *dev, size_t size) if (IS_ERR(obj->filp)) goto free; + /* Basically we want to disable the OOM killer and handle ENOMEM + * ourselves by sacrificing pages from cached buffers. + * XXX shmem_file_[gs]et_gfp_mask() + */ + mapping_set_gfp_mask(obj->filp->f_path.dentry->d_inode->i_mapping, + GFP_HIGHUSER | + __GFP_COLD | + __GFP_FS | + __GFP_RECLAIMABLE | + __GFP_NORETRY | + __GFP_NOWARN | + __GFP_NOMEMALLOC); + kref_init(&obj->refcount); kref_init(&obj->handlecount); obj->size = size; @@ -539,7 +552,7 @@ int drm_gem_mmap(struct file *filp, struct vm_area_struct *vma) vma->vm_flags |= VM_RESERVED | VM_IO | VM_PFNMAP | VM_DONTEXPAND; vma->vm_ops = obj->dev->driver->gem_vm_ops; vma->vm_private_data = map->handle; - vma->vm_page_prot = pgprot_writecombine(vma->vm_page_prot); + vma->vm_page_prot = pgprot_writecombine(vm_get_page_prot(vma->vm_flags)); /* Take a ref for this mapping of the object, so that the fault * handler can dereference the mmap offset's pointer to the object. diff --git a/drivers/gpu/drm/drm_mm.c b/drivers/gpu/drm/drm_mm.c index c861d80..97dc5a4 100644 --- a/drivers/gpu/drm/drm_mm.c +++ b/drivers/gpu/drm/drm_mm.c @@ -103,6 +103,11 @@ static struct drm_mm_node *drm_mm_kmalloc(struct drm_mm *mm, int atomic) return child; } +/* drm_mm_pre_get() - pre allocate drm_mm_node structure + * drm_mm: memory manager struct we are pre-allocating for + * + * Returns 0 on success or -ENOMEM if allocation fails. + */ int drm_mm_pre_get(struct drm_mm *mm) { struct drm_mm_node *node; @@ -253,12 +258,14 @@ void drm_mm_put_block(struct drm_mm_node *cur) prev_node->size += next_node->size; list_del(&next_node->ml_entry); list_del(&next_node->fl_entry); + spin_lock(&mm->unused_lock); if (mm->num_unused < MM_UNUSED_TARGET) { list_add(&next_node->fl_entry, &mm->unused_nodes); ++mm->num_unused; } else kfree(next_node); + spin_unlock(&mm->unused_lock); } else { next_node->size += cur->size; next_node->start = cur->start; @@ -271,11 +278,13 @@ void drm_mm_put_block(struct drm_mm_node *cur) list_add(&cur->fl_entry, &mm->fl_entry); } else { list_del(&cur->ml_entry); + spin_lock(&mm->unused_lock); if (mm->num_unused < MM_UNUSED_TARGET) { list_add(&cur->fl_entry, &mm->unused_nodes); ++mm->num_unused; } else kfree(cur); + spin_unlock(&mm->unused_lock); } } diff --git a/drivers/gpu/drm/drm_modes.c b/drivers/gpu/drm/drm_modes.c index 49404ce..51f6772 100644 --- a/drivers/gpu/drm/drm_modes.c +++ b/drivers/gpu/drm/drm_modes.c @@ -88,7 +88,7 @@ EXPORT_SYMBOL(drm_mode_debug_printmodeline); #define HV_FACTOR 1000 struct drm_display_mode *drm_cvt_mode(struct drm_device *dev, int hdisplay, int vdisplay, int vrefresh, - bool reduced, bool interlaced) + bool reduced, bool interlaced, bool margins) { /* 1) top/bottom margin size (% of height) - default: 1.8, */ #define CVT_MARGIN_PERCENTAGE 18 @@ -101,7 +101,6 @@ struct drm_display_mode *drm_cvt_mode(struct drm_device *dev, int hdisplay, /* Pixel Clock step (kHz) */ #define CVT_CLOCK_STEP 250 struct drm_display_mode *drm_mode; - bool margins = false; unsigned int vfieldrate, hperiod; int hdisplay_rnd, hmargin, vdisplay_rnd, vmargin, vsync; int interlace; diff --git a/drivers/gpu/drm/drm_vm.c b/drivers/gpu/drm/drm_vm.c index 7e1fbe5..4ac900f 100644 --- a/drivers/gpu/drm/drm_vm.c +++ b/drivers/gpu/drm/drm_vm.c @@ -369,28 +369,28 @@ static int drm_vm_sg_fault(struct vm_area_struct *vma, struct vm_fault *vmf) } /** AGP virtual memory operations */ -static struct vm_operations_struct drm_vm_ops = { +static const struct vm_operations_struct drm_vm_ops = { .fault = drm_vm_fault, .open = drm_vm_open, .close = drm_vm_close, }; /** Shared virtual memory operations */ -static struct vm_operations_struct drm_vm_shm_ops = { +static const struct vm_operations_struct drm_vm_shm_ops = { .fault = drm_vm_shm_fault, .open = drm_vm_open, .close = drm_vm_shm_close, }; /** DMA virtual memory operations */ -static struct vm_operations_struct drm_vm_dma_ops = { +static const struct vm_operations_struct drm_vm_dma_ops = { .fault = drm_vm_dma_fault, .open = drm_vm_open, .close = drm_vm_close, }; /** Scatter-gather virtual memory operations */ -static struct vm_operations_struct drm_vm_sg_ops = { +static const struct vm_operations_struct drm_vm_sg_ops = { .fault = drm_vm_sg_fault, .open = drm_vm_open, .close = drm_vm_close, diff --git a/drivers/gpu/drm/i915/Makefile b/drivers/gpu/drm/i915/Makefile index 5269dfa..fa7b9be 100644 --- a/drivers/gpu/drm/i915/Makefile +++ b/drivers/gpu/drm/i915/Makefile @@ -9,6 +9,7 @@ i915-y := i915_drv.o i915_dma.o i915_irq.o i915_mem.o \ i915_gem.o \ i915_gem_debug.o \ i915_gem_tiling.o \ + i915_trace_points.o \ intel_display.o \ intel_crt.o \ intel_lvds.o \ diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c index 1e3bdce..26bf055 100644 --- a/drivers/gpu/drm/i915/i915_debugfs.c +++ b/drivers/gpu/drm/i915/i915_debugfs.c @@ -96,11 +96,13 @@ static int i915_gem_object_list_info(struct seq_file *m, void *data) { struct drm_gem_object *obj = obj_priv->obj; - seq_printf(m, " %p: %s %08x %08x %d", + seq_printf(m, " %p: %s %8zd %08x %08x %d %s", obj, get_pin_flag(obj_priv), + obj->size, obj->read_domains, obj->write_domain, - obj_priv->last_rendering_seqno); + obj_priv->last_rendering_seqno, + obj_priv->dirty ? "dirty" : ""); if (obj->name) seq_printf(m, " (name: %d)", obj->name); @@ -265,10 +267,10 @@ static void i915_dump_pages(struct seq_file *m, struct page **pages, int page_co uint32_t *mem; for (page = 0; page < page_count; page++) { - mem = kmap(pages[page]); + mem = kmap_atomic(pages[page], KM_USER0); for (i = 0; i < PAGE_SIZE; i += 4) seq_printf(m, "%08x : %08x\n", i, mem[i / 4]); - kunmap(pages[page]); + kunmap_atomic(pages[page], KM_USER0); } } diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c index 5a49a18..e5b138b 100644 --- a/drivers/gpu/drm/i915/i915_dma.c +++ b/drivers/gpu/drm/i915/i915_dma.c @@ -33,6 +33,7 @@ #include "intel_drv.h" #include "i915_drm.h" #include "i915_drv.h" +#include "i915_trace.h" #include <linux/vgaarb.h> /* Really want an OS-independent resettable timer. Would like to have @@ -50,14 +51,18 @@ int i915_wait_ring(struct drm_device * dev, int n, const char *caller) u32 last_head = I915_READ(PRB0_HEAD) & HEAD_ADDR; int i; + trace_i915_ring_wait_begin (dev); + for (i = 0; i < 100000; i++) { ring->head = I915_READ(PRB0_HEAD) & HEAD_ADDR; acthd = I915_READ(acthd_reg); ring->space = ring->head - (ring->tail + 8); if (ring->space < 0) ring->space += ring->Size; - if (ring->space >= n) + if (ring->space >= n) { + trace_i915_ring_wait_end (dev); return 0; + } if (dev->primary->master) { struct drm_i915_master_private *master_priv = dev->primary->master->driver_priv; @@ -77,6 +82,7 @@ int i915_wait_ring(struct drm_device * dev, int n, const char *caller) } + trace_i915_ring_wait_end (dev); return -EBUSY; } @@ -922,7 +928,8 @@ static int i915_get_bridge_dev(struct drm_device *dev) * how much was set aside so we can use it for our own purposes. */ static int i915_probe_agp(struct drm_device *dev, uint32_t *aperture_size, - uint32_t *preallocated_size) + uint32_t *preallocated_size, + uint32_t *start) { struct drm_i915_private *dev_priv = dev->dev_private; u16 tmp = 0; @@ -1009,10 +1016,159 @@ static int i915_probe_agp(struct drm_device *dev, uint32_t *aperture_size, return -1; } *preallocated_size = stolen - overhead; + *start = overhead; return 0; } +#define PTE_ADDRESS_MASK 0xfffff000 +#define PTE_ADDRESS_MASK_HIGH 0x000000f0 /* i915+ */ +#define PTE_MAPPING_TYPE_UNCACHED (0 << 1) +#define PTE_MAPPING_TYPE_DCACHE (1 << 1) /* i830 only */ +#define PTE_MAPPING_TYPE_CACHED (3 << 1) +#define PTE_MAPPING_TYPE_MASK (3 << 1) +#define PTE_VALID (1 << 0) + +/** + * i915_gtt_to_phys - take a GTT address and turn it into a physical one + * @dev: drm device + * @gtt_addr: address to translate + * + * Some chip functions require allocations from stolen space but need the + * physical address of the memory in question. We use this routine + * to get a physical address suitable for register programming from a given + * GTT address. + */ +static unsigned long i915_gtt_to_phys(struct drm_device *dev, + unsigned long gtt_addr) +{ + unsigned long *gtt; + unsigned long entry, phys; + int gtt_bar = IS_I9XX(dev) ? 0 : 1; + int gtt_offset, gtt_size; + + if (IS_I965G(dev)) { + if (IS_G4X(dev) || IS_IGDNG(dev)) { + gtt_offset = 2*1024*1024; + gtt_size = 2*1024*1024; + } else { + gtt_offset = 512*1024; + gtt_size = 512*1024; + } + } else { + gtt_bar = 3; + gtt_offset = 0; + gtt_size = pci_resource_len(dev->pdev, gtt_bar); + } + + gtt = ioremap_wc(pci_resource_start(dev->pdev, gtt_bar) + gtt_offset, + gtt_size); + if (!gtt) { + DRM_ERROR("ioremap of GTT failed\n"); + return 0; + } + + entry = *(volatile u32 *)(gtt + (gtt_addr / 1024)); + + DRM_DEBUG("GTT addr: 0x%08lx, PTE: 0x%08lx\n", gtt_addr, entry); + + /* Mask out these reserved bits on this hardware. */ + if (!IS_I9XX(dev) || IS_I915G(dev) || IS_I915GM(dev) || + IS_I945G(dev) || IS_I945GM(dev)) { + entry &= ~PTE_ADDRESS_MASK_HIGH; + } + + /* If it's not a mapping type we know, then bail. */ + if ((entry & PTE_MAPPING_TYPE_MASK) != PTE_MAPPING_TYPE_UNCACHED && + (entry & PTE_MAPPING_TYPE_MASK) != PTE_MAPPING_TYPE_CACHED) { + iounmap(gtt); + return 0; + } + + if (!(entry & PTE_VALID)) { + DRM_ERROR("bad GTT entry in stolen space\n"); + iounmap(gtt); + return 0; + } + + iounmap(gtt); + + phys =(entry & PTE_ADDRESS_MASK) | + ((uint64_t)(entry & PTE_ADDRESS_MASK_HIGH) << (32 - 4)); + + DRM_DEBUG("GTT addr: 0x%08lx, phys addr: 0x%08lx\n", gtt_addr, phys); + + return phys; +} + +static void i915_warn_stolen(struct drm_device *dev) +{ + DRM_ERROR("not enough stolen space for compressed buffer, disabling\n"); + DRM_ERROR("hint: you may be able to increase stolen memory size in the BIOS to avoid this\n"); +} + +static void i915_setup_compression(struct drm_device *dev, int size) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_mm_node *compressed_fb, *compressed_llb; + unsigned long cfb_base, ll_base; + + /* Leave 1M for line length buffer & misc. */ + compressed_fb = drm_mm_search_free(&dev_priv->vram, size, 4096, 0); + if (!compressed_fb) { + i915_warn_stolen(dev); + return; + } + + compressed_fb = drm_mm_get_block(compressed_fb, size, 4096); + if (!compressed_fb) { + i915_warn_stolen(dev); + return; + } + + cfb_base = i915_gtt_to_phys(dev, compressed_fb->start); + if (!cfb_base) { + DRM_ERROR("failed to get stolen phys addr, disabling FBC\n"); + drm_mm_put_block(compressed_fb); + } + + if (!IS_GM45(dev)) { + compressed_llb = drm_mm_search_free(&dev_priv->vram, 4096, + 4096, 0); + if (!compressed_llb) { + i915_warn_stolen(dev); + return; + } + + compressed_llb = drm_mm_get_block(compressed_llb, 4096, 4096); + if (!compressed_llb) { + i915_warn_stolen(dev); + return; + } + + ll_base = i915_gtt_to_phys(dev, compressed_llb->start); + if (!ll_base) { + DRM_ERROR("failed to get stolen phys addr, disabling FBC\n"); + drm_mm_put_block(compressed_fb); + drm_mm_put_block(compressed_llb); + } + } + + dev_priv->cfb_size = size; + + if (IS_GM45(dev)) { + g4x_disable_fbc(dev); + I915_WRITE(DPFC_CB_BASE, compressed_fb->start); + } else { + i8xx_disable_fbc(dev); + I915_WRITE(FBC_CFB_BASE, cfb_base); + I915_WRITE(FBC_LL_BASE, ll_base); + } + + DRM_DEBUG("FBC base 0x%08lx, ll base 0x%08lx, size %dM\n", cfb_base, + ll_base, size >> 20); +} + /* true = enable decode, false = disable decoder */ static unsigned int i915_vga_set_decode(void *cookie, bool state) { @@ -1027,6 +1183,7 @@ static unsigned int i915_vga_set_decode(void *cookie, bool state) } static int i915_load_modeset_init(struct drm_device *dev, + unsigned long prealloc_start, unsigned long prealloc_size, unsigned long agp_size) { @@ -1047,6 +1204,10 @@ static int i915_load_modeset_init(struct drm_device *dev, /* Basic memrange allocator for stolen space (aka vram) */ drm_mm_init(&dev_priv->vram, 0, prealloc_size); + DRM_INFO("set up %ldM of stolen space\n", prealloc_size / (1024*1024)); + + /* We're off and running w/KMS */ + dev_priv->mm.suspended = 0; /* Let GEM Manage from end of prealloc space to end of aperture. * @@ -1059,10 +1220,24 @@ static int i915_load_modeset_init(struct drm_device *dev, */ i915_gem_do_init(dev, prealloc_size, agp_size - 4096); + mutex_lock(&dev->struct_mutex); ret = i915_gem_init_ringbuffer(dev); + mutex_unlock(&dev->struct_mutex); if (ret) goto out; + /* Try to set up FBC with a reasonable compressed buffer size */ + if (I915_HAS_FBC(dev) && i915_powersave) { + int cfb_size; + + /* Try to get an 8M buffer... */ + if (prealloc_size > (9*1024*1024)) + cfb_size = 8*1024*1024; + else /* fall back to 7/8 of the stolen space */ + cfb_size = prealloc_size * 7 / 8; + i915_setup_compression(dev, cfb_size); + } + /* Allow hardware batchbuffers unless told otherwise. */ dev_priv->allow_batchbuffer = 1; @@ -1180,7 +1355,7 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags) struct drm_i915_private *dev_priv = dev->dev_private; resource_size_t base, size; int ret = 0, mmio_bar = IS_I9XX(dev) ? 0 : 1; - uint32_t agp_size, prealloc_size; + uint32_t agp_size, prealloc_size, prealloc_start; /* i915 has 4 more counters */ dev->counters += 4; @@ -1234,7 +1409,7 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags) "performance may suffer.\n"); } - ret = i915_probe_agp(dev, &agp_size, &prealloc_size); + ret = i915_probe_agp(dev, &agp_size, &prealloc_size, &prealloc_start); if (ret) goto out_iomapfree; @@ -1292,6 +1467,7 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags) spin_lock_init(&dev_priv->user_irq_lock); spin_lock_init(&dev_priv->error_lock); dev_priv->user_irq_refcount = 0; + dev_priv->trace_irq_seqno = 0; ret = drm_vblank_init(dev, I915_NUM_PIPE); @@ -1300,8 +1476,12 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags) return ret; } + /* Start out suspended */ + dev_priv->mm.suspended = 1; + if (drm_core_check_feature(dev, DRIVER_MODESET)) { - ret = i915_load_modeset_init(dev, prealloc_size, agp_size); + ret = i915_load_modeset_init(dev, prealloc_start, + prealloc_size, agp_size); if (ret < 0) { DRM_ERROR("failed to init modeset\n"); goto out_workqueue_free; @@ -1313,6 +1493,8 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags) if (!IS_IGDNG(dev)) intel_opregion_init(dev, 0); + setup_timer(&dev_priv->hangcheck_timer, i915_hangcheck_elapsed, + (unsigned long) dev); return 0; out_workqueue_free: @@ -1333,6 +1515,7 @@ int i915_driver_unload(struct drm_device *dev) struct drm_i915_private *dev_priv = dev->dev_private; destroy_workqueue(dev_priv->wq); + del_timer_sync(&dev_priv->hangcheck_timer); io_mapping_free(dev_priv->mm.gtt_mapping); if (dev_priv->mm.gtt_mtrr >= 0) { @@ -1472,6 +1655,7 @@ struct drm_ioctl_desc i915_ioctls[] = { DRM_IOCTL_DEF(DRM_I915_GEM_GET_TILING, i915_gem_get_tiling, 0), DRM_IOCTL_DEF(DRM_I915_GEM_GET_APERTURE, i915_gem_get_aperture_ioctl, 0), DRM_IOCTL_DEF(DRM_I915_GET_PIPE_FROM_CRTC_ID, intel_get_pipe_from_crtc_id, 0), + DRM_IOCTL_DEF(DRM_I915_GEM_MADVISE, i915_gem_madvise_ioctl, 0), }; int i915_max_ioctl = DRM_ARRAY_SIZE(i915_ioctls); diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c index dbe568c..7f436ec 100644 --- a/drivers/gpu/drm/i915/i915_drv.c +++ b/drivers/gpu/drm/i915/i915_drv.c @@ -89,6 +89,9 @@ static int i915_suspend(struct drm_device *dev, pm_message_t state) pci_set_power_state(dev->pdev, PCI_D3hot); } + /* Modeset on resume, not lid events */ + dev_priv->modeset_on_lid = 0; + return 0; } @@ -97,8 +100,6 @@ static int i915_resume(struct drm_device *dev) struct drm_i915_private *dev_priv = dev->dev_private; int ret = 0; - pci_set_power_state(dev->pdev, PCI_D0); - pci_restore_state(dev->pdev); if (pci_enable_device(dev->pdev)) return -1; pci_set_master(dev->pdev); @@ -124,9 +125,135 @@ static int i915_resume(struct drm_device *dev) drm_helper_resume_force_mode(dev); } + dev_priv->modeset_on_lid = 0; + return ret; } +/** + * i965_reset - reset chip after a hang + * @dev: drm device to reset + * @flags: reset domains + * + * Reset the chip. Useful if a hang is detected. Returns zero on successful + * reset or otherwise an error code. + * + * Procedure is fairly simple: + * - reset the chip using the reset reg + * - re-init context state + * - re-init hardware status page + * - re-init ring buffer + * - re-init interrupt state + * - re-init display + */ +int i965_reset(struct drm_device *dev, u8 flags) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + unsigned long timeout; + u8 gdrst; + /* + * We really should only reset the display subsystem if we actually + * need to + */ + bool need_display = true; + + mutex_lock(&dev->struct_mutex); + + /* + * Clear request list + */ + i915_gem_retire_requests(dev); + + if (need_display) + i915_save_display(dev); + + if (IS_I965G(dev) || IS_G4X(dev)) { + /* + * Set the domains we want to reset, then the reset bit (bit 0). + * Clear the reset bit after a while and wait for hardware status + * bit (bit 1) to be set + */ + pci_read_config_byte(dev->pdev, GDRST, &gdrst); + pci_write_config_byte(dev->pdev, GDRST, gdrst | flags | ((flags == GDRST_FULL) ? 0x1 : 0x0)); + udelay(50); + pci_write_config_byte(dev->pdev, GDRST, gdrst & 0xfe); + + /* ...we don't want to loop forever though, 500ms should be plenty */ + timeout = jiffies + msecs_to_jiffies(500); + do { + udelay(100); + pci_read_config_byte(dev->pdev, GDRST, &gdrst); + } while ((gdrst & 0x1) && time_after(timeout, jiffies)); + + if (gdrst & 0x1) { + WARN(true, "i915: Failed to reset chip\n"); + mutex_unlock(&dev->struct_mutex); + return -EIO; + } + } else { + DRM_ERROR("Error occurred. Don't know how to reset this chip.\n"); + return -ENODEV; + } + + /* Ok, now get things going again... */ + + /* + * Everything depends on having the GTT running, so we need to start + * there. Fortunately we don't need to do this unless we reset the + * chip at a PCI level. + * + * Next we need to restore the context, but we don't use those + * yet either... + * + * Ring buffer needs to be re-initialized in the KMS case, or if X + * was running at the time of the reset (i.e. we weren't VT + * switched away). + */ + if (drm_core_check_feature(dev, DRIVER_MODESET) || + !dev_priv->mm.suspended) { + drm_i915_ring_buffer_t *ring = &dev_priv->ring; + struct drm_gem_object *obj = ring->ring_obj; + struct drm_i915_gem_object *obj_priv = obj->driver_private; + dev_priv->mm.suspended = 0; + + /* Stop the ring if it's running. */ + I915_WRITE(PRB0_CTL, 0); + I915_WRITE(PRB0_TAIL, 0); + I915_WRITE(PRB0_HEAD, 0); + + /* Initialize the ring. */ + I915_WRITE(PRB0_START, obj_priv->gtt_offset); + I915_WRITE(PRB0_CTL, + ((obj->size - 4096) & RING_NR_PAGES) | + RING_NO_REPORT | + RING_VALID); + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + i915_kernel_lost_context(dev); + else { + ring->head = I915_READ(PRB0_HEAD) & HEAD_ADDR; + ring->tail = I915_READ(PRB0_TAIL) & TAIL_ADDR; + ring->space = ring->head - (ring->tail + 8); + if (ring->space < 0) + ring->space += ring->Size; + } + + mutex_unlock(&dev->struct_mutex); + drm_irq_uninstall(dev); + drm_irq_install(dev); + mutex_lock(&dev->struct_mutex); + } + + /* + * Display needs restore too... + */ + if (need_display) + i915_restore_display(dev); + + mutex_unlock(&dev->struct_mutex); + return 0; +} + + static int __devinit i915_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent) { @@ -234,6 +361,8 @@ static int __init i915_init(void) { driver.num_ioctls = i915_max_ioctl; + i915_gem_shrinker_init(); + /* * If CONFIG_DRM_I915_KMS is set, default to KMS unless * explicitly disabled with the module pararmeter. @@ -260,6 +389,7 @@ static int __init i915_init(void) static void __exit i915_exit(void) { + i915_gem_shrinker_exit(); drm_exit(&driver); } diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h index a0632f8..a725f659 100644 --- a/drivers/gpu/drm/i915/i915_drv.h +++ b/drivers/gpu/drm/i915/i915_drv.h @@ -48,6 +48,11 @@ enum pipe { PIPE_B, }; +enum plane { + PLANE_A = 0, + PLANE_B, +}; + #define I915_NUM_PIPE 2 /* Interface history: @@ -148,6 +153,23 @@ struct drm_i915_error_state { struct timeval time; }; +struct drm_i915_display_funcs { + void (*dpms)(struct drm_crtc *crtc, int mode); + bool (*fbc_enabled)(struct drm_crtc *crtc); + void (*enable_fbc)(struct drm_crtc *crtc, unsigned long interval); + void (*disable_fbc)(struct drm_device *dev); + int (*get_display_clock_speed)(struct drm_device *dev); + int (*get_fifo_size)(struct drm_device *dev, int plane); + void (*update_wm)(struct drm_device *dev, int planea_clock, + int planeb_clock, int sr_hdisplay, int pixel_size); + /* clock updates for mode set */ + /* cursor updates */ + /* render clock increase/decrease */ + /* display clock increase/decrease */ + /* pll clock increase/decrease */ + /* clock gating init */ +}; + typedef struct drm_i915_private { struct drm_device *dev; @@ -180,6 +202,7 @@ typedef struct drm_i915_private { spinlock_t user_irq_lock; /** Refcount for i915_user_irq_get() versus i915_user_irq_put(). */ int user_irq_refcount; + u32 trace_irq_seqno; /** Cached value of IMR to avoid reads in updating the bitfield */ u32 irq_mask_reg; u32 pipestat[2]; @@ -198,10 +221,21 @@ typedef struct drm_i915_private { unsigned int sr01, adpa, ppcr, dvob, dvoc, lvds; int vblank_pipe; + /* For hangcheck timer */ +#define DRM_I915_HANGCHECK_PERIOD 75 /* in jiffies */ + struct timer_list hangcheck_timer; + int hangcheck_count; + uint32_t last_acthd; + bool cursor_needs_physical; struct drm_mm vram; + unsigned long cfb_size; + unsigned long cfb_pitch; + int cfb_fence; + int cfb_plane; + int irq_enabled; struct intel_opregion opregion; @@ -222,6 +256,8 @@ typedef struct drm_i915_private { unsigned int edp_support:1; int lvds_ssc_freq; + struct notifier_block lid_notifier; + int crt_ddc_bus; /* -1 = unknown, else GPIO to use for CRT DDC */ struct drm_i915_fence_reg fence_regs[16]; /* assume 965 */ int fence_reg_start; /* 4 if userland hasn't ioctl'd us yet */ @@ -234,7 +270,11 @@ typedef struct drm_i915_private { struct work_struct error_work; struct workqueue_struct *wq; + /* Display functions */ + struct drm_i915_display_funcs display; + /* Register state */ + bool modeset_on_lid; u8 saveLBB; u32 saveDSPACNTR; u32 saveDSPBCNTR; @@ -256,6 +296,13 @@ typedef struct drm_i915_private { u32 saveVBLANK_A; u32 saveVSYNC_A; u32 saveBCLRPAT_A; + u32 saveTRANSACONF; + u32 saveTRANS_HTOTAL_A; + u32 saveTRANS_HBLANK_A; + u32 saveTRANS_HSYNC_A; + u32 saveTRANS_VTOTAL_A; + u32 saveTRANS_VBLANK_A; + u32 saveTRANS_VSYNC_A; u32 savePIPEASTAT; u32 saveDSPASTRIDE; u32 saveDSPASIZE; @@ -264,8 +311,11 @@ typedef struct drm_i915_private { u32 saveDSPASURF; u32 saveDSPATILEOFF; u32 savePFIT_PGM_RATIOS; + u32 saveBLC_HIST_CTL; u32 saveBLC_PWM_CTL; u32 saveBLC_PWM_CTL2; + u32 saveBLC_CPU_PWM_CTL; + u32 saveBLC_CPU_PWM_CTL2; u32 saveFPB0; u32 saveFPB1; u32 saveDPLL_B; @@ -277,6 +327,13 @@ typedef struct drm_i915_private { u32 saveVBLANK_B; u32 saveVSYNC_B; u32 saveBCLRPAT_B; + u32 saveTRANSBCONF; + u32 saveTRANS_HTOTAL_B; + u32 saveTRANS_HBLANK_B; + u32 saveTRANS_HSYNC_B; + u32 saveTRANS_VTOTAL_B; + u32 saveTRANS_VBLANK_B; + u32 saveTRANS_VSYNC_B; u32 savePIPEBSTAT; u32 saveDSPBSTRIDE; u32 saveDSPBSIZE; @@ -302,6 +359,7 @@ typedef struct drm_i915_private { u32 savePFIT_CONTROL; u32 save_palette_a[256]; u32 save_palette_b[256]; + u32 saveDPFC_CB_BASE; u32 saveFBC_CFB_BASE; u32 saveFBC_LL_BASE; u32 saveFBC_CONTROL; @@ -309,6 +367,12 @@ typedef struct drm_i915_private { u32 saveIER; u32 saveIIR; u32 saveIMR; + u32 saveDEIER; + u32 saveDEIMR; + u32 saveGTIER; + u32 saveGTIMR; + u32 saveFDI_RXA_IMR; + u32 saveFDI_RXB_IMR; u32 saveCACHE_MODE_0; u32 saveD_STATE; u32 saveDSPCLK_GATE_D; @@ -342,6 +406,26 @@ typedef struct drm_i915_private { u32 savePIPEB_DP_LINK_M; u32 savePIPEA_DP_LINK_N; u32 savePIPEB_DP_LINK_N; + u32 saveFDI_RXA_CTL; + u32 saveFDI_TXA_CTL; + u32 saveFDI_RXB_CTL; + u32 saveFDI_TXB_CTL; + u32 savePFA_CTL_1; + u32 savePFB_CTL_1; + u32 savePFA_WIN_SZ; + u32 savePFB_WIN_SZ; + u32 savePFA_WIN_POS; + u32 savePFB_WIN_POS; + u32 savePCH_DREF_CONTROL; + u32 saveDISP_ARB_CTL; + u32 savePIPEA_DATA_M1; + u32 savePIPEA_DATA_N1; + u32 savePIPEA_LINK_M1; + u32 savePIPEA_LINK_N1; + u32 savePIPEB_DATA_M1; + u32 savePIPEB_DATA_N1; + u32 savePIPEB_LINK_M1; + u32 savePIPEB_LINK_N1; struct { struct drm_mm gtt_space; @@ -350,6 +434,15 @@ typedef struct drm_i915_private { int gtt_mtrr; /** + * Membership on list of all loaded devices, used to evict + * inactive buffers under memory pressure. + * + * Modifications should only be done whilst holding the + * shrink_list_lock spinlock. + */ + struct list_head shrink_list; + + /** * List of objects currently involved in rendering from the * ringbuffer. * @@ -432,7 +525,7 @@ typedef struct drm_i915_private { * It prevents command submission from occuring and makes * every pending request fail */ - int wedged; + atomic_t wedged; /** Bit 6 swizzling required for X tiling */ uint32_t bit_6_swizzle_x; @@ -443,6 +536,8 @@ typedef struct drm_i915_private { struct drm_i915_gem_phys_object *phys_objs[I915_MAX_PHYS_OBJECT]; } mm; struct sdvo_device_mapping sdvo_mappings[2]; + /* indicate whether the LVDS_BORDER should be enabled or not */ + unsigned int lvds_border_bits; /* Reclocking support */ bool render_reclock_avail; @@ -491,10 +586,7 @@ struct drm_i915_gem_object { * This is the same as gtt_space->start */ uint32_t gtt_offset; - /** - * Required alignment for the object - */ - uint32_t gtt_alignment; + /** * Fake offset for use by mmap(2) */ @@ -541,6 +633,11 @@ struct drm_i915_gem_object { * in an execbuffer object list. */ int in_execbuffer; + + /** + * Advice: are the backing pages purgeable? + */ + int madv; }; /** @@ -585,6 +682,8 @@ extern int i915_max_ioctl; extern unsigned int i915_fbpercrtc; extern unsigned int i915_powersave; +extern void i915_save_display(struct drm_device *dev); +extern void i915_restore_display(struct drm_device *dev); extern int i915_master_create(struct drm_device *dev, struct drm_master *master); extern void i915_master_destroy(struct drm_device *dev, struct drm_master *master); @@ -604,13 +703,16 @@ extern long i915_compat_ioctl(struct file *filp, unsigned int cmd, extern int i915_emit_box(struct drm_device *dev, struct drm_clip_rect *boxes, int i, int DR1, int DR4); +extern int i965_reset(struct drm_device *dev, u8 flags); /* i915_irq.c */ +void i915_hangcheck_elapsed(unsigned long data); extern int i915_irq_emit(struct drm_device *dev, void *data, struct drm_file *file_priv); extern int i915_irq_wait(struct drm_device *dev, void *data, struct drm_file *file_priv); void i915_user_irq_get(struct drm_device *dev); +void i915_trace_irq_get(struct drm_device *dev, u32 seqno); void i915_user_irq_put(struct drm_device *dev); extern void i915_enable_interrupt (struct drm_device *dev); @@ -676,6 +778,8 @@ int i915_gem_busy_ioctl(struct drm_device *dev, void *data, struct drm_file *file_priv); int i915_gem_throttle_ioctl(struct drm_device *dev, void *data, struct drm_file *file_priv); +int i915_gem_madvise_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv); int i915_gem_entervt_ioctl(struct drm_device *dev, void *data, struct drm_file *file_priv); int i915_gem_leavevt_ioctl(struct drm_device *dev, void *data, @@ -695,6 +799,7 @@ int i915_gem_object_unbind(struct drm_gem_object *obj); void i915_gem_release_mmap(struct drm_gem_object *obj); void i915_gem_lastclose(struct drm_device *dev); uint32_t i915_get_gem_seqno(struct drm_device *dev); +bool i915_seqno_passed(uint32_t seq1, uint32_t seq2); int i915_gem_object_get_fence_reg(struct drm_gem_object *obj); int i915_gem_object_put_fence_reg(struct drm_gem_object *obj); void i915_gem_retire_requests(struct drm_device *dev); @@ -720,6 +825,9 @@ int i915_gem_object_get_pages(struct drm_gem_object *obj); void i915_gem_object_put_pages(struct drm_gem_object *obj); void i915_gem_release(struct drm_device * dev, struct drm_file *file_priv); +void i915_gem_shrinker_init(void); +void i915_gem_shrinker_exit(void); + /* i915_gem_tiling.c */ void i915_gem_detect_bit_6_swizzle(struct drm_device *dev); void i915_gem_object_do_bit_17_swizzle(struct drm_gem_object *obj); @@ -767,6 +875,8 @@ static inline void opregion_enable_asle(struct drm_device *dev) { return; } extern void intel_modeset_init(struct drm_device *dev); extern void intel_modeset_cleanup(struct drm_device *dev); extern int intel_modeset_vga_set_state(struct drm_device *dev, bool state); +extern void i8xx_disable_fbc(struct drm_device *dev); +extern void g4x_disable_fbc(struct drm_device *dev); /** * Lock test for when it's just for synchronization of ring access. @@ -864,6 +974,7 @@ extern int i915_wait_ring(struct drm_device * dev, int n, const char *caller); (dev)->pci_device == 0x2E12 || \ (dev)->pci_device == 0x2E22 || \ (dev)->pci_device == 0x2E32 || \ + (dev)->pci_device == 0x2E42 || \ (dev)->pci_device == 0x0042 || \ (dev)->pci_device == 0x0046) @@ -876,6 +987,7 @@ extern int i915_wait_ring(struct drm_device * dev, int n, const char *caller); (dev)->pci_device == 0x2E12 || \ (dev)->pci_device == 0x2E22 || \ (dev)->pci_device == 0x2E32 || \ + (dev)->pci_device == 0x2E42 || \ IS_GM45(dev)) #define IS_IGDG(dev) ((dev)->pci_device == 0xa001) @@ -909,12 +1021,16 @@ extern int i915_wait_ring(struct drm_device * dev, int n, const char *caller); #define SUPPORTS_INTEGRATED_HDMI(dev) (IS_G4X(dev) || IS_IGDNG(dev)) #define SUPPORTS_INTEGRATED_DP(dev) (IS_G4X(dev) || IS_IGDNG(dev)) #define SUPPORTS_EDP(dev) (IS_IGDNG_M(dev)) -#define I915_HAS_HOTPLUG(dev) (IS_I945G(dev) || IS_I945GM(dev) || IS_I965G(dev)) +#define I915_HAS_HOTPLUG(dev) (IS_I945G(dev) || IS_I945GM(dev) || IS_G33(dev) || IS_I965G(dev)) /* dsparb controlled by hw only */ #define DSPARB_HWCONTROL(dev) (IS_G4X(dev) || IS_IGDNG(dev)) #define HAS_FW_BLC(dev) (IS_I9XX(dev) || IS_G4X(dev) || IS_IGDNG(dev)) #define HAS_PIPE_CXSR(dev) (IS_G4X(dev) || IS_IGDNG(dev)) +#define I915_HAS_FBC(dev) (IS_MOBILE(dev) && \ + (IS_I9XX(dev) || IS_GM45(dev)) && \ + !IS_IGD(dev) && \ + !IS_IGDNG(dev)) #define PRIMARY_RINGBUFFER_SIZE (128*1024) diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index c673171..abfc27b 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -29,6 +29,7 @@ #include "drm.h" #include "i915_drm.h" #include "i915_drv.h" +#include "i915_trace.h" #include "intel_drv.h" #include <linux/swap.h> #include <linux/pci.h> @@ -48,11 +49,15 @@ static int i915_gem_object_wait_rendering(struct drm_gem_object *obj); static int i915_gem_object_bind_to_gtt(struct drm_gem_object *obj, unsigned alignment); static void i915_gem_clear_fence_reg(struct drm_gem_object *obj); -static int i915_gem_evict_something(struct drm_device *dev); +static int i915_gem_evict_something(struct drm_device *dev, int min_size); +static int i915_gem_evict_from_inactive_list(struct drm_device *dev); static int i915_gem_phys_pwrite(struct drm_device *dev, struct drm_gem_object *obj, struct drm_i915_gem_pwrite *args, struct drm_file *file_priv); +static LIST_HEAD(shrink_list); +static DEFINE_SPINLOCK(shrink_list_lock); + int i915_gem_do_init(struct drm_device *dev, unsigned long start, unsigned long end) { @@ -316,6 +321,45 @@ fail_unlock: return ret; } +static inline gfp_t +i915_gem_object_get_page_gfp_mask (struct drm_gem_object *obj) +{ + return mapping_gfp_mask(obj->filp->f_path.dentry->d_inode->i_mapping); +} + +static inline void +i915_gem_object_set_page_gfp_mask (struct drm_gem_object *obj, gfp_t gfp) +{ + mapping_set_gfp_mask(obj->filp->f_path.dentry->d_inode->i_mapping, gfp); +} + +static int +i915_gem_object_get_pages_or_evict(struct drm_gem_object *obj) +{ + int ret; + + ret = i915_gem_object_get_pages(obj); + + /* If we've insufficient memory to map in the pages, attempt + * to make some space by throwing out some old buffers. + */ + if (ret == -ENOMEM) { + struct drm_device *dev = obj->dev; + gfp_t gfp; + + ret = i915_gem_evict_something(dev, obj->size); + if (ret) + return ret; + + gfp = i915_gem_object_get_page_gfp_mask(obj); + i915_gem_object_set_page_gfp_mask(obj, gfp & ~__GFP_NORETRY); + ret = i915_gem_object_get_pages(obj); + i915_gem_object_set_page_gfp_mask (obj, gfp); + } + + return ret; +} + /** * This is the fallback shmem pread path, which allocates temporary storage * in kernel space to copy_to_user into outside of the struct_mutex, so we @@ -367,8 +411,8 @@ i915_gem_shmem_pread_slow(struct drm_device *dev, struct drm_gem_object *obj, mutex_lock(&dev->struct_mutex); - ret = i915_gem_object_get_pages(obj); - if (ret != 0) + ret = i915_gem_object_get_pages_or_evict(obj); + if (ret) goto fail_unlock; ret = i915_gem_object_set_cpu_read_domain_range(obj, args->offset, @@ -842,8 +886,8 @@ i915_gem_shmem_pwrite_slow(struct drm_device *dev, struct drm_gem_object *obj, mutex_lock(&dev->struct_mutex); - ret = i915_gem_object_get_pages(obj); - if (ret != 0) + ret = i915_gem_object_get_pages_or_evict(obj); + if (ret) goto fail_unlock; ret = i915_gem_object_set_to_cpu_domain(obj, 1); @@ -1155,28 +1199,22 @@ int i915_gem_fault(struct vm_area_struct *vma, struct vm_fault *vmf) /* Now bind it into the GTT if needed */ mutex_lock(&dev->struct_mutex); if (!obj_priv->gtt_space) { - ret = i915_gem_object_bind_to_gtt(obj, obj_priv->gtt_alignment); - if (ret) { - mutex_unlock(&dev->struct_mutex); - return VM_FAULT_SIGBUS; - } - - ret = i915_gem_object_set_to_gtt_domain(obj, write); - if (ret) { - mutex_unlock(&dev->struct_mutex); - return VM_FAULT_SIGBUS; - } + ret = i915_gem_object_bind_to_gtt(obj, 0); + if (ret) + goto unlock; list_add_tail(&obj_priv->list, &dev_priv->mm.inactive_list); + + ret = i915_gem_object_set_to_gtt_domain(obj, write); + if (ret) + goto unlock; } /* Need a new fence register? */ if (obj_priv->tiling_mode != I915_TILING_NONE) { ret = i915_gem_object_get_fence_reg(obj); - if (ret) { - mutex_unlock(&dev->struct_mutex); - return VM_FAULT_SIGBUS; - } + if (ret) + goto unlock; } pfn = ((dev->agp->base + obj_priv->gtt_offset) >> PAGE_SHIFT) + @@ -1184,18 +1222,18 @@ int i915_gem_fault(struct vm_area_struct *vma, struct vm_fault *vmf) /* Finally, remap it using the new GTT offset */ ret = vm_insert_pfn(vma, (unsigned long)vmf->virtual_address, pfn); - +unlock: mutex_unlock(&dev->struct_mutex); switch (ret) { + case 0: + case -ERESTARTSYS: + return VM_FAULT_NOPAGE; case -ENOMEM: case -EAGAIN: return VM_FAULT_OOM; - case -EFAULT: - case -EINVAL: - return VM_FAULT_SIGBUS; default: - return VM_FAULT_NOPAGE; + return VM_FAULT_SIGBUS; } } @@ -1388,6 +1426,14 @@ i915_gem_mmap_gtt_ioctl(struct drm_device *dev, void *data, obj_priv = obj->driver_private; + if (obj_priv->madv != I915_MADV_WILLNEED) { + DRM_ERROR("Attempting to mmap a purgeable buffer\n"); + drm_gem_object_unreference(obj); + mutex_unlock(&dev->struct_mutex); + return -EINVAL; + } + + if (!obj_priv->mmap_offset) { ret = i915_gem_create_mmap_offset(obj); if (ret) { @@ -1399,22 +1445,12 @@ i915_gem_mmap_gtt_ioctl(struct drm_device *dev, void *data, args->offset = obj_priv->mmap_offset; - obj_priv->gtt_alignment = i915_gem_get_gtt_alignment(obj); - - /* Make sure the alignment is correct for fence regs etc */ - if (obj_priv->agp_mem && - (obj_priv->gtt_offset & (obj_priv->gtt_alignment - 1))) { - drm_gem_object_unreference(obj); - mutex_unlock(&dev->struct_mutex); - return -EINVAL; - } - /* * Pull it into the GTT so that we have a page list (makes the * initial fault faster and any subsequent flushing possible). */ if (!obj_priv->agp_mem) { - ret = i915_gem_object_bind_to_gtt(obj, obj_priv->gtt_alignment); + ret = i915_gem_object_bind_to_gtt(obj, 0); if (ret) { drm_gem_object_unreference(obj); mutex_unlock(&dev->struct_mutex); @@ -1437,6 +1473,7 @@ i915_gem_object_put_pages(struct drm_gem_object *obj) int i; BUG_ON(obj_priv->pages_refcount == 0); + BUG_ON(obj_priv->madv == __I915_MADV_PURGED); if (--obj_priv->pages_refcount != 0) return; @@ -1444,13 +1481,21 @@ i915_gem_object_put_pages(struct drm_gem_object *obj) if (obj_priv->tiling_mode != I915_TILING_NONE) i915_gem_object_save_bit_17_swizzle(obj); - for (i = 0; i < page_count; i++) - if (obj_priv->pages[i] != NULL) { - if (obj_priv->dirty) - set_page_dirty(obj_priv->pages[i]); + if (obj_priv->madv == I915_MADV_DONTNEED) + obj_priv->dirty = 0; + + for (i = 0; i < page_count; i++) { + if (obj_priv->pages[i] == NULL) + break; + + if (obj_priv->dirty) + set_page_dirty(obj_priv->pages[i]); + + if (obj_priv->madv == I915_MADV_WILLNEED) mark_page_accessed(obj_priv->pages[i]); - page_cache_release(obj_priv->pages[i]); - } + + page_cache_release(obj_priv->pages[i]); + } obj_priv->dirty = 0; drm_free_large(obj_priv->pages); @@ -1489,6 +1534,26 @@ i915_gem_object_move_to_flushing(struct drm_gem_object *obj) obj_priv->last_rendering_seqno = 0; } +/* Immediately discard the backing storage */ +static void +i915_gem_object_truncate(struct drm_gem_object *obj) +{ + struct drm_i915_gem_object *obj_priv = obj->driver_private; + struct inode *inode; + + inode = obj->filp->f_path.dentry->d_inode; + if (inode->i_op->truncate) + inode->i_op->truncate (inode); + + obj_priv->madv = __I915_MADV_PURGED; +} + +static inline int +i915_gem_object_is_purgeable(struct drm_i915_gem_object *obj_priv) +{ + return obj_priv->madv == I915_MADV_DONTNEED; +} + static void i915_gem_object_move_to_inactive(struct drm_gem_object *obj) { @@ -1577,15 +1642,24 @@ i915_add_request(struct drm_device *dev, struct drm_file *file_priv, if ((obj->write_domain & flush_domains) == obj->write_domain) { + uint32_t old_write_domain = obj->write_domain; + obj->write_domain = 0; i915_gem_object_move_to_active(obj, seqno); + + trace_i915_gem_object_change_domain(obj, + obj->read_domains, + old_write_domain); } } } - if (was_empty && !dev_priv->mm.suspended) - queue_delayed_work(dev_priv->wq, &dev_priv->mm.retire_work, HZ); + if (!dev_priv->mm.suspended) { + mod_timer(&dev_priv->hangcheck_timer, jiffies + DRM_I915_HANGCHECK_PERIOD); + if (was_empty) + queue_delayed_work(dev_priv->wq, &dev_priv->mm.retire_work, HZ); + } return seqno; } @@ -1623,6 +1697,8 @@ i915_gem_retire_request(struct drm_device *dev, { drm_i915_private_t *dev_priv = dev->dev_private; + trace_i915_gem_request_retire(dev, request->seqno); + /* Move any buffers on the active list that are no longer referenced * by the ringbuffer to the flushing/inactive lists as appropriate. */ @@ -1671,7 +1747,7 @@ out: /** * Returns true if seq1 is later than seq2. */ -static int +bool i915_seqno_passed(uint32_t seq1, uint32_t seq2) { return (int32_t)(seq1 - seq2) >= 0; @@ -1694,7 +1770,7 @@ i915_gem_retire_requests(struct drm_device *dev) drm_i915_private_t *dev_priv = dev->dev_private; uint32_t seqno; - if (!dev_priv->hw_status_page) + if (!dev_priv->hw_status_page || list_empty(&dev_priv->mm.request_list)) return; seqno = i915_get_gem_seqno(dev); @@ -1709,7 +1785,7 @@ i915_gem_retire_requests(struct drm_device *dev) retiring_seqno = request->seqno; if (i915_seqno_passed(seqno, retiring_seqno) || - dev_priv->mm.wedged) { + atomic_read(&dev_priv->mm.wedged)) { i915_gem_retire_request(dev, request); list_del(&request->list); @@ -1718,6 +1794,12 @@ i915_gem_retire_requests(struct drm_device *dev) } else break; } + + if (unlikely (dev_priv->trace_irq_seqno && + i915_seqno_passed(dev_priv->trace_irq_seqno, seqno))) { + i915_user_irq_put(dev); + dev_priv->trace_irq_seqno = 0; + } } void @@ -1751,6 +1833,9 @@ i915_wait_request(struct drm_device *dev, uint32_t seqno) BUG_ON(seqno == 0); + if (atomic_read(&dev_priv->mm.wedged)) + return -EIO; + if (!i915_seqno_passed(i915_get_gem_seqno(dev), seqno)) { if (IS_IGDNG(dev)) ier = I915_READ(DEIER) | I915_READ(GTIER); @@ -1763,16 +1848,20 @@ i915_wait_request(struct drm_device *dev, uint32_t seqno) i915_driver_irq_postinstall(dev); } + trace_i915_gem_request_wait_begin(dev, seqno); + dev_priv->mm.waiting_gem_seqno = seqno; i915_user_irq_get(dev); ret = wait_event_interruptible(dev_priv->irq_queue, i915_seqno_passed(i915_get_gem_seqno(dev), seqno) || - dev_priv->mm.wedged); + atomic_read(&dev_priv->mm.wedged)); i915_user_irq_put(dev); dev_priv->mm.waiting_gem_seqno = 0; + + trace_i915_gem_request_wait_end(dev, seqno); } - if (dev_priv->mm.wedged) + if (atomic_read(&dev_priv->mm.wedged)) ret = -EIO; if (ret && ret != -ERESTARTSYS) @@ -1803,6 +1892,8 @@ i915_gem_flush(struct drm_device *dev, DRM_INFO("%s: invalidate %08x flush %08x\n", __func__, invalidate_domains, flush_domains); #endif + trace_i915_gem_request_flush(dev, dev_priv->mm.next_gem_seqno, + invalidate_domains, flush_domains); if (flush_domains & I915_GEM_DOMAIN_CPU) drm_agp_chipset_flush(dev); @@ -1915,6 +2006,12 @@ i915_gem_object_unbind(struct drm_gem_object *obj) return -EINVAL; } + /* blow away mappings if mapped through GTT */ + i915_gem_release_mmap(obj); + + if (obj_priv->fence_reg != I915_FENCE_REG_NONE) + i915_gem_clear_fence_reg(obj); + /* Move the object to the CPU domain to ensure that * any possible CPU writes while it's not in the GTT * are flushed when we go to remap it. This will @@ -1928,21 +2025,16 @@ i915_gem_object_unbind(struct drm_gem_object *obj) return ret; } + BUG_ON(obj_priv->active); + if (obj_priv->agp_mem != NULL) { drm_unbind_agp(obj_priv->agp_mem); drm_free_agp(obj_priv->agp_mem, obj->size / PAGE_SIZE); obj_priv->agp_mem = NULL; } - BUG_ON(obj_priv->active); - - /* blow away mappings if mapped through GTT */ - i915_gem_release_mmap(obj); - - if (obj_priv->fence_reg != I915_FENCE_REG_NONE) - i915_gem_clear_fence_reg(obj); - i915_gem_object_put_pages(obj); + BUG_ON(obj_priv->pages_refcount); if (obj_priv->gtt_space) { atomic_dec(&dev->gtt_count); @@ -1956,40 +2048,113 @@ i915_gem_object_unbind(struct drm_gem_object *obj) if (!list_empty(&obj_priv->list)) list_del_init(&obj_priv->list); + if (i915_gem_object_is_purgeable(obj_priv)) + i915_gem_object_truncate(obj); + + trace_i915_gem_object_unbind(obj); + + return 0; +} + +static struct drm_gem_object * +i915_gem_find_inactive_object(struct drm_device *dev, int min_size) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_gem_object *obj_priv; + struct drm_gem_object *best = NULL; + struct drm_gem_object *first = NULL; + + /* Try to find the smallest clean object */ + list_for_each_entry(obj_priv, &dev_priv->mm.inactive_list, list) { + struct drm_gem_object *obj = obj_priv->obj; + if (obj->size >= min_size) { + if ((!obj_priv->dirty || + i915_gem_object_is_purgeable(obj_priv)) && + (!best || obj->size < best->size)) { + best = obj; + if (best->size == min_size) + return best; + } + if (!first) + first = obj; + } + } + + return best ? best : first; +} + +static int +i915_gem_evict_everything(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + uint32_t seqno; + int ret; + bool lists_empty; + + spin_lock(&dev_priv->mm.active_list_lock); + lists_empty = (list_empty(&dev_priv->mm.inactive_list) && + list_empty(&dev_priv->mm.flushing_list) && + list_empty(&dev_priv->mm.active_list)); + spin_unlock(&dev_priv->mm.active_list_lock); + + if (lists_empty) + return -ENOSPC; + + /* Flush everything (on to the inactive lists) and evict */ + i915_gem_flush(dev, I915_GEM_GPU_DOMAINS, I915_GEM_GPU_DOMAINS); + seqno = i915_add_request(dev, NULL, I915_GEM_GPU_DOMAINS); + if (seqno == 0) + return -ENOMEM; + + ret = i915_wait_request(dev, seqno); + if (ret) + return ret; + + ret = i915_gem_evict_from_inactive_list(dev); + if (ret) + return ret; + + spin_lock(&dev_priv->mm.active_list_lock); + lists_empty = (list_empty(&dev_priv->mm.inactive_list) && + list_empty(&dev_priv->mm.flushing_list) && + list_empty(&dev_priv->mm.active_list)); + spin_unlock(&dev_priv->mm.active_list_lock); + BUG_ON(!lists_empty); + return 0; } static int -i915_gem_evict_something(struct drm_device *dev) +i915_gem_evict_something(struct drm_device *dev, int min_size) { drm_i915_private_t *dev_priv = dev->dev_private; struct drm_gem_object *obj; - struct drm_i915_gem_object *obj_priv; - int ret = 0; + int ret; for (;;) { + i915_gem_retire_requests(dev); + /* If there's an inactive buffer available now, grab it * and be done. */ - if (!list_empty(&dev_priv->mm.inactive_list)) { - obj_priv = list_first_entry(&dev_priv->mm.inactive_list, - struct drm_i915_gem_object, - list); - obj = obj_priv->obj; - BUG_ON(obj_priv->pin_count != 0); + obj = i915_gem_find_inactive_object(dev, min_size); + if (obj) { + struct drm_i915_gem_object *obj_priv; + #if WATCH_LRU DRM_INFO("%s: evicting %p\n", __func__, obj); #endif + obj_priv = obj->driver_private; + BUG_ON(obj_priv->pin_count != 0); BUG_ON(obj_priv->active); /* Wait on the rendering and unbind the buffer. */ - ret = i915_gem_object_unbind(obj); - break; + return i915_gem_object_unbind(obj); } /* If we didn't get anything, but the ring is still processing - * things, wait for one of those things to finish and hopefully - * leave us a buffer to evict. + * things, wait for the next to finish and hopefully leave us + * a buffer to evict. */ if (!list_empty(&dev_priv->mm.request_list)) { struct drm_i915_gem_request *request; @@ -2000,16 +2165,9 @@ i915_gem_evict_something(struct drm_device *dev) ret = i915_wait_request(dev, request->seqno); if (ret) - break; + return ret; - /* if waiting caused an object to become inactive, - * then loop around and wait for it. Otherwise, we - * assume that waiting freed and unbound something, - * so there should now be some space in the GTT - */ - if (!list_empty(&dev_priv->mm.inactive_list)) - continue; - break; + continue; } /* If we didn't have anything on the request list but there @@ -2018,46 +2176,44 @@ i915_gem_evict_something(struct drm_device *dev) * will get moved to inactive. */ if (!list_empty(&dev_priv->mm.flushing_list)) { - obj_priv = list_first_entry(&dev_priv->mm.flushing_list, - struct drm_i915_gem_object, - list); - obj = obj_priv->obj; + struct drm_i915_gem_object *obj_priv; - i915_gem_flush(dev, - obj->write_domain, - obj->write_domain); - i915_add_request(dev, NULL, obj->write_domain); + /* Find an object that we can immediately reuse */ + list_for_each_entry(obj_priv, &dev_priv->mm.flushing_list, list) { + obj = obj_priv->obj; + if (obj->size >= min_size) + break; - obj = NULL; - continue; + obj = NULL; + } + + if (obj != NULL) { + uint32_t seqno; + + i915_gem_flush(dev, + obj->write_domain, + obj->write_domain); + seqno = i915_add_request(dev, NULL, obj->write_domain); + if (seqno == 0) + return -ENOMEM; + + ret = i915_wait_request(dev, seqno); + if (ret) + return ret; + + continue; + } } - DRM_ERROR("inactive empty %d request empty %d " - "flushing empty %d\n", - list_empty(&dev_priv->mm.inactive_list), - list_empty(&dev_priv->mm.request_list), - list_empty(&dev_priv->mm.flushing_list)); - /* If we didn't do any of the above, there's nothing to be done - * and we just can't fit it in. + /* If we didn't do any of the above, there's no single buffer + * large enough to swap out for the new one, so just evict + * everything and start again. (This should be rare.) */ - return -ENOSPC; + if (!list_empty (&dev_priv->mm.inactive_list)) + return i915_gem_evict_from_inactive_list(dev); + else + return i915_gem_evict_everything(dev); } - return ret; -} - -static int -i915_gem_evict_everything(struct drm_device *dev) -{ - int ret; - - for (;;) { - ret = i915_gem_evict_something(dev); - if (ret != 0) - break; - } - if (ret == -ENOSPC) - return 0; - return ret; } int @@ -2080,7 +2236,6 @@ i915_gem_object_get_pages(struct drm_gem_object *obj) BUG_ON(obj_priv->pages != NULL); obj_priv->pages = drm_calloc_large(page_count, sizeof(struct page *)); if (obj_priv->pages == NULL) { - DRM_ERROR("Faled to allocate page list\n"); obj_priv->pages_refcount--; return -ENOMEM; } @@ -2091,7 +2246,6 @@ i915_gem_object_get_pages(struct drm_gem_object *obj) page = read_mapping_page(mapping, i, NULL); if (IS_ERR(page)) { ret = PTR_ERR(page); - DRM_ERROR("read_mapping_page failed: %d\n", ret); i915_gem_object_put_pages(obj); return ret; } @@ -2328,6 +2482,8 @@ i915_gem_object_get_fence_reg(struct drm_gem_object *obj) else i830_write_fence_reg(reg); + trace_i915_gem_object_get_fence(obj, i, obj_priv->tiling_mode); + return 0; } @@ -2410,10 +2566,17 @@ i915_gem_object_bind_to_gtt(struct drm_gem_object *obj, unsigned alignment) drm_i915_private_t *dev_priv = dev->dev_private; struct drm_i915_gem_object *obj_priv = obj->driver_private; struct drm_mm_node *free_space; - int page_count, ret; + bool retry_alloc = false; + int ret; if (dev_priv->mm.suspended) return -EBUSY; + + if (obj_priv->madv != I915_MADV_WILLNEED) { + DRM_ERROR("Attempting to bind a purgeable object\n"); + return -EINVAL; + } + if (alignment == 0) alignment = i915_gem_get_gtt_alignment(obj); if (alignment & (i915_gem_get_gtt_alignment(obj) - 1)) { @@ -2433,30 +2596,16 @@ i915_gem_object_bind_to_gtt(struct drm_gem_object *obj, unsigned alignment) } } if (obj_priv->gtt_space == NULL) { - bool lists_empty; - /* If the gtt is empty and we're still having trouble * fitting our object in, we're out of memory. */ #if WATCH_LRU DRM_INFO("%s: GTT full, evicting something\n", __func__); #endif - spin_lock(&dev_priv->mm.active_list_lock); - lists_empty = (list_empty(&dev_priv->mm.inactive_list) && - list_empty(&dev_priv->mm.flushing_list) && - list_empty(&dev_priv->mm.active_list)); - spin_unlock(&dev_priv->mm.active_list_lock); - if (lists_empty) { - DRM_ERROR("GTT full, but LRU list empty\n"); - return -ENOSPC; - } - - ret = i915_gem_evict_something(dev); - if (ret != 0) { - if (ret != -ERESTARTSYS) - DRM_ERROR("Failed to evict a buffer %d\n", ret); + ret = i915_gem_evict_something(dev, obj->size); + if (ret) return ret; - } + goto search_free; } @@ -2464,27 +2613,56 @@ i915_gem_object_bind_to_gtt(struct drm_gem_object *obj, unsigned alignment) DRM_INFO("Binding object of size %zd at 0x%08x\n", obj->size, obj_priv->gtt_offset); #endif + if (retry_alloc) { + i915_gem_object_set_page_gfp_mask (obj, + i915_gem_object_get_page_gfp_mask (obj) & ~__GFP_NORETRY); + } ret = i915_gem_object_get_pages(obj); + if (retry_alloc) { + i915_gem_object_set_page_gfp_mask (obj, + i915_gem_object_get_page_gfp_mask (obj) | __GFP_NORETRY); + } if (ret) { drm_mm_put_block(obj_priv->gtt_space); obj_priv->gtt_space = NULL; + + if (ret == -ENOMEM) { + /* first try to clear up some space from the GTT */ + ret = i915_gem_evict_something(dev, obj->size); + if (ret) { + /* now try to shrink everyone else */ + if (! retry_alloc) { + retry_alloc = true; + goto search_free; + } + + return ret; + } + + goto search_free; + } + return ret; } - page_count = obj->size / PAGE_SIZE; /* Create an AGP memory structure pointing at our pages, and bind it * into the GTT. */ obj_priv->agp_mem = drm_agp_bind_pages(dev, obj_priv->pages, - page_count, + obj->size >> PAGE_SHIFT, obj_priv->gtt_offset, obj_priv->agp_type); if (obj_priv->agp_mem == NULL) { i915_gem_object_put_pages(obj); drm_mm_put_block(obj_priv->gtt_space); obj_priv->gtt_space = NULL; - return -ENOMEM; + + ret = i915_gem_evict_something(dev, obj->size); + if (ret) + return ret; + + goto search_free; } atomic_inc(&dev->gtt_count); atomic_add(obj->size, &dev->gtt_memory); @@ -2496,6 +2674,8 @@ i915_gem_object_bind_to_gtt(struct drm_gem_object *obj, unsigned alignment) BUG_ON(obj->read_domains & I915_GEM_GPU_DOMAINS); BUG_ON(obj->write_domain & I915_GEM_GPU_DOMAINS); + trace_i915_gem_object_bind(obj, obj_priv->gtt_offset); + return 0; } @@ -2511,15 +2691,7 @@ i915_gem_clflush_object(struct drm_gem_object *obj) if (obj_priv->pages == NULL) return; - /* XXX: The 865 in particular appears to be weird in how it handles - * cache flushing. We haven't figured it out, but the - * clflush+agp_chipset_flush doesn't appear to successfully get the - * data visible to the PGU, while wbinvd + agp_chipset_flush does. - */ - if (IS_I865G(obj->dev)) { - wbinvd(); - return; - } + trace_i915_gem_object_clflush(obj); drm_clflush_pages(obj_priv->pages, obj->size / PAGE_SIZE); } @@ -2530,21 +2702,29 @@ i915_gem_object_flush_gpu_write_domain(struct drm_gem_object *obj) { struct drm_device *dev = obj->dev; uint32_t seqno; + uint32_t old_write_domain; if ((obj->write_domain & I915_GEM_GPU_DOMAINS) == 0) return; /* Queue the GPU write cache flushing we need. */ + old_write_domain = obj->write_domain; i915_gem_flush(dev, 0, obj->write_domain); seqno = i915_add_request(dev, NULL, obj->write_domain); obj->write_domain = 0; i915_gem_object_move_to_active(obj, seqno); + + trace_i915_gem_object_change_domain(obj, + obj->read_domains, + old_write_domain); } /** Flushes the GTT write domain for the object if it's dirty. */ static void i915_gem_object_flush_gtt_write_domain(struct drm_gem_object *obj) { + uint32_t old_write_domain; + if (obj->write_domain != I915_GEM_DOMAIN_GTT) return; @@ -2552,7 +2732,12 @@ i915_gem_object_flush_gtt_write_domain(struct drm_gem_object *obj) * to it immediately go to main memory as far as we know, so there's * no chipset flush. It also doesn't land in render cache. */ + old_write_domain = obj->write_domain; obj->write_domain = 0; + + trace_i915_gem_object_change_domain(obj, + obj->read_domains, + old_write_domain); } /** Flushes the CPU write domain for the object if it's dirty. */ @@ -2560,13 +2745,19 @@ static void i915_gem_object_flush_cpu_write_domain(struct drm_gem_object *obj) { struct drm_device *dev = obj->dev; + uint32_t old_write_domain; if (obj->write_domain != I915_GEM_DOMAIN_CPU) return; i915_gem_clflush_object(obj); drm_agp_chipset_flush(dev); + old_write_domain = obj->write_domain; obj->write_domain = 0; + + trace_i915_gem_object_change_domain(obj, + obj->read_domains, + old_write_domain); } /** @@ -2579,6 +2770,7 @@ int i915_gem_object_set_to_gtt_domain(struct drm_gem_object *obj, int write) { struct drm_i915_gem_object *obj_priv = obj->driver_private; + uint32_t old_write_domain, old_read_domains; int ret; /* Not valid to be called on unbound objects. */ @@ -2591,6 +2783,9 @@ i915_gem_object_set_to_gtt_domain(struct drm_gem_object *obj, int write) if (ret != 0) return ret; + old_write_domain = obj->write_domain; + old_read_domains = obj->read_domains; + /* If we're writing through the GTT domain, then CPU and GPU caches * will need to be invalidated at next use. */ @@ -2609,6 +2804,10 @@ i915_gem_object_set_to_gtt_domain(struct drm_gem_object *obj, int write) obj_priv->dirty = 1; } + trace_i915_gem_object_change_domain(obj, + old_read_domains, + old_write_domain); + return 0; } @@ -2621,6 +2820,7 @@ i915_gem_object_set_to_gtt_domain(struct drm_gem_object *obj, int write) static int i915_gem_object_set_to_cpu_domain(struct drm_gem_object *obj, int write) { + uint32_t old_write_domain, old_read_domains; int ret; i915_gem_object_flush_gpu_write_domain(obj); @@ -2636,6 +2836,9 @@ i915_gem_object_set_to_cpu_domain(struct drm_gem_object *obj, int write) */ i915_gem_object_set_to_full_cpu_read_domain(obj); + old_write_domain = obj->write_domain; + old_read_domains = obj->read_domains; + /* Flush the CPU cache if it's still invalid. */ if ((obj->read_domains & I915_GEM_DOMAIN_CPU) == 0) { i915_gem_clflush_object(obj); @@ -2656,6 +2859,10 @@ i915_gem_object_set_to_cpu_domain(struct drm_gem_object *obj, int write) obj->write_domain = I915_GEM_DOMAIN_CPU; } + trace_i915_gem_object_change_domain(obj, + old_read_domains, + old_write_domain); + return 0; } @@ -2777,6 +2984,7 @@ i915_gem_object_set_to_gpu_domain(struct drm_gem_object *obj) struct drm_i915_gem_object *obj_priv = obj->driver_private; uint32_t invalidate_domains = 0; uint32_t flush_domains = 0; + uint32_t old_read_domains; BUG_ON(obj->pending_read_domains & I915_GEM_DOMAIN_CPU); BUG_ON(obj->pending_write_domain == I915_GEM_DOMAIN_CPU); @@ -2823,6 +3031,8 @@ i915_gem_object_set_to_gpu_domain(struct drm_gem_object *obj) i915_gem_clflush_object(obj); } + old_read_domains = obj->read_domains; + /* The actual obj->write_domain will be updated with * pending_write_domain after we emit the accumulated flush for all * of our domain changes in execbuffers (which clears objects' @@ -2841,6 +3051,10 @@ i915_gem_object_set_to_gpu_domain(struct drm_gem_object *obj) obj->read_domains, obj->write_domain, dev->invalidate_domains, dev->flush_domains); #endif + + trace_i915_gem_object_change_domain(obj, + old_read_domains, + obj->write_domain); } /** @@ -2893,6 +3107,7 @@ i915_gem_object_set_cpu_read_domain_range(struct drm_gem_object *obj, uint64_t offset, uint64_t size) { struct drm_i915_gem_object *obj_priv = obj->driver_private; + uint32_t old_read_domains; int i, ret; if (offset == 0 && size == obj->size) @@ -2939,8 +3154,13 @@ i915_gem_object_set_cpu_read_domain_range(struct drm_gem_object *obj, */ BUG_ON((obj->write_domain & ~I915_GEM_DOMAIN_CPU) != 0); + old_read_domains = obj->read_domains; obj->read_domains |= I915_GEM_DOMAIN_CPU; + trace_i915_gem_object_change_domain(obj, + old_read_domains, + obj->write_domain); + return 0; } @@ -2984,6 +3204,21 @@ i915_gem_object_pin_and_relocate(struct drm_gem_object *obj, } target_obj_priv = target_obj->driver_private; +#if WATCH_RELOC + DRM_INFO("%s: obj %p offset %08x target %d " + "read %08x write %08x gtt %08x " + "presumed %08x delta %08x\n", + __func__, + obj, + (int) reloc->offset, + (int) reloc->target_handle, + (int) reloc->read_domains, + (int) reloc->write_domain, + (int) target_obj_priv->gtt_offset, + (int) reloc->presumed_offset, + reloc->delta); +#endif + /* The target buffer should have appeared before us in the * exec_object list, so it should have a GTT space bound by now. */ @@ -2995,25 +3230,7 @@ i915_gem_object_pin_and_relocate(struct drm_gem_object *obj, return -EINVAL; } - if (reloc->offset > obj->size - 4) { - DRM_ERROR("Relocation beyond object bounds: " - "obj %p target %d offset %d size %d.\n", - obj, reloc->target_handle, - (int) reloc->offset, (int) obj->size); - drm_gem_object_unreference(target_obj); - i915_gem_object_unpin(obj); - return -EINVAL; - } - if (reloc->offset & 3) { - DRM_ERROR("Relocation not 4-byte aligned: " - "obj %p target %d offset %d.\n", - obj, reloc->target_handle, - (int) reloc->offset); - drm_gem_object_unreference(target_obj); - i915_gem_object_unpin(obj); - return -EINVAL; - } - + /* Validate that the target is in a valid r/w GPU domain */ if (reloc->write_domain & I915_GEM_DOMAIN_CPU || reloc->read_domains & I915_GEM_DOMAIN_CPU) { DRM_ERROR("reloc with read/write CPU domains: " @@ -3027,7 +3244,6 @@ i915_gem_object_pin_and_relocate(struct drm_gem_object *obj, i915_gem_object_unpin(obj); return -EINVAL; } - if (reloc->write_domain && target_obj->pending_write_domain && reloc->write_domain != target_obj->pending_write_domain) { DRM_ERROR("Write domain conflict: " @@ -3042,21 +3258,6 @@ i915_gem_object_pin_and_relocate(struct drm_gem_object *obj, return -EINVAL; } -#if WATCH_RELOC - DRM_INFO("%s: obj %p offset %08x target %d " - "read %08x write %08x gtt %08x " - "presumed %08x delta %08x\n", - __func__, - obj, - (int) reloc->offset, - (int) reloc->target_handle, - (int) reloc->read_domains, - (int) reloc->write_domain, - (int) target_obj_priv->gtt_offset, - (int) reloc->presumed_offset, - reloc->delta); -#endif - target_obj->pending_read_domains |= reloc->read_domains; target_obj->pending_write_domain |= reloc->write_domain; @@ -3068,6 +3269,37 @@ i915_gem_object_pin_and_relocate(struct drm_gem_object *obj, continue; } + /* Check that the relocation address is valid... */ + if (reloc->offset > obj->size - 4) { + DRM_ERROR("Relocation beyond object bounds: " + "obj %p target %d offset %d size %d.\n", + obj, reloc->target_handle, + (int) reloc->offset, (int) obj->size); + drm_gem_object_unreference(target_obj); + i915_gem_object_unpin(obj); + return -EINVAL; + } + if (reloc->offset & 3) { + DRM_ERROR("Relocation not 4-byte aligned: " + "obj %p target %d offset %d.\n", + obj, reloc->target_handle, + (int) reloc->offset); + drm_gem_object_unreference(target_obj); + i915_gem_object_unpin(obj); + return -EINVAL; + } + + /* and points to somewhere within the target object. */ + if (reloc->delta >= target_obj->size) { + DRM_ERROR("Relocation beyond target object bounds: " + "obj %p target %d delta %d size %d.\n", + obj, reloc->target_handle, + (int) reloc->delta, (int) target_obj->size); + drm_gem_object_unreference(target_obj); + i915_gem_object_unpin(obj); + return -EINVAL; + } + ret = i915_gem_object_set_to_gtt_domain(obj, 1); if (ret != 0) { drm_gem_object_unreference(target_obj); @@ -3126,6 +3358,8 @@ i915_dispatch_gem_execbuffer(struct drm_device *dev, exec_start = (uint32_t) exec_offset + exec->batch_start_offset; exec_len = (uint32_t) exec->batch_len; + trace_i915_gem_request_submit(dev, dev_priv->mm.next_gem_seqno + 1); + count = nbox ? nbox : 1; for (i = 0; i < count; i++) { @@ -3363,7 +3597,7 @@ i915_gem_execbuffer(struct drm_device *dev, void *data, i915_verify_inactive(dev, __FILE__, __LINE__); - if (dev_priv->mm.wedged) { + if (atomic_read(&dev_priv->mm.wedged)) { DRM_ERROR("Execbuf while wedged\n"); mutex_unlock(&dev->struct_mutex); ret = -EIO; @@ -3421,8 +3655,23 @@ i915_gem_execbuffer(struct drm_device *dev, void *data, /* error other than GTT full, or we've already tried again */ if (ret != -ENOSPC || pin_tries >= 1) { - if (ret != -ERESTARTSYS) - DRM_ERROR("Failed to pin buffers %d\n", ret); + if (ret != -ERESTARTSYS) { + unsigned long long total_size = 0; + for (i = 0; i < args->buffer_count; i++) + total_size += object_list[i]->size; + DRM_ERROR("Failed to pin buffer %d of %d, total %llu bytes: %d\n", + pinned+1, args->buffer_count, + total_size, ret); + DRM_ERROR("%d objects [%d pinned], " + "%d object bytes [%d pinned], " + "%d/%d gtt bytes\n", + atomic_read(&dev->object_count), + atomic_read(&dev->pin_count), + atomic_read(&dev->object_memory), + atomic_read(&dev->pin_memory), + atomic_read(&dev->gtt_memory), + dev->gtt_total); + } goto err; } @@ -3433,7 +3682,7 @@ i915_gem_execbuffer(struct drm_device *dev, void *data, /* evict everyone we can from the aperture */ ret = i915_gem_evict_everything(dev); - if (ret) + if (ret && ret != -ENOSPC) goto err; } @@ -3489,8 +3738,12 @@ i915_gem_execbuffer(struct drm_device *dev, void *data, for (i = 0; i < args->buffer_count; i++) { struct drm_gem_object *obj = object_list[i]; + uint32_t old_write_domain = obj->write_domain; obj->write_domain = obj->pending_write_domain; + trace_i915_gem_object_change_domain(obj, + obj->read_domains, + old_write_domain); } i915_verify_inactive(dev, __FILE__, __LINE__); @@ -3607,11 +3860,8 @@ i915_gem_object_pin(struct drm_gem_object *obj, uint32_t alignment) i915_verify_inactive(dev, __FILE__, __LINE__); if (obj_priv->gtt_space == NULL) { ret = i915_gem_object_bind_to_gtt(obj, alignment); - if (ret != 0) { - if (ret != -EBUSY && ret != -ERESTARTSYS) - DRM_ERROR("Failure to bind: %d\n", ret); + if (ret) return ret; - } } /* * Pre-965 chips need a fence register set up in order to @@ -3691,6 +3941,13 @@ i915_gem_pin_ioctl(struct drm_device *dev, void *data, } obj_priv = obj->driver_private; + if (obj_priv->madv != I915_MADV_WILLNEED) { + DRM_ERROR("Attempting to pin a purgeable buffer\n"); + drm_gem_object_unreference(obj); + mutex_unlock(&dev->struct_mutex); + return -EINVAL; + } + if (obj_priv->pin_filp != NULL && obj_priv->pin_filp != file_priv) { DRM_ERROR("Already pinned in i915_gem_pin_ioctl(): %d\n", args->handle); @@ -3803,6 +4060,56 @@ i915_gem_throttle_ioctl(struct drm_device *dev, void *data, return i915_gem_ring_throttle(dev, file_priv); } +int +i915_gem_madvise_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_i915_gem_madvise *args = data; + struct drm_gem_object *obj; + struct drm_i915_gem_object *obj_priv; + + switch (args->madv) { + case I915_MADV_DONTNEED: + case I915_MADV_WILLNEED: + break; + default: + return -EINVAL; + } + + obj = drm_gem_object_lookup(dev, file_priv, args->handle); + if (obj == NULL) { + DRM_ERROR("Bad handle in i915_gem_madvise_ioctl(): %d\n", + args->handle); + return -EBADF; + } + + mutex_lock(&dev->struct_mutex); + obj_priv = obj->driver_private; + + if (obj_priv->pin_count) { + drm_gem_object_unreference(obj); + mutex_unlock(&dev->struct_mutex); + + DRM_ERROR("Attempted i915_gem_madvise_ioctl() on a pinned object\n"); + return -EINVAL; + } + + if (obj_priv->madv != __I915_MADV_PURGED) + obj_priv->madv = args->madv; + + /* if the object is no longer bound, discard its backing storage */ + if (i915_gem_object_is_purgeable(obj_priv) && + obj_priv->gtt_space == NULL) + i915_gem_object_truncate(obj); + + args->retained = obj_priv->madv != __I915_MADV_PURGED; + + drm_gem_object_unreference(obj); + mutex_unlock(&dev->struct_mutex); + + return 0; +} + int i915_gem_init_object(struct drm_gem_object *obj) { struct drm_i915_gem_object *obj_priv; @@ -3827,6 +4134,9 @@ int i915_gem_init_object(struct drm_gem_object *obj) obj_priv->fence_reg = I915_FENCE_REG_NONE; INIT_LIST_HEAD(&obj_priv->list); INIT_LIST_HEAD(&obj_priv->fence_list); + obj_priv->madv = I915_MADV_WILLNEED; + + trace_i915_gem_object_create(obj); return 0; } @@ -3836,6 +4146,8 @@ void i915_gem_free_object(struct drm_gem_object *obj) struct drm_device *dev = obj->dev; struct drm_i915_gem_object *obj_priv = obj->driver_private; + trace_i915_gem_object_destroy(obj); + while (obj_priv->pin_count > 0) i915_gem_object_unpin(obj); @@ -3844,43 +4156,35 @@ void i915_gem_free_object(struct drm_gem_object *obj) i915_gem_object_unbind(obj); - i915_gem_free_mmap_offset(obj); + if (obj_priv->mmap_offset) + i915_gem_free_mmap_offset(obj); kfree(obj_priv->page_cpu_valid); kfree(obj_priv->bit_17); kfree(obj->driver_private); } -/** Unbinds all objects that are on the given buffer list. */ +/** Unbinds all inactive objects. */ static int -i915_gem_evict_from_list(struct drm_device *dev, struct list_head *head) +i915_gem_evict_from_inactive_list(struct drm_device *dev) { - struct drm_gem_object *obj; - struct drm_i915_gem_object *obj_priv; - int ret; + drm_i915_private_t *dev_priv = dev->dev_private; - while (!list_empty(head)) { - obj_priv = list_first_entry(head, - struct drm_i915_gem_object, - list); - obj = obj_priv->obj; + while (!list_empty(&dev_priv->mm.inactive_list)) { + struct drm_gem_object *obj; + int ret; - if (obj_priv->pin_count != 0) { - DRM_ERROR("Pinned object in unbind list\n"); - mutex_unlock(&dev->struct_mutex); - return -EINVAL; - } + obj = list_first_entry(&dev_priv->mm.inactive_list, + struct drm_i915_gem_object, + list)->obj; ret = i915_gem_object_unbind(obj); if (ret != 0) { - DRM_ERROR("Error unbinding object in LeaveVT: %d\n", - ret); - mutex_unlock(&dev->struct_mutex); + DRM_ERROR("Error unbinding object: %d\n", ret); return ret; } } - return 0; } @@ -3902,6 +4206,7 @@ i915_gem_idle(struct drm_device *dev) * We need to replace this with a semaphore, or something. */ dev_priv->mm.suspended = 1; + del_timer(&dev_priv->hangcheck_timer); /* Cancel the retire work handler, wait for it to finish if running */ @@ -3931,7 +4236,7 @@ i915_gem_idle(struct drm_device *dev) if (last_seqno == cur_seqno) { if (stuck++ > 100) { DRM_ERROR("hardware wedged\n"); - dev_priv->mm.wedged = 1; + atomic_set(&dev_priv->mm.wedged, 1); DRM_WAKEUP(&dev_priv->irq_queue); break; } @@ -3944,7 +4249,7 @@ i915_gem_idle(struct drm_device *dev) i915_gem_retire_requests(dev); spin_lock(&dev_priv->mm.active_list_lock); - if (!dev_priv->mm.wedged) { + if (!atomic_read(&dev_priv->mm.wedged)) { /* Active and flushing should now be empty as we've * waited for a sequence higher than any pending execbuffer */ @@ -3962,29 +4267,41 @@ i915_gem_idle(struct drm_device *dev) * the GPU domains and just stuff them onto inactive. */ while (!list_empty(&dev_priv->mm.active_list)) { - struct drm_i915_gem_object *obj_priv; + struct drm_gem_object *obj; + uint32_t old_write_domain; - obj_priv = list_first_entry(&dev_priv->mm.active_list, - struct drm_i915_gem_object, - list); - obj_priv->obj->write_domain &= ~I915_GEM_GPU_DOMAINS; - i915_gem_object_move_to_inactive(obj_priv->obj); + obj = list_first_entry(&dev_priv->mm.active_list, + struct drm_i915_gem_object, + list)->obj; + old_write_domain = obj->write_domain; + obj->write_domain &= ~I915_GEM_GPU_DOMAINS; + i915_gem_object_move_to_inactive(obj); + + trace_i915_gem_object_change_domain(obj, + obj->read_domains, + old_write_domain); } spin_unlock(&dev_priv->mm.active_list_lock); while (!list_empty(&dev_priv->mm.flushing_list)) { - struct drm_i915_gem_object *obj_priv; + struct drm_gem_object *obj; + uint32_t old_write_domain; - obj_priv = list_first_entry(&dev_priv->mm.flushing_list, - struct drm_i915_gem_object, - list); - obj_priv->obj->write_domain &= ~I915_GEM_GPU_DOMAINS; - i915_gem_object_move_to_inactive(obj_priv->obj); + obj = list_first_entry(&dev_priv->mm.flushing_list, + struct drm_i915_gem_object, + list)->obj; + old_write_domain = obj->write_domain; + obj->write_domain &= ~I915_GEM_GPU_DOMAINS; + i915_gem_object_move_to_inactive(obj); + + trace_i915_gem_object_change_domain(obj, + obj->read_domains, + old_write_domain); } /* Move all inactive buffers out of the GTT. */ - ret = i915_gem_evict_from_list(dev, &dev_priv->mm.inactive_list); + ret = i915_gem_evict_from_inactive_list(dev); WARN_ON(!list_empty(&dev_priv->mm.inactive_list)); if (ret) { mutex_unlock(&dev->struct_mutex); @@ -4206,9 +4523,9 @@ i915_gem_entervt_ioctl(struct drm_device *dev, void *data, if (drm_core_check_feature(dev, DRIVER_MODESET)) return 0; - if (dev_priv->mm.wedged) { + if (atomic_read(&dev_priv->mm.wedged)) { DRM_ERROR("Reenabling wedged hardware, good luck\n"); - dev_priv->mm.wedged = 0; + atomic_set(&dev_priv->mm.wedged, 0); } mutex_lock(&dev->struct_mutex); @@ -4274,6 +4591,10 @@ i915_gem_load(struct drm_device *dev) i915_gem_retire_work_handler); dev_priv->mm.next_gem_seqno = 1; + spin_lock(&shrink_list_lock); + list_add(&dev_priv->mm.shrink_list, &shrink_list); + spin_unlock(&shrink_list_lock); + /* Old X drivers will take 0-2 for front, back, depth buffers */ dev_priv->fence_reg_start = 3; @@ -4491,3 +4812,116 @@ void i915_gem_release(struct drm_device * dev, struct drm_file *file_priv) list_del_init(i915_file_priv->mm.request_list.next); mutex_unlock(&dev->struct_mutex); } + +static int +i915_gem_shrink(int nr_to_scan, gfp_t gfp_mask) +{ + drm_i915_private_t *dev_priv, *next_dev; + struct drm_i915_gem_object *obj_priv, *next_obj; + int cnt = 0; + int would_deadlock = 1; + + /* "fast-path" to count number of available objects */ + if (nr_to_scan == 0) { + spin_lock(&shrink_list_lock); + list_for_each_entry(dev_priv, &shrink_list, mm.shrink_list) { + struct drm_device *dev = dev_priv->dev; + + if (mutex_trylock(&dev->struct_mutex)) { + list_for_each_entry(obj_priv, + &dev_priv->mm.inactive_list, + list) + cnt++; + mutex_unlock(&dev->struct_mutex); + } + } + spin_unlock(&shrink_list_lock); + + return (cnt / 100) * sysctl_vfs_cache_pressure; + } + + spin_lock(&shrink_list_lock); + + /* first scan for clean buffers */ + list_for_each_entry_safe(dev_priv, next_dev, + &shrink_list, mm.shrink_list) { + struct drm_device *dev = dev_priv->dev; + + if (! mutex_trylock(&dev->struct_mutex)) + continue; + + spin_unlock(&shrink_list_lock); + + i915_gem_retire_requests(dev); + + list_for_each_entry_safe(obj_priv, next_obj, + &dev_priv->mm.inactive_list, + list) { + if (i915_gem_object_is_purgeable(obj_priv)) { + i915_gem_object_unbind(obj_priv->obj); + if (--nr_to_scan <= 0) + break; + } + } + + spin_lock(&shrink_list_lock); + mutex_unlock(&dev->struct_mutex); + + would_deadlock = 0; + + if (nr_to_scan <= 0) + break; + } + + /* second pass, evict/count anything still on the inactive list */ + list_for_each_entry_safe(dev_priv, next_dev, + &shrink_list, mm.shrink_list) { + struct drm_device *dev = dev_priv->dev; + + if (! mutex_trylock(&dev->struct_mutex)) + continue; + + spin_unlock(&shrink_list_lock); + + list_for_each_entry_safe(obj_priv, next_obj, + &dev_priv->mm.inactive_list, + list) { + if (nr_to_scan > 0) { + i915_gem_object_unbind(obj_priv->obj); + nr_to_scan--; + } else + cnt++; + } + + spin_lock(&shrink_list_lock); + mutex_unlock(&dev->struct_mutex); + + would_deadlock = 0; + } + + spin_unlock(&shrink_list_lock); + + if (would_deadlock) + return -1; + else if (cnt > 0) + return (cnt / 100) * sysctl_vfs_cache_pressure; + else + return 0; +} + +static struct shrinker shrinker = { + .shrink = i915_gem_shrink, + .seeks = DEFAULT_SEEKS, +}; + +__init void +i915_gem_shrinker_init(void) +{ + register_shrinker(&shrinker); +} + +__exit void +i915_gem_shrinker_exit(void) +{ + unregister_shrinker(&shrinker); +} diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c index 6c89f2f..aa7fd82 100644 --- a/drivers/gpu/drm/i915/i915_irq.c +++ b/drivers/gpu/drm/i915/i915_irq.c @@ -31,6 +31,7 @@ #include "drm.h" #include "i915_drm.h" #include "i915_drv.h" +#include "i915_trace.h" #include "intel_drv.h" #define MAX_NOPID ((u32)~0) @@ -253,10 +254,15 @@ irqreturn_t igdng_irq_handler(struct drm_device *dev) { drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; int ret = IRQ_NONE; - u32 de_iir, gt_iir; + u32 de_iir, gt_iir, de_ier; u32 new_de_iir, new_gt_iir; struct drm_i915_master_private *master_priv; + /* disable master interrupt before clearing iir */ + de_ier = I915_READ(DEIER); + I915_WRITE(DEIER, de_ier & ~DE_MASTER_IRQ_CONTROL); + (void)I915_READ(DEIER); + de_iir = I915_READ(DEIIR); gt_iir = I915_READ(GTIIR); @@ -279,7 +285,9 @@ irqreturn_t igdng_irq_handler(struct drm_device *dev) } if (gt_iir & GT_USER_INTERRUPT) { - dev_priv->mm.irq_gem_seqno = i915_get_gem_seqno(dev); + u32 seqno = i915_get_gem_seqno(dev); + dev_priv->mm.irq_gem_seqno = seqno; + trace_i915_gem_request_complete(dev, seqno); DRM_WAKEUP(&dev_priv->irq_queue); } @@ -287,6 +295,9 @@ irqreturn_t igdng_irq_handler(struct drm_device *dev) gt_iir = new_gt_iir; } + I915_WRITE(DEIER, de_ier); + (void)I915_READ(DEIER); + return ret; } @@ -302,12 +313,25 @@ static void i915_error_work_func(struct work_struct *work) drm_i915_private_t *dev_priv = container_of(work, drm_i915_private_t, error_work); struct drm_device *dev = dev_priv->dev; - char *event_string = "ERROR=1"; - char *envp[] = { event_string, NULL }; + char *error_event[] = { "ERROR=1", NULL }; + char *reset_event[] = { "RESET=1", NULL }; + char *reset_done_event[] = { "ERROR=0", NULL }; DRM_DEBUG("generating error event\n"); - - kobject_uevent_env(&dev->primary->kdev.kobj, KOBJ_CHANGE, envp); + kobject_uevent_env(&dev->primary->kdev.kobj, KOBJ_CHANGE, error_event); + + if (atomic_read(&dev_priv->mm.wedged)) { + if (IS_I965G(dev)) { + DRM_DEBUG("resetting chip\n"); + kobject_uevent_env(&dev->primary->kdev.kobj, KOBJ_CHANGE, reset_event); + if (!i965_reset(dev, GDRST_RENDER)) { + atomic_set(&dev_priv->mm.wedged, 0); + kobject_uevent_env(&dev->primary->kdev.kobj, KOBJ_CHANGE, reset_done_event); + } + } else { + printk("reboot required\n"); + } + } } /** @@ -372,7 +396,7 @@ out: * so userspace knows something bad happened (should trigger collection * of a ring dump etc.). */ -static void i915_handle_error(struct drm_device *dev) +static void i915_handle_error(struct drm_device *dev, bool wedged) { struct drm_i915_private *dev_priv = dev->dev_private; u32 eir = I915_READ(EIR); @@ -482,6 +506,16 @@ static void i915_handle_error(struct drm_device *dev) I915_WRITE(IIR, I915_RENDER_COMMAND_PARSER_ERROR_INTERRUPT); } + if (wedged) { + atomic_set(&dev_priv->mm.wedged, 1); + + /* + * Wakeup waiting processes so they don't hang + */ + printk("i915: Waking up sleeping processes\n"); + DRM_WAKEUP(&dev_priv->irq_queue); + } + queue_work(dev_priv->wq, &dev_priv->error_work); } @@ -527,7 +561,7 @@ irqreturn_t i915_driver_irq_handler(DRM_IRQ_ARGS) pipeb_stats = I915_READ(PIPEBSTAT); if (iir & I915_RENDER_COMMAND_PARSER_ERROR_INTERRUPT) - i915_handle_error(dev); + i915_handle_error(dev, false); /* * Clear the PIPE(A|B)STAT regs before the IIR @@ -599,8 +633,12 @@ irqreturn_t i915_driver_irq_handler(DRM_IRQ_ARGS) } if (iir & I915_USER_INTERRUPT) { - dev_priv->mm.irq_gem_seqno = i915_get_gem_seqno(dev); + u32 seqno = i915_get_gem_seqno(dev); + dev_priv->mm.irq_gem_seqno = seqno; + trace_i915_gem_request_complete(dev, seqno); DRM_WAKEUP(&dev_priv->irq_queue); + dev_priv->hangcheck_count = 0; + mod_timer(&dev_priv->hangcheck_timer, jiffies + DRM_I915_HANGCHECK_PERIOD); } if (pipea_stats & vblank_status) { @@ -695,6 +733,16 @@ void i915_user_irq_put(struct drm_device *dev) spin_unlock_irqrestore(&dev_priv->user_irq_lock, irqflags); } +void i915_trace_irq_get(struct drm_device *dev, u32 seqno) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + + if (dev_priv->trace_irq_seqno == 0) + i915_user_irq_get(dev); + + dev_priv->trace_irq_seqno = seqno; +} + static int i915_wait_irq(struct drm_device * dev, int irq_nr) { drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; @@ -880,6 +928,52 @@ int i915_vblank_swap(struct drm_device *dev, void *data, return -EINVAL; } +struct drm_i915_gem_request *i915_get_tail_request(struct drm_device *dev) { + drm_i915_private_t *dev_priv = dev->dev_private; + return list_entry(dev_priv->mm.request_list.prev, struct drm_i915_gem_request, list); +} + +/** + * This is called when the chip hasn't reported back with completed + * batchbuffers in a long time. The first time this is called we simply record + * ACTHD. If ACTHD hasn't changed by the time the hangcheck timer elapses + * again, we assume the chip is wedged and try to fix it. + */ +void i915_hangcheck_elapsed(unsigned long data) +{ + struct drm_device *dev = (struct drm_device *)data; + drm_i915_private_t *dev_priv = dev->dev_private; + uint32_t acthd; + + if (!IS_I965G(dev)) + acthd = I915_READ(ACTHD); + else + acthd = I915_READ(ACTHD_I965); + + /* If all work is done then ACTHD clearly hasn't advanced. */ + if (list_empty(&dev_priv->mm.request_list) || + i915_seqno_passed(i915_get_gem_seqno(dev), i915_get_tail_request(dev)->seqno)) { + dev_priv->hangcheck_count = 0; + return; + } + + if (dev_priv->last_acthd == acthd && dev_priv->hangcheck_count > 0) { + DRM_ERROR("Hangcheck timer elapsed... GPU hung\n"); + i915_handle_error(dev, true); + return; + } + + /* Reset timer case chip hangs without another request being added */ + mod_timer(&dev_priv->hangcheck_timer, jiffies + DRM_I915_HANGCHECK_PERIOD); + + if (acthd != dev_priv->last_acthd) + dev_priv->hangcheck_count = 0; + else + dev_priv->hangcheck_count++; + + dev_priv->last_acthd = acthd; +} + /* drm_dma.h hooks */ static void igdng_irq_preinstall(struct drm_device *dev) diff --git a/drivers/gpu/drm/i915/i915_opregion.c b/drivers/gpu/drm/i915/i915_opregion.c index e4b4e88..2d51935 100644 --- a/drivers/gpu/drm/i915/i915_opregion.c +++ b/drivers/gpu/drm/i915/i915_opregion.c @@ -148,6 +148,7 @@ static u32 asle_set_backlight(struct drm_device *dev, u32 bclp) struct drm_i915_private *dev_priv = dev->dev_private; struct opregion_asle *asle = dev_priv->opregion.asle; u32 blc_pwm_ctl, blc_pwm_ctl2; + u32 max_backlight, level, shift; if (!(bclp & ASLE_BCLP_VALID)) return ASLE_BACKLIGHT_FAIL; @@ -157,14 +158,25 @@ static u32 asle_set_backlight(struct drm_device *dev, u32 bclp) return ASLE_BACKLIGHT_FAIL; blc_pwm_ctl = I915_READ(BLC_PWM_CTL); - blc_pwm_ctl &= ~BACKLIGHT_DUTY_CYCLE_MASK; blc_pwm_ctl2 = I915_READ(BLC_PWM_CTL2); - if (blc_pwm_ctl2 & BLM_COMBINATION_MODE) + if (IS_I965G(dev) && (blc_pwm_ctl2 & BLM_COMBINATION_MODE)) pci_write_config_dword(dev->pdev, PCI_LBPC, bclp); - else - I915_WRITE(BLC_PWM_CTL, blc_pwm_ctl | ((bclp * 0x101)-1)); - + else { + if (IS_IGD(dev)) { + blc_pwm_ctl &= ~(BACKLIGHT_DUTY_CYCLE_MASK - 1); + max_backlight = (blc_pwm_ctl & BACKLIGHT_MODULATION_FREQ_MASK) >> + BACKLIGHT_MODULATION_FREQ_SHIFT; + shift = BACKLIGHT_DUTY_CYCLE_SHIFT + 1; + } else { + blc_pwm_ctl &= ~BACKLIGHT_DUTY_CYCLE_MASK; + max_backlight = ((blc_pwm_ctl & BACKLIGHT_MODULATION_FREQ_MASK) >> + BACKLIGHT_MODULATION_FREQ_SHIFT) * 2; + shift = BACKLIGHT_DUTY_CYCLE_SHIFT; + } + level = (bclp * max_backlight) / 255; + I915_WRITE(BLC_PWM_CTL, blc_pwm_ctl | (level << shift)); + } asle->cblv = (bclp*0x64)/0xff | ASLE_CBLV_VALID; return 0; diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h index 3f79635..1687edf 100644 --- a/drivers/gpu/drm/i915/i915_reg.h +++ b/drivers/gpu/drm/i915/i915_reg.h @@ -86,6 +86,10 @@ #define I915_GC_RENDER_CLOCK_200_MHZ (1 << 0) #define I915_GC_RENDER_CLOCK_333_MHZ (4 << 0) #define LBB 0xf4 +#define GDRST 0xc0 +#define GDRST_FULL (0<<2) +#define GDRST_RENDER (1<<2) +#define GDRST_MEDIA (3<<2) /* VGA stuff */ @@ -344,9 +348,37 @@ #define FBC_CTL_PLANEA (0<<0) #define FBC_CTL_PLANEB (1<<0) #define FBC_FENCE_OFF 0x0321b +#define FBC_TAG 0x03300 #define FBC_LL_SIZE (1536) +/* Framebuffer compression for GM45+ */ +#define DPFC_CB_BASE 0x3200 +#define DPFC_CONTROL 0x3208 +#define DPFC_CTL_EN (1<<31) +#define DPFC_CTL_PLANEA (0<<30) +#define DPFC_CTL_PLANEB (1<<30) +#define DPFC_CTL_FENCE_EN (1<<29) +#define DPFC_SR_EN (1<<10) +#define DPFC_CTL_LIMIT_1X (0<<6) +#define DPFC_CTL_LIMIT_2X (1<<6) +#define DPFC_CTL_LIMIT_4X (2<<6) +#define DPFC_RECOMP_CTL 0x320c +#define DPFC_RECOMP_STALL_EN (1<<27) +#define DPFC_RECOMP_STALL_WM_SHIFT (16) +#define DPFC_RECOMP_STALL_WM_MASK (0x07ff0000) +#define DPFC_RECOMP_TIMER_COUNT_SHIFT (0) +#define DPFC_RECOMP_TIMER_COUNT_MASK (0x0000003f) +#define DPFC_STATUS 0x3210 +#define DPFC_INVAL_SEG_SHIFT (16) +#define DPFC_INVAL_SEG_MASK (0x07ff0000) +#define DPFC_COMP_SEG_SHIFT (0) +#define DPFC_COMP_SEG_MASK (0x000003ff) +#define DPFC_STATUS2 0x3214 +#define DPFC_FENCE_YOFF 0x3218 +#define DPFC_CHICKEN 0x3224 +#define DPFC_HT_MODIFY (1<<31) + /* * GPIO regs */ @@ -936,6 +968,8 @@ #define LVDS_PORT_EN (1 << 31) /* Selects pipe B for LVDS data. Must be set on pre-965. */ #define LVDS_PIPEB_SELECT (1 << 30) +/* Enable border for unscaled (or aspect-scaled) display */ +#define LVDS_BORDER_ENABLE (1 << 15) /* * Enables the A0-A2 data pairs and CLKA, containing 18 bits of color data per * pixel. @@ -1046,6 +1080,8 @@ #define BACKLIGHT_DUTY_CYCLE_SHIFT (0) #define BACKLIGHT_DUTY_CYCLE_MASK (0xffff) +#define BLC_HIST_CTL 0x61260 + /* TV port control */ #define TV_CTL 0x68000 /** Enables the TV encoder */ @@ -1748,6 +1784,11 @@ #define PIPE_START_VBLANK_INTERRUPT_STATUS (1UL<<2) /* 965 or later */ #define PIPE_VBLANK_INTERRUPT_STATUS (1UL<<1) #define PIPE_OVERLAY_UPDATED_STATUS (1UL<<0) +#define PIPE_BPC_MASK (7 << 5) /* Ironlake */ +#define PIPE_8BPC (0 << 5) +#define PIPE_10BPC (1 << 5) +#define PIPE_6BPC (2 << 5) +#define PIPE_12BPC (3 << 5) #define DSPARB 0x70030 #define DSPARB_CSTART_MASK (0x7f << 7) @@ -1758,17 +1799,29 @@ #define DSPARB_AEND_SHIFT 0 #define DSPFW1 0x70034 +#define DSPFW_SR_SHIFT 23 +#define DSPFW_CURSORB_SHIFT 16 +#define DSPFW_PLANEB_SHIFT 8 #define DSPFW2 0x70038 +#define DSPFW_CURSORA_MASK 0x00003f00 +#define DSPFW_CURSORA_SHIFT 16 #define DSPFW3 0x7003c +#define DSPFW_HPLL_SR_EN (1<<31) +#define DSPFW_CURSOR_SR_SHIFT 24 #define IGD_SELF_REFRESH_EN (1<<30) /* FIFO watermark sizes etc */ +#define G4X_FIFO_LINE_SIZE 64 #define I915_FIFO_LINE_SIZE 64 #define I830_FIFO_LINE_SIZE 32 + +#define G4X_FIFO_SIZE 127 #define I945_FIFO_SIZE 127 /* 945 & 965 */ #define I915_FIFO_SIZE 95 #define I855GM_FIFO_SIZE 127 /* In cachelines */ #define I830_FIFO_SIZE 95 + +#define G4X_MAX_WM 0x3f #define I915_MAX_WM 0x3f #define IGD_DISPLAY_FIFO 512 /* in 64byte unit */ @@ -1998,8 +2051,15 @@ #define PFA_CTL_1 0x68080 #define PFB_CTL_1 0x68880 #define PF_ENABLE (1<<31) +#define PF_FILTER_MASK (3<<23) +#define PF_FILTER_PROGRAMMED (0<<23) +#define PF_FILTER_MED_3x3 (1<<23) +#define PF_FILTER_EDGE_ENHANCE (2<<23) +#define PF_FILTER_EDGE_SOFTEN (3<<23) #define PFA_WIN_SZ 0x68074 #define PFB_WIN_SZ 0x68874 +#define PFA_WIN_POS 0x68070 +#define PFB_WIN_POS 0x68870 /* legacy palette */ #define LGC_PALETTE_A 0x4a000 @@ -2115,11 +2175,11 @@ #define DREF_CPU_SOURCE_OUTPUT_MASK (3<<13) #define DREF_SSC_SOURCE_DISABLE (0<<11) #define DREF_SSC_SOURCE_ENABLE (2<<11) -#define DREF_SSC_SOURCE_MASK (2<<11) +#define DREF_SSC_SOURCE_MASK (3<<11) #define DREF_NONSPREAD_SOURCE_DISABLE (0<<9) #define DREF_NONSPREAD_CK505_ENABLE (1<<9) #define DREF_NONSPREAD_SOURCE_ENABLE (2<<9) -#define DREF_NONSPREAD_SOURCE_MASK (2<<9) +#define DREF_NONSPREAD_SOURCE_MASK (3<<9) #define DREF_SUPERSPREAD_SOURCE_DISABLE (0<<7) #define DREF_SUPERSPREAD_SOURCE_ENABLE (2<<7) #define DREF_SSC4_DOWNSPREAD (0<<6) diff --git a/drivers/gpu/drm/i915/i915_suspend.c b/drivers/gpu/drm/i915/i915_suspend.c index 20d4d19..6eec817 100644 --- a/drivers/gpu/drm/i915/i915_suspend.c +++ b/drivers/gpu/drm/i915/i915_suspend.c @@ -32,11 +32,15 @@ static bool i915_pipe_enabled(struct drm_device *dev, enum pipe pipe) { struct drm_i915_private *dev_priv = dev->dev_private; + u32 dpll_reg; - if (pipe == PIPE_A) - return (I915_READ(DPLL_A) & DPLL_VCO_ENABLE); - else - return (I915_READ(DPLL_B) & DPLL_VCO_ENABLE); + if (IS_IGDNG(dev)) { + dpll_reg = (pipe == PIPE_A) ? PCH_DPLL_A: PCH_DPLL_B; + } else { + dpll_reg = (pipe == PIPE_A) ? DPLL_A: DPLL_B; + } + + return (I915_READ(dpll_reg) & DPLL_VCO_ENABLE); } static void i915_save_palette(struct drm_device *dev, enum pipe pipe) @@ -49,6 +53,9 @@ static void i915_save_palette(struct drm_device *dev, enum pipe pipe) if (!i915_pipe_enabled(dev, pipe)) return; + if (IS_IGDNG(dev)) + reg = (pipe == PIPE_A) ? LGC_PALETTE_A : LGC_PALETTE_B; + if (pipe == PIPE_A) array = dev_priv->save_palette_a; else @@ -68,6 +75,9 @@ static void i915_restore_palette(struct drm_device *dev, enum pipe pipe) if (!i915_pipe_enabled(dev, pipe)) return; + if (IS_IGDNG(dev)) + reg = (pipe == PIPE_A) ? LGC_PALETTE_A : LGC_PALETTE_B; + if (pipe == PIPE_A) array = dev_priv->save_palette_a; else @@ -228,13 +238,25 @@ static void i915_save_modeset_reg(struct drm_device *dev) if (drm_core_check_feature(dev, DRIVER_MODESET)) return; + + if (IS_IGDNG(dev)) { + dev_priv->savePCH_DREF_CONTROL = I915_READ(PCH_DREF_CONTROL); + dev_priv->saveDISP_ARB_CTL = I915_READ(DISP_ARB_CTL); + } + /* Pipe & plane A info */ dev_priv->savePIPEACONF = I915_READ(PIPEACONF); dev_priv->savePIPEASRC = I915_READ(PIPEASRC); - dev_priv->saveFPA0 = I915_READ(FPA0); - dev_priv->saveFPA1 = I915_READ(FPA1); - dev_priv->saveDPLL_A = I915_READ(DPLL_A); - if (IS_I965G(dev)) + if (IS_IGDNG(dev)) { + dev_priv->saveFPA0 = I915_READ(PCH_FPA0); + dev_priv->saveFPA1 = I915_READ(PCH_FPA1); + dev_priv->saveDPLL_A = I915_READ(PCH_DPLL_A); + } else { + dev_priv->saveFPA0 = I915_READ(FPA0); + dev_priv->saveFPA1 = I915_READ(FPA1); + dev_priv->saveDPLL_A = I915_READ(DPLL_A); + } + if (IS_I965G(dev) && !IS_IGDNG(dev)) dev_priv->saveDPLL_A_MD = I915_READ(DPLL_A_MD); dev_priv->saveHTOTAL_A = I915_READ(HTOTAL_A); dev_priv->saveHBLANK_A = I915_READ(HBLANK_A); @@ -242,7 +264,30 @@ static void i915_save_modeset_reg(struct drm_device *dev) dev_priv->saveVTOTAL_A = I915_READ(VTOTAL_A); dev_priv->saveVBLANK_A = I915_READ(VBLANK_A); dev_priv->saveVSYNC_A = I915_READ(VSYNC_A); - dev_priv->saveBCLRPAT_A = I915_READ(BCLRPAT_A); + if (!IS_IGDNG(dev)) + dev_priv->saveBCLRPAT_A = I915_READ(BCLRPAT_A); + + if (IS_IGDNG(dev)) { + dev_priv->savePIPEA_DATA_M1 = I915_READ(PIPEA_DATA_M1); + dev_priv->savePIPEA_DATA_N1 = I915_READ(PIPEA_DATA_N1); + dev_priv->savePIPEA_LINK_M1 = I915_READ(PIPEA_LINK_M1); + dev_priv->savePIPEA_LINK_N1 = I915_READ(PIPEA_LINK_N1); + + dev_priv->saveFDI_TXA_CTL = I915_READ(FDI_TXA_CTL); + dev_priv->saveFDI_RXA_CTL = I915_READ(FDI_RXA_CTL); + + dev_priv->savePFA_CTL_1 = I915_READ(PFA_CTL_1); + dev_priv->savePFA_WIN_SZ = I915_READ(PFA_WIN_SZ); + dev_priv->savePFA_WIN_POS = I915_READ(PFA_WIN_POS); + + dev_priv->saveTRANSACONF = I915_READ(TRANSACONF); + dev_priv->saveTRANS_HTOTAL_A = I915_READ(TRANS_HTOTAL_A); + dev_priv->saveTRANS_HBLANK_A = I915_READ(TRANS_HBLANK_A); + dev_priv->saveTRANS_HSYNC_A = I915_READ(TRANS_HSYNC_A); + dev_priv->saveTRANS_VTOTAL_A = I915_READ(TRANS_VTOTAL_A); + dev_priv->saveTRANS_VBLANK_A = I915_READ(TRANS_VBLANK_A); + dev_priv->saveTRANS_VSYNC_A = I915_READ(TRANS_VSYNC_A); + } dev_priv->saveDSPACNTR = I915_READ(DSPACNTR); dev_priv->saveDSPASTRIDE = I915_READ(DSPASTRIDE); @@ -259,10 +304,16 @@ static void i915_save_modeset_reg(struct drm_device *dev) /* Pipe & plane B info */ dev_priv->savePIPEBCONF = I915_READ(PIPEBCONF); dev_priv->savePIPEBSRC = I915_READ(PIPEBSRC); - dev_priv->saveFPB0 = I915_READ(FPB0); - dev_priv->saveFPB1 = I915_READ(FPB1); - dev_priv->saveDPLL_B = I915_READ(DPLL_B); - if (IS_I965G(dev)) + if (IS_IGDNG(dev)) { + dev_priv->saveFPB0 = I915_READ(PCH_FPB0); + dev_priv->saveFPB1 = I915_READ(PCH_FPB1); + dev_priv->saveDPLL_B = I915_READ(PCH_DPLL_B); + } else { + dev_priv->saveFPB0 = I915_READ(FPB0); + dev_priv->saveFPB1 = I915_READ(FPB1); + dev_priv->saveDPLL_B = I915_READ(DPLL_B); + } + if (IS_I965G(dev) && !IS_IGDNG(dev)) dev_priv->saveDPLL_B_MD = I915_READ(DPLL_B_MD); dev_priv->saveHTOTAL_B = I915_READ(HTOTAL_B); dev_priv->saveHBLANK_B = I915_READ(HBLANK_B); @@ -270,7 +321,30 @@ static void i915_save_modeset_reg(struct drm_device *dev) dev_priv->saveVTOTAL_B = I915_READ(VTOTAL_B); dev_priv->saveVBLANK_B = I915_READ(VBLANK_B); dev_priv->saveVSYNC_B = I915_READ(VSYNC_B); - dev_priv->saveBCLRPAT_A = I915_READ(BCLRPAT_A); + if (!IS_IGDNG(dev)) + dev_priv->saveBCLRPAT_B = I915_READ(BCLRPAT_B); + + if (IS_IGDNG(dev)) { + dev_priv->savePIPEB_DATA_M1 = I915_READ(PIPEB_DATA_M1); + dev_priv->savePIPEB_DATA_N1 = I915_READ(PIPEB_DATA_N1); + dev_priv->savePIPEB_LINK_M1 = I915_READ(PIPEB_LINK_M1); + dev_priv->savePIPEB_LINK_N1 = I915_READ(PIPEB_LINK_N1); + + dev_priv->saveFDI_TXB_CTL = I915_READ(FDI_TXB_CTL); + dev_priv->saveFDI_RXB_CTL = I915_READ(FDI_RXB_CTL); + + dev_priv->savePFB_CTL_1 = I915_READ(PFB_CTL_1); + dev_priv->savePFB_WIN_SZ = I915_READ(PFB_WIN_SZ); + dev_priv->savePFB_WIN_POS = I915_READ(PFB_WIN_POS); + + dev_priv->saveTRANSBCONF = I915_READ(TRANSBCONF); + dev_priv->saveTRANS_HTOTAL_B = I915_READ(TRANS_HTOTAL_B); + dev_priv->saveTRANS_HBLANK_B = I915_READ(TRANS_HBLANK_B); + dev_priv->saveTRANS_HSYNC_B = I915_READ(TRANS_HSYNC_B); + dev_priv->saveTRANS_VTOTAL_B = I915_READ(TRANS_VTOTAL_B); + dev_priv->saveTRANS_VBLANK_B = I915_READ(TRANS_VBLANK_B); + dev_priv->saveTRANS_VSYNC_B = I915_READ(TRANS_VSYNC_B); + } dev_priv->saveDSPBCNTR = I915_READ(DSPBCNTR); dev_priv->saveDSPBSTRIDE = I915_READ(DSPBSTRIDE); @@ -285,26 +359,50 @@ static void i915_save_modeset_reg(struct drm_device *dev) dev_priv->savePIPEBSTAT = I915_READ(PIPEBSTAT); return; } + static void i915_restore_modeset_reg(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; + int dpll_a_reg, fpa0_reg, fpa1_reg; + int dpll_b_reg, fpb0_reg, fpb1_reg; if (drm_core_check_feature(dev, DRIVER_MODESET)) return; + if (IS_IGDNG(dev)) { + dpll_a_reg = PCH_DPLL_A; + dpll_b_reg = PCH_DPLL_B; + fpa0_reg = PCH_FPA0; + fpb0_reg = PCH_FPB0; + fpa1_reg = PCH_FPA1; + fpb1_reg = PCH_FPB1; + } else { + dpll_a_reg = DPLL_A; + dpll_b_reg = DPLL_B; + fpa0_reg = FPA0; + fpb0_reg = FPB0; + fpa1_reg = FPA1; + fpb1_reg = FPB1; + } + + if (IS_IGDNG(dev)) { + I915_WRITE(PCH_DREF_CONTROL, dev_priv->savePCH_DREF_CONTROL); + I915_WRITE(DISP_ARB_CTL, dev_priv->saveDISP_ARB_CTL); + } + /* Pipe & plane A info */ /* Prime the clock */ if (dev_priv->saveDPLL_A & DPLL_VCO_ENABLE) { - I915_WRITE(DPLL_A, dev_priv->saveDPLL_A & + I915_WRITE(dpll_a_reg, dev_priv->saveDPLL_A & ~DPLL_VCO_ENABLE); DRM_UDELAY(150); } - I915_WRITE(FPA0, dev_priv->saveFPA0); - I915_WRITE(FPA1, dev_priv->saveFPA1); + I915_WRITE(fpa0_reg, dev_priv->saveFPA0); + I915_WRITE(fpa1_reg, dev_priv->saveFPA1); /* Actually enable it */ - I915_WRITE(DPLL_A, dev_priv->saveDPLL_A); + I915_WRITE(dpll_a_reg, dev_priv->saveDPLL_A); DRM_UDELAY(150); - if (IS_I965G(dev)) + if (IS_I965G(dev) && !IS_IGDNG(dev)) I915_WRITE(DPLL_A_MD, dev_priv->saveDPLL_A_MD); DRM_UDELAY(150); @@ -315,7 +413,30 @@ static void i915_restore_modeset_reg(struct drm_device *dev) I915_WRITE(VTOTAL_A, dev_priv->saveVTOTAL_A); I915_WRITE(VBLANK_A, dev_priv->saveVBLANK_A); I915_WRITE(VSYNC_A, dev_priv->saveVSYNC_A); - I915_WRITE(BCLRPAT_A, dev_priv->saveBCLRPAT_A); + if (!IS_IGDNG(dev)) + I915_WRITE(BCLRPAT_A, dev_priv->saveBCLRPAT_A); + + if (IS_IGDNG(dev)) { + I915_WRITE(PIPEA_DATA_M1, dev_priv->savePIPEA_DATA_M1); + I915_WRITE(PIPEA_DATA_N1, dev_priv->savePIPEA_DATA_N1); + I915_WRITE(PIPEA_LINK_M1, dev_priv->savePIPEA_LINK_M1); + I915_WRITE(PIPEA_LINK_N1, dev_priv->savePIPEA_LINK_N1); + + I915_WRITE(FDI_RXA_CTL, dev_priv->saveFDI_RXA_CTL); + I915_WRITE(FDI_TXA_CTL, dev_priv->saveFDI_TXA_CTL); + + I915_WRITE(PFA_CTL_1, dev_priv->savePFA_CTL_1); + I915_WRITE(PFA_WIN_SZ, dev_priv->savePFA_WIN_SZ); + I915_WRITE(PFA_WIN_POS, dev_priv->savePFA_WIN_POS); + + I915_WRITE(TRANSACONF, dev_priv->saveTRANSACONF); + I915_WRITE(TRANS_HTOTAL_A, dev_priv->saveTRANS_HTOTAL_A); + I915_WRITE(TRANS_HBLANK_A, dev_priv->saveTRANS_HBLANK_A); + I915_WRITE(TRANS_HSYNC_A, dev_priv->saveTRANS_HSYNC_A); + I915_WRITE(TRANS_VTOTAL_A, dev_priv->saveTRANS_VTOTAL_A); + I915_WRITE(TRANS_VBLANK_A, dev_priv->saveTRANS_VBLANK_A); + I915_WRITE(TRANS_VSYNC_A, dev_priv->saveTRANS_VSYNC_A); + } /* Restore plane info */ I915_WRITE(DSPASIZE, dev_priv->saveDSPASIZE); @@ -337,16 +458,16 @@ static void i915_restore_modeset_reg(struct drm_device *dev) /* Pipe & plane B info */ if (dev_priv->saveDPLL_B & DPLL_VCO_ENABLE) { - I915_WRITE(DPLL_B, dev_priv->saveDPLL_B & + I915_WRITE(dpll_b_reg, dev_priv->saveDPLL_B & ~DPLL_VCO_ENABLE); DRM_UDELAY(150); } - I915_WRITE(FPB0, dev_priv->saveFPB0); - I915_WRITE(FPB1, dev_priv->saveFPB1); + I915_WRITE(fpb0_reg, dev_priv->saveFPB0); + I915_WRITE(fpb1_reg, dev_priv->saveFPB1); /* Actually enable it */ - I915_WRITE(DPLL_B, dev_priv->saveDPLL_B); + I915_WRITE(dpll_b_reg, dev_priv->saveDPLL_B); DRM_UDELAY(150); - if (IS_I965G(dev)) + if (IS_I965G(dev) && !IS_IGDNG(dev)) I915_WRITE(DPLL_B_MD, dev_priv->saveDPLL_B_MD); DRM_UDELAY(150); @@ -357,7 +478,30 @@ static void i915_restore_modeset_reg(struct drm_device *dev) I915_WRITE(VTOTAL_B, dev_priv->saveVTOTAL_B); I915_WRITE(VBLANK_B, dev_priv->saveVBLANK_B); I915_WRITE(VSYNC_B, dev_priv->saveVSYNC_B); - I915_WRITE(BCLRPAT_B, dev_priv->saveBCLRPAT_B); + if (!IS_IGDNG(dev)) + I915_WRITE(BCLRPAT_B, dev_priv->saveBCLRPAT_B); + + if (IS_IGDNG(dev)) { + I915_WRITE(PIPEB_DATA_M1, dev_priv->savePIPEB_DATA_M1); + I915_WRITE(PIPEB_DATA_N1, dev_priv->savePIPEB_DATA_N1); + I915_WRITE(PIPEB_LINK_M1, dev_priv->savePIPEB_LINK_M1); + I915_WRITE(PIPEB_LINK_N1, dev_priv->savePIPEB_LINK_N1); + + I915_WRITE(FDI_RXB_CTL, dev_priv->saveFDI_RXB_CTL); + I915_WRITE(FDI_TXB_CTL, dev_priv->saveFDI_TXB_CTL); + + I915_WRITE(PFB_CTL_1, dev_priv->savePFB_CTL_1); + I915_WRITE(PFB_WIN_SZ, dev_priv->savePFB_WIN_SZ); + I915_WRITE(PFB_WIN_POS, dev_priv->savePFB_WIN_POS); + + I915_WRITE(TRANSBCONF, dev_priv->saveTRANSBCONF); + I915_WRITE(TRANS_HTOTAL_B, dev_priv->saveTRANS_HTOTAL_B); + I915_WRITE(TRANS_HBLANK_B, dev_priv->saveTRANS_HBLANK_B); + I915_WRITE(TRANS_HSYNC_B, dev_priv->saveTRANS_HSYNC_B); + I915_WRITE(TRANS_VTOTAL_B, dev_priv->saveTRANS_VTOTAL_B); + I915_WRITE(TRANS_VBLANK_B, dev_priv->saveTRANS_VBLANK_B); + I915_WRITE(TRANS_VSYNC_B, dev_priv->saveTRANS_VSYNC_B); + } /* Restore plane info */ I915_WRITE(DSPBSIZE, dev_priv->saveDSPBSIZE); @@ -379,19 +523,10 @@ static void i915_restore_modeset_reg(struct drm_device *dev) return; } -int i915_save_state(struct drm_device *dev) + +void i915_save_display(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; - int i; - - pci_read_config_byte(dev->pdev, LBB, &dev_priv->saveLBB); - - /* Render Standby */ - if (IS_I965G(dev) && IS_MOBILE(dev)) - dev_priv->saveRENDERSTANDBY = I915_READ(MCHBAR_RENDER_STANDBY); - - /* Hardware status page */ - dev_priv->saveHWS = I915_READ(HWS_PGA); /* Display arbitration control */ dev_priv->saveDSPARB = I915_READ(DSPARB); @@ -399,6 +534,7 @@ int i915_save_state(struct drm_device *dev) /* This is only meaningful in non-KMS mode */ /* Don't save them in KMS mode */ i915_save_modeset_reg(dev); + /* Cursor state */ dev_priv->saveCURACNTR = I915_READ(CURACNTR); dev_priv->saveCURAPOS = I915_READ(CURAPOS); @@ -410,21 +546,43 @@ int i915_save_state(struct drm_device *dev) dev_priv->saveCURSIZE = I915_READ(CURSIZE); /* CRT state */ - dev_priv->saveADPA = I915_READ(ADPA); + if (IS_IGDNG(dev)) { + dev_priv->saveADPA = I915_READ(PCH_ADPA); + } else { + dev_priv->saveADPA = I915_READ(ADPA); + } /* LVDS state */ - dev_priv->savePP_CONTROL = I915_READ(PP_CONTROL); - dev_priv->savePFIT_PGM_RATIOS = I915_READ(PFIT_PGM_RATIOS); - dev_priv->saveBLC_PWM_CTL = I915_READ(BLC_PWM_CTL); - if (IS_I965G(dev)) - dev_priv->saveBLC_PWM_CTL2 = I915_READ(BLC_PWM_CTL2); - if (IS_MOBILE(dev) && !IS_I830(dev)) - dev_priv->saveLVDS = I915_READ(LVDS); - if (!IS_I830(dev) && !IS_845G(dev)) + if (IS_IGDNG(dev)) { + dev_priv->savePP_CONTROL = I915_READ(PCH_PP_CONTROL); + dev_priv->saveBLC_PWM_CTL = I915_READ(BLC_PWM_PCH_CTL1); + dev_priv->saveBLC_PWM_CTL2 = I915_READ(BLC_PWM_PCH_CTL2); + dev_priv->saveBLC_CPU_PWM_CTL = I915_READ(BLC_PWM_CPU_CTL); + dev_priv->saveBLC_CPU_PWM_CTL2 = I915_READ(BLC_PWM_CPU_CTL2); + dev_priv->saveLVDS = I915_READ(PCH_LVDS); + } else { + dev_priv->savePP_CONTROL = I915_READ(PP_CONTROL); + dev_priv->savePFIT_PGM_RATIOS = I915_READ(PFIT_PGM_RATIOS); + dev_priv->saveBLC_PWM_CTL = I915_READ(BLC_PWM_CTL); + dev_priv->saveBLC_HIST_CTL = I915_READ(BLC_HIST_CTL); + if (IS_I965G(dev)) + dev_priv->saveBLC_PWM_CTL2 = I915_READ(BLC_PWM_CTL2); + if (IS_MOBILE(dev) && !IS_I830(dev)) + dev_priv->saveLVDS = I915_READ(LVDS); + } + + if (!IS_I830(dev) && !IS_845G(dev) && !IS_IGDNG(dev)) dev_priv->savePFIT_CONTROL = I915_READ(PFIT_CONTROL); - dev_priv->savePP_ON_DELAYS = I915_READ(PP_ON_DELAYS); - dev_priv->savePP_OFF_DELAYS = I915_READ(PP_OFF_DELAYS); - dev_priv->savePP_DIVISOR = I915_READ(PP_DIVISOR); + + if (IS_IGDNG(dev)) { + dev_priv->savePP_ON_DELAYS = I915_READ(PCH_PP_ON_DELAYS); + dev_priv->savePP_OFF_DELAYS = I915_READ(PCH_PP_OFF_DELAYS); + dev_priv->savePP_DIVISOR = I915_READ(PCH_PP_DIVISOR); + } else { + dev_priv->savePP_ON_DELAYS = I915_READ(PP_ON_DELAYS); + dev_priv->savePP_OFF_DELAYS = I915_READ(PP_OFF_DELAYS); + dev_priv->savePP_DIVISOR = I915_READ(PP_DIVISOR); + } /* Display Port state */ if (SUPPORTS_INTEGRATED_DP(dev)) { @@ -443,25 +601,162 @@ int i915_save_state(struct drm_device *dev) /* FIXME: save TV & SDVO state */ /* FBC state */ - dev_priv->saveFBC_CFB_BASE = I915_READ(FBC_CFB_BASE); - dev_priv->saveFBC_LL_BASE = I915_READ(FBC_LL_BASE); - dev_priv->saveFBC_CONTROL2 = I915_READ(FBC_CONTROL2); - dev_priv->saveFBC_CONTROL = I915_READ(FBC_CONTROL); - - /* Interrupt state */ - dev_priv->saveIIR = I915_READ(IIR); - dev_priv->saveIER = I915_READ(IER); - dev_priv->saveIMR = I915_READ(IMR); + if (IS_GM45(dev)) { + dev_priv->saveDPFC_CB_BASE = I915_READ(DPFC_CB_BASE); + } else { + dev_priv->saveFBC_CFB_BASE = I915_READ(FBC_CFB_BASE); + dev_priv->saveFBC_LL_BASE = I915_READ(FBC_LL_BASE); + dev_priv->saveFBC_CONTROL2 = I915_READ(FBC_CONTROL2); + dev_priv->saveFBC_CONTROL = I915_READ(FBC_CONTROL); + } /* VGA state */ dev_priv->saveVGA0 = I915_READ(VGA0); dev_priv->saveVGA1 = I915_READ(VGA1); dev_priv->saveVGA_PD = I915_READ(VGA_PD); - dev_priv->saveVGACNTRL = I915_READ(VGACNTRL); + if (IS_IGDNG(dev)) + dev_priv->saveVGACNTRL = I915_READ(CPU_VGACNTRL); + else + dev_priv->saveVGACNTRL = I915_READ(VGACNTRL); + + i915_save_vga(dev); +} + +void i915_restore_display(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + /* Display arbitration */ + I915_WRITE(DSPARB, dev_priv->saveDSPARB); + + /* Display port ratios (must be done before clock is set) */ + if (SUPPORTS_INTEGRATED_DP(dev)) { + I915_WRITE(PIPEA_GMCH_DATA_M, dev_priv->savePIPEA_GMCH_DATA_M); + I915_WRITE(PIPEB_GMCH_DATA_M, dev_priv->savePIPEB_GMCH_DATA_M); + I915_WRITE(PIPEA_GMCH_DATA_N, dev_priv->savePIPEA_GMCH_DATA_N); + I915_WRITE(PIPEB_GMCH_DATA_N, dev_priv->savePIPEB_GMCH_DATA_N); + I915_WRITE(PIPEA_DP_LINK_M, dev_priv->savePIPEA_DP_LINK_M); + I915_WRITE(PIPEB_DP_LINK_M, dev_priv->savePIPEB_DP_LINK_M); + I915_WRITE(PIPEA_DP_LINK_N, dev_priv->savePIPEA_DP_LINK_N); + I915_WRITE(PIPEB_DP_LINK_N, dev_priv->savePIPEB_DP_LINK_N); + } + + /* This is only meaningful in non-KMS mode */ + /* Don't restore them in KMS mode */ + i915_restore_modeset_reg(dev); + + /* Cursor state */ + I915_WRITE(CURAPOS, dev_priv->saveCURAPOS); + I915_WRITE(CURACNTR, dev_priv->saveCURACNTR); + I915_WRITE(CURABASE, dev_priv->saveCURABASE); + I915_WRITE(CURBPOS, dev_priv->saveCURBPOS); + I915_WRITE(CURBCNTR, dev_priv->saveCURBCNTR); + I915_WRITE(CURBBASE, dev_priv->saveCURBBASE); + if (!IS_I9XX(dev)) + I915_WRITE(CURSIZE, dev_priv->saveCURSIZE); + + /* CRT state */ + if (IS_IGDNG(dev)) + I915_WRITE(PCH_ADPA, dev_priv->saveADPA); + else + I915_WRITE(ADPA, dev_priv->saveADPA); + + /* LVDS state */ + if (IS_I965G(dev) && !IS_IGDNG(dev)) + I915_WRITE(BLC_PWM_CTL2, dev_priv->saveBLC_PWM_CTL2); + + if (IS_IGDNG(dev)) { + I915_WRITE(PCH_LVDS, dev_priv->saveLVDS); + } else if (IS_MOBILE(dev) && !IS_I830(dev)) + I915_WRITE(LVDS, dev_priv->saveLVDS); + + if (!IS_I830(dev) && !IS_845G(dev) && !IS_IGDNG(dev)) + I915_WRITE(PFIT_CONTROL, dev_priv->savePFIT_CONTROL); + + if (IS_IGDNG(dev)) { + I915_WRITE(BLC_PWM_PCH_CTL1, dev_priv->saveBLC_PWM_CTL); + I915_WRITE(BLC_PWM_PCH_CTL2, dev_priv->saveBLC_PWM_CTL2); + I915_WRITE(BLC_PWM_CPU_CTL, dev_priv->saveBLC_CPU_PWM_CTL); + I915_WRITE(BLC_PWM_CPU_CTL2, dev_priv->saveBLC_CPU_PWM_CTL2); + I915_WRITE(PCH_PP_ON_DELAYS, dev_priv->savePP_ON_DELAYS); + I915_WRITE(PCH_PP_OFF_DELAYS, dev_priv->savePP_OFF_DELAYS); + I915_WRITE(PCH_PP_DIVISOR, dev_priv->savePP_DIVISOR); + I915_WRITE(PCH_PP_CONTROL, dev_priv->savePP_CONTROL); + } else { + I915_WRITE(PFIT_PGM_RATIOS, dev_priv->savePFIT_PGM_RATIOS); + I915_WRITE(BLC_PWM_CTL, dev_priv->saveBLC_PWM_CTL); + I915_WRITE(BLC_HIST_CTL, dev_priv->saveBLC_HIST_CTL); + I915_WRITE(PP_ON_DELAYS, dev_priv->savePP_ON_DELAYS); + I915_WRITE(PP_OFF_DELAYS, dev_priv->savePP_OFF_DELAYS); + I915_WRITE(PP_DIVISOR, dev_priv->savePP_DIVISOR); + I915_WRITE(PP_CONTROL, dev_priv->savePP_CONTROL); + } + + /* Display Port state */ + if (SUPPORTS_INTEGRATED_DP(dev)) { + I915_WRITE(DP_B, dev_priv->saveDP_B); + I915_WRITE(DP_C, dev_priv->saveDP_C); + I915_WRITE(DP_D, dev_priv->saveDP_D); + } + /* FIXME: restore TV & SDVO state */ + + /* FBC info */ + if (IS_GM45(dev)) { + g4x_disable_fbc(dev); + I915_WRITE(DPFC_CB_BASE, dev_priv->saveDPFC_CB_BASE); + } else { + i8xx_disable_fbc(dev); + I915_WRITE(FBC_CFB_BASE, dev_priv->saveFBC_CFB_BASE); + I915_WRITE(FBC_LL_BASE, dev_priv->saveFBC_LL_BASE); + I915_WRITE(FBC_CONTROL2, dev_priv->saveFBC_CONTROL2); + I915_WRITE(FBC_CONTROL, dev_priv->saveFBC_CONTROL); + } + + /* VGA state */ + if (IS_IGDNG(dev)) + I915_WRITE(CPU_VGACNTRL, dev_priv->saveVGACNTRL); + else + I915_WRITE(VGACNTRL, dev_priv->saveVGACNTRL); + I915_WRITE(VGA0, dev_priv->saveVGA0); + I915_WRITE(VGA1, dev_priv->saveVGA1); + I915_WRITE(VGA_PD, dev_priv->saveVGA_PD); + DRM_UDELAY(150); + + i915_restore_vga(dev); +} + +int i915_save_state(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int i; + + pci_read_config_byte(dev->pdev, LBB, &dev_priv->saveLBB); + + /* Render Standby */ + if (IS_I965G(dev) && IS_MOBILE(dev)) + dev_priv->saveRENDERSTANDBY = I915_READ(MCHBAR_RENDER_STANDBY); + + /* Hardware status page */ + dev_priv->saveHWS = I915_READ(HWS_PGA); + + i915_save_display(dev); + + /* Interrupt state */ + if (IS_IGDNG(dev)) { + dev_priv->saveDEIER = I915_READ(DEIER); + dev_priv->saveDEIMR = I915_READ(DEIMR); + dev_priv->saveGTIER = I915_READ(GTIER); + dev_priv->saveGTIMR = I915_READ(GTIMR); + dev_priv->saveFDI_RXA_IMR = I915_READ(FDI_RXA_IMR); + dev_priv->saveFDI_RXB_IMR = I915_READ(FDI_RXB_IMR); + } else { + dev_priv->saveIER = I915_READ(IER); + dev_priv->saveIMR = I915_READ(IMR); + } /* Clock gating state */ dev_priv->saveD_STATE = I915_READ(D_STATE); - dev_priv->saveDSPCLK_GATE_D = I915_READ(DSPCLK_GATE_D); + dev_priv->saveDSPCLK_GATE_D = I915_READ(DSPCLK_GATE_D); /* Not sure about this */ /* Cache mode state */ dev_priv->saveCACHE_MODE_0 = I915_READ(CACHE_MODE_0); @@ -489,7 +784,6 @@ int i915_save_state(struct drm_device *dev) for (i = 0; i < 8; i++) dev_priv->saveFENCE[i+8] = I915_READ(FENCE_REG_945_8 + (i * 4)); } - i915_save_vga(dev); return 0; } @@ -508,9 +802,6 @@ int i915_restore_state(struct drm_device *dev) /* Hardware status page */ I915_WRITE(HWS_PGA, dev_priv->saveHWS); - /* Display arbitration */ - I915_WRITE(DSPARB, dev_priv->saveDSPARB); - /* Fences */ if (IS_I965G(dev)) { for (i = 0; i < 16; i++) @@ -522,69 +813,21 @@ int i915_restore_state(struct drm_device *dev) for (i = 0; i < 8; i++) I915_WRITE(FENCE_REG_945_8 + (i * 4), dev_priv->saveFENCE[i+8]); } - - /* Display port ratios (must be done before clock is set) */ - if (SUPPORTS_INTEGRATED_DP(dev)) { - I915_WRITE(PIPEA_GMCH_DATA_M, dev_priv->savePIPEA_GMCH_DATA_M); - I915_WRITE(PIPEB_GMCH_DATA_M, dev_priv->savePIPEB_GMCH_DATA_M); - I915_WRITE(PIPEA_GMCH_DATA_N, dev_priv->savePIPEA_GMCH_DATA_N); - I915_WRITE(PIPEB_GMCH_DATA_N, dev_priv->savePIPEB_GMCH_DATA_N); - I915_WRITE(PIPEA_DP_LINK_M, dev_priv->savePIPEA_DP_LINK_M); - I915_WRITE(PIPEB_DP_LINK_M, dev_priv->savePIPEB_DP_LINK_M); - I915_WRITE(PIPEA_DP_LINK_N, dev_priv->savePIPEA_DP_LINK_N); - I915_WRITE(PIPEB_DP_LINK_N, dev_priv->savePIPEB_DP_LINK_N); - } - /* This is only meaningful in non-KMS mode */ - /* Don't restore them in KMS mode */ - i915_restore_modeset_reg(dev); - /* Cursor state */ - I915_WRITE(CURAPOS, dev_priv->saveCURAPOS); - I915_WRITE(CURACNTR, dev_priv->saveCURACNTR); - I915_WRITE(CURABASE, dev_priv->saveCURABASE); - I915_WRITE(CURBPOS, dev_priv->saveCURBPOS); - I915_WRITE(CURBCNTR, dev_priv->saveCURBCNTR); - I915_WRITE(CURBBASE, dev_priv->saveCURBBASE); - if (!IS_I9XX(dev)) - I915_WRITE(CURSIZE, dev_priv->saveCURSIZE); - /* CRT state */ - I915_WRITE(ADPA, dev_priv->saveADPA); + i915_restore_display(dev); - /* LVDS state */ - if (IS_I965G(dev)) - I915_WRITE(BLC_PWM_CTL2, dev_priv->saveBLC_PWM_CTL2); - if (IS_MOBILE(dev) && !IS_I830(dev)) - I915_WRITE(LVDS, dev_priv->saveLVDS); - if (!IS_I830(dev) && !IS_845G(dev)) - I915_WRITE(PFIT_CONTROL, dev_priv->savePFIT_CONTROL); - - I915_WRITE(PFIT_PGM_RATIOS, dev_priv->savePFIT_PGM_RATIOS); - I915_WRITE(BLC_PWM_CTL, dev_priv->saveBLC_PWM_CTL); - I915_WRITE(PP_ON_DELAYS, dev_priv->savePP_ON_DELAYS); - I915_WRITE(PP_OFF_DELAYS, dev_priv->savePP_OFF_DELAYS); - I915_WRITE(PP_DIVISOR, dev_priv->savePP_DIVISOR); - I915_WRITE(PP_CONTROL, dev_priv->savePP_CONTROL); - - /* Display Port state */ - if (SUPPORTS_INTEGRATED_DP(dev)) { - I915_WRITE(DP_B, dev_priv->saveDP_B); - I915_WRITE(DP_C, dev_priv->saveDP_C); - I915_WRITE(DP_D, dev_priv->saveDP_D); + /* Interrupt state */ + if (IS_IGDNG(dev)) { + I915_WRITE(DEIER, dev_priv->saveDEIER); + I915_WRITE(DEIMR, dev_priv->saveDEIMR); + I915_WRITE(GTIER, dev_priv->saveGTIER); + I915_WRITE(GTIMR, dev_priv->saveGTIMR); + I915_WRITE(FDI_RXA_IMR, dev_priv->saveFDI_RXA_IMR); + I915_WRITE(FDI_RXB_IMR, dev_priv->saveFDI_RXB_IMR); + } else { + I915_WRITE (IER, dev_priv->saveIER); + I915_WRITE (IMR, dev_priv->saveIMR); } - /* FIXME: restore TV & SDVO state */ - - /* FBC info */ - I915_WRITE(FBC_CFB_BASE, dev_priv->saveFBC_CFB_BASE); - I915_WRITE(FBC_LL_BASE, dev_priv->saveFBC_LL_BASE); - I915_WRITE(FBC_CONTROL2, dev_priv->saveFBC_CONTROL2); - I915_WRITE(FBC_CONTROL, dev_priv->saveFBC_CONTROL); - - /* VGA state */ - I915_WRITE(VGACNTRL, dev_priv->saveVGACNTRL); - I915_WRITE(VGA0, dev_priv->saveVGA0); - I915_WRITE(VGA1, dev_priv->saveVGA1); - I915_WRITE(VGA_PD, dev_priv->saveVGA_PD); - DRM_UDELAY(150); /* Clock gating state */ I915_WRITE (D_STATE, dev_priv->saveD_STATE); @@ -603,8 +846,6 @@ int i915_restore_state(struct drm_device *dev) for (i = 0; i < 3; i++) I915_WRITE(SWF30 + (i << 2), dev_priv->saveSWF2[i]); - i915_restore_vga(dev); - return 0; } diff --git a/drivers/gpu/drm/i915/i915_trace.h b/drivers/gpu/drm/i915/i915_trace.h new file mode 100644 index 0000000..01840d9 --- /dev/null +++ b/drivers/gpu/drm/i915/i915_trace.h @@ -0,0 +1,316 @@ +#if !defined(_I915_TRACE_H_) || defined(TRACE_HEADER_MULTI_READ) +#define _I915_TRACE_H_ + +#include <linux/stringify.h> +#include <linux/types.h> +#include <linux/tracepoint.h> + +#include <drm/drmP.h> + +#undef TRACE_SYSTEM +#define TRACE_SYSTEM i915 +#define TRACE_SYSTEM_STRING __stringify(TRACE_SYSTEM) +#define TRACE_INCLUDE_FILE i915_trace + +/* object tracking */ + +TRACE_EVENT(i915_gem_object_create, + + TP_PROTO(struct drm_gem_object *obj), + + TP_ARGS(obj), + + TP_STRUCT__entry( + __field(struct drm_gem_object *, obj) + __field(u32, size) + ), + + TP_fast_assign( + __entry->obj = obj; + __entry->size = obj->size; + ), + + TP_printk("obj=%p, size=%u", __entry->obj, __entry->size) +); + +TRACE_EVENT(i915_gem_object_bind, + + TP_PROTO(struct drm_gem_object *obj, u32 gtt_offset), + + TP_ARGS(obj, gtt_offset), + + TP_STRUCT__entry( + __field(struct drm_gem_object *, obj) + __field(u32, gtt_offset) + ), + + TP_fast_assign( + __entry->obj = obj; + __entry->gtt_offset = gtt_offset; + ), + + TP_printk("obj=%p, gtt_offset=%08x", + __entry->obj, __entry->gtt_offset) +); + +TRACE_EVENT(i915_gem_object_clflush, + + TP_PROTO(struct drm_gem_object *obj), + + TP_ARGS(obj), + + TP_STRUCT__entry( + __field(struct drm_gem_object *, obj) + ), + + TP_fast_assign( + __entry->obj = obj; + ), + + TP_printk("obj=%p", __entry->obj) +); + +TRACE_EVENT(i915_gem_object_change_domain, + + TP_PROTO(struct drm_gem_object *obj, uint32_t old_read_domains, uint32_t old_write_domain), + + TP_ARGS(obj, old_read_domains, old_write_domain), + + TP_STRUCT__entry( + __field(struct drm_gem_object *, obj) + __field(u32, read_domains) + __field(u32, write_domain) + ), + + TP_fast_assign( + __entry->obj = obj; + __entry->read_domains = obj->read_domains | (old_read_domains << 16); + __entry->write_domain = obj->write_domain | (old_write_domain << 16); + ), + + TP_printk("obj=%p, read=%04x, write=%04x", + __entry->obj, + __entry->read_domains, __entry->write_domain) +); + +TRACE_EVENT(i915_gem_object_get_fence, + + TP_PROTO(struct drm_gem_object *obj, int fence, int tiling_mode), + + TP_ARGS(obj, fence, tiling_mode), + + TP_STRUCT__entry( + __field(struct drm_gem_object *, obj) + __field(int, fence) + __field(int, tiling_mode) + ), + + TP_fast_assign( + __entry->obj = obj; + __entry->fence = fence; + __entry->tiling_mode = tiling_mode; + ), + + TP_printk("obj=%p, fence=%d, tiling=%d", + __entry->obj, __entry->fence, __entry->tiling_mode) +); + +TRACE_EVENT(i915_gem_object_unbind, + + TP_PROTO(struct drm_gem_object *obj), + + TP_ARGS(obj), + + TP_STRUCT__entry( + __field(struct drm_gem_object *, obj) + ), + + TP_fast_assign( + __entry->obj = obj; + ), + + TP_printk("obj=%p", __entry->obj) +); + +TRACE_EVENT(i915_gem_object_destroy, + + TP_PROTO(struct drm_gem_object *obj), + + TP_ARGS(obj), + + TP_STRUCT__entry( + __field(struct drm_gem_object *, obj) + ), + + TP_fast_assign( + __entry->obj = obj; + ), + + TP_printk("obj=%p", __entry->obj) +); + +/* batch tracing */ + +TRACE_EVENT(i915_gem_request_submit, + + TP_PROTO(struct drm_device *dev, u32 seqno), + + TP_ARGS(dev, seqno), + + TP_STRUCT__entry( + __field(u32, dev) + __field(u32, seqno) + ), + + TP_fast_assign( + __entry->dev = dev->primary->index; + __entry->seqno = seqno; + i915_trace_irq_get(dev, seqno); + ), + + TP_printk("dev=%u, seqno=%u", __entry->dev, __entry->seqno) +); + +TRACE_EVENT(i915_gem_request_flush, + + TP_PROTO(struct drm_device *dev, u32 seqno, + u32 flush_domains, u32 invalidate_domains), + + TP_ARGS(dev, seqno, flush_domains, invalidate_domains), + + TP_STRUCT__entry( + __field(u32, dev) + __field(u32, seqno) + __field(u32, flush_domains) + __field(u32, invalidate_domains) + ), + + TP_fast_assign( + __entry->dev = dev->primary->index; + __entry->seqno = seqno; + __entry->flush_domains = flush_domains; + __entry->invalidate_domains = invalidate_domains; + ), + + TP_printk("dev=%u, seqno=%u, flush=%04x, invalidate=%04x", + __entry->dev, __entry->seqno, + __entry->flush_domains, __entry->invalidate_domains) +); + + +TRACE_EVENT(i915_gem_request_complete, + + TP_PROTO(struct drm_device *dev, u32 seqno), + + TP_ARGS(dev, seqno), + + TP_STRUCT__entry( + __field(u32, dev) + __field(u32, seqno) + ), + + TP_fast_assign( + __entry->dev = dev->primary->index; + __entry->seqno = seqno; + ), + + TP_printk("dev=%u, seqno=%u", __entry->dev, __entry->seqno) +); + +TRACE_EVENT(i915_gem_request_retire, + + TP_PROTO(struct drm_device *dev, u32 seqno), + + TP_ARGS(dev, seqno), + + TP_STRUCT__entry( + __field(u32, dev) + __field(u32, seqno) + ), + + TP_fast_assign( + __entry->dev = dev->primary->index; + __entry->seqno = seqno; + ), + + TP_printk("dev=%u, seqno=%u", __entry->dev, __entry->seqno) +); + +TRACE_EVENT(i915_gem_request_wait_begin, + + TP_PROTO(struct drm_device *dev, u32 seqno), + + TP_ARGS(dev, seqno), + + TP_STRUCT__entry( + __field(u32, dev) + __field(u32, seqno) + ), + + TP_fast_assign( + __entry->dev = dev->primary->index; + __entry->seqno = seqno; + ), + + TP_printk("dev=%u, seqno=%u", __entry->dev, __entry->seqno) +); + +TRACE_EVENT(i915_gem_request_wait_end, + + TP_PROTO(struct drm_device *dev, u32 seqno), + + TP_ARGS(dev, seqno), + + TP_STRUCT__entry( + __field(u32, dev) + __field(u32, seqno) + ), + + TP_fast_assign( + __entry->dev = dev->primary->index; + __entry->seqno = seqno; + ), + + TP_printk("dev=%u, seqno=%u", __entry->dev, __entry->seqno) +); + +TRACE_EVENT(i915_ring_wait_begin, + + TP_PROTO(struct drm_device *dev), + + TP_ARGS(dev), + + TP_STRUCT__entry( + __field(u32, dev) + ), + + TP_fast_assign( + __entry->dev = dev->primary->index; + ), + + TP_printk("dev=%u", __entry->dev) +); + +TRACE_EVENT(i915_ring_wait_end, + + TP_PROTO(struct drm_device *dev), + + TP_ARGS(dev), + + TP_STRUCT__entry( + __field(u32, dev) + ), + + TP_fast_assign( + __entry->dev = dev->primary->index; + ), + + TP_printk("dev=%u", __entry->dev) +); + +#endif /* _I915_TRACE_H_ */ + +/* This part must be outside protection */ +#undef TRACE_INCLUDE_PATH +#define TRACE_INCLUDE_PATH ../../drivers/gpu/drm/i915 +#include <trace/define_trace.h> diff --git a/drivers/gpu/drm/i915/i915_trace_points.c b/drivers/gpu/drm/i915/i915_trace_points.c new file mode 100644 index 0000000..ead876e --- /dev/null +++ b/drivers/gpu/drm/i915/i915_trace_points.c @@ -0,0 +1,11 @@ +/* + * Copyright © 2009 Intel Corporation + * + * Authors: + * Chris Wilson <chris@chris-wilson.co.uk> + */ + +#include "i915_drv.h" + +#define CREATE_TRACE_POINTS +#include "i915_trace.h" diff --git a/drivers/gpu/drm/i915/intel_bios.c b/drivers/gpu/drm/i915/intel_bios.c index 1e28c16..96cd256 100644 --- a/drivers/gpu/drm/i915/intel_bios.c +++ b/drivers/gpu/drm/i915/intel_bios.c @@ -217,6 +217,9 @@ parse_general_features(struct drm_i915_private *dev_priv, if (IS_I85X(dev_priv->dev)) dev_priv->lvds_ssc_freq = general->ssc_freq ? 66 : 48; + else if (IS_IGDNG(dev_priv->dev)) + dev_priv->lvds_ssc_freq = + general->ssc_freq ? 100 : 120; else dev_priv->lvds_ssc_freq = general->ssc_freq ? 100 : 96; @@ -348,20 +351,18 @@ parse_driver_features(struct drm_i915_private *dev_priv, struct drm_device *dev = dev_priv->dev; struct bdb_driver_features *driver; - /* set default for chips without eDP */ - if (!SUPPORTS_EDP(dev)) { - dev_priv->edp_support = 0; - return; - } - driver = find_section(bdb, BDB_DRIVER_FEATURES); if (!driver) return; - if (driver->lvds_config == BDB_DRIVER_FEATURE_EDP) + if (driver && SUPPORTS_EDP(dev) && + driver->lvds_config == BDB_DRIVER_FEATURE_EDP) { dev_priv->edp_support = 1; + } else { + dev_priv->edp_support = 0; + } - if (driver->dual_frequency) + if (driver && driver->dual_frequency) dev_priv->render_reclock_avail = true; } diff --git a/drivers/gpu/drm/i915/intel_crt.c b/drivers/gpu/drm/i915/intel_crt.c index 88814fa..e505144 100644 --- a/drivers/gpu/drm/i915/intel_crt.c +++ b/drivers/gpu/drm/i915/intel_crt.c @@ -179,13 +179,10 @@ static bool intel_igdng_crt_detect_hotplug(struct drm_connector *connector) { struct drm_device *dev = connector->dev; struct drm_i915_private *dev_priv = dev->dev_private; - u32 adpa, temp; + u32 adpa; bool ret; - temp = adpa = I915_READ(PCH_ADPA); - - adpa &= ~ADPA_DAC_ENABLE; - I915_WRITE(PCH_ADPA, adpa); + adpa = I915_READ(PCH_ADPA); adpa &= ~ADPA_CRT_HOTPLUG_MASK; @@ -212,8 +209,6 @@ static bool intel_igdng_crt_detect_hotplug(struct drm_connector *connector) else ret = false; - /* restore origin register */ - I915_WRITE(PCH_ADPA, temp); return ret; } @@ -267,8 +262,8 @@ static bool intel_crt_detect_hotplug(struct drm_connector *connector) } while (time_after(timeout, jiffies)); } - if ((I915_READ(PORT_HOTPLUG_STAT) & CRT_HOTPLUG_MONITOR_MASK) == - CRT_HOTPLUG_MONITOR_COLOR) + if ((I915_READ(PORT_HOTPLUG_STAT) & CRT_HOTPLUG_MONITOR_MASK) != + CRT_HOTPLUG_MONITOR_NONE) return true; return false; diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c index 0227b16..099f420 100644 --- a/drivers/gpu/drm/i915/intel_display.c +++ b/drivers/gpu/drm/i915/intel_display.c @@ -24,6 +24,8 @@ * Eric Anholt <eric@anholt.net> */ +#include <linux/module.h> +#include <linux/input.h> #include <linux/i2c.h> #include <linux/kernel.h> #include "drmP.h" @@ -861,10 +863,8 @@ intel_igdng_find_best_PLL(const intel_limit_t *limit, struct drm_crtc *crtc, struct drm_device *dev = crtc->dev; struct drm_i915_private *dev_priv = dev->dev_private; intel_clock_t clock; - int max_n; - bool found; int err_most = 47; - found = false; + int err_min = 10000; /* eDP has only 2 clock choice, no n/m/p setting */ if (HAS_eDP) @@ -875,7 +875,7 @@ intel_igdng_find_best_PLL(const intel_limit_t *limit, struct drm_crtc *crtc, refclk, best_clock); if (intel_pipe_has_type(crtc, INTEL_OUTPUT_LVDS)) { - if ((I915_READ(LVDS) & LVDS_CLKB_POWER_MASK) == + if ((I915_READ(PCH_LVDS) & LVDS_CLKB_POWER_MASK) == LVDS_CLKB_POWER_UP) clock.p2 = limit->p2.p2_fast; else @@ -888,10 +888,9 @@ intel_igdng_find_best_PLL(const intel_limit_t *limit, struct drm_crtc *crtc, } memset(best_clock, 0, sizeof(*best_clock)); - max_n = limit->n.max; for (clock.p1 = limit->p1.max; clock.p1 >= limit->p1.min; clock.p1--) { /* based on hardware requriment prefer smaller n to precision */ - for (clock.n = limit->n.min; clock.n <= max_n; clock.n++) { + for (clock.n = limit->n.min; clock.n <= limit->n.max; clock.n++) { /* based on hardware requirment prefere larger m1,m2 */ for (clock.m1 = limit->m1.max; clock.m1 >= limit->m1.min; clock.m1--) { @@ -905,18 +904,18 @@ intel_igdng_find_best_PLL(const intel_limit_t *limit, struct drm_crtc *crtc, this_err = abs((10000 - (target*10000/clock.dot))); if (this_err < err_most) { *best_clock = clock; - err_most = this_err; - max_n = clock.n; - found = true; /* found on first matching */ goto out; + } else if (this_err < err_min) { + *best_clock = clock; + err_min = this_err; } } } } } out: - return found; + return true; } /* DisplayPort has only two frequencies, 162MHz and 270MHz */ @@ -941,6 +940,7 @@ intel_find_pll_g4x_dp(const intel_limit_t *limit, struct drm_crtc *crtc, clock.m = 5 * (clock.m1 + 2) + (clock.m2 + 2); clock.p = (clock.p1 * clock.p2); clock.dot = 96000 * clock.m / (clock.n + 2) / clock.p; + clock.vco = 0; memcpy(best_clock, &clock, sizeof(intel_clock_t)); return true; } @@ -952,6 +952,241 @@ intel_wait_for_vblank(struct drm_device *dev) mdelay(20); } +/* Parameters have changed, update FBC info */ +static void i8xx_enable_fbc(struct drm_crtc *crtc, unsigned long interval) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_framebuffer *fb = crtc->fb; + struct intel_framebuffer *intel_fb = to_intel_framebuffer(fb); + struct drm_i915_gem_object *obj_priv = intel_fb->obj->driver_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int plane, i; + u32 fbc_ctl, fbc_ctl2; + + dev_priv->cfb_pitch = dev_priv->cfb_size / FBC_LL_SIZE; + + if (fb->pitch < dev_priv->cfb_pitch) + dev_priv->cfb_pitch = fb->pitch; + + /* FBC_CTL wants 64B units */ + dev_priv->cfb_pitch = (dev_priv->cfb_pitch / 64) - 1; + dev_priv->cfb_fence = obj_priv->fence_reg; + dev_priv->cfb_plane = intel_crtc->plane; + plane = dev_priv->cfb_plane == 0 ? FBC_CTL_PLANEA : FBC_CTL_PLANEB; + + /* Clear old tags */ + for (i = 0; i < (FBC_LL_SIZE / 32) + 1; i++) + I915_WRITE(FBC_TAG + (i * 4), 0); + + /* Set it up... */ + fbc_ctl2 = FBC_CTL_FENCE_DBL | FBC_CTL_IDLE_IMM | plane; + if (obj_priv->tiling_mode != I915_TILING_NONE) + fbc_ctl2 |= FBC_CTL_CPU_FENCE; + I915_WRITE(FBC_CONTROL2, fbc_ctl2); + I915_WRITE(FBC_FENCE_OFF, crtc->y); + + /* enable it... */ + fbc_ctl = FBC_CTL_EN | FBC_CTL_PERIODIC; + fbc_ctl |= (dev_priv->cfb_pitch & 0xff) << FBC_CTL_STRIDE_SHIFT; + fbc_ctl |= (interval & 0x2fff) << FBC_CTL_INTERVAL_SHIFT; + if (obj_priv->tiling_mode != I915_TILING_NONE) + fbc_ctl |= dev_priv->cfb_fence; + I915_WRITE(FBC_CONTROL, fbc_ctl); + + DRM_DEBUG("enabled FBC, pitch %ld, yoff %d, plane %d, ", + dev_priv->cfb_pitch, crtc->y, dev_priv->cfb_plane); +} + +void i8xx_disable_fbc(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 fbc_ctl; + + if (!I915_HAS_FBC(dev)) + return; + + /* Disable compression */ + fbc_ctl = I915_READ(FBC_CONTROL); + fbc_ctl &= ~FBC_CTL_EN; + I915_WRITE(FBC_CONTROL, fbc_ctl); + + /* Wait for compressing bit to clear */ + while (I915_READ(FBC_STATUS) & FBC_STAT_COMPRESSING) + ; /* nothing */ + + intel_wait_for_vblank(dev); + + DRM_DEBUG("disabled FBC\n"); +} + +static bool i8xx_fbc_enabled(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + + return I915_READ(FBC_CONTROL) & FBC_CTL_EN; +} + +static void g4x_enable_fbc(struct drm_crtc *crtc, unsigned long interval) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_framebuffer *fb = crtc->fb; + struct intel_framebuffer *intel_fb = to_intel_framebuffer(fb); + struct drm_i915_gem_object *obj_priv = intel_fb->obj->driver_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int plane = (intel_crtc->plane == 0 ? DPFC_CTL_PLANEA : + DPFC_CTL_PLANEB); + unsigned long stall_watermark = 200; + u32 dpfc_ctl; + + dev_priv->cfb_pitch = (dev_priv->cfb_pitch / 64) - 1; + dev_priv->cfb_fence = obj_priv->fence_reg; + dev_priv->cfb_plane = intel_crtc->plane; + + dpfc_ctl = plane | DPFC_SR_EN | DPFC_CTL_LIMIT_1X; + if (obj_priv->tiling_mode != I915_TILING_NONE) { + dpfc_ctl |= DPFC_CTL_FENCE_EN | dev_priv->cfb_fence; + I915_WRITE(DPFC_CHICKEN, DPFC_HT_MODIFY); + } else { + I915_WRITE(DPFC_CHICKEN, ~DPFC_HT_MODIFY); + } + + I915_WRITE(DPFC_CONTROL, dpfc_ctl); + I915_WRITE(DPFC_RECOMP_CTL, DPFC_RECOMP_STALL_EN | + (stall_watermark << DPFC_RECOMP_STALL_WM_SHIFT) | + (interval << DPFC_RECOMP_TIMER_COUNT_SHIFT)); + I915_WRITE(DPFC_FENCE_YOFF, crtc->y); + + /* enable it... */ + I915_WRITE(DPFC_CONTROL, I915_READ(DPFC_CONTROL) | DPFC_CTL_EN); + + DRM_DEBUG("enabled fbc on plane %d\n", intel_crtc->plane); +} + +void g4x_disable_fbc(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 dpfc_ctl; + + /* Disable compression */ + dpfc_ctl = I915_READ(DPFC_CONTROL); + dpfc_ctl &= ~DPFC_CTL_EN; + I915_WRITE(DPFC_CONTROL, dpfc_ctl); + intel_wait_for_vblank(dev); + + DRM_DEBUG("disabled FBC\n"); +} + +static bool g4x_fbc_enabled(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + + return I915_READ(DPFC_CONTROL) & DPFC_CTL_EN; +} + +/** + * intel_update_fbc - enable/disable FBC as needed + * @crtc: CRTC to point the compressor at + * @mode: mode in use + * + * Set up the framebuffer compression hardware at mode set time. We + * enable it if possible: + * - plane A only (on pre-965) + * - no pixel mulitply/line duplication + * - no alpha buffer discard + * - no dual wide + * - framebuffer <= 2048 in width, 1536 in height + * + * We can't assume that any compression will take place (worst case), + * so the compressed buffer has to be the same size as the uncompressed + * one. It also must reside (along with the line length buffer) in + * stolen memory. + * + * We need to enable/disable FBC on a global basis. + */ +static void intel_update_fbc(struct drm_crtc *crtc, + struct drm_display_mode *mode) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_framebuffer *fb = crtc->fb; + struct intel_framebuffer *intel_fb; + struct drm_i915_gem_object *obj_priv; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int plane = intel_crtc->plane; + + if (!i915_powersave) + return; + + if (!dev_priv->display.fbc_enabled || + !dev_priv->display.enable_fbc || + !dev_priv->display.disable_fbc) + return; + + if (!crtc->fb) + return; + + intel_fb = to_intel_framebuffer(fb); + obj_priv = intel_fb->obj->driver_private; + + /* + * If FBC is already on, we just have to verify that we can + * keep it that way... + * Need to disable if: + * - changing FBC params (stride, fence, mode) + * - new fb is too large to fit in compressed buffer + * - going to an unsupported config (interlace, pixel multiply, etc.) + */ + if (intel_fb->obj->size > dev_priv->cfb_size) { + DRM_DEBUG("framebuffer too large, disabling compression\n"); + goto out_disable; + } + if ((mode->flags & DRM_MODE_FLAG_INTERLACE) || + (mode->flags & DRM_MODE_FLAG_DBLSCAN)) { + DRM_DEBUG("mode incompatible with compression, disabling\n"); + goto out_disable; + } + if ((mode->hdisplay > 2048) || + (mode->vdisplay > 1536)) { + DRM_DEBUG("mode too large for compression, disabling\n"); + goto out_disable; + } + if ((IS_I915GM(dev) || IS_I945GM(dev)) && plane != 0) { + DRM_DEBUG("plane not 0, disabling compression\n"); + goto out_disable; + } + if (obj_priv->tiling_mode != I915_TILING_X) { + DRM_DEBUG("framebuffer not tiled, disabling compression\n"); + goto out_disable; + } + + if (dev_priv->display.fbc_enabled(crtc)) { + /* We can re-enable it in this case, but need to update pitch */ + if (fb->pitch > dev_priv->cfb_pitch) + dev_priv->display.disable_fbc(dev); + if (obj_priv->fence_reg != dev_priv->cfb_fence) + dev_priv->display.disable_fbc(dev); + if (plane != dev_priv->cfb_plane) + dev_priv->display.disable_fbc(dev); + } + + if (!dev_priv->display.fbc_enabled(crtc)) { + /* Now try to turn it back on if possible */ + dev_priv->display.enable_fbc(crtc, 500); + } + + return; + +out_disable: + DRM_DEBUG("unsupported config, disabling FBC\n"); + /* Multiple disables should be harmless */ + if (dev_priv->display.fbc_enabled(crtc)) + dev_priv->display.disable_fbc(dev); +} + static int intel_pipe_set_base(struct drm_crtc *crtc, int x, int y, struct drm_framebuffer *old_fb) @@ -964,12 +1199,13 @@ intel_pipe_set_base(struct drm_crtc *crtc, int x, int y, struct drm_i915_gem_object *obj_priv; struct drm_gem_object *obj; int pipe = intel_crtc->pipe; + int plane = intel_crtc->plane; unsigned long Start, Offset; - int dspbase = (pipe == 0 ? DSPAADDR : DSPBADDR); - int dspsurf = (pipe == 0 ? DSPASURF : DSPBSURF); - int dspstride = (pipe == 0) ? DSPASTRIDE : DSPBSTRIDE; - int dsptileoff = (pipe == 0 ? DSPATILEOFF : DSPBTILEOFF); - int dspcntr_reg = (pipe == 0) ? DSPACNTR : DSPBCNTR; + int dspbase = (plane == 0 ? DSPAADDR : DSPBADDR); + int dspsurf = (plane == 0 ? DSPASURF : DSPBSURF); + int dspstride = (plane == 0) ? DSPASTRIDE : DSPBSTRIDE; + int dsptileoff = (plane == 0 ? DSPATILEOFF : DSPBTILEOFF); + int dspcntr_reg = (plane == 0) ? DSPACNTR : DSPBCNTR; u32 dspcntr, alignment; int ret; @@ -979,12 +1215,12 @@ intel_pipe_set_base(struct drm_crtc *crtc, int x, int y, return 0; } - switch (pipe) { + switch (plane) { case 0: case 1: break; default: - DRM_ERROR("Can't update pipe %d in SAREA\n", pipe); + DRM_ERROR("Can't update plane %d in SAREA\n", plane); return -EINVAL; } @@ -1022,9 +1258,11 @@ intel_pipe_set_base(struct drm_crtc *crtc, int x, int y, return ret; } - /* Pre-i965 needs to install a fence for tiled scan-out */ - if (!IS_I965G(dev) && - obj_priv->fence_reg == I915_FENCE_REG_NONE && + /* Install a fence for tiled scan-out. Pre-i965 always needs a fence, + * whereas 965+ only requires a fence if using framebuffer compression. + * For simplicity, we always install a fence as the cost is not that onerous. + */ + if (obj_priv->fence_reg == I915_FENCE_REG_NONE && obj_priv->tiling_mode != I915_TILING_NONE) { ret = i915_gem_object_get_fence_reg(obj); if (ret != 0) { @@ -1086,6 +1324,9 @@ intel_pipe_set_base(struct drm_crtc *crtc, int x, int y, I915_READ(dspbase); } + if ((IS_I965G(dev) || plane == 0)) + intel_update_fbc(crtc, &crtc->mode); + intel_wait_for_vblank(dev); if (old_fb) { @@ -1217,6 +1458,7 @@ static void igdng_crtc_dpms(struct drm_crtc *crtc, int mode) int transconf_reg = (pipe == 0) ? TRANSACONF : TRANSBCONF; int pf_ctl_reg = (pipe == 0) ? PFA_CTL_1 : PFB_CTL_1; int pf_win_size = (pipe == 0) ? PFA_WIN_SZ : PFB_WIN_SZ; + int pf_win_pos = (pipe == 0) ? PFA_WIN_POS : PFB_WIN_POS; int cpu_htot_reg = (pipe == 0) ? HTOTAL_A : HTOTAL_B; int cpu_hblank_reg = (pipe == 0) ? HBLANK_A : HBLANK_B; int cpu_hsync_reg = (pipe == 0) ? HSYNC_A : HSYNC_B; @@ -1268,6 +1510,19 @@ static void igdng_crtc_dpms(struct drm_crtc *crtc, int mode) } } + /* Enable panel fitting for LVDS */ + if (intel_pipe_has_type(crtc, INTEL_OUTPUT_LVDS)) { + temp = I915_READ(pf_ctl_reg); + I915_WRITE(pf_ctl_reg, temp | PF_ENABLE | PF_FILTER_MED_3x3); + + /* currently full aspect */ + I915_WRITE(pf_win_pos, 0); + + I915_WRITE(pf_win_size, + (dev_priv->panel_fixed_mode->hdisplay << 16) | + (dev_priv->panel_fixed_mode->vdisplay)); + } + /* Enable CPU pipe */ temp = I915_READ(pipeconf_reg); if ((temp & PIPEACONF_ENABLE) == 0) { @@ -1532,9 +1787,10 @@ static void i9xx_crtc_dpms(struct drm_crtc *crtc, int mode) struct drm_i915_private *dev_priv = dev->dev_private; struct intel_crtc *intel_crtc = to_intel_crtc(crtc); int pipe = intel_crtc->pipe; + int plane = intel_crtc->plane; int dpll_reg = (pipe == 0) ? DPLL_A : DPLL_B; - int dspcntr_reg = (pipe == 0) ? DSPACNTR : DSPBCNTR; - int dspbase_reg = (pipe == 0) ? DSPAADDR : DSPBADDR; + int dspcntr_reg = (plane == 0) ? DSPACNTR : DSPBCNTR; + int dspbase_reg = (plane == 0) ? DSPAADDR : DSPBADDR; int pipeconf_reg = (pipe == 0) ? PIPEACONF : PIPEBCONF; u32 temp; @@ -1545,6 +1801,8 @@ static void i9xx_crtc_dpms(struct drm_crtc *crtc, int mode) case DRM_MODE_DPMS_ON: case DRM_MODE_DPMS_STANDBY: case DRM_MODE_DPMS_SUSPEND: + intel_update_watermarks(dev); + /* Enable the DPLL */ temp = I915_READ(dpll_reg); if ((temp & DPLL_VCO_ENABLE) == 0) { @@ -1577,15 +1835,21 @@ static void i9xx_crtc_dpms(struct drm_crtc *crtc, int mode) intel_crtc_load_lut(crtc); + if ((IS_I965G(dev) || plane == 0)) + intel_update_fbc(crtc, &crtc->mode); + /* Give the overlay scaler a chance to enable if it's on this pipe */ //intel_crtc_dpms_video(crtc, true); TODO - intel_update_watermarks(dev); break; case DRM_MODE_DPMS_OFF: intel_update_watermarks(dev); /* Give the overlay scaler a chance to disable if it's on this pipe */ //intel_crtc_dpms_video(crtc, FALSE); TODO + if (dev_priv->cfb_plane == plane && + dev_priv->display.disable_fbc) + dev_priv->display.disable_fbc(dev); + /* Disable the VGA plane that we never use */ i915_disable_vga(dev); @@ -1634,15 +1898,13 @@ static void i9xx_crtc_dpms(struct drm_crtc *crtc, int mode) static void intel_crtc_dpms(struct drm_crtc *crtc, int mode) { struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; struct drm_i915_master_private *master_priv; struct intel_crtc *intel_crtc = to_intel_crtc(crtc); int pipe = intel_crtc->pipe; bool enabled; - if (IS_IGDNG(dev)) - igdng_crtc_dpms(crtc, mode); - else - i9xx_crtc_dpms(crtc, mode); + dev_priv->display.dpms(crtc, mode); intel_crtc->dpms_mode = mode; @@ -1709,56 +1971,68 @@ static bool intel_crtc_mode_fixup(struct drm_crtc *crtc, return true; } +static int i945_get_display_clock_speed(struct drm_device *dev) +{ + return 400000; +} -/** Returns the core display clock speed for i830 - i945 */ -static int intel_get_core_clock_speed(struct drm_device *dev) +static int i915_get_display_clock_speed(struct drm_device *dev) { + return 333000; +} - /* Core clock values taken from the published datasheets. - * The 830 may go up to 166 Mhz, which we should check. - */ - if (IS_I945G(dev)) - return 400000; - else if (IS_I915G(dev)) - return 333000; - else if (IS_I945GM(dev) || IS_845G(dev) || IS_IGDGM(dev)) - return 200000; - else if (IS_I915GM(dev)) { - u16 gcfgc = 0; +static int i9xx_misc_get_display_clock_speed(struct drm_device *dev) +{ + return 200000; +} - pci_read_config_word(dev->pdev, GCFGC, &gcfgc); +static int i915gm_get_display_clock_speed(struct drm_device *dev) +{ + u16 gcfgc = 0; - if (gcfgc & GC_LOW_FREQUENCY_ENABLE) - return 133000; - else { - switch (gcfgc & GC_DISPLAY_CLOCK_MASK) { - case GC_DISPLAY_CLOCK_333_MHZ: - return 333000; - default: - case GC_DISPLAY_CLOCK_190_200_MHZ: - return 190000; - } - } - } else if (IS_I865G(dev)) - return 266000; - else if (IS_I855(dev)) { - u16 hpllcc = 0; - /* Assume that the hardware is in the high speed state. This - * should be the default. - */ - switch (hpllcc & GC_CLOCK_CONTROL_MASK) { - case GC_CLOCK_133_200: - case GC_CLOCK_100_200: - return 200000; - case GC_CLOCK_166_250: - return 250000; - case GC_CLOCK_100_133: - return 133000; + pci_read_config_word(dev->pdev, GCFGC, &gcfgc); + + if (gcfgc & GC_LOW_FREQUENCY_ENABLE) + return 133000; + else { + switch (gcfgc & GC_DISPLAY_CLOCK_MASK) { + case GC_DISPLAY_CLOCK_333_MHZ: + return 333000; + default: + case GC_DISPLAY_CLOCK_190_200_MHZ: + return 190000; } - } else /* 852, 830 */ + } +} + +static int i865_get_display_clock_speed(struct drm_device *dev) +{ + return 266000; +} + +static int i855_get_display_clock_speed(struct drm_device *dev) +{ + u16 hpllcc = 0; + /* Assume that the hardware is in the high speed state. This + * should be the default. + */ + switch (hpllcc & GC_CLOCK_CONTROL_MASK) { + case GC_CLOCK_133_200: + case GC_CLOCK_100_200: + return 200000; + case GC_CLOCK_166_250: + return 250000; + case GC_CLOCK_100_133: return 133000; + } - return 0; /* Silence gcc warning */ + /* Shouldn't happen */ + return 0; +} + +static int i830_get_display_clock_speed(struct drm_device *dev) +{ + return 133000; } /** @@ -1809,7 +2083,7 @@ fdi_reduce_ratio(u32 *num, u32 *den) #define LINK_N 0x80000 static void -igdng_compute_m_n(int bytes_per_pixel, int nlanes, +igdng_compute_m_n(int bits_per_pixel, int nlanes, int pixel_clock, int link_clock, struct fdi_m_n *m_n) { @@ -1819,7 +2093,8 @@ igdng_compute_m_n(int bytes_per_pixel, int nlanes, temp = (u64) DATA_N * pixel_clock; temp = div_u64(temp, link_clock); - m_n->gmch_m = div_u64(temp * bytes_per_pixel, nlanes); + m_n->gmch_m = div_u64(temp * bits_per_pixel, nlanes); + m_n->gmch_m >>= 3; /* convert to bytes_per_pixel */ m_n->gmch_n = DATA_N; fdi_reduce_ratio(&m_n->gmch_m, &m_n->gmch_n); @@ -1867,6 +2142,13 @@ static struct intel_watermark_params igd_cursor_hplloff_wm = { IGD_CURSOR_GUARD_WM, IGD_FIFO_LINE_SIZE }; +static struct intel_watermark_params g4x_wm_info = { + G4X_FIFO_SIZE, + G4X_MAX_WM, + G4X_MAX_WM, + 2, + G4X_FIFO_LINE_SIZE, +}; static struct intel_watermark_params i945_wm_info = { I945_FIFO_SIZE, I915_MAX_WM, @@ -1921,7 +2203,14 @@ static unsigned long intel_calculate_wm(unsigned long clock_in_khz, { long entries_required, wm_size; - entries_required = (clock_in_khz * pixel_size * latency_ns) / 1000000; + /* + * Note: we need to make sure we don't overflow for various clock & + * latency values. + * clocks go from a few thousand to several hundred thousand. + * latency is usually a few thousand + */ + entries_required = ((clock_in_khz / 1000) * pixel_size * latency_ns) / + 1000; entries_required /= wm->cacheline_size; DRM_DEBUG("FIFO entries required for mode: %d\n", entries_required); @@ -1986,14 +2275,13 @@ static struct cxsr_latency *intel_get_cxsr_latency(int is_desktop, int fsb, for (i = 0; i < ARRAY_SIZE(cxsr_latency_table); i++) { latency = &cxsr_latency_table[i]; if (is_desktop == latency->is_desktop && - fsb == latency->fsb_freq && mem == latency->mem_freq) - break; + fsb == latency->fsb_freq && mem == latency->mem_freq) + return latency; } - if (i >= ARRAY_SIZE(cxsr_latency_table)) { - DRM_DEBUG("Unknown FSB/MEM found, disable CxSR\n"); - return NULL; - } - return latency; + + DRM_DEBUG("Unknown FSB/MEM found, disable CxSR\n"); + + return NULL; } static void igd_disable_cxsr(struct drm_device *dev) @@ -2084,32 +2372,17 @@ static void igd_enable_cxsr(struct drm_device *dev, unsigned long clock, */ const static int latency_ns = 5000; -static int intel_get_fifo_size(struct drm_device *dev, int plane) +static int i9xx_get_fifo_size(struct drm_device *dev, int plane) { struct drm_i915_private *dev_priv = dev->dev_private; uint32_t dsparb = I915_READ(DSPARB); int size; - if (IS_I9XX(dev)) { - if (plane == 0) - size = dsparb & 0x7f; - else - size = ((dsparb >> DSPARB_CSTART_SHIFT) & 0x7f) - - (dsparb & 0x7f); - } else if (IS_I85X(dev)) { - if (plane == 0) - size = dsparb & 0x1ff; - else - size = ((dsparb >> DSPARB_BEND_SHIFT) & 0x1ff) - - (dsparb & 0x1ff); - size >>= 1; /* Convert to cachelines */ - } else if (IS_845G(dev)) { + if (plane == 0) size = dsparb & 0x7f; - size >>= 2; /* Convert to cachelines */ - } else { - size = dsparb & 0x7f; - size >>= 1; /* Convert to cachelines */ - } + else + size = ((dsparb >> DSPARB_CSTART_SHIFT) & 0x7f) - + (dsparb & 0x7f); DRM_DEBUG("FIFO size - (0x%08x) %s: %d\n", dsparb, plane ? "B" : "A", size); @@ -2117,19 +2390,127 @@ static int intel_get_fifo_size(struct drm_device *dev, int plane) return size; } -static void g4x_update_wm(struct drm_device *dev) +static int i85x_get_fifo_size(struct drm_device *dev, int plane) { struct drm_i915_private *dev_priv = dev->dev_private; - u32 fw_blc_self = I915_READ(FW_BLC_SELF); + uint32_t dsparb = I915_READ(DSPARB); + int size; - if (i915_powersave) - fw_blc_self |= FW_BLC_SELF_EN; + if (plane == 0) + size = dsparb & 0x1ff; else - fw_blc_self &= ~FW_BLC_SELF_EN; - I915_WRITE(FW_BLC_SELF, fw_blc_self); + size = ((dsparb >> DSPARB_BEND_SHIFT) & 0x1ff) - + (dsparb & 0x1ff); + size >>= 1; /* Convert to cachelines */ + + DRM_DEBUG("FIFO size - (0x%08x) %s: %d\n", dsparb, plane ? "B" : "A", + size); + + return size; +} + +static int i845_get_fifo_size(struct drm_device *dev, int plane) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + uint32_t dsparb = I915_READ(DSPARB); + int size; + + size = dsparb & 0x7f; + size >>= 2; /* Convert to cachelines */ + + DRM_DEBUG("FIFO size - (0x%08x) %s: %d\n", dsparb, plane ? "B" : "A", + size); + + return size; +} + +static int i830_get_fifo_size(struct drm_device *dev, int plane) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + uint32_t dsparb = I915_READ(DSPARB); + int size; + + size = dsparb & 0x7f; + size >>= 1; /* Convert to cachelines */ + + DRM_DEBUG("FIFO size - (0x%08x) %s: %d\n", dsparb, plane ? "B" : "A", + size); + + return size; +} + +static void g4x_update_wm(struct drm_device *dev, int planea_clock, + int planeb_clock, int sr_hdisplay, int pixel_size) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int total_size, cacheline_size; + int planea_wm, planeb_wm, cursora_wm, cursorb_wm, cursor_sr; + struct intel_watermark_params planea_params, planeb_params; + unsigned long line_time_us; + int sr_clock, sr_entries = 0, entries_required; + + /* Create copies of the base settings for each pipe */ + planea_params = planeb_params = g4x_wm_info; + + /* Grab a couple of global values before we overwrite them */ + total_size = planea_params.fifo_size; + cacheline_size = planea_params.cacheline_size; + + /* + * Note: we need to make sure we don't overflow for various clock & + * latency values. + * clocks go from a few thousand to several hundred thousand. + * latency is usually a few thousand + */ + entries_required = ((planea_clock / 1000) * pixel_size * latency_ns) / + 1000; + entries_required /= G4X_FIFO_LINE_SIZE; + planea_wm = entries_required + planea_params.guard_size; + + entries_required = ((planeb_clock / 1000) * pixel_size * latency_ns) / + 1000; + entries_required /= G4X_FIFO_LINE_SIZE; + planeb_wm = entries_required + planeb_params.guard_size; + + cursora_wm = cursorb_wm = 16; + cursor_sr = 32; + + DRM_DEBUG("FIFO watermarks - A: %d, B: %d\n", planea_wm, planeb_wm); + + /* Calc sr entries for one plane configs */ + if (sr_hdisplay && (!planea_clock || !planeb_clock)) { + /* self-refresh has much higher latency */ + const static int sr_latency_ns = 12000; + + sr_clock = planea_clock ? planea_clock : planeb_clock; + line_time_us = ((sr_hdisplay * 1000) / sr_clock); + + /* Use ns/us then divide to preserve precision */ + sr_entries = (((sr_latency_ns / line_time_us) + 1) * + pixel_size * sr_hdisplay) / 1000; + sr_entries = roundup(sr_entries / cacheline_size, 1); + DRM_DEBUG("self-refresh entries: %d\n", sr_entries); + I915_WRITE(FW_BLC_SELF, FW_BLC_SELF_EN); + } + + DRM_DEBUG("Setting FIFO watermarks - A: %d, B: %d, SR %d\n", + planea_wm, planeb_wm, sr_entries); + + planea_wm &= 0x3f; + planeb_wm &= 0x3f; + + I915_WRITE(DSPFW1, (sr_entries << DSPFW_SR_SHIFT) | + (cursorb_wm << DSPFW_CURSORB_SHIFT) | + (planeb_wm << DSPFW_PLANEB_SHIFT) | planea_wm); + I915_WRITE(DSPFW2, (I915_READ(DSPFW2) & DSPFW_CURSORA_MASK) | + (cursora_wm << DSPFW_CURSORA_SHIFT)); + /* HPLL off in SR has some issues on G4x... disable it */ + I915_WRITE(DSPFW3, (I915_READ(DSPFW3) & ~DSPFW_HPLL_SR_EN) | + (cursor_sr << DSPFW_CURSOR_SR_SHIFT)); } -static void i965_update_wm(struct drm_device *dev) +static void i965_update_wm(struct drm_device *dev, int unused, int unused2, + int unused3, int unused4) { struct drm_i915_private *dev_priv = dev->dev_private; @@ -2165,8 +2546,8 @@ static void i9xx_update_wm(struct drm_device *dev, int planea_clock, cacheline_size = planea_params.cacheline_size; /* Update per-plane FIFO sizes */ - planea_params.fifo_size = intel_get_fifo_size(dev, 0); - planeb_params.fifo_size = intel_get_fifo_size(dev, 1); + planea_params.fifo_size = dev_priv->display.get_fifo_size(dev, 0); + planeb_params.fifo_size = dev_priv->display.get_fifo_size(dev, 1); planea_wm = intel_calculate_wm(planea_clock, &planea_params, pixel_size, latency_ns); @@ -2213,14 +2594,14 @@ static void i9xx_update_wm(struct drm_device *dev, int planea_clock, I915_WRITE(FW_BLC2, fwater_hi); } -static void i830_update_wm(struct drm_device *dev, int planea_clock, - int pixel_size) +static void i830_update_wm(struct drm_device *dev, int planea_clock, int unused, + int unused2, int pixel_size) { struct drm_i915_private *dev_priv = dev->dev_private; uint32_t fwater_lo = I915_READ(FW_BLC) & ~0xfff; int planea_wm; - i830_wm_info.fifo_size = intel_get_fifo_size(dev, 0); + i830_wm_info.fifo_size = dev_priv->display.get_fifo_size(dev, 0); planea_wm = intel_calculate_wm(planea_clock, &i830_wm_info, pixel_size, latency_ns); @@ -2264,12 +2645,16 @@ static void i830_update_wm(struct drm_device *dev, int planea_clock, */ static void intel_update_watermarks(struct drm_device *dev) { + struct drm_i915_private *dev_priv = dev->dev_private; struct drm_crtc *crtc; struct intel_crtc *intel_crtc; int sr_hdisplay = 0; unsigned long planea_clock = 0, planeb_clock = 0, sr_clock = 0; int enabled = 0, pixel_size = 0; + if (!dev_priv->display.update_wm) + return; + /* Get the clock config from both planes */ list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { intel_crtc = to_intel_crtc(crtc); @@ -2302,15 +2687,8 @@ static void intel_update_watermarks(struct drm_device *dev) else if (IS_IGD(dev)) igd_disable_cxsr(dev); - if (IS_G4X(dev)) - g4x_update_wm(dev); - else if (IS_I965G(dev)) - i965_update_wm(dev); - else if (IS_I9XX(dev) || IS_MOBILE(dev)) - i9xx_update_wm(dev, planea_clock, planeb_clock, sr_hdisplay, - pixel_size); - else - i830_update_wm(dev, planea_clock, pixel_size); + dev_priv->display.update_wm(dev, planea_clock, planeb_clock, + sr_hdisplay, pixel_size); } static int intel_crtc_mode_set(struct drm_crtc *crtc, @@ -2323,10 +2701,11 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, struct drm_i915_private *dev_priv = dev->dev_private; struct intel_crtc *intel_crtc = to_intel_crtc(crtc); int pipe = intel_crtc->pipe; + int plane = intel_crtc->plane; int fp_reg = (pipe == 0) ? FPA0 : FPB0; int dpll_reg = (pipe == 0) ? DPLL_A : DPLL_B; int dpll_md_reg = (intel_crtc->pipe == 0) ? DPLL_A_MD : DPLL_B_MD; - int dspcntr_reg = (pipe == 0) ? DSPACNTR : DSPBCNTR; + int dspcntr_reg = (plane == 0) ? DSPACNTR : DSPBCNTR; int pipeconf_reg = (pipe == 0) ? PIPEACONF : PIPEBCONF; int htot_reg = (pipe == 0) ? HTOTAL_A : HTOTAL_B; int hblank_reg = (pipe == 0) ? HBLANK_A : HBLANK_B; @@ -2334,8 +2713,8 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, int vtot_reg = (pipe == 0) ? VTOTAL_A : VTOTAL_B; int vblank_reg = (pipe == 0) ? VBLANK_A : VBLANK_B; int vsync_reg = (pipe == 0) ? VSYNC_A : VSYNC_B; - int dspsize_reg = (pipe == 0) ? DSPASIZE : DSPBSIZE; - int dsppos_reg = (pipe == 0) ? DSPAPOS : DSPBPOS; + int dspsize_reg = (plane == 0) ? DSPASIZE : DSPBSIZE; + int dsppos_reg = (plane == 0) ? DSPAPOS : DSPBPOS; int pipesrc_reg = (pipe == 0) ? PIPEASRC : PIPEBSRC; int refclk, num_outputs = 0; intel_clock_t clock, reduced_clock; @@ -2453,7 +2832,7 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, /* FDI link */ if (IS_IGDNG(dev)) { - int lane, link_bw; + int lane, link_bw, bpp; /* eDP doesn't require FDI link, so just set DP M/N according to current link config */ if (is_edp) { @@ -2472,10 +2851,72 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, lane = 4; link_bw = 270000; } - igdng_compute_m_n(3, lane, target_clock, + + /* determine panel color depth */ + temp = I915_READ(pipeconf_reg); + + switch (temp & PIPE_BPC_MASK) { + case PIPE_8BPC: + bpp = 24; + break; + case PIPE_10BPC: + bpp = 30; + break; + case PIPE_6BPC: + bpp = 18; + break; + case PIPE_12BPC: + bpp = 36; + break; + default: + DRM_ERROR("unknown pipe bpc value\n"); + bpp = 24; + } + + igdng_compute_m_n(bpp, lane, target_clock, link_bw, &m_n); } + /* Ironlake: try to setup display ref clock before DPLL + * enabling. This is only under driver's control after + * PCH B stepping, previous chipset stepping should be + * ignoring this setting. + */ + if (IS_IGDNG(dev)) { + temp = I915_READ(PCH_DREF_CONTROL); + /* Always enable nonspread source */ + temp &= ~DREF_NONSPREAD_SOURCE_MASK; + temp |= DREF_NONSPREAD_SOURCE_ENABLE; + I915_WRITE(PCH_DREF_CONTROL, temp); + POSTING_READ(PCH_DREF_CONTROL); + + temp &= ~DREF_SSC_SOURCE_MASK; + temp |= DREF_SSC_SOURCE_ENABLE; + I915_WRITE(PCH_DREF_CONTROL, temp); + POSTING_READ(PCH_DREF_CONTROL); + + udelay(200); + + if (is_edp) { + if (dev_priv->lvds_use_ssc) { + temp |= DREF_SSC1_ENABLE; + I915_WRITE(PCH_DREF_CONTROL, temp); + POSTING_READ(PCH_DREF_CONTROL); + + udelay(200); + + temp &= ~DREF_CPU_SOURCE_OUTPUT_MASK; + temp |= DREF_CPU_SOURCE_OUTPUT_DOWNSPREAD; + I915_WRITE(PCH_DREF_CONTROL, temp); + POSTING_READ(PCH_DREF_CONTROL); + } else { + temp |= DREF_CPU_SOURCE_OUTPUT_NONSPREAD; + I915_WRITE(PCH_DREF_CONTROL, temp); + POSTING_READ(PCH_DREF_CONTROL); + } + } + } + if (IS_IGD(dev)) { fp = (1 << clock.n) << 16 | clock.m1 << 8 | clock.m2; if (has_reduced_clock) @@ -2568,7 +3009,7 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, enable color space conversion */ if (!IS_IGDNG(dev)) { if (pipe == 0) - dspcntr |= DISPPLANE_SEL_PIPE_A; + dspcntr &= ~DISPPLANE_SEL_PIPE_MASK; else dspcntr |= DISPPLANE_SEL_PIPE_B; } @@ -2580,7 +3021,8 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, * XXX: No double-wide on 915GM pipe B. Is that the only reason for the * pipe == 0 check? */ - if (mode->clock > intel_get_core_clock_speed(dev) * 9 / 10) + if (mode->clock > + dev_priv->display.get_display_clock_speed(dev) * 9 / 10) pipeconf |= PIPEACONF_DOUBLE_WIDE; else pipeconf &= ~PIPEACONF_DOUBLE_WIDE; @@ -2625,6 +3067,8 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, lvds = I915_READ(lvds_reg); lvds |= LVDS_PORT_EN | LVDS_A0A2_CLKA_POWER_UP | LVDS_PIPEB_SELECT; + /* set the corresponsding LVDS_BORDER bit */ + lvds |= dev_priv->lvds_border_bits; /* Set the B0-B3 data pairs corresponding to whether we're going to * set the DPLLs for dual-channel mode or not. */ @@ -2652,9 +3096,12 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, udelay(150); if (IS_I965G(dev) && !IS_IGDNG(dev)) { - sdvo_pixel_multiply = adjusted_mode->clock / mode->clock; - I915_WRITE(dpll_md_reg, (0 << DPLL_MD_UDI_DIVIDER_SHIFT) | + if (is_sdvo) { + sdvo_pixel_multiply = adjusted_mode->clock / mode->clock; + I915_WRITE(dpll_md_reg, (0 << DPLL_MD_UDI_DIVIDER_SHIFT) | ((sdvo_pixel_multiply - 1) << DPLL_MD_UDI_MULTIPLIER_SHIFT)); + } else + I915_WRITE(dpll_md_reg, 0); } else { /* write it again -- the BIOS does, after all */ I915_WRITE(dpll_reg, dpll); @@ -2734,6 +3181,9 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, /* Flush the plane changes */ ret = intel_pipe_set_base(crtc, x, y, old_fb); + if ((IS_I965G(dev) || plane == 0)) + intel_update_fbc(crtc, &crtc->mode); + intel_update_watermarks(dev); drm_vblank_post_modeset(dev, pipe); @@ -2863,6 +3313,7 @@ static int intel_crtc_cursor_set(struct drm_crtc *crtc, i915_gem_object_unpin(intel_crtc->cursor_bo); drm_gem_object_unreference(intel_crtc->cursor_bo); } + mutex_unlock(&dev->struct_mutex); intel_crtc->cursor_addr = addr; @@ -2922,6 +3373,16 @@ void intel_crtc_fb_gamma_set(struct drm_crtc *crtc, u16 red, u16 green, intel_crtc->lut_b[regno] = blue >> 8; } +void intel_crtc_fb_gamma_get(struct drm_crtc *crtc, u16 *red, u16 *green, + u16 *blue, int regno) +{ + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + + *red = intel_crtc->lut_r[regno] << 8; + *green = intel_crtc->lut_g[regno] << 8; + *blue = intel_crtc->lut_b[regno] << 8; +} + static void intel_crtc_gamma_set(struct drm_crtc *crtc, u16 *red, u16 *green, u16 *blue, uint32_t size) { @@ -3513,6 +3974,7 @@ static const struct drm_crtc_helper_funcs intel_helper_funcs = { .mode_set_base = intel_pipe_set_base, .prepare = intel_crtc_prepare, .commit = intel_crtc_commit, + .load_lut = intel_crtc_load_lut, }; static const struct drm_crtc_funcs intel_crtc_funcs = { @@ -3544,6 +4006,14 @@ static void intel_crtc_init(struct drm_device *dev, int pipe) intel_crtc->lut_b[i] = i; } + /* Swap pipes & planes for FBC on pre-965 */ + intel_crtc->pipe = pipe; + intel_crtc->plane = pipe; + if (IS_MOBILE(dev) && (IS_I9XX(dev) && !IS_I965G(dev))) { + DRM_DEBUG("swapping pipes & planes for FBC\n"); + intel_crtc->plane = ((pipe == 0) ? 1 : 0); + } + intel_crtc->cursor_addr = 0; intel_crtc->dpms_mode = DRM_MODE_DPMS_OFF; drm_crtc_helper_add(&intel_crtc->base, &intel_helper_funcs); @@ -3787,7 +4257,9 @@ void intel_init_clock_gating(struct drm_device *dev) * Disable clock gating reported to work incorrectly according to the * specs, but enable as much else as we can. */ - if (IS_G4X(dev)) { + if (IS_IGDNG(dev)) { + return; + } else if (IS_G4X(dev)) { uint32_t dspclk_gate; I915_WRITE(RENCLK_GATE_D1, 0); I915_WRITE(RENCLK_GATE_D2, VF_UNIT_CLOCK_GATE_DISABLE | @@ -3826,6 +4298,75 @@ void intel_init_clock_gating(struct drm_device *dev) } } +/* Set up chip specific display functions */ +static void intel_init_display(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + /* We always want a DPMS function */ + if (IS_IGDNG(dev)) + dev_priv->display.dpms = igdng_crtc_dpms; + else + dev_priv->display.dpms = i9xx_crtc_dpms; + + /* Only mobile has FBC, leave pointers NULL for other chips */ + if (IS_MOBILE(dev)) { + if (IS_GM45(dev)) { + dev_priv->display.fbc_enabled = g4x_fbc_enabled; + dev_priv->display.enable_fbc = g4x_enable_fbc; + dev_priv->display.disable_fbc = g4x_disable_fbc; + } else if (IS_I965GM(dev) || IS_I945GM(dev) || IS_I915GM(dev)) { + dev_priv->display.fbc_enabled = i8xx_fbc_enabled; + dev_priv->display.enable_fbc = i8xx_enable_fbc; + dev_priv->display.disable_fbc = i8xx_disable_fbc; + } + /* 855GM needs testing */ + } + + /* Returns the core display clock speed */ + if (IS_I945G(dev)) + dev_priv->display.get_display_clock_speed = + i945_get_display_clock_speed; + else if (IS_I915G(dev)) + dev_priv->display.get_display_clock_speed = + i915_get_display_clock_speed; + else if (IS_I945GM(dev) || IS_845G(dev) || IS_IGDGM(dev)) + dev_priv->display.get_display_clock_speed = + i9xx_misc_get_display_clock_speed; + else if (IS_I915GM(dev)) + dev_priv->display.get_display_clock_speed = + i915gm_get_display_clock_speed; + else if (IS_I865G(dev)) + dev_priv->display.get_display_clock_speed = + i865_get_display_clock_speed; + else if (IS_I855(dev)) + dev_priv->display.get_display_clock_speed = + i855_get_display_clock_speed; + else /* 852, 830 */ + dev_priv->display.get_display_clock_speed = + i830_get_display_clock_speed; + + /* For FIFO watermark updates */ + if (IS_IGDNG(dev)) + dev_priv->display.update_wm = NULL; + else if (IS_G4X(dev)) + dev_priv->display.update_wm = g4x_update_wm; + else if (IS_I965G(dev)) + dev_priv->display.update_wm = i965_update_wm; + else if (IS_I9XX(dev) || IS_MOBILE(dev)) { + dev_priv->display.update_wm = i9xx_update_wm; + dev_priv->display.get_fifo_size = i9xx_get_fifo_size; + } else { + if (IS_I85X(dev)) + dev_priv->display.get_fifo_size = i85x_get_fifo_size; + else if (IS_845G(dev)) + dev_priv->display.get_fifo_size = i845_get_fifo_size; + else + dev_priv->display.get_fifo_size = i830_get_fifo_size; + dev_priv->display.update_wm = i830_update_wm; + } +} + void intel_modeset_init(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; @@ -3839,6 +4380,8 @@ void intel_modeset_init(struct drm_device *dev) dev->mode_config.funcs = (void *)&intel_mode_funcs; + intel_init_display(dev); + if (IS_I965G(dev)) { dev->mode_config.max_width = 8192; dev->mode_config.max_height = 8192; @@ -3904,6 +4447,9 @@ void intel_modeset_cleanup(struct drm_device *dev) mutex_unlock(&dev->struct_mutex); + if (dev_priv->display.disable_fbc) + dev_priv->display.disable_fbc(dev); + drm_mode_config_cleanup(dev); } diff --git a/drivers/gpu/drm/i915/intel_dp.c b/drivers/gpu/drm/i915/intel_dp.c index f4856a5..d834475 100644 --- a/drivers/gpu/drm/i915/intel_dp.c +++ b/drivers/gpu/drm/i915/intel_dp.c @@ -400,7 +400,7 @@ intel_dp_i2c_init(struct intel_output *intel_output, const char *name) { struct intel_dp_priv *dp_priv = intel_output->dev_priv; - DRM_ERROR("i2c_init %s\n", name); + DRM_DEBUG_KMS("i2c_init %s\n", name); dp_priv->algo.running = false; dp_priv->algo.address = 0; dp_priv->algo.aux_ch = intel_dp_i2c_aux_ch; diff --git a/drivers/gpu/drm/i915/intel_drv.h b/drivers/gpu/drm/i915/intel_drv.h index 3ebbbab..ef61fe9 100644 --- a/drivers/gpu/drm/i915/intel_drv.h +++ b/drivers/gpu/drm/i915/intel_drv.h @@ -28,6 +28,7 @@ #include <linux/i2c.h> #include <linux/i2c-id.h> #include <linux/i2c-algo-bit.h> +#include "i915_drv.h" #include "drm_crtc.h" #include "drm_crtc_helper.h" @@ -111,8 +112,8 @@ struct intel_output { struct intel_crtc { struct drm_crtc base; - int pipe; - int plane; + enum pipe pipe; + enum plane plane; struct drm_gem_object *cursor_bo; uint32_t cursor_addr; u8 lut_r[256], lut_g[256], lut_b[256]; @@ -174,6 +175,8 @@ extern int intelfb_resize(struct drm_device *dev, struct drm_crtc *crtc); extern void intelfb_restore(void); extern void intel_crtc_fb_gamma_set(struct drm_crtc *crtc, u16 red, u16 green, u16 blue, int regno); +extern void intel_crtc_fb_gamma_get(struct drm_crtc *crtc, u16 *red, u16 *green, + u16 *blue, int regno); extern int intel_framebuffer_create(struct drm_device *dev, struct drm_mode_fb_cmd *mode_cmd, diff --git a/drivers/gpu/drm/i915/intel_fb.c b/drivers/gpu/drm/i915/intel_fb.c index 7ba4a23..2b0fe54 100644 --- a/drivers/gpu/drm/i915/intel_fb.c +++ b/drivers/gpu/drm/i915/intel_fb.c @@ -60,10 +60,12 @@ static struct fb_ops intelfb_ops = { .fb_imageblit = cfb_imageblit, .fb_pan_display = drm_fb_helper_pan_display, .fb_blank = drm_fb_helper_blank, + .fb_setcmap = drm_fb_helper_setcmap, }; static struct drm_fb_helper_funcs intel_fb_helper_funcs = { .gamma_set = intel_crtc_fb_gamma_set, + .gamma_get = intel_crtc_fb_gamma_get, }; @@ -110,6 +112,7 @@ EXPORT_SYMBOL(intelfb_resize); static int intelfb_create(struct drm_device *dev, uint32_t fb_width, uint32_t fb_height, uint32_t surface_width, uint32_t surface_height, + uint32_t surface_depth, uint32_t surface_bpp, struct drm_framebuffer **fb_p) { struct fb_info *info; @@ -122,12 +125,16 @@ static int intelfb_create(struct drm_device *dev, uint32_t fb_width, struct device *device = &dev->pdev->dev; int size, ret, mmio_bar = IS_I9XX(dev) ? 0 : 1; + /* we don't do packed 24bpp */ + if (surface_bpp == 24) + surface_bpp = 32; + mode_cmd.width = surface_width; mode_cmd.height = surface_height; - mode_cmd.bpp = 32; + mode_cmd.bpp = surface_bpp; mode_cmd.pitch = ALIGN(mode_cmd.width * ((mode_cmd.bpp + 1) / 8), 64); - mode_cmd.depth = 24; + mode_cmd.depth = surface_depth; size = mode_cmd.pitch * mode_cmd.height; size = ALIGN(size, PAGE_SIZE); @@ -205,7 +212,7 @@ static int intelfb_create(struct drm_device *dev, uint32_t fb_width, // memset(info->screen_base, 0, size); - drm_fb_helper_fill_fix(info, fb->pitch); + drm_fb_helper_fill_fix(info, fb->pitch, fb->depth); drm_fb_helper_fill_var(info, fb, fb_width, fb_height); /* FIXME: we really shouldn't expose mmio space at all */ @@ -243,7 +250,7 @@ int intelfb_probe(struct drm_device *dev) int ret; DRM_DEBUG("\n"); - ret = drm_fb_helper_single_fb_probe(dev, intelfb_create); + ret = drm_fb_helper_single_fb_probe(dev, 32, intelfb_create); return ret; } EXPORT_SYMBOL(intelfb_probe); diff --git a/drivers/gpu/drm/i915/intel_hdmi.c b/drivers/gpu/drm/i915/intel_hdmi.c index fa304e1..c33451a 100644 --- a/drivers/gpu/drm/i915/intel_hdmi.c +++ b/drivers/gpu/drm/i915/intel_hdmi.c @@ -77,14 +77,32 @@ static void intel_hdmi_dpms(struct drm_encoder *encoder, int mode) struct intel_hdmi_priv *hdmi_priv = intel_output->dev_priv; u32 temp; - if (mode != DRM_MODE_DPMS_ON) { - temp = I915_READ(hdmi_priv->sdvox_reg); + temp = I915_READ(hdmi_priv->sdvox_reg); + + /* HW workaround, need to toggle enable bit off and on for 12bpc, but + * we do this anyway which shows more stable in testing. + */ + if (IS_IGDNG(dev)) { I915_WRITE(hdmi_priv->sdvox_reg, temp & ~SDVO_ENABLE); + POSTING_READ(hdmi_priv->sdvox_reg); + } + + if (mode != DRM_MODE_DPMS_ON) { + temp &= ~SDVO_ENABLE; } else { - temp = I915_READ(hdmi_priv->sdvox_reg); - I915_WRITE(hdmi_priv->sdvox_reg, temp | SDVO_ENABLE); + temp |= SDVO_ENABLE; } + + I915_WRITE(hdmi_priv->sdvox_reg, temp); POSTING_READ(hdmi_priv->sdvox_reg); + + /* HW workaround, need to write this twice for issue that may result + * in first write getting masked. + */ + if (IS_IGDNG(dev)) { + I915_WRITE(hdmi_priv->sdvox_reg, temp); + POSTING_READ(hdmi_priv->sdvox_reg); + } } static void intel_hdmi_save(struct drm_connector *connector) @@ -223,7 +241,7 @@ void intel_hdmi_init(struct drm_device *dev, int sdvox_reg) connector = &intel_output->base; drm_connector_init(dev, connector, &intel_hdmi_connector_funcs, - DRM_MODE_CONNECTOR_DVID); + DRM_MODE_CONNECTOR_HDMIA); drm_connector_helper_add(connector, &intel_hdmi_connector_helper_funcs); intel_output->type = INTEL_OUTPUT_HDMI; diff --git a/drivers/gpu/drm/i915/intel_lvds.c b/drivers/gpu/drm/i915/intel_lvds.c index dafc0da..05598ae 100644 --- a/drivers/gpu/drm/i915/intel_lvds.c +++ b/drivers/gpu/drm/i915/intel_lvds.c @@ -27,6 +27,7 @@ * Jesse Barnes <jesse.barnes@intel.com> */ +#include <acpi/button.h> #include <linux/dmi.h> #include <linux/i2c.h> #include "drmP.h" @@ -295,6 +296,10 @@ static bool intel_lvds_mode_fixup(struct drm_encoder *encoder, goto out; } + /* full screen scale for now */ + if (IS_IGDNG(dev)) + goto out; + /* 965+ wants fuzzy fitting */ if (IS_I965G(dev)) pfit_control |= (intel_crtc->pipe << PFIT_PIPE_SHIFT) | @@ -322,8 +327,10 @@ static bool intel_lvds_mode_fixup(struct drm_encoder *encoder, * to register description and PRM. * Change the value here to see the borders for debugging */ - I915_WRITE(BCLRPAT_A, 0); - I915_WRITE(BCLRPAT_B, 0); + if (!IS_IGDNG(dev)) { + I915_WRITE(BCLRPAT_A, 0); + I915_WRITE(BCLRPAT_B, 0); + } switch (lvds_priv->fitting_mode) { case DRM_MODE_SCALE_CENTER: @@ -373,7 +380,7 @@ static bool intel_lvds_mode_fixup(struct drm_encoder *encoder, adjusted_mode->crtc_vblank_start + vsync_pos; /* keep the vsync width constant */ adjusted_mode->crtc_vsync_end = - adjusted_mode->crtc_vblank_start + vsync_width; + adjusted_mode->crtc_vsync_start + vsync_width; border = 1; break; case DRM_MODE_SCALE_ASPECT: @@ -519,6 +526,14 @@ out: lvds_priv->pfit_control = pfit_control; lvds_priv->pfit_pgm_ratios = pfit_pgm_ratios; /* + * When there exists the border, it means that the LVDS_BORDR + * should be enabled. + */ + if (border) + dev_priv->lvds_border_bits |= LVDS_BORDER_ENABLE; + else + dev_priv->lvds_border_bits &= ~(LVDS_BORDER_ENABLE); + /* * XXX: It would be nice to support lower refresh rates on the * panels to reduce power consumption, and perhaps match the * user's requested refresh rate. @@ -572,7 +587,6 @@ static void intel_lvds_mode_set(struct drm_encoder *encoder, * settings. */ - /* No panel fitting yet, fixme */ if (IS_IGDNG(dev)) return; @@ -585,15 +599,33 @@ static void intel_lvds_mode_set(struct drm_encoder *encoder, I915_WRITE(PFIT_CONTROL, lvds_priv->pfit_control); } +/* Some lid devices report incorrect lid status, assume they're connected */ +static const struct dmi_system_id bad_lid_status[] = { + { + .ident = "Aspire One", + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "Acer"), + DMI_MATCH(DMI_PRODUCT_NAME, "Aspire one"), + }, + }, + { } +}; + /** * Detect the LVDS connection. * - * This always returns CONNECTOR_STATUS_CONNECTED. This connector should only have - * been set up if the LVDS was actually connected anyway. + * Since LVDS doesn't have hotlug, we use the lid as a proxy. Open means + * connected and closed means disconnected. We also send hotplug events as + * needed, using lid status notification from the input layer. */ static enum drm_connector_status intel_lvds_detect(struct drm_connector *connector) { - return connector_status_connected; + enum drm_connector_status status = connector_status_connected; + + if (!acpi_lid_open() && !dmi_check_system(bad_lid_status)) + status = connector_status_disconnected; + + return status; } /** @@ -632,6 +664,39 @@ static int intel_lvds_get_modes(struct drm_connector *connector) return 0; } +/* + * Lid events. Note the use of 'modeset_on_lid': + * - we set it on lid close, and reset it on open + * - we use it as a "only once" bit (ie we ignore + * duplicate events where it was already properly + * set/reset) + * - the suspend/resume paths will also set it to + * zero, since they restore the mode ("lid open"). + */ +static int intel_lid_notify(struct notifier_block *nb, unsigned long val, + void *unused) +{ + struct drm_i915_private *dev_priv = + container_of(nb, struct drm_i915_private, lid_notifier); + struct drm_device *dev = dev_priv->dev; + + if (!acpi_lid_open()) { + dev_priv->modeset_on_lid = 1; + return NOTIFY_OK; + } + + if (!dev_priv->modeset_on_lid) + return NOTIFY_OK; + + dev_priv->modeset_on_lid = 0; + + mutex_lock(&dev->mode_config.mutex); + drm_helper_resume_force_mode(dev); + mutex_unlock(&dev->mode_config.mutex); + + return NOTIFY_OK; +} + /** * intel_lvds_destroy - unregister and free LVDS structures * @connector: connector to free @@ -641,10 +706,14 @@ static int intel_lvds_get_modes(struct drm_connector *connector) */ static void intel_lvds_destroy(struct drm_connector *connector) { + struct drm_device *dev = connector->dev; struct intel_output *intel_output = to_intel_output(connector); + struct drm_i915_private *dev_priv = dev->dev_private; if (intel_output->ddc_bus) intel_i2c_destroy(intel_output->ddc_bus); + if (dev_priv->lid_notifier.notifier_call) + acpi_lid_notifier_unregister(&dev_priv->lid_notifier); drm_sysfs_connector_remove(connector); drm_connector_cleanup(connector); kfree(connector); @@ -1011,6 +1080,11 @@ out: pwm |= PWM_PCH_ENABLE; I915_WRITE(BLC_PWM_PCH_CTL1, pwm); } + dev_priv->lid_notifier.notifier_call = intel_lid_notify; + if (acpi_lid_notifier_register(&dev_priv->lid_notifier)) { + DRM_DEBUG("lid notifier registration failed\n"); + dev_priv->lid_notifier.notifier_call = NULL; + } drm_sysfs_connector_add(connector); return; diff --git a/drivers/gpu/drm/i915/intel_sdvo.c b/drivers/gpu/drm/i915/intel_sdvo.c index 0bf28ef..083bec2 100644 --- a/drivers/gpu/drm/i915/intel_sdvo.c +++ b/drivers/gpu/drm/i915/intel_sdvo.c @@ -135,6 +135,30 @@ struct intel_sdvo_priv { struct intel_sdvo_dtd save_input_dtd_1, save_input_dtd_2; struct intel_sdvo_dtd save_output_dtd[16]; u32 save_SDVOX; + /* add the property for the SDVO-TV */ + struct drm_property *left_property; + struct drm_property *right_property; + struct drm_property *top_property; + struct drm_property *bottom_property; + struct drm_property *hpos_property; + struct drm_property *vpos_property; + + /* add the property for the SDVO-TV/LVDS */ + struct drm_property *brightness_property; + struct drm_property *contrast_property; + struct drm_property *saturation_property; + struct drm_property *hue_property; + + /* Add variable to record current setting for the above property */ + u32 left_margin, right_margin, top_margin, bottom_margin; + /* this is to get the range of margin.*/ + u32 max_hscan, max_vscan; + u32 max_hpos, cur_hpos; + u32 max_vpos, cur_vpos; + u32 cur_brightness, max_brightness; + u32 cur_contrast, max_contrast; + u32 cur_saturation, max_saturation; + u32 cur_hue, max_hue; }; static bool @@ -281,6 +305,31 @@ static const struct _sdvo_cmd_name { SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_SDTV_RESOLUTION_SUPPORT), SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_SCALED_HDTV_RESOLUTION_SUPPORT), SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_SUPPORTED_ENHANCEMENTS), + /* Add the op code for SDVO enhancements */ + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_MAX_POSITION_H), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_POSITION_H), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_POSITION_H), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_MAX_POSITION_V), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_POSITION_V), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_POSITION_V), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_MAX_SATURATION), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_SATURATION), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_SATURATION), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_MAX_HUE), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_HUE), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_HUE), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_MAX_CONTRAST), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_CONTRAST), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_CONTRAST), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_MAX_BRIGHTNESS), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_BRIGHTNESS), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_BRIGHTNESS), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_MAX_OVERSCAN_H), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_OVERSCAN_H), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_OVERSCAN_H), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_MAX_OVERSCAN_V), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_OVERSCAN_V), + SDVO_CMD_NAME_ENTRY(SDVO_CMD_SET_OVERSCAN_V), /* HDMI op code */ SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_SUPP_ENCODE), SDVO_CMD_NAME_ENTRY(SDVO_CMD_GET_ENCODE), @@ -981,7 +1030,7 @@ static void intel_sdvo_set_tv_format(struct intel_output *output) status = intel_sdvo_read_response(output, NULL, 0); if (status != SDVO_CMD_STATUS_SUCCESS) - DRM_DEBUG("%s: Failed to set TV format\n", + DRM_DEBUG_KMS("%s: Failed to set TV format\n", SDVO_NAME(sdvo_priv)); } @@ -1792,6 +1841,45 @@ static int intel_sdvo_get_modes(struct drm_connector *connector) return 1; } +static +void intel_sdvo_destroy_enhance_property(struct drm_connector *connector) +{ + struct intel_output *intel_output = to_intel_output(connector); + struct intel_sdvo_priv *sdvo_priv = intel_output->dev_priv; + struct drm_device *dev = connector->dev; + + if (sdvo_priv->is_tv) { + if (sdvo_priv->left_property) + drm_property_destroy(dev, sdvo_priv->left_property); + if (sdvo_priv->right_property) + drm_property_destroy(dev, sdvo_priv->right_property); + if (sdvo_priv->top_property) + drm_property_destroy(dev, sdvo_priv->top_property); + if (sdvo_priv->bottom_property) + drm_property_destroy(dev, sdvo_priv->bottom_property); + if (sdvo_priv->hpos_property) + drm_property_destroy(dev, sdvo_priv->hpos_property); + if (sdvo_priv->vpos_property) + drm_property_destroy(dev, sdvo_priv->vpos_property); + } + if (sdvo_priv->is_tv) { + if (sdvo_priv->saturation_property) + drm_property_destroy(dev, + sdvo_priv->saturation_property); + if (sdvo_priv->contrast_property) + drm_property_destroy(dev, + sdvo_priv->contrast_property); + if (sdvo_priv->hue_property) + drm_property_destroy(dev, sdvo_priv->hue_property); + } + if (sdvo_priv->is_tv || sdvo_priv->is_lvds) { + if (sdvo_priv->brightness_property) + drm_property_destroy(dev, + sdvo_priv->brightness_property); + } + return; +} + static void intel_sdvo_destroy(struct drm_connector *connector) { struct intel_output *intel_output = to_intel_output(connector); @@ -1812,6 +1900,9 @@ static void intel_sdvo_destroy(struct drm_connector *connector) drm_property_destroy(connector->dev, sdvo_priv->tv_format_property); + if (sdvo_priv->is_tv || sdvo_priv->is_lvds) + intel_sdvo_destroy_enhance_property(connector); + drm_sysfs_connector_remove(connector); drm_connector_cleanup(connector); @@ -1829,6 +1920,8 @@ intel_sdvo_set_property(struct drm_connector *connector, struct drm_crtc *crtc = encoder->crtc; int ret = 0; bool changed = false; + uint8_t cmd, status; + uint16_t temp_value; ret = drm_connector_property_set_value(connector, property, val); if (ret < 0) @@ -1845,11 +1938,102 @@ intel_sdvo_set_property(struct drm_connector *connector, sdvo_priv->tv_format_name = sdvo_priv->tv_format_supported[val]; changed = true; - } else { - ret = -EINVAL; - goto out; } + if (sdvo_priv->is_tv || sdvo_priv->is_lvds) { + cmd = 0; + temp_value = val; + if (sdvo_priv->left_property == property) { + drm_connector_property_set_value(connector, + sdvo_priv->right_property, val); + if (sdvo_priv->left_margin == temp_value) + goto out; + + sdvo_priv->left_margin = temp_value; + sdvo_priv->right_margin = temp_value; + temp_value = sdvo_priv->max_hscan - + sdvo_priv->left_margin; + cmd = SDVO_CMD_SET_OVERSCAN_H; + } else if (sdvo_priv->right_property == property) { + drm_connector_property_set_value(connector, + sdvo_priv->left_property, val); + if (sdvo_priv->right_margin == temp_value) + goto out; + + sdvo_priv->left_margin = temp_value; + sdvo_priv->right_margin = temp_value; + temp_value = sdvo_priv->max_hscan - + sdvo_priv->left_margin; + cmd = SDVO_CMD_SET_OVERSCAN_H; + } else if (sdvo_priv->top_property == property) { + drm_connector_property_set_value(connector, + sdvo_priv->bottom_property, val); + if (sdvo_priv->top_margin == temp_value) + goto out; + + sdvo_priv->top_margin = temp_value; + sdvo_priv->bottom_margin = temp_value; + temp_value = sdvo_priv->max_vscan - + sdvo_priv->top_margin; + cmd = SDVO_CMD_SET_OVERSCAN_V; + } else if (sdvo_priv->bottom_property == property) { + drm_connector_property_set_value(connector, + sdvo_priv->top_property, val); + if (sdvo_priv->bottom_margin == temp_value) + goto out; + sdvo_priv->top_margin = temp_value; + sdvo_priv->bottom_margin = temp_value; + temp_value = sdvo_priv->max_vscan - + sdvo_priv->top_margin; + cmd = SDVO_CMD_SET_OVERSCAN_V; + } else if (sdvo_priv->hpos_property == property) { + if (sdvo_priv->cur_hpos == temp_value) + goto out; + + cmd = SDVO_CMD_SET_POSITION_H; + sdvo_priv->cur_hpos = temp_value; + } else if (sdvo_priv->vpos_property == property) { + if (sdvo_priv->cur_vpos == temp_value) + goto out; + + cmd = SDVO_CMD_SET_POSITION_V; + sdvo_priv->cur_vpos = temp_value; + } else if (sdvo_priv->saturation_property == property) { + if (sdvo_priv->cur_saturation == temp_value) + goto out; + + cmd = SDVO_CMD_SET_SATURATION; + sdvo_priv->cur_saturation = temp_value; + } else if (sdvo_priv->contrast_property == property) { + if (sdvo_priv->cur_contrast == temp_value) + goto out; + + cmd = SDVO_CMD_SET_CONTRAST; + sdvo_priv->cur_contrast = temp_value; + } else if (sdvo_priv->hue_property == property) { + if (sdvo_priv->cur_hue == temp_value) + goto out; + + cmd = SDVO_CMD_SET_HUE; + sdvo_priv->cur_hue = temp_value; + } else if (sdvo_priv->brightness_property == property) { + if (sdvo_priv->cur_brightness == temp_value) + goto out; + + cmd = SDVO_CMD_SET_BRIGHTNESS; + sdvo_priv->cur_brightness = temp_value; + } + if (cmd) { + intel_sdvo_write_cmd(intel_output, cmd, &temp_value, 2); + status = intel_sdvo_read_response(intel_output, + NULL, 0); + if (status != SDVO_CMD_STATUS_SUCCESS) { + DRM_DEBUG_KMS("Incorrect SDVO command \n"); + return -EINVAL; + } + changed = true; + } + } if (changed && crtc) drm_crtc_helper_set_mode(crtc, &crtc->mode, crtc->x, crtc->y, crtc->fb); @@ -2090,6 +2274,8 @@ intel_sdvo_output_setup(struct intel_output *intel_output, uint16_t flags) sdvo_priv->controlled_output = SDVO_OUTPUT_RGB1; encoder->encoder_type = DRM_MODE_ENCODER_DAC; connector->connector_type = DRM_MODE_CONNECTOR_VGA; + intel_output->clone_mask = (1 << INTEL_SDVO_NON_TV_CLONE_BIT) | + (1 << INTEL_ANALOG_CLONE_BIT); } else if (flags & SDVO_OUTPUT_LVDS0) { sdvo_priv->controlled_output = SDVO_OUTPUT_LVDS0; @@ -2176,6 +2362,310 @@ static void intel_sdvo_tv_create_property(struct drm_connector *connector) } +static void intel_sdvo_create_enhance_property(struct drm_connector *connector) +{ + struct intel_output *intel_output = to_intel_output(connector); + struct intel_sdvo_priv *sdvo_priv = intel_output->dev_priv; + struct intel_sdvo_enhancements_reply sdvo_data; + struct drm_device *dev = connector->dev; + uint8_t status; + uint16_t response, data_value[2]; + + intel_sdvo_write_cmd(intel_output, SDVO_CMD_GET_SUPPORTED_ENHANCEMENTS, + NULL, 0); + status = intel_sdvo_read_response(intel_output, &sdvo_data, + sizeof(sdvo_data)); + if (status != SDVO_CMD_STATUS_SUCCESS) { + DRM_DEBUG_KMS(" incorrect response is returned\n"); + return; + } + response = *((uint16_t *)&sdvo_data); + if (!response) { + DRM_DEBUG_KMS("No enhancement is supported\n"); + return; + } + if (sdvo_priv->is_tv) { + /* when horizontal overscan is supported, Add the left/right + * property + */ + if (sdvo_data.overscan_h) { + intel_sdvo_write_cmd(intel_output, + SDVO_CMD_GET_MAX_OVERSCAN_H, NULL, 0); + status = intel_sdvo_read_response(intel_output, + &data_value, 4); + if (status != SDVO_CMD_STATUS_SUCCESS) { + DRM_DEBUG_KMS("Incorrect SDVO max " + "h_overscan\n"); + return; + } + intel_sdvo_write_cmd(intel_output, + SDVO_CMD_GET_OVERSCAN_H, NULL, 0); + status = intel_sdvo_read_response(intel_output, + &response, 2); + if (status != SDVO_CMD_STATUS_SUCCESS) { + DRM_DEBUG_KMS("Incorrect SDVO h_overscan\n"); + return; + } + sdvo_priv->max_hscan = data_value[0]; + sdvo_priv->left_margin = data_value[0] - response; + sdvo_priv->right_margin = sdvo_priv->left_margin; + sdvo_priv->left_property = + drm_property_create(dev, DRM_MODE_PROP_RANGE, + "left_margin", 2); + sdvo_priv->left_property->values[0] = 0; + sdvo_priv->left_property->values[1] = data_value[0]; + drm_connector_attach_property(connector, + sdvo_priv->left_property, + sdvo_priv->left_margin); + sdvo_priv->right_property = + drm_property_create(dev, DRM_MODE_PROP_RANGE, + "right_margin", 2); + sdvo_priv->right_property->values[0] = 0; + sdvo_priv->right_property->values[1] = data_value[0]; + drm_connector_attach_property(connector, + sdvo_priv->right_property, + sdvo_priv->right_margin); + DRM_DEBUG_KMS("h_overscan: max %d, " + "default %d, current %d\n", + data_value[0], data_value[1], response); + } + if (sdvo_data.overscan_v) { + intel_sdvo_write_cmd(intel_output, + SDVO_CMD_GET_MAX_OVERSCAN_V, NULL, 0); + status = intel_sdvo_read_response(intel_output, + &data_value, 4); + if (status != SDVO_CMD_STATUS_SUCCESS) { + DRM_DEBUG_KMS("Incorrect SDVO max " + "v_overscan\n"); + return; + } + intel_sdvo_write_cmd(intel_output, + SDVO_CMD_GET_OVERSCAN_V, NULL, 0); + status = intel_sdvo_read_response(intel_output, + &response, 2); + if (status != SDVO_CMD_STATUS_SUCCESS) { + DRM_DEBUG_KMS("Incorrect SDVO v_overscan\n"); + return; + } + sdvo_priv->max_vscan = data_value[0]; + sdvo_priv->top_margin = data_value[0] - response; + sdvo_priv->bottom_margin = sdvo_priv->top_margin; + sdvo_priv->top_property = + drm_property_create(dev, DRM_MODE_PROP_RANGE, + "top_margin", 2); + sdvo_priv->top_property->values[0] = 0; + sdvo_priv->top_property->values[1] = data_value[0]; + drm_connector_attach_property(connector, + sdvo_priv->top_property, + sdvo_priv->top_margin); + sdvo_priv->bottom_property = + drm_property_create(dev, DRM_MODE_PROP_RANGE, + "bottom_margin", 2); + sdvo_priv->bottom_property->values[0] = 0; + sdvo_priv->bottom_property->values[1] = data_value[0]; + drm_connector_attach_property(connector, + sdvo_priv->bottom_property, + sdvo_priv->bottom_margin); + DRM_DEBUG_KMS("v_overscan: max %d, " + "default %d, current %d\n", + data_value[0], data_value[1], response); + } + if (sdvo_data.position_h) { + intel_sdvo_write_cmd(intel_output, + SDVO_CMD_GET_MAX_POSITION_H, NULL, 0); + status = intel_sdvo_read_response(intel_output, + &data_value, 4); + if (status != SDVO_CMD_STATUS_SUCCESS) { + DRM_DEBUG_KMS("Incorrect SDVO Max h_pos\n"); + return; + } + intel_sdvo_write_cmd(intel_output, + SDVO_CMD_GET_POSITION_H, NULL, 0); + status = intel_sdvo_read_response(intel_output, + &response, 2); + if (status != SDVO_CMD_STATUS_SUCCESS) { + DRM_DEBUG_KMS("Incorrect SDVO get h_postion\n"); + return; + } + sdvo_priv->max_hpos = data_value[0]; + sdvo_priv->cur_hpos = response; + sdvo_priv->hpos_property = + drm_property_create(dev, DRM_MODE_PROP_RANGE, + "hpos", 2); + sdvo_priv->hpos_property->values[0] = 0; + sdvo_priv->hpos_property->values[1] = data_value[0]; + drm_connector_attach_property(connector, + sdvo_priv->hpos_property, + sdvo_priv->cur_hpos); + DRM_DEBUG_KMS("h_position: max %d, " + "default %d, current %d\n", + data_value[0], data_value[1], response); + } + if (sdvo_data.position_v) { + intel_sdvo_write_cmd(intel_output, + SDVO_CMD_GET_MAX_POSITION_V, NULL, 0); + status = intel_sdvo_read_response(intel_output, + &data_value, 4); + if (status != SDVO_CMD_STATUS_SUCCESS) { + DRM_DEBUG_KMS("Incorrect SDVO Max v_pos\n"); + return; + } + intel_sdvo_write_cmd(intel_output, + SDVO_CMD_GET_POSITION_V, NULL, 0); + status = intel_sdvo_read_response(intel_output, + &response, 2); + if (status != SDVO_CMD_STATUS_SUCCESS) { + DRM_DEBUG_KMS("Incorrect SDVO get v_postion\n"); + return; + } + sdvo_priv->max_vpos = data_value[0]; + sdvo_priv->cur_vpos = response; + sdvo_priv->vpos_property = + drm_property_create(dev, DRM_MODE_PROP_RANGE, + "vpos", 2); + sdvo_priv->vpos_property->values[0] = 0; + sdvo_priv->vpos_property->values[1] = data_value[0]; + drm_connector_attach_property(connector, + sdvo_priv->vpos_property, + sdvo_priv->cur_vpos); + DRM_DEBUG_KMS("v_position: max %d, " + "default %d, current %d\n", + data_value[0], data_value[1], response); + } + } + if (sdvo_priv->is_tv) { + if (sdvo_data.saturation) { + intel_sdvo_write_cmd(intel_output, + SDVO_CMD_GET_MAX_SATURATION, NULL, 0); + status = intel_sdvo_read_response(intel_output, + &data_value, 4); + if (status != SDVO_CMD_STATUS_SUCCESS) { + DRM_DEBUG_KMS("Incorrect SDVO Max sat\n"); + return; + } + intel_sdvo_write_cmd(intel_output, + SDVO_CMD_GET_SATURATION, NULL, 0); + status = intel_sdvo_read_response(intel_output, + &response, 2); + if (status != SDVO_CMD_STATUS_SUCCESS) { + DRM_DEBUG_KMS("Incorrect SDVO get sat\n"); + return; + } + sdvo_priv->max_saturation = data_value[0]; + sdvo_priv->cur_saturation = response; + sdvo_priv->saturation_property = + drm_property_create(dev, DRM_MODE_PROP_RANGE, + "saturation", 2); + sdvo_priv->saturation_property->values[0] = 0; + sdvo_priv->saturation_property->values[1] = + data_value[0]; + drm_connector_attach_property(connector, + sdvo_priv->saturation_property, + sdvo_priv->cur_saturation); + DRM_DEBUG_KMS("saturation: max %d, " + "default %d, current %d\n", + data_value[0], data_value[1], response); + } + if (sdvo_data.contrast) { + intel_sdvo_write_cmd(intel_output, + SDVO_CMD_GET_MAX_CONTRAST, NULL, 0); + status = intel_sdvo_read_response(intel_output, + &data_value, 4); + if (status != SDVO_CMD_STATUS_SUCCESS) { + DRM_DEBUG_KMS("Incorrect SDVO Max contrast\n"); + return; + } + intel_sdvo_write_cmd(intel_output, + SDVO_CMD_GET_CONTRAST, NULL, 0); + status = intel_sdvo_read_response(intel_output, + &response, 2); + if (status != SDVO_CMD_STATUS_SUCCESS) { + DRM_DEBUG_KMS("Incorrect SDVO get contrast\n"); + return; + } + sdvo_priv->max_contrast = data_value[0]; + sdvo_priv->cur_contrast = response; + sdvo_priv->contrast_property = + drm_property_create(dev, DRM_MODE_PROP_RANGE, + "contrast", 2); + sdvo_priv->contrast_property->values[0] = 0; + sdvo_priv->contrast_property->values[1] = data_value[0]; + drm_connector_attach_property(connector, + sdvo_priv->contrast_property, + sdvo_priv->cur_contrast); + DRM_DEBUG_KMS("contrast: max %d, " + "default %d, current %d\n", + data_value[0], data_value[1], response); + } + if (sdvo_data.hue) { + intel_sdvo_write_cmd(intel_output, + SDVO_CMD_GET_MAX_HUE, NULL, 0); + status = intel_sdvo_read_response(intel_output, + &data_value, 4); + if (status != SDVO_CMD_STATUS_SUCCESS) { + DRM_DEBUG_KMS("Incorrect SDVO Max hue\n"); + return; + } + intel_sdvo_write_cmd(intel_output, + SDVO_CMD_GET_HUE, NULL, 0); + status = intel_sdvo_read_response(intel_output, + &response, 2); + if (status != SDVO_CMD_STATUS_SUCCESS) { + DRM_DEBUG_KMS("Incorrect SDVO get hue\n"); + return; + } + sdvo_priv->max_hue = data_value[0]; + sdvo_priv->cur_hue = response; + sdvo_priv->hue_property = + drm_property_create(dev, DRM_MODE_PROP_RANGE, + "hue", 2); + sdvo_priv->hue_property->values[0] = 0; + sdvo_priv->hue_property->values[1] = + data_value[0]; + drm_connector_attach_property(connector, + sdvo_priv->hue_property, + sdvo_priv->cur_hue); + DRM_DEBUG_KMS("hue: max %d, default %d, current %d\n", + data_value[0], data_value[1], response); + } + } + if (sdvo_priv->is_tv || sdvo_priv->is_lvds) { + if (sdvo_data.brightness) { + intel_sdvo_write_cmd(intel_output, + SDVO_CMD_GET_MAX_BRIGHTNESS, NULL, 0); + status = intel_sdvo_read_response(intel_output, + &data_value, 4); + if (status != SDVO_CMD_STATUS_SUCCESS) { + DRM_DEBUG_KMS("Incorrect SDVO Max bright\n"); + return; + } + intel_sdvo_write_cmd(intel_output, + SDVO_CMD_GET_BRIGHTNESS, NULL, 0); + status = intel_sdvo_read_response(intel_output, + &response, 2); + if (status != SDVO_CMD_STATUS_SUCCESS) { + DRM_DEBUG_KMS("Incorrect SDVO get brigh\n"); + return; + } + sdvo_priv->max_brightness = data_value[0]; + sdvo_priv->cur_brightness = response; + sdvo_priv->brightness_property = + drm_property_create(dev, DRM_MODE_PROP_RANGE, + "brightness", 2); + sdvo_priv->brightness_property->values[0] = 0; + sdvo_priv->brightness_property->values[1] = + data_value[0]; + drm_connector_attach_property(connector, + sdvo_priv->brightness_property, + sdvo_priv->cur_brightness); + DRM_DEBUG_KMS("brightness: max %d, " + "default %d, current %d\n", + data_value[0], data_value[1], response); + } + } + return; +} + bool intel_sdvo_init(struct drm_device *dev, int output_device) { struct drm_connector *connector; @@ -2264,6 +2754,10 @@ bool intel_sdvo_init(struct drm_device *dev, int output_device) drm_mode_connector_attach_encoder(&intel_output->base, &intel_output->enc); if (sdvo_priv->is_tv) intel_sdvo_tv_create_property(connector); + + if (sdvo_priv->is_tv || sdvo_priv->is_lvds) + intel_sdvo_create_enhance_property(connector); + drm_sysfs_connector_add(connector); intel_sdvo_select_ddc_bus(sdvo_priv); diff --git a/drivers/gpu/drm/i915/intel_tv.c b/drivers/gpu/drm/i915/intel_tv.c index c64eab4..9ca9179 100644 --- a/drivers/gpu/drm/i915/intel_tv.c +++ b/drivers/gpu/drm/i915/intel_tv.c @@ -1082,7 +1082,8 @@ intel_tv_mode_valid(struct drm_connector *connector, struct drm_display_mode *mo const struct tv_mode *tv_mode = intel_tv_mode_find(intel_output); /* Ensure TV refresh is close to desired refresh */ - if (tv_mode && abs(tv_mode->refresh - drm_mode_vrefresh(mode)) < 10) + if (tv_mode && abs(tv_mode->refresh - drm_mode_vrefresh(mode) * 1000) + < 1000) return MODE_OK; return MODE_CLOCK_RANGE; } diff --git a/drivers/gpu/drm/radeon/.gitignore b/drivers/gpu/drm/radeon/.gitignore new file mode 100644 index 0000000..403eb3a --- /dev/null +++ b/drivers/gpu/drm/radeon/.gitignore @@ -0,0 +1,3 @@ +mkregtable +*_reg_safe.h + diff --git a/drivers/gpu/drm/radeon/Makefile b/drivers/gpu/drm/radeon/Makefile index 09a28923..b5713ee 100644 --- a/drivers/gpu/drm/radeon/Makefile +++ b/drivers/gpu/drm/radeon/Makefile @@ -49,7 +49,7 @@ radeon-y += radeon_device.o radeon_kms.o \ radeon_cs.o radeon_bios.o radeon_benchmark.o r100.o r300.o r420.o \ rs400.o rs600.o rs690.o rv515.o r520.o r600.o rv770.o radeon_test.o \ r200.o radeon_legacy_tv.o r600_cs.o r600_blit.o r600_blit_shaders.o \ - r600_blit_kms.o + r600_blit_kms.o radeon_pm.o radeon-$(CONFIG_COMPAT) += radeon_ioc32.o diff --git a/drivers/gpu/drm/radeon/atom.c b/drivers/gpu/drm/radeon/atom.c index 901befe..d67c425 100644 --- a/drivers/gpu/drm/radeon/atom.c +++ b/drivers/gpu/drm/radeon/atom.c @@ -107,6 +107,7 @@ static uint32_t atom_iio_execute(struct atom_context *ctx, int base, base += 3; break; case ATOM_IIO_WRITE: + (void)ctx->card->reg_read(ctx->card, CU16(base + 1)); ctx->card->reg_write(ctx->card, CU16(base + 1), temp); base += 3; break; diff --git a/drivers/gpu/drm/radeon/atombios.h b/drivers/gpu/drm/radeon/atombios.h index 5d40208..c11dddd 100644 --- a/drivers/gpu/drm/radeon/atombios.h +++ b/drivers/gpu/drm/radeon/atombios.h @@ -2314,7 +2314,7 @@ typedef struct _ATOM_SPREAD_SPECTRUM_ASSIGNMENT { UCHAR ucSS_Step; UCHAR ucSS_Delay; UCHAR ucSS_Id; - UCHAR ucRecommandedRef_Div; + UCHAR ucRecommendedRef_Div; UCHAR ucSS_Range; /* it was reserved for V11 */ } ATOM_SPREAD_SPECTRUM_ASSIGNMENT; diff --git a/drivers/gpu/drm/radeon/atombios_crtc.c b/drivers/gpu/drm/radeon/atombios_crtc.c index 6a01592..c15287a 100644 --- a/drivers/gpu/drm/radeon/atombios_crtc.c +++ b/drivers/gpu/drm/radeon/atombios_crtc.c @@ -31,10 +31,6 @@ #include "atom.h" #include "atom-bits.h" -/* evil but including atombios.h is much worse */ -bool radeon_atom_get_tv_timings(struct radeon_device *rdev, int index, - SET_CRTC_TIMING_PARAMETERS_PS_ALLOCATION *crtc_timing, - int32_t *pixel_clock); static void atombios_overscan_setup(struct drm_crtc *crtc, struct drm_display_mode *mode, struct drm_display_mode *adjusted_mode) @@ -248,18 +244,18 @@ void atombios_crtc_dpms(struct drm_crtc *crtc, int mode) switch (mode) { case DRM_MODE_DPMS_ON: + atombios_enable_crtc(crtc, 1); if (ASIC_IS_DCE3(rdev)) atombios_enable_crtc_memreq(crtc, 1); - atombios_enable_crtc(crtc, 1); atombios_blank_crtc(crtc, 0); break; case DRM_MODE_DPMS_STANDBY: case DRM_MODE_DPMS_SUSPEND: case DRM_MODE_DPMS_OFF: atombios_blank_crtc(crtc, 1); - atombios_enable_crtc(crtc, 0); if (ASIC_IS_DCE3(rdev)) atombios_enable_crtc_memreq(crtc, 0); + atombios_enable_crtc(crtc, 0); break; } @@ -270,59 +266,147 @@ void atombios_crtc_dpms(struct drm_crtc *crtc, int mode) static void atombios_set_crtc_dtd_timing(struct drm_crtc *crtc, - SET_CRTC_USING_DTD_TIMING_PARAMETERS * crtc_param) + struct drm_display_mode *mode) { + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); struct drm_device *dev = crtc->dev; struct radeon_device *rdev = dev->dev_private; - SET_CRTC_USING_DTD_TIMING_PARAMETERS conv_param; + SET_CRTC_USING_DTD_TIMING_PARAMETERS args; int index = GetIndexIntoMasterTable(COMMAND, SetCRTC_UsingDTDTiming); + u16 misc = 0; - conv_param.usH_Size = cpu_to_le16(crtc_param->usH_Size); - conv_param.usH_Blanking_Time = - cpu_to_le16(crtc_param->usH_Blanking_Time); - conv_param.usV_Size = cpu_to_le16(crtc_param->usV_Size); - conv_param.usV_Blanking_Time = - cpu_to_le16(crtc_param->usV_Blanking_Time); - conv_param.usH_SyncOffset = cpu_to_le16(crtc_param->usH_SyncOffset); - conv_param.usH_SyncWidth = cpu_to_le16(crtc_param->usH_SyncWidth); - conv_param.usV_SyncOffset = cpu_to_le16(crtc_param->usV_SyncOffset); - conv_param.usV_SyncWidth = cpu_to_le16(crtc_param->usV_SyncWidth); - conv_param.susModeMiscInfo.usAccess = - cpu_to_le16(crtc_param->susModeMiscInfo.usAccess); - conv_param.ucCRTC = crtc_param->ucCRTC; + memset(&args, 0, sizeof(args)); + args.usH_Size = cpu_to_le16(mode->crtc_hdisplay); + args.usH_Blanking_Time = + cpu_to_le16(mode->crtc_hblank_end - mode->crtc_hdisplay); + args.usV_Size = cpu_to_le16(mode->crtc_vdisplay); + args.usV_Blanking_Time = + cpu_to_le16(mode->crtc_vblank_end - mode->crtc_vdisplay); + args.usH_SyncOffset = + cpu_to_le16(mode->crtc_hsync_start - mode->crtc_hdisplay); + args.usH_SyncWidth = + cpu_to_le16(mode->crtc_hsync_end - mode->crtc_hsync_start); + args.usV_SyncOffset = + cpu_to_le16(mode->crtc_vsync_start - mode->crtc_vdisplay); + args.usV_SyncWidth = + cpu_to_le16(mode->crtc_vsync_end - mode->crtc_vsync_start); + /*args.ucH_Border = mode->hborder;*/ + /*args.ucV_Border = mode->vborder;*/ + + if (mode->flags & DRM_MODE_FLAG_NVSYNC) + misc |= ATOM_VSYNC_POLARITY; + if (mode->flags & DRM_MODE_FLAG_NHSYNC) + misc |= ATOM_HSYNC_POLARITY; + if (mode->flags & DRM_MODE_FLAG_CSYNC) + misc |= ATOM_COMPOSITESYNC; + if (mode->flags & DRM_MODE_FLAG_INTERLACE) + misc |= ATOM_INTERLACE; + if (mode->flags & DRM_MODE_FLAG_DBLSCAN) + misc |= ATOM_DOUBLE_CLOCK_MODE; + + args.susModeMiscInfo.usAccess = cpu_to_le16(misc); + args.ucCRTC = radeon_crtc->crtc_id; printk("executing set crtc dtd timing\n"); - atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&conv_param); + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); } -void atombios_crtc_set_timing(struct drm_crtc *crtc, - SET_CRTC_TIMING_PARAMETERS_PS_ALLOCATION * - crtc_param) +static void atombios_crtc_set_timing(struct drm_crtc *crtc, + struct drm_display_mode *mode) { + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); struct drm_device *dev = crtc->dev; struct radeon_device *rdev = dev->dev_private; - SET_CRTC_TIMING_PARAMETERS_PS_ALLOCATION conv_param; + SET_CRTC_TIMING_PARAMETERS_PS_ALLOCATION args; int index = GetIndexIntoMasterTable(COMMAND, SetCRTC_Timing); + u16 misc = 0; - conv_param.usH_Total = cpu_to_le16(crtc_param->usH_Total); - conv_param.usH_Disp = cpu_to_le16(crtc_param->usH_Disp); - conv_param.usH_SyncStart = cpu_to_le16(crtc_param->usH_SyncStart); - conv_param.usH_SyncWidth = cpu_to_le16(crtc_param->usH_SyncWidth); - conv_param.usV_Total = cpu_to_le16(crtc_param->usV_Total); - conv_param.usV_Disp = cpu_to_le16(crtc_param->usV_Disp); - conv_param.usV_SyncStart = cpu_to_le16(crtc_param->usV_SyncStart); - conv_param.usV_SyncWidth = cpu_to_le16(crtc_param->usV_SyncWidth); - conv_param.susModeMiscInfo.usAccess = - cpu_to_le16(crtc_param->susModeMiscInfo.usAccess); - conv_param.ucCRTC = crtc_param->ucCRTC; - conv_param.ucOverscanRight = crtc_param->ucOverscanRight; - conv_param.ucOverscanLeft = crtc_param->ucOverscanLeft; - conv_param.ucOverscanBottom = crtc_param->ucOverscanBottom; - conv_param.ucOverscanTop = crtc_param->ucOverscanTop; - conv_param.ucReserved = crtc_param->ucReserved; + memset(&args, 0, sizeof(args)); + args.usH_Total = cpu_to_le16(mode->crtc_htotal); + args.usH_Disp = cpu_to_le16(mode->crtc_hdisplay); + args.usH_SyncStart = cpu_to_le16(mode->crtc_hsync_start); + args.usH_SyncWidth = + cpu_to_le16(mode->crtc_hsync_end - mode->crtc_hsync_start); + args.usV_Total = cpu_to_le16(mode->crtc_vtotal); + args.usV_Disp = cpu_to_le16(mode->crtc_vdisplay); + args.usV_SyncStart = cpu_to_le16(mode->crtc_vsync_start); + args.usV_SyncWidth = + cpu_to_le16(mode->crtc_vsync_end - mode->crtc_vsync_start); + + if (mode->flags & DRM_MODE_FLAG_NVSYNC) + misc |= ATOM_VSYNC_POLARITY; + if (mode->flags & DRM_MODE_FLAG_NHSYNC) + misc |= ATOM_HSYNC_POLARITY; + if (mode->flags & DRM_MODE_FLAG_CSYNC) + misc |= ATOM_COMPOSITESYNC; + if (mode->flags & DRM_MODE_FLAG_INTERLACE) + misc |= ATOM_INTERLACE; + if (mode->flags & DRM_MODE_FLAG_DBLSCAN) + misc |= ATOM_DOUBLE_CLOCK_MODE; + + args.susModeMiscInfo.usAccess = cpu_to_le16(misc); + args.ucCRTC = radeon_crtc->crtc_id; printk("executing set crtc timing\n"); - atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&conv_param); + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); +} + +static void atombios_set_ss(struct drm_crtc *crtc, int enable) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct drm_device *dev = crtc->dev; + struct radeon_device *rdev = dev->dev_private; + struct drm_encoder *encoder = NULL; + struct radeon_encoder *radeon_encoder = NULL; + struct radeon_encoder_atom_dig *dig = NULL; + int index = GetIndexIntoMasterTable(COMMAND, EnableSpreadSpectrumOnPPLL); + ENABLE_SPREAD_SPECTRUM_ON_PPLL_PS_ALLOCATION args; + ENABLE_LVDS_SS_PARAMETERS legacy_args; + uint16_t percentage = 0; + uint8_t type = 0, step = 0, delay = 0, range = 0; + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + if (encoder->crtc == crtc) { + radeon_encoder = to_radeon_encoder(encoder); + /* only enable spread spectrum on LVDS */ + if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) { + dig = radeon_encoder->enc_priv; + if (dig && dig->ss) { + percentage = dig->ss->percentage; + type = dig->ss->type; + step = dig->ss->step; + delay = dig->ss->delay; + range = dig->ss->range; + } else if (enable) + return; + } else if (enable) + return; + break; + } + } + + if (!radeon_encoder) + return; + + if (ASIC_IS_AVIVO(rdev)) { + memset(&args, 0, sizeof(args)); + args.usSpreadSpectrumPercentage = cpu_to_le16(percentage); + args.ucSpreadSpectrumType = type; + args.ucSpreadSpectrumStep = step; + args.ucSpreadSpectrumDelay = delay; + args.ucSpreadSpectrumRange = range; + args.ucPpll = radeon_crtc->crtc_id ? ATOM_PPLL2 : ATOM_PPLL1; + args.ucEnable = enable; + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + } else { + memset(&legacy_args, 0, sizeof(legacy_args)); + legacy_args.usSpreadSpectrumPercentage = cpu_to_le16(percentage); + legacy_args.ucSpreadSpectrumType = type; + legacy_args.ucSpreadSpectrumStepSize_Delay = (step & 3) << 2; + legacy_args.ucSpreadSpectrumStepSize_Delay |= (delay & 7) << 4; + legacy_args.ucEnable = enable; + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&legacy_args); + } } void atombios_crtc_set_pll(struct drm_crtc *crtc, struct drm_display_mode *mode) @@ -333,12 +417,13 @@ void atombios_crtc_set_pll(struct drm_crtc *crtc, struct drm_display_mode *mode) struct drm_encoder *encoder = NULL; struct radeon_encoder *radeon_encoder = NULL; uint8_t frev, crev; - int index = GetIndexIntoMasterTable(COMMAND, SetPixelClock); + int index; SET_PIXEL_CLOCK_PS_ALLOCATION args; PIXEL_CLOCK_PARAMETERS *spc1_ptr; PIXEL_CLOCK_PARAMETERS_V2 *spc2_ptr; PIXEL_CLOCK_PARAMETERS_V3 *spc3_ptr; - uint32_t sclock = mode->clock; + uint32_t pll_clock = mode->clock; + uint32_t adjusted_clock; uint32_t ref_div = 0, fb_div = 0, frac_fb_div = 0, post_div = 0; struct radeon_pll *pll; int pll_flags = 0; @@ -346,8 +431,6 @@ void atombios_crtc_set_pll(struct drm_crtc *crtc, struct drm_display_mode *mode) memset(&args, 0, sizeof(args)); if (ASIC_IS_AVIVO(rdev)) { - uint32_t ss_cntl; - if ((rdev->family == CHIP_RS600) || (rdev->family == CHIP_RS690) || (rdev->family == CHIP_RS740)) @@ -358,15 +441,6 @@ void atombios_crtc_set_pll(struct drm_crtc *crtc, struct drm_display_mode *mode) pll_flags |= RADEON_PLL_PREFER_HIGH_FB_DIV; else pll_flags |= RADEON_PLL_PREFER_LOW_REF_DIV; - - /* disable spread spectrum clocking for now -- thanks Hedy Lamarr */ - if (radeon_crtc->crtc_id == 0) { - ss_cntl = RREG32(AVIVO_P1PLL_INT_SS_CNTL); - WREG32(AVIVO_P1PLL_INT_SS_CNTL, ss_cntl & ~1); - } else { - ss_cntl = RREG32(AVIVO_P2PLL_INT_SS_CNTL); - WREG32(AVIVO_P2PLL_INT_SS_CNTL, ss_cntl & ~1); - } } else { pll_flags |= RADEON_PLL_LEGACY; @@ -393,14 +467,43 @@ void atombios_crtc_set_pll(struct drm_crtc *crtc, struct drm_display_mode *mode) } } + /* DCE3+ has an AdjustDisplayPll that will adjust the pixel clock + * accordingly based on the encoder/transmitter to work around + * special hw requirements. + */ + if (ASIC_IS_DCE3(rdev)) { + ADJUST_DISPLAY_PLL_PS_ALLOCATION adjust_pll_args; + + if (!encoder) + return; + + memset(&adjust_pll_args, 0, sizeof(adjust_pll_args)); + adjust_pll_args.usPixelClock = cpu_to_le16(mode->clock / 10); + adjust_pll_args.ucTransmitterID = radeon_encoder->encoder_id; + adjust_pll_args.ucEncodeMode = atombios_get_encoder_mode(encoder); + + index = GetIndexIntoMasterTable(COMMAND, AdjustDisplayPll); + atom_execute_table(rdev->mode_info.atom_context, + index, (uint32_t *)&adjust_pll_args); + adjusted_clock = le16_to_cpu(adjust_pll_args.usPixelClock) * 10; + } else { + /* DVO wants 2x pixel clock if the DVO chip is in 12 bit mode */ + if (ASIC_IS_AVIVO(rdev) && + (radeon_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1)) + adjusted_clock = mode->clock * 2; + else + adjusted_clock = mode->clock; + } + if (radeon_crtc->crtc_id == 0) pll = &rdev->clock.p1pll; else pll = &rdev->clock.p2pll; - radeon_compute_pll(pll, mode->clock, &sclock, &fb_div, &frac_fb_div, + radeon_compute_pll(pll, adjusted_clock, &pll_clock, &fb_div, &frac_fb_div, &ref_div, &post_div, pll_flags); + index = GetIndexIntoMasterTable(COMMAND, SetPixelClock); atom_parse_cmd_header(rdev->mode_info.atom_context, index, &frev, &crev); @@ -409,7 +512,7 @@ void atombios_crtc_set_pll(struct drm_crtc *crtc, struct drm_display_mode *mode) switch (crev) { case 1: spc1_ptr = (PIXEL_CLOCK_PARAMETERS *) & args.sPCLKInput; - spc1_ptr->usPixelClock = cpu_to_le16(sclock); + spc1_ptr->usPixelClock = cpu_to_le16(mode->clock / 10); spc1_ptr->usRefDiv = cpu_to_le16(ref_div); spc1_ptr->usFbDiv = cpu_to_le16(fb_div); spc1_ptr->ucFracFbDiv = frac_fb_div; @@ -422,7 +525,7 @@ void atombios_crtc_set_pll(struct drm_crtc *crtc, struct drm_display_mode *mode) case 2: spc2_ptr = (PIXEL_CLOCK_PARAMETERS_V2 *) & args.sPCLKInput; - spc2_ptr->usPixelClock = cpu_to_le16(sclock); + spc2_ptr->usPixelClock = cpu_to_le16(mode->clock / 10); spc2_ptr->usRefDiv = cpu_to_le16(ref_div); spc2_ptr->usFbDiv = cpu_to_le16(fb_div); spc2_ptr->ucFracFbDiv = frac_fb_div; @@ -437,7 +540,7 @@ void atombios_crtc_set_pll(struct drm_crtc *crtc, struct drm_display_mode *mode) return; spc3_ptr = (PIXEL_CLOCK_PARAMETERS_V3 *) & args.sPCLKInput; - spc3_ptr->usPixelClock = cpu_to_le16(sclock); + spc3_ptr->usPixelClock = cpu_to_le16(mode->clock / 10); spc3_ptr->usRefDiv = cpu_to_le16(ref_div); spc3_ptr->usFbDiv = cpu_to_le16(fb_div); spc3_ptr->ucFracFbDiv = frac_fb_div; @@ -527,6 +630,16 @@ int atombios_crtc_set_base(struct drm_crtc *crtc, int x, int y, WREG32(AVIVO_D1VGA_CONTROL, 0); else WREG32(AVIVO_D2VGA_CONTROL, 0); + + if (rdev->family >= CHIP_RV770) { + if (radeon_crtc->crtc_id) { + WREG32(R700_D2GRPH_PRIMARY_SURFACE_ADDRESS_HIGH, 0); + WREG32(R700_D2GRPH_SECONDARY_SURFACE_ADDRESS_HIGH, 0); + } else { + WREG32(R700_D1GRPH_PRIMARY_SURFACE_ADDRESS_HIGH, 0); + WREG32(R700_D1GRPH_SECONDARY_SURFACE_ADDRESS_HIGH, 0); + } + } WREG32(AVIVO_D1GRPH_PRIMARY_SURFACE_ADDRESS + radeon_crtc->crtc_offset, (u32) fb_location); WREG32(AVIVO_D1GRPH_SECONDARY_SURFACE_ADDRESS + @@ -563,6 +676,10 @@ int atombios_crtc_set_base(struct drm_crtc *crtc, int x, int y, radeon_fb = to_radeon_framebuffer(old_fb); radeon_gem_object_unpin(radeon_fb->obj); } + + /* Bytes per pixel may have changed */ + radeon_bandwidth_update(rdev); + return 0; } @@ -574,134 +691,24 @@ int atombios_crtc_mode_set(struct drm_crtc *crtc, struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); struct drm_device *dev = crtc->dev; struct radeon_device *rdev = dev->dev_private; - struct drm_encoder *encoder; - SET_CRTC_TIMING_PARAMETERS_PS_ALLOCATION crtc_timing; - int need_tv_timings = 0; - bool ret; /* TODO color tiling */ - memset(&crtc_timing, 0, sizeof(crtc_timing)); - - list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { - /* find tv std */ - if (encoder->crtc == crtc) { - struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); - - if (radeon_encoder->active_device & ATOM_DEVICE_TV_SUPPORT) { - struct radeon_encoder_atom_dac *tv_dac = radeon_encoder->enc_priv; - if (tv_dac) { - if (tv_dac->tv_std == TV_STD_NTSC || - tv_dac->tv_std == TV_STD_NTSC_J || - tv_dac->tv_std == TV_STD_PAL_M) - need_tv_timings = 1; - else - need_tv_timings = 2; - break; - } - } - } - } - - crtc_timing.ucCRTC = radeon_crtc->crtc_id; - if (need_tv_timings) { - ret = radeon_atom_get_tv_timings(rdev, need_tv_timings - 1, - &crtc_timing, &adjusted_mode->clock); - if (ret == false) - need_tv_timings = 0; - } - - if (!need_tv_timings) { - crtc_timing.usH_Total = adjusted_mode->crtc_htotal; - crtc_timing.usH_Disp = adjusted_mode->crtc_hdisplay; - crtc_timing.usH_SyncStart = adjusted_mode->crtc_hsync_start; - crtc_timing.usH_SyncWidth = - adjusted_mode->crtc_hsync_end - adjusted_mode->crtc_hsync_start; - - crtc_timing.usV_Total = adjusted_mode->crtc_vtotal; - crtc_timing.usV_Disp = adjusted_mode->crtc_vdisplay; - crtc_timing.usV_SyncStart = adjusted_mode->crtc_vsync_start; - crtc_timing.usV_SyncWidth = - adjusted_mode->crtc_vsync_end - adjusted_mode->crtc_vsync_start; - - if (adjusted_mode->flags & DRM_MODE_FLAG_NVSYNC) - crtc_timing.susModeMiscInfo.usAccess |= ATOM_VSYNC_POLARITY; - - if (adjusted_mode->flags & DRM_MODE_FLAG_NHSYNC) - crtc_timing.susModeMiscInfo.usAccess |= ATOM_HSYNC_POLARITY; - - if (adjusted_mode->flags & DRM_MODE_FLAG_CSYNC) - crtc_timing.susModeMiscInfo.usAccess |= ATOM_COMPOSITESYNC; - - if (adjusted_mode->flags & DRM_MODE_FLAG_INTERLACE) - crtc_timing.susModeMiscInfo.usAccess |= ATOM_INTERLACE; - - if (adjusted_mode->flags & DRM_MODE_FLAG_DBLSCAN) - crtc_timing.susModeMiscInfo.usAccess |= ATOM_DOUBLE_CLOCK_MODE; - } + atombios_set_ss(crtc, 0); atombios_crtc_set_pll(crtc, adjusted_mode); - atombios_crtc_set_timing(crtc, &crtc_timing); + atombios_set_ss(crtc, 1); + atombios_crtc_set_timing(crtc, adjusted_mode); if (ASIC_IS_AVIVO(rdev)) atombios_crtc_set_base(crtc, x, y, old_fb); else { - if (radeon_crtc->crtc_id == 0) { - SET_CRTC_USING_DTD_TIMING_PARAMETERS crtc_dtd_timing; - memset(&crtc_dtd_timing, 0, sizeof(crtc_dtd_timing)); - - /* setup FP shadow regs on R4xx */ - crtc_dtd_timing.ucCRTC = radeon_crtc->crtc_id; - crtc_dtd_timing.usH_Size = adjusted_mode->crtc_hdisplay; - crtc_dtd_timing.usV_Size = adjusted_mode->crtc_vdisplay; - crtc_dtd_timing.usH_Blanking_Time = - adjusted_mode->crtc_hblank_end - - adjusted_mode->crtc_hdisplay; - crtc_dtd_timing.usV_Blanking_Time = - adjusted_mode->crtc_vblank_end - - adjusted_mode->crtc_vdisplay; - crtc_dtd_timing.usH_SyncOffset = - adjusted_mode->crtc_hsync_start - - adjusted_mode->crtc_hdisplay; - crtc_dtd_timing.usV_SyncOffset = - adjusted_mode->crtc_vsync_start - - adjusted_mode->crtc_vdisplay; - crtc_dtd_timing.usH_SyncWidth = - adjusted_mode->crtc_hsync_end - - adjusted_mode->crtc_hsync_start; - crtc_dtd_timing.usV_SyncWidth = - adjusted_mode->crtc_vsync_end - - adjusted_mode->crtc_vsync_start; - /* crtc_dtd_timing.ucH_Border = adjusted_mode->crtc_hborder; */ - /* crtc_dtd_timing.ucV_Border = adjusted_mode->crtc_vborder; */ - - if (adjusted_mode->flags & DRM_MODE_FLAG_NVSYNC) - crtc_dtd_timing.susModeMiscInfo.usAccess |= - ATOM_VSYNC_POLARITY; - - if (adjusted_mode->flags & DRM_MODE_FLAG_NHSYNC) - crtc_dtd_timing.susModeMiscInfo.usAccess |= - ATOM_HSYNC_POLARITY; - - if (adjusted_mode->flags & DRM_MODE_FLAG_CSYNC) - crtc_dtd_timing.susModeMiscInfo.usAccess |= - ATOM_COMPOSITESYNC; - - if (adjusted_mode->flags & DRM_MODE_FLAG_INTERLACE) - crtc_dtd_timing.susModeMiscInfo.usAccess |= - ATOM_INTERLACE; - - if (adjusted_mode->flags & DRM_MODE_FLAG_DBLSCAN) - crtc_dtd_timing.susModeMiscInfo.usAccess |= - ATOM_DOUBLE_CLOCK_MODE; - - atombios_set_crtc_dtd_timing(crtc, &crtc_dtd_timing); - } + if (radeon_crtc->crtc_id == 0) + atombios_set_crtc_dtd_timing(crtc, adjusted_mode); radeon_crtc_set_base(crtc, x, y, old_fb); radeon_legacy_atom_set_surface(crtc); } atombios_overscan_setup(crtc, mode, adjusted_mode); atombios_scaler_setup(crtc); - radeon_bandwidth_update(rdev); return 0; } @@ -733,6 +740,7 @@ static const struct drm_crtc_helper_funcs atombios_helper_funcs = { .mode_set_base = atombios_crtc_set_base, .prepare = atombios_crtc_prepare, .commit = atombios_crtc_commit, + .load_lut = radeon_crtc_load_lut, }; void radeon_atombios_init_crtc(struct drm_device *dev, diff --git a/drivers/gpu/drm/radeon/avivod.h b/drivers/gpu/drm/radeon/avivod.h index e2b92c4..d4e6e6e 100644 --- a/drivers/gpu/drm/radeon/avivod.h +++ b/drivers/gpu/drm/radeon/avivod.h @@ -57,13 +57,4 @@ #define VGA_RENDER_CONTROL 0x0300 #define VGA_VSTATUS_CNTL_MASK 0x00030000 -/* AVIVO disable VGA rendering */ -static inline void radeon_avivo_vga_render_disable(struct radeon_device *rdev) -{ - u32 vga_render; - vga_render = RREG32(VGA_RENDER_CONTROL); - vga_render &= ~VGA_VSTATUS_CNTL_MASK; - WREG32(VGA_RENDER_CONTROL, vga_render); -} - #endif diff --git a/drivers/gpu/drm/radeon/mkregtable.c b/drivers/gpu/drm/radeon/mkregtable.c index fb211e5..0d79577 100644 --- a/drivers/gpu/drm/radeon/mkregtable.c +++ b/drivers/gpu/drm/radeon/mkregtable.c @@ -561,7 +561,7 @@ struct table { char *gpu_prefix; }; -struct offset *offset_new(unsigned o) +static struct offset *offset_new(unsigned o) { struct offset *offset; @@ -573,12 +573,12 @@ struct offset *offset_new(unsigned o) return offset; } -void table_offset_add(struct table *t, struct offset *offset) +static void table_offset_add(struct table *t, struct offset *offset) { list_add_tail(&offset->list, &t->offsets); } -void table_init(struct table *t) +static void table_init(struct table *t) { INIT_LIST_HEAD(&t->offsets); t->offset_max = 0; @@ -586,7 +586,7 @@ void table_init(struct table *t) t->table = NULL; } -void table_print(struct table *t) +static void table_print(struct table *t) { unsigned nlloop, i, j, n, c, id; @@ -611,7 +611,7 @@ void table_print(struct table *t) printf("};\n"); } -int table_build(struct table *t) +static int table_build(struct table *t) { struct offset *offset; unsigned i, m; @@ -631,7 +631,7 @@ int table_build(struct table *t) } static char gpu_name[10]; -int parser_auth(struct table *t, const char *filename) +static int parser_auth(struct table *t, const char *filename) { FILE *file; regex_t mask_rex; diff --git a/drivers/gpu/drm/radeon/r100.c b/drivers/gpu/drm/radeon/r100.c index be51c5f..c9e93ea 100644 --- a/drivers/gpu/drm/radeon/r100.c +++ b/drivers/gpu/drm/radeon/r100.c @@ -32,6 +32,9 @@ #include "radeon_reg.h" #include "radeon.h" #include "r100d.h" +#include "rs100d.h" +#include "rv200d.h" +#include "rv250d.h" #include <linux/firmware.h> #include <linux/platform_device.h> @@ -60,18 +63,7 @@ MODULE_FIRMWARE(FIRMWARE_R520); /* This files gather functions specifics to: * r100,rv100,rs100,rv200,rs200,r200,rv250,rs300,rv280 - * - * Some of these functions might be used by newer ASICs. */ -int r200_init(struct radeon_device *rdev); -void r100_hdp_reset(struct radeon_device *rdev); -void r100_gpu_init(struct radeon_device *rdev); -int r100_gui_wait_for_idle(struct radeon_device *rdev); -int r100_mc_wait_for_idle(struct radeon_device *rdev); -void r100_gpu_wait_for_vsync(struct radeon_device *rdev); -void r100_gpu_wait_for_vsync2(struct radeon_device *rdev); -int r100_debugfs_mc_info_init(struct radeon_device *rdev); - /* * PCI GART @@ -152,136 +144,6 @@ void r100_pci_gart_fini(struct radeon_device *rdev) radeon_gart_fini(rdev); } - -/* - * MC - */ -void r100_mc_disable_clients(struct radeon_device *rdev) -{ - uint32_t ov0_scale_cntl, crtc_ext_cntl, crtc_gen_cntl, crtc2_gen_cntl; - - /* FIXME: is this function correct for rs100,rs200,rs300 ? */ - if (r100_gui_wait_for_idle(rdev)) { - printk(KERN_WARNING "Failed to wait GUI idle while " - "programming pipes. Bad things might happen.\n"); - } - - /* stop display and memory access */ - ov0_scale_cntl = RREG32(RADEON_OV0_SCALE_CNTL); - WREG32(RADEON_OV0_SCALE_CNTL, ov0_scale_cntl & ~RADEON_SCALER_ENABLE); - crtc_ext_cntl = RREG32(RADEON_CRTC_EXT_CNTL); - WREG32(RADEON_CRTC_EXT_CNTL, crtc_ext_cntl | RADEON_CRTC_DISPLAY_DIS); - crtc_gen_cntl = RREG32(RADEON_CRTC_GEN_CNTL); - - r100_gpu_wait_for_vsync(rdev); - - WREG32(RADEON_CRTC_GEN_CNTL, - (crtc_gen_cntl & ~(RADEON_CRTC_CUR_EN | RADEON_CRTC_ICON_EN)) | - RADEON_CRTC_DISP_REQ_EN_B | RADEON_CRTC_EXT_DISP_EN); - - if (!(rdev->flags & RADEON_SINGLE_CRTC)) { - crtc2_gen_cntl = RREG32(RADEON_CRTC2_GEN_CNTL); - - r100_gpu_wait_for_vsync2(rdev); - WREG32(RADEON_CRTC2_GEN_CNTL, - (crtc2_gen_cntl & - ~(RADEON_CRTC2_CUR_EN | RADEON_CRTC2_ICON_EN)) | - RADEON_CRTC2_DISP_REQ_EN_B); - } - - udelay(500); -} - -void r100_mc_setup(struct radeon_device *rdev) -{ - uint32_t tmp; - int r; - - r = r100_debugfs_mc_info_init(rdev); - if (r) { - DRM_ERROR("Failed to register debugfs file for R100 MC !\n"); - } - /* Write VRAM size in case we are limiting it */ - WREG32(RADEON_CONFIG_MEMSIZE, rdev->mc.real_vram_size); - /* Novell bug 204882 for RN50/M6/M7 with 8/16/32MB VRAM, - * if the aperture is 64MB but we have 32MB VRAM - * we report only 32MB VRAM but we have to set MC_FB_LOCATION - * to 64MB, otherwise the gpu accidentially dies */ - tmp = rdev->mc.vram_location + rdev->mc.mc_vram_size - 1; - tmp = REG_SET(RADEON_MC_FB_TOP, tmp >> 16); - tmp |= REG_SET(RADEON_MC_FB_START, rdev->mc.vram_location >> 16); - WREG32(RADEON_MC_FB_LOCATION, tmp); - - /* Enable bus mastering */ - tmp = RREG32(RADEON_BUS_CNTL) & ~RADEON_BUS_MASTER_DIS; - WREG32(RADEON_BUS_CNTL, tmp); - - if (rdev->flags & RADEON_IS_AGP) { - tmp = rdev->mc.gtt_location + rdev->mc.gtt_size - 1; - tmp = REG_SET(RADEON_MC_AGP_TOP, tmp >> 16); - tmp |= REG_SET(RADEON_MC_AGP_START, rdev->mc.gtt_location >> 16); - WREG32(RADEON_MC_AGP_LOCATION, tmp); - WREG32(RADEON_AGP_BASE, rdev->mc.agp_base); - } else { - WREG32(RADEON_MC_AGP_LOCATION, 0x0FFFFFFF); - WREG32(RADEON_AGP_BASE, 0); - } - - tmp = RREG32(RADEON_HOST_PATH_CNTL) & RADEON_HDP_APER_CNTL; - tmp |= (7 << 28); - WREG32(RADEON_HOST_PATH_CNTL, tmp | RADEON_HDP_SOFT_RESET | RADEON_HDP_READ_BUFFER_INVALIDATE); - (void)RREG32(RADEON_HOST_PATH_CNTL); - WREG32(RADEON_HOST_PATH_CNTL, tmp); - (void)RREG32(RADEON_HOST_PATH_CNTL); -} - -int r100_mc_init(struct radeon_device *rdev) -{ - int r; - - if (r100_debugfs_rbbm_init(rdev)) { - DRM_ERROR("Failed to register debugfs file for RBBM !\n"); - } - - r100_gpu_init(rdev); - /* Disable gart which also disable out of gart access */ - r100_pci_gart_disable(rdev); - - /* Setup GPU memory space */ - rdev->mc.gtt_location = 0xFFFFFFFFUL; - if (rdev->flags & RADEON_IS_AGP) { - r = radeon_agp_init(rdev); - if (r) { - printk(KERN_WARNING "[drm] Disabling AGP\n"); - rdev->flags &= ~RADEON_IS_AGP; - rdev->mc.gtt_size = radeon_gart_size * 1024 * 1024; - } else { - rdev->mc.gtt_location = rdev->mc.agp_base; - } - } - r = radeon_mc_setup(rdev); - if (r) { - return r; - } - - r100_mc_disable_clients(rdev); - if (r100_mc_wait_for_idle(rdev)) { - printk(KERN_WARNING "Failed to wait MC idle while " - "programming pipes. Bad things might happen.\n"); - } - - r100_mc_setup(rdev); - return 0; -} - -void r100_mc_fini(struct radeon_device *rdev) -{ -} - - -/* - * Interrupts - */ int r100_irq_set(struct radeon_device *rdev) { uint32_t tmp = 0; @@ -324,7 +186,7 @@ static inline uint32_t r100_irq_ack(struct radeon_device *rdev) int r100_irq_process(struct radeon_device *rdev) { - uint32_t status; + uint32_t status, msi_rearm; status = r100_irq_ack(rdev); if (!status) { @@ -347,6 +209,21 @@ int r100_irq_process(struct radeon_device *rdev) } status = r100_irq_ack(rdev); } + if (rdev->msi_enabled) { + switch (rdev->family) { + case CHIP_RS400: + case CHIP_RS480: + msi_rearm = RREG32(RADEON_AIC_CNTL) & ~RS400_MSI_REARM; + WREG32(RADEON_AIC_CNTL, msi_rearm); + WREG32(RADEON_AIC_CNTL, msi_rearm | RS400_MSI_REARM); + break; + default: + msi_rearm = RREG32(RADEON_MSI_REARM_EN) & ~RV370_MSI_REARM_EN; + WREG32(RADEON_MSI_REARM_EN, msi_rearm); + WREG32(RADEON_MSI_REARM_EN, msi_rearm | RV370_MSI_REARM_EN); + break; + } + } return IRQ_HANDLED; } @@ -358,10 +235,6 @@ u32 r100_get_vblank_counter(struct radeon_device *rdev, int crtc) return RREG32(RADEON_CRTC2_CRNT_FRAME); } - -/* - * Fence emission - */ void r100_fence_ring_emit(struct radeon_device *rdev, struct radeon_fence *fence) { @@ -377,16 +250,12 @@ void r100_fence_ring_emit(struct radeon_device *rdev, radeon_ring_write(rdev, RADEON_SW_INT_FIRE); } - -/* - * Writeback - */ int r100_wb_init(struct radeon_device *rdev) { int r; if (rdev->wb.wb_obj == NULL) { - r = radeon_object_create(rdev, NULL, 4096, + r = radeon_object_create(rdev, NULL, RADEON_GPU_PAGE_SIZE, true, RADEON_GEM_DOMAIN_GTT, false, &rdev->wb.wb_obj); @@ -504,10 +373,6 @@ int r100_copy_blit(struct radeon_device *rdev, return r; } - -/* - * CP - */ static int r100_cp_wait_for_idle(struct radeon_device *rdev) { unsigned i; @@ -612,6 +477,7 @@ static int r100_cp_init_microcode(struct radeon_device *rdev) } return err; } + static void r100_cp_load_microcode(struct radeon_device *rdev) { const __be32 *fw_data; @@ -712,19 +578,19 @@ int r100_cp_init(struct radeon_device *rdev, unsigned ring_size) indirect1_start = 16; /* cp setup */ WREG32(0x718, pre_write_timer | (pre_write_limit << 28)); - WREG32(RADEON_CP_RB_CNTL, -#ifdef __BIG_ENDIAN - RADEON_BUF_SWAP_32BIT | -#endif - REG_SET(RADEON_RB_BUFSZ, rb_bufsz) | + tmp = (REG_SET(RADEON_RB_BUFSZ, rb_bufsz) | REG_SET(RADEON_RB_BLKSZ, rb_blksz) | REG_SET(RADEON_MAX_FETCH, max_fetch) | RADEON_RB_NO_UPDATE); +#ifdef __BIG_ENDIAN + tmp |= RADEON_BUF_SWAP_32BIT; +#endif + WREG32(RADEON_CP_RB_CNTL, tmp); + /* Set ring address */ DRM_INFO("radeon: ring at 0x%016lX\n", (unsigned long)rdev->cp.gpu_addr); WREG32(RADEON_CP_RB_BASE, rdev->cp.gpu_addr); /* Force read & write ptr to 0 */ - tmp = RREG32(RADEON_CP_RB_CNTL); WREG32(RADEON_CP_RB_CNTL, tmp | RADEON_RB_RPTR_WR_ENA); WREG32(RADEON_CP_RB_RPTR_WR, 0); WREG32(RADEON_CP_RB_WPTR, 0); @@ -863,13 +729,11 @@ int r100_cs_parse_packet0(struct radeon_cs_parser *p, void r100_cs_dump_packet(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt) { - struct radeon_cs_chunk *ib_chunk; volatile uint32_t *ib; unsigned i; unsigned idx; ib = p->ib->ptr; - ib_chunk = &p->chunks[p->chunk_ib_idx]; idx = pkt->idx; for (i = 0; i <= (pkt->count + 1); i++, idx++) { DRM_INFO("ib[%d]=0x%08X\n", idx, ib[idx]); @@ -896,7 +760,7 @@ int r100_cs_packet_parse(struct radeon_cs_parser *p, idx, ib_chunk->length_dw); return -EINVAL; } - header = ib_chunk->kdata[idx]; + header = radeon_get_ib_value(p, idx); pkt->idx = idx; pkt->type = CP_PACKET_GET_TYPE(header); pkt->count = CP_PACKET_GET_COUNT(header); @@ -939,7 +803,6 @@ int r100_cs_packet_parse(struct radeon_cs_parser *p, */ int r100_cs_packet_parse_vline(struct radeon_cs_parser *p) { - struct radeon_cs_chunk *ib_chunk; struct drm_mode_object *obj; struct drm_crtc *crtc; struct radeon_crtc *radeon_crtc; @@ -947,8 +810,9 @@ int r100_cs_packet_parse_vline(struct radeon_cs_parser *p) int crtc_id; int r; uint32_t header, h_idx, reg; + volatile uint32_t *ib; - ib_chunk = &p->chunks[p->chunk_ib_idx]; + ib = p->ib->ptr; /* parse the wait until */ r = r100_cs_packet_parse(p, &waitreloc, p->idx); @@ -963,24 +827,24 @@ int r100_cs_packet_parse_vline(struct radeon_cs_parser *p) return r; } - if (ib_chunk->kdata[waitreloc.idx + 1] != RADEON_WAIT_CRTC_VLINE) { + if (radeon_get_ib_value(p, waitreloc.idx + 1) != RADEON_WAIT_CRTC_VLINE) { DRM_ERROR("vline wait had illegal wait until\n"); r = -EINVAL; return r; } /* jump over the NOP */ - r = r100_cs_packet_parse(p, &p3reloc, p->idx); + r = r100_cs_packet_parse(p, &p3reloc, p->idx + waitreloc.count + 2); if (r) return r; h_idx = p->idx - 2; - p->idx += waitreloc.count; - p->idx += p3reloc.count; + p->idx += waitreloc.count + 2; + p->idx += p3reloc.count + 2; - header = ib_chunk->kdata[h_idx]; - crtc_id = ib_chunk->kdata[h_idx + 5]; - reg = ib_chunk->kdata[h_idx] >> 2; + header = radeon_get_ib_value(p, h_idx); + crtc_id = radeon_get_ib_value(p, h_idx + 5); + reg = CP_PACKET0_GET_REG(header); mutex_lock(&p->rdev->ddev->mode_config.mutex); obj = drm_mode_object_find(p->rdev->ddev, crtc_id, DRM_MODE_OBJECT_CRTC); if (!obj) { @@ -994,16 +858,16 @@ int r100_cs_packet_parse_vline(struct radeon_cs_parser *p) if (!crtc->enabled) { /* if the CRTC isn't enabled - we need to nop out the wait until */ - ib_chunk->kdata[h_idx + 2] = PACKET2(0); - ib_chunk->kdata[h_idx + 3] = PACKET2(0); + ib[h_idx + 2] = PACKET2(0); + ib[h_idx + 3] = PACKET2(0); } else if (crtc_id == 1) { switch (reg) { case AVIVO_D1MODE_VLINE_START_END: - header &= R300_CP_PACKET0_REG_MASK; + header &= ~R300_CP_PACKET0_REG_MASK; header |= AVIVO_D2MODE_VLINE_START_END >> 2; break; case RADEON_CRTC_GUI_TRIG_VLINE: - header &= R300_CP_PACKET0_REG_MASK; + header &= ~R300_CP_PACKET0_REG_MASK; header |= RADEON_CRTC2_GUI_TRIG_VLINE >> 2; break; default: @@ -1011,8 +875,8 @@ int r100_cs_packet_parse_vline(struct radeon_cs_parser *p) r = -EINVAL; goto out; } - ib_chunk->kdata[h_idx] = header; - ib_chunk->kdata[h_idx + 3] |= RADEON_ENG_DISPLAY_SELECT_CRTC1; + ib[h_idx] = header; + ib[h_idx + 3] |= RADEON_ENG_DISPLAY_SELECT_CRTC1; } out: mutex_unlock(&p->rdev->ddev->mode_config.mutex); @@ -1033,7 +897,6 @@ out: int r100_cs_packet_next_reloc(struct radeon_cs_parser *p, struct radeon_cs_reloc **cs_reloc) { - struct radeon_cs_chunk *ib_chunk; struct radeon_cs_chunk *relocs_chunk; struct radeon_cs_packet p3reloc; unsigned idx; @@ -1044,7 +907,6 @@ int r100_cs_packet_next_reloc(struct radeon_cs_parser *p, return -EINVAL; } *cs_reloc = NULL; - ib_chunk = &p->chunks[p->chunk_ib_idx]; relocs_chunk = &p->chunks[p->chunk_relocs_idx]; r = r100_cs_packet_parse(p, &p3reloc, p->idx); if (r) { @@ -1057,7 +919,7 @@ int r100_cs_packet_next_reloc(struct radeon_cs_parser *p, r100_cs_dump_packet(p, &p3reloc); return -EINVAL; } - idx = ib_chunk->kdata[p3reloc.idx + 1]; + idx = radeon_get_ib_value(p, p3reloc.idx + 1); if (idx >= relocs_chunk->length_dw) { DRM_ERROR("Relocs at %d after relocations chunk end %d !\n", idx, relocs_chunk->length_dw); @@ -1126,7 +988,6 @@ static int r100_packet0_check(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt, unsigned idx, unsigned reg) { - struct radeon_cs_chunk *ib_chunk; struct radeon_cs_reloc *reloc; struct r100_cs_track *track; volatile uint32_t *ib; @@ -1134,11 +995,13 @@ static int r100_packet0_check(struct radeon_cs_parser *p, int r; int i, face; u32 tile_flags = 0; + u32 idx_value; ib = p->ib->ptr; - ib_chunk = &p->chunks[p->chunk_ib_idx]; track = (struct r100_cs_track *)p->track; + idx_value = radeon_get_ib_value(p, idx); + switch (reg) { case RADEON_CRTC_GUI_TRIG_VLINE: r = r100_cs_packet_parse_vline(p); @@ -1166,8 +1029,8 @@ static int r100_packet0_check(struct radeon_cs_parser *p, return r; } track->zb.robj = reloc->robj; - track->zb.offset = ib_chunk->kdata[idx]; - ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset); + track->zb.offset = idx_value; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); break; case RADEON_RB3D_COLOROFFSET: r = r100_cs_packet_next_reloc(p, &reloc); @@ -1178,8 +1041,8 @@ static int r100_packet0_check(struct radeon_cs_parser *p, return r; } track->cb[0].robj = reloc->robj; - track->cb[0].offset = ib_chunk->kdata[idx]; - ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset); + track->cb[0].offset = idx_value; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); break; case RADEON_PP_TXOFFSET_0: case RADEON_PP_TXOFFSET_1: @@ -1192,7 +1055,7 @@ static int r100_packet0_check(struct radeon_cs_parser *p, r100_cs_dump_packet(p, pkt); return r; } - ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset); + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); track->textures[i].robj = reloc->robj; break; case RADEON_PP_CUBIC_OFFSET_T0_0: @@ -1208,8 +1071,8 @@ static int r100_packet0_check(struct radeon_cs_parser *p, r100_cs_dump_packet(p, pkt); return r; } - track->textures[0].cube_info[i].offset = ib_chunk->kdata[idx]; - ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset); + track->textures[0].cube_info[i].offset = idx_value; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); track->textures[0].cube_info[i].robj = reloc->robj; break; case RADEON_PP_CUBIC_OFFSET_T1_0: @@ -1225,8 +1088,8 @@ static int r100_packet0_check(struct radeon_cs_parser *p, r100_cs_dump_packet(p, pkt); return r; } - track->textures[1].cube_info[i].offset = ib_chunk->kdata[idx]; - ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset); + track->textures[1].cube_info[i].offset = idx_value; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); track->textures[1].cube_info[i].robj = reloc->robj; break; case RADEON_PP_CUBIC_OFFSET_T2_0: @@ -1242,12 +1105,12 @@ static int r100_packet0_check(struct radeon_cs_parser *p, r100_cs_dump_packet(p, pkt); return r; } - track->textures[2].cube_info[i].offset = ib_chunk->kdata[idx]; - ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset); + track->textures[2].cube_info[i].offset = idx_value; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); track->textures[2].cube_info[i].robj = reloc->robj; break; case RADEON_RE_WIDTH_HEIGHT: - track->maxy = ((ib_chunk->kdata[idx] >> 16) & 0x7FF); + track->maxy = ((idx_value >> 16) & 0x7FF); break; case RADEON_RB3D_COLORPITCH: r = r100_cs_packet_next_reloc(p, &reloc); @@ -1263,17 +1126,17 @@ static int r100_packet0_check(struct radeon_cs_parser *p, if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) tile_flags |= RADEON_COLOR_MICROTILE_ENABLE; - tmp = ib_chunk->kdata[idx] & ~(0x7 << 16); + tmp = idx_value & ~(0x7 << 16); tmp |= tile_flags; ib[idx] = tmp; - track->cb[0].pitch = ib_chunk->kdata[idx] & RADEON_COLORPITCH_MASK; + track->cb[0].pitch = idx_value & RADEON_COLORPITCH_MASK; break; case RADEON_RB3D_DEPTHPITCH: - track->zb.pitch = ib_chunk->kdata[idx] & RADEON_DEPTHPITCH_MASK; + track->zb.pitch = idx_value & RADEON_DEPTHPITCH_MASK; break; case RADEON_RB3D_CNTL: - switch ((ib_chunk->kdata[idx] >> RADEON_RB3D_COLOR_FORMAT_SHIFT) & 0x1f) { + switch ((idx_value >> RADEON_RB3D_COLOR_FORMAT_SHIFT) & 0x1f) { case 7: case 8: case 9: @@ -1291,13 +1154,13 @@ static int r100_packet0_check(struct radeon_cs_parser *p, break; default: DRM_ERROR("Invalid color buffer format (%d) !\n", - ((ib_chunk->kdata[idx] >> RADEON_RB3D_COLOR_FORMAT_SHIFT) & 0x1f)); + ((idx_value >> RADEON_RB3D_COLOR_FORMAT_SHIFT) & 0x1f)); return -EINVAL; } - track->z_enabled = !!(ib_chunk->kdata[idx] & RADEON_Z_ENABLE); + track->z_enabled = !!(idx_value & RADEON_Z_ENABLE); break; case RADEON_RB3D_ZSTENCILCNTL: - switch (ib_chunk->kdata[idx] & 0xf) { + switch (idx_value & 0xf) { case 0: track->zb.cpp = 2; break; @@ -1321,44 +1184,44 @@ static int r100_packet0_check(struct radeon_cs_parser *p, r100_cs_dump_packet(p, pkt); return r; } - ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset); + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); break; case RADEON_PP_CNTL: { - uint32_t temp = ib_chunk->kdata[idx] >> 4; + uint32_t temp = idx_value >> 4; for (i = 0; i < track->num_texture; i++) track->textures[i].enabled = !!(temp & (1 << i)); } break; case RADEON_SE_VF_CNTL: - track->vap_vf_cntl = ib_chunk->kdata[idx]; + track->vap_vf_cntl = idx_value; break; case RADEON_SE_VTX_FMT: - track->vtx_size = r100_get_vtx_size(ib_chunk->kdata[idx]); + track->vtx_size = r100_get_vtx_size(idx_value); break; case RADEON_PP_TEX_SIZE_0: case RADEON_PP_TEX_SIZE_1: case RADEON_PP_TEX_SIZE_2: i = (reg - RADEON_PP_TEX_SIZE_0) / 8; - track->textures[i].width = (ib_chunk->kdata[idx] & RADEON_TEX_USIZE_MASK) + 1; - track->textures[i].height = ((ib_chunk->kdata[idx] & RADEON_TEX_VSIZE_MASK) >> RADEON_TEX_VSIZE_SHIFT) + 1; + track->textures[i].width = (idx_value & RADEON_TEX_USIZE_MASK) + 1; + track->textures[i].height = ((idx_value & RADEON_TEX_VSIZE_MASK) >> RADEON_TEX_VSIZE_SHIFT) + 1; break; case RADEON_PP_TEX_PITCH_0: case RADEON_PP_TEX_PITCH_1: case RADEON_PP_TEX_PITCH_2: i = (reg - RADEON_PP_TEX_PITCH_0) / 8; - track->textures[i].pitch = ib_chunk->kdata[idx] + 32; + track->textures[i].pitch = idx_value + 32; break; case RADEON_PP_TXFILTER_0: case RADEON_PP_TXFILTER_1: case RADEON_PP_TXFILTER_2: i = (reg - RADEON_PP_TXFILTER_0) / 24; - track->textures[i].num_levels = ((ib_chunk->kdata[idx] & RADEON_MAX_MIP_LEVEL_MASK) + track->textures[i].num_levels = ((idx_value & RADEON_MAX_MIP_LEVEL_MASK) >> RADEON_MAX_MIP_LEVEL_SHIFT); - tmp = (ib_chunk->kdata[idx] >> 23) & 0x7; + tmp = (idx_value >> 23) & 0x7; if (tmp == 2 || tmp == 6) track->textures[i].roundup_w = false; - tmp = (ib_chunk->kdata[idx] >> 27) & 0x7; + tmp = (idx_value >> 27) & 0x7; if (tmp == 2 || tmp == 6) track->textures[i].roundup_h = false; break; @@ -1366,16 +1229,16 @@ static int r100_packet0_check(struct radeon_cs_parser *p, case RADEON_PP_TXFORMAT_1: case RADEON_PP_TXFORMAT_2: i = (reg - RADEON_PP_TXFORMAT_0) / 24; - if (ib_chunk->kdata[idx] & RADEON_TXFORMAT_NON_POWER2) { + if (idx_value & RADEON_TXFORMAT_NON_POWER2) { track->textures[i].use_pitch = 1; } else { track->textures[i].use_pitch = 0; - track->textures[i].width = 1 << ((ib_chunk->kdata[idx] >> RADEON_TXFORMAT_WIDTH_SHIFT) & RADEON_TXFORMAT_WIDTH_MASK); - track->textures[i].height = 1 << ((ib_chunk->kdata[idx] >> RADEON_TXFORMAT_HEIGHT_SHIFT) & RADEON_TXFORMAT_HEIGHT_MASK); + track->textures[i].width = 1 << ((idx_value >> RADEON_TXFORMAT_WIDTH_SHIFT) & RADEON_TXFORMAT_WIDTH_MASK); + track->textures[i].height = 1 << ((idx_value >> RADEON_TXFORMAT_HEIGHT_SHIFT) & RADEON_TXFORMAT_HEIGHT_MASK); } - if (ib_chunk->kdata[idx] & RADEON_TXFORMAT_CUBIC_MAP_ENABLE) + if (idx_value & RADEON_TXFORMAT_CUBIC_MAP_ENABLE) track->textures[i].tex_coord_type = 2; - switch ((ib_chunk->kdata[idx] & RADEON_TXFORMAT_FORMAT_MASK)) { + switch ((idx_value & RADEON_TXFORMAT_FORMAT_MASK)) { case RADEON_TXFORMAT_I8: case RADEON_TXFORMAT_RGB332: case RADEON_TXFORMAT_Y8: @@ -1402,13 +1265,13 @@ static int r100_packet0_check(struct radeon_cs_parser *p, track->textures[i].cpp = 4; break; } - track->textures[i].cube_info[4].width = 1 << ((ib_chunk->kdata[idx] >> 16) & 0xf); - track->textures[i].cube_info[4].height = 1 << ((ib_chunk->kdata[idx] >> 20) & 0xf); + track->textures[i].cube_info[4].width = 1 << ((idx_value >> 16) & 0xf); + track->textures[i].cube_info[4].height = 1 << ((idx_value >> 20) & 0xf); break; case RADEON_PP_CUBIC_FACES_0: case RADEON_PP_CUBIC_FACES_1: case RADEON_PP_CUBIC_FACES_2: - tmp = ib_chunk->kdata[idx]; + tmp = idx_value; i = (reg - RADEON_PP_CUBIC_FACES_0) / 4; for (face = 0; face < 4; face++) { track->textures[i].cube_info[face].width = 1 << ((tmp >> (face * 8)) & 0xf); @@ -1427,15 +1290,14 @@ int r100_cs_track_check_pkt3_indx_buffer(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt, struct radeon_object *robj) { - struct radeon_cs_chunk *ib_chunk; unsigned idx; - - ib_chunk = &p->chunks[p->chunk_ib_idx]; + u32 value; idx = pkt->idx + 1; - if ((ib_chunk->kdata[idx+2] + 1) > radeon_object_size(robj)) { + value = radeon_get_ib_value(p, idx + 2); + if ((value + 1) > radeon_object_size(robj)) { DRM_ERROR("[drm] Buffer too small for PACKET3 INDX_BUFFER " "(need %u have %lu) !\n", - ib_chunk->kdata[idx+2] + 1, + value + 1, radeon_object_size(robj)); return -EINVAL; } @@ -1445,59 +1307,20 @@ int r100_cs_track_check_pkt3_indx_buffer(struct radeon_cs_parser *p, static int r100_packet3_check(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt) { - struct radeon_cs_chunk *ib_chunk; struct radeon_cs_reloc *reloc; struct r100_cs_track *track; unsigned idx; - unsigned i, c; volatile uint32_t *ib; int r; ib = p->ib->ptr; - ib_chunk = &p->chunks[p->chunk_ib_idx]; idx = pkt->idx + 1; track = (struct r100_cs_track *)p->track; switch (pkt->opcode) { case PACKET3_3D_LOAD_VBPNTR: - c = ib_chunk->kdata[idx++]; - track->num_arrays = c; - for (i = 0; i < (c - 1); i += 2, idx += 3) { - r = r100_cs_packet_next_reloc(p, &reloc); - if (r) { - DRM_ERROR("No reloc for packet3 %d\n", - pkt->opcode); - r100_cs_dump_packet(p, pkt); - return r; - } - ib[idx+1] = ib_chunk->kdata[idx+1] + ((u32)reloc->lobj.gpu_offset); - track->arrays[i + 0].robj = reloc->robj; - track->arrays[i + 0].esize = ib_chunk->kdata[idx] >> 8; - track->arrays[i + 0].esize &= 0x7F; - r = r100_cs_packet_next_reloc(p, &reloc); - if (r) { - DRM_ERROR("No reloc for packet3 %d\n", - pkt->opcode); - r100_cs_dump_packet(p, pkt); - return r; - } - ib[idx+2] = ib_chunk->kdata[idx+2] + ((u32)reloc->lobj.gpu_offset); - track->arrays[i + 1].robj = reloc->robj; - track->arrays[i + 1].esize = ib_chunk->kdata[idx] >> 24; - track->arrays[i + 1].esize &= 0x7F; - } - if (c & 1) { - r = r100_cs_packet_next_reloc(p, &reloc); - if (r) { - DRM_ERROR("No reloc for packet3 %d\n", - pkt->opcode); - r100_cs_dump_packet(p, pkt); - return r; - } - ib[idx+1] = ib_chunk->kdata[idx+1] + ((u32)reloc->lobj.gpu_offset); - track->arrays[i + 0].robj = reloc->robj; - track->arrays[i + 0].esize = ib_chunk->kdata[idx] >> 8; - track->arrays[i + 0].esize &= 0x7F; - } + r = r100_packet3_load_vbpntr(p, pkt, idx); + if (r) + return r; break; case PACKET3_INDX_BUFFER: r = r100_cs_packet_next_reloc(p, &reloc); @@ -1506,7 +1329,7 @@ static int r100_packet3_check(struct radeon_cs_parser *p, r100_cs_dump_packet(p, pkt); return r; } - ib[idx+1] = ib_chunk->kdata[idx+1] + ((u32)reloc->lobj.gpu_offset); + ib[idx+1] = radeon_get_ib_value(p, idx+1) + ((u32)reloc->lobj.gpu_offset); r = r100_cs_track_check_pkt3_indx_buffer(p, pkt, reloc->robj); if (r) { return r; @@ -1520,27 +1343,27 @@ static int r100_packet3_check(struct radeon_cs_parser *p, r100_cs_dump_packet(p, pkt); return r; } - ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset); + ib[idx] = radeon_get_ib_value(p, idx) + ((u32)reloc->lobj.gpu_offset); track->num_arrays = 1; - track->vtx_size = r100_get_vtx_size(ib_chunk->kdata[idx+2]); + track->vtx_size = r100_get_vtx_size(radeon_get_ib_value(p, idx + 2)); track->arrays[0].robj = reloc->robj; track->arrays[0].esize = track->vtx_size; - track->max_indx = ib_chunk->kdata[idx+1]; + track->max_indx = radeon_get_ib_value(p, idx+1); - track->vap_vf_cntl = ib_chunk->kdata[idx+3]; + track->vap_vf_cntl = radeon_get_ib_value(p, idx+3); track->immd_dwords = pkt->count - 1; r = r100_cs_track_check(p->rdev, track); if (r) return r; break; case PACKET3_3D_DRAW_IMMD: - if (((ib_chunk->kdata[idx+1] >> 4) & 0x3) != 3) { + if (((radeon_get_ib_value(p, idx + 1) >> 4) & 0x3) != 3) { DRM_ERROR("PRIM_WALK must be 3 for IMMD draw\n"); return -EINVAL; } - track->vap_vf_cntl = ib_chunk->kdata[idx+1]; + track->vap_vf_cntl = radeon_get_ib_value(p, idx + 1); track->immd_dwords = pkt->count - 1; r = r100_cs_track_check(p->rdev, track); if (r) @@ -1548,11 +1371,11 @@ static int r100_packet3_check(struct radeon_cs_parser *p, break; /* triggers drawing using in-packet vertex data */ case PACKET3_3D_DRAW_IMMD_2: - if (((ib_chunk->kdata[idx] >> 4) & 0x3) != 3) { + if (((radeon_get_ib_value(p, idx) >> 4) & 0x3) != 3) { DRM_ERROR("PRIM_WALK must be 3 for IMMD draw\n"); return -EINVAL; } - track->vap_vf_cntl = ib_chunk->kdata[idx]; + track->vap_vf_cntl = radeon_get_ib_value(p, idx); track->immd_dwords = pkt->count; r = r100_cs_track_check(p->rdev, track); if (r) @@ -1560,28 +1383,28 @@ static int r100_packet3_check(struct radeon_cs_parser *p, break; /* triggers drawing using in-packet vertex data */ case PACKET3_3D_DRAW_VBUF_2: - track->vap_vf_cntl = ib_chunk->kdata[idx]; + track->vap_vf_cntl = radeon_get_ib_value(p, idx); r = r100_cs_track_check(p->rdev, track); if (r) return r; break; /* triggers drawing of vertex buffers setup elsewhere */ case PACKET3_3D_DRAW_INDX_2: - track->vap_vf_cntl = ib_chunk->kdata[idx]; + track->vap_vf_cntl = radeon_get_ib_value(p, idx); r = r100_cs_track_check(p->rdev, track); if (r) return r; break; /* triggers drawing using indices to vertex buffer */ case PACKET3_3D_DRAW_VBUF: - track->vap_vf_cntl = ib_chunk->kdata[idx + 1]; + track->vap_vf_cntl = radeon_get_ib_value(p, idx + 1); r = r100_cs_track_check(p->rdev, track); if (r) return r; break; /* triggers drawing of vertex buffers setup elsewhere */ case PACKET3_3D_DRAW_INDX: - track->vap_vf_cntl = ib_chunk->kdata[idx + 1]; + track->vap_vf_cntl = radeon_get_ib_value(p, idx + 1); r = r100_cs_track_check(p->rdev, track); if (r) return r; @@ -2033,7 +1856,7 @@ void r100_pll_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v) r100_pll_errata_after_data(rdev); } -int r100_init(struct radeon_device *rdev) +void r100_set_safe_registers(struct radeon_device *rdev) { if (ASIC_IS_RN50(rdev)) { rdev->config.r100.reg_safe_bm = rn50_reg_safe_bm; @@ -2042,9 +1865,8 @@ int r100_init(struct radeon_device *rdev) rdev->config.r100.reg_safe_bm = r100_reg_safe_bm; rdev->config.r100.reg_safe_bm_size = ARRAY_SIZE(r100_reg_safe_bm); } else { - return r200_init(rdev); + r200_set_safe_registers(rdev); } - return 0; } /* @@ -2342,9 +2164,11 @@ void r100_bandwidth_update(struct radeon_device *rdev) mode1 = &rdev->mode_info.crtcs[0]->base.mode; pixel_bytes1 = rdev->mode_info.crtcs[0]->base.fb->bits_per_pixel / 8; } - if (rdev->mode_info.crtcs[1]->base.enabled) { - mode2 = &rdev->mode_info.crtcs[1]->base.mode; - pixel_bytes2 = rdev->mode_info.crtcs[1]->base.fb->bits_per_pixel / 8; + if (!(rdev->flags & RADEON_SINGLE_CRTC)) { + if (rdev->mode_info.crtcs[1]->base.enabled) { + mode2 = &rdev->mode_info.crtcs[1]->base.mode; + pixel_bytes2 = rdev->mode_info.crtcs[1]->base.fb->bits_per_pixel / 8; + } } min_mem_eff.full = rfixed_const_8(0); @@ -2555,7 +2379,7 @@ void r100_bandwidth_update(struct radeon_device *rdev) /* Find the total latency for the display data. */ - disp_latency_overhead.full = rfixed_const(80); + disp_latency_overhead.full = rfixed_const(8); disp_latency_overhead.full = rfixed_div(disp_latency_overhead, sclk_ff); mc_latency_mclk.full += disp_latency_overhead.full + cur_latency_mclk.full; mc_latency_sclk.full += disp_latency_overhead.full + cur_latency_sclk.full; @@ -2753,8 +2577,11 @@ void r100_bandwidth_update(struct radeon_device *rdev) static inline void r100_cs_track_texture_print(struct r100_cs_track_texture *t) { DRM_ERROR("pitch %d\n", t->pitch); + DRM_ERROR("use_pitch %d\n", t->use_pitch); DRM_ERROR("width %d\n", t->width); + DRM_ERROR("width_11 %d\n", t->width_11); DRM_ERROR("height %d\n", t->height); + DRM_ERROR("height_11 %d\n", t->height_11); DRM_ERROR("num levels %d\n", t->num_levels); DRM_ERROR("depth %d\n", t->txdepth); DRM_ERROR("bpp %d\n", t->cpp); @@ -2814,15 +2641,17 @@ static int r100_cs_track_texture_check(struct radeon_device *rdev, else w = track->textures[u].pitch / (1 << i); } else { - w = track->textures[u].width / (1 << i); + w = track->textures[u].width; if (rdev->family >= CHIP_RV515) w |= track->textures[u].width_11; + w = w / (1 << i); if (track->textures[u].roundup_w) w = roundup_pow_of_two(w); } - h = track->textures[u].height / (1 << i); + h = track->textures[u].height; if (rdev->family >= CHIP_RV515) h |= track->textures[u].height_11; + h = h / (1 << i); if (track->textures[u].roundup_h) h = roundup_pow_of_two(h); size += w * h; @@ -3157,7 +2986,7 @@ void r100_mc_stop(struct radeon_device *rdev, struct r100_mc_save *save) WREG32(R_000740_CP_CSQ_CNTL, 0); /* Save few CRTC registers */ - save->GENMO_WT = RREG32(R_0003C0_GENMO_WT); + save->GENMO_WT = RREG8(R_0003C2_GENMO_WT); save->CRTC_EXT_CNTL = RREG32(R_000054_CRTC_EXT_CNTL); save->CRTC_GEN_CNTL = RREG32(R_000050_CRTC_GEN_CNTL); save->CUR_OFFSET = RREG32(R_000260_CUR_OFFSET); @@ -3167,7 +2996,7 @@ void r100_mc_stop(struct radeon_device *rdev, struct r100_mc_save *save) } /* Disable VGA aperture access */ - WREG32(R_0003C0_GENMO_WT, C_0003C0_VGA_RAM_EN & save->GENMO_WT); + WREG8(R_0003C2_GENMO_WT, C_0003C2_VGA_RAM_EN & save->GENMO_WT); /* Disable cursor, overlay, crtc */ WREG32(R_000260_CUR_OFFSET, save->CUR_OFFSET | S_000260_CUR_LOCK(1)); WREG32(R_000054_CRTC_EXT_CNTL, save->CRTC_EXT_CNTL | @@ -3199,10 +3028,264 @@ void r100_mc_resume(struct radeon_device *rdev, struct r100_mc_save *save) rdev->mc.vram_location); } /* Restore CRTC registers */ - WREG32(R_0003C0_GENMO_WT, save->GENMO_WT); + WREG8(R_0003C2_GENMO_WT, save->GENMO_WT); WREG32(R_000054_CRTC_EXT_CNTL, save->CRTC_EXT_CNTL); WREG32(R_000050_CRTC_GEN_CNTL, save->CRTC_GEN_CNTL); if (!(rdev->flags & RADEON_SINGLE_CRTC)) { WREG32(R_0003F8_CRTC2_GEN_CNTL, save->CRTC2_GEN_CNTL); } } + +void r100_vga_render_disable(struct radeon_device *rdev) +{ + u32 tmp; + + tmp = RREG8(R_0003C2_GENMO_WT); + WREG8(R_0003C2_GENMO_WT, C_0003C2_VGA_RAM_EN & tmp); +} + +static void r100_debugfs(struct radeon_device *rdev) +{ + int r; + + r = r100_debugfs_mc_info_init(rdev); + if (r) + dev_warn(rdev->dev, "Failed to create r100_mc debugfs file.\n"); +} + +static void r100_mc_program(struct radeon_device *rdev) +{ + struct r100_mc_save save; + + /* Stops all mc clients */ + r100_mc_stop(rdev, &save); + if (rdev->flags & RADEON_IS_AGP) { + WREG32(R_00014C_MC_AGP_LOCATION, + S_00014C_MC_AGP_START(rdev->mc.gtt_start >> 16) | + S_00014C_MC_AGP_TOP(rdev->mc.gtt_end >> 16)); + WREG32(R_000170_AGP_BASE, lower_32_bits(rdev->mc.agp_base)); + if (rdev->family > CHIP_RV200) + WREG32(R_00015C_AGP_BASE_2, + upper_32_bits(rdev->mc.agp_base) & 0xff); + } else { + WREG32(R_00014C_MC_AGP_LOCATION, 0x0FFFFFFF); + WREG32(R_000170_AGP_BASE, 0); + if (rdev->family > CHIP_RV200) + WREG32(R_00015C_AGP_BASE_2, 0); + } + /* Wait for mc idle */ + if (r100_mc_wait_for_idle(rdev)) + dev_warn(rdev->dev, "Wait for MC idle timeout.\n"); + /* Program MC, should be a 32bits limited address space */ + WREG32(R_000148_MC_FB_LOCATION, + S_000148_MC_FB_START(rdev->mc.vram_start >> 16) | + S_000148_MC_FB_TOP(rdev->mc.vram_end >> 16)); + r100_mc_resume(rdev, &save); +} + +void r100_clock_startup(struct radeon_device *rdev) +{ + u32 tmp; + + if (radeon_dynclks != -1 && radeon_dynclks) + radeon_legacy_set_clock_gating(rdev, 1); + /* We need to force on some of the block */ + tmp = RREG32_PLL(R_00000D_SCLK_CNTL); + tmp |= S_00000D_FORCE_CP(1) | S_00000D_FORCE_VIP(1); + if ((rdev->family == CHIP_RV250) || (rdev->family == CHIP_RV280)) + tmp |= S_00000D_FORCE_DISP1(1) | S_00000D_FORCE_DISP2(1); + WREG32_PLL(R_00000D_SCLK_CNTL, tmp); +} + +static int r100_startup(struct radeon_device *rdev) +{ + int r; + + r100_mc_program(rdev); + /* Resume clock */ + r100_clock_startup(rdev); + /* Initialize GPU configuration (# pipes, ...) */ + r100_gpu_init(rdev); + /* Initialize GART (initialize after TTM so we can allocate + * memory through TTM but finalize after TTM) */ + if (rdev->flags & RADEON_IS_PCI) { + r = r100_pci_gart_enable(rdev); + if (r) + return r; + } + /* Enable IRQ */ + rdev->irq.sw_int = true; + r100_irq_set(rdev); + /* 1M ring buffer */ + r = r100_cp_init(rdev, 1024 * 1024); + if (r) { + dev_err(rdev->dev, "failled initializing CP (%d).\n", r); + return r; + } + r = r100_wb_init(rdev); + if (r) + dev_err(rdev->dev, "failled initializing WB (%d).\n", r); + r = r100_ib_init(rdev); + if (r) { + dev_err(rdev->dev, "failled initializing IB (%d).\n", r); + return r; + } + return 0; +} + +int r100_resume(struct radeon_device *rdev) +{ + /* Make sur GART are not working */ + if (rdev->flags & RADEON_IS_PCI) + r100_pci_gart_disable(rdev); + /* Resume clock before doing reset */ + r100_clock_startup(rdev); + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_gpu_reset(rdev)) { + dev_warn(rdev->dev, "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* post */ + radeon_combios_asic_init(rdev->ddev); + /* Resume clock after posting */ + r100_clock_startup(rdev); + return r100_startup(rdev); +} + +int r100_suspend(struct radeon_device *rdev) +{ + r100_cp_disable(rdev); + r100_wb_disable(rdev); + r100_irq_disable(rdev); + if (rdev->flags & RADEON_IS_PCI) + r100_pci_gart_disable(rdev); + return 0; +} + +void r100_fini(struct radeon_device *rdev) +{ + r100_suspend(rdev); + r100_cp_fini(rdev); + r100_wb_fini(rdev); + r100_ib_fini(rdev); + radeon_gem_fini(rdev); + if (rdev->flags & RADEON_IS_PCI) + r100_pci_gart_fini(rdev); + radeon_irq_kms_fini(rdev); + radeon_fence_driver_fini(rdev); + radeon_object_fini(rdev); + radeon_atombios_fini(rdev); + kfree(rdev->bios); + rdev->bios = NULL; +} + +int r100_mc_init(struct radeon_device *rdev) +{ + int r; + u32 tmp; + + /* Setup GPU memory space */ + rdev->mc.vram_location = 0xFFFFFFFFUL; + rdev->mc.gtt_location = 0xFFFFFFFFUL; + if (rdev->flags & RADEON_IS_IGP) { + tmp = G_00015C_MC_FB_START(RREG32(R_00015C_NB_TOM)); + rdev->mc.vram_location = tmp << 16; + } + if (rdev->flags & RADEON_IS_AGP) { + r = radeon_agp_init(rdev); + if (r) { + printk(KERN_WARNING "[drm] Disabling AGP\n"); + rdev->flags &= ~RADEON_IS_AGP; + rdev->mc.gtt_size = radeon_gart_size * 1024 * 1024; + } else { + rdev->mc.gtt_location = rdev->mc.agp_base; + } + } + r = radeon_mc_setup(rdev); + if (r) + return r; + return 0; +} + +int r100_init(struct radeon_device *rdev) +{ + int r; + + /* Register debugfs file specific to this group of asics */ + r100_debugfs(rdev); + /* Disable VGA */ + r100_vga_render_disable(rdev); + /* Initialize scratch registers */ + radeon_scratch_init(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + /* TODO: disable VGA need to use VGA request */ + /* BIOS*/ + if (!radeon_get_bios(rdev)) { + if (ASIC_IS_AVIVO(rdev)) + return -EINVAL; + } + if (rdev->is_atom_bios) { + dev_err(rdev->dev, "Expecting combios for RS400/RS480 GPU\n"); + return -EINVAL; + } else { + r = radeon_combios_init(rdev); + if (r) + return r; + } + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_gpu_reset(rdev)) { + dev_warn(rdev->dev, + "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* check if cards are posted or not */ + if (!radeon_card_posted(rdev) && rdev->bios) { + DRM_INFO("GPU not posted. posting now...\n"); + radeon_combios_asic_init(rdev->ddev); + } + /* Set asic errata */ + r100_errata(rdev); + /* Initialize clocks */ + radeon_get_clock_info(rdev->ddev); + /* Get vram informations */ + r100_vram_info(rdev); + /* Initialize memory controller (also test AGP) */ + r = r100_mc_init(rdev); + if (r) + return r; + /* Fence driver */ + r = radeon_fence_driver_init(rdev); + if (r) + return r; + r = radeon_irq_kms_init(rdev); + if (r) + return r; + /* Memory manager */ + r = radeon_object_init(rdev); + if (r) + return r; + if (rdev->flags & RADEON_IS_PCI) { + r = r100_pci_gart_init(rdev); + if (r) + return r; + } + r100_set_safe_registers(rdev); + rdev->accel_working = true; + r = r100_startup(rdev); + if (r) { + /* Somethings want wront with the accel init stop accel */ + dev_err(rdev->dev, "Disabling GPU acceleration\n"); + r100_suspend(rdev); + r100_cp_fini(rdev); + r100_wb_fini(rdev); + r100_ib_fini(rdev); + if (rdev->flags & RADEON_IS_PCI) + r100_pci_gart_fini(rdev); + radeon_irq_kms_fini(rdev); + rdev->accel_working = false; + } + return 0; +} diff --git a/drivers/gpu/drm/radeon/r100_track.h b/drivers/gpu/drm/radeon/r100_track.h index 70a82ed..0daf0d7 100644 --- a/drivers/gpu/drm/radeon/r100_track.h +++ b/drivers/gpu/drm/radeon/r100_track.h @@ -84,6 +84,8 @@ int r200_packet0_check(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt, unsigned idx, unsigned reg); + + static inline int r100_reloc_pitch_offset(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt, unsigned idx, @@ -93,9 +95,7 @@ static inline int r100_reloc_pitch_offset(struct radeon_cs_parser *p, u32 tile_flags = 0; u32 tmp; struct radeon_cs_reloc *reloc; - struct radeon_cs_chunk *ib_chunk; - - ib_chunk = &p->chunks[p->chunk_ib_idx]; + u32 value; r = r100_cs_packet_next_reloc(p, &reloc); if (r) { @@ -104,7 +104,8 @@ static inline int r100_reloc_pitch_offset(struct radeon_cs_parser *p, r100_cs_dump_packet(p, pkt); return r; } - tmp = ib_chunk->kdata[idx] & 0x003fffff; + value = radeon_get_ib_value(p, idx); + tmp = value & 0x003fffff; tmp += (((u32)reloc->lobj.gpu_offset) >> 10); if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) @@ -119,6 +120,64 @@ static inline int r100_reloc_pitch_offset(struct radeon_cs_parser *p, } tmp |= tile_flags; - p->ib->ptr[idx] = (ib_chunk->kdata[idx] & 0x3fc00000) | tmp; + p->ib->ptr[idx] = (value & 0x3fc00000) | tmp; return 0; } + +static inline int r100_packet3_load_vbpntr(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt, + int idx) +{ + unsigned c, i; + struct radeon_cs_reloc *reloc; + struct r100_cs_track *track; + int r = 0; + volatile uint32_t *ib; + u32 idx_value; + + ib = p->ib->ptr; + track = (struct r100_cs_track *)p->track; + c = radeon_get_ib_value(p, idx++) & 0x1F; + track->num_arrays = c; + for (i = 0; i < (c - 1); i+=2, idx+=3) { + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for packet3 %d\n", + pkt->opcode); + r100_cs_dump_packet(p, pkt); + return r; + } + idx_value = radeon_get_ib_value(p, idx); + ib[idx+1] = radeon_get_ib_value(p, idx + 1) + ((u32)reloc->lobj.gpu_offset); + + track->arrays[i + 0].esize = idx_value >> 8; + track->arrays[i + 0].robj = reloc->robj; + track->arrays[i + 0].esize &= 0x7F; + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for packet3 %d\n", + pkt->opcode); + r100_cs_dump_packet(p, pkt); + return r; + } + ib[idx+2] = radeon_get_ib_value(p, idx + 2) + ((u32)reloc->lobj.gpu_offset); + track->arrays[i + 1].robj = reloc->robj; + track->arrays[i + 1].esize = idx_value >> 24; + track->arrays[i + 1].esize &= 0x7F; + } + if (c & 1) { + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for packet3 %d\n", + pkt->opcode); + r100_cs_dump_packet(p, pkt); + return r; + } + idx_value = radeon_get_ib_value(p, idx); + ib[idx+1] = radeon_get_ib_value(p, idx + 1) + ((u32)reloc->lobj.gpu_offset); + track->arrays[i + 0].robj = reloc->robj; + track->arrays[i + 0].esize = idx_value >> 8; + track->arrays[i + 0].esize &= 0x7F; + } + return r; +} diff --git a/drivers/gpu/drm/radeon/r100d.h b/drivers/gpu/drm/radeon/r100d.h index c4b257e..df29a63 100644 --- a/drivers/gpu/drm/radeon/r100d.h +++ b/drivers/gpu/drm/radeon/r100d.h @@ -381,6 +381,24 @@ #define S_000054_VCRTC_IDX_MASTER(x) (((x) & 0x7F) << 24) #define G_000054_VCRTC_IDX_MASTER(x) (((x) >> 24) & 0x7F) #define C_000054_VCRTC_IDX_MASTER 0x80FFFFFF +#define R_000148_MC_FB_LOCATION 0x000148 +#define S_000148_MC_FB_START(x) (((x) & 0xFFFF) << 0) +#define G_000148_MC_FB_START(x) (((x) >> 0) & 0xFFFF) +#define C_000148_MC_FB_START 0xFFFF0000 +#define S_000148_MC_FB_TOP(x) (((x) & 0xFFFF) << 16) +#define G_000148_MC_FB_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_000148_MC_FB_TOP 0x0000FFFF +#define R_00014C_MC_AGP_LOCATION 0x00014C +#define S_00014C_MC_AGP_START(x) (((x) & 0xFFFF) << 0) +#define G_00014C_MC_AGP_START(x) (((x) >> 0) & 0xFFFF) +#define C_00014C_MC_AGP_START 0xFFFF0000 +#define S_00014C_MC_AGP_TOP(x) (((x) & 0xFFFF) << 16) +#define G_00014C_MC_AGP_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_00014C_MC_AGP_TOP 0x0000FFFF +#define R_000170_AGP_BASE 0x000170 +#define S_000170_AGP_BASE_ADDR(x) (((x) & 0xFFFFFFFF) << 0) +#define G_000170_AGP_BASE_ADDR(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_000170_AGP_BASE_ADDR 0x00000000 #define R_00023C_DISPLAY_BASE_ADDR 0x00023C #define S_00023C_DISPLAY_BASE_ADDR(x) (((x) & 0xFFFFFFFF) << 0) #define G_00023C_DISPLAY_BASE_ADDR(x) (((x) >> 0) & 0xFFFFFFFF) @@ -403,25 +421,25 @@ #define S_000360_CUR2_LOCK(x) (((x) & 0x1) << 31) #define G_000360_CUR2_LOCK(x) (((x) >> 31) & 0x1) #define C_000360_CUR2_LOCK 0x7FFFFFFF -#define R_0003C0_GENMO_WT 0x0003C0 -#define S_0003C0_GENMO_MONO_ADDRESS_B(x) (((x) & 0x1) << 0) -#define G_0003C0_GENMO_MONO_ADDRESS_B(x) (((x) >> 0) & 0x1) -#define C_0003C0_GENMO_MONO_ADDRESS_B 0xFFFFFFFE -#define S_0003C0_VGA_RAM_EN(x) (((x) & 0x1) << 1) -#define G_0003C0_VGA_RAM_EN(x) (((x) >> 1) & 0x1) -#define C_0003C0_VGA_RAM_EN 0xFFFFFFFD -#define S_0003C0_VGA_CKSEL(x) (((x) & 0x3) << 2) -#define G_0003C0_VGA_CKSEL(x) (((x) >> 2) & 0x3) -#define C_0003C0_VGA_CKSEL 0xFFFFFFF3 -#define S_0003C0_ODD_EVEN_MD_PGSEL(x) (((x) & 0x1) << 5) -#define G_0003C0_ODD_EVEN_MD_PGSEL(x) (((x) >> 5) & 0x1) -#define C_0003C0_ODD_EVEN_MD_PGSEL 0xFFFFFFDF -#define S_0003C0_VGA_HSYNC_POL(x) (((x) & 0x1) << 6) -#define G_0003C0_VGA_HSYNC_POL(x) (((x) >> 6) & 0x1) -#define C_0003C0_VGA_HSYNC_POL 0xFFFFFFBF -#define S_0003C0_VGA_VSYNC_POL(x) (((x) & 0x1) << 7) -#define G_0003C0_VGA_VSYNC_POL(x) (((x) >> 7) & 0x1) -#define C_0003C0_VGA_VSYNC_POL 0xFFFFFF7F +#define R_0003C2_GENMO_WT 0x0003C0 +#define S_0003C2_GENMO_MONO_ADDRESS_B(x) (((x) & 0x1) << 0) +#define G_0003C2_GENMO_MONO_ADDRESS_B(x) (((x) >> 0) & 0x1) +#define C_0003C2_GENMO_MONO_ADDRESS_B 0xFE +#define S_0003C2_VGA_RAM_EN(x) (((x) & 0x1) << 1) +#define G_0003C2_VGA_RAM_EN(x) (((x) >> 1) & 0x1) +#define C_0003C2_VGA_RAM_EN 0xFD +#define S_0003C2_VGA_CKSEL(x) (((x) & 0x3) << 2) +#define G_0003C2_VGA_CKSEL(x) (((x) >> 2) & 0x3) +#define C_0003C2_VGA_CKSEL 0xF3 +#define S_0003C2_ODD_EVEN_MD_PGSEL(x) (((x) & 0x1) << 5) +#define G_0003C2_ODD_EVEN_MD_PGSEL(x) (((x) >> 5) & 0x1) +#define C_0003C2_ODD_EVEN_MD_PGSEL 0xDF +#define S_0003C2_VGA_HSYNC_POL(x) (((x) & 0x1) << 6) +#define G_0003C2_VGA_HSYNC_POL(x) (((x) >> 6) & 0x1) +#define C_0003C2_VGA_HSYNC_POL 0xBF +#define S_0003C2_VGA_VSYNC_POL(x) (((x) & 0x1) << 7) +#define G_0003C2_VGA_VSYNC_POL(x) (((x) >> 7) & 0x1) +#define C_0003C2_VGA_VSYNC_POL 0x7F #define R_0003F8_CRTC2_GEN_CNTL 0x0003F8 #define S_0003F8_CRTC2_DBL_SCAN_EN(x) (((x) & 0x1) << 0) #define G_0003F8_CRTC2_DBL_SCAN_EN(x) (((x) >> 0) & 0x1) @@ -545,6 +563,46 @@ #define S_000774_SCRATCH_ADDR(x) (((x) & 0x7FFFFFF) << 5) #define G_000774_SCRATCH_ADDR(x) (((x) >> 5) & 0x7FFFFFF) #define C_000774_SCRATCH_ADDR 0x0000001F +#define R_0007C0_CP_STAT 0x0007C0 +#define S_0007C0_MRU_BUSY(x) (((x) & 0x1) << 0) +#define G_0007C0_MRU_BUSY(x) (((x) >> 0) & 0x1) +#define C_0007C0_MRU_BUSY 0xFFFFFFFE +#define S_0007C0_MWU_BUSY(x) (((x) & 0x1) << 1) +#define G_0007C0_MWU_BUSY(x) (((x) >> 1) & 0x1) +#define C_0007C0_MWU_BUSY 0xFFFFFFFD +#define S_0007C0_RSIU_BUSY(x) (((x) & 0x1) << 2) +#define G_0007C0_RSIU_BUSY(x) (((x) >> 2) & 0x1) +#define C_0007C0_RSIU_BUSY 0xFFFFFFFB +#define S_0007C0_RCIU_BUSY(x) (((x) & 0x1) << 3) +#define G_0007C0_RCIU_BUSY(x) (((x) >> 3) & 0x1) +#define C_0007C0_RCIU_BUSY 0xFFFFFFF7 +#define S_0007C0_CSF_PRIMARY_BUSY(x) (((x) & 0x1) << 9) +#define G_0007C0_CSF_PRIMARY_BUSY(x) (((x) >> 9) & 0x1) +#define C_0007C0_CSF_PRIMARY_BUSY 0xFFFFFDFF +#define S_0007C0_CSF_INDIRECT_BUSY(x) (((x) & 0x1) << 10) +#define G_0007C0_CSF_INDIRECT_BUSY(x) (((x) >> 10) & 0x1) +#define C_0007C0_CSF_INDIRECT_BUSY 0xFFFFFBFF +#define S_0007C0_CSQ_PRIMARY_BUSY(x) (((x) & 0x1) << 11) +#define G_0007C0_CSQ_PRIMARY_BUSY(x) (((x) >> 11) & 0x1) +#define C_0007C0_CSQ_PRIMARY_BUSY 0xFFFFF7FF +#define S_0007C0_CSQ_INDIRECT_BUSY(x) (((x) & 0x1) << 12) +#define G_0007C0_CSQ_INDIRECT_BUSY(x) (((x) >> 12) & 0x1) +#define C_0007C0_CSQ_INDIRECT_BUSY 0xFFFFEFFF +#define S_0007C0_CSI_BUSY(x) (((x) & 0x1) << 13) +#define G_0007C0_CSI_BUSY(x) (((x) >> 13) & 0x1) +#define C_0007C0_CSI_BUSY 0xFFFFDFFF +#define S_0007C0_GUIDMA_BUSY(x) (((x) & 0x1) << 28) +#define G_0007C0_GUIDMA_BUSY(x) (((x) >> 28) & 0x1) +#define C_0007C0_GUIDMA_BUSY 0xEFFFFFFF +#define S_0007C0_VIDDMA_BUSY(x) (((x) & 0x1) << 29) +#define G_0007C0_VIDDMA_BUSY(x) (((x) >> 29) & 0x1) +#define C_0007C0_VIDDMA_BUSY 0xDFFFFFFF +#define S_0007C0_CMDSTRM_BUSY(x) (((x) & 0x1) << 30) +#define G_0007C0_CMDSTRM_BUSY(x) (((x) >> 30) & 0x1) +#define C_0007C0_CMDSTRM_BUSY 0xBFFFFFFF +#define S_0007C0_CP_BUSY(x) (((x) & 0x1) << 31) +#define G_0007C0_CP_BUSY(x) (((x) >> 31) & 0x1) +#define C_0007C0_CP_BUSY 0x7FFFFFFF #define R_000E40_RBBM_STATUS 0x000E40 #define S_000E40_CMDFIFO_AVAIL(x) (((x) & 0x7F) << 0) #define G_000E40_CMDFIFO_AVAIL(x) (((x) >> 0) & 0x7F) @@ -604,4 +662,53 @@ #define G_000E40_GUI_ACTIVE(x) (((x) >> 31) & 0x1) #define C_000E40_GUI_ACTIVE 0x7FFFFFFF + +#define R_00000D_SCLK_CNTL 0x00000D +#define S_00000D_SCLK_SRC_SEL(x) (((x) & 0x7) << 0) +#define G_00000D_SCLK_SRC_SEL(x) (((x) >> 0) & 0x7) +#define C_00000D_SCLK_SRC_SEL 0xFFFFFFF8 +#define S_00000D_TCLK_SRC_SEL(x) (((x) & 0x7) << 8) +#define G_00000D_TCLK_SRC_SEL(x) (((x) >> 8) & 0x7) +#define C_00000D_TCLK_SRC_SEL 0xFFFFF8FF +#define S_00000D_FORCE_CP(x) (((x) & 0x1) << 16) +#define G_00000D_FORCE_CP(x) (((x) >> 16) & 0x1) +#define C_00000D_FORCE_CP 0xFFFEFFFF +#define S_00000D_FORCE_HDP(x) (((x) & 0x1) << 17) +#define G_00000D_FORCE_HDP(x) (((x) >> 17) & 0x1) +#define C_00000D_FORCE_HDP 0xFFFDFFFF +#define S_00000D_FORCE_DISP(x) (((x) & 0x1) << 18) +#define G_00000D_FORCE_DISP(x) (((x) >> 18) & 0x1) +#define C_00000D_FORCE_DISP 0xFFFBFFFF +#define S_00000D_FORCE_TOP(x) (((x) & 0x1) << 19) +#define G_00000D_FORCE_TOP(x) (((x) >> 19) & 0x1) +#define C_00000D_FORCE_TOP 0xFFF7FFFF +#define S_00000D_FORCE_E2(x) (((x) & 0x1) << 20) +#define G_00000D_FORCE_E2(x) (((x) >> 20) & 0x1) +#define C_00000D_FORCE_E2 0xFFEFFFFF +#define S_00000D_FORCE_SE(x) (((x) & 0x1) << 21) +#define G_00000D_FORCE_SE(x) (((x) >> 21) & 0x1) +#define C_00000D_FORCE_SE 0xFFDFFFFF +#define S_00000D_FORCE_IDCT(x) (((x) & 0x1) << 22) +#define G_00000D_FORCE_IDCT(x) (((x) >> 22) & 0x1) +#define C_00000D_FORCE_IDCT 0xFFBFFFFF +#define S_00000D_FORCE_VIP(x) (((x) & 0x1) << 23) +#define G_00000D_FORCE_VIP(x) (((x) >> 23) & 0x1) +#define C_00000D_FORCE_VIP 0xFF7FFFFF +#define S_00000D_FORCE_RE(x) (((x) & 0x1) << 24) +#define G_00000D_FORCE_RE(x) (((x) >> 24) & 0x1) +#define C_00000D_FORCE_RE 0xFEFFFFFF +#define S_00000D_FORCE_PB(x) (((x) & 0x1) << 25) +#define G_00000D_FORCE_PB(x) (((x) >> 25) & 0x1) +#define C_00000D_FORCE_PB 0xFDFFFFFF +#define S_00000D_FORCE_TAM(x) (((x) & 0x1) << 26) +#define G_00000D_FORCE_TAM(x) (((x) >> 26) & 0x1) +#define C_00000D_FORCE_TAM 0xFBFFFFFF +#define S_00000D_FORCE_TDM(x) (((x) & 0x1) << 27) +#define G_00000D_FORCE_TDM(x) (((x) >> 27) & 0x1) +#define C_00000D_FORCE_TDM 0xF7FFFFFF +#define S_00000D_FORCE_RB(x) (((x) & 0x1) << 28) +#define G_00000D_FORCE_RB(x) (((x) >> 28) & 0x1) +#define C_00000D_FORCE_RB 0xEFFFFFFF + + #endif diff --git a/drivers/gpu/drm/radeon/r200.c b/drivers/gpu/drm/radeon/r200.c index 568c74b..eb740fc 100644 --- a/drivers/gpu/drm/radeon/r200.c +++ b/drivers/gpu/drm/radeon/r200.c @@ -96,7 +96,6 @@ int r200_packet0_check(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt, unsigned idx, unsigned reg) { - struct radeon_cs_chunk *ib_chunk; struct radeon_cs_reloc *reloc; struct r100_cs_track *track; volatile uint32_t *ib; @@ -105,11 +104,11 @@ int r200_packet0_check(struct radeon_cs_parser *p, int i; int face; u32 tile_flags = 0; + u32 idx_value; ib = p->ib->ptr; - ib_chunk = &p->chunks[p->chunk_ib_idx]; track = (struct r100_cs_track *)p->track; - + idx_value = radeon_get_ib_value(p, idx); switch (reg) { case RADEON_CRTC_GUI_TRIG_VLINE: r = r100_cs_packet_parse_vline(p); @@ -137,8 +136,8 @@ int r200_packet0_check(struct radeon_cs_parser *p, return r; } track->zb.robj = reloc->robj; - track->zb.offset = ib_chunk->kdata[idx]; - ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset); + track->zb.offset = idx_value; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); break; case RADEON_RB3D_COLOROFFSET: r = r100_cs_packet_next_reloc(p, &reloc); @@ -149,8 +148,8 @@ int r200_packet0_check(struct radeon_cs_parser *p, return r; } track->cb[0].robj = reloc->robj; - track->cb[0].offset = ib_chunk->kdata[idx]; - ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset); + track->cb[0].offset = idx_value; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); break; case R200_PP_TXOFFSET_0: case R200_PP_TXOFFSET_1: @@ -166,7 +165,7 @@ int r200_packet0_check(struct radeon_cs_parser *p, r100_cs_dump_packet(p, pkt); return r; } - ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset); + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); track->textures[i].robj = reloc->robj; break; case R200_PP_CUBIC_OFFSET_F1_0: @@ -208,12 +207,12 @@ int r200_packet0_check(struct radeon_cs_parser *p, r100_cs_dump_packet(p, pkt); return r; } - track->textures[i].cube_info[face - 1].offset = ib_chunk->kdata[idx]; - ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset); + track->textures[i].cube_info[face - 1].offset = idx_value; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); track->textures[i].cube_info[face - 1].robj = reloc->robj; break; case RADEON_RE_WIDTH_HEIGHT: - track->maxy = ((ib_chunk->kdata[idx] >> 16) & 0x7FF); + track->maxy = ((idx_value >> 16) & 0x7FF); break; case RADEON_RB3D_COLORPITCH: r = r100_cs_packet_next_reloc(p, &reloc); @@ -229,17 +228,17 @@ int r200_packet0_check(struct radeon_cs_parser *p, if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) tile_flags |= RADEON_COLOR_MICROTILE_ENABLE; - tmp = ib_chunk->kdata[idx] & ~(0x7 << 16); + tmp = idx_value & ~(0x7 << 16); tmp |= tile_flags; ib[idx] = tmp; - track->cb[0].pitch = ib_chunk->kdata[idx] & RADEON_COLORPITCH_MASK; + track->cb[0].pitch = idx_value & RADEON_COLORPITCH_MASK; break; case RADEON_RB3D_DEPTHPITCH: - track->zb.pitch = ib_chunk->kdata[idx] & RADEON_DEPTHPITCH_MASK; + track->zb.pitch = idx_value & RADEON_DEPTHPITCH_MASK; break; case RADEON_RB3D_CNTL: - switch ((ib_chunk->kdata[idx] >> RADEON_RB3D_COLOR_FORMAT_SHIFT) & 0x1f) { + switch ((idx_value >> RADEON_RB3D_COLOR_FORMAT_SHIFT) & 0x1f) { case 7: case 8: case 9: @@ -257,18 +256,18 @@ int r200_packet0_check(struct radeon_cs_parser *p, break; default: DRM_ERROR("Invalid color buffer format (%d) !\n", - ((ib_chunk->kdata[idx] >> RADEON_RB3D_COLOR_FORMAT_SHIFT) & 0x1f)); + ((idx_value >> RADEON_RB3D_COLOR_FORMAT_SHIFT) & 0x1f)); return -EINVAL; } - if (ib_chunk->kdata[idx] & RADEON_DEPTHXY_OFFSET_ENABLE) { + if (idx_value & RADEON_DEPTHXY_OFFSET_ENABLE) { DRM_ERROR("No support for depth xy offset in kms\n"); return -EINVAL; } - track->z_enabled = !!(ib_chunk->kdata[idx] & RADEON_Z_ENABLE); + track->z_enabled = !!(idx_value & RADEON_Z_ENABLE); break; case RADEON_RB3D_ZSTENCILCNTL: - switch (ib_chunk->kdata[idx] & 0xf) { + switch (idx_value & 0xf) { case 0: track->zb.cpp = 2; break; @@ -292,27 +291,27 @@ int r200_packet0_check(struct radeon_cs_parser *p, r100_cs_dump_packet(p, pkt); return r; } - ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset); + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); break; case RADEON_PP_CNTL: { - uint32_t temp = ib_chunk->kdata[idx] >> 4; + uint32_t temp = idx_value >> 4; for (i = 0; i < track->num_texture; i++) track->textures[i].enabled = !!(temp & (1 << i)); } break; case RADEON_SE_VF_CNTL: - track->vap_vf_cntl = ib_chunk->kdata[idx]; + track->vap_vf_cntl = idx_value; break; case 0x210c: /* VAP_VF_MAX_VTX_INDX */ - track->max_indx = ib_chunk->kdata[idx] & 0x00FFFFFFUL; + track->max_indx = idx_value & 0x00FFFFFFUL; break; case R200_SE_VTX_FMT_0: - track->vtx_size = r200_get_vtx_size_0(ib_chunk->kdata[idx]); + track->vtx_size = r200_get_vtx_size_0(idx_value); break; case R200_SE_VTX_FMT_1: - track->vtx_size += r200_get_vtx_size_1(ib_chunk->kdata[idx]); + track->vtx_size += r200_get_vtx_size_1(idx_value); break; case R200_PP_TXSIZE_0: case R200_PP_TXSIZE_1: @@ -321,8 +320,8 @@ int r200_packet0_check(struct radeon_cs_parser *p, case R200_PP_TXSIZE_4: case R200_PP_TXSIZE_5: i = (reg - R200_PP_TXSIZE_0) / 32; - track->textures[i].width = (ib_chunk->kdata[idx] & RADEON_TEX_USIZE_MASK) + 1; - track->textures[i].height = ((ib_chunk->kdata[idx] & RADEON_TEX_VSIZE_MASK) >> RADEON_TEX_VSIZE_SHIFT) + 1; + track->textures[i].width = (idx_value & RADEON_TEX_USIZE_MASK) + 1; + track->textures[i].height = ((idx_value & RADEON_TEX_VSIZE_MASK) >> RADEON_TEX_VSIZE_SHIFT) + 1; break; case R200_PP_TXPITCH_0: case R200_PP_TXPITCH_1: @@ -331,7 +330,7 @@ int r200_packet0_check(struct radeon_cs_parser *p, case R200_PP_TXPITCH_4: case R200_PP_TXPITCH_5: i = (reg - R200_PP_TXPITCH_0) / 32; - track->textures[i].pitch = ib_chunk->kdata[idx] + 32; + track->textures[i].pitch = idx_value + 32; break; case R200_PP_TXFILTER_0: case R200_PP_TXFILTER_1: @@ -340,12 +339,12 @@ int r200_packet0_check(struct radeon_cs_parser *p, case R200_PP_TXFILTER_4: case R200_PP_TXFILTER_5: i = (reg - R200_PP_TXFILTER_0) / 32; - track->textures[i].num_levels = ((ib_chunk->kdata[idx] & R200_MAX_MIP_LEVEL_MASK) + track->textures[i].num_levels = ((idx_value & R200_MAX_MIP_LEVEL_MASK) >> R200_MAX_MIP_LEVEL_SHIFT); - tmp = (ib_chunk->kdata[idx] >> 23) & 0x7; + tmp = (idx_value >> 23) & 0x7; if (tmp == 2 || tmp == 6) track->textures[i].roundup_w = false; - tmp = (ib_chunk->kdata[idx] >> 27) & 0x7; + tmp = (idx_value >> 27) & 0x7; if (tmp == 2 || tmp == 6) track->textures[i].roundup_h = false; break; @@ -364,8 +363,8 @@ int r200_packet0_check(struct radeon_cs_parser *p, case R200_PP_TXFORMAT_X_4: case R200_PP_TXFORMAT_X_5: i = (reg - R200_PP_TXFORMAT_X_0) / 32; - track->textures[i].txdepth = ib_chunk->kdata[idx] & 0x7; - tmp = (ib_chunk->kdata[idx] >> 16) & 0x3; + track->textures[i].txdepth = idx_value & 0x7; + tmp = (idx_value >> 16) & 0x3; /* 2D, 3D, CUBE */ switch (tmp) { case 0: @@ -389,14 +388,14 @@ int r200_packet0_check(struct radeon_cs_parser *p, case R200_PP_TXFORMAT_4: case R200_PP_TXFORMAT_5: i = (reg - R200_PP_TXFORMAT_0) / 32; - if (ib_chunk->kdata[idx] & R200_TXFORMAT_NON_POWER2) { + if (idx_value & R200_TXFORMAT_NON_POWER2) { track->textures[i].use_pitch = 1; } else { track->textures[i].use_pitch = 0; - track->textures[i].width = 1 << ((ib_chunk->kdata[idx] >> RADEON_TXFORMAT_WIDTH_SHIFT) & RADEON_TXFORMAT_WIDTH_MASK); - track->textures[i].height = 1 << ((ib_chunk->kdata[idx] >> RADEON_TXFORMAT_HEIGHT_SHIFT) & RADEON_TXFORMAT_HEIGHT_MASK); + track->textures[i].width = 1 << ((idx_value >> RADEON_TXFORMAT_WIDTH_SHIFT) & RADEON_TXFORMAT_WIDTH_MASK); + track->textures[i].height = 1 << ((idx_value >> RADEON_TXFORMAT_HEIGHT_SHIFT) & RADEON_TXFORMAT_HEIGHT_MASK); } - switch ((ib_chunk->kdata[idx] & RADEON_TXFORMAT_FORMAT_MASK)) { + switch ((idx_value & RADEON_TXFORMAT_FORMAT_MASK)) { case R200_TXFORMAT_I8: case R200_TXFORMAT_RGB332: case R200_TXFORMAT_Y8: @@ -424,8 +423,8 @@ int r200_packet0_check(struct radeon_cs_parser *p, track->textures[i].cpp = 4; break; } - track->textures[i].cube_info[4].width = 1 << ((ib_chunk->kdata[idx] >> 16) & 0xf); - track->textures[i].cube_info[4].height = 1 << ((ib_chunk->kdata[idx] >> 20) & 0xf); + track->textures[i].cube_info[4].width = 1 << ((idx_value >> 16) & 0xf); + track->textures[i].cube_info[4].height = 1 << ((idx_value >> 20) & 0xf); break; case R200_PP_CUBIC_FACES_0: case R200_PP_CUBIC_FACES_1: @@ -433,7 +432,7 @@ int r200_packet0_check(struct radeon_cs_parser *p, case R200_PP_CUBIC_FACES_3: case R200_PP_CUBIC_FACES_4: case R200_PP_CUBIC_FACES_5: - tmp = ib_chunk->kdata[idx]; + tmp = idx_value; i = (reg - R200_PP_CUBIC_FACES_0) / 32; for (face = 0; face < 4; face++) { track->textures[i].cube_info[face].width = 1 << ((tmp >> (face * 8)) & 0xf); @@ -448,9 +447,8 @@ int r200_packet0_check(struct radeon_cs_parser *p, return 0; } -int r200_init(struct radeon_device *rdev) +void r200_set_safe_registers(struct radeon_device *rdev) { rdev->config.r100.reg_safe_bm = r200_reg_safe_bm; rdev->config.r100.reg_safe_bm_size = ARRAY_SIZE(r200_reg_safe_bm); - return 0; } diff --git a/drivers/gpu/drm/radeon/r300.c b/drivers/gpu/drm/radeon/r300.c index bb151ec..2f43ee8 100644 --- a/drivers/gpu/drm/radeon/r300.c +++ b/drivers/gpu/drm/radeon/r300.c @@ -33,43 +33,16 @@ #include "radeon_drm.h" #include "r100_track.h" #include "r300d.h" - +#include "rv350d.h" #include "r300_reg_safe.h" -/* r300,r350,rv350,rv370,rv380 depends on : */ -void r100_hdp_reset(struct radeon_device *rdev); -int r100_cp_reset(struct radeon_device *rdev); -int r100_rb2d_reset(struct radeon_device *rdev); -int r100_cp_init(struct radeon_device *rdev, unsigned ring_size); -int r100_pci_gart_enable(struct radeon_device *rdev); -void r100_mc_setup(struct radeon_device *rdev); -void r100_mc_disable_clients(struct radeon_device *rdev); -int r100_gui_wait_for_idle(struct radeon_device *rdev); -int r100_cs_packet_parse(struct radeon_cs_parser *p, - struct radeon_cs_packet *pkt, - unsigned idx); -int r100_cs_packet_parse_vline(struct radeon_cs_parser *p); -int r100_cs_parse_packet0(struct radeon_cs_parser *p, - struct radeon_cs_packet *pkt, - const unsigned *auth, unsigned n, - radeon_packet0_check_t check); -int r100_cs_track_check_pkt3_indx_buffer(struct radeon_cs_parser *p, - struct radeon_cs_packet *pkt, - struct radeon_object *robj); - -/* This files gather functions specifics to: - * r300,r350,rv350,rv370,rv380 - * - * Some of these functions might be used by newer ASICs. - */ -void r300_gpu_init(struct radeon_device *rdev); -int r300_mc_wait_for_idle(struct radeon_device *rdev); -int rv370_debugfs_pcie_gart_info_init(struct radeon_device *rdev); - +/* This files gather functions specifics to: r300,r350,rv350,rv370,rv380 */ /* * rv370,rv380 PCIE GART */ +static int rv370_debugfs_pcie_gart_info_init(struct radeon_device *rdev); + void rv370_pcie_gart_tlb_flush(struct radeon_device *rdev) { uint32_t tmp; @@ -140,7 +113,7 @@ int rv370_pcie_gart_enable(struct radeon_device *rdev) tmp = RADEON_PCIE_TX_GART_UNMAPPED_ACCESS_DISCARD; WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp); WREG32_PCIE(RADEON_PCIE_TX_GART_START_LO, rdev->mc.gtt_location); - tmp = rdev->mc.gtt_location + rdev->mc.gtt_size - 4096; + tmp = rdev->mc.gtt_location + rdev->mc.gtt_size - RADEON_GPU_PAGE_SIZE; WREG32_PCIE(RADEON_PCIE_TX_GART_END_LO, tmp); WREG32_PCIE(RADEON_PCIE_TX_GART_START_HI, 0); WREG32_PCIE(RADEON_PCIE_TX_GART_END_HI, 0); @@ -182,59 +155,6 @@ void rv370_pcie_gart_fini(struct radeon_device *rdev) radeon_gart_fini(rdev); } -/* - * MC - */ -int r300_mc_init(struct radeon_device *rdev) -{ - int r; - - if (r100_debugfs_rbbm_init(rdev)) { - DRM_ERROR("Failed to register debugfs file for RBBM !\n"); - } - - r300_gpu_init(rdev); - r100_pci_gart_disable(rdev); - if (rdev->flags & RADEON_IS_PCIE) { - rv370_pcie_gart_disable(rdev); - } - - /* Setup GPU memory space */ - rdev->mc.vram_location = 0xFFFFFFFFUL; - rdev->mc.gtt_location = 0xFFFFFFFFUL; - if (rdev->flags & RADEON_IS_AGP) { - r = radeon_agp_init(rdev); - if (r) { - printk(KERN_WARNING "[drm] Disabling AGP\n"); - rdev->flags &= ~RADEON_IS_AGP; - rdev->mc.gtt_size = radeon_gart_size * 1024 * 1024; - } else { - rdev->mc.gtt_location = rdev->mc.agp_base; - } - } - r = radeon_mc_setup(rdev); - if (r) { - return r; - } - - /* Program GPU memory space */ - r100_mc_disable_clients(rdev); - if (r300_mc_wait_for_idle(rdev)) { - printk(KERN_WARNING "Failed to wait MC idle while " - "programming pipes. Bad things might happen.\n"); - } - r100_mc_setup(rdev); - return 0; -} - -void r300_mc_fini(struct radeon_device *rdev) -{ -} - - -/* - * Fence emission - */ void r300_fence_ring_emit(struct radeon_device *rdev, struct radeon_fence *fence) { @@ -260,10 +180,6 @@ void r300_fence_ring_emit(struct radeon_device *rdev, radeon_ring_write(rdev, RADEON_SW_INT_FIRE); } - -/* - * Global GPU functions - */ int r300_copy_dma(struct radeon_device *rdev, uint64_t src_offset, uint64_t dst_offset, @@ -582,11 +498,6 @@ void r300_vram_info(struct radeon_device *rdev) r100_vram_init_sizes(rdev); } - -/* - * PCIE Lanes - */ - void rv370_set_pcie_lanes(struct radeon_device *rdev, int lanes) { uint32_t link_width_cntl, mask; @@ -646,10 +557,6 @@ void rv370_set_pcie_lanes(struct radeon_device *rdev, int lanes) } - -/* - * Debugfs info - */ #if defined(CONFIG_DEBUG_FS) static int rv370_debugfs_pcie_gart_info(struct seq_file *m, void *data) { @@ -680,7 +587,7 @@ static struct drm_info_list rv370_pcie_gart_info_list[] = { }; #endif -int rv370_debugfs_pcie_gart_info_init(struct radeon_device *rdev) +static int rv370_debugfs_pcie_gart_info_init(struct radeon_device *rdev) { #if defined(CONFIG_DEBUG_FS) return radeon_debugfs_add_files(rdev, rv370_pcie_gart_info_list, 1); @@ -689,25 +596,22 @@ int rv370_debugfs_pcie_gart_info_init(struct radeon_device *rdev) #endif } - -/* - * CS functions - */ static int r300_packet0_check(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt, unsigned idx, unsigned reg) { - struct radeon_cs_chunk *ib_chunk; struct radeon_cs_reloc *reloc; struct r100_cs_track *track; volatile uint32_t *ib; uint32_t tmp, tile_flags = 0; unsigned i; int r; + u32 idx_value; ib = p->ib->ptr; - ib_chunk = &p->chunks[p->chunk_ib_idx]; track = (struct r100_cs_track *)p->track; + idx_value = radeon_get_ib_value(p, idx); + switch(reg) { case AVIVO_D1MODE_VLINE_START_END: case RADEON_CRTC_GUI_TRIG_VLINE: @@ -738,8 +642,8 @@ static int r300_packet0_check(struct radeon_cs_parser *p, return r; } track->cb[i].robj = reloc->robj; - track->cb[i].offset = ib_chunk->kdata[idx]; - ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset); + track->cb[i].offset = idx_value; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); break; case R300_ZB_DEPTHOFFSET: r = r100_cs_packet_next_reloc(p, &reloc); @@ -750,8 +654,8 @@ static int r300_packet0_check(struct radeon_cs_parser *p, return r; } track->zb.robj = reloc->robj; - track->zb.offset = ib_chunk->kdata[idx]; - ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset); + track->zb.offset = idx_value; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); break; case R300_TX_OFFSET_0: case R300_TX_OFFSET_0+4: @@ -777,32 +681,32 @@ static int r300_packet0_check(struct radeon_cs_parser *p, r100_cs_dump_packet(p, pkt); return r; } - ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset); + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); track->textures[i].robj = reloc->robj; break; /* Tracked registers */ case 0x2084: /* VAP_VF_CNTL */ - track->vap_vf_cntl = ib_chunk->kdata[idx]; + track->vap_vf_cntl = idx_value; break; case 0x20B4: /* VAP_VTX_SIZE */ - track->vtx_size = ib_chunk->kdata[idx] & 0x7F; + track->vtx_size = idx_value & 0x7F; break; case 0x2134: /* VAP_VF_MAX_VTX_INDX */ - track->max_indx = ib_chunk->kdata[idx] & 0x00FFFFFFUL; + track->max_indx = idx_value & 0x00FFFFFFUL; break; case 0x43E4: /* SC_SCISSOR1 */ - track->maxy = ((ib_chunk->kdata[idx] >> 13) & 0x1FFF) + 1; + track->maxy = ((idx_value >> 13) & 0x1FFF) + 1; if (p->rdev->family < CHIP_RV515) { track->maxy -= 1440; } break; case 0x4E00: /* RB3D_CCTL */ - track->num_cb = ((ib_chunk->kdata[idx] >> 5) & 0x3) + 1; + track->num_cb = ((idx_value >> 5) & 0x3) + 1; break; case 0x4E38: case 0x4E3C: @@ -825,13 +729,13 @@ static int r300_packet0_check(struct radeon_cs_parser *p, if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) tile_flags |= R300_COLOR_MICROTILE_ENABLE; - tmp = ib_chunk->kdata[idx] & ~(0x7 << 16); + tmp = idx_value & ~(0x7 << 16); tmp |= tile_flags; ib[idx] = tmp; i = (reg - 0x4E38) >> 2; - track->cb[i].pitch = ib_chunk->kdata[idx] & 0x3FFE; - switch (((ib_chunk->kdata[idx] >> 21) & 0xF)) { + track->cb[i].pitch = idx_value & 0x3FFE; + switch (((idx_value >> 21) & 0xF)) { case 9: case 11: case 12: @@ -854,13 +758,13 @@ static int r300_packet0_check(struct radeon_cs_parser *p, break; default: DRM_ERROR("Invalid color buffer format (%d) !\n", - ((ib_chunk->kdata[idx] >> 21) & 0xF)); + ((idx_value >> 21) & 0xF)); return -EINVAL; } break; case 0x4F00: /* ZB_CNTL */ - if (ib_chunk->kdata[idx] & 2) { + if (idx_value & 2) { track->z_enabled = true; } else { track->z_enabled = false; @@ -868,7 +772,7 @@ static int r300_packet0_check(struct radeon_cs_parser *p, break; case 0x4F10: /* ZB_FORMAT */ - switch ((ib_chunk->kdata[idx] & 0xF)) { + switch ((idx_value & 0xF)) { case 0: case 1: track->zb.cpp = 2; @@ -878,7 +782,7 @@ static int r300_packet0_check(struct radeon_cs_parser *p, break; default: DRM_ERROR("Invalid z buffer format (%d) !\n", - (ib_chunk->kdata[idx] & 0xF)); + (idx_value & 0xF)); return -EINVAL; } break; @@ -897,17 +801,17 @@ static int r300_packet0_check(struct radeon_cs_parser *p, if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) tile_flags |= R300_DEPTHMICROTILE_TILED;; - tmp = ib_chunk->kdata[idx] & ~(0x7 << 16); + tmp = idx_value & ~(0x7 << 16); tmp |= tile_flags; ib[idx] = tmp; - track->zb.pitch = ib_chunk->kdata[idx] & 0x3FFC; + track->zb.pitch = idx_value & 0x3FFC; break; case 0x4104: for (i = 0; i < 16; i++) { bool enabled; - enabled = !!(ib_chunk->kdata[idx] & (1 << i)); + enabled = !!(idx_value & (1 << i)); track->textures[i].enabled = enabled; } break; @@ -929,9 +833,9 @@ static int r300_packet0_check(struct radeon_cs_parser *p, case 0x44FC: /* TX_FORMAT1_[0-15] */ i = (reg - 0x44C0) >> 2; - tmp = (ib_chunk->kdata[idx] >> 25) & 0x3; + tmp = (idx_value >> 25) & 0x3; track->textures[i].tex_coord_type = tmp; - switch ((ib_chunk->kdata[idx] & 0x1F)) { + switch ((idx_value & 0x1F)) { case R300_TX_FORMAT_X8: case R300_TX_FORMAT_Y4X4: case R300_TX_FORMAT_Z3Y3X2: @@ -971,7 +875,7 @@ static int r300_packet0_check(struct radeon_cs_parser *p, break; default: DRM_ERROR("Invalid texture format %u\n", - (ib_chunk->kdata[idx] & 0x1F)); + (idx_value & 0x1F)); return -EINVAL; break; } @@ -994,11 +898,11 @@ static int r300_packet0_check(struct radeon_cs_parser *p, case 0x443C: /* TX_FILTER0_[0-15] */ i = (reg - 0x4400) >> 2; - tmp = ib_chunk->kdata[idx] & 0x7; + tmp = idx_value & 0x7; if (tmp == 2 || tmp == 4 || tmp == 6) { track->textures[i].roundup_w = false; } - tmp = (ib_chunk->kdata[idx] >> 3) & 0x7; + tmp = (idx_value >> 3) & 0x7; if (tmp == 2 || tmp == 4 || tmp == 6) { track->textures[i].roundup_h = false; } @@ -1021,12 +925,12 @@ static int r300_packet0_check(struct radeon_cs_parser *p, case 0x453C: /* TX_FORMAT2_[0-15] */ i = (reg - 0x4500) >> 2; - tmp = ib_chunk->kdata[idx] & 0x3FFF; + tmp = idx_value & 0x3FFF; track->textures[i].pitch = tmp + 1; if (p->rdev->family >= CHIP_RV515) { - tmp = ((ib_chunk->kdata[idx] >> 15) & 1) << 11; + tmp = ((idx_value >> 15) & 1) << 11; track->textures[i].width_11 = tmp; - tmp = ((ib_chunk->kdata[idx] >> 16) & 1) << 11; + tmp = ((idx_value >> 16) & 1) << 11; track->textures[i].height_11 = tmp; } break; @@ -1048,15 +952,15 @@ static int r300_packet0_check(struct radeon_cs_parser *p, case 0x44BC: /* TX_FORMAT0_[0-15] */ i = (reg - 0x4480) >> 2; - tmp = ib_chunk->kdata[idx] & 0x7FF; + tmp = idx_value & 0x7FF; track->textures[i].width = tmp + 1; - tmp = (ib_chunk->kdata[idx] >> 11) & 0x7FF; + tmp = (idx_value >> 11) & 0x7FF; track->textures[i].height = tmp + 1; - tmp = (ib_chunk->kdata[idx] >> 26) & 0xF; + tmp = (idx_value >> 26) & 0xF; track->textures[i].num_levels = tmp; - tmp = ib_chunk->kdata[idx] & (1 << 31); + tmp = idx_value & (1 << 31); track->textures[i].use_pitch = !!tmp; - tmp = (ib_chunk->kdata[idx] >> 22) & 0xF; + tmp = (idx_value >> 22) & 0xF; track->textures[i].txdepth = tmp; break; case R300_ZB_ZPASS_ADDR: @@ -1067,7 +971,7 @@ static int r300_packet0_check(struct radeon_cs_parser *p, r100_cs_dump_packet(p, pkt); return r; } - ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset); + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); break; case 0x4be8: /* valid register only on RV530 */ @@ -1085,60 +989,20 @@ static int r300_packet0_check(struct radeon_cs_parser *p, static int r300_packet3_check(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt) { - struct radeon_cs_chunk *ib_chunk; - struct radeon_cs_reloc *reloc; struct r100_cs_track *track; volatile uint32_t *ib; unsigned idx; - unsigned i, c; int r; ib = p->ib->ptr; - ib_chunk = &p->chunks[p->chunk_ib_idx]; idx = pkt->idx + 1; track = (struct r100_cs_track *)p->track; switch(pkt->opcode) { case PACKET3_3D_LOAD_VBPNTR: - c = ib_chunk->kdata[idx++] & 0x1F; - track->num_arrays = c; - for (i = 0; i < (c - 1); i+=2, idx+=3) { - r = r100_cs_packet_next_reloc(p, &reloc); - if (r) { - DRM_ERROR("No reloc for packet3 %d\n", - pkt->opcode); - r100_cs_dump_packet(p, pkt); - return r; - } - ib[idx+1] = ib_chunk->kdata[idx+1] + ((u32)reloc->lobj.gpu_offset); - track->arrays[i + 0].robj = reloc->robj; - track->arrays[i + 0].esize = ib_chunk->kdata[idx] >> 8; - track->arrays[i + 0].esize &= 0x7F; - r = r100_cs_packet_next_reloc(p, &reloc); - if (r) { - DRM_ERROR("No reloc for packet3 %d\n", - pkt->opcode); - r100_cs_dump_packet(p, pkt); - return r; - } - ib[idx+2] = ib_chunk->kdata[idx+2] + ((u32)reloc->lobj.gpu_offset); - track->arrays[i + 1].robj = reloc->robj; - track->arrays[i + 1].esize = ib_chunk->kdata[idx] >> 24; - track->arrays[i + 1].esize &= 0x7F; - } - if (c & 1) { - r = r100_cs_packet_next_reloc(p, &reloc); - if (r) { - DRM_ERROR("No reloc for packet3 %d\n", - pkt->opcode); - r100_cs_dump_packet(p, pkt); - return r; - } - ib[idx+1] = ib_chunk->kdata[idx+1] + ((u32)reloc->lobj.gpu_offset); - track->arrays[i + 0].robj = reloc->robj; - track->arrays[i + 0].esize = ib_chunk->kdata[idx] >> 8; - track->arrays[i + 0].esize &= 0x7F; - } + r = r100_packet3_load_vbpntr(p, pkt, idx); + if (r) + return r; break; case PACKET3_INDX_BUFFER: r = r100_cs_packet_next_reloc(p, &reloc); @@ -1147,7 +1011,7 @@ static int r300_packet3_check(struct radeon_cs_parser *p, r100_cs_dump_packet(p, pkt); return r; } - ib[idx+1] = ib_chunk->kdata[idx+1] + ((u32)reloc->lobj.gpu_offset); + ib[idx+1] = radeon_get_ib_value(p, idx + 1) + ((u32)reloc->lobj.gpu_offset); r = r100_cs_track_check_pkt3_indx_buffer(p, pkt, reloc->robj); if (r) { return r; @@ -1158,11 +1022,11 @@ static int r300_packet3_check(struct radeon_cs_parser *p, /* Number of dwords is vtx_size * (num_vertices - 1) * PRIM_WALK must be equal to 3 vertex data in embedded * in cmd stream */ - if (((ib_chunk->kdata[idx+1] >> 4) & 0x3) != 3) { + if (((radeon_get_ib_value(p, idx + 1) >> 4) & 0x3) != 3) { DRM_ERROR("PRIM_WALK must be 3 for IMMD draw\n"); return -EINVAL; } - track->vap_vf_cntl = ib_chunk->kdata[idx+1]; + track->vap_vf_cntl = radeon_get_ib_value(p, idx + 1); track->immd_dwords = pkt->count - 1; r = r100_cs_track_check(p->rdev, track); if (r) { @@ -1173,11 +1037,11 @@ static int r300_packet3_check(struct radeon_cs_parser *p, /* Number of dwords is vtx_size * (num_vertices - 1) * PRIM_WALK must be equal to 3 vertex data in embedded * in cmd stream */ - if (((ib_chunk->kdata[idx] >> 4) & 0x3) != 3) { + if (((radeon_get_ib_value(p, idx) >> 4) & 0x3) != 3) { DRM_ERROR("PRIM_WALK must be 3 for IMMD draw\n"); return -EINVAL; } - track->vap_vf_cntl = ib_chunk->kdata[idx]; + track->vap_vf_cntl = radeon_get_ib_value(p, idx); track->immd_dwords = pkt->count; r = r100_cs_track_check(p->rdev, track); if (r) { @@ -1185,28 +1049,28 @@ static int r300_packet3_check(struct radeon_cs_parser *p, } break; case PACKET3_3D_DRAW_VBUF: - track->vap_vf_cntl = ib_chunk->kdata[idx + 1]; + track->vap_vf_cntl = radeon_get_ib_value(p, idx + 1); r = r100_cs_track_check(p->rdev, track); if (r) { return r; } break; case PACKET3_3D_DRAW_VBUF_2: - track->vap_vf_cntl = ib_chunk->kdata[idx]; + track->vap_vf_cntl = radeon_get_ib_value(p, idx); r = r100_cs_track_check(p->rdev, track); if (r) { return r; } break; case PACKET3_3D_DRAW_INDX: - track->vap_vf_cntl = ib_chunk->kdata[idx + 1]; + track->vap_vf_cntl = radeon_get_ib_value(p, idx + 1); r = r100_cs_track_check(p->rdev, track); if (r) { return r; } break; case PACKET3_3D_DRAW_INDX_2: - track->vap_vf_cntl = ib_chunk->kdata[idx]; + track->vap_vf_cntl = radeon_get_ib_value(p, idx); r = r100_cs_track_check(p->rdev, track); if (r) { return r; @@ -1265,12 +1129,6 @@ void r300_set_reg_safe(struct radeon_device *rdev) rdev->config.r300.reg_safe_bm_size = ARRAY_SIZE(r300_reg_safe_bm); } -int r300_init(struct radeon_device *rdev) -{ - r300_set_reg_safe(rdev); - return 0; -} - void r300_mc_program(struct radeon_device *rdev) { struct r100_mc_save save; @@ -1304,3 +1162,198 @@ void r300_mc_program(struct radeon_device *rdev) S_000148_MC_FB_TOP(rdev->mc.vram_end >> 16)); r100_mc_resume(rdev, &save); } + +void r300_clock_startup(struct radeon_device *rdev) +{ + u32 tmp; + + if (radeon_dynclks != -1 && radeon_dynclks) + radeon_legacy_set_clock_gating(rdev, 1); + /* We need to force on some of the block */ + tmp = RREG32_PLL(R_00000D_SCLK_CNTL); + tmp |= S_00000D_FORCE_CP(1) | S_00000D_FORCE_VIP(1); + if ((rdev->family == CHIP_RV350) || (rdev->family == CHIP_RV380)) + tmp |= S_00000D_FORCE_VAP(1); + WREG32_PLL(R_00000D_SCLK_CNTL, tmp); +} + +static int r300_startup(struct radeon_device *rdev) +{ + int r; + + r300_mc_program(rdev); + /* Resume clock */ + r300_clock_startup(rdev); + /* Initialize GPU configuration (# pipes, ...) */ + r300_gpu_init(rdev); + /* Initialize GART (initialize after TTM so we can allocate + * memory through TTM but finalize after TTM) */ + if (rdev->flags & RADEON_IS_PCIE) { + r = rv370_pcie_gart_enable(rdev); + if (r) + return r; + } + if (rdev->flags & RADEON_IS_PCI) { + r = r100_pci_gart_enable(rdev); + if (r) + return r; + } + /* Enable IRQ */ + rdev->irq.sw_int = true; + r100_irq_set(rdev); + /* 1M ring buffer */ + r = r100_cp_init(rdev, 1024 * 1024); + if (r) { + dev_err(rdev->dev, "failled initializing CP (%d).\n", r); + return r; + } + r = r100_wb_init(rdev); + if (r) + dev_err(rdev->dev, "failled initializing WB (%d).\n", r); + r = r100_ib_init(rdev); + if (r) { + dev_err(rdev->dev, "failled initializing IB (%d).\n", r); + return r; + } + return 0; +} + +int r300_resume(struct radeon_device *rdev) +{ + /* Make sur GART are not working */ + if (rdev->flags & RADEON_IS_PCIE) + rv370_pcie_gart_disable(rdev); + if (rdev->flags & RADEON_IS_PCI) + r100_pci_gart_disable(rdev); + /* Resume clock before doing reset */ + r300_clock_startup(rdev); + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_gpu_reset(rdev)) { + dev_warn(rdev->dev, "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* post */ + radeon_combios_asic_init(rdev->ddev); + /* Resume clock after posting */ + r300_clock_startup(rdev); + return r300_startup(rdev); +} + +int r300_suspend(struct radeon_device *rdev) +{ + r100_cp_disable(rdev); + r100_wb_disable(rdev); + r100_irq_disable(rdev); + if (rdev->flags & RADEON_IS_PCIE) + rv370_pcie_gart_disable(rdev); + if (rdev->flags & RADEON_IS_PCI) + r100_pci_gart_disable(rdev); + return 0; +} + +void r300_fini(struct radeon_device *rdev) +{ + r300_suspend(rdev); + r100_cp_fini(rdev); + r100_wb_fini(rdev); + r100_ib_fini(rdev); + radeon_gem_fini(rdev); + if (rdev->flags & RADEON_IS_PCIE) + rv370_pcie_gart_fini(rdev); + if (rdev->flags & RADEON_IS_PCI) + r100_pci_gart_fini(rdev); + radeon_irq_kms_fini(rdev); + radeon_fence_driver_fini(rdev); + radeon_object_fini(rdev); + radeon_atombios_fini(rdev); + kfree(rdev->bios); + rdev->bios = NULL; +} + +int r300_init(struct radeon_device *rdev) +{ + int r; + + /* Disable VGA */ + r100_vga_render_disable(rdev); + /* Initialize scratch registers */ + radeon_scratch_init(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + /* TODO: disable VGA need to use VGA request */ + /* BIOS*/ + if (!radeon_get_bios(rdev)) { + if (ASIC_IS_AVIVO(rdev)) + return -EINVAL; + } + if (rdev->is_atom_bios) { + dev_err(rdev->dev, "Expecting combios for RS400/RS480 GPU\n"); + return -EINVAL; + } else { + r = radeon_combios_init(rdev); + if (r) + return r; + } + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_gpu_reset(rdev)) { + dev_warn(rdev->dev, + "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* check if cards are posted or not */ + if (!radeon_card_posted(rdev) && rdev->bios) { + DRM_INFO("GPU not posted. posting now...\n"); + radeon_combios_asic_init(rdev->ddev); + } + /* Set asic errata */ + r300_errata(rdev); + /* Initialize clocks */ + radeon_get_clock_info(rdev->ddev); + /* Get vram informations */ + r300_vram_info(rdev); + /* Initialize memory controller (also test AGP) */ + r = r420_mc_init(rdev); + if (r) + return r; + /* Fence driver */ + r = radeon_fence_driver_init(rdev); + if (r) + return r; + r = radeon_irq_kms_init(rdev); + if (r) + return r; + /* Memory manager */ + r = radeon_object_init(rdev); + if (r) + return r; + if (rdev->flags & RADEON_IS_PCIE) { + r = rv370_pcie_gart_init(rdev); + if (r) + return r; + } + if (rdev->flags & RADEON_IS_PCI) { + r = r100_pci_gart_init(rdev); + if (r) + return r; + } + r300_set_reg_safe(rdev); + rdev->accel_working = true; + r = r300_startup(rdev); + if (r) { + /* Somethings want wront with the accel init stop accel */ + dev_err(rdev->dev, "Disabling GPU acceleration\n"); + r300_suspend(rdev); + r100_cp_fini(rdev); + r100_wb_fini(rdev); + r100_ib_fini(rdev); + if (rdev->flags & RADEON_IS_PCIE) + rv370_pcie_gart_fini(rdev); + if (rdev->flags & RADEON_IS_PCI) + r100_pci_gart_fini(rdev); + radeon_irq_kms_fini(rdev); + rdev->accel_working = false; + } + return 0; +} diff --git a/drivers/gpu/drm/radeon/r300d.h b/drivers/gpu/drm/radeon/r300d.h index d4fa3eb..4c73114 100644 --- a/drivers/gpu/drm/radeon/r300d.h +++ b/drivers/gpu/drm/radeon/r300d.h @@ -96,6 +96,211 @@ #define S_000170_AGP_BASE_ADDR(x) (((x) & 0xFFFFFFFF) << 0) #define G_000170_AGP_BASE_ADDR(x) (((x) >> 0) & 0xFFFFFFFF) #define C_000170_AGP_BASE_ADDR 0x00000000 +#define R_0007C0_CP_STAT 0x0007C0 +#define S_0007C0_MRU_BUSY(x) (((x) & 0x1) << 0) +#define G_0007C0_MRU_BUSY(x) (((x) >> 0) & 0x1) +#define C_0007C0_MRU_BUSY 0xFFFFFFFE +#define S_0007C0_MWU_BUSY(x) (((x) & 0x1) << 1) +#define G_0007C0_MWU_BUSY(x) (((x) >> 1) & 0x1) +#define C_0007C0_MWU_BUSY 0xFFFFFFFD +#define S_0007C0_RSIU_BUSY(x) (((x) & 0x1) << 2) +#define G_0007C0_RSIU_BUSY(x) (((x) >> 2) & 0x1) +#define C_0007C0_RSIU_BUSY 0xFFFFFFFB +#define S_0007C0_RCIU_BUSY(x) (((x) & 0x1) << 3) +#define G_0007C0_RCIU_BUSY(x) (((x) >> 3) & 0x1) +#define C_0007C0_RCIU_BUSY 0xFFFFFFF7 +#define S_0007C0_CSF_PRIMARY_BUSY(x) (((x) & 0x1) << 9) +#define G_0007C0_CSF_PRIMARY_BUSY(x) (((x) >> 9) & 0x1) +#define C_0007C0_CSF_PRIMARY_BUSY 0xFFFFFDFF +#define S_0007C0_CSF_INDIRECT_BUSY(x) (((x) & 0x1) << 10) +#define G_0007C0_CSF_INDIRECT_BUSY(x) (((x) >> 10) & 0x1) +#define C_0007C0_CSF_INDIRECT_BUSY 0xFFFFFBFF +#define S_0007C0_CSQ_PRIMARY_BUSY(x) (((x) & 0x1) << 11) +#define G_0007C0_CSQ_PRIMARY_BUSY(x) (((x) >> 11) & 0x1) +#define C_0007C0_CSQ_PRIMARY_BUSY 0xFFFFF7FF +#define S_0007C0_CSQ_INDIRECT_BUSY(x) (((x) & 0x1) << 12) +#define G_0007C0_CSQ_INDIRECT_BUSY(x) (((x) >> 12) & 0x1) +#define C_0007C0_CSQ_INDIRECT_BUSY 0xFFFFEFFF +#define S_0007C0_CSI_BUSY(x) (((x) & 0x1) << 13) +#define G_0007C0_CSI_BUSY(x) (((x) >> 13) & 0x1) +#define C_0007C0_CSI_BUSY 0xFFFFDFFF +#define S_0007C0_CSF_INDIRECT2_BUSY(x) (((x) & 0x1) << 14) +#define G_0007C0_CSF_INDIRECT2_BUSY(x) (((x) >> 14) & 0x1) +#define C_0007C0_CSF_INDIRECT2_BUSY 0xFFFFBFFF +#define S_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) & 0x1) << 15) +#define G_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) >> 15) & 0x1) +#define C_0007C0_CSQ_INDIRECT2_BUSY 0xFFFF7FFF +#define S_0007C0_GUIDMA_BUSY(x) (((x) & 0x1) << 28) +#define G_0007C0_GUIDMA_BUSY(x) (((x) >> 28) & 0x1) +#define C_0007C0_GUIDMA_BUSY 0xEFFFFFFF +#define S_0007C0_VIDDMA_BUSY(x) (((x) & 0x1) << 29) +#define G_0007C0_VIDDMA_BUSY(x) (((x) >> 29) & 0x1) +#define C_0007C0_VIDDMA_BUSY 0xDFFFFFFF +#define S_0007C0_CMDSTRM_BUSY(x) (((x) & 0x1) << 30) +#define G_0007C0_CMDSTRM_BUSY(x) (((x) >> 30) & 0x1) +#define C_0007C0_CMDSTRM_BUSY 0xBFFFFFFF +#define S_0007C0_CP_BUSY(x) (((x) & 0x1) << 31) +#define G_0007C0_CP_BUSY(x) (((x) >> 31) & 0x1) +#define C_0007C0_CP_BUSY 0x7FFFFFFF +#define R_000E40_RBBM_STATUS 0x000E40 +#define S_000E40_CMDFIFO_AVAIL(x) (((x) & 0x7F) << 0) +#define G_000E40_CMDFIFO_AVAIL(x) (((x) >> 0) & 0x7F) +#define C_000E40_CMDFIFO_AVAIL 0xFFFFFF80 +#define S_000E40_HIRQ_ON_RBB(x) (((x) & 0x1) << 8) +#define G_000E40_HIRQ_ON_RBB(x) (((x) >> 8) & 0x1) +#define C_000E40_HIRQ_ON_RBB 0xFFFFFEFF +#define S_000E40_CPRQ_ON_RBB(x) (((x) & 0x1) << 9) +#define G_000E40_CPRQ_ON_RBB(x) (((x) >> 9) & 0x1) +#define C_000E40_CPRQ_ON_RBB 0xFFFFFDFF +#define S_000E40_CFRQ_ON_RBB(x) (((x) & 0x1) << 10) +#define G_000E40_CFRQ_ON_RBB(x) (((x) >> 10) & 0x1) +#define C_000E40_CFRQ_ON_RBB 0xFFFFFBFF +#define S_000E40_HIRQ_IN_RTBUF(x) (((x) & 0x1) << 11) +#define G_000E40_HIRQ_IN_RTBUF(x) (((x) >> 11) & 0x1) +#define C_000E40_HIRQ_IN_RTBUF 0xFFFFF7FF +#define S_000E40_CPRQ_IN_RTBUF(x) (((x) & 0x1) << 12) +#define G_000E40_CPRQ_IN_RTBUF(x) (((x) >> 12) & 0x1) +#define C_000E40_CPRQ_IN_RTBUF 0xFFFFEFFF +#define S_000E40_CFRQ_IN_RTBUF(x) (((x) & 0x1) << 13) +#define G_000E40_CFRQ_IN_RTBUF(x) (((x) >> 13) & 0x1) +#define C_000E40_CFRQ_IN_RTBUF 0xFFFFDFFF +#define S_000E40_CF_PIPE_BUSY(x) (((x) & 0x1) << 14) +#define G_000E40_CF_PIPE_BUSY(x) (((x) >> 14) & 0x1) +#define C_000E40_CF_PIPE_BUSY 0xFFFFBFFF +#define S_000E40_ENG_EV_BUSY(x) (((x) & 0x1) << 15) +#define G_000E40_ENG_EV_BUSY(x) (((x) >> 15) & 0x1) +#define C_000E40_ENG_EV_BUSY 0xFFFF7FFF +#define S_000E40_CP_CMDSTRM_BUSY(x) (((x) & 0x1) << 16) +#define G_000E40_CP_CMDSTRM_BUSY(x) (((x) >> 16) & 0x1) +#define C_000E40_CP_CMDSTRM_BUSY 0xFFFEFFFF +#define S_000E40_E2_BUSY(x) (((x) & 0x1) << 17) +#define G_000E40_E2_BUSY(x) (((x) >> 17) & 0x1) +#define C_000E40_E2_BUSY 0xFFFDFFFF +#define S_000E40_RB2D_BUSY(x) (((x) & 0x1) << 18) +#define G_000E40_RB2D_BUSY(x) (((x) >> 18) & 0x1) +#define C_000E40_RB2D_BUSY 0xFFFBFFFF +#define S_000E40_RB3D_BUSY(x) (((x) & 0x1) << 19) +#define G_000E40_RB3D_BUSY(x) (((x) >> 19) & 0x1) +#define C_000E40_RB3D_BUSY 0xFFF7FFFF +#define S_000E40_VAP_BUSY(x) (((x) & 0x1) << 20) +#define G_000E40_VAP_BUSY(x) (((x) >> 20) & 0x1) +#define C_000E40_VAP_BUSY 0xFFEFFFFF +#define S_000E40_RE_BUSY(x) (((x) & 0x1) << 21) +#define G_000E40_RE_BUSY(x) (((x) >> 21) & 0x1) +#define C_000E40_RE_BUSY 0xFFDFFFFF +#define S_000E40_TAM_BUSY(x) (((x) & 0x1) << 22) +#define G_000E40_TAM_BUSY(x) (((x) >> 22) & 0x1) +#define C_000E40_TAM_BUSY 0xFFBFFFFF +#define S_000E40_TDM_BUSY(x) (((x) & 0x1) << 23) +#define G_000E40_TDM_BUSY(x) (((x) >> 23) & 0x1) +#define C_000E40_TDM_BUSY 0xFF7FFFFF +#define S_000E40_PB_BUSY(x) (((x) & 0x1) << 24) +#define G_000E40_PB_BUSY(x) (((x) >> 24) & 0x1) +#define C_000E40_PB_BUSY 0xFEFFFFFF +#define S_000E40_TIM_BUSY(x) (((x) & 0x1) << 25) +#define G_000E40_TIM_BUSY(x) (((x) >> 25) & 0x1) +#define C_000E40_TIM_BUSY 0xFDFFFFFF +#define S_000E40_GA_BUSY(x) (((x) & 0x1) << 26) +#define G_000E40_GA_BUSY(x) (((x) >> 26) & 0x1) +#define C_000E40_GA_BUSY 0xFBFFFFFF +#define S_000E40_CBA2D_BUSY(x) (((x) & 0x1) << 27) +#define G_000E40_CBA2D_BUSY(x) (((x) >> 27) & 0x1) +#define C_000E40_CBA2D_BUSY 0xF7FFFFFF +#define S_000E40_GUI_ACTIVE(x) (((x) & 0x1) << 31) +#define G_000E40_GUI_ACTIVE(x) (((x) >> 31) & 0x1) +#define C_000E40_GUI_ACTIVE 0x7FFFFFFF +#define R_00000D_SCLK_CNTL 0x00000D +#define S_00000D_SCLK_SRC_SEL(x) (((x) & 0x7) << 0) +#define G_00000D_SCLK_SRC_SEL(x) (((x) >> 0) & 0x7) +#define C_00000D_SCLK_SRC_SEL 0xFFFFFFF8 +#define S_00000D_CP_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 3) +#define G_00000D_CP_MAX_DYN_STOP_LAT(x) (((x) >> 3) & 0x1) +#define C_00000D_CP_MAX_DYN_STOP_LAT 0xFFFFFFF7 +#define S_00000D_HDP_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 4) +#define G_00000D_HDP_MAX_DYN_STOP_LAT(x) (((x) >> 4) & 0x1) +#define C_00000D_HDP_MAX_DYN_STOP_LAT 0xFFFFFFEF +#define S_00000D_TV_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 5) +#define G_00000D_TV_MAX_DYN_STOP_LAT(x) (((x) >> 5) & 0x1) +#define C_00000D_TV_MAX_DYN_STOP_LAT 0xFFFFFFDF +#define S_00000D_E2_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 6) +#define G_00000D_E2_MAX_DYN_STOP_LAT(x) (((x) >> 6) & 0x1) +#define C_00000D_E2_MAX_DYN_STOP_LAT 0xFFFFFFBF +#define S_00000D_SE_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 7) +#define G_00000D_SE_MAX_DYN_STOP_LAT(x) (((x) >> 7) & 0x1) +#define C_00000D_SE_MAX_DYN_STOP_LAT 0xFFFFFF7F +#define S_00000D_IDCT_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 8) +#define G_00000D_IDCT_MAX_DYN_STOP_LAT(x) (((x) >> 8) & 0x1) +#define C_00000D_IDCT_MAX_DYN_STOP_LAT 0xFFFFFEFF +#define S_00000D_VIP_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 9) +#define G_00000D_VIP_MAX_DYN_STOP_LAT(x) (((x) >> 9) & 0x1) +#define C_00000D_VIP_MAX_DYN_STOP_LAT 0xFFFFFDFF +#define S_00000D_RE_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 10) +#define G_00000D_RE_MAX_DYN_STOP_LAT(x) (((x) >> 10) & 0x1) +#define C_00000D_RE_MAX_DYN_STOP_LAT 0xFFFFFBFF +#define S_00000D_PB_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 11) +#define G_00000D_PB_MAX_DYN_STOP_LAT(x) (((x) >> 11) & 0x1) +#define C_00000D_PB_MAX_DYN_STOP_LAT 0xFFFFF7FF +#define S_00000D_TAM_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 12) +#define G_00000D_TAM_MAX_DYN_STOP_LAT(x) (((x) >> 12) & 0x1) +#define C_00000D_TAM_MAX_DYN_STOP_LAT 0xFFFFEFFF +#define S_00000D_TDM_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 13) +#define G_00000D_TDM_MAX_DYN_STOP_LAT(x) (((x) >> 13) & 0x1) +#define C_00000D_TDM_MAX_DYN_STOP_LAT 0xFFFFDFFF +#define S_00000D_RB_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 14) +#define G_00000D_RB_MAX_DYN_STOP_LAT(x) (((x) >> 14) & 0x1) +#define C_00000D_RB_MAX_DYN_STOP_LAT 0xFFFFBFFF +#define S_00000D_FORCE_DISP2(x) (((x) & 0x1) << 15) +#define G_00000D_FORCE_DISP2(x) (((x) >> 15) & 0x1) +#define C_00000D_FORCE_DISP2 0xFFFF7FFF +#define S_00000D_FORCE_CP(x) (((x) & 0x1) << 16) +#define G_00000D_FORCE_CP(x) (((x) >> 16) & 0x1) +#define C_00000D_FORCE_CP 0xFFFEFFFF +#define S_00000D_FORCE_HDP(x) (((x) & 0x1) << 17) +#define G_00000D_FORCE_HDP(x) (((x) >> 17) & 0x1) +#define C_00000D_FORCE_HDP 0xFFFDFFFF +#define S_00000D_FORCE_DISP1(x) (((x) & 0x1) << 18) +#define G_00000D_FORCE_DISP1(x) (((x) >> 18) & 0x1) +#define C_00000D_FORCE_DISP1 0xFFFBFFFF +#define S_00000D_FORCE_TOP(x) (((x) & 0x1) << 19) +#define G_00000D_FORCE_TOP(x) (((x) >> 19) & 0x1) +#define C_00000D_FORCE_TOP 0xFFF7FFFF +#define S_00000D_FORCE_E2(x) (((x) & 0x1) << 20) +#define G_00000D_FORCE_E2(x) (((x) >> 20) & 0x1) +#define C_00000D_FORCE_E2 0xFFEFFFFF +#define S_00000D_FORCE_SE(x) (((x) & 0x1) << 21) +#define G_00000D_FORCE_SE(x) (((x) >> 21) & 0x1) +#define C_00000D_FORCE_SE 0xFFDFFFFF +#define S_00000D_FORCE_IDCT(x) (((x) & 0x1) << 22) +#define G_00000D_FORCE_IDCT(x) (((x) >> 22) & 0x1) +#define C_00000D_FORCE_IDCT 0xFFBFFFFF +#define S_00000D_FORCE_VIP(x) (((x) & 0x1) << 23) +#define G_00000D_FORCE_VIP(x) (((x) >> 23) & 0x1) +#define C_00000D_FORCE_VIP 0xFF7FFFFF +#define S_00000D_FORCE_RE(x) (((x) & 0x1) << 24) +#define G_00000D_FORCE_RE(x) (((x) >> 24) & 0x1) +#define C_00000D_FORCE_RE 0xFEFFFFFF +#define S_00000D_FORCE_PB(x) (((x) & 0x1) << 25) +#define G_00000D_FORCE_PB(x) (((x) >> 25) & 0x1) +#define C_00000D_FORCE_PB 0xFDFFFFFF +#define S_00000D_FORCE_TAM(x) (((x) & 0x1) << 26) +#define G_00000D_FORCE_TAM(x) (((x) >> 26) & 0x1) +#define C_00000D_FORCE_TAM 0xFBFFFFFF +#define S_00000D_FORCE_TDM(x) (((x) & 0x1) << 27) +#define G_00000D_FORCE_TDM(x) (((x) >> 27) & 0x1) +#define C_00000D_FORCE_TDM 0xF7FFFFFF +#define S_00000D_FORCE_RB(x) (((x) & 0x1) << 28) +#define G_00000D_FORCE_RB(x) (((x) >> 28) & 0x1) +#define C_00000D_FORCE_RB 0xEFFFFFFF +#define S_00000D_FORCE_TV_SCLK(x) (((x) & 0x1) << 29) +#define G_00000D_FORCE_TV_SCLK(x) (((x) >> 29) & 0x1) +#define C_00000D_FORCE_TV_SCLK 0xDFFFFFFF +#define S_00000D_FORCE_SUBPIC(x) (((x) & 0x1) << 30) +#define G_00000D_FORCE_SUBPIC(x) (((x) >> 30) & 0x1) +#define C_00000D_FORCE_SUBPIC 0xBFFFFFFF +#define S_00000D_FORCE_OV0(x) (((x) & 0x1) << 31) +#define G_00000D_FORCE_OV0(x) (((x) >> 31) & 0x1) +#define C_00000D_FORCE_OV0 0x7FFFFFFF + #endif diff --git a/drivers/gpu/drm/radeon/r420.c b/drivers/gpu/drm/radeon/r420.c index 49a2fdc..1cefdbc 100644 --- a/drivers/gpu/drm/radeon/r420.c +++ b/drivers/gpu/drm/radeon/r420.c @@ -155,6 +155,9 @@ static void r420_debugfs(struct radeon_device *rdev) static void r420_clock_resume(struct radeon_device *rdev) { u32 sclk_cntl; + + if (radeon_dynclks != -1 && radeon_dynclks) + radeon_atom_set_clock_gating(rdev, 1); sclk_cntl = RREG32_PLL(R_00000D_SCLK_CNTL); sclk_cntl |= S_00000D_FORCE_CP(1) | S_00000D_FORCE_VIP(1); if (rdev->family == CHIP_R420) @@ -167,6 +170,8 @@ static int r420_startup(struct radeon_device *rdev) int r; r300_mc_program(rdev); + /* Resume clock */ + r420_clock_resume(rdev); /* Initialize GART (initialize after TTM so we can allocate * memory through TTM but finalize after TTM) */ if (rdev->flags & RADEON_IS_PCIE) { @@ -267,7 +272,6 @@ int r420_init(struct radeon_device *rdev) { int r; - rdev->new_init_path = true; /* Initialize scratch registers */ radeon_scratch_init(rdev); /* Initialize surface registers */ @@ -307,6 +311,8 @@ int r420_init(struct radeon_device *rdev) } /* Initialize clocks */ radeon_get_clock_info(rdev->ddev); + /* Initialize power management */ + radeon_pm_init(rdev); /* Get vram informations */ r300_vram_info(rdev); /* Initialize memory controller (also test AGP) */ diff --git a/drivers/gpu/drm/radeon/r420d.h b/drivers/gpu/drm/radeon/r420d.h index a48a7db..fc78d31 100644 --- a/drivers/gpu/drm/radeon/r420d.h +++ b/drivers/gpu/drm/radeon/r420d.h @@ -212,9 +212,9 @@ #define S_00000D_FORCE_E2(x) (((x) & 0x1) << 20) #define G_00000D_FORCE_E2(x) (((x) >> 20) & 0x1) #define C_00000D_FORCE_E2 0xFFEFFFFF -#define S_00000D_FORCE_SE(x) (((x) & 0x1) << 21) -#define G_00000D_FORCE_SE(x) (((x) >> 21) & 0x1) -#define C_00000D_FORCE_SE 0xFFDFFFFF +#define S_00000D_FORCE_VAP(x) (((x) & 0x1) << 21) +#define G_00000D_FORCE_VAP(x) (((x) >> 21) & 0x1) +#define C_00000D_FORCE_VAP 0xFFDFFFFF #define S_00000D_FORCE_IDCT(x) (((x) & 0x1) << 22) #define G_00000D_FORCE_IDCT(x) (((x) >> 22) & 0x1) #define C_00000D_FORCE_IDCT 0xFFBFFFFF @@ -224,24 +224,24 @@ #define S_00000D_FORCE_RE(x) (((x) & 0x1) << 24) #define G_00000D_FORCE_RE(x) (((x) >> 24) & 0x1) #define C_00000D_FORCE_RE 0xFEFFFFFF -#define S_00000D_FORCE_PB(x) (((x) & 0x1) << 25) -#define G_00000D_FORCE_PB(x) (((x) >> 25) & 0x1) -#define C_00000D_FORCE_PB 0xFDFFFFFF +#define S_00000D_FORCE_SR(x) (((x) & 0x1) << 25) +#define G_00000D_FORCE_SR(x) (((x) >> 25) & 0x1) +#define C_00000D_FORCE_SR 0xFDFFFFFF #define S_00000D_FORCE_PX(x) (((x) & 0x1) << 26) #define G_00000D_FORCE_PX(x) (((x) >> 26) & 0x1) #define C_00000D_FORCE_PX 0xFBFFFFFF #define S_00000D_FORCE_TX(x) (((x) & 0x1) << 27) #define G_00000D_FORCE_TX(x) (((x) >> 27) & 0x1) #define C_00000D_FORCE_TX 0xF7FFFFFF -#define S_00000D_FORCE_RB(x) (((x) & 0x1) << 28) -#define G_00000D_FORCE_RB(x) (((x) >> 28) & 0x1) -#define C_00000D_FORCE_RB 0xEFFFFFFF +#define S_00000D_FORCE_US(x) (((x) & 0x1) << 28) +#define G_00000D_FORCE_US(x) (((x) >> 28) & 0x1) +#define C_00000D_FORCE_US 0xEFFFFFFF #define S_00000D_FORCE_TV_SCLK(x) (((x) & 0x1) << 29) #define G_00000D_FORCE_TV_SCLK(x) (((x) >> 29) & 0x1) #define C_00000D_FORCE_TV_SCLK 0xDFFFFFFF -#define S_00000D_FORCE_SUBPIC(x) (((x) & 0x1) << 30) -#define G_00000D_FORCE_SUBPIC(x) (((x) >> 30) & 0x1) -#define C_00000D_FORCE_SUBPIC 0xBFFFFFFF +#define S_00000D_FORCE_SU(x) (((x) & 0x1) << 30) +#define G_00000D_FORCE_SU(x) (((x) >> 30) & 0x1) +#define C_00000D_FORCE_SU 0xBFFFFFFF #define S_00000D_FORCE_OV0(x) (((x) & 0x1) << 31) #define G_00000D_FORCE_OV0(x) (((x) >> 31) & 0x1) #define C_00000D_FORCE_OV0 0x7FFFFFFF diff --git a/drivers/gpu/drm/radeon/r500_reg.h b/drivers/gpu/drm/radeon/r500_reg.h index e1d5e03..7baa739 100644 --- a/drivers/gpu/drm/radeon/r500_reg.h +++ b/drivers/gpu/drm/radeon/r500_reg.h @@ -384,9 +384,16 @@ # define AVIVO_D1GRPH_TILED (1 << 20) # define AVIVO_D1GRPH_MACRO_ADDRESS_MODE (1 << 21) +/* The R7xx *_HIGH surface regs are backwards; the D1 regs are in the D2 + * block and vice versa. This applies to GRPH, CUR, etc. + */ #define AVIVO_D1GRPH_LUT_SEL 0x6108 #define AVIVO_D1GRPH_PRIMARY_SURFACE_ADDRESS 0x6110 +#define R700_D1GRPH_PRIMARY_SURFACE_ADDRESS_HIGH 0x6914 +#define R700_D2GRPH_PRIMARY_SURFACE_ADDRESS_HIGH 0x6114 #define AVIVO_D1GRPH_SECONDARY_SURFACE_ADDRESS 0x6118 +#define R700_D1GRPH_SECONDARY_SURFACE_ADDRESS_HIGH 0x691c +#define R700_D2GRPH_SECONDARY_SURFACE_ADDRESS_HIGH 0x611c #define AVIVO_D1GRPH_PITCH 0x6120 #define AVIVO_D1GRPH_SURFACE_OFFSET_X 0x6124 #define AVIVO_D1GRPH_SURFACE_OFFSET_Y 0x6128 @@ -404,6 +411,8 @@ # define AVIVO_D1CURSOR_MODE_MASK (3 << 8) # define AVIVO_D1CURSOR_MODE_24BPP 2 #define AVIVO_D1CUR_SURFACE_ADDRESS 0x6408 +#define R700_D1CUR_SURFACE_ADDRESS_HIGH 0x6c0c +#define R700_D2CUR_SURFACE_ADDRESS_HIGH 0x640c #define AVIVO_D1CUR_SIZE 0x6410 #define AVIVO_D1CUR_POSITION 0x6414 #define AVIVO_D1CUR_HOT_SPOT 0x6418 @@ -445,6 +454,8 @@ #define AVIVO_D1MODE_VBLANK_STATUS 0x6534 # define AVIVO_VBLANK_ACK (1 << 4) #define AVIVO_D1MODE_VLINE_START_END 0x6538 +#define AVIVO_D1MODE_VLINE_STATUS 0x653c +# define AVIVO_D1MODE_VLINE_STAT (1 << 12) #define AVIVO_DxMODE_INT_MASK 0x6540 # define AVIVO_D1MODE_INT_MASK (1 << 0) # define AVIVO_D2MODE_INT_MASK (1 << 8) @@ -502,6 +513,7 @@ #define AVIVO_D2MODE_VBLANK_STATUS 0x6d34 #define AVIVO_D2MODE_VLINE_START_END 0x6d38 +#define AVIVO_D2MODE_VLINE_STATUS 0x6d3c #define AVIVO_D2MODE_VIEWPORT_START 0x6d80 #define AVIVO_D2MODE_VIEWPORT_SIZE 0x6d84 #define AVIVO_D2MODE_EXT_OVERSCAN_LEFT_RIGHT 0x6d88 diff --git a/drivers/gpu/drm/radeon/r520.c b/drivers/gpu/drm/radeon/r520.c index d4b0b9d..f743518 100644 --- a/drivers/gpu/drm/radeon/r520.c +++ b/drivers/gpu/drm/radeon/r520.c @@ -26,108 +26,13 @@ * Jerome Glisse */ #include "drmP.h" -#include "radeon_reg.h" #include "radeon.h" +#include "atom.h" +#include "r520d.h" -/* r520,rv530,rv560,rv570,r580 depends on : */ -void r100_hdp_reset(struct radeon_device *rdev); -void r420_pipes_init(struct radeon_device *rdev); -void rs600_mc_disable_clients(struct radeon_device *rdev); -void rs600_disable_vga(struct radeon_device *rdev); -int rv515_debugfs_pipes_info_init(struct radeon_device *rdev); -int rv515_debugfs_ga_info_init(struct radeon_device *rdev); +/* This files gather functions specifics to: r520,rv530,rv560,rv570,r580 */ -/* This files gather functions specifics to: - * r520,rv530,rv560,rv570,r580 - * - * Some of these functions might be used by newer ASICs. - */ -void r520_gpu_init(struct radeon_device *rdev); -int r520_mc_wait_for_idle(struct radeon_device *rdev); - - -/* - * MC - */ -int r520_mc_init(struct radeon_device *rdev) -{ - uint32_t tmp; - int r; - - if (r100_debugfs_rbbm_init(rdev)) { - DRM_ERROR("Failed to register debugfs file for RBBM !\n"); - } - if (rv515_debugfs_pipes_info_init(rdev)) { - DRM_ERROR("Failed to register debugfs file for pipes !\n"); - } - if (rv515_debugfs_ga_info_init(rdev)) { - DRM_ERROR("Failed to register debugfs file for pipes !\n"); - } - - r520_gpu_init(rdev); - rv370_pcie_gart_disable(rdev); - - /* Setup GPU memory space */ - rdev->mc.vram_location = 0xFFFFFFFFUL; - rdev->mc.gtt_location = 0xFFFFFFFFUL; - if (rdev->flags & RADEON_IS_AGP) { - r = radeon_agp_init(rdev); - if (r) { - printk(KERN_WARNING "[drm] Disabling AGP\n"); - rdev->flags &= ~RADEON_IS_AGP; - rdev->mc.gtt_size = radeon_gart_size * 1024 * 1024; - } else { - rdev->mc.gtt_location = rdev->mc.agp_base; - } - } - r = radeon_mc_setup(rdev); - if (r) { - return r; - } - - /* Program GPU memory space */ - rs600_mc_disable_clients(rdev); - if (r520_mc_wait_for_idle(rdev)) { - printk(KERN_WARNING "Failed to wait MC idle while " - "programming pipes. Bad things might happen.\n"); - } - /* Write VRAM size in case we are limiting it */ - WREG32(RADEON_CONFIG_MEMSIZE, rdev->mc.real_vram_size); - tmp = rdev->mc.vram_location + rdev->mc.mc_vram_size - 1; - tmp = REG_SET(R520_MC_FB_TOP, tmp >> 16); - tmp |= REG_SET(R520_MC_FB_START, rdev->mc.vram_location >> 16); - WREG32_MC(R520_MC_FB_LOCATION, tmp); - WREG32(RS690_HDP_FB_LOCATION, rdev->mc.vram_location >> 16); - WREG32(0x310, rdev->mc.vram_location); - if (rdev->flags & RADEON_IS_AGP) { - tmp = rdev->mc.gtt_location + rdev->mc.gtt_size - 1; - tmp = REG_SET(R520_MC_AGP_TOP, tmp >> 16); - tmp |= REG_SET(R520_MC_AGP_START, rdev->mc.gtt_location >> 16); - WREG32_MC(R520_MC_AGP_LOCATION, tmp); - WREG32_MC(R520_MC_AGP_BASE, rdev->mc.agp_base); - WREG32_MC(R520_MC_AGP_BASE_2, 0); - } else { - WREG32_MC(R520_MC_AGP_LOCATION, 0x0FFFFFFF); - WREG32_MC(R520_MC_AGP_BASE, 0); - WREG32_MC(R520_MC_AGP_BASE_2, 0); - } - return 0; -} - -void r520_mc_fini(struct radeon_device *rdev) -{ -} - - -/* - * Global GPU functions - */ -void r520_errata(struct radeon_device *rdev) -{ - rdev->pll_errata = 0; -} - -int r520_mc_wait_for_idle(struct radeon_device *rdev) +static int r520_mc_wait_for_idle(struct radeon_device *rdev) { unsigned i; uint32_t tmp; @@ -143,12 +48,12 @@ int r520_mc_wait_for_idle(struct radeon_device *rdev) return -1; } -void r520_gpu_init(struct radeon_device *rdev) +static void r520_gpu_init(struct radeon_device *rdev) { unsigned pipe_select_current, gb_pipe_select, tmp; r100_hdp_reset(rdev); - rs600_disable_vga(rdev); + rv515_vga_render_disable(rdev); /* * DST_PIPE_CONFIG 0x170C * GB_TILE_CONFIG 0x4018 @@ -186,10 +91,6 @@ void r520_gpu_init(struct radeon_device *rdev) } } - -/* - * VRAM info - */ static void r520_vram_get_type(struct radeon_device *rdev) { uint32_t tmp; @@ -233,7 +134,169 @@ void r520_vram_info(struct radeon_device *rdev) rdev->pm.sclk.full = rfixed_div(rdev->pm.sclk, a); } -void r520_bandwidth_update(struct radeon_device *rdev) +void r520_mc_program(struct radeon_device *rdev) +{ + struct rv515_mc_save save; + + /* Stops all mc clients */ + rv515_mc_stop(rdev, &save); + + /* Wait for mc idle */ + if (r520_mc_wait_for_idle(rdev)) + dev_warn(rdev->dev, "Wait MC idle timeout before updating MC.\n"); + /* Write VRAM size in case we are limiting it */ + WREG32(R_0000F8_CONFIG_MEMSIZE, rdev->mc.real_vram_size); + /* Program MC, should be a 32bits limited address space */ + WREG32_MC(R_000004_MC_FB_LOCATION, + S_000004_MC_FB_START(rdev->mc.vram_start >> 16) | + S_000004_MC_FB_TOP(rdev->mc.vram_end >> 16)); + WREG32(R_000134_HDP_FB_LOCATION, + S_000134_HDP_FB_START(rdev->mc.vram_start >> 16)); + if (rdev->flags & RADEON_IS_AGP) { + WREG32_MC(R_000005_MC_AGP_LOCATION, + S_000005_MC_AGP_START(rdev->mc.gtt_start >> 16) | + S_000005_MC_AGP_TOP(rdev->mc.gtt_end >> 16)); + WREG32_MC(R_000006_AGP_BASE, lower_32_bits(rdev->mc.agp_base)); + WREG32_MC(R_000007_AGP_BASE_2, + S_000007_AGP_BASE_ADDR_2(upper_32_bits(rdev->mc.agp_base))); + } else { + WREG32_MC(R_000005_MC_AGP_LOCATION, 0xFFFFFFFF); + WREG32_MC(R_000006_AGP_BASE, 0); + WREG32_MC(R_000007_AGP_BASE_2, 0); + } + + rv515_mc_resume(rdev, &save); +} + +static int r520_startup(struct radeon_device *rdev) +{ + int r; + + r520_mc_program(rdev); + /* Resume clock */ + rv515_clock_startup(rdev); + /* Initialize GPU configuration (# pipes, ...) */ + r520_gpu_init(rdev); + /* Initialize GART (initialize after TTM so we can allocate + * memory through TTM but finalize after TTM) */ + if (rdev->flags & RADEON_IS_PCIE) { + r = rv370_pcie_gart_enable(rdev); + if (r) + return r; + } + /* Enable IRQ */ + rdev->irq.sw_int = true; + rs600_irq_set(rdev); + /* 1M ring buffer */ + r = r100_cp_init(rdev, 1024 * 1024); + if (r) { + dev_err(rdev->dev, "failled initializing CP (%d).\n", r); + return r; + } + r = r100_wb_init(rdev); + if (r) + dev_err(rdev->dev, "failled initializing WB (%d).\n", r); + r = r100_ib_init(rdev); + if (r) { + dev_err(rdev->dev, "failled initializing IB (%d).\n", r); + return r; + } + return 0; +} + +int r520_resume(struct radeon_device *rdev) { - rv515_bandwidth_avivo_update(rdev); + /* Make sur GART are not working */ + if (rdev->flags & RADEON_IS_PCIE) + rv370_pcie_gart_disable(rdev); + /* Resume clock before doing reset */ + rv515_clock_startup(rdev); + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_gpu_reset(rdev)) { + dev_warn(rdev->dev, "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* post */ + atom_asic_init(rdev->mode_info.atom_context); + /* Resume clock after posting */ + rv515_clock_startup(rdev); + return r520_startup(rdev); +} + +int r520_init(struct radeon_device *rdev) +{ + int r; + + /* Initialize scratch registers */ + radeon_scratch_init(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + /* TODO: disable VGA need to use VGA request */ + /* BIOS*/ + if (!radeon_get_bios(rdev)) { + if (ASIC_IS_AVIVO(rdev)) + return -EINVAL; + } + if (rdev->is_atom_bios) { + r = radeon_atombios_init(rdev); + if (r) + return r; + } else { + dev_err(rdev->dev, "Expecting atombios for RV515 GPU\n"); + return -EINVAL; + } + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_gpu_reset(rdev)) { + dev_warn(rdev->dev, + "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* check if cards are posted or not */ + if (!radeon_card_posted(rdev) && rdev->bios) { + DRM_INFO("GPU not posted. posting now...\n"); + atom_asic_init(rdev->mode_info.atom_context); + } + /* Initialize clocks */ + radeon_get_clock_info(rdev->ddev); + /* Initialize power management */ + radeon_pm_init(rdev); + /* Get vram informations */ + r520_vram_info(rdev); + /* Initialize memory controller (also test AGP) */ + r = r420_mc_init(rdev); + if (r) + return r; + rv515_debugfs(rdev); + /* Fence driver */ + r = radeon_fence_driver_init(rdev); + if (r) + return r; + r = radeon_irq_kms_init(rdev); + if (r) + return r; + /* Memory manager */ + r = radeon_object_init(rdev); + if (r) + return r; + r = rv370_pcie_gart_init(rdev); + if (r) + return r; + rv515_set_safe_registers(rdev); + rdev->accel_working = true; + r = r520_startup(rdev); + if (r) { + /* Somethings want wront with the accel init stop accel */ + dev_err(rdev->dev, "Disabling GPU acceleration\n"); + rv515_suspend(rdev); + r100_cp_fini(rdev); + r100_wb_fini(rdev); + r100_ib_fini(rdev); + rv370_pcie_gart_fini(rdev); + radeon_agp_fini(rdev); + radeon_irq_kms_fini(rdev); + rdev->accel_working = false; + } + return 0; } diff --git a/drivers/gpu/drm/radeon/r520d.h b/drivers/gpu/drm/radeon/r520d.h new file mode 100644 index 0000000..61af61f --- /dev/null +++ b/drivers/gpu/drm/radeon/r520d.h @@ -0,0 +1,187 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef __R520D_H__ +#define __R520D_H__ + +/* Registers */ +#define R_0000F8_CONFIG_MEMSIZE 0x0000F8 +#define S_0000F8_CONFIG_MEMSIZE(x) (((x) & 0xFFFFFFFF) << 0) +#define G_0000F8_CONFIG_MEMSIZE(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_0000F8_CONFIG_MEMSIZE 0x00000000 +#define R_000134_HDP_FB_LOCATION 0x000134 +#define S_000134_HDP_FB_START(x) (((x) & 0xFFFF) << 0) +#define G_000134_HDP_FB_START(x) (((x) >> 0) & 0xFFFF) +#define C_000134_HDP_FB_START 0xFFFF0000 +#define R_0007C0_CP_STAT 0x0007C0 +#define S_0007C0_MRU_BUSY(x) (((x) & 0x1) << 0) +#define G_0007C0_MRU_BUSY(x) (((x) >> 0) & 0x1) +#define C_0007C0_MRU_BUSY 0xFFFFFFFE +#define S_0007C0_MWU_BUSY(x) (((x) & 0x1) << 1) +#define G_0007C0_MWU_BUSY(x) (((x) >> 1) & 0x1) +#define C_0007C0_MWU_BUSY 0xFFFFFFFD +#define S_0007C0_RSIU_BUSY(x) (((x) & 0x1) << 2) +#define G_0007C0_RSIU_BUSY(x) (((x) >> 2) & 0x1) +#define C_0007C0_RSIU_BUSY 0xFFFFFFFB +#define S_0007C0_RCIU_BUSY(x) (((x) & 0x1) << 3) +#define G_0007C0_RCIU_BUSY(x) (((x) >> 3) & 0x1) +#define C_0007C0_RCIU_BUSY 0xFFFFFFF7 +#define S_0007C0_CSF_PRIMARY_BUSY(x) (((x) & 0x1) << 9) +#define G_0007C0_CSF_PRIMARY_BUSY(x) (((x) >> 9) & 0x1) +#define C_0007C0_CSF_PRIMARY_BUSY 0xFFFFFDFF +#define S_0007C0_CSF_INDIRECT_BUSY(x) (((x) & 0x1) << 10) +#define G_0007C0_CSF_INDIRECT_BUSY(x) (((x) >> 10) & 0x1) +#define C_0007C0_CSF_INDIRECT_BUSY 0xFFFFFBFF +#define S_0007C0_CSQ_PRIMARY_BUSY(x) (((x) & 0x1) << 11) +#define G_0007C0_CSQ_PRIMARY_BUSY(x) (((x) >> 11) & 0x1) +#define C_0007C0_CSQ_PRIMARY_BUSY 0xFFFFF7FF +#define S_0007C0_CSQ_INDIRECT_BUSY(x) (((x) & 0x1) << 12) +#define G_0007C0_CSQ_INDIRECT_BUSY(x) (((x) >> 12) & 0x1) +#define C_0007C0_CSQ_INDIRECT_BUSY 0xFFFFEFFF +#define S_0007C0_CSI_BUSY(x) (((x) & 0x1) << 13) +#define G_0007C0_CSI_BUSY(x) (((x) >> 13) & 0x1) +#define C_0007C0_CSI_BUSY 0xFFFFDFFF +#define S_0007C0_CSF_INDIRECT2_BUSY(x) (((x) & 0x1) << 14) +#define G_0007C0_CSF_INDIRECT2_BUSY(x) (((x) >> 14) & 0x1) +#define C_0007C0_CSF_INDIRECT2_BUSY 0xFFFFBFFF +#define S_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) & 0x1) << 15) +#define G_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) >> 15) & 0x1) +#define C_0007C0_CSQ_INDIRECT2_BUSY 0xFFFF7FFF +#define S_0007C0_GUIDMA_BUSY(x) (((x) & 0x1) << 28) +#define G_0007C0_GUIDMA_BUSY(x) (((x) >> 28) & 0x1) +#define C_0007C0_GUIDMA_BUSY 0xEFFFFFFF +#define S_0007C0_VIDDMA_BUSY(x) (((x) & 0x1) << 29) +#define G_0007C0_VIDDMA_BUSY(x) (((x) >> 29) & 0x1) +#define C_0007C0_VIDDMA_BUSY 0xDFFFFFFF +#define S_0007C0_CMDSTRM_BUSY(x) (((x) & 0x1) << 30) +#define G_0007C0_CMDSTRM_BUSY(x) (((x) >> 30) & 0x1) +#define C_0007C0_CMDSTRM_BUSY 0xBFFFFFFF +#define S_0007C0_CP_BUSY(x) (((x) & 0x1) << 31) +#define G_0007C0_CP_BUSY(x) (((x) >> 31) & 0x1) +#define C_0007C0_CP_BUSY 0x7FFFFFFF +#define R_000E40_RBBM_STATUS 0x000E40 +#define S_000E40_CMDFIFO_AVAIL(x) (((x) & 0x7F) << 0) +#define G_000E40_CMDFIFO_AVAIL(x) (((x) >> 0) & 0x7F) +#define C_000E40_CMDFIFO_AVAIL 0xFFFFFF80 +#define S_000E40_HIRQ_ON_RBB(x) (((x) & 0x1) << 8) +#define G_000E40_HIRQ_ON_RBB(x) (((x) >> 8) & 0x1) +#define C_000E40_HIRQ_ON_RBB 0xFFFFFEFF +#define S_000E40_CPRQ_ON_RBB(x) (((x) & 0x1) << 9) +#define G_000E40_CPRQ_ON_RBB(x) (((x) >> 9) & 0x1) +#define C_000E40_CPRQ_ON_RBB 0xFFFFFDFF +#define S_000E40_CFRQ_ON_RBB(x) (((x) & 0x1) << 10) +#define G_000E40_CFRQ_ON_RBB(x) (((x) >> 10) & 0x1) +#define C_000E40_CFRQ_ON_RBB 0xFFFFFBFF +#define S_000E40_HIRQ_IN_RTBUF(x) (((x) & 0x1) << 11) +#define G_000E40_HIRQ_IN_RTBUF(x) (((x) >> 11) & 0x1) +#define C_000E40_HIRQ_IN_RTBUF 0xFFFFF7FF +#define S_000E40_CPRQ_IN_RTBUF(x) (((x) & 0x1) << 12) +#define G_000E40_CPRQ_IN_RTBUF(x) (((x) >> 12) & 0x1) +#define C_000E40_CPRQ_IN_RTBUF 0xFFFFEFFF +#define S_000E40_CFRQ_IN_RTBUF(x) (((x) & 0x1) << 13) +#define G_000E40_CFRQ_IN_RTBUF(x) (((x) >> 13) & 0x1) +#define C_000E40_CFRQ_IN_RTBUF 0xFFFFDFFF +#define S_000E40_CF_PIPE_BUSY(x) (((x) & 0x1) << 14) +#define G_000E40_CF_PIPE_BUSY(x) (((x) >> 14) & 0x1) +#define C_000E40_CF_PIPE_BUSY 0xFFFFBFFF +#define S_000E40_ENG_EV_BUSY(x) (((x) & 0x1) << 15) +#define G_000E40_ENG_EV_BUSY(x) (((x) >> 15) & 0x1) +#define C_000E40_ENG_EV_BUSY 0xFFFF7FFF +#define S_000E40_CP_CMDSTRM_BUSY(x) (((x) & 0x1) << 16) +#define G_000E40_CP_CMDSTRM_BUSY(x) (((x) >> 16) & 0x1) +#define C_000E40_CP_CMDSTRM_BUSY 0xFFFEFFFF +#define S_000E40_E2_BUSY(x) (((x) & 0x1) << 17) +#define G_000E40_E2_BUSY(x) (((x) >> 17) & 0x1) +#define C_000E40_E2_BUSY 0xFFFDFFFF +#define S_000E40_RB2D_BUSY(x) (((x) & 0x1) << 18) +#define G_000E40_RB2D_BUSY(x) (((x) >> 18) & 0x1) +#define C_000E40_RB2D_BUSY 0xFFFBFFFF +#define S_000E40_RB3D_BUSY(x) (((x) & 0x1) << 19) +#define G_000E40_RB3D_BUSY(x) (((x) >> 19) & 0x1) +#define C_000E40_RB3D_BUSY 0xFFF7FFFF +#define S_000E40_VAP_BUSY(x) (((x) & 0x1) << 20) +#define G_000E40_VAP_BUSY(x) (((x) >> 20) & 0x1) +#define C_000E40_VAP_BUSY 0xFFEFFFFF +#define S_000E40_RE_BUSY(x) (((x) & 0x1) << 21) +#define G_000E40_RE_BUSY(x) (((x) >> 21) & 0x1) +#define C_000E40_RE_BUSY 0xFFDFFFFF +#define S_000E40_TAM_BUSY(x) (((x) & 0x1) << 22) +#define G_000E40_TAM_BUSY(x) (((x) >> 22) & 0x1) +#define C_000E40_TAM_BUSY 0xFFBFFFFF +#define S_000E40_TDM_BUSY(x) (((x) & 0x1) << 23) +#define G_000E40_TDM_BUSY(x) (((x) >> 23) & 0x1) +#define C_000E40_TDM_BUSY 0xFF7FFFFF +#define S_000E40_PB_BUSY(x) (((x) & 0x1) << 24) +#define G_000E40_PB_BUSY(x) (((x) >> 24) & 0x1) +#define C_000E40_PB_BUSY 0xFEFFFFFF +#define S_000E40_TIM_BUSY(x) (((x) & 0x1) << 25) +#define G_000E40_TIM_BUSY(x) (((x) >> 25) & 0x1) +#define C_000E40_TIM_BUSY 0xFDFFFFFF +#define S_000E40_GA_BUSY(x) (((x) & 0x1) << 26) +#define G_000E40_GA_BUSY(x) (((x) >> 26) & 0x1) +#define C_000E40_GA_BUSY 0xFBFFFFFF +#define S_000E40_CBA2D_BUSY(x) (((x) & 0x1) << 27) +#define G_000E40_CBA2D_BUSY(x) (((x) >> 27) & 0x1) +#define C_000E40_CBA2D_BUSY 0xF7FFFFFF +#define S_000E40_RBBM_HIBUSY(x) (((x) & 0x1) << 28) +#define G_000E40_RBBM_HIBUSY(x) (((x) >> 28) & 0x1) +#define C_000E40_RBBM_HIBUSY 0xEFFFFFFF +#define S_000E40_SKID_CFBUSY(x) (((x) & 0x1) << 29) +#define G_000E40_SKID_CFBUSY(x) (((x) >> 29) & 0x1) +#define C_000E40_SKID_CFBUSY 0xDFFFFFFF +#define S_000E40_VAP_VF_BUSY(x) (((x) & 0x1) << 30) +#define G_000E40_VAP_VF_BUSY(x) (((x) >> 30) & 0x1) +#define C_000E40_VAP_VF_BUSY 0xBFFFFFFF +#define S_000E40_GUI_ACTIVE(x) (((x) & 0x1) << 31) +#define G_000E40_GUI_ACTIVE(x) (((x) >> 31) & 0x1) +#define C_000E40_GUI_ACTIVE 0x7FFFFFFF + + +#define R_000004_MC_FB_LOCATION 0x000004 +#define S_000004_MC_FB_START(x) (((x) & 0xFFFF) << 0) +#define G_000004_MC_FB_START(x) (((x) >> 0) & 0xFFFF) +#define C_000004_MC_FB_START 0xFFFF0000 +#define S_000004_MC_FB_TOP(x) (((x) & 0xFFFF) << 16) +#define G_000004_MC_FB_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_000004_MC_FB_TOP 0x0000FFFF +#define R_000005_MC_AGP_LOCATION 0x000005 +#define S_000005_MC_AGP_START(x) (((x) & 0xFFFF) << 0) +#define G_000005_MC_AGP_START(x) (((x) >> 0) & 0xFFFF) +#define C_000005_MC_AGP_START 0xFFFF0000 +#define S_000005_MC_AGP_TOP(x) (((x) & 0xFFFF) << 16) +#define G_000005_MC_AGP_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_000005_MC_AGP_TOP 0x0000FFFF +#define R_000006_AGP_BASE 0x000006 +#define S_000006_AGP_BASE_ADDR(x) (((x) & 0xFFFFFFFF) << 0) +#define G_000006_AGP_BASE_ADDR(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_000006_AGP_BASE_ADDR 0x00000000 +#define R_000007_AGP_BASE_2 0x000007 +#define S_000007_AGP_BASE_ADDR_2(x) (((x) & 0xF) << 0) +#define G_000007_AGP_BASE_ADDR_2(x) (((x) >> 0) & 0xF) +#define C_000007_AGP_BASE_ADDR_2 0xFFFFFFF0 + +#endif diff --git a/drivers/gpu/drm/radeon/r600.c b/drivers/gpu/drm/radeon/r600.c index eab31c1..278f646 100644 --- a/drivers/gpu/drm/radeon/r600.c +++ b/drivers/gpu/drm/radeon/r600.c @@ -33,8 +33,8 @@ #include "radeon.h" #include "radeon_mode.h" #include "r600d.h" -#include "avivod.h" #include "atom.h" +#include "avivod.h" #define PFP_UCODE_SIZE 576 #define PM4_UCODE_SIZE 1792 @@ -65,16 +65,11 @@ MODULE_FIRMWARE("radeon/RV710_me.bin"); int r600_debugfs_mc_info_init(struct radeon_device *rdev); -/* This files gather functions specifics to: - * r600,rv610,rv630,rv620,rv635,rv670 - * - * Some of these functions might be used by newer ASICs. - */ +/* r600,rv610,rv630,rv620,rv635,rv670 */ int r600_mc_wait_for_idle(struct radeon_device *rdev); void r600_gpu_init(struct radeon_device *rdev); void r600_fini(struct radeon_device *rdev); - /* * R600 PCIE GART */ @@ -168,7 +163,7 @@ int r600_pcie_gart_enable(struct radeon_device *rdev) WREG32(MC_VM_L1_TLB_MCB_RD_SEM_CNTL, tmp | ENABLE_SEMAPHORE_MODE); WREG32(MC_VM_L1_TLB_MCB_WR_SEM_CNTL, tmp | ENABLE_SEMAPHORE_MODE); WREG32(VM_CONTEXT0_PAGE_TABLE_START_ADDR, rdev->mc.gtt_start >> 12); - WREG32(VM_CONTEXT0_PAGE_TABLE_END_ADDR, (rdev->mc.gtt_end - 1) >> 12); + WREG32(VM_CONTEXT0_PAGE_TABLE_END_ADDR, rdev->mc.gtt_end >> 12); WREG32(VM_CONTEXT0_PAGE_TABLE_BASE_ADDR, rdev->gart.table_addr >> 12); WREG32(VM_CONTEXT0_CNTL, ENABLE_CONTEXT | PAGE_TABLE_DEPTH(0) | RANGE_PROTECTION_FAULT_ENABLE_DEFAULT); @@ -225,6 +220,40 @@ void r600_pcie_gart_fini(struct radeon_device *rdev) radeon_gart_fini(rdev); } +void r600_agp_enable(struct radeon_device *rdev) +{ + u32 tmp; + int i; + + /* Setup L2 cache */ + WREG32(VM_L2_CNTL, ENABLE_L2_CACHE | ENABLE_L2_FRAGMENT_PROCESSING | + ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE | + EFFECTIVE_L2_QUEUE_SIZE(7)); + WREG32(VM_L2_CNTL2, 0); + WREG32(VM_L2_CNTL3, BANK_SELECT_0(0) | BANK_SELECT_1(1)); + /* Setup TLB control */ + tmp = ENABLE_L1_TLB | ENABLE_L1_FRAGMENT_PROCESSING | + SYSTEM_ACCESS_MODE_NOT_IN_SYS | + EFFECTIVE_L1_TLB_SIZE(5) | EFFECTIVE_L1_QUEUE_SIZE(5) | + ENABLE_WAIT_L2_QUERY; + WREG32(MC_VM_L1_TLB_MCB_RD_SYS_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_WR_SYS_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_RD_HDP_CNTL, tmp | ENABLE_L1_STRICT_ORDERING); + WREG32(MC_VM_L1_TLB_MCB_WR_HDP_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCD_RD_A_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCD_WR_A_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCD_RD_B_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCD_WR_B_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_RD_GFX_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_WR_GFX_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_RD_PDMA_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_WR_PDMA_CNTL, tmp); + WREG32(MC_VM_L1_TLB_MCB_RD_SEM_CNTL, tmp | ENABLE_SEMAPHORE_MODE); + WREG32(MC_VM_L1_TLB_MCB_WR_SEM_CNTL, tmp | ENABLE_SEMAPHORE_MODE); + for (i = 0; i < 7; i++) + WREG32(VM_CONTEXT0_CNTL + (i * 4), 0); +} + int r600_mc_wait_for_idle(struct radeon_device *rdev) { unsigned i; @@ -240,14 +269,9 @@ int r600_mc_wait_for_idle(struct radeon_device *rdev) return -1; } -static void r600_mc_resume(struct radeon_device *rdev) +static void r600_mc_program(struct radeon_device *rdev) { - u32 d1vga_control, d2vga_control; - u32 vga_render_control, vga_hdp_control; - u32 d1crtc_control, d2crtc_control; - u32 new_d1grph_primary, new_d1grph_secondary; - u32 new_d2grph_primary, new_d2grph_secondary; - u64 old_vram_start; + struct rv515_mc_save save; u32 tmp; int i, j; @@ -261,99 +285,64 @@ static void r600_mc_resume(struct radeon_device *rdev) } WREG32(HDP_REG_COHERENCY_FLUSH_CNTL, 0); - d1vga_control = RREG32(D1VGA_CONTROL); - d2vga_control = RREG32(D2VGA_CONTROL); - vga_render_control = RREG32(VGA_RENDER_CONTROL); - vga_hdp_control = RREG32(VGA_HDP_CONTROL); - d1crtc_control = RREG32(D1CRTC_CONTROL); - d2crtc_control = RREG32(D2CRTC_CONTROL); - old_vram_start = (u64)(RREG32(MC_VM_FB_LOCATION) & 0xFFFF) << 24; - new_d1grph_primary = RREG32(D1GRPH_PRIMARY_SURFACE_ADDRESS); - new_d1grph_secondary = RREG32(D1GRPH_SECONDARY_SURFACE_ADDRESS); - new_d1grph_primary += rdev->mc.vram_start - old_vram_start; - new_d1grph_secondary += rdev->mc.vram_start - old_vram_start; - new_d2grph_primary = RREG32(D2GRPH_PRIMARY_SURFACE_ADDRESS); - new_d2grph_secondary = RREG32(D2GRPH_SECONDARY_SURFACE_ADDRESS); - new_d2grph_primary += rdev->mc.vram_start - old_vram_start; - new_d2grph_secondary += rdev->mc.vram_start - old_vram_start; - - /* Stop all video */ - WREG32(D1VGA_CONTROL, 0); - WREG32(D2VGA_CONTROL, 0); - WREG32(VGA_RENDER_CONTROL, 0); - WREG32(D1CRTC_UPDATE_LOCK, 1); - WREG32(D2CRTC_UPDATE_LOCK, 1); - WREG32(D1CRTC_CONTROL, 0); - WREG32(D2CRTC_CONTROL, 0); - WREG32(D1CRTC_UPDATE_LOCK, 0); - WREG32(D2CRTC_UPDATE_LOCK, 0); - - mdelay(1); + rv515_mc_stop(rdev, &save); if (r600_mc_wait_for_idle(rdev)) { - printk(KERN_WARNING "[drm] MC not idle !\n"); + dev_warn(rdev->dev, "Wait for MC idle timedout !\n"); } - - /* Lockout access through VGA aperture*/ + /* Lockout access through VGA aperture (doesn't exist before R600) */ WREG32(VGA_HDP_CONTROL, VGA_MEMORY_DISABLE); - /* Update configuration */ - WREG32(MC_VM_SYSTEM_APERTURE_LOW_ADDR, rdev->mc.vram_start >> 12); - WREG32(MC_VM_SYSTEM_APERTURE_HIGH_ADDR, (rdev->mc.vram_end - 1) >> 12); + if (rdev->flags & RADEON_IS_AGP) { + if (rdev->mc.vram_start < rdev->mc.gtt_start) { + /* VRAM before AGP */ + WREG32(MC_VM_SYSTEM_APERTURE_LOW_ADDR, + rdev->mc.vram_start >> 12); + WREG32(MC_VM_SYSTEM_APERTURE_HIGH_ADDR, + rdev->mc.gtt_end >> 12); + } else { + /* VRAM after AGP */ + WREG32(MC_VM_SYSTEM_APERTURE_LOW_ADDR, + rdev->mc.gtt_start >> 12); + WREG32(MC_VM_SYSTEM_APERTURE_HIGH_ADDR, + rdev->mc.vram_end >> 12); + } + } else { + WREG32(MC_VM_SYSTEM_APERTURE_LOW_ADDR, rdev->mc.vram_start >> 12); + WREG32(MC_VM_SYSTEM_APERTURE_HIGH_ADDR, rdev->mc.vram_end >> 12); + } WREG32(MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR, 0); - tmp = (((rdev->mc.vram_end - 1) >> 24) & 0xFFFF) << 16; + tmp = ((rdev->mc.vram_end >> 24) & 0xFFFF) << 16; tmp |= ((rdev->mc.vram_start >> 24) & 0xFFFF); WREG32(MC_VM_FB_LOCATION, tmp); WREG32(HDP_NONSURFACE_BASE, (rdev->mc.vram_start >> 8)); WREG32(HDP_NONSURFACE_INFO, (2 << 7)); - WREG32(HDP_NONSURFACE_SIZE, (rdev->mc.mc_vram_size - 1) | 0x3FF); + WREG32(HDP_NONSURFACE_SIZE, rdev->mc.mc_vram_size | 0x3FF); if (rdev->flags & RADEON_IS_AGP) { - WREG32(MC_VM_AGP_TOP, (rdev->mc.gtt_end - 1) >> 16); - WREG32(MC_VM_AGP_BOT, rdev->mc.gtt_start >> 16); + WREG32(MC_VM_AGP_TOP, rdev->mc.gtt_end >> 22); + WREG32(MC_VM_AGP_BOT, rdev->mc.gtt_start >> 22); WREG32(MC_VM_AGP_BASE, rdev->mc.agp_base >> 22); } else { WREG32(MC_VM_AGP_BASE, 0); WREG32(MC_VM_AGP_TOP, 0x0FFFFFFF); WREG32(MC_VM_AGP_BOT, 0x0FFFFFFF); } - WREG32(D1GRPH_PRIMARY_SURFACE_ADDRESS, new_d1grph_primary); - WREG32(D1GRPH_SECONDARY_SURFACE_ADDRESS, new_d1grph_secondary); - WREG32(D2GRPH_PRIMARY_SURFACE_ADDRESS, new_d2grph_primary); - WREG32(D2GRPH_SECONDARY_SURFACE_ADDRESS, new_d2grph_secondary); - WREG32(VGA_MEMORY_BASE_ADDRESS, rdev->mc.vram_start); - - /* Unlock host access */ - WREG32(VGA_HDP_CONTROL, vga_hdp_control); - - mdelay(1); if (r600_mc_wait_for_idle(rdev)) { - printk(KERN_WARNING "[drm] MC not idle !\n"); + dev_warn(rdev->dev, "Wait for MC idle timedout !\n"); } - - /* Restore video state */ - WREG32(D1CRTC_UPDATE_LOCK, 1); - WREG32(D2CRTC_UPDATE_LOCK, 1); - WREG32(D1CRTC_CONTROL, d1crtc_control); - WREG32(D2CRTC_CONTROL, d2crtc_control); - WREG32(D1CRTC_UPDATE_LOCK, 0); - WREG32(D2CRTC_UPDATE_LOCK, 0); - WREG32(D1VGA_CONTROL, d1vga_control); - WREG32(D2VGA_CONTROL, d2vga_control); - WREG32(VGA_RENDER_CONTROL, vga_render_control); - + rv515_mc_resume(rdev, &save); /* we need to own VRAM, so turn off the VGA renderer here * to stop it overwriting our objects */ - radeon_avivo_vga_render_disable(rdev); + rv515_vga_render_disable(rdev); } int r600_mc_init(struct radeon_device *rdev) { fixed20_12 a; u32 tmp; - int chansize; + int chansize, numchan; int r; /* Get VRAM informations */ - rdev->mc.vram_width = 128; rdev->mc.vram_is_ddr = true; tmp = RREG32(RAMCFG); if (tmp & CHANSIZE_OVERRIDE) { @@ -363,23 +352,36 @@ int r600_mc_init(struct radeon_device *rdev) } else { chansize = 32; } - if (rdev->family == CHIP_R600) { - rdev->mc.vram_width = 8 * chansize; - } else if (rdev->family == CHIP_RV670) { - rdev->mc.vram_width = 4 * chansize; - } else if ((rdev->family == CHIP_RV610) || - (rdev->family == CHIP_RV620)) { - rdev->mc.vram_width = chansize; - } else if ((rdev->family == CHIP_RV630) || - (rdev->family == CHIP_RV635)) { - rdev->mc.vram_width = 2 * chansize; + tmp = RREG32(CHMAP); + switch ((tmp & NOOFCHAN_MASK) >> NOOFCHAN_SHIFT) { + case 0: + default: + numchan = 1; + break; + case 1: + numchan = 2; + break; + case 2: + numchan = 4; + break; + case 3: + numchan = 8; + break; } + rdev->mc.vram_width = numchan * chansize; /* Could aper size report 0 ? */ rdev->mc.aper_base = drm_get_resource_start(rdev->ddev, 0); rdev->mc.aper_size = drm_get_resource_len(rdev->ddev, 0); /* Setup GPU memory space */ rdev->mc.mc_vram_size = RREG32(CONFIG_MEMSIZE); rdev->mc.real_vram_size = RREG32(CONFIG_MEMSIZE); + + if (rdev->mc.mc_vram_size > rdev->mc.aper_size) + rdev->mc.mc_vram_size = rdev->mc.aper_size; + + if (rdev->mc.real_vram_size > rdev->mc.aper_size) + rdev->mc.real_vram_size = rdev->mc.aper_size; + if (rdev->flags & RADEON_IS_AGP) { r = radeon_agp_init(rdev); if (r) @@ -407,40 +409,34 @@ int r600_mc_init(struct radeon_device *rdev) rdev->mc.gtt_location = rdev->mc.mc_vram_size; } } else { - if (rdev->family == CHIP_RS780 || rdev->family == CHIP_RS880) { - rdev->mc.vram_location = (RREG32(MC_VM_FB_LOCATION) & - 0xFFFF) << 24; - rdev->mc.gtt_size = radeon_gart_size * 1024 * 1024; - tmp = rdev->mc.vram_location + rdev->mc.mc_vram_size; - if ((0xFFFFFFFFUL - tmp) >= rdev->mc.gtt_size) { - /* Enough place after vram */ - rdev->mc.gtt_location = tmp; - } else if (rdev->mc.vram_location >= rdev->mc.gtt_size) { - /* Enough place before vram */ + rdev->mc.gtt_size = radeon_gart_size * 1024 * 1024; + rdev->mc.vram_location = (RREG32(MC_VM_FB_LOCATION) & + 0xFFFF) << 24; + tmp = rdev->mc.vram_location + rdev->mc.mc_vram_size; + if ((0xFFFFFFFFUL - tmp) >= rdev->mc.gtt_size) { + /* Enough place after vram */ + rdev->mc.gtt_location = tmp; + } else if (rdev->mc.vram_location >= rdev->mc.gtt_size) { + /* Enough place before vram */ + rdev->mc.gtt_location = 0; + } else { + /* Not enough place after or before shrink + * gart size + */ + if (rdev->mc.vram_location > (0xFFFFFFFFUL - tmp)) { rdev->mc.gtt_location = 0; + rdev->mc.gtt_size = rdev->mc.vram_location; } else { - /* Not enough place after or before shrink - * gart size - */ - if (rdev->mc.vram_location > (0xFFFFFFFFUL - tmp)) { - rdev->mc.gtt_location = 0; - rdev->mc.gtt_size = rdev->mc.vram_location; - } else { - rdev->mc.gtt_location = tmp; - rdev->mc.gtt_size = 0xFFFFFFFFUL - tmp; - } + rdev->mc.gtt_location = tmp; + rdev->mc.gtt_size = 0xFFFFFFFFUL - tmp; } - rdev->mc.gtt_location = rdev->mc.mc_vram_size; - } else { - rdev->mc.vram_location = 0x00000000UL; - rdev->mc.gtt_location = rdev->mc.mc_vram_size; - rdev->mc.gtt_size = radeon_gart_size * 1024 * 1024; } + rdev->mc.gtt_location = rdev->mc.mc_vram_size; } rdev->mc.vram_start = rdev->mc.vram_location; - rdev->mc.vram_end = rdev->mc.vram_location + rdev->mc.mc_vram_size; + rdev->mc.vram_end = rdev->mc.vram_location + rdev->mc.mc_vram_size - 1; rdev->mc.gtt_start = rdev->mc.gtt_location; - rdev->mc.gtt_end = rdev->mc.gtt_location + rdev->mc.gtt_size; + rdev->mc.gtt_end = rdev->mc.gtt_location + rdev->mc.gtt_size - 1; /* FIXME: we should enforce default clock in case GPU is not in * default setup */ @@ -456,6 +452,7 @@ int r600_mc_init(struct radeon_device *rdev) */ int r600_gpu_soft_reset(struct radeon_device *rdev) { + struct rv515_mc_save save; u32 grbm_busy_mask = S_008010_VC_BUSY(1) | S_008010_VGT_BUSY_NO_DMA(1) | S_008010_VGT_BUSY(1) | S_008010_TA03_BUSY(1) | S_008010_TC_BUSY(1) | S_008010_SX_BUSY(1) | @@ -473,13 +470,25 @@ int r600_gpu_soft_reset(struct radeon_device *rdev) S_008014_CB0_BUSY(1) | S_008014_CB1_BUSY(1) | S_008014_CB2_BUSY(1) | S_008014_CB3_BUSY(1); u32 srbm_reset = 0; + u32 tmp; + dev_info(rdev->dev, "GPU softreset \n"); + dev_info(rdev->dev, " R_008010_GRBM_STATUS=0x%08X\n", + RREG32(R_008010_GRBM_STATUS)); + dev_info(rdev->dev, " R_008014_GRBM_STATUS2=0x%08X\n", + RREG32(R_008014_GRBM_STATUS2)); + dev_info(rdev->dev, " R_000E50_SRBM_STATUS=0x%08X\n", + RREG32(R_000E50_SRBM_STATUS)); + rv515_mc_stop(rdev, &save); + if (r600_mc_wait_for_idle(rdev)) { + dev_warn(rdev->dev, "Wait for MC idle timedout !\n"); + } /* Disable CP parsing/prefetching */ WREG32(R_0086D8_CP_ME_CNTL, S_0086D8_CP_ME_HALT(0xff)); /* Check if any of the rendering block is busy and reset it */ if ((RREG32(R_008010_GRBM_STATUS) & grbm_busy_mask) || (RREG32(R_008014_GRBM_STATUS2) & grbm2_busy_mask)) { - WREG32(R_008020_GRBM_SOFT_RESET, S_008020_SOFT_RESET_CR(1) | + tmp = S_008020_SOFT_RESET_CR(1) | S_008020_SOFT_RESET_DB(1) | S_008020_SOFT_RESET_CB(1) | S_008020_SOFT_RESET_PA(1) | @@ -491,14 +500,18 @@ int r600_gpu_soft_reset(struct radeon_device *rdev) S_008020_SOFT_RESET_TC(1) | S_008020_SOFT_RESET_TA(1) | S_008020_SOFT_RESET_VC(1) | - S_008020_SOFT_RESET_VGT(1)); + S_008020_SOFT_RESET_VGT(1); + dev_info(rdev->dev, " R_008020_GRBM_SOFT_RESET=0x%08X\n", tmp); + WREG32(R_008020_GRBM_SOFT_RESET, tmp); (void)RREG32(R_008020_GRBM_SOFT_RESET); udelay(50); WREG32(R_008020_GRBM_SOFT_RESET, 0); (void)RREG32(R_008020_GRBM_SOFT_RESET); } /* Reset CP (we always reset CP) */ - WREG32(R_008020_GRBM_SOFT_RESET, S_008020_SOFT_RESET_CP(1)); + tmp = S_008020_SOFT_RESET_CP(1); + dev_info(rdev->dev, "R_008020_GRBM_SOFT_RESET=0x%08X\n", tmp); + WREG32(R_008020_GRBM_SOFT_RESET, tmp); (void)RREG32(R_008020_GRBM_SOFT_RESET); udelay(50); WREG32(R_008020_GRBM_SOFT_RESET, 0); @@ -526,6 +539,14 @@ int r600_gpu_soft_reset(struct radeon_device *rdev) srbm_reset |= S_000E60_SOFT_RESET_RLC(1); if (G_000E50_SEM_BUSY(RREG32(R_000E50_SRBM_STATUS))) srbm_reset |= S_000E60_SOFT_RESET_SEM(1); + if (G_000E50_BIF_BUSY(RREG32(R_000E50_SRBM_STATUS))) + srbm_reset |= S_000E60_SOFT_RESET_BIF(1); + dev_info(rdev->dev, " R_000E60_SRBM_SOFT_RESET=0x%08X\n", srbm_reset); + WREG32(R_000E60_SRBM_SOFT_RESET, srbm_reset); + (void)RREG32(R_000E60_SRBM_SOFT_RESET); + udelay(50); + WREG32(R_000E60_SRBM_SOFT_RESET, 0); + (void)RREG32(R_000E60_SRBM_SOFT_RESET); WREG32(R_000E60_SRBM_SOFT_RESET, srbm_reset); (void)RREG32(R_000E60_SRBM_SOFT_RESET); udelay(50); @@ -533,6 +554,17 @@ int r600_gpu_soft_reset(struct radeon_device *rdev) (void)RREG32(R_000E60_SRBM_SOFT_RESET); /* Wait a little for things to settle down */ udelay(50); + dev_info(rdev->dev, " R_008010_GRBM_STATUS=0x%08X\n", + RREG32(R_008010_GRBM_STATUS)); + dev_info(rdev->dev, " R_008014_GRBM_STATUS2=0x%08X\n", + RREG32(R_008014_GRBM_STATUS2)); + dev_info(rdev->dev, " R_000E50_SRBM_STATUS=0x%08X\n", + RREG32(R_000E50_SRBM_STATUS)); + /* After reset we need to reinit the asic as GPU often endup in an + * incoherent state. + */ + atom_asic_init(rdev->mode_info.atom_context); + rv515_mc_resume(rdev, &save); return 0; } @@ -826,7 +858,8 @@ void r600_gpu_init(struct radeon_device *rdev) ((rdev->family) == CHIP_RV630) || ((rdev->family) == CHIP_RV610) || ((rdev->family) == CHIP_RV620) || - ((rdev->family) == CHIP_RS780)) { + ((rdev->family) == CHIP_RS780) || + ((rdev->family) == CHIP_RS880)) { WREG32(DB_DEBUG, PREZ_MUST_WAIT_FOR_POSTZ_DONE); } else { WREG32(DB_DEBUG, 0); @@ -843,7 +876,8 @@ void r600_gpu_init(struct radeon_device *rdev) tmp = RREG32(SQ_MS_FIFO_SIZES); if (((rdev->family) == CHIP_RV610) || ((rdev->family) == CHIP_RV620) || - ((rdev->family) == CHIP_RS780)) { + ((rdev->family) == CHIP_RS780) || + ((rdev->family) == CHIP_RS880)) { tmp = (CACHE_FIFO_SIZE(0xa) | FETCH_FIFO_HIWATER(0xa) | DONE_FIFO_HIWATER(0xe0) | @@ -886,7 +920,8 @@ void r600_gpu_init(struct radeon_device *rdev) NUM_ES_STACK_ENTRIES(0)); } else if (((rdev->family) == CHIP_RV610) || ((rdev->family) == CHIP_RV620) || - ((rdev->family) == CHIP_RS780)) { + ((rdev->family) == CHIP_RS780) || + ((rdev->family) == CHIP_RS880)) { /* no vertex cache */ sq_config &= ~VC_ENABLE; @@ -943,7 +978,8 @@ void r600_gpu_init(struct radeon_device *rdev) if (((rdev->family) == CHIP_RV610) || ((rdev->family) == CHIP_RV620) || - ((rdev->family) == CHIP_RS780)) { + ((rdev->family) == CHIP_RS780) || + ((rdev->family) == CHIP_RS880)) { WREG32(VGT_CACHE_INVALIDATION, CACHE_INVALIDATION(TC_ONLY)); } else { WREG32(VGT_CACHE_INVALIDATION, CACHE_INVALIDATION(VC_AND_TC)); @@ -969,8 +1005,9 @@ void r600_gpu_init(struct radeon_device *rdev) tmp = rdev->config.r600.max_pipes * 16; switch (rdev->family) { case CHIP_RV610: - case CHIP_RS780: case CHIP_RV620: + case CHIP_RS780: + case CHIP_RS880: tmp += 32; break; case CHIP_RV670: @@ -1011,8 +1048,9 @@ void r600_gpu_init(struct radeon_device *rdev) switch (rdev->family) { case CHIP_RV610: - case CHIP_RS780: case CHIP_RV620: + case CHIP_RS780: + case CHIP_RS880: tmp = TC_L2_SIZE(8); break; case CHIP_RV630: @@ -1234,19 +1272,17 @@ int r600_cp_resume(struct radeon_device *rdev) /* Set ring buffer size */ rb_bufsz = drm_order(rdev->cp.ring_size / 8); + tmp = RB_NO_UPDATE | (drm_order(RADEON_GPU_PAGE_SIZE/8) << 8) | rb_bufsz; #ifdef __BIG_ENDIAN - WREG32(CP_RB_CNTL, BUF_SWAP_32BIT | RB_NO_UPDATE | - (drm_order(4096/8) << 8) | rb_bufsz); -#else - WREG32(CP_RB_CNTL, RB_NO_UPDATE | (drm_order(4096/8) << 8) | rb_bufsz); + tmp |= BUF_SWAP_32BIT; #endif + WREG32(CP_RB_CNTL, tmp); WREG32(CP_SEM_WAIT_TIMER, 0x4); /* Set the write pointer delay */ WREG32(CP_RB_WPTR_DELAY, 0); /* Initialize the ring buffer's read and write pointers */ - tmp = RREG32(CP_RB_CNTL); WREG32(CP_RB_CNTL, tmp | RB_RPTR_WR_ENA); WREG32(CP_RB_RPTR_WR, 0); WREG32(CP_RB_WPTR, 0); @@ -1343,32 +1379,47 @@ int r600_ring_test(struct radeon_device *rdev) return r; } -/* - * Writeback - */ -int r600_wb_init(struct radeon_device *rdev) +void r600_wb_disable(struct radeon_device *rdev) +{ + WREG32(SCRATCH_UMSK, 0); + if (rdev->wb.wb_obj) { + radeon_object_kunmap(rdev->wb.wb_obj); + radeon_object_unpin(rdev->wb.wb_obj); + } +} + +void r600_wb_fini(struct radeon_device *rdev) +{ + r600_wb_disable(rdev); + if (rdev->wb.wb_obj) { + radeon_object_unref(&rdev->wb.wb_obj); + rdev->wb.wb = NULL; + rdev->wb.wb_obj = NULL; + } +} + +int r600_wb_enable(struct radeon_device *rdev) { int r; if (rdev->wb.wb_obj == NULL) { - r = radeon_object_create(rdev, NULL, 4096, - true, - RADEON_GEM_DOMAIN_GTT, - false, &rdev->wb.wb_obj); + r = radeon_object_create(rdev, NULL, RADEON_GPU_PAGE_SIZE, true, + RADEON_GEM_DOMAIN_GTT, false, &rdev->wb.wb_obj); if (r) { - DRM_ERROR("radeon: failed to create WB buffer (%d).\n", r); + dev_warn(rdev->dev, "failed to create WB buffer (%d).\n", r); return r; } - r = radeon_object_pin(rdev->wb.wb_obj, - RADEON_GEM_DOMAIN_GTT, - &rdev->wb.gpu_addr); + r = radeon_object_pin(rdev->wb.wb_obj, RADEON_GEM_DOMAIN_GTT, + &rdev->wb.gpu_addr); if (r) { - DRM_ERROR("radeon: failed to pin WB buffer (%d).\n", r); + dev_warn(rdev->dev, "failed to pin WB buffer (%d).\n", r); + r600_wb_fini(rdev); return r; } r = radeon_object_kmap(rdev->wb.wb_obj, (void **)&rdev->wb.wb); if (r) { - DRM_ERROR("radeon: failed to map WB buffer (%d).\n", r); + dev_warn(rdev->dev, "failed to map WB buffer (%d).\n", r); + r600_wb_fini(rdev); return r; } } @@ -1379,21 +1430,6 @@ int r600_wb_init(struct radeon_device *rdev) return 0; } -void r600_wb_fini(struct radeon_device *rdev) -{ - if (rdev->wb.wb_obj) { - radeon_object_kunmap(rdev->wb.wb_obj); - radeon_object_unpin(rdev->wb.wb_obj); - radeon_object_unref(&rdev->wb.wb_obj); - rdev->wb.wb = NULL; - rdev->wb.wb_obj = NULL; - } -} - - -/* - * CS - */ void r600_fence_ring_emit(struct radeon_device *rdev, struct radeon_fence *fence) { @@ -1417,8 +1453,8 @@ int r600_copy_blit(struct radeon_device *rdev, uint64_t src_offset, uint64_t dst_offset, unsigned num_pages, struct radeon_fence *fence) { - r600_blit_prepare_copy(rdev, num_pages * 4096); - r600_kms_blit_copy(rdev, src_offset, dst_offset, num_pages * 4096); + r600_blit_prepare_copy(rdev, num_pages * RADEON_GPU_PAGE_SIZE); + r600_kms_blit_copy(rdev, src_offset, dst_offset, num_pages * RADEON_GPU_PAGE_SIZE); r600_blit_done_copy(rdev, fence); return 0; } @@ -1470,11 +1506,14 @@ int r600_startup(struct radeon_device *rdev) { int r; - r600_gpu_reset(rdev); - r600_mc_resume(rdev); - r = r600_pcie_gart_enable(rdev); - if (r) - return r; + r600_mc_program(rdev); + if (rdev->flags & RADEON_IS_AGP) { + r600_agp_enable(rdev); + } else { + r = r600_pcie_gart_enable(rdev); + if (r) + return r; + } r600_gpu_init(rdev); r = radeon_object_pin(rdev->r600_blit.shader_obj, RADEON_GEM_DOMAIN_VRAM, @@ -1493,9 +1532,8 @@ int r600_startup(struct radeon_device *rdev) r = r600_cp_resume(rdev); if (r) return r; - r = r600_wb_init(rdev); - if (r) - return r; + /* write back buffer are not vital so don't worry about failure */ + r600_wb_enable(rdev); return 0; } @@ -1517,15 +1555,12 @@ int r600_resume(struct radeon_device *rdev) { int r; - if (radeon_gpu_reset(rdev)) { - /* FIXME: what do we want to do here ? */ - } + /* Do not reset GPU before posting, on r600 hw unlike on r500 hw, + * posting will perform necessary task to bring back GPU into good + * shape. + */ /* post card */ - if (rdev->is_atom_bios) { - atom_asic_init(rdev->mode_info.atom_context); - } else { - radeon_combios_asic_init(rdev->ddev); - } + atom_asic_init(rdev->mode_info.atom_context); /* Initialize clocks */ r = radeon_clocks_init(rdev); if (r) { @@ -1538,7 +1573,7 @@ int r600_resume(struct radeon_device *rdev) return r; } - r = radeon_ib_test(rdev); + r = r600_ib_test(rdev); if (r) { DRM_ERROR("radeon: failled testing IB (%d).\n", r); return r; @@ -1546,13 +1581,12 @@ int r600_resume(struct radeon_device *rdev) return r; } - int r600_suspend(struct radeon_device *rdev) { /* FIXME: we should wait for ring to be empty */ r600_cp_stop(rdev); rdev->cp.ready = false; - + r600_wb_disable(rdev); r600_pcie_gart_disable(rdev); /* unpin shaders bo */ radeon_object_unpin(rdev->r600_blit.shader_obj); @@ -1569,7 +1603,6 @@ int r600_init(struct radeon_device *rdev) { int r; - rdev->new_init_path = true; r = radeon_dummy_page_init(rdev); if (r) return r; @@ -1586,8 +1619,10 @@ int r600_init(struct radeon_device *rdev) return -EINVAL; } /* Must be an ATOMBIOS */ - if (!rdev->is_atom_bios) + if (!rdev->is_atom_bios) { + dev_err(rdev->dev, "Expecting atombios for R600 GPU\n"); return -EINVAL; + } r = radeon_atombios_init(rdev); if (r) return r; @@ -1600,24 +1635,20 @@ int r600_init(struct radeon_device *rdev) r600_scratch_init(rdev); /* Initialize surface registers */ radeon_surface_init(rdev); + /* Initialize clocks */ radeon_get_clock_info(rdev->ddev); r = radeon_clocks_init(rdev); if (r) return r; + /* Initialize power management */ + radeon_pm_init(rdev); /* Fence driver */ r = radeon_fence_driver_init(rdev); if (r) return r; r = r600_mc_init(rdev); - if (r) { - if (rdev->flags & RADEON_IS_AGP) { - /* Retry with disabling AGP */ - r600_fini(rdev); - rdev->flags &= ~RADEON_IS_AGP; - return r600_init(rdev); - } + if (r) return r; - } /* Memory manager */ r = radeon_object_init(rdev); if (r) @@ -1646,12 +1677,10 @@ int r600_init(struct radeon_device *rdev) r = r600_startup(rdev); if (r) { - if (rdev->flags & RADEON_IS_AGP) { - /* Retry with disabling AGP */ - r600_fini(rdev); - rdev->flags &= ~RADEON_IS_AGP; - return r600_init(rdev); - } + r600_suspend(rdev); + r600_wb_fini(rdev); + radeon_ring_fini(rdev); + r600_pcie_gart_fini(rdev); rdev->accel_working = false; } if (rdev->accel_working) { @@ -1660,7 +1689,7 @@ int r600_init(struct radeon_device *rdev) DRM_ERROR("radeon: failled initializing IB pool (%d).\n", r); rdev->accel_working = false; } - r = radeon_ib_test(rdev); + r = r600_ib_test(rdev); if (r) { DRM_ERROR("radeon: failled testing IB (%d).\n", r); rdev->accel_working = false; @@ -1676,19 +1705,15 @@ void r600_fini(struct radeon_device *rdev) r600_blit_fini(rdev); radeon_ring_fini(rdev); + r600_wb_fini(rdev); r600_pcie_gart_fini(rdev); radeon_gem_fini(rdev); radeon_fence_driver_fini(rdev); radeon_clocks_fini(rdev); -#if __OS_HAS_AGP if (rdev->flags & RADEON_IS_AGP) radeon_agp_fini(rdev); -#endif radeon_object_fini(rdev); - if (rdev->is_atom_bios) - radeon_atombios_fini(rdev); - else - radeon_combios_fini(rdev); + radeon_atombios_fini(rdev); kfree(rdev->bios); rdev->bios = NULL; radeon_dummy_page_fini(rdev); diff --git a/drivers/gpu/drm/radeon/r600_blit.c b/drivers/gpu/drm/radeon/r600_blit.c index dde2ccb..5ea4323 100644 --- a/drivers/gpu/drm/radeon/r600_blit.c +++ b/drivers/gpu/drm/radeon/r600_blit.c @@ -737,7 +737,7 @@ r600_blit_copy(struct drm_device *dev, /* dst */ set_render_target(dev_priv, COLOR_8_8_8_8, - dst_x + cur_size, h, + (dst_x + cur_size) / 4, h, dst_gpu_addr); /* scissors */ @@ -774,11 +774,10 @@ r600_blit_swap(struct drm_device *dev, { drm_radeon_private_t *dev_priv = dev->dev_private; int cb_format, tex_format; + int sx2, sy2, dx2, dy2; u64 vb_addr; u32 *vb; - vb = r600_nomm_get_vb_ptr(dev); - if ((dev_priv->blit_vb->used + 48) > dev_priv->blit_vb->total) { r600_nomm_put_vb(dev); @@ -787,19 +786,13 @@ r600_blit_swap(struct drm_device *dev, return; set_shaders(dev); - vb = r600_nomm_get_vb_ptr(dev); } + vb = r600_nomm_get_vb_ptr(dev); - if (cpp == 4) { - cb_format = COLOR_8_8_8_8; - tex_format = FMT_8_8_8_8; - } else if (cpp == 2) { - cb_format = COLOR_5_6_5; - tex_format = FMT_5_6_5; - } else { - cb_format = COLOR_8; - tex_format = FMT_8; - } + sx2 = sx + w; + sy2 = sy + h; + dx2 = dx + w; + dy2 = dy + h; vb[0] = i2f(dx); vb[1] = i2f(dy); @@ -807,31 +800,46 @@ r600_blit_swap(struct drm_device *dev, vb[3] = i2f(sy); vb[4] = i2f(dx); - vb[5] = i2f(dy + h); + vb[5] = i2f(dy2); vb[6] = i2f(sx); - vb[7] = i2f(sy + h); + vb[7] = i2f(sy2); - vb[8] = i2f(dx + w); - vb[9] = i2f(dy + h); - vb[10] = i2f(sx + w); - vb[11] = i2f(sy + h); + vb[8] = i2f(dx2); + vb[9] = i2f(dy2); + vb[10] = i2f(sx2); + vb[11] = i2f(sy2); + + switch(cpp) { + case 4: + cb_format = COLOR_8_8_8_8; + tex_format = FMT_8_8_8_8; + break; + case 2: + cb_format = COLOR_5_6_5; + tex_format = FMT_5_6_5; + break; + default: + cb_format = COLOR_8; + tex_format = FMT_8; + break; + } /* src */ set_tex_resource(dev_priv, tex_format, src_pitch / cpp, - sy + h, src_pitch / cpp, + sy2, src_pitch / cpp, src_gpu_addr); cp_set_surface_sync(dev_priv, - R600_TC_ACTION_ENA, (src_pitch * (sy + h)), src_gpu_addr); + R600_TC_ACTION_ENA, src_pitch * sy2, src_gpu_addr); /* dst */ set_render_target(dev_priv, cb_format, - dst_pitch / cpp, dy + h, + dst_pitch / cpp, dy2, dst_gpu_addr); /* scissors */ - set_scissors(dev_priv, dx, dy, dx + w, dy + h); + set_scissors(dev_priv, dx, dy, dx2, dy2); /* Vertex buffer setup */ vb_addr = dev_priv->gart_buffers_offset + @@ -844,7 +852,7 @@ r600_blit_swap(struct drm_device *dev, cp_set_surface_sync(dev_priv, R600_CB_ACTION_ENA | R600_CB0_DEST_BASE_ENA, - dst_pitch * (dy + h), dst_gpu_addr); + dst_pitch * dy2, dst_gpu_addr); dev_priv->blit_vb->used += 12 * 4; } diff --git a/drivers/gpu/drm/radeon/r600_blit_kms.c b/drivers/gpu/drm/radeon/r600_blit_kms.c index 0a6f468..dbf716e 100644 --- a/drivers/gpu/drm/radeon/r600_blit_kms.c +++ b/drivers/gpu/drm/radeon/r600_blit_kms.c @@ -368,7 +368,7 @@ set_default_state(struct radeon_device *rdev) if ((rdev->family == CHIP_RV610) || (rdev->family == CHIP_RV620) || (rdev->family == CHIP_RS780) || - (rdev->family == CHIP_RS780) || + (rdev->family == CHIP_RS880) || (rdev->family == CHIP_RV710)) sq_config = 0; else @@ -774,7 +774,7 @@ void r600_kms_blit_copy(struct radeon_device *rdev, /* dst 23 */ set_render_target(rdev, COLOR_8_8_8_8, - dst_x + cur_size, h, + (dst_x + cur_size) / 4, h, dst_gpu_addr); /* scissors 12 */ diff --git a/drivers/gpu/drm/radeon/r600_cs.c b/drivers/gpu/drm/radeon/r600_cs.c index 33b89cd..0d82076 100644 --- a/drivers/gpu/drm/radeon/r600_cs.c +++ b/drivers/gpu/drm/radeon/r600_cs.c @@ -28,7 +28,6 @@ #include "drmP.h" #include "radeon.h" #include "r600d.h" -#include "avivod.h" static int r600_cs_packet_next_reloc_mm(struct radeon_cs_parser *p, struct radeon_cs_reloc **cs_reloc); @@ -57,7 +56,7 @@ int r600_cs_packet_parse(struct radeon_cs_parser *p, idx, ib_chunk->length_dw); return -EINVAL; } - header = ib_chunk->kdata[idx]; + header = radeon_get_ib_value(p, idx); pkt->idx = idx; pkt->type = CP_PACKET_GET_TYPE(header); pkt->count = CP_PACKET_GET_COUNT(header); @@ -98,7 +97,6 @@ int r600_cs_packet_parse(struct radeon_cs_parser *p, static int r600_cs_packet_next_reloc_mm(struct radeon_cs_parser *p, struct radeon_cs_reloc **cs_reloc) { - struct radeon_cs_chunk *ib_chunk; struct radeon_cs_chunk *relocs_chunk; struct radeon_cs_packet p3reloc; unsigned idx; @@ -109,7 +107,6 @@ static int r600_cs_packet_next_reloc_mm(struct radeon_cs_parser *p, return -EINVAL; } *cs_reloc = NULL; - ib_chunk = &p->chunks[p->chunk_ib_idx]; relocs_chunk = &p->chunks[p->chunk_relocs_idx]; r = r600_cs_packet_parse(p, &p3reloc, p->idx); if (r) { @@ -121,7 +118,7 @@ static int r600_cs_packet_next_reloc_mm(struct radeon_cs_parser *p, p3reloc.idx); return -EINVAL; } - idx = ib_chunk->kdata[p3reloc.idx + 1]; + idx = radeon_get_ib_value(p, p3reloc.idx + 1); if (idx >= relocs_chunk->length_dw) { DRM_ERROR("Relocs at %d after relocations chunk end %d !\n", idx, relocs_chunk->length_dw); @@ -146,7 +143,6 @@ static int r600_cs_packet_next_reloc_mm(struct radeon_cs_parser *p, static int r600_cs_packet_next_reloc_nomm(struct radeon_cs_parser *p, struct radeon_cs_reloc **cs_reloc) { - struct radeon_cs_chunk *ib_chunk; struct radeon_cs_chunk *relocs_chunk; struct radeon_cs_packet p3reloc; unsigned idx; @@ -157,7 +153,6 @@ static int r600_cs_packet_next_reloc_nomm(struct radeon_cs_parser *p, return -EINVAL; } *cs_reloc = NULL; - ib_chunk = &p->chunks[p->chunk_ib_idx]; relocs_chunk = &p->chunks[p->chunk_relocs_idx]; r = r600_cs_packet_parse(p, &p3reloc, p->idx); if (r) { @@ -169,7 +164,7 @@ static int r600_cs_packet_next_reloc_nomm(struct radeon_cs_parser *p, p3reloc.idx); return -EINVAL; } - idx = ib_chunk->kdata[p3reloc.idx + 1]; + idx = radeon_get_ib_value(p, p3reloc.idx + 1); if (idx >= relocs_chunk->length_dw) { DRM_ERROR("Relocs at %d after relocations chunk end %d !\n", idx, relocs_chunk->length_dw); @@ -181,13 +176,136 @@ static int r600_cs_packet_next_reloc_nomm(struct radeon_cs_parser *p, return 0; } +/** + * r600_cs_packet_next_vline() - parse userspace VLINE packet + * @parser: parser structure holding parsing context. + * + * Userspace sends a special sequence for VLINE waits. + * PACKET0 - VLINE_START_END + value + * PACKET3 - WAIT_REG_MEM poll vline status reg + * RELOC (P3) - crtc_id in reloc. + * + * This function parses this and relocates the VLINE START END + * and WAIT_REG_MEM packets to the correct crtc. + * It also detects a switched off crtc and nulls out the + * wait in that case. + */ +static int r600_cs_packet_parse_vline(struct radeon_cs_parser *p) +{ + struct drm_mode_object *obj; + struct drm_crtc *crtc; + struct radeon_crtc *radeon_crtc; + struct radeon_cs_packet p3reloc, wait_reg_mem; + int crtc_id; + int r; + uint32_t header, h_idx, reg, wait_reg_mem_info; + volatile uint32_t *ib; + + ib = p->ib->ptr; + + /* parse the WAIT_REG_MEM */ + r = r600_cs_packet_parse(p, &wait_reg_mem, p->idx); + if (r) + return r; + + /* check its a WAIT_REG_MEM */ + if (wait_reg_mem.type != PACKET_TYPE3 || + wait_reg_mem.opcode != PACKET3_WAIT_REG_MEM) { + DRM_ERROR("vline wait missing WAIT_REG_MEM segment\n"); + r = -EINVAL; + return r; + } + + wait_reg_mem_info = radeon_get_ib_value(p, wait_reg_mem.idx + 1); + /* bit 4 is reg (0) or mem (1) */ + if (wait_reg_mem_info & 0x10) { + DRM_ERROR("vline WAIT_REG_MEM waiting on MEM rather than REG\n"); + r = -EINVAL; + return r; + } + /* waiting for value to be equal */ + if ((wait_reg_mem_info & 0x7) != 0x3) { + DRM_ERROR("vline WAIT_REG_MEM function not equal\n"); + r = -EINVAL; + return r; + } + if ((radeon_get_ib_value(p, wait_reg_mem.idx + 2) << 2) != AVIVO_D1MODE_VLINE_STATUS) { + DRM_ERROR("vline WAIT_REG_MEM bad reg\n"); + r = -EINVAL; + return r; + } + + if (radeon_get_ib_value(p, wait_reg_mem.idx + 5) != AVIVO_D1MODE_VLINE_STAT) { + DRM_ERROR("vline WAIT_REG_MEM bad bit mask\n"); + r = -EINVAL; + return r; + } + + /* jump over the NOP */ + r = r600_cs_packet_parse(p, &p3reloc, p->idx + wait_reg_mem.count + 2); + if (r) + return r; + + h_idx = p->idx - 2; + p->idx += wait_reg_mem.count + 2; + p->idx += p3reloc.count + 2; + + header = radeon_get_ib_value(p, h_idx); + crtc_id = radeon_get_ib_value(p, h_idx + 2 + 7 + 1); + reg = CP_PACKET0_GET_REG(header); + mutex_lock(&p->rdev->ddev->mode_config.mutex); + obj = drm_mode_object_find(p->rdev->ddev, crtc_id, DRM_MODE_OBJECT_CRTC); + if (!obj) { + DRM_ERROR("cannot find crtc %d\n", crtc_id); + r = -EINVAL; + goto out; + } + crtc = obj_to_crtc(obj); + radeon_crtc = to_radeon_crtc(crtc); + crtc_id = radeon_crtc->crtc_id; + + if (!crtc->enabled) { + /* if the CRTC isn't enabled - we need to nop out the WAIT_REG_MEM */ + ib[h_idx + 2] = PACKET2(0); + ib[h_idx + 3] = PACKET2(0); + ib[h_idx + 4] = PACKET2(0); + ib[h_idx + 5] = PACKET2(0); + ib[h_idx + 6] = PACKET2(0); + ib[h_idx + 7] = PACKET2(0); + ib[h_idx + 8] = PACKET2(0); + } else if (crtc_id == 1) { + switch (reg) { + case AVIVO_D1MODE_VLINE_START_END: + header &= ~R600_CP_PACKET0_REG_MASK; + header |= AVIVO_D2MODE_VLINE_START_END >> 2; + break; + default: + DRM_ERROR("unknown crtc reloc\n"); + r = -EINVAL; + goto out; + } + ib[h_idx] = header; + ib[h_idx + 4] = AVIVO_D2MODE_VLINE_STATUS >> 2; + } +out: + mutex_unlock(&p->rdev->ddev->mode_config.mutex); + return r; +} + static int r600_packet0_check(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt, unsigned idx, unsigned reg) { + int r; + switch (reg) { case AVIVO_D1MODE_VLINE_START_END: - case AVIVO_D2MODE_VLINE_START_END: + r = r600_cs_packet_parse_vline(p); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + return r; + } break; default: printk(KERN_ERR "Forbidden register 0x%04X in cs at %d\n", @@ -218,17 +336,18 @@ static int r600_cs_parse_packet0(struct radeon_cs_parser *p, static int r600_packet3_check(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt) { - struct radeon_cs_chunk *ib_chunk; struct radeon_cs_reloc *reloc; volatile u32 *ib; unsigned idx; unsigned i; unsigned start_reg, end_reg, reg; int r; + u32 idx_value; ib = p->ib->ptr; - ib_chunk = &p->chunks[p->chunk_ib_idx]; idx = pkt->idx + 1; + idx_value = radeon_get_ib_value(p, idx); + switch (pkt->opcode) { case PACKET3_START_3D_CMDBUF: if (p->family >= CHIP_RV770 || pkt->count) { @@ -259,8 +378,8 @@ static int r600_packet3_check(struct radeon_cs_parser *p, DRM_ERROR("bad DRAW_INDEX\n"); return -EINVAL; } - ib[idx+0] += (u32)(reloc->lobj.gpu_offset & 0xffffffff); - ib[idx+1] = upper_32_bits(reloc->lobj.gpu_offset) & 0xff; + ib[idx+0] = idx_value + (u32)(reloc->lobj.gpu_offset & 0xffffffff); + ib[idx+1] += upper_32_bits(reloc->lobj.gpu_offset) & 0xff; break; case PACKET3_DRAW_INDEX_AUTO: if (pkt->count != 1) { @@ -281,14 +400,14 @@ static int r600_packet3_check(struct radeon_cs_parser *p, return -EINVAL; } /* bit 4 is reg (0) or mem (1) */ - if (ib_chunk->kdata[idx+0] & 0x10) { + if (idx_value & 0x10) { r = r600_cs_packet_next_reloc(p, &reloc); if (r) { DRM_ERROR("bad WAIT_REG_MEM\n"); return -EINVAL; } ib[idx+1] += (u32)(reloc->lobj.gpu_offset & 0xffffffff); - ib[idx+2] = upper_32_bits(reloc->lobj.gpu_offset) & 0xff; + ib[idx+2] += upper_32_bits(reloc->lobj.gpu_offset) & 0xff; } break; case PACKET3_SURFACE_SYNC: @@ -297,8 +416,8 @@ static int r600_packet3_check(struct radeon_cs_parser *p, return -EINVAL; } /* 0xffffffff/0x0 is flush all cache flag */ - if (ib_chunk->kdata[idx+1] != 0xffffffff || - ib_chunk->kdata[idx+2] != 0) { + if (radeon_get_ib_value(p, idx + 1) != 0xffffffff || + radeon_get_ib_value(p, idx + 2) != 0) { r = r600_cs_packet_next_reloc(p, &reloc); if (r) { DRM_ERROR("bad SURFACE_SYNC\n"); @@ -319,7 +438,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, return -EINVAL; } ib[idx+1] += (u32)(reloc->lobj.gpu_offset & 0xffffffff); - ib[idx+2] |= upper_32_bits(reloc->lobj.gpu_offset) & 0xff; + ib[idx+2] += upper_32_bits(reloc->lobj.gpu_offset) & 0xff; } break; case PACKET3_EVENT_WRITE_EOP: @@ -333,10 +452,10 @@ static int r600_packet3_check(struct radeon_cs_parser *p, return -EINVAL; } ib[idx+1] += (u32)(reloc->lobj.gpu_offset & 0xffffffff); - ib[idx+2] |= upper_32_bits(reloc->lobj.gpu_offset) & 0xff; + ib[idx+2] += upper_32_bits(reloc->lobj.gpu_offset) & 0xff; break; case PACKET3_SET_CONFIG_REG: - start_reg = (ib[idx+0] << 2) + PACKET3_SET_CONFIG_REG_OFFSET; + start_reg = (idx_value << 2) + PACKET3_SET_CONFIG_REG_OFFSET; end_reg = 4 * pkt->count + start_reg - 4; if ((start_reg < PACKET3_SET_CONFIG_REG_OFFSET) || (start_reg >= PACKET3_SET_CONFIG_REG_END) || @@ -347,6 +466,23 @@ static int r600_packet3_check(struct radeon_cs_parser *p, for (i = 0; i < pkt->count; i++) { reg = start_reg + (4 * i); switch (reg) { + case SQ_ESGS_RING_BASE: + case SQ_GSVS_RING_BASE: + case SQ_ESTMP_RING_BASE: + case SQ_GSTMP_RING_BASE: + case SQ_VSTMP_RING_BASE: + case SQ_PSTMP_RING_BASE: + case SQ_FBUF_RING_BASE: + case SQ_REDUC_RING_BASE: + case SX_MEMORY_EXPORT_BASE: + r = r600_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad SET_CONFIG_REG " + "0x%04X\n", reg); + return -EINVAL; + } + ib[idx+1+i] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + break; case CP_COHER_BASE: /* use PACKET3_SURFACE_SYNC */ return -EINVAL; @@ -356,7 +492,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, } break; case PACKET3_SET_CONTEXT_REG: - start_reg = (ib[idx+0] << 2) + PACKET3_SET_CONTEXT_REG_OFFSET; + start_reg = (idx_value << 2) + PACKET3_SET_CONTEXT_REG_OFFSET; end_reg = 4 * pkt->count + start_reg - 4; if ((start_reg < PACKET3_SET_CONTEXT_REG_OFFSET) || (start_reg >= PACKET3_SET_CONTEXT_REG_END) || @@ -368,6 +504,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, reg = start_reg + (4 * i); switch (reg) { case DB_DEPTH_BASE: + case DB_HTILE_DATA_BASE: case CB_COLOR0_BASE: case CB_COLOR1_BASE: case CB_COLOR2_BASE: @@ -421,7 +558,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, DRM_ERROR("bad SET_RESOURCE\n"); return -EINVAL; } - start_reg = (ib[idx+0] << 2) + PACKET3_SET_RESOURCE_OFFSET; + start_reg = (idx_value << 2) + PACKET3_SET_RESOURCE_OFFSET; end_reg = 4 * pkt->count + start_reg - 4; if ((start_reg < PACKET3_SET_RESOURCE_OFFSET) || (start_reg >= PACKET3_SET_RESOURCE_END) || @@ -430,7 +567,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, return -EINVAL; } for (i = 0; i < (pkt->count / 7); i++) { - switch (G__SQ_VTX_CONSTANT_TYPE(ib[idx+(i*7)+6+1])) { + switch (G__SQ_VTX_CONSTANT_TYPE(radeon_get_ib_value(p, idx+(i*7)+6+1))) { case SQ_TEX_VTX_VALID_TEXTURE: /* tex base */ r = r600_cs_packet_next_reloc(p, &reloc); @@ -455,7 +592,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, return -EINVAL; } ib[idx+1+(i*7)+0] += (u32)((reloc->lobj.gpu_offset) & 0xffffffff); - ib[idx+1+(i*7)+2] |= upper_32_bits(reloc->lobj.gpu_offset) & 0xff; + ib[idx+1+(i*7)+2] += upper_32_bits(reloc->lobj.gpu_offset) & 0xff; break; case SQ_TEX_VTX_INVALID_TEXTURE: case SQ_TEX_VTX_INVALID_BUFFER: @@ -466,7 +603,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, } break; case PACKET3_SET_ALU_CONST: - start_reg = (ib[idx+0] << 2) + PACKET3_SET_ALU_CONST_OFFSET; + start_reg = (idx_value << 2) + PACKET3_SET_ALU_CONST_OFFSET; end_reg = 4 * pkt->count + start_reg - 4; if ((start_reg < PACKET3_SET_ALU_CONST_OFFSET) || (start_reg >= PACKET3_SET_ALU_CONST_END) || @@ -476,7 +613,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, } break; case PACKET3_SET_BOOL_CONST: - start_reg = (ib[idx+0] << 2) + PACKET3_SET_BOOL_CONST_OFFSET; + start_reg = (idx_value << 2) + PACKET3_SET_BOOL_CONST_OFFSET; end_reg = 4 * pkt->count + start_reg - 4; if ((start_reg < PACKET3_SET_BOOL_CONST_OFFSET) || (start_reg >= PACKET3_SET_BOOL_CONST_END) || @@ -486,7 +623,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, } break; case PACKET3_SET_LOOP_CONST: - start_reg = (ib[idx+0] << 2) + PACKET3_SET_LOOP_CONST_OFFSET; + start_reg = (idx_value << 2) + PACKET3_SET_LOOP_CONST_OFFSET; end_reg = 4 * pkt->count + start_reg - 4; if ((start_reg < PACKET3_SET_LOOP_CONST_OFFSET) || (start_reg >= PACKET3_SET_LOOP_CONST_END) || @@ -496,7 +633,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, } break; case PACKET3_SET_CTL_CONST: - start_reg = (ib[idx+0] << 2) + PACKET3_SET_CTL_CONST_OFFSET; + start_reg = (idx_value << 2) + PACKET3_SET_CTL_CONST_OFFSET; end_reg = 4 * pkt->count + start_reg - 4; if ((start_reg < PACKET3_SET_CTL_CONST_OFFSET) || (start_reg >= PACKET3_SET_CTL_CONST_END) || @@ -510,7 +647,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, DRM_ERROR("bad SET_SAMPLER\n"); return -EINVAL; } - start_reg = (ib[idx+0] << 2) + PACKET3_SET_SAMPLER_OFFSET; + start_reg = (idx_value << 2) + PACKET3_SET_SAMPLER_OFFSET; end_reg = 4 * pkt->count + start_reg - 4; if ((start_reg < PACKET3_SET_SAMPLER_OFFSET) || (start_reg >= PACKET3_SET_SAMPLER_END) || @@ -602,6 +739,8 @@ static void r600_cs_parser_fini(struct radeon_cs_parser *parser, int error) kfree(parser->relocs); for (i = 0; i < parser->nchunks; i++) { kfree(parser->chunks[i].kdata); + kfree(parser->chunks[i].kpage[0]); + kfree(parser->chunks[i].kpage[1]); } kfree(parser->chunks); kfree(parser->chunks_array); @@ -639,7 +778,6 @@ int r600_cs_legacy(struct drm_device *dev, void *data, struct drm_file *filp, * uncached). */ ib_chunk = &parser.chunks[parser.chunk_ib_idx]; parser.ib->length_dw = ib_chunk->length_dw; - memcpy((void *)parser.ib->ptr, ib_chunk->kdata, ib_chunk->length_dw*4); *l = parser.ib->length_dw; r = r600_cs_parse(&parser); if (r) { @@ -647,6 +785,12 @@ int r600_cs_legacy(struct drm_device *dev, void *data, struct drm_file *filp, r600_cs_parser_fini(&parser, r); return r; } + r = radeon_cs_finish_pages(&parser); + if (r) { + DRM_ERROR("Invalid command stream !\n"); + r600_cs_parser_fini(&parser, r); + return r; + } r600_cs_parser_fini(&parser, r); return r; } diff --git a/drivers/gpu/drm/radeon/r600d.h b/drivers/gpu/drm/radeon/r600d.h index 4a9028a..27ab428 100644 --- a/drivers/gpu/drm/radeon/r600d.h +++ b/drivers/gpu/drm/radeon/r600d.h @@ -119,6 +119,7 @@ #define DB_DEBUG 0x9830 #define PREZ_MUST_WAIT_FOR_POSTZ_DONE (1 << 31) #define DB_DEPTH_BASE 0x2800C +#define DB_HTILE_DATA_BASE 0x28014 #define DB_WATERMARKS 0x9838 #define DEPTH_FREE(x) ((x) << 0) #define DEPTH_FLUSH(x) ((x) << 5) @@ -171,6 +172,14 @@ #define SQ_STACK_RESOURCE_MGMT_2 0x8c14 # define NUM_GS_STACK_ENTRIES(x) ((x) << 0) # define NUM_ES_STACK_ENTRIES(x) ((x) << 16) +#define SQ_ESGS_RING_BASE 0x8c40 +#define SQ_GSVS_RING_BASE 0x8c48 +#define SQ_ESTMP_RING_BASE 0x8c50 +#define SQ_GSTMP_RING_BASE 0x8c58 +#define SQ_VSTMP_RING_BASE 0x8c60 +#define SQ_PSTMP_RING_BASE 0x8c68 +#define SQ_FBUF_RING_BASE 0x8c70 +#define SQ_REDUC_RING_BASE 0x8c78 #define GRBM_CNTL 0x8000 # define GRBM_READ_TIMEOUT(x) ((x) << 0) @@ -271,6 +280,10 @@ #define PCIE_PORT_INDEX 0x0038 #define PCIE_PORT_DATA 0x003C +#define CHMAP 0x2004 +#define NOOFCHAN_SHIFT 12 +#define NOOFCHAN_MASK 0x00003000 + #define RAMCFG 0x2408 #define NOOFBANK_SHIFT 0 #define NOOFBANK_MASK 0x00000001 @@ -352,6 +365,7 @@ #define SX_MISC 0x28350 +#define SX_MEMORY_EXPORT_BASE 0x9010 #define SX_DEBUG_1 0x9054 #define SMX_EVENT_RELEASE (1 << 0) #define ENABLE_NEW_SMX_ADDRESS (1 << 16) @@ -643,6 +657,7 @@ #define G_000E50_MCDW_BUSY(x) (((x) >> 13) & 1) #define G_000E50_SEM_BUSY(x) (((x) >> 14) & 1) #define G_000E50_RLC_BUSY(x) (((x) >> 15) & 1) +#define G_000E50_BIF_BUSY(x) (((x) >> 29) & 1) #define R_000E60_SRBM_SOFT_RESET 0x0E60 #define S_000E60_SOFT_RESET_BIF(x) (((x) & 1) << 1) #define S_000E60_SOFT_RESET_CG(x) (((x) & 1) << 2) diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h index c839b60..224506a2 100644 --- a/drivers/gpu/drm/radeon/radeon.h +++ b/drivers/gpu/drm/radeon/radeon.h @@ -44,11 +44,30 @@ * - TESTING, TESTING, TESTING */ +/* Initialization path: + * We expect that acceleration initialization might fail for various + * reasons even thought we work hard to make it works on most + * configurations. In order to still have a working userspace in such + * situation the init path must succeed up to the memory controller + * initialization point. Failure before this point are considered as + * fatal error. Here is the init callchain : + * radeon_device_init perform common structure, mutex initialization + * asic_init setup the GPU memory layout and perform all + * one time initialization (failure in this + * function are considered fatal) + * asic_startup setup the GPU acceleration, in order to + * follow guideline the first thing this + * function should do is setting the GPU + * memory controller (only MC setup failure + * are considered as fatal) + */ + #include <asm/atomic.h> #include <linux/wait.h> #include <linux/list.h> #include <linux/kref.h> +#include "radeon_family.h" #include "radeon_mode.h" #include "radeon_reg.h" @@ -77,64 +96,6 @@ extern int radeon_tv; #define RADEONFB_CONN_LIMIT 4 #define RADEON_BIOS_NUM_SCRATCH 8 -enum radeon_family { - CHIP_R100, - CHIP_RV100, - CHIP_RS100, - CHIP_RV200, - CHIP_RS200, - CHIP_R200, - CHIP_RV250, - CHIP_RS300, - CHIP_RV280, - CHIP_R300, - CHIP_R350, - CHIP_RV350, - CHIP_RV380, - CHIP_R420, - CHIP_R423, - CHIP_RV410, - CHIP_RS400, - CHIP_RS480, - CHIP_RS600, - CHIP_RS690, - CHIP_RS740, - CHIP_RV515, - CHIP_R520, - CHIP_RV530, - CHIP_RV560, - CHIP_RV570, - CHIP_R580, - CHIP_R600, - CHIP_RV610, - CHIP_RV630, - CHIP_RV670, - CHIP_RV620, - CHIP_RV635, - CHIP_RS780, - CHIP_RS880, - CHIP_RV770, - CHIP_RV730, - CHIP_RV710, - CHIP_RV740, - CHIP_LAST, -}; - -enum radeon_chip_flags { - RADEON_FAMILY_MASK = 0x0000ffffUL, - RADEON_FLAGS_MASK = 0xffff0000UL, - RADEON_IS_MOBILITY = 0x00010000UL, - RADEON_IS_IGP = 0x00020000UL, - RADEON_SINGLE_CRTC = 0x00040000UL, - RADEON_IS_AGP = 0x00080000UL, - RADEON_HAS_HIERZ = 0x00100000UL, - RADEON_IS_PCIE = 0x00200000UL, - RADEON_NEW_MEMMAP = 0x00400000UL, - RADEON_IS_PCI = 0x00800000UL, - RADEON_IS_IGPGART = 0x01000000UL, -}; - - /* * Errata workarounds. */ @@ -178,6 +139,10 @@ struct radeon_clock { uint32_t default_sclk; }; +/* + * Power management + */ +int radeon_pm_init(struct radeon_device *rdev); /* * Fences. @@ -315,6 +280,8 @@ union radeon_gart_table { struct radeon_gart_table_vram vram; }; +#define RADEON_GPU_PAGE_SIZE 4096 + struct radeon_gart { dma_addr_t table_addr; unsigned num_gpu_pages; @@ -399,7 +366,7 @@ struct radeon_ib { unsigned long idx; uint64_t gpu_addr; struct radeon_fence *fence; - volatile uint32_t *ptr; + uint32_t *ptr; uint32_t length_dw; }; @@ -472,7 +439,12 @@ struct radeon_cs_reloc { struct radeon_cs_chunk { uint32_t chunk_id; uint32_t length_dw; + int kpage_idx[2]; + uint32_t *kpage[2]; uint32_t *kdata; + void __user *user_ptr; + int last_copied_page; + int last_page_index; }; struct radeon_cs_parser { @@ -495,8 +467,38 @@ struct radeon_cs_parser { struct radeon_ib *ib; void *track; unsigned family; + int parser_error; }; +extern int radeon_cs_update_pages(struct radeon_cs_parser *p, int pg_idx); +extern int radeon_cs_finish_pages(struct radeon_cs_parser *p); + + +static inline u32 radeon_get_ib_value(struct radeon_cs_parser *p, int idx) +{ + struct radeon_cs_chunk *ibc = &p->chunks[p->chunk_ib_idx]; + u32 pg_idx, pg_offset; + u32 idx_value = 0; + int new_page; + + pg_idx = (idx * 4) / PAGE_SIZE; + pg_offset = (idx * 4) % PAGE_SIZE; + + if (ibc->kpage_idx[0] == pg_idx) + return ibc->kpage[0][pg_offset/4]; + if (ibc->kpage_idx[1] == pg_idx) + return ibc->kpage[1][pg_offset/4]; + + new_page = radeon_cs_update_pages(p, pg_idx); + if (new_page < 0) { + p->parser_error = new_page; + return 0; + } + + idx_value = ibc->kpage[new_page][pg_offset/4]; + return idx_value; +} + struct radeon_cs_packet { unsigned idx; unsigned type; @@ -517,6 +519,7 @@ typedef int (*radeon_packet3_check_t)(struct radeon_cs_parser *p, * AGP */ int radeon_agp_init(struct radeon_device *rdev); +void radeon_agp_resume(struct radeon_device *rdev); void radeon_agp_fini(struct radeon_device *rdev); @@ -594,18 +597,8 @@ struct radeon_asic { void (*fini)(struct radeon_device *rdev); int (*resume)(struct radeon_device *rdev); int (*suspend)(struct radeon_device *rdev); - void (*errata)(struct radeon_device *rdev); - void (*vram_info)(struct radeon_device *rdev); void (*vga_set_state)(struct radeon_device *rdev, bool state); int (*gpu_reset)(struct radeon_device *rdev); - int (*mc_init)(struct radeon_device *rdev); - void (*mc_fini)(struct radeon_device *rdev); - int (*wb_init)(struct radeon_device *rdev); - void (*wb_fini)(struct radeon_device *rdev); - int (*gart_init)(struct radeon_device *rdev); - void (*gart_fini)(struct radeon_device *rdev); - int (*gart_enable)(struct radeon_device *rdev); - void (*gart_disable)(struct radeon_device *rdev); void (*gart_tlb_flush)(struct radeon_device *rdev); int (*gart_set_page)(struct radeon_device *rdev, int i, uint64_t addr); int (*cp_init)(struct radeon_device *rdev, unsigned ring_size); @@ -615,7 +608,6 @@ struct radeon_asic { void (*ring_start)(struct radeon_device *rdev); int (*ring_test)(struct radeon_device *rdev); void (*ring_ib_execute)(struct radeon_device *rdev, struct radeon_ib *ib); - int (*ib_test)(struct radeon_device *rdev); int (*irq_set)(struct radeon_device *rdev); int (*irq_process)(struct radeon_device *rdev); u32 (*get_vblank_counter)(struct radeon_device *rdev, int crtc); @@ -636,7 +628,9 @@ struct radeon_asic { uint64_t dst_offset, unsigned num_pages, struct radeon_fence *fence); + uint32_t (*get_engine_clock)(struct radeon_device *rdev); void (*set_engine_clock)(struct radeon_device *rdev, uint32_t eng_clock); + uint32_t (*get_memory_clock)(struct radeon_device *rdev); void (*set_memory_clock)(struct radeon_device *rdev, uint32_t mem_clock); void (*set_pcie_lanes)(struct radeon_device *rdev, int lanes); void (*set_clock_gating)(struct radeon_device *rdev, int enable); @@ -793,12 +787,12 @@ struct radeon_device { bool shutdown; bool suspend; bool need_dma32; - bool new_init_path; bool accel_working; struct radeon_surface_reg surface_regs[RADEON_GEM_MAX_SURFACES]; const struct firmware *me_fw; /* all family ME firmware */ const struct firmware *pfp_fw; /* r6/700 PFP firmware */ struct r600_blit r600_blit; + int msi_enabled; /* msi enabled */ }; int radeon_device_init(struct radeon_device *rdev, @@ -953,28 +947,14 @@ static inline void radeon_ring_write(struct radeon_device *rdev, uint32_t v) #define radeon_resume(rdev) (rdev)->asic->resume((rdev)) #define radeon_suspend(rdev) (rdev)->asic->suspend((rdev)) #define radeon_cs_parse(p) rdev->asic->cs_parse((p)) -#define radeon_errata(rdev) (rdev)->asic->errata((rdev)) -#define radeon_vram_info(rdev) (rdev)->asic->vram_info((rdev)) #define radeon_vga_set_state(rdev, state) (rdev)->asic->vga_set_state((rdev), (state)) #define radeon_gpu_reset(rdev) (rdev)->asic->gpu_reset((rdev)) -#define radeon_mc_init(rdev) (rdev)->asic->mc_init((rdev)) -#define radeon_mc_fini(rdev) (rdev)->asic->mc_fini((rdev)) -#define radeon_wb_init(rdev) (rdev)->asic->wb_init((rdev)) -#define radeon_wb_fini(rdev) (rdev)->asic->wb_fini((rdev)) -#define radeon_gpu_gart_init(rdev) (rdev)->asic->gart_init((rdev)) -#define radeon_gpu_gart_fini(rdev) (rdev)->asic->gart_fini((rdev)) -#define radeon_gart_enable(rdev) (rdev)->asic->gart_enable((rdev)) -#define radeon_gart_disable(rdev) (rdev)->asic->gart_disable((rdev)) #define radeon_gart_tlb_flush(rdev) (rdev)->asic->gart_tlb_flush((rdev)) #define radeon_gart_set_page(rdev, i, p) (rdev)->asic->gart_set_page((rdev), (i), (p)) -#define radeon_cp_init(rdev,rsize) (rdev)->asic->cp_init((rdev), (rsize)) -#define radeon_cp_fini(rdev) (rdev)->asic->cp_fini((rdev)) -#define radeon_cp_disable(rdev) (rdev)->asic->cp_disable((rdev)) #define radeon_cp_commit(rdev) (rdev)->asic->cp_commit((rdev)) #define radeon_ring_start(rdev) (rdev)->asic->ring_start((rdev)) #define radeon_ring_test(rdev) (rdev)->asic->ring_test((rdev)) #define radeon_ring_ib_execute(rdev, ib) (rdev)->asic->ring_ib_execute((rdev), (ib)) -#define radeon_ib_test(rdev) (rdev)->asic->ib_test((rdev)) #define radeon_irq_set(rdev) (rdev)->asic->irq_set((rdev)) #define radeon_irq_process(rdev) (rdev)->asic->irq_process((rdev)) #define radeon_get_vblank_counter(rdev, crtc) (rdev)->asic->get_vblank_counter((rdev), (crtc)) @@ -982,7 +962,9 @@ static inline void radeon_ring_write(struct radeon_device *rdev, uint32_t v) #define radeon_copy_blit(rdev, s, d, np, f) (rdev)->asic->copy_blit((rdev), (s), (d), (np), (f)) #define radeon_copy_dma(rdev, s, d, np, f) (rdev)->asic->copy_dma((rdev), (s), (d), (np), (f)) #define radeon_copy(rdev, s, d, np, f) (rdev)->asic->copy((rdev), (s), (d), (np), (f)) +#define radeon_get_engine_clock(rdev) (rdev)->asic->get_engine_clock((rdev)) #define radeon_set_engine_clock(rdev, e) (rdev)->asic->set_engine_clock((rdev), (e)) +#define radeon_get_memory_clock(rdev) (rdev)->asic->get_memory_clock((rdev)) #define radeon_set_memory_clock(rdev, e) (rdev)->asic->set_engine_clock((rdev), (e)) #define radeon_set_pcie_lanes(rdev, l) (rdev)->asic->set_pcie_lanes((rdev), (l)) #define radeon_set_clock_gating(rdev, e) (rdev)->asic->set_clock_gating((rdev), (e)) @@ -1000,6 +982,8 @@ extern void radeon_clocks_fini(struct radeon_device *rdev); extern void radeon_scratch_init(struct radeon_device *rdev); extern void radeon_surface_init(struct radeon_device *rdev); extern int radeon_cs_parser_init(struct radeon_cs_parser *p, void *data); +extern void radeon_legacy_set_clock_gating(struct radeon_device *rdev, int enable); +extern void radeon_atom_set_clock_gating(struct radeon_device *rdev, int enable); /* r100,rv100,rs100,rv200,rs200,r200,rv250,rs300,rv280 */ struct r100_mc_save { @@ -1031,23 +1015,71 @@ extern void r100_vram_init_sizes(struct radeon_device *rdev); extern void r100_wb_disable(struct radeon_device *rdev); extern void r100_wb_fini(struct radeon_device *rdev); extern int r100_wb_init(struct radeon_device *rdev); +extern void r100_hdp_reset(struct radeon_device *rdev); +extern int r100_rb2d_reset(struct radeon_device *rdev); +extern int r100_cp_reset(struct radeon_device *rdev); +extern void r100_vga_render_disable(struct radeon_device *rdev); +extern int r100_cs_track_check_pkt3_indx_buffer(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt, + struct radeon_object *robj); +extern int r100_cs_parse_packet0(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt, + const unsigned *auth, unsigned n, + radeon_packet0_check_t check); +extern int r100_cs_packet_parse(struct radeon_cs_parser *p, + struct radeon_cs_packet *pkt, + unsigned idx); + +/* rv200,rv250,rv280 */ +extern void r200_set_safe_registers(struct radeon_device *rdev); /* r300,r350,rv350,rv370,rv380 */ extern void r300_set_reg_safe(struct radeon_device *rdev); extern void r300_mc_program(struct radeon_device *rdev); extern void r300_vram_info(struct radeon_device *rdev); +extern void r300_clock_startup(struct radeon_device *rdev); +extern int r300_mc_wait_for_idle(struct radeon_device *rdev); extern int rv370_pcie_gart_init(struct radeon_device *rdev); extern void rv370_pcie_gart_fini(struct radeon_device *rdev); extern int rv370_pcie_gart_enable(struct radeon_device *rdev); extern void rv370_pcie_gart_disable(struct radeon_device *rdev); /* r420,r423,rv410 */ +extern int r420_mc_init(struct radeon_device *rdev); extern u32 r420_mc_rreg(struct radeon_device *rdev, u32 reg); extern void r420_mc_wreg(struct radeon_device *rdev, u32 reg, u32 v); extern int r420_debugfs_pipes_info_init(struct radeon_device *rdev); +extern void r420_pipes_init(struct radeon_device *rdev); /* rv515 */ +struct rv515_mc_save { + u32 d1vga_control; + u32 d2vga_control; + u32 vga_render_control; + u32 vga_hdp_control; + u32 d1crtc_control; + u32 d2crtc_control; +}; extern void rv515_bandwidth_avivo_update(struct radeon_device *rdev); +extern void rv515_vga_render_disable(struct radeon_device *rdev); +extern void rv515_set_safe_registers(struct radeon_device *rdev); +extern void rv515_mc_stop(struct radeon_device *rdev, struct rv515_mc_save *save); +extern void rv515_mc_resume(struct radeon_device *rdev, struct rv515_mc_save *save); +extern void rv515_clock_startup(struct radeon_device *rdev); +extern void rv515_debugfs(struct radeon_device *rdev); +extern int rv515_suspend(struct radeon_device *rdev); + +/* rs400 */ +extern int rs400_gart_init(struct radeon_device *rdev); +extern int rs400_gart_enable(struct radeon_device *rdev); +extern void rs400_gart_adjust_size(struct radeon_device *rdev); +extern void rs400_gart_disable(struct radeon_device *rdev); +extern void rs400_gart_fini(struct radeon_device *rdev); + +/* rs600 */ +extern void rs600_set_safe_registers(struct radeon_device *rdev); +extern int rs600_irq_set(struct radeon_device *rdev); +extern void rs600_irq_disable(struct radeon_device *rdev); /* rs690, rs740 */ extern void rs690_line_buffer_adjust(struct radeon_device *rdev, @@ -1066,8 +1098,9 @@ extern int r600_pcie_gart_init(struct radeon_device *rdev); extern void r600_pcie_gart_tlb_flush(struct radeon_device *rdev); extern int r600_ib_test(struct radeon_device *rdev); extern int r600_ring_test(struct radeon_device *rdev); -extern int r600_wb_init(struct radeon_device *rdev); extern void r600_wb_fini(struct radeon_device *rdev); +extern int r600_wb_enable(struct radeon_device *rdev); +extern void r600_wb_disable(struct radeon_device *rdev); extern void r600_scratch_init(struct radeon_device *rdev); extern int r600_blit_init(struct radeon_device *rdev); extern void r600_blit_fini(struct radeon_device *rdev); diff --git a/drivers/gpu/drm/radeon/radeon_agp.c b/drivers/gpu/drm/radeon/radeon_agp.c index 23ea995..54bf49a 100644 --- a/drivers/gpu/drm/radeon/radeon_agp.c +++ b/drivers/gpu/drm/radeon/radeon_agp.c @@ -237,6 +237,18 @@ int radeon_agp_init(struct radeon_device *rdev) #endif } +void radeon_agp_resume(struct radeon_device *rdev) +{ +#if __OS_HAS_AGP + int r; + if (rdev->flags & RADEON_IS_AGP) { + r = radeon_agp_init(rdev); + if (r) + dev_warn(rdev->dev, "radeon AGP reinit failed\n"); + } +#endif +} + void radeon_agp_fini(struct radeon_device *rdev) { #if __OS_HAS_AGP diff --git a/drivers/gpu/drm/radeon/radeon_asic.h b/drivers/gpu/drm/radeon/radeon_asic.h index 8968f78..c18fbee 100644 --- a/drivers/gpu/drm/radeon/radeon_asic.h +++ b/drivers/gpu/drm/radeon/radeon_asic.h @@ -31,38 +31,30 @@ /* * common functions */ +uint32_t radeon_legacy_get_engine_clock(struct radeon_device *rdev); void radeon_legacy_set_engine_clock(struct radeon_device *rdev, uint32_t eng_clock); void radeon_legacy_set_clock_gating(struct radeon_device *rdev, int enable); +uint32_t radeon_atom_get_engine_clock(struct radeon_device *rdev); void radeon_atom_set_engine_clock(struct radeon_device *rdev, uint32_t eng_clock); +uint32_t radeon_atom_get_memory_clock(struct radeon_device *rdev); void radeon_atom_set_memory_clock(struct radeon_device *rdev, uint32_t mem_clock); void radeon_atom_set_clock_gating(struct radeon_device *rdev, int enable); /* * r100,rv100,rs100,rv200,rs200,r200,rv250,rs300,rv280 */ -int r100_init(struct radeon_device *rdev); -int r200_init(struct radeon_device *rdev); +extern int r100_init(struct radeon_device *rdev); +extern void r100_fini(struct radeon_device *rdev); +extern int r100_suspend(struct radeon_device *rdev); +extern int r100_resume(struct radeon_device *rdev); uint32_t r100_mm_rreg(struct radeon_device *rdev, uint32_t reg); void r100_mm_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v); -void r100_errata(struct radeon_device *rdev); -void r100_vram_info(struct radeon_device *rdev); void r100_vga_set_state(struct radeon_device *rdev, bool state); int r100_gpu_reset(struct radeon_device *rdev); -int r100_mc_init(struct radeon_device *rdev); -void r100_mc_fini(struct radeon_device *rdev); u32 r100_get_vblank_counter(struct radeon_device *rdev, int crtc); -int r100_wb_init(struct radeon_device *rdev); -void r100_wb_fini(struct radeon_device *rdev); -int r100_pci_gart_init(struct radeon_device *rdev); -void r100_pci_gart_fini(struct radeon_device *rdev); -int r100_pci_gart_enable(struct radeon_device *rdev); -void r100_pci_gart_disable(struct radeon_device *rdev); void r100_pci_gart_tlb_flush(struct radeon_device *rdev); int r100_pci_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr); -int r100_cp_init(struct radeon_device *rdev, unsigned ring_size); -void r100_cp_fini(struct radeon_device *rdev); -void r100_cp_disable(struct radeon_device *rdev); void r100_cp_commit(struct radeon_device *rdev); void r100_ring_start(struct radeon_device *rdev); int r100_irq_set(struct radeon_device *rdev); @@ -83,33 +75,21 @@ int r100_set_surface_reg(struct radeon_device *rdev, int reg, int r100_clear_surface_reg(struct radeon_device *rdev, int reg); void r100_bandwidth_update(struct radeon_device *rdev); void r100_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib); -int r100_ib_test(struct radeon_device *rdev); int r100_ring_test(struct radeon_device *rdev); static struct radeon_asic r100_asic = { .init = &r100_init, - .errata = &r100_errata, - .vram_info = &r100_vram_info, + .fini = &r100_fini, + .suspend = &r100_suspend, + .resume = &r100_resume, .vga_set_state = &r100_vga_set_state, .gpu_reset = &r100_gpu_reset, - .mc_init = &r100_mc_init, - .mc_fini = &r100_mc_fini, - .wb_init = &r100_wb_init, - .wb_fini = &r100_wb_fini, - .gart_init = &r100_pci_gart_init, - .gart_fini = &r100_pci_gart_fini, - .gart_enable = &r100_pci_gart_enable, - .gart_disable = &r100_pci_gart_disable, .gart_tlb_flush = &r100_pci_gart_tlb_flush, .gart_set_page = &r100_pci_gart_set_page, - .cp_init = &r100_cp_init, - .cp_fini = &r100_cp_fini, - .cp_disable = &r100_cp_disable, .cp_commit = &r100_cp_commit, .ring_start = &r100_ring_start, .ring_test = &r100_ring_test, .ring_ib_execute = &r100_ring_ib_execute, - .ib_test = &r100_ib_test, .irq_set = &r100_irq_set, .irq_process = &r100_irq_process, .get_vblank_counter = &r100_get_vblank_counter, @@ -118,7 +98,9 @@ static struct radeon_asic r100_asic = { .copy_blit = &r100_copy_blit, .copy_dma = NULL, .copy = &r100_copy_blit, + .get_engine_clock = &radeon_legacy_get_engine_clock, .set_engine_clock = &radeon_legacy_set_engine_clock, + .get_memory_clock = NULL, .set_memory_clock = NULL, .set_pcie_lanes = NULL, .set_clock_gating = &radeon_legacy_set_clock_gating, @@ -131,55 +113,38 @@ static struct radeon_asic r100_asic = { /* * r300,r350,rv350,rv380 */ -int r300_init(struct radeon_device *rdev); -void r300_errata(struct radeon_device *rdev); -void r300_vram_info(struct radeon_device *rdev); -int r300_gpu_reset(struct radeon_device *rdev); -int r300_mc_init(struct radeon_device *rdev); -void r300_mc_fini(struct radeon_device *rdev); -void r300_ring_start(struct radeon_device *rdev); -void r300_fence_ring_emit(struct radeon_device *rdev, - struct radeon_fence *fence); -int r300_cs_parse(struct radeon_cs_parser *p); -int rv370_pcie_gart_init(struct radeon_device *rdev); -void rv370_pcie_gart_fini(struct radeon_device *rdev); -int rv370_pcie_gart_enable(struct radeon_device *rdev); -void rv370_pcie_gart_disable(struct radeon_device *rdev); -void rv370_pcie_gart_tlb_flush(struct radeon_device *rdev); -int rv370_pcie_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr); -uint32_t rv370_pcie_rreg(struct radeon_device *rdev, uint32_t reg); -void rv370_pcie_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v); -void rv370_set_pcie_lanes(struct radeon_device *rdev, int lanes); -int r300_copy_dma(struct radeon_device *rdev, - uint64_t src_offset, - uint64_t dst_offset, - unsigned num_pages, - struct radeon_fence *fence); - +extern int r300_init(struct radeon_device *rdev); +extern void r300_fini(struct radeon_device *rdev); +extern int r300_suspend(struct radeon_device *rdev); +extern int r300_resume(struct radeon_device *rdev); +extern int r300_gpu_reset(struct radeon_device *rdev); +extern void r300_ring_start(struct radeon_device *rdev); +extern void r300_fence_ring_emit(struct radeon_device *rdev, + struct radeon_fence *fence); +extern int r300_cs_parse(struct radeon_cs_parser *p); +extern void rv370_pcie_gart_tlb_flush(struct radeon_device *rdev); +extern int rv370_pcie_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr); +extern uint32_t rv370_pcie_rreg(struct radeon_device *rdev, uint32_t reg); +extern void rv370_pcie_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v); +extern void rv370_set_pcie_lanes(struct radeon_device *rdev, int lanes); +extern int r300_copy_dma(struct radeon_device *rdev, + uint64_t src_offset, + uint64_t dst_offset, + unsigned num_pages, + struct radeon_fence *fence); static struct radeon_asic r300_asic = { .init = &r300_init, - .errata = &r300_errata, - .vram_info = &r300_vram_info, + .fini = &r300_fini, + .suspend = &r300_suspend, + .resume = &r300_resume, .vga_set_state = &r100_vga_set_state, .gpu_reset = &r300_gpu_reset, - .mc_init = &r300_mc_init, - .mc_fini = &r300_mc_fini, - .wb_init = &r100_wb_init, - .wb_fini = &r100_wb_fini, - .gart_init = &r100_pci_gart_init, - .gart_fini = &r100_pci_gart_fini, - .gart_enable = &r100_pci_gart_enable, - .gart_disable = &r100_pci_gart_disable, .gart_tlb_flush = &r100_pci_gart_tlb_flush, .gart_set_page = &r100_pci_gart_set_page, - .cp_init = &r100_cp_init, - .cp_fini = &r100_cp_fini, - .cp_disable = &r100_cp_disable, .cp_commit = &r100_cp_commit, .ring_start = &r300_ring_start, .ring_test = &r100_ring_test, .ring_ib_execute = &r100_ring_ib_execute, - .ib_test = &r100_ib_test, .irq_set = &r100_irq_set, .irq_process = &r100_irq_process, .get_vblank_counter = &r100_get_vblank_counter, @@ -188,7 +153,9 @@ static struct radeon_asic r300_asic = { .copy_blit = &r100_copy_blit, .copy_dma = &r300_copy_dma, .copy = &r100_copy_blit, + .get_engine_clock = &radeon_legacy_get_engine_clock, .set_engine_clock = &radeon_legacy_set_engine_clock, + .get_memory_clock = NULL, .set_memory_clock = NULL, .set_pcie_lanes = &rv370_set_pcie_lanes, .set_clock_gating = &radeon_legacy_set_clock_gating, @@ -209,26 +176,14 @@ static struct radeon_asic r420_asic = { .fini = &r420_fini, .suspend = &r420_suspend, .resume = &r420_resume, - .errata = NULL, - .vram_info = NULL, .vga_set_state = &r100_vga_set_state, .gpu_reset = &r300_gpu_reset, - .mc_init = NULL, - .mc_fini = NULL, - .wb_init = NULL, - .wb_fini = NULL, - .gart_enable = NULL, - .gart_disable = NULL, .gart_tlb_flush = &rv370_pcie_gart_tlb_flush, .gart_set_page = &rv370_pcie_gart_set_page, - .cp_init = NULL, - .cp_fini = NULL, - .cp_disable = NULL, .cp_commit = &r100_cp_commit, .ring_start = &r300_ring_start, .ring_test = &r100_ring_test, .ring_ib_execute = &r100_ring_ib_execute, - .ib_test = NULL, .irq_set = &r100_irq_set, .irq_process = &r100_irq_process, .get_vblank_counter = &r100_get_vblank_counter, @@ -237,7 +192,9 @@ static struct radeon_asic r420_asic = { .copy_blit = &r100_copy_blit, .copy_dma = &r300_copy_dma, .copy = &r100_copy_blit, + .get_engine_clock = &radeon_atom_get_engine_clock, .set_engine_clock = &radeon_atom_set_engine_clock, + .get_memory_clock = &radeon_atom_get_memory_clock, .set_memory_clock = &radeon_atom_set_memory_clock, .set_pcie_lanes = &rv370_set_pcie_lanes, .set_clock_gating = &radeon_atom_set_clock_gating, @@ -250,42 +207,27 @@ static struct radeon_asic r420_asic = { /* * rs400,rs480 */ -void rs400_errata(struct radeon_device *rdev); -void rs400_vram_info(struct radeon_device *rdev); -int rs400_mc_init(struct radeon_device *rdev); -void rs400_mc_fini(struct radeon_device *rdev); -int rs400_gart_init(struct radeon_device *rdev); -void rs400_gart_fini(struct radeon_device *rdev); -int rs400_gart_enable(struct radeon_device *rdev); -void rs400_gart_disable(struct radeon_device *rdev); +extern int rs400_init(struct radeon_device *rdev); +extern void rs400_fini(struct radeon_device *rdev); +extern int rs400_suspend(struct radeon_device *rdev); +extern int rs400_resume(struct radeon_device *rdev); void rs400_gart_tlb_flush(struct radeon_device *rdev); int rs400_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr); uint32_t rs400_mc_rreg(struct radeon_device *rdev, uint32_t reg); void rs400_mc_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v); static struct radeon_asic rs400_asic = { - .init = &r300_init, - .errata = &rs400_errata, - .vram_info = &rs400_vram_info, + .init = &rs400_init, + .fini = &rs400_fini, + .suspend = &rs400_suspend, + .resume = &rs400_resume, .vga_set_state = &r100_vga_set_state, .gpu_reset = &r300_gpu_reset, - .mc_init = &rs400_mc_init, - .mc_fini = &rs400_mc_fini, - .wb_init = &r100_wb_init, - .wb_fini = &r100_wb_fini, - .gart_init = &rs400_gart_init, - .gart_fini = &rs400_gart_fini, - .gart_enable = &rs400_gart_enable, - .gart_disable = &rs400_gart_disable, .gart_tlb_flush = &rs400_gart_tlb_flush, .gart_set_page = &rs400_gart_set_page, - .cp_init = &r100_cp_init, - .cp_fini = &r100_cp_fini, - .cp_disable = &r100_cp_disable, .cp_commit = &r100_cp_commit, .ring_start = &r300_ring_start, .ring_test = &r100_ring_test, .ring_ib_execute = &r100_ring_ib_execute, - .ib_test = &r100_ib_test, .irq_set = &r100_irq_set, .irq_process = &r100_irq_process, .get_vblank_counter = &r100_get_vblank_counter, @@ -294,7 +236,9 @@ static struct radeon_asic rs400_asic = { .copy_blit = &r100_copy_blit, .copy_dma = &r300_copy_dma, .copy = &r100_copy_blit, + .get_engine_clock = &radeon_legacy_get_engine_clock, .set_engine_clock = &radeon_legacy_set_engine_clock, + .get_memory_clock = NULL, .set_memory_clock = NULL, .set_pcie_lanes = NULL, .set_clock_gating = &radeon_legacy_set_clock_gating, @@ -307,18 +251,13 @@ static struct radeon_asic rs400_asic = { /* * rs600. */ -int rs600_init(struct radeon_device *rdev); -void rs600_errata(struct radeon_device *rdev); -void rs600_vram_info(struct radeon_device *rdev); -int rs600_mc_init(struct radeon_device *rdev); -void rs600_mc_fini(struct radeon_device *rdev); +extern int rs600_init(struct radeon_device *rdev); +extern void rs600_fini(struct radeon_device *rdev); +extern int rs600_suspend(struct radeon_device *rdev); +extern int rs600_resume(struct radeon_device *rdev); int rs600_irq_set(struct radeon_device *rdev); int rs600_irq_process(struct radeon_device *rdev); u32 rs600_get_vblank_counter(struct radeon_device *rdev, int crtc); -int rs600_gart_init(struct radeon_device *rdev); -void rs600_gart_fini(struct radeon_device *rdev); -int rs600_gart_enable(struct radeon_device *rdev); -void rs600_gart_disable(struct radeon_device *rdev); void rs600_gart_tlb_flush(struct radeon_device *rdev); int rs600_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr); uint32_t rs600_mc_rreg(struct radeon_device *rdev, uint32_t reg); @@ -326,28 +265,17 @@ void rs600_mc_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v); void rs600_bandwidth_update(struct radeon_device *rdev); static struct radeon_asic rs600_asic = { .init = &rs600_init, - .errata = &rs600_errata, - .vram_info = &rs600_vram_info, + .fini = &rs600_fini, + .suspend = &rs600_suspend, + .resume = &rs600_resume, .vga_set_state = &r100_vga_set_state, .gpu_reset = &r300_gpu_reset, - .mc_init = &rs600_mc_init, - .mc_fini = &rs600_mc_fini, - .wb_init = &r100_wb_init, - .wb_fini = &r100_wb_fini, - .gart_init = &rs600_gart_init, - .gart_fini = &rs600_gart_fini, - .gart_enable = &rs600_gart_enable, - .gart_disable = &rs600_gart_disable, .gart_tlb_flush = &rs600_gart_tlb_flush, .gart_set_page = &rs600_gart_set_page, - .cp_init = &r100_cp_init, - .cp_fini = &r100_cp_fini, - .cp_disable = &r100_cp_disable, .cp_commit = &r100_cp_commit, .ring_start = &r300_ring_start, .ring_test = &r100_ring_test, .ring_ib_execute = &r100_ring_ib_execute, - .ib_test = &r100_ib_test, .irq_set = &rs600_irq_set, .irq_process = &rs600_irq_process, .get_vblank_counter = &rs600_get_vblank_counter, @@ -356,7 +284,9 @@ static struct radeon_asic rs600_asic = { .copy_blit = &r100_copy_blit, .copy_dma = &r300_copy_dma, .copy = &r100_copy_blit, + .get_engine_clock = &radeon_atom_get_engine_clock, .set_engine_clock = &radeon_atom_set_engine_clock, + .get_memory_clock = &radeon_atom_get_memory_clock, .set_memory_clock = &radeon_atom_set_memory_clock, .set_pcie_lanes = NULL, .set_clock_gating = &radeon_atom_set_clock_gating, @@ -367,37 +297,26 @@ static struct radeon_asic rs600_asic = { /* * rs690,rs740 */ -void rs690_errata(struct radeon_device *rdev); -void rs690_vram_info(struct radeon_device *rdev); -int rs690_mc_init(struct radeon_device *rdev); -void rs690_mc_fini(struct radeon_device *rdev); +int rs690_init(struct radeon_device *rdev); +void rs690_fini(struct radeon_device *rdev); +int rs690_resume(struct radeon_device *rdev); +int rs690_suspend(struct radeon_device *rdev); uint32_t rs690_mc_rreg(struct radeon_device *rdev, uint32_t reg); void rs690_mc_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v); void rs690_bandwidth_update(struct radeon_device *rdev); static struct radeon_asic rs690_asic = { - .init = &rs600_init, - .errata = &rs690_errata, - .vram_info = &rs690_vram_info, + .init = &rs690_init, + .fini = &rs690_fini, + .suspend = &rs690_suspend, + .resume = &rs690_resume, .vga_set_state = &r100_vga_set_state, .gpu_reset = &r300_gpu_reset, - .mc_init = &rs690_mc_init, - .mc_fini = &rs690_mc_fini, - .wb_init = &r100_wb_init, - .wb_fini = &r100_wb_fini, - .gart_init = &rs400_gart_init, - .gart_fini = &rs400_gart_fini, - .gart_enable = &rs400_gart_enable, - .gart_disable = &rs400_gart_disable, .gart_tlb_flush = &rs400_gart_tlb_flush, .gart_set_page = &rs400_gart_set_page, - .cp_init = &r100_cp_init, - .cp_fini = &r100_cp_fini, - .cp_disable = &r100_cp_disable, .cp_commit = &r100_cp_commit, .ring_start = &r300_ring_start, .ring_test = &r100_ring_test, .ring_ib_execute = &r100_ring_ib_execute, - .ib_test = &r100_ib_test, .irq_set = &rs600_irq_set, .irq_process = &rs600_irq_process, .get_vblank_counter = &rs600_get_vblank_counter, @@ -406,7 +325,9 @@ static struct radeon_asic rs690_asic = { .copy_blit = &r100_copy_blit, .copy_dma = &r300_copy_dma, .copy = &r300_copy_dma, + .get_engine_clock = &radeon_atom_get_engine_clock, .set_engine_clock = &radeon_atom_set_engine_clock, + .get_memory_clock = &radeon_atom_get_memory_clock, .set_memory_clock = &radeon_atom_set_memory_clock, .set_pcie_lanes = NULL, .set_clock_gating = &radeon_atom_set_clock_gating, @@ -420,41 +341,29 @@ static struct radeon_asic rs690_asic = { * rv515 */ int rv515_init(struct radeon_device *rdev); -void rv515_errata(struct radeon_device *rdev); -void rv515_vram_info(struct radeon_device *rdev); +void rv515_fini(struct radeon_device *rdev); int rv515_gpu_reset(struct radeon_device *rdev); -int rv515_mc_init(struct radeon_device *rdev); -void rv515_mc_fini(struct radeon_device *rdev); uint32_t rv515_mc_rreg(struct radeon_device *rdev, uint32_t reg); void rv515_mc_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v); void rv515_ring_start(struct radeon_device *rdev); uint32_t rv515_pcie_rreg(struct radeon_device *rdev, uint32_t reg); void rv515_pcie_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v); void rv515_bandwidth_update(struct radeon_device *rdev); +int rv515_resume(struct radeon_device *rdev); +int rv515_suspend(struct radeon_device *rdev); static struct radeon_asic rv515_asic = { .init = &rv515_init, - .errata = &rv515_errata, - .vram_info = &rv515_vram_info, + .fini = &rv515_fini, + .suspend = &rv515_suspend, + .resume = &rv515_resume, .vga_set_state = &r100_vga_set_state, .gpu_reset = &rv515_gpu_reset, - .mc_init = &rv515_mc_init, - .mc_fini = &rv515_mc_fini, - .wb_init = &r100_wb_init, - .wb_fini = &r100_wb_fini, - .gart_init = &rv370_pcie_gart_init, - .gart_fini = &rv370_pcie_gart_fini, - .gart_enable = &rv370_pcie_gart_enable, - .gart_disable = &rv370_pcie_gart_disable, .gart_tlb_flush = &rv370_pcie_gart_tlb_flush, .gart_set_page = &rv370_pcie_gart_set_page, - .cp_init = &r100_cp_init, - .cp_fini = &r100_cp_fini, - .cp_disable = &r100_cp_disable, .cp_commit = &r100_cp_commit, .ring_start = &rv515_ring_start, .ring_test = &r100_ring_test, .ring_ib_execute = &r100_ring_ib_execute, - .ib_test = &r100_ib_test, .irq_set = &rs600_irq_set, .irq_process = &rs600_irq_process, .get_vblank_counter = &rs600_get_vblank_counter, @@ -463,7 +372,9 @@ static struct radeon_asic rv515_asic = { .copy_blit = &r100_copy_blit, .copy_dma = &r300_copy_dma, .copy = &r100_copy_blit, + .get_engine_clock = &radeon_atom_get_engine_clock, .set_engine_clock = &radeon_atom_set_engine_clock, + .get_memory_clock = &radeon_atom_get_memory_clock, .set_memory_clock = &radeon_atom_set_memory_clock, .set_pcie_lanes = &rv370_set_pcie_lanes, .set_clock_gating = &radeon_atom_set_clock_gating, @@ -476,35 +387,21 @@ static struct radeon_asic rv515_asic = { /* * r520,rv530,rv560,rv570,r580 */ -void r520_errata(struct radeon_device *rdev); -void r520_vram_info(struct radeon_device *rdev); -int r520_mc_init(struct radeon_device *rdev); -void r520_mc_fini(struct radeon_device *rdev); -void r520_bandwidth_update(struct radeon_device *rdev); +int r520_init(struct radeon_device *rdev); +int r520_resume(struct radeon_device *rdev); static struct radeon_asic r520_asic = { - .init = &rv515_init, - .errata = &r520_errata, - .vram_info = &r520_vram_info, + .init = &r520_init, + .fini = &rv515_fini, + .suspend = &rv515_suspend, + .resume = &r520_resume, .vga_set_state = &r100_vga_set_state, .gpu_reset = &rv515_gpu_reset, - .mc_init = &r520_mc_init, - .mc_fini = &r520_mc_fini, - .wb_init = &r100_wb_init, - .wb_fini = &r100_wb_fini, - .gart_init = &rv370_pcie_gart_init, - .gart_fini = &rv370_pcie_gart_fini, - .gart_enable = &rv370_pcie_gart_enable, - .gart_disable = &rv370_pcie_gart_disable, .gart_tlb_flush = &rv370_pcie_gart_tlb_flush, .gart_set_page = &rv370_pcie_gart_set_page, - .cp_init = &r100_cp_init, - .cp_fini = &r100_cp_fini, - .cp_disable = &r100_cp_disable, .cp_commit = &r100_cp_commit, .ring_start = &rv515_ring_start, .ring_test = &r100_ring_test, .ring_ib_execute = &r100_ring_ib_execute, - .ib_test = &r100_ib_test, .irq_set = &rs600_irq_set, .irq_process = &rs600_irq_process, .get_vblank_counter = &rs600_get_vblank_counter, @@ -513,13 +410,15 @@ static struct radeon_asic r520_asic = { .copy_blit = &r100_copy_blit, .copy_dma = &r300_copy_dma, .copy = &r100_copy_blit, + .get_engine_clock = &radeon_atom_get_engine_clock, .set_engine_clock = &radeon_atom_set_engine_clock, + .get_memory_clock = &radeon_atom_get_memory_clock, .set_memory_clock = &radeon_atom_set_memory_clock, .set_pcie_lanes = &rv370_set_pcie_lanes, .set_clock_gating = &radeon_atom_set_clock_gating, .set_surface_reg = r100_set_surface_reg, .clear_surface_reg = r100_clear_surface_reg, - .bandwidth_update = &r520_bandwidth_update, + .bandwidth_update = &rv515_bandwidth_update, }; /* @@ -552,37 +451,23 @@ int r600_set_surface_reg(struct radeon_device *rdev, int reg, uint32_t offset, uint32_t obj_size); int r600_clear_surface_reg(struct radeon_device *rdev, int reg); void r600_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib); -int r600_ib_test(struct radeon_device *rdev); int r600_ring_test(struct radeon_device *rdev); int r600_copy_blit(struct radeon_device *rdev, uint64_t src_offset, uint64_t dst_offset, unsigned num_pages, struct radeon_fence *fence); static struct radeon_asic r600_asic = { - .errata = NULL, .init = &r600_init, .fini = &r600_fini, .suspend = &r600_suspend, .resume = &r600_resume, .cp_commit = &r600_cp_commit, - .vram_info = NULL, .vga_set_state = &r600_vga_set_state, .gpu_reset = &r600_gpu_reset, - .mc_init = NULL, - .mc_fini = NULL, - .wb_init = &r600_wb_init, - .wb_fini = &r600_wb_fini, - .gart_enable = NULL, - .gart_disable = NULL, .gart_tlb_flush = &r600_pcie_gart_tlb_flush, .gart_set_page = &rs600_gart_set_page, - .cp_init = NULL, - .cp_fini = NULL, - .cp_disable = NULL, - .ring_start = NULL, .ring_test = &r600_ring_test, .ring_ib_execute = &r600_ring_ib_execute, - .ib_test = &r600_ib_test, .irq_set = &r600_irq_set, .irq_process = &r600_irq_process, .fence_ring_emit = &r600_fence_ring_emit, @@ -590,13 +475,15 @@ static struct radeon_asic r600_asic = { .copy_blit = &r600_copy_blit, .copy_dma = &r600_copy_blit, .copy = &r600_copy_blit, + .get_engine_clock = &radeon_atom_get_engine_clock, .set_engine_clock = &radeon_atom_set_engine_clock, + .get_memory_clock = &radeon_atom_get_memory_clock, .set_memory_clock = &radeon_atom_set_memory_clock, .set_pcie_lanes = NULL, .set_clock_gating = &radeon_atom_set_clock_gating, .set_surface_reg = r600_set_surface_reg, .clear_surface_reg = r600_clear_surface_reg, - .bandwidth_update = &r520_bandwidth_update, + .bandwidth_update = &rv515_bandwidth_update, }; /* @@ -609,30 +496,17 @@ int rv770_resume(struct radeon_device *rdev); int rv770_gpu_reset(struct radeon_device *rdev); static struct radeon_asic rv770_asic = { - .errata = NULL, .init = &rv770_init, .fini = &rv770_fini, .suspend = &rv770_suspend, .resume = &rv770_resume, .cp_commit = &r600_cp_commit, - .vram_info = NULL, .gpu_reset = &rv770_gpu_reset, .vga_set_state = &r600_vga_set_state, - .mc_init = NULL, - .mc_fini = NULL, - .wb_init = &r600_wb_init, - .wb_fini = &r600_wb_fini, - .gart_enable = NULL, - .gart_disable = NULL, .gart_tlb_flush = &r600_pcie_gart_tlb_flush, .gart_set_page = &rs600_gart_set_page, - .cp_init = NULL, - .cp_fini = NULL, - .cp_disable = NULL, - .ring_start = NULL, .ring_test = &r600_ring_test, .ring_ib_execute = &r600_ring_ib_execute, - .ib_test = &r600_ib_test, .irq_set = &r600_irq_set, .irq_process = &r600_irq_process, .fence_ring_emit = &r600_fence_ring_emit, @@ -640,13 +514,15 @@ static struct radeon_asic rv770_asic = { .copy_blit = &r600_copy_blit, .copy_dma = &r600_copy_blit, .copy = &r600_copy_blit, + .get_engine_clock = &radeon_atom_get_engine_clock, .set_engine_clock = &radeon_atom_set_engine_clock, + .get_memory_clock = &radeon_atom_get_memory_clock, .set_memory_clock = &radeon_atom_set_memory_clock, .set_pcie_lanes = NULL, .set_clock_gating = &radeon_atom_set_clock_gating, .set_surface_reg = r600_set_surface_reg, .clear_surface_reg = r600_clear_surface_reg, - .bandwidth_update = &r520_bandwidth_update, + .bandwidth_update = &rv515_bandwidth_update, }; #endif diff --git a/drivers/gpu/drm/radeon/radeon_atombios.c b/drivers/gpu/drm/radeon/radeon_atombios.c index 7437421..2ed88a8 100644 --- a/drivers/gpu/drm/radeon/radeon_atombios.c +++ b/drivers/gpu/drm/radeon/radeon_atombios.c @@ -46,7 +46,8 @@ radeon_add_atom_connector(struct drm_device *dev, uint32_t supported_device, int connector_type, struct radeon_i2c_bus_rec *i2c_bus, - bool linkb, uint32_t igp_lane_info); + bool linkb, uint32_t igp_lane_info, + uint16_t connector_object_id); /* from radeon_legacy_encoder.c */ extern void @@ -193,6 +194,23 @@ const int supported_devices_connector_convert[] = { DRM_MODE_CONNECTOR_DisplayPort }; +const uint16_t supported_devices_connector_object_id_convert[] = { + CONNECTOR_OBJECT_ID_NONE, + CONNECTOR_OBJECT_ID_VGA, + CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I, /* not all boards support DL */ + CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D, /* not all boards support DL */ + CONNECTOR_OBJECT_ID_VGA, /* technically DVI-A */ + CONNECTOR_OBJECT_ID_COMPOSITE, + CONNECTOR_OBJECT_ID_SVIDEO, + CONNECTOR_OBJECT_ID_LVDS, + CONNECTOR_OBJECT_ID_9PIN_DIN, + CONNECTOR_OBJECT_ID_9PIN_DIN, + CONNECTOR_OBJECT_ID_DISPLAYPORT, + CONNECTOR_OBJECT_ID_HDMI_TYPE_A, + CONNECTOR_OBJECT_ID_HDMI_TYPE_B, + CONNECTOR_OBJECT_ID_SVIDEO +}; + const int object_connector_convert[] = { DRM_MODE_CONNECTOR_Unknown, DRM_MODE_CONNECTOR_DVII, @@ -229,7 +247,7 @@ bool radeon_get_atom_connector_info_from_object_table(struct drm_device *dev) ATOM_OBJECT_HEADER *obj_header; int i, j, path_size, device_support; int connector_type; - uint16_t igp_lane_info, conn_id; + uint16_t igp_lane_info, conn_id, connector_object_id; bool linkb; struct radeon_i2c_bus_rec ddc_bus; @@ -272,15 +290,13 @@ bool radeon_get_atom_connector_info_from_object_table(struct drm_device *dev) (le16_to_cpu(path->usConnObjectId) & OBJECT_TYPE_MASK) >> OBJECT_TYPE_SHIFT; - if ((le16_to_cpu(path->usDeviceTag) == - ATOM_DEVICE_TV1_SUPPORT) - || (le16_to_cpu(path->usDeviceTag) == - ATOM_DEVICE_TV2_SUPPORT) - || (le16_to_cpu(path->usDeviceTag) == - ATOM_DEVICE_CV_SUPPORT)) + /* TODO CV support */ + if (le16_to_cpu(path->usDeviceTag) == + ATOM_DEVICE_CV_SUPPORT) continue; - if ((rdev->family == CHIP_RS780) && + /* IGP chips */ + if ((rdev->flags & RADEON_IS_IGP) && (con_obj_id == CONNECTOR_OBJECT_ID_PCIE_CONNECTOR)) { uint16_t igp_offset = 0; @@ -314,6 +330,7 @@ bool radeon_get_atom_connector_info_from_object_table(struct drm_device *dev) connector_type = object_connector_convert [ct]; + connector_object_id = ct; igp_lane_info = slot_config & 0xffff; } else @@ -324,6 +341,7 @@ bool radeon_get_atom_connector_info_from_object_table(struct drm_device *dev) igp_lane_info = 0; connector_type = object_connector_convert[con_obj_id]; + connector_object_id = con_obj_id; } if (connector_type == DRM_MODE_CONNECTOR_Unknown) @@ -428,7 +446,8 @@ bool radeon_get_atom_connector_info_from_object_table(struct drm_device *dev) le16_to_cpu(path-> usDeviceTag), connector_type, &ddc_bus, - linkb, igp_lane_info); + linkb, igp_lane_info, + connector_object_id); } } @@ -438,6 +457,45 @@ bool radeon_get_atom_connector_info_from_object_table(struct drm_device *dev) return true; } +static uint16_t atombios_get_connector_object_id(struct drm_device *dev, + int connector_type, + uint16_t devices) +{ + struct radeon_device *rdev = dev->dev_private; + + if (rdev->flags & RADEON_IS_IGP) { + return supported_devices_connector_object_id_convert + [connector_type]; + } else if (((connector_type == DRM_MODE_CONNECTOR_DVII) || + (connector_type == DRM_MODE_CONNECTOR_DVID)) && + (devices & ATOM_DEVICE_DFP2_SUPPORT)) { + struct radeon_mode_info *mode_info = &rdev->mode_info; + struct atom_context *ctx = mode_info->atom_context; + int index = GetIndexIntoMasterTable(DATA, XTMDS_Info); + uint16_t size, data_offset; + uint8_t frev, crev; + ATOM_XTMDS_INFO *xtmds; + + atom_parse_data_header(ctx, index, &size, &frev, &crev, &data_offset); + xtmds = (ATOM_XTMDS_INFO *)(ctx->bios + data_offset); + + if (xtmds->ucSupportedLink & ATOM_XTMDS_SUPPORTED_DUALLINK) { + if (connector_type == DRM_MODE_CONNECTOR_DVII) + return CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I; + else + return CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D; + } else { + if (connector_type == DRM_MODE_CONNECTOR_DVII) + return CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I; + else + return CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D; + } + } else { + return supported_devices_connector_object_id_convert + [connector_type]; + } +} + struct bios_connector { bool valid; uint16_t line_mux; @@ -596,14 +654,20 @@ bool radeon_get_atom_connector_info_from_supported_devices_table(struct /* add the connectors */ for (i = 0; i < ATOM_MAX_SUPPORTED_DEVICE; i++) { - if (bios_connectors[i].valid) + if (bios_connectors[i].valid) { + uint16_t connector_object_id = + atombios_get_connector_object_id(dev, + bios_connectors[i].connector_type, + bios_connectors[i].devices); radeon_add_atom_connector(dev, bios_connectors[i].line_mux, bios_connectors[i].devices, bios_connectors[i]. connector_type, &bios_connectors[i].ddc_bus, - false, 0); + false, 0, + connector_object_id); + } } radeon_link_encoder_connector(dev); @@ -644,8 +708,12 @@ bool radeon_atom_get_clock_info(struct drm_device *dev) le16_to_cpu(firmware_info->info.usReferenceClock); p1pll->reference_div = 0; - p1pll->pll_out_min = - le16_to_cpu(firmware_info->info.usMinPixelClockPLL_Output); + if (crev < 2) + p1pll->pll_out_min = + le16_to_cpu(firmware_info->info.usMinPixelClockPLL_Output); + else + p1pll->pll_out_min = + le32_to_cpu(firmware_info->info_12.ulMinPixelClockPLL_Output); p1pll->pll_out_max = le32_to_cpu(firmware_info->info.ulMaxPixelClockPLL_Output); @@ -654,6 +722,16 @@ bool radeon_atom_get_clock_info(struct drm_device *dev) p1pll->pll_out_min = 64800; else p1pll->pll_out_min = 20000; + } else if (p1pll->pll_out_min > 64800) { + /* Limiting the pll output range is a good thing generally as + * it limits the number of possible pll combinations for a given + * frequency presumably to the ones that work best on each card. + * However, certain duallink DVI monitors seem to like + * pll combinations that would be limited by this at least on + * pre-DCE 3.0 r6xx hardware. This might need to be adjusted per + * family. + */ + p1pll->pll_out_min = 64800; } p1pll->pll_in_min = @@ -770,6 +848,46 @@ bool radeon_atombios_get_tmds_info(struct radeon_encoder *encoder, return false; } +static struct radeon_atom_ss *radeon_atombios_get_ss_info(struct + radeon_encoder + *encoder, + int id) +{ + struct drm_device *dev = encoder->base.dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_mode_info *mode_info = &rdev->mode_info; + int index = GetIndexIntoMasterTable(DATA, PPLL_SS_Info); + uint16_t data_offset; + struct _ATOM_SPREAD_SPECTRUM_INFO *ss_info; + uint8_t frev, crev; + struct radeon_atom_ss *ss = NULL; + + if (id > ATOM_MAX_SS_ENTRY) + return NULL; + + atom_parse_data_header(mode_info->atom_context, index, NULL, &frev, + &crev, &data_offset); + + ss_info = + (struct _ATOM_SPREAD_SPECTRUM_INFO *)(mode_info->atom_context->bios + data_offset); + + if (ss_info) { + ss = + kzalloc(sizeof(struct radeon_atom_ss), GFP_KERNEL); + + if (!ss) + return NULL; + + ss->percentage = le16_to_cpu(ss_info->asSS_Info[id].usSpreadSpectrumPercentage); + ss->type = ss_info->asSS_Info[id].ucSpreadSpectrumType; + ss->step = ss_info->asSS_Info[id].ucSS_Step; + ss->delay = ss_info->asSS_Info[id].ucSS_Delay; + ss->range = ss_info->asSS_Info[id].ucSS_Range; + ss->refdiv = ss_info->asSS_Info[id].ucRecommendedRef_Div; + } + return ss; +} + union lvds_info { struct _ATOM_LVDS_INFO info; struct _ATOM_LVDS_INFO_V12 info_12; @@ -801,27 +919,31 @@ struct radeon_encoder_atom_dig *radeon_atombios_get_lvds_info(struct if (!lvds) return NULL; - lvds->native_mode.dotclock = + lvds->native_mode.clock = le16_to_cpu(lvds_info->info.sLCDTiming.usPixClk) * 10; - lvds->native_mode.panel_xres = + lvds->native_mode.hdisplay = le16_to_cpu(lvds_info->info.sLCDTiming.usHActive); - lvds->native_mode.panel_yres = + lvds->native_mode.vdisplay = le16_to_cpu(lvds_info->info.sLCDTiming.usVActive); - lvds->native_mode.hblank = - le16_to_cpu(lvds_info->info.sLCDTiming.usHBlanking_Time); - lvds->native_mode.hoverplus = - le16_to_cpu(lvds_info->info.sLCDTiming.usHSyncOffset); - lvds->native_mode.hsync_width = - le16_to_cpu(lvds_info->info.sLCDTiming.usHSyncWidth); - lvds->native_mode.vblank = - le16_to_cpu(lvds_info->info.sLCDTiming.usVBlanking_Time); - lvds->native_mode.voverplus = - le16_to_cpu(lvds_info->info.sLCDTiming.usVSyncOffset); - lvds->native_mode.vsync_width = - le16_to_cpu(lvds_info->info.sLCDTiming.usVSyncWidth); + lvds->native_mode.htotal = lvds->native_mode.hdisplay + + le16_to_cpu(lvds_info->info.sLCDTiming.usHBlanking_Time); + lvds->native_mode.hsync_start = lvds->native_mode.hdisplay + + le16_to_cpu(lvds_info->info.sLCDTiming.usHSyncOffset); + lvds->native_mode.hsync_end = lvds->native_mode.hsync_start + + le16_to_cpu(lvds_info->info.sLCDTiming.usHSyncWidth); + lvds->native_mode.vtotal = lvds->native_mode.vdisplay + + le16_to_cpu(lvds_info->info.sLCDTiming.usVBlanking_Time); + lvds->native_mode.vsync_start = lvds->native_mode.vdisplay + + le16_to_cpu(lvds_info->info.sLCDTiming.usVSyncWidth); + lvds->native_mode.vsync_end = lvds->native_mode.vsync_start + + le16_to_cpu(lvds_info->info.sLCDTiming.usVSyncWidth); lvds->panel_pwr_delay = le16_to_cpu(lvds_info->info.usOffDelayInMs); lvds->lvds_misc = lvds_info->info.ucLVDS_Misc; + /* set crtc values */ + drm_mode_set_crtcinfo(&lvds->native_mode, CRTC_INTERLACE_HALVE_V); + + lvds->ss = radeon_atombios_get_ss_info(encoder, lvds_info->info.ucSS_Id); encoder->native_mode = lvds->native_mode; } @@ -860,8 +982,7 @@ radeon_atombios_get_primary_dac_info(struct radeon_encoder *encoder) } bool radeon_atom_get_tv_timings(struct radeon_device *rdev, int index, - SET_CRTC_TIMING_PARAMETERS_PS_ALLOCATION *crtc_timing, - int32_t *pixel_clock) + struct drm_display_mode *mode) { struct radeon_mode_info *mode_info = &rdev->mode_info; ATOM_ANALOG_TV_INFO *tv_info; @@ -869,7 +990,7 @@ bool radeon_atom_get_tv_timings(struct radeon_device *rdev, int index, ATOM_DTD_FORMAT *dtd_timings; int data_index = GetIndexIntoMasterTable(DATA, AnalogTV_Info); u8 frev, crev; - uint16_t data_offset; + u16 data_offset, misc; atom_parse_data_header(mode_info->atom_context, data_index, NULL, &frev, &crev, &data_offset); @@ -879,28 +1000,37 @@ bool radeon_atom_get_tv_timings(struct radeon_device *rdev, int index, if (index > MAX_SUPPORTED_TV_TIMING) return false; - crtc_timing->usH_Total = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_H_Total); - crtc_timing->usH_Disp = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_H_Disp); - crtc_timing->usH_SyncStart = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_H_SyncStart); - crtc_timing->usH_SyncWidth = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_H_SyncWidth); - - crtc_timing->usV_Total = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_V_Total); - crtc_timing->usV_Disp = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_V_Disp); - crtc_timing->usV_SyncStart = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_V_SyncStart); - crtc_timing->usV_SyncWidth = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_V_SyncWidth); - - crtc_timing->susModeMiscInfo = tv_info->aModeTimings[index].susModeMiscInfo; - - crtc_timing->ucOverscanRight = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_OverscanRight); - crtc_timing->ucOverscanLeft = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_OverscanLeft); - crtc_timing->ucOverscanBottom = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_OverscanBottom); - crtc_timing->ucOverscanTop = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_OverscanTop); - *pixel_clock = le16_to_cpu(tv_info->aModeTimings[index].usPixelClock) * 10; + mode->crtc_htotal = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_H_Total); + mode->crtc_hdisplay = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_H_Disp); + mode->crtc_hsync_start = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_H_SyncStart); + mode->crtc_hsync_end = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_H_SyncStart) + + le16_to_cpu(tv_info->aModeTimings[index].usCRTC_H_SyncWidth); + + mode->crtc_vtotal = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_V_Total); + mode->crtc_vdisplay = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_V_Disp); + mode->crtc_vsync_start = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_V_SyncStart); + mode->crtc_vsync_end = le16_to_cpu(tv_info->aModeTimings[index].usCRTC_V_SyncStart) + + le16_to_cpu(tv_info->aModeTimings[index].usCRTC_V_SyncWidth); + + mode->flags = 0; + misc = le16_to_cpu(tv_info->aModeTimings[index].susModeMiscInfo.usAccess); + if (misc & ATOM_VSYNC_POLARITY) + mode->flags |= DRM_MODE_FLAG_NVSYNC; + if (misc & ATOM_HSYNC_POLARITY) + mode->flags |= DRM_MODE_FLAG_NHSYNC; + if (misc & ATOM_COMPOSITESYNC) + mode->flags |= DRM_MODE_FLAG_CSYNC; + if (misc & ATOM_INTERLACE) + mode->flags |= DRM_MODE_FLAG_INTERLACE; + if (misc & ATOM_DOUBLE_CLOCK_MODE) + mode->flags |= DRM_MODE_FLAG_DBLSCAN; + + mode->clock = le16_to_cpu(tv_info->aModeTimings[index].usPixelClock) * 10; if (index == 1) { /* PAL timings appear to have wrong values for totals */ - crtc_timing->usH_Total -= 1; - crtc_timing->usV_Total -= 1; + mode->crtc_htotal -= 1; + mode->crtc_vtotal -= 1; } break; case 2: @@ -909,17 +1039,36 @@ bool radeon_atom_get_tv_timings(struct radeon_device *rdev, int index, return false; dtd_timings = &tv_info_v1_2->aModeTimings[index]; - crtc_timing->usH_Total = le16_to_cpu(dtd_timings->usHActive) + le16_to_cpu(dtd_timings->usHBlanking_Time); - crtc_timing->usH_Disp = le16_to_cpu(dtd_timings->usHActive); - crtc_timing->usH_SyncStart = le16_to_cpu(dtd_timings->usHActive) + le16_to_cpu(dtd_timings->usHSyncOffset); - crtc_timing->usH_SyncWidth = le16_to_cpu(dtd_timings->usHSyncWidth); - crtc_timing->usV_Total = le16_to_cpu(dtd_timings->usVActive) + le16_to_cpu(dtd_timings->usVBlanking_Time); - crtc_timing->usV_Disp = le16_to_cpu(dtd_timings->usVActive); - crtc_timing->usV_SyncStart = le16_to_cpu(dtd_timings->usVActive) + le16_to_cpu(dtd_timings->usVSyncOffset); - crtc_timing->usV_SyncWidth = le16_to_cpu(dtd_timings->usVSyncWidth); - - crtc_timing->susModeMiscInfo.usAccess = le16_to_cpu(dtd_timings->susModeMiscInfo.usAccess); - *pixel_clock = le16_to_cpu(dtd_timings->usPixClk) * 10; + mode->crtc_htotal = le16_to_cpu(dtd_timings->usHActive) + + le16_to_cpu(dtd_timings->usHBlanking_Time); + mode->crtc_hdisplay = le16_to_cpu(dtd_timings->usHActive); + mode->crtc_hsync_start = le16_to_cpu(dtd_timings->usHActive) + + le16_to_cpu(dtd_timings->usHSyncOffset); + mode->crtc_hsync_end = mode->crtc_hsync_start + + le16_to_cpu(dtd_timings->usHSyncWidth); + + mode->crtc_vtotal = le16_to_cpu(dtd_timings->usVActive) + + le16_to_cpu(dtd_timings->usVBlanking_Time); + mode->crtc_vdisplay = le16_to_cpu(dtd_timings->usVActive); + mode->crtc_vsync_start = le16_to_cpu(dtd_timings->usVActive) + + le16_to_cpu(dtd_timings->usVSyncOffset); + mode->crtc_vsync_end = mode->crtc_vsync_start + + le16_to_cpu(dtd_timings->usVSyncWidth); + + mode->flags = 0; + misc = le16_to_cpu(dtd_timings->susModeMiscInfo.usAccess); + if (misc & ATOM_VSYNC_POLARITY) + mode->flags |= DRM_MODE_FLAG_NVSYNC; + if (misc & ATOM_HSYNC_POLARITY) + mode->flags |= DRM_MODE_FLAG_NHSYNC; + if (misc & ATOM_COMPOSITESYNC) + mode->flags |= DRM_MODE_FLAG_CSYNC; + if (misc & ATOM_INTERLACE) + mode->flags |= DRM_MODE_FLAG_INTERLACE; + if (misc & ATOM_DOUBLE_CLOCK_MODE) + mode->flags |= DRM_MODE_FLAG_DBLSCAN; + + mode->clock = le16_to_cpu(dtd_timings->usPixClk) * 10; break; } return true; @@ -984,6 +1133,24 @@ void radeon_atom_static_pwrmgt_setup(struct radeon_device *rdev, int enable) atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); } +uint32_t radeon_atom_get_engine_clock(struct radeon_device *rdev) +{ + GET_ENGINE_CLOCK_PS_ALLOCATION args; + int index = GetIndexIntoMasterTable(COMMAND, GetEngineClock); + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + return args.ulReturnEngineClock; +} + +uint32_t radeon_atom_get_memory_clock(struct radeon_device *rdev) +{ + GET_MEMORY_CLOCK_PS_ALLOCATION args; + int index = GetIndexIntoMasterTable(COMMAND, GetMemoryClock); + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + return args.ulReturnMemoryClock; +} + void radeon_atom_set_engine_clock(struct radeon_device *rdev, uint32_t eng_clock) { diff --git a/drivers/gpu/drm/radeon/radeon_benchmark.c b/drivers/gpu/drm/radeon/radeon_benchmark.c index 2e938f7..10bd50a 100644 --- a/drivers/gpu/drm/radeon/radeon_benchmark.c +++ b/drivers/gpu/drm/radeon/radeon_benchmark.c @@ -63,7 +63,7 @@ void radeon_benchmark_move(struct radeon_device *rdev, unsigned bsize, if (r) { goto out_cleanup; } - r = radeon_copy_dma(rdev, saddr, daddr, size / 4096, fence); + r = radeon_copy_dma(rdev, saddr, daddr, size / RADEON_GPU_PAGE_SIZE, fence); if (r) { goto out_cleanup; } @@ -88,7 +88,7 @@ void radeon_benchmark_move(struct radeon_device *rdev, unsigned bsize, if (r) { goto out_cleanup; } - r = radeon_copy_blit(rdev, saddr, daddr, size / 4096, fence); + r = radeon_copy_blit(rdev, saddr, daddr, size / RADEON_GPU_PAGE_SIZE, fence); if (r) { goto out_cleanup; } diff --git a/drivers/gpu/drm/radeon/radeon_bios.c b/drivers/gpu/drm/radeon/radeon_bios.c index 96e37a6..9069217 100644 --- a/drivers/gpu/drm/radeon/radeon_bios.c +++ b/drivers/gpu/drm/radeon/radeon_bios.c @@ -33,12 +33,47 @@ /* * BIOS. */ + +/* If you boot an IGP board with a discrete card as the primary, + * the IGP rom is not accessible via the rom bar as the IGP rom is + * part of the system bios. On boot, the system bios puts a + * copy of the igp rom at the start of vram if a discrete card is + * present. + */ +static bool igp_read_bios_from_vram(struct radeon_device *rdev) +{ + uint8_t __iomem *bios; + resource_size_t vram_base; + resource_size_t size = 256 * 1024; /* ??? */ + + rdev->bios = NULL; + vram_base = drm_get_resource_start(rdev->ddev, 0); + bios = ioremap(vram_base, size); + if (!bios) { + return false; + } + + if (size == 0 || bios[0] != 0x55 || bios[1] != 0xaa) { + iounmap(bios); + return false; + } + rdev->bios = kmalloc(size, GFP_KERNEL); + if (rdev->bios == NULL) { + iounmap(bios); + return false; + } + memcpy(rdev->bios, bios, size); + iounmap(bios); + return true; +} + static bool radeon_read_bios(struct radeon_device *rdev) { uint8_t __iomem *bios; size_t size; rdev->bios = NULL; + /* XXX: some cards may return 0 for rom size? ddx has a workaround */ bios = pci_map_rom(rdev->pdev, &size); if (!bios) { return false; @@ -341,7 +376,9 @@ static bool legacy_read_disabled_bios(struct radeon_device *rdev) static bool radeon_read_disabled_bios(struct radeon_device *rdev) { - if (rdev->family >= CHIP_RV770) + if (rdev->flags & RADEON_IS_IGP) + return igp_read_bios_from_vram(rdev); + else if (rdev->family >= CHIP_RV770) return r700_read_disabled_bios(rdev); else if (rdev->family >= CHIP_R600) return r600_read_disabled_bios(rdev); @@ -356,7 +393,12 @@ bool radeon_get_bios(struct radeon_device *rdev) bool r; uint16_t tmp; - r = radeon_read_bios(rdev); + if (rdev->flags & RADEON_IS_IGP) { + r = igp_read_bios_from_vram(rdev); + if (r == false) + r = radeon_read_bios(rdev); + } else + r = radeon_read_bios(rdev); if (r == false) { r = radeon_read_disabled_bios(rdev); } diff --git a/drivers/gpu/drm/radeon/radeon_clocks.c b/drivers/gpu/drm/radeon/radeon_clocks.c index 152eef1..a813541 100644 --- a/drivers/gpu/drm/radeon/radeon_clocks.c +++ b/drivers/gpu/drm/radeon/radeon_clocks.c @@ -32,7 +32,7 @@ #include "atom.h" /* 10 khz */ -static uint32_t radeon_legacy_get_engine_clock(struct radeon_device *rdev) +uint32_t radeon_legacy_get_engine_clock(struct radeon_device *rdev) { struct radeon_pll *spll = &rdev->clock.spll; uint32_t fb_div, ref_div, post_div, sclk; @@ -411,7 +411,7 @@ void radeon_legacy_set_clock_gating(struct radeon_device *rdev, int enable) R300_PIXCLK_TRANS_ALWAYS_ONb | R300_PIXCLK_TVO_ALWAYS_ONb | R300_P2G2CLK_ALWAYS_ONb | - R300_P2G2CLK_ALWAYS_ONb); + R300_P2G2CLK_DAC_ALWAYS_ONb); WREG32_PLL(RADEON_PIXCLKS_CNTL, tmp); } else if (rdev->family >= CHIP_RV350) { tmp = RREG32_PLL(R300_SCLK_CNTL2); @@ -464,7 +464,7 @@ void radeon_legacy_set_clock_gating(struct radeon_device *rdev, int enable) R300_PIXCLK_TRANS_ALWAYS_ONb | R300_PIXCLK_TVO_ALWAYS_ONb | R300_P2G2CLK_ALWAYS_ONb | - R300_P2G2CLK_ALWAYS_ONb); + R300_P2G2CLK_DAC_ALWAYS_ONb); WREG32_PLL(RADEON_PIXCLKS_CNTL, tmp); tmp = RREG32_PLL(RADEON_MCLK_MISC); @@ -654,7 +654,7 @@ void radeon_legacy_set_clock_gating(struct radeon_device *rdev, int enable) R300_PIXCLK_TRANS_ALWAYS_ONb | R300_PIXCLK_TVO_ALWAYS_ONb | R300_P2G2CLK_ALWAYS_ONb | - R300_P2G2CLK_ALWAYS_ONb | + R300_P2G2CLK_DAC_ALWAYS_ONb | R300_DISP_DAC_PIXCLK_DAC2_BLANK_OFF); WREG32_PLL(RADEON_PIXCLKS_CNTL, tmp); } else if (rdev->family >= CHIP_RV350) { @@ -705,7 +705,7 @@ void radeon_legacy_set_clock_gating(struct radeon_device *rdev, int enable) R300_PIXCLK_TRANS_ALWAYS_ONb | R300_PIXCLK_TVO_ALWAYS_ONb | R300_P2G2CLK_ALWAYS_ONb | - R300_P2G2CLK_ALWAYS_ONb | + R300_P2G2CLK_DAC_ALWAYS_ONb | R300_DISP_DAC_PIXCLK_DAC2_BLANK_OFF); WREG32_PLL(RADEON_PIXCLKS_CNTL, tmp); } else { diff --git a/drivers/gpu/drm/radeon/radeon_combios.c b/drivers/gpu/drm/radeon/radeon_combios.c index 748265a..5253cbf 100644 --- a/drivers/gpu/drm/radeon/radeon_combios.c +++ b/drivers/gpu/drm/radeon/radeon_combios.c @@ -49,7 +49,8 @@ radeon_add_legacy_connector(struct drm_device *dev, uint32_t connector_id, uint32_t supported_device, int connector_type, - struct radeon_i2c_bus_rec *i2c_bus); + struct radeon_i2c_bus_rec *i2c_bus, + uint16_t connector_object_id); /* from radeon_legacy_encoder.c */ extern void @@ -808,25 +809,25 @@ static struct radeon_encoder_lvds *radeon_legacy_get_lvds_info_from_regs(struct lvds->panel_blon_delay = (lvds_ss_gen_cntl >> RADEON_LVDS_PWRSEQ_DELAY2_SHIFT) & 0xf; if (fp_vert_stretch & RADEON_VERT_STRETCH_ENABLE) - lvds->native_mode.panel_yres = + lvds->native_mode.vdisplay = ((fp_vert_stretch & RADEON_VERT_PANEL_SIZE) >> RADEON_VERT_PANEL_SHIFT) + 1; else - lvds->native_mode.panel_yres = + lvds->native_mode.vdisplay = (RREG32(RADEON_CRTC_V_TOTAL_DISP) >> 16) + 1; if (fp_horz_stretch & RADEON_HORZ_STRETCH_ENABLE) - lvds->native_mode.panel_xres = + lvds->native_mode.hdisplay = (((fp_horz_stretch & RADEON_HORZ_PANEL_SIZE) >> RADEON_HORZ_PANEL_SHIFT) + 1) * 8; else - lvds->native_mode.panel_xres = + lvds->native_mode.hdisplay = ((RREG32(RADEON_CRTC_H_TOTAL_DISP) >> 16) + 1) * 8; - if ((lvds->native_mode.panel_xres < 640) || - (lvds->native_mode.panel_yres < 480)) { - lvds->native_mode.panel_xres = 640; - lvds->native_mode.panel_yres = 480; + if ((lvds->native_mode.hdisplay < 640) || + (lvds->native_mode.vdisplay < 480)) { + lvds->native_mode.hdisplay = 640; + lvds->native_mode.vdisplay = 480; } ppll_div_sel = RREG8(RADEON_CLOCK_CNTL_INDEX + 1) & 0x3; @@ -846,8 +847,8 @@ static struct radeon_encoder_lvds *radeon_legacy_get_lvds_info_from_regs(struct lvds->panel_vcc_delay = 200; DRM_INFO("Panel info derived from registers\n"); - DRM_INFO("Panel Size %dx%d\n", lvds->native_mode.panel_xres, - lvds->native_mode.panel_yres); + DRM_INFO("Panel Size %dx%d\n", lvds->native_mode.hdisplay, + lvds->native_mode.vdisplay); return lvds; } @@ -882,11 +883,11 @@ struct radeon_encoder_lvds *radeon_combios_get_lvds_info(struct radeon_encoder DRM_INFO("Panel ID String: %s\n", stmp); - lvds->native_mode.panel_xres = RBIOS16(lcd_info + 0x19); - lvds->native_mode.panel_yres = RBIOS16(lcd_info + 0x1b); + lvds->native_mode.hdisplay = RBIOS16(lcd_info + 0x19); + lvds->native_mode.vdisplay = RBIOS16(lcd_info + 0x1b); - DRM_INFO("Panel Size %dx%d\n", lvds->native_mode.panel_xres, - lvds->native_mode.panel_yres); + DRM_INFO("Panel Size %dx%d\n", lvds->native_mode.hdisplay, + lvds->native_mode.vdisplay); lvds->panel_vcc_delay = RBIOS16(lcd_info + 0x2c); if (lvds->panel_vcc_delay > 2000 || lvds->panel_vcc_delay < 0) @@ -944,27 +945,25 @@ struct radeon_encoder_lvds *radeon_combios_get_lvds_info(struct radeon_encoder if (tmp == 0) break; - if ((RBIOS16(tmp) == lvds->native_mode.panel_xres) && + if ((RBIOS16(tmp) == lvds->native_mode.hdisplay) && (RBIOS16(tmp + 2) == - lvds->native_mode.panel_yres)) { - lvds->native_mode.hblank = - (RBIOS16(tmp + 17) - RBIOS16(tmp + 19)) * 8; - lvds->native_mode.hoverplus = - (RBIOS16(tmp + 21) - RBIOS16(tmp + 19) - - 1) * 8; - lvds->native_mode.hsync_width = - RBIOS8(tmp + 23) * 8; - - lvds->native_mode.vblank = (RBIOS16(tmp + 24) - - RBIOS16(tmp + 26)); - lvds->native_mode.voverplus = - ((RBIOS16(tmp + 28) & 0x7ff) - - RBIOS16(tmp + 26)); - lvds->native_mode.vsync_width = - ((RBIOS16(tmp + 28) & 0xf800) >> 11); - lvds->native_mode.dotclock = - RBIOS16(tmp + 9) * 10; + lvds->native_mode.vdisplay)) { + lvds->native_mode.htotal = RBIOS16(tmp + 17) * 8; + lvds->native_mode.hsync_start = RBIOS16(tmp + 21) * 8; + lvds->native_mode.hsync_end = (RBIOS8(tmp + 23) + + RBIOS16(tmp + 21)) * 8; + + lvds->native_mode.vtotal = RBIOS16(tmp + 24); + lvds->native_mode.vsync_start = RBIOS16(tmp + 28) & 0x7ff; + lvds->native_mode.vsync_end = + ((RBIOS16(tmp + 28) & 0xf800) >> 11) + + (RBIOS16(tmp + 28) & 0x7ff); + + lvds->native_mode.clock = RBIOS16(tmp + 9) * 10; lvds->native_mode.flags = 0; + /* set crtc values */ + drm_mode_set_crtcinfo(&lvds->native_mode, CRTC_INTERLACE_HALVE_V); + } } } else { @@ -1178,7 +1177,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) radeon_add_legacy_connector(dev, 0, ATOM_DEVICE_CRT1_SUPPORT, DRM_MODE_CONNECTOR_VGA, - &ddc_i2c); + &ddc_i2c, + CONNECTOR_OBJECT_ID_VGA); } else if (rdev->flags & RADEON_IS_MOBILITY) { /* LVDS */ ddc_i2c = combios_setup_i2c_bus(RADEON_LCD_GPIO_MASK); @@ -1190,7 +1190,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) radeon_add_legacy_connector(dev, 0, ATOM_DEVICE_LCD1_SUPPORT, DRM_MODE_CONNECTOR_LVDS, - &ddc_i2c); + &ddc_i2c, + CONNECTOR_OBJECT_ID_LVDS); /* VGA - primary dac */ ddc_i2c = combios_setup_i2c_bus(RADEON_GPIO_VGA_DDC); @@ -1202,7 +1203,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) radeon_add_legacy_connector(dev, 1, ATOM_DEVICE_CRT1_SUPPORT, DRM_MODE_CONNECTOR_VGA, - &ddc_i2c); + &ddc_i2c, + CONNECTOR_OBJECT_ID_VGA); } else { /* DVI-I - tv dac, int tmds */ ddc_i2c = combios_setup_i2c_bus(RADEON_GPIO_DVI_DDC); @@ -1220,7 +1222,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) ATOM_DEVICE_DFP1_SUPPORT | ATOM_DEVICE_CRT2_SUPPORT, DRM_MODE_CONNECTOR_DVII, - &ddc_i2c); + &ddc_i2c, + CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I); /* VGA - primary dac */ ddc_i2c = combios_setup_i2c_bus(RADEON_GPIO_VGA_DDC); @@ -1232,7 +1235,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) radeon_add_legacy_connector(dev, 1, ATOM_DEVICE_CRT1_SUPPORT, DRM_MODE_CONNECTOR_VGA, - &ddc_i2c); + &ddc_i2c, + CONNECTOR_OBJECT_ID_VGA); } if (rdev->family != CHIP_R100 && rdev->family != CHIP_R200) { @@ -1245,7 +1249,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) radeon_add_legacy_connector(dev, 2, ATOM_DEVICE_TV1_SUPPORT, DRM_MODE_CONNECTOR_SVIDEO, - &ddc_i2c); + &ddc_i2c, + CONNECTOR_OBJECT_ID_SVIDEO); } break; case CT_IBOOK: @@ -1259,7 +1264,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) 0), ATOM_DEVICE_LCD1_SUPPORT); radeon_add_legacy_connector(dev, 0, ATOM_DEVICE_LCD1_SUPPORT, - DRM_MODE_CONNECTOR_LVDS, &ddc_i2c); + DRM_MODE_CONNECTOR_LVDS, &ddc_i2c, + CONNECTOR_OBJECT_ID_LVDS); /* VGA - TV DAC */ ddc_i2c = combios_setup_i2c_bus(RADEON_GPIO_VGA_DDC); radeon_add_legacy_encoder(dev, @@ -1268,7 +1274,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) 2), ATOM_DEVICE_CRT2_SUPPORT); radeon_add_legacy_connector(dev, 1, ATOM_DEVICE_CRT2_SUPPORT, - DRM_MODE_CONNECTOR_VGA, &ddc_i2c); + DRM_MODE_CONNECTOR_VGA, &ddc_i2c, + CONNECTOR_OBJECT_ID_VGA); /* TV - TV DAC */ radeon_add_legacy_encoder(dev, radeon_get_encoder_id(dev, @@ -1277,7 +1284,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) ATOM_DEVICE_TV1_SUPPORT); radeon_add_legacy_connector(dev, 2, ATOM_DEVICE_TV1_SUPPORT, DRM_MODE_CONNECTOR_SVIDEO, - &ddc_i2c); + &ddc_i2c, + CONNECTOR_OBJECT_ID_SVIDEO); break; case CT_POWERBOOK_EXTERNAL: DRM_INFO("Connector Table: %d (powerbook external tmds)\n", @@ -1290,7 +1298,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) 0), ATOM_DEVICE_LCD1_SUPPORT); radeon_add_legacy_connector(dev, 0, ATOM_DEVICE_LCD1_SUPPORT, - DRM_MODE_CONNECTOR_LVDS, &ddc_i2c); + DRM_MODE_CONNECTOR_LVDS, &ddc_i2c, + CONNECTOR_OBJECT_ID_LVDS); /* DVI-I - primary dac, ext tmds */ ddc_i2c = combios_setup_i2c_bus(RADEON_GPIO_VGA_DDC); radeon_add_legacy_encoder(dev, @@ -1303,10 +1312,12 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) ATOM_DEVICE_CRT1_SUPPORT, 1), ATOM_DEVICE_CRT1_SUPPORT); + /* XXX some are SL */ radeon_add_legacy_connector(dev, 1, ATOM_DEVICE_DFP2_SUPPORT | ATOM_DEVICE_CRT1_SUPPORT, - DRM_MODE_CONNECTOR_DVII, &ddc_i2c); + DRM_MODE_CONNECTOR_DVII, &ddc_i2c, + CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I); /* TV - TV DAC */ radeon_add_legacy_encoder(dev, radeon_get_encoder_id(dev, @@ -1315,7 +1326,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) ATOM_DEVICE_TV1_SUPPORT); radeon_add_legacy_connector(dev, 2, ATOM_DEVICE_TV1_SUPPORT, DRM_MODE_CONNECTOR_SVIDEO, - &ddc_i2c); + &ddc_i2c, + CONNECTOR_OBJECT_ID_SVIDEO); break; case CT_POWERBOOK_INTERNAL: DRM_INFO("Connector Table: %d (powerbook internal tmds)\n", @@ -1328,7 +1340,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) 0), ATOM_DEVICE_LCD1_SUPPORT); radeon_add_legacy_connector(dev, 0, ATOM_DEVICE_LCD1_SUPPORT, - DRM_MODE_CONNECTOR_LVDS, &ddc_i2c); + DRM_MODE_CONNECTOR_LVDS, &ddc_i2c, + CONNECTOR_OBJECT_ID_LVDS); /* DVI-I - primary dac, int tmds */ ddc_i2c = combios_setup_i2c_bus(RADEON_GPIO_VGA_DDC); radeon_add_legacy_encoder(dev, @@ -1344,7 +1357,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) radeon_add_legacy_connector(dev, 1, ATOM_DEVICE_DFP1_SUPPORT | ATOM_DEVICE_CRT1_SUPPORT, - DRM_MODE_CONNECTOR_DVII, &ddc_i2c); + DRM_MODE_CONNECTOR_DVII, &ddc_i2c, + CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I); /* TV - TV DAC */ radeon_add_legacy_encoder(dev, radeon_get_encoder_id(dev, @@ -1353,7 +1367,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) ATOM_DEVICE_TV1_SUPPORT); radeon_add_legacy_connector(dev, 2, ATOM_DEVICE_TV1_SUPPORT, DRM_MODE_CONNECTOR_SVIDEO, - &ddc_i2c); + &ddc_i2c, + CONNECTOR_OBJECT_ID_SVIDEO); break; case CT_POWERBOOK_VGA: DRM_INFO("Connector Table: %d (powerbook vga)\n", @@ -1366,7 +1381,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) 0), ATOM_DEVICE_LCD1_SUPPORT); radeon_add_legacy_connector(dev, 0, ATOM_DEVICE_LCD1_SUPPORT, - DRM_MODE_CONNECTOR_LVDS, &ddc_i2c); + DRM_MODE_CONNECTOR_LVDS, &ddc_i2c, + CONNECTOR_OBJECT_ID_LVDS); /* VGA - primary dac */ ddc_i2c = combios_setup_i2c_bus(RADEON_GPIO_VGA_DDC); radeon_add_legacy_encoder(dev, @@ -1375,7 +1391,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) 1), ATOM_DEVICE_CRT1_SUPPORT); radeon_add_legacy_connector(dev, 1, ATOM_DEVICE_CRT1_SUPPORT, - DRM_MODE_CONNECTOR_VGA, &ddc_i2c); + DRM_MODE_CONNECTOR_VGA, &ddc_i2c, + CONNECTOR_OBJECT_ID_VGA); /* TV - TV DAC */ radeon_add_legacy_encoder(dev, radeon_get_encoder_id(dev, @@ -1384,7 +1401,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) ATOM_DEVICE_TV1_SUPPORT); radeon_add_legacy_connector(dev, 2, ATOM_DEVICE_TV1_SUPPORT, DRM_MODE_CONNECTOR_SVIDEO, - &ddc_i2c); + &ddc_i2c, + CONNECTOR_OBJECT_ID_SVIDEO); break; case CT_MINI_EXTERNAL: DRM_INFO("Connector Table: %d (mini external tmds)\n", @@ -1401,10 +1419,12 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) ATOM_DEVICE_CRT2_SUPPORT, 2), ATOM_DEVICE_CRT2_SUPPORT); + /* XXX are any DL? */ radeon_add_legacy_connector(dev, 0, ATOM_DEVICE_DFP2_SUPPORT | ATOM_DEVICE_CRT2_SUPPORT, - DRM_MODE_CONNECTOR_DVII, &ddc_i2c); + DRM_MODE_CONNECTOR_DVII, &ddc_i2c, + CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I); /* TV - TV DAC */ radeon_add_legacy_encoder(dev, radeon_get_encoder_id(dev, @@ -1413,7 +1433,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) ATOM_DEVICE_TV1_SUPPORT); radeon_add_legacy_connector(dev, 1, ATOM_DEVICE_TV1_SUPPORT, DRM_MODE_CONNECTOR_SVIDEO, - &ddc_i2c); + &ddc_i2c, + CONNECTOR_OBJECT_ID_SVIDEO); break; case CT_MINI_INTERNAL: DRM_INFO("Connector Table: %d (mini internal tmds)\n", @@ -1433,7 +1454,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) radeon_add_legacy_connector(dev, 0, ATOM_DEVICE_DFP1_SUPPORT | ATOM_DEVICE_CRT2_SUPPORT, - DRM_MODE_CONNECTOR_DVII, &ddc_i2c); + DRM_MODE_CONNECTOR_DVII, &ddc_i2c, + CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I); /* TV - TV DAC */ radeon_add_legacy_encoder(dev, radeon_get_encoder_id(dev, @@ -1442,7 +1464,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) ATOM_DEVICE_TV1_SUPPORT); radeon_add_legacy_connector(dev, 1, ATOM_DEVICE_TV1_SUPPORT, DRM_MODE_CONNECTOR_SVIDEO, - &ddc_i2c); + &ddc_i2c, + CONNECTOR_OBJECT_ID_SVIDEO); break; case CT_IMAC_G5_ISIGHT: DRM_INFO("Connector Table: %d (imac g5 isight)\n", @@ -1455,7 +1478,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) 0), ATOM_DEVICE_DFP1_SUPPORT); radeon_add_legacy_connector(dev, 0, ATOM_DEVICE_DFP1_SUPPORT, - DRM_MODE_CONNECTOR_DVID, &ddc_i2c); + DRM_MODE_CONNECTOR_DVID, &ddc_i2c, + CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D); /* VGA - tv dac */ ddc_i2c = combios_setup_i2c_bus(RADEON_GPIO_DVI_DDC); radeon_add_legacy_encoder(dev, @@ -1464,7 +1488,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) 2), ATOM_DEVICE_CRT2_SUPPORT); radeon_add_legacy_connector(dev, 1, ATOM_DEVICE_CRT2_SUPPORT, - DRM_MODE_CONNECTOR_VGA, &ddc_i2c); + DRM_MODE_CONNECTOR_VGA, &ddc_i2c, + CONNECTOR_OBJECT_ID_VGA); /* TV - TV DAC */ radeon_add_legacy_encoder(dev, radeon_get_encoder_id(dev, @@ -1473,7 +1498,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) ATOM_DEVICE_TV1_SUPPORT); radeon_add_legacy_connector(dev, 2, ATOM_DEVICE_TV1_SUPPORT, DRM_MODE_CONNECTOR_SVIDEO, - &ddc_i2c); + &ddc_i2c, + CONNECTOR_OBJECT_ID_SVIDEO); break; case CT_EMAC: DRM_INFO("Connector Table: %d (emac)\n", @@ -1486,7 +1512,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) 1), ATOM_DEVICE_CRT1_SUPPORT); radeon_add_legacy_connector(dev, 0, ATOM_DEVICE_CRT1_SUPPORT, - DRM_MODE_CONNECTOR_VGA, &ddc_i2c); + DRM_MODE_CONNECTOR_VGA, &ddc_i2c, + CONNECTOR_OBJECT_ID_VGA); /* VGA - tv dac */ ddc_i2c = combios_setup_i2c_bus(RADEON_GPIO_CRT2_DDC); radeon_add_legacy_encoder(dev, @@ -1495,7 +1522,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) 2), ATOM_DEVICE_CRT2_SUPPORT); radeon_add_legacy_connector(dev, 1, ATOM_DEVICE_CRT2_SUPPORT, - DRM_MODE_CONNECTOR_VGA, &ddc_i2c); + DRM_MODE_CONNECTOR_VGA, &ddc_i2c, + CONNECTOR_OBJECT_ID_VGA); /* TV - TV DAC */ radeon_add_legacy_encoder(dev, radeon_get_encoder_id(dev, @@ -1504,7 +1532,8 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) ATOM_DEVICE_TV1_SUPPORT); radeon_add_legacy_connector(dev, 2, ATOM_DEVICE_TV1_SUPPORT, DRM_MODE_CONNECTOR_SVIDEO, - &ddc_i2c); + &ddc_i2c, + CONNECTOR_OBJECT_ID_SVIDEO); break; default: DRM_INFO("Connector table: %d (invalid)\n", @@ -1581,11 +1610,63 @@ static bool radeon_apply_legacy_quirks(struct drm_device *dev, return true; } +static bool radeon_apply_legacy_tv_quirks(struct drm_device *dev) +{ + /* Acer 5102 has non-existent TV port */ + if (dev->pdev->device == 0x5975 && + dev->pdev->subsystem_vendor == 0x1025 && + dev->pdev->subsystem_device == 0x009f) + return false; + + /* HP dc5750 has non-existent TV port */ + if (dev->pdev->device == 0x5974 && + dev->pdev->subsystem_vendor == 0x103c && + dev->pdev->subsystem_device == 0x280a) + return false; + + return true; +} + +static uint16_t combios_check_dl_dvi(struct drm_device *dev, int is_dvi_d) +{ + struct radeon_device *rdev = dev->dev_private; + uint32_t ext_tmds_info; + + if (rdev->flags & RADEON_IS_IGP) { + if (is_dvi_d) + return CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D; + else + return CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I; + } + ext_tmds_info = combios_get_table_offset(dev, COMBIOS_EXT_TMDS_INFO_TABLE); + if (ext_tmds_info) { + uint8_t rev = RBIOS8(ext_tmds_info); + uint8_t flags = RBIOS8(ext_tmds_info + 4 + 5); + if (rev >= 3) { + if (is_dvi_d) + return CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D; + else + return CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I; + } else { + if (flags & 1) { + if (is_dvi_d) + return CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D; + else + return CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I; + } + } + } + if (is_dvi_d) + return CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D; + else + return CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I; +} + bool radeon_get_legacy_connector_info_from_bios(struct drm_device *dev) { struct radeon_device *rdev = dev->dev_private; uint32_t conn_info, entry, devices; - uint16_t tmp; + uint16_t tmp, connector_object_id; enum radeon_combios_ddc ddc_type; enum radeon_combios_connector connector; int i = 0; @@ -1628,8 +1709,9 @@ bool radeon_get_legacy_connector_info_from_bios(struct drm_device *dev) break; } - radeon_apply_legacy_quirks(dev, i, &connector, - &ddc_i2c); + if (!radeon_apply_legacy_quirks(dev, i, &connector, + &ddc_i2c)) + continue; switch (connector) { case CONNECTOR_PROPRIETARY_LEGACY: @@ -1644,7 +1726,8 @@ bool radeon_get_legacy_connector_info_from_bios(struct drm_device *dev) radeon_add_legacy_connector(dev, i, devices, legacy_connector_convert [connector], - &ddc_i2c); + &ddc_i2c, + CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D); break; case CONNECTOR_CRT_LEGACY: if (tmp & 0x1) { @@ -1669,7 +1752,8 @@ bool radeon_get_legacy_connector_info_from_bios(struct drm_device *dev) devices, legacy_connector_convert [connector], - &ddc_i2c); + &ddc_i2c, + CONNECTOR_OBJECT_ID_VGA); break; case CONNECTOR_DVI_I_LEGACY: devices = 0; @@ -1698,6 +1782,7 @@ bool radeon_get_legacy_connector_info_from_bios(struct drm_device *dev) ATOM_DEVICE_DFP2_SUPPORT, 0), ATOM_DEVICE_DFP2_SUPPORT); + connector_object_id = combios_check_dl_dvi(dev, 0); } else { devices |= ATOM_DEVICE_DFP1_SUPPORT; radeon_add_legacy_encoder(dev, @@ -1706,19 +1791,24 @@ bool radeon_get_legacy_connector_info_from_bios(struct drm_device *dev) ATOM_DEVICE_DFP1_SUPPORT, 0), ATOM_DEVICE_DFP1_SUPPORT); + connector_object_id = CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I; } radeon_add_legacy_connector(dev, i, devices, legacy_connector_convert [connector], - &ddc_i2c); + &ddc_i2c, + connector_object_id); break; case CONNECTOR_DVI_D_LEGACY: - if ((tmp >> 4) & 0x1) + if ((tmp >> 4) & 0x1) { devices = ATOM_DEVICE_DFP2_SUPPORT; - else + connector_object_id = combios_check_dl_dvi(dev, 1); + } else { devices = ATOM_DEVICE_DFP1_SUPPORT; + connector_object_id = CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I; + } radeon_add_legacy_encoder(dev, radeon_get_encoder_id (dev, devices, 0), @@ -1726,7 +1816,8 @@ bool radeon_get_legacy_connector_info_from_bios(struct drm_device *dev) radeon_add_legacy_connector(dev, i, devices, legacy_connector_convert [connector], - &ddc_i2c); + &ddc_i2c, + connector_object_id); break; case CONNECTOR_CTV_LEGACY: case CONNECTOR_STV_LEGACY: @@ -1740,7 +1831,8 @@ bool radeon_get_legacy_connector_info_from_bios(struct drm_device *dev) ATOM_DEVICE_TV1_SUPPORT, legacy_connector_convert [connector], - &ddc_i2c); + &ddc_i2c, + CONNECTOR_OBJECT_ID_SVIDEO); break; default: DRM_ERROR("Unknown connector type: %d\n", @@ -1772,10 +1864,29 @@ bool radeon_get_legacy_connector_info_from_bios(struct drm_device *dev) ATOM_DEVICE_CRT1_SUPPORT | ATOM_DEVICE_DFP1_SUPPORT, DRM_MODE_CONNECTOR_DVII, - &ddc_i2c); + &ddc_i2c, + CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I); } else { - DRM_DEBUG("No connector info found\n"); - return false; + uint16_t crt_info = + combios_get_table_offset(dev, COMBIOS_CRT_INFO_TABLE); + DRM_DEBUG("Found CRT table, assuming VGA connector\n"); + if (crt_info) { + radeon_add_legacy_encoder(dev, + radeon_get_encoder_id(dev, + ATOM_DEVICE_CRT1_SUPPORT, + 1), + ATOM_DEVICE_CRT1_SUPPORT); + ddc_i2c = combios_setup_i2c_bus(RADEON_GPIO_VGA_DDC); + radeon_add_legacy_connector(dev, + 0, + ATOM_DEVICE_CRT1_SUPPORT, + DRM_MODE_CONNECTOR_VGA, + &ddc_i2c, + CONNECTOR_OBJECT_ID_VGA); + } else { + DRM_DEBUG("No connector info found\n"); + return false; + } } } @@ -1870,7 +1981,8 @@ bool radeon_get_legacy_connector_info_from_bios(struct drm_device *dev) 5, ATOM_DEVICE_LCD1_SUPPORT, DRM_MODE_CONNECTOR_LVDS, - &ddc_i2c); + &ddc_i2c, + CONNECTOR_OBJECT_ID_LVDS); } } @@ -1880,16 +1992,19 @@ bool radeon_get_legacy_connector_info_from_bios(struct drm_device *dev) combios_get_table_offset(dev, COMBIOS_TV_INFO_TABLE); if (tv_info) { if (RBIOS8(tv_info + 6) == 'T') { - radeon_add_legacy_encoder(dev, - radeon_get_encoder_id - (dev, - ATOM_DEVICE_TV1_SUPPORT, - 2), - ATOM_DEVICE_TV1_SUPPORT); - radeon_add_legacy_connector(dev, 6, - ATOM_DEVICE_TV1_SUPPORT, - DRM_MODE_CONNECTOR_SVIDEO, - &ddc_i2c); + if (radeon_apply_legacy_tv_quirks(dev)) { + radeon_add_legacy_encoder(dev, + radeon_get_encoder_id + (dev, + ATOM_DEVICE_TV1_SUPPORT, + 2), + ATOM_DEVICE_TV1_SUPPORT); + radeon_add_legacy_connector(dev, 6, + ATOM_DEVICE_TV1_SUPPORT, + DRM_MODE_CONNECTOR_SVIDEO, + &ddc_i2c, + CONNECTOR_OBJECT_ID_SVIDEO); + } } } } diff --git a/drivers/gpu/drm/radeon/radeon_connectors.c b/drivers/gpu/drm/radeon/radeon_connectors.c index af1d551..29763ce 100644 --- a/drivers/gpu/drm/radeon/radeon_connectors.c +++ b/drivers/gpu/drm/radeon/radeon_connectors.c @@ -26,6 +26,7 @@ #include "drmP.h" #include "drm_edid.h" #include "drm_crtc_helper.h" +#include "drm_fb_helper.h" #include "radeon_drm.h" #include "radeon.h" #include "atom.h" @@ -177,25 +178,12 @@ static struct drm_display_mode *radeon_fp_native_mode(struct drm_encoder *encode struct drm_device *dev = encoder->dev; struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); struct drm_display_mode *mode = NULL; - struct radeon_native_mode *native_mode = &radeon_encoder->native_mode; - - if (native_mode->panel_xres != 0 && - native_mode->panel_yres != 0 && - native_mode->dotclock != 0) { - mode = drm_mode_create(dev); - - mode->hdisplay = native_mode->panel_xres; - mode->vdisplay = native_mode->panel_yres; - - mode->htotal = mode->hdisplay + native_mode->hblank; - mode->hsync_start = mode->hdisplay + native_mode->hoverplus; - mode->hsync_end = mode->hsync_start + native_mode->hsync_width; - mode->vtotal = mode->vdisplay + native_mode->vblank; - mode->vsync_start = mode->vdisplay + native_mode->voverplus; - mode->vsync_end = mode->vsync_start + native_mode->vsync_width; - mode->clock = native_mode->dotclock; - mode->flags = 0; + struct drm_display_mode *native_mode = &radeon_encoder->native_mode; + if (native_mode->hdisplay != 0 && + native_mode->vdisplay != 0 && + native_mode->clock != 0) { + mode = drm_mode_duplicate(dev, native_mode); mode->type = DRM_MODE_TYPE_PREFERRED | DRM_MODE_TYPE_DRIVER; drm_mode_set_name(mode); @@ -209,7 +197,7 @@ static void radeon_add_common_modes(struct drm_encoder *encoder, struct drm_conn struct drm_device *dev = encoder->dev; struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); struct drm_display_mode *mode = NULL; - struct radeon_native_mode *native_mode = &radeon_encoder->native_mode; + struct drm_display_mode *native_mode = &radeon_encoder->native_mode; int i; struct mode_size { int w; @@ -235,17 +223,22 @@ static void radeon_add_common_modes(struct drm_encoder *encoder, struct drm_conn }; for (i = 0; i < 17; i++) { + if (radeon_encoder->devices & (ATOM_DEVICE_TV_SUPPORT)) { + if (common_modes[i].w > 1024 || + common_modes[i].h > 768) + continue; + } if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) { - if (common_modes[i].w > native_mode->panel_xres || - common_modes[i].h > native_mode->panel_yres || - (common_modes[i].w == native_mode->panel_xres && - common_modes[i].h == native_mode->panel_yres)) + if (common_modes[i].w > native_mode->hdisplay || + common_modes[i].h > native_mode->vdisplay || + (common_modes[i].w == native_mode->hdisplay && + common_modes[i].h == native_mode->vdisplay)) continue; } if (common_modes[i].w < 320 || common_modes[i].h < 200) continue; - mode = drm_cvt_mode(dev, common_modes[i].w, common_modes[i].h, 60, false, false); + mode = drm_cvt_mode(dev, common_modes[i].w, common_modes[i].h, 60, false, false, false); drm_mode_probed_add(connector, mode); } } @@ -343,28 +336,23 @@ static void radeon_fixup_lvds_native_mode(struct drm_encoder *encoder, struct drm_connector *connector) { struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); - struct radeon_native_mode *native_mode = &radeon_encoder->native_mode; + struct drm_display_mode *native_mode = &radeon_encoder->native_mode; /* Try to get native mode details from EDID if necessary */ - if (!native_mode->dotclock) { + if (!native_mode->clock) { struct drm_display_mode *t, *mode; list_for_each_entry_safe(mode, t, &connector->probed_modes, head) { - if (mode->hdisplay == native_mode->panel_xres && - mode->vdisplay == native_mode->panel_yres) { - native_mode->hblank = mode->htotal - mode->hdisplay; - native_mode->hoverplus = mode->hsync_start - mode->hdisplay; - native_mode->hsync_width = mode->hsync_end - mode->hsync_start; - native_mode->vblank = mode->vtotal - mode->vdisplay; - native_mode->voverplus = mode->vsync_start - mode->vdisplay; - native_mode->vsync_width = mode->vsync_end - mode->vsync_start; - native_mode->dotclock = mode->clock; + if (mode->hdisplay == native_mode->hdisplay && + mode->vdisplay == native_mode->vdisplay) { + *native_mode = *mode; + drm_mode_set_crtcinfo(native_mode, CRTC_INTERLACE_HALVE_V); DRM_INFO("Determined LVDS native mode details from EDID\n"); break; } } } - if (!native_mode->dotclock) { + if (!native_mode->clock) { DRM_INFO("No LVDS native mode details, disabling RMX\n"); radeon_encoder->rmx_type = RMX_OFF; } @@ -409,13 +397,64 @@ static int radeon_lvds_get_modes(struct drm_connector *connector) static int radeon_lvds_mode_valid(struct drm_connector *connector, struct drm_display_mode *mode) { + struct drm_encoder *encoder = radeon_best_single_encoder(connector); + + if ((mode->hdisplay < 320) || (mode->vdisplay < 240)) + return MODE_PANEL; + + if (encoder) { + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_display_mode *native_mode = &radeon_encoder->native_mode; + + /* AVIVO hardware supports downscaling modes larger than the panel + * to the panel size, but I'm not sure this is desirable. + */ + if ((mode->hdisplay > native_mode->hdisplay) || + (mode->vdisplay > native_mode->vdisplay)) + return MODE_PANEL; + + /* if scaling is disabled, block non-native modes */ + if (radeon_encoder->rmx_type == RMX_OFF) { + if ((mode->hdisplay != native_mode->hdisplay) || + (mode->vdisplay != native_mode->vdisplay)) + return MODE_PANEL; + } + } + return MODE_OK; } static enum drm_connector_status radeon_lvds_detect(struct drm_connector *connector) { - enum drm_connector_status ret = connector_status_connected; + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + struct drm_encoder *encoder = radeon_best_single_encoder(connector); + enum drm_connector_status ret = connector_status_disconnected; + + if (encoder) { + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_display_mode *native_mode = &radeon_encoder->native_mode; + + /* check if panel is valid */ + if (native_mode->hdisplay >= 320 && native_mode->vdisplay >= 240) + ret = connector_status_connected; + + } + + /* check for edid as well */ + if (radeon_connector->edid) + ret = connector_status_connected; + else { + if (radeon_connector->ddc_bus) { + radeon_i2c_do_lock(radeon_connector, 1); + radeon_connector->edid = drm_get_edid(&radeon_connector->base, + &radeon_connector->ddc_bus->adapter); + radeon_i2c_do_lock(radeon_connector, 0); + if (radeon_connector->edid) + ret = connector_status_connected; + } + } /* check acpi lid status ??? */ + radeon_connector_update_scratch_regs(connector, ret); return ret; } @@ -426,6 +465,8 @@ static void radeon_connector_destroy(struct drm_connector *connector) if (radeon_connector->ddc_bus) radeon_i2c_destroy(radeon_connector->ddc_bus); + if (radeon_connector->edid) + kfree(radeon_connector->edid); kfree(radeon_connector->con_priv); drm_sysfs_connector_remove(connector); drm_connector_cleanup(connector); @@ -495,6 +536,8 @@ static int radeon_vga_get_modes(struct drm_connector *connector) static int radeon_vga_mode_valid(struct drm_connector *connector, struct drm_display_mode *mode) { + /* XXX check mode bandwidth */ + /* XXX verify against max DAC output frequency */ return MODE_OK; } @@ -513,9 +556,33 @@ static enum drm_connector_status radeon_vga_detect(struct drm_connector *connect radeon_i2c_do_lock(radeon_connector, 1); dret = radeon_ddc_probe(radeon_connector); radeon_i2c_do_lock(radeon_connector, 0); - if (dret) - ret = connector_status_connected; - else { + if (dret) { + if (radeon_connector->edid) { + kfree(radeon_connector->edid); + radeon_connector->edid = NULL; + } + radeon_i2c_do_lock(radeon_connector, 1); + radeon_connector->edid = drm_get_edid(&radeon_connector->base, &radeon_connector->ddc_bus->adapter); + radeon_i2c_do_lock(radeon_connector, 0); + + if (!radeon_connector->edid) { + DRM_ERROR("%s: probed a monitor but no|invalid EDID\n", + drm_get_connector_name(connector)); + ret = connector_status_connected; + } else { + radeon_connector->use_digital = !!(radeon_connector->edid->input & DRM_EDID_INPUT_DIGITAL); + + /* some oems have boards with separate digital and analog connectors + * with a shared ddc line (often vga + hdmi) + */ + if (radeon_connector->use_digital && radeon_connector->shared_ddc) { + kfree(radeon_connector->edid); + radeon_connector->edid = NULL; + ret = connector_status_disconnected; + } else + ret = connector_status_connected; + } + } else { if (radeon_connector->dac_load_detect) { encoder_funcs = encoder->helper_private; ret = encoder_funcs->detect(encoder, connector); @@ -559,7 +626,7 @@ static int radeon_tv_get_modes(struct drm_connector *connector) radeon_add_common_modes(encoder, connector); else { /* only 800x600 is supported right now on pre-avivo chips */ - tv_mode = drm_cvt_mode(dev, 800, 600, 60, false, false); + tv_mode = drm_cvt_mode(dev, 800, 600, 60, false, false, false); tv_mode->type = DRM_MODE_TYPE_DRIVER | DRM_MODE_TYPE_PREFERRED; drm_mode_probed_add(connector, tv_mode); } @@ -569,6 +636,8 @@ static int radeon_tv_get_modes(struct drm_connector *connector) static int radeon_tv_mode_valid(struct drm_connector *connector, struct drm_display_mode *mode) { + if ((mode->hdisplay > 1024) || (mode->vdisplay > 768)) + return MODE_CLOCK_RANGE; return MODE_OK; } @@ -643,20 +712,29 @@ static enum drm_connector_status radeon_dvi_detect(struct drm_connector *connect dret = radeon_ddc_probe(radeon_connector); radeon_i2c_do_lock(radeon_connector, 0); if (dret) { + if (radeon_connector->edid) { + kfree(radeon_connector->edid); + radeon_connector->edid = NULL; + } radeon_i2c_do_lock(radeon_connector, 1); radeon_connector->edid = drm_get_edid(&radeon_connector->base, &radeon_connector->ddc_bus->adapter); radeon_i2c_do_lock(radeon_connector, 0); if (!radeon_connector->edid) { - DRM_ERROR("DDC responded but not EDID found for %s\n", - drm_get_connector_name(connector)); + DRM_ERROR("%s: probed a monitor but no|invalid EDID\n", + drm_get_connector_name(connector)); } else { radeon_connector->use_digital = !!(radeon_connector->edid->input & DRM_EDID_INPUT_DIGITAL); - /* if this isn't a digital monitor - then we need to make sure we don't have any - TV conflicts */ - ret = connector_status_connected; + /* some oems have boards with separate digital and analog connectors + * with a shared ddc line (often vga + hdmi) + */ + if ((!radeon_connector->use_digital) && radeon_connector->shared_ddc) { + kfree(radeon_connector->edid); + radeon_connector->edid = NULL; + ret = connector_status_disconnected; + } else + ret = connector_status_connected; } } @@ -743,9 +821,36 @@ struct drm_encoder *radeon_dvi_encoder(struct drm_connector *connector) return NULL; } +static void radeon_dvi_force(struct drm_connector *connector) +{ + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + if (connector->force == DRM_FORCE_ON) + radeon_connector->use_digital = false; + if (connector->force == DRM_FORCE_ON_DIGITAL) + radeon_connector->use_digital = true; +} + +static int radeon_dvi_mode_valid(struct drm_connector *connector, + struct drm_display_mode *mode) +{ + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + + /* XXX check mode bandwidth */ + + if (radeon_connector->use_digital && (mode->clock > 165000)) { + if ((radeon_connector->connector_object_id == CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I) || + (radeon_connector->connector_object_id == CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D) || + (radeon_connector->connector_object_id == CONNECTOR_OBJECT_ID_HDMI_TYPE_B)) + return MODE_OK; + else + return MODE_CLOCK_HIGH; + } + return MODE_OK; +} + struct drm_connector_helper_funcs radeon_dvi_connector_helper_funcs = { .get_modes = radeon_dvi_get_modes, - .mode_valid = radeon_vga_mode_valid, + .mode_valid = radeon_dvi_mode_valid, .best_encoder = radeon_dvi_encoder, }; @@ -755,6 +860,7 @@ struct drm_connector_funcs radeon_dvi_connector_funcs = { .fill_modes = drm_helper_probe_single_connector_modes, .set_property = radeon_connector_set_property, .destroy = radeon_connector_destroy, + .force = radeon_dvi_force, }; void @@ -764,13 +870,16 @@ radeon_add_atom_connector(struct drm_device *dev, int connector_type, struct radeon_i2c_bus_rec *i2c_bus, bool linkb, - uint32_t igp_lane_info) + uint32_t igp_lane_info, + uint16_t connector_object_id) { struct radeon_device *rdev = dev->dev_private; struct drm_connector *connector; struct radeon_connector *radeon_connector; struct radeon_connector_atom_dig *radeon_dig_connector; uint32_t subpixel_order = SubPixelNone; + bool shared_ddc = false; + int ret; /* fixme - tv/cv/din */ if (connector_type == DRM_MODE_CONNECTOR_Unknown) @@ -783,6 +892,13 @@ radeon_add_atom_connector(struct drm_device *dev, radeon_connector->devices |= supported_device; return; } + if (radeon_connector->ddc_bus && i2c_bus->valid) { + if (memcmp(&radeon_connector->ddc_bus->rec, i2c_bus, + sizeof(struct radeon_i2c_bus_rec)) == 0) { + radeon_connector->shared_ddc = true; + shared_ddc = true; + } + } } radeon_connector = kzalloc(sizeof(struct radeon_connector), GFP_KERNEL); @@ -793,27 +909,35 @@ radeon_add_atom_connector(struct drm_device *dev, radeon_connector->connector_id = connector_id; radeon_connector->devices = supported_device; + radeon_connector->shared_ddc = shared_ddc; + radeon_connector->connector_object_id = connector_object_id; switch (connector_type) { case DRM_MODE_CONNECTOR_VGA: drm_connector_init(dev, &radeon_connector->base, &radeon_vga_connector_funcs, connector_type); - drm_connector_helper_add(&radeon_connector->base, &radeon_vga_connector_helper_funcs); + ret = drm_connector_helper_add(&radeon_connector->base, &radeon_vga_connector_helper_funcs); + if (ret) + goto failed; if (i2c_bus->valid) { radeon_connector->ddc_bus = radeon_i2c_create(dev, i2c_bus, "VGA"); if (!radeon_connector->ddc_bus) goto failed; } + radeon_connector->dac_load_detect = true; drm_connector_attach_property(&radeon_connector->base, rdev->mode_info.load_detect_property, 1); break; case DRM_MODE_CONNECTOR_DVIA: drm_connector_init(dev, &radeon_connector->base, &radeon_vga_connector_funcs, connector_type); - drm_connector_helper_add(&radeon_connector->base, &radeon_vga_connector_helper_funcs); + ret = drm_connector_helper_add(&radeon_connector->base, &radeon_vga_connector_helper_funcs); + if (ret) + goto failed; if (i2c_bus->valid) { radeon_connector->ddc_bus = radeon_i2c_create(dev, i2c_bus, "DVI"); if (!radeon_connector->ddc_bus) goto failed; } + radeon_connector->dac_load_detect = true; drm_connector_attach_property(&radeon_connector->base, rdev->mode_info.load_detect_property, 1); @@ -827,7 +951,9 @@ radeon_add_atom_connector(struct drm_device *dev, radeon_dig_connector->igp_lane_info = igp_lane_info; radeon_connector->con_priv = radeon_dig_connector; drm_connector_init(dev, &radeon_connector->base, &radeon_dvi_connector_funcs, connector_type); - drm_connector_helper_add(&radeon_connector->base, &radeon_dvi_connector_helper_funcs); + ret = drm_connector_helper_add(&radeon_connector->base, &radeon_dvi_connector_helper_funcs); + if (ret) + goto failed; if (i2c_bus->valid) { radeon_connector->ddc_bus = radeon_i2c_create(dev, i2c_bus, "DVI"); if (!radeon_connector->ddc_bus) @@ -837,6 +963,7 @@ radeon_add_atom_connector(struct drm_device *dev, drm_connector_attach_property(&radeon_connector->base, rdev->mode_info.coherent_mode_property, 1); + radeon_connector->dac_load_detect = true; drm_connector_attach_property(&radeon_connector->base, rdev->mode_info.load_detect_property, 1); @@ -850,7 +977,9 @@ radeon_add_atom_connector(struct drm_device *dev, radeon_dig_connector->igp_lane_info = igp_lane_info; radeon_connector->con_priv = radeon_dig_connector; drm_connector_init(dev, &radeon_connector->base, &radeon_dvi_connector_funcs, connector_type); - drm_connector_helper_add(&radeon_connector->base, &radeon_dvi_connector_helper_funcs); + ret = drm_connector_helper_add(&radeon_connector->base, &radeon_dvi_connector_helper_funcs); + if (ret) + goto failed; if (i2c_bus->valid) { radeon_connector->ddc_bus = radeon_i2c_create(dev, i2c_bus, "HDMI"); if (!radeon_connector->ddc_bus) @@ -869,7 +998,9 @@ radeon_add_atom_connector(struct drm_device *dev, radeon_dig_connector->igp_lane_info = igp_lane_info; radeon_connector->con_priv = radeon_dig_connector; drm_connector_init(dev, &radeon_connector->base, &radeon_dvi_connector_funcs, connector_type); - drm_connector_helper_add(&radeon_connector->base, &radeon_dvi_connector_helper_funcs); + ret = drm_connector_helper_add(&radeon_connector->base, &radeon_dvi_connector_helper_funcs); + if (ret) + goto failed; if (i2c_bus->valid) { radeon_connector->ddc_bus = radeon_i2c_create(dev, i2c_bus, "DP"); if (!radeon_connector->ddc_bus) @@ -882,11 +1013,14 @@ radeon_add_atom_connector(struct drm_device *dev, case DRM_MODE_CONNECTOR_9PinDIN: if (radeon_tv == 1) { drm_connector_init(dev, &radeon_connector->base, &radeon_tv_connector_funcs, connector_type); - drm_connector_helper_add(&radeon_connector->base, &radeon_tv_connector_helper_funcs); + ret = drm_connector_helper_add(&radeon_connector->base, &radeon_tv_connector_helper_funcs); + if (ret) + goto failed; + radeon_connector->dac_load_detect = true; + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.load_detect_property, + 1); } - drm_connector_attach_property(&radeon_connector->base, - rdev->mode_info.load_detect_property, - 1); break; case DRM_MODE_CONNECTOR_LVDS: radeon_dig_connector = kzalloc(sizeof(struct radeon_connector_atom_dig), GFP_KERNEL); @@ -896,7 +1030,9 @@ radeon_add_atom_connector(struct drm_device *dev, radeon_dig_connector->igp_lane_info = igp_lane_info; radeon_connector->con_priv = radeon_dig_connector; drm_connector_init(dev, &radeon_connector->base, &radeon_lvds_connector_funcs, connector_type); - drm_connector_helper_add(&radeon_connector->base, &radeon_lvds_connector_helper_funcs); + ret = drm_connector_helper_add(&radeon_connector->base, &radeon_lvds_connector_helper_funcs); + if (ret) + goto failed; if (i2c_bus->valid) { radeon_connector->ddc_bus = radeon_i2c_create(dev, i2c_bus, "LVDS"); if (!radeon_connector->ddc_bus) @@ -926,12 +1062,14 @@ radeon_add_legacy_connector(struct drm_device *dev, uint32_t connector_id, uint32_t supported_device, int connector_type, - struct radeon_i2c_bus_rec *i2c_bus) + struct radeon_i2c_bus_rec *i2c_bus, + uint16_t connector_object_id) { struct radeon_device *rdev = dev->dev_private; struct drm_connector *connector; struct radeon_connector *radeon_connector; uint32_t subpixel_order = SubPixelNone; + int ret; /* fixme - tv/cv/din */ if (connector_type == DRM_MODE_CONNECTOR_Unknown) @@ -954,27 +1092,34 @@ radeon_add_legacy_connector(struct drm_device *dev, radeon_connector->connector_id = connector_id; radeon_connector->devices = supported_device; + radeon_connector->connector_object_id = connector_object_id; switch (connector_type) { case DRM_MODE_CONNECTOR_VGA: drm_connector_init(dev, &radeon_connector->base, &radeon_vga_connector_funcs, connector_type); - drm_connector_helper_add(&radeon_connector->base, &radeon_vga_connector_helper_funcs); + ret = drm_connector_helper_add(&radeon_connector->base, &radeon_vga_connector_helper_funcs); + if (ret) + goto failed; if (i2c_bus->valid) { radeon_connector->ddc_bus = radeon_i2c_create(dev, i2c_bus, "VGA"); if (!radeon_connector->ddc_bus) goto failed; } + radeon_connector->dac_load_detect = true; drm_connector_attach_property(&radeon_connector->base, rdev->mode_info.load_detect_property, 1); break; case DRM_MODE_CONNECTOR_DVIA: drm_connector_init(dev, &radeon_connector->base, &radeon_vga_connector_funcs, connector_type); - drm_connector_helper_add(&radeon_connector->base, &radeon_vga_connector_helper_funcs); + ret = drm_connector_helper_add(&radeon_connector->base, &radeon_vga_connector_helper_funcs); + if (ret) + goto failed; if (i2c_bus->valid) { radeon_connector->ddc_bus = radeon_i2c_create(dev, i2c_bus, "DVI"); if (!radeon_connector->ddc_bus) goto failed; } + radeon_connector->dac_load_detect = true; drm_connector_attach_property(&radeon_connector->base, rdev->mode_info.load_detect_property, 1); @@ -982,11 +1127,14 @@ radeon_add_legacy_connector(struct drm_device *dev, case DRM_MODE_CONNECTOR_DVII: case DRM_MODE_CONNECTOR_DVID: drm_connector_init(dev, &radeon_connector->base, &radeon_dvi_connector_funcs, connector_type); - drm_connector_helper_add(&radeon_connector->base, &radeon_dvi_connector_helper_funcs); + ret = drm_connector_helper_add(&radeon_connector->base, &radeon_dvi_connector_helper_funcs); + if (ret) + goto failed; if (i2c_bus->valid) { radeon_connector->ddc_bus = radeon_i2c_create(dev, i2c_bus, "DVI"); if (!radeon_connector->ddc_bus) goto failed; + radeon_connector->dac_load_detect = true; drm_connector_attach_property(&radeon_connector->base, rdev->mode_info.load_detect_property, 1); @@ -998,7 +1146,17 @@ radeon_add_legacy_connector(struct drm_device *dev, case DRM_MODE_CONNECTOR_9PinDIN: if (radeon_tv == 1) { drm_connector_init(dev, &radeon_connector->base, &radeon_tv_connector_funcs, connector_type); - drm_connector_helper_add(&radeon_connector->base, &radeon_tv_connector_helper_funcs); + ret = drm_connector_helper_add(&radeon_connector->base, &radeon_tv_connector_helper_funcs); + if (ret) + goto failed; + radeon_connector->dac_load_detect = true; + /* RS400,RC410,RS480 chipset seems to report a lot + * of false positive on load detect, we haven't yet + * found a way to make load detect reliable on those + * chipset, thus just disable it for TV. + */ + if (rdev->family == CHIP_RS400 || rdev->family == CHIP_RS480) + radeon_connector->dac_load_detect = false; drm_connector_attach_property(&radeon_connector->base, rdev->mode_info.load_detect_property, 1); @@ -1006,7 +1164,9 @@ radeon_add_legacy_connector(struct drm_device *dev, break; case DRM_MODE_CONNECTOR_LVDS: drm_connector_init(dev, &radeon_connector->base, &radeon_lvds_connector_funcs, connector_type); - drm_connector_helper_add(&radeon_connector->base, &radeon_lvds_connector_helper_funcs); + ret = drm_connector_helper_add(&radeon_connector->base, &radeon_lvds_connector_helper_funcs); + if (ret) + goto failed; if (i2c_bus->valid) { radeon_connector->ddc_bus = radeon_i2c_create(dev, i2c_bus, "LVDS"); if (!radeon_connector->ddc_bus) diff --git a/drivers/gpu/drm/radeon/radeon_cs.c b/drivers/gpu/drm/radeon/radeon_cs.c index 12f5990..5ab2cf9 100644 --- a/drivers/gpu/drm/radeon/radeon_cs.c +++ b/drivers/gpu/drm/radeon/radeon_cs.c @@ -142,15 +142,31 @@ int radeon_cs_parser_init(struct radeon_cs_parser *p, void *data) } p->chunks[i].length_dw = user_chunk.length_dw; - cdata = (uint32_t *)(unsigned long)user_chunk.chunk_data; + p->chunks[i].user_ptr = (void __user *)(unsigned long)user_chunk.chunk_data; - size = p->chunks[i].length_dw * sizeof(uint32_t); - p->chunks[i].kdata = kmalloc(size, GFP_KERNEL); - if (p->chunks[i].kdata == NULL) { - return -ENOMEM; - } - if (DRM_COPY_FROM_USER(p->chunks[i].kdata, cdata, size)) { - return -EFAULT; + cdata = (uint32_t *)(unsigned long)user_chunk.chunk_data; + if (p->chunks[i].chunk_id != RADEON_CHUNK_ID_IB) { + size = p->chunks[i].length_dw * sizeof(uint32_t); + p->chunks[i].kdata = kmalloc(size, GFP_KERNEL); + if (p->chunks[i].kdata == NULL) { + return -ENOMEM; + } + if (DRM_COPY_FROM_USER(p->chunks[i].kdata, + p->chunks[i].user_ptr, size)) { + return -EFAULT; + } + } else { + p->chunks[i].kpage[0] = kmalloc(PAGE_SIZE, GFP_KERNEL); + p->chunks[i].kpage[1] = kmalloc(PAGE_SIZE, GFP_KERNEL); + if (p->chunks[i].kpage[0] == NULL || p->chunks[i].kpage[1] == NULL) { + kfree(p->chunks[i].kpage[0]); + kfree(p->chunks[i].kpage[1]); + return -ENOMEM; + } + p->chunks[i].kpage_idx[0] = -1; + p->chunks[i].kpage_idx[1] = -1; + p->chunks[i].last_copied_page = -1; + p->chunks[i].last_page_index = ((p->chunks[i].length_dw * 4) - 1) / PAGE_SIZE; } } if (p->chunks[p->chunk_ib_idx].length_dw > (16 * 1024)) { @@ -190,6 +206,8 @@ static void radeon_cs_parser_fini(struct radeon_cs_parser *parser, int error) kfree(parser->relocs_ptr); for (i = 0; i < parser->nchunks; i++) { kfree(parser->chunks[i].kdata); + kfree(parser->chunks[i].kpage[0]); + kfree(parser->chunks[i].kpage[1]); } kfree(parser->chunks); kfree(parser->chunks_array); @@ -238,8 +256,14 @@ int radeon_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) * uncached). */ ib_chunk = &parser.chunks[parser.chunk_ib_idx]; parser.ib->length_dw = ib_chunk->length_dw; - memcpy((void *)parser.ib->ptr, ib_chunk->kdata, ib_chunk->length_dw*4); r = radeon_cs_parse(&parser); + if (r || parser.parser_error) { + DRM_ERROR("Invalid command stream !\n"); + radeon_cs_parser_fini(&parser, r); + mutex_unlock(&rdev->cs_mutex); + return r; + } + r = radeon_cs_finish_pages(&parser); if (r) { DRM_ERROR("Invalid command stream !\n"); radeon_cs_parser_fini(&parser, r); @@ -254,3 +278,64 @@ int radeon_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) mutex_unlock(&rdev->cs_mutex); return r; } + +int radeon_cs_finish_pages(struct radeon_cs_parser *p) +{ + struct radeon_cs_chunk *ibc = &p->chunks[p->chunk_ib_idx]; + int i; + int size = PAGE_SIZE; + + for (i = ibc->last_copied_page + 1; i <= ibc->last_page_index; i++) { + if (i == ibc->last_page_index) { + size = (ibc->length_dw * 4) % PAGE_SIZE; + if (size == 0) + size = PAGE_SIZE; + } + + if (DRM_COPY_FROM_USER(p->ib->ptr + (i * (PAGE_SIZE/4)), + ibc->user_ptr + (i * PAGE_SIZE), + size)) + return -EFAULT; + } + return 0; +} + +int radeon_cs_update_pages(struct radeon_cs_parser *p, int pg_idx) +{ + int new_page; + struct radeon_cs_chunk *ibc = &p->chunks[p->chunk_ib_idx]; + int i; + int size = PAGE_SIZE; + + for (i = ibc->last_copied_page + 1; i < pg_idx; i++) { + if (DRM_COPY_FROM_USER(p->ib->ptr + (i * (PAGE_SIZE/4)), + ibc->user_ptr + (i * PAGE_SIZE), + PAGE_SIZE)) { + p->parser_error = -EFAULT; + return 0; + } + } + + new_page = ibc->kpage_idx[0] < ibc->kpage_idx[1] ? 0 : 1; + + if (pg_idx == ibc->last_page_index) { + size = (ibc->length_dw * 4) % PAGE_SIZE; + if (size == 0) + size = PAGE_SIZE; + } + + if (DRM_COPY_FROM_USER(ibc->kpage[new_page], + ibc->user_ptr + (pg_idx * PAGE_SIZE), + size)) { + p->parser_error = -EFAULT; + return 0; + } + + /* copy to IB here */ + memcpy((void *)(p->ib->ptr+(pg_idx*(PAGE_SIZE/4))), ibc->kpage[new_page], size); + + ibc->last_copied_page = pg_idx; + ibc->kpage_idx[new_page] = pg_idx; + + return new_page; +} diff --git a/drivers/gpu/drm/radeon/radeon_cursor.c b/drivers/gpu/drm/radeon/radeon_cursor.c index b13c79e..28772a3 100644 --- a/drivers/gpu/drm/radeon/radeon_cursor.c +++ b/drivers/gpu/drm/radeon/radeon_cursor.c @@ -109,9 +109,15 @@ static void radeon_set_cursor(struct drm_crtc *crtc, struct drm_gem_object *obj, struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); struct radeon_device *rdev = crtc->dev->dev_private; - if (ASIC_IS_AVIVO(rdev)) + if (ASIC_IS_AVIVO(rdev)) { + if (rdev->family >= CHIP_RV770) { + if (radeon_crtc->crtc_id) + WREG32(R700_D2CUR_SURFACE_ADDRESS_HIGH, 0); + else + WREG32(R700_D1CUR_SURFACE_ADDRESS_HIGH, 0); + } WREG32(AVIVO_D1CUR_SURFACE_ADDRESS + radeon_crtc->crtc_offset, gpu_addr); - else { + } else { radeon_crtc->legacy_cursor_offset = gpu_addr - radeon_crtc->legacy_display_base_addr; /* offset is from DISP(2)_BASE_ADDRESS */ WREG32(RADEON_CUR_OFFSET + radeon_crtc->crtc_offset, radeon_crtc->legacy_cursor_offset); diff --git a/drivers/gpu/drm/radeon/radeon_device.c b/drivers/gpu/drm/radeon/radeon_device.c index daf5db7..41bb76f 100644 --- a/drivers/gpu/drm/radeon/radeon_device.c +++ b/drivers/gpu/drm/radeon/radeon_device.c @@ -322,10 +322,6 @@ int radeon_asic_init(struct radeon_device *rdev) case CHIP_RV380: rdev->asic = &r300_asic; if (rdev->flags & RADEON_IS_PCIE) { - rdev->asic->gart_init = &rv370_pcie_gart_init; - rdev->asic->gart_fini = &rv370_pcie_gart_fini; - rdev->asic->gart_enable = &rv370_pcie_gart_enable; - rdev->asic->gart_disable = &rv370_pcie_gart_disable; rdev->asic->gart_tlb_flush = &rv370_pcie_gart_tlb_flush; rdev->asic->gart_set_page = &rv370_pcie_gart_set_page; } @@ -448,20 +444,24 @@ static uint32_t cail_reg_read(struct card_info *info, uint32_t reg) return r; } -static struct card_info atom_card_info = { - .dev = NULL, - .reg_read = cail_reg_read, - .reg_write = cail_reg_write, - .mc_read = cail_mc_read, - .mc_write = cail_mc_write, - .pll_read = cail_pll_read, - .pll_write = cail_pll_write, -}; - int radeon_atombios_init(struct radeon_device *rdev) { - atom_card_info.dev = rdev->ddev; - rdev->mode_info.atom_context = atom_parse(&atom_card_info, rdev->bios); + struct card_info *atom_card_info = + kzalloc(sizeof(struct card_info), GFP_KERNEL); + + if (!atom_card_info) + return -ENOMEM; + + rdev->mode_info.atom_card_info = atom_card_info; + atom_card_info->dev = rdev->ddev; + atom_card_info->reg_read = cail_reg_read; + atom_card_info->reg_write = cail_reg_write; + atom_card_info->mc_read = cail_mc_read; + atom_card_info->mc_write = cail_mc_write; + atom_card_info->pll_read = cail_pll_read; + atom_card_info->pll_write = cail_pll_write; + + rdev->mode_info.atom_context = atom_parse(atom_card_info, rdev->bios); radeon_atom_initialize_bios_scratch_regs(rdev->ddev); return 0; } @@ -469,6 +469,7 @@ int radeon_atombios_init(struct radeon_device *rdev) void radeon_atombios_fini(struct radeon_device *rdev) { kfree(rdev->mode_info.atom_context); + kfree(rdev->mode_info.atom_card_info); } int radeon_combios_init(struct radeon_device *rdev) @@ -485,7 +486,6 @@ void radeon_combios_fini(struct radeon_device *rdev) static unsigned int radeon_vga_set_decode(void *cookie, bool state) { struct radeon_device *rdev = cookie; - radeon_vga_set_state(rdev, state); if (state) return VGA_RSRC_LEGACY_IO | VGA_RSRC_LEGACY_MEM | @@ -493,6 +493,29 @@ static unsigned int radeon_vga_set_decode(void *cookie, bool state) else return VGA_RSRC_NORMAL_IO | VGA_RSRC_NORMAL_MEM; } + +void radeon_agp_disable(struct radeon_device *rdev) +{ + rdev->flags &= ~RADEON_IS_AGP; + if (rdev->family >= CHIP_R600) { + DRM_INFO("Forcing AGP to PCIE mode\n"); + rdev->flags |= RADEON_IS_PCIE; + } else if (rdev->family >= CHIP_RV515 || + rdev->family == CHIP_RV380 || + rdev->family == CHIP_RV410 || + rdev->family == CHIP_R423) { + DRM_INFO("Forcing AGP to PCIE mode\n"); + rdev->flags |= RADEON_IS_PCIE; + rdev->asic->gart_tlb_flush = &rv370_pcie_gart_tlb_flush; + rdev->asic->gart_set_page = &rv370_pcie_gart_set_page; + } else { + DRM_INFO("Forcing AGP to PCI mode\n"); + rdev->flags |= RADEON_IS_PCI; + rdev->asic->gart_tlb_flush = &r100_pci_gart_tlb_flush; + rdev->asic->gart_set_page = &r100_pci_gart_set_page; + } +} + /* * Radeon device. */ @@ -531,29 +554,7 @@ int radeon_device_init(struct radeon_device *rdev, } if (radeon_agpmode == -1) { - rdev->flags &= ~RADEON_IS_AGP; - if (rdev->family >= CHIP_RV515 || - rdev->family == CHIP_RV380 || - rdev->family == CHIP_RV410 || - rdev->family == CHIP_R423) { - DRM_INFO("Forcing AGP to PCIE mode\n"); - rdev->flags |= RADEON_IS_PCIE; - rdev->asic->gart_init = &rv370_pcie_gart_init; - rdev->asic->gart_fini = &rv370_pcie_gart_fini; - rdev->asic->gart_enable = &rv370_pcie_gart_enable; - rdev->asic->gart_disable = &rv370_pcie_gart_disable; - rdev->asic->gart_tlb_flush = &rv370_pcie_gart_tlb_flush; - rdev->asic->gart_set_page = &rv370_pcie_gart_set_page; - } else { - DRM_INFO("Forcing AGP to PCI mode\n"); - rdev->flags |= RADEON_IS_PCI; - rdev->asic->gart_init = &r100_pci_gart_init; - rdev->asic->gart_fini = &r100_pci_gart_fini; - rdev->asic->gart_enable = &r100_pci_gart_enable; - rdev->asic->gart_disable = &r100_pci_gart_disable; - rdev->asic->gart_tlb_flush = &r100_pci_gart_tlb_flush; - rdev->asic->gart_set_page = &r100_pci_gart_set_page; - } + radeon_agp_disable(rdev); } /* set DMA mask + need_dma32 flags. @@ -585,111 +586,26 @@ int radeon_device_init(struct radeon_device *rdev, DRM_INFO("register mmio base: 0x%08X\n", (uint32_t)rdev->rmmio_base); DRM_INFO("register mmio size: %u\n", (unsigned)rdev->rmmio_size); - rdev->new_init_path = false; - r = radeon_init(rdev); - if (r) { - return r; - } - /* if we have > 1 VGA cards, then disable the radeon VGA resources */ - r = vga_client_register(rdev->pdev, rdev, NULL, radeon_vga_set_decode); - if (r) { - return -EINVAL; - } + /* this will fail for cards that aren't VGA class devices, just + * ignore it */ + vga_client_register(rdev->pdev, rdev, NULL, radeon_vga_set_decode); - if (!rdev->new_init_path) { - /* Setup errata flags */ - radeon_errata(rdev); - /* Initialize scratch registers */ - radeon_scratch_init(rdev); - /* Initialize surface registers */ - radeon_surface_init(rdev); - - /* BIOS*/ - if (!radeon_get_bios(rdev)) { - if (ASIC_IS_AVIVO(rdev)) - return -EINVAL; - } - if (rdev->is_atom_bios) { - r = radeon_atombios_init(rdev); - if (r) { - return r; - } - } else { - r = radeon_combios_init(rdev); - if (r) { - return r; - } - } - /* Reset gpu before posting otherwise ATOM will enter infinite loop */ - if (radeon_gpu_reset(rdev)) { - /* FIXME: what do we want to do here ? */ - } - /* check if cards are posted or not */ - if (!radeon_card_posted(rdev) && rdev->bios) { - DRM_INFO("GPU not posted. posting now...\n"); - if (rdev->is_atom_bios) { - atom_asic_init(rdev->mode_info.atom_context); - } else { - radeon_combios_asic_init(rdev->ddev); - } - } - /* Get clock & vram information */ - radeon_get_clock_info(rdev->ddev); - radeon_vram_info(rdev); - /* Initialize clocks */ - r = radeon_clocks_init(rdev); - if (r) { - return r; - } + r = radeon_init(rdev); + if (r) + return r; - /* Initialize memory controller (also test AGP) */ - r = radeon_mc_init(rdev); - if (r) { - return r; - } - /* Fence driver */ - r = radeon_fence_driver_init(rdev); - if (r) { - return r; - } - r = radeon_irq_kms_init(rdev); - if (r) { - return r; - } - /* Memory manager */ - r = radeon_object_init(rdev); - if (r) { - return r; - } - r = radeon_gpu_gart_init(rdev); + if (rdev->flags & RADEON_IS_AGP && !rdev->accel_working) { + /* Acceleration not working on AGP card try again + * with fallback to PCI or PCIE GART + */ + radeon_gpu_reset(rdev); + radeon_fini(rdev); + radeon_agp_disable(rdev); + r = radeon_init(rdev); if (r) return r; - /* Initialize GART (initialize after TTM so we can allocate - * memory through TTM but finalize after TTM) */ - r = radeon_gart_enable(rdev); - if (r) - return 0; - r = radeon_gem_init(rdev); - if (r) - return 0; - - /* 1M ring buffer */ - r = radeon_cp_init(rdev, 1024 * 1024); - if (r) - return 0; - r = radeon_wb_init(rdev); - if (r) - DRM_ERROR("radeon: failled initializing WB (%d).\n", r); - r = radeon_ib_pool_init(rdev); - if (r) - return 0; - r = radeon_ib_test(rdev); - if (r) - return 0; - rdev->accel_working = true; } - DRM_INFO("radeon: kernel modesetting successfully initialized.\n"); if (radeon_testing) { radeon_test_moves(rdev); } @@ -703,32 +619,8 @@ void radeon_device_fini(struct radeon_device *rdev) { DRM_INFO("radeon: finishing device.\n"); rdev->shutdown = true; - /* Order matter so becarefull if you rearrange anythings */ - if (!rdev->new_init_path) { - radeon_ib_pool_fini(rdev); - radeon_cp_fini(rdev); - radeon_wb_fini(rdev); - radeon_gpu_gart_fini(rdev); - radeon_gem_fini(rdev); - radeon_mc_fini(rdev); -#if __OS_HAS_AGP - radeon_agp_fini(rdev); -#endif - radeon_irq_kms_fini(rdev); - vga_client_register(rdev->pdev, NULL, NULL, NULL); - radeon_fence_driver_fini(rdev); - radeon_clocks_fini(rdev); - radeon_object_fini(rdev); - if (rdev->is_atom_bios) { - radeon_atombios_fini(rdev); - } else { - radeon_combios_fini(rdev); - } - kfree(rdev->bios); - rdev->bios = NULL; - } else { - radeon_fini(rdev); - } + radeon_fini(rdev); + vga_client_register(rdev->pdev, NULL, NULL, NULL); iounmap(rdev->rmmio); rdev->rmmio = NULL; } @@ -768,14 +660,7 @@ int radeon_suspend_kms(struct drm_device *dev, pm_message_t state) radeon_save_bios_scratch_regs(rdev); - if (!rdev->new_init_path) { - radeon_cp_disable(rdev); - radeon_gart_disable(rdev); - rdev->irq.sw_int = false; - radeon_irq_set(rdev); - } else { - radeon_suspend(rdev); - } + radeon_suspend(rdev); /* evict remaining vram memory */ radeon_object_evict_vram(rdev); @@ -794,7 +679,6 @@ int radeon_suspend_kms(struct drm_device *dev, pm_message_t state) int radeon_resume_kms(struct drm_device *dev) { struct radeon_device *rdev = dev->dev_private; - int r; acquire_console_sem(); pci_set_power_state(dev->pdev, PCI_D0); @@ -804,43 +688,9 @@ int radeon_resume_kms(struct drm_device *dev) return -1; } pci_set_master(dev->pdev); - /* Reset gpu before posting otherwise ATOM will enter infinite loop */ - if (!rdev->new_init_path) { - if (radeon_gpu_reset(rdev)) { - /* FIXME: what do we want to do here ? */ - } - /* post card */ - if (rdev->is_atom_bios) { - atom_asic_init(rdev->mode_info.atom_context); - } else { - radeon_combios_asic_init(rdev->ddev); - } - /* Initialize clocks */ - r = radeon_clocks_init(rdev); - if (r) { - release_console_sem(); - return r; - } - /* Enable IRQ */ - rdev->irq.sw_int = true; - radeon_irq_set(rdev); - /* Initialize GPU Memory Controller */ - r = radeon_mc_init(rdev); - if (r) { - goto out; - } - r = radeon_gart_enable(rdev); - if (r) { - goto out; - } - r = radeon_cp_init(rdev, rdev->cp.ring_size); - if (r) { - goto out; - } - } else { - radeon_resume(rdev); - } -out: + /* resume AGP if in use */ + radeon_agp_resume(rdev); + radeon_resume(rdev); radeon_restore_bios_scratch_regs(rdev); fb_set_suspend(rdev->fbdev_info, 0); release_console_sem(); diff --git a/drivers/gpu/drm/radeon/radeon_display.c b/drivers/gpu/drm/radeon/radeon_display.c index 5d8141b..c85df4a 100644 --- a/drivers/gpu/drm/radeon/radeon_display.c +++ b/drivers/gpu/drm/radeon/radeon_display.c @@ -106,51 +106,44 @@ void radeon_crtc_load_lut(struct drm_crtc *crtc) legacy_crtc_load_lut(crtc); } -/** Sets the color ramps on behalf of RandR */ +/** Sets the color ramps on behalf of fbcon */ void radeon_crtc_fb_gamma_set(struct drm_crtc *crtc, u16 red, u16 green, u16 blue, int regno) { struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); - if (regno == 0) - DRM_DEBUG("gamma set %d\n", radeon_crtc->crtc_id); radeon_crtc->lut_r[regno] = red >> 6; radeon_crtc->lut_g[regno] = green >> 6; radeon_crtc->lut_b[regno] = blue >> 6; } +/** Gets the color ramps on behalf of fbcon */ +void radeon_crtc_fb_gamma_get(struct drm_crtc *crtc, u16 *red, u16 *green, + u16 *blue, int regno) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + + *red = radeon_crtc->lut_r[regno] << 6; + *green = radeon_crtc->lut_g[regno] << 6; + *blue = radeon_crtc->lut_b[regno] << 6; +} + static void radeon_crtc_gamma_set(struct drm_crtc *crtc, u16 *red, u16 *green, u16 *blue, uint32_t size) { struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); - int i, j; + int i; if (size != 256) { return; } - if (crtc->fb == NULL) { - return; - } - if (crtc->fb->depth == 16) { - for (i = 0; i < 64; i++) { - if (i <= 31) { - for (j = 0; j < 8; j++) { - radeon_crtc->lut_r[i * 8 + j] = red[i] >> 6; - radeon_crtc->lut_b[i * 8 + j] = blue[i] >> 6; - } - } - for (j = 0; j < 4; j++) - radeon_crtc->lut_g[i * 4 + j] = green[i] >> 6; - } - } else { - for (i = 0; i < 256; i++) { - radeon_crtc->lut_r[i] = red[i] >> 6; - radeon_crtc->lut_g[i] = green[i] >> 6; - radeon_crtc->lut_b[i] = blue[i] >> 6; - } + /* userspace palettes are always correct as is */ + for (i = 0; i < 256; i++) { + radeon_crtc->lut_r[i] = red[i] >> 6; + radeon_crtc->lut_g[i] = green[i] >> 6; + radeon_crtc->lut_b[i] = blue[i] >> 6; } - radeon_crtc_load_lut(crtc); } @@ -341,27 +334,19 @@ static bool radeon_setup_enc_conn(struct drm_device *dev) int radeon_ddc_get_modes(struct radeon_connector *radeon_connector) { - struct edid *edid; int ret = 0; if (!radeon_connector->ddc_bus) return -1; if (!radeon_connector->edid) { radeon_i2c_do_lock(radeon_connector, 1); - edid = drm_get_edid(&radeon_connector->base, &radeon_connector->ddc_bus->adapter); + radeon_connector->edid = drm_get_edid(&radeon_connector->base, &radeon_connector->ddc_bus->adapter); radeon_i2c_do_lock(radeon_connector, 0); - } else - edid = radeon_connector->edid; + } - if (edid) { - /* update digital bits here */ - if (edid->input & DRM_EDID_INPUT_DIGITAL) - radeon_connector->use_digital = 1; - else - radeon_connector->use_digital = 0; - drm_mode_connector_update_edid_property(&radeon_connector->base, edid); - ret = drm_add_edid_modes(&radeon_connector->base, edid); - kfree(edid); + if (radeon_connector->edid) { + drm_mode_connector_update_edid_property(&radeon_connector->base, radeon_connector->edid); + ret = drm_add_edid_modes(&radeon_connector->base, radeon_connector->edid); return ret; } drm_mode_connector_update_edid_property(&radeon_connector->base, NULL); @@ -724,7 +709,11 @@ int radeon_modeset_init(struct radeon_device *rdev) if (ret) { return ret; } - /* allocate crtcs - TODO single crtc */ + + if (rdev->flags & RADEON_SINGLE_CRTC) + num_crtc = 1; + + /* allocate crtcs */ for (i = 0; i < num_crtc; i++) { radeon_crtc_init(rdev->ddev, i); } @@ -764,7 +753,7 @@ bool radeon_crtc_scaling_mode_fixup(struct drm_crtc *crtc, radeon_crtc->rmx_type = radeon_encoder->rmx_type; memcpy(&radeon_crtc->native_mode, &radeon_encoder->native_mode, - sizeof(struct radeon_native_mode)); + sizeof(struct drm_display_mode)); first = false; } else { if (radeon_crtc->rmx_type != radeon_encoder->rmx_type) { @@ -782,10 +771,10 @@ bool radeon_crtc_scaling_mode_fixup(struct drm_crtc *crtc, if (radeon_crtc->rmx_type != RMX_OFF) { fixed20_12 a, b; a.full = rfixed_const(crtc->mode.vdisplay); - b.full = rfixed_const(radeon_crtc->native_mode.panel_xres); + b.full = rfixed_const(radeon_crtc->native_mode.hdisplay); radeon_crtc->vsc.full = rfixed_div(a, b); a.full = rfixed_const(crtc->mode.hdisplay); - b.full = rfixed_const(radeon_crtc->native_mode.panel_yres); + b.full = rfixed_const(radeon_crtc->native_mode.vdisplay); radeon_crtc->hsc.full = rfixed_div(a, b); } else { radeon_crtc->vsc.full = rfixed_const(1); diff --git a/drivers/gpu/drm/radeon/radeon_drv.c b/drivers/gpu/drm/radeon/radeon_drv.c index 50fce49..7f50fb8 100644 --- a/drivers/gpu/drm/radeon/radeon_drv.c +++ b/drivers/gpu/drm/radeon/radeon_drv.c @@ -62,9 +62,6 @@ void radeon_driver_irq_preinstall_kms(struct drm_device *dev); int radeon_driver_irq_postinstall_kms(struct drm_device *dev); void radeon_driver_irq_uninstall_kms(struct drm_device *dev); irqreturn_t radeon_driver_irq_handler_kms(DRM_IRQ_ARGS); -int radeon_master_create_kms(struct drm_device *dev, struct drm_master *master); -void radeon_master_destroy_kms(struct drm_device *dev, - struct drm_master *master); int radeon_dma_ioctl_kms(struct drm_device *dev, void *data, struct drm_file *file_priv); int radeon_gem_object_init(struct drm_gem_object *obj); @@ -260,8 +257,6 @@ static struct drm_driver kms_driver = { .get_vblank_counter = radeon_get_vblank_counter_kms, .enable_vblank = radeon_enable_vblank_kms, .disable_vblank = radeon_disable_vblank_kms, - .master_create = radeon_master_create_kms, - .master_destroy = radeon_master_destroy_kms, #if defined(CONFIG_DEBUG_FS) .debugfs_init = radeon_debugfs_init, .debugfs_cleanup = radeon_debugfs_cleanup, diff --git a/drivers/gpu/drm/radeon/radeon_drv.h b/drivers/gpu/drm/radeon/radeon_drv.h index cb0cfe4..350962e 100644 --- a/drivers/gpu/drm/radeon/radeon_drv.h +++ b/drivers/gpu/drm/radeon/radeon_drv.h @@ -34,6 +34,8 @@ #include <linux/firmware.h> #include <linux/platform_device.h> +#include "radeon_family.h" + /* General customization: */ @@ -109,75 +111,12 @@ #define DRIVER_MINOR 31 #define DRIVER_PATCHLEVEL 0 -/* - * Radeon chip families - */ -enum radeon_family { - CHIP_R100, - CHIP_RV100, - CHIP_RS100, - CHIP_RV200, - CHIP_RS200, - CHIP_R200, - CHIP_RV250, - CHIP_RS300, - CHIP_RV280, - CHIP_R300, - CHIP_R350, - CHIP_RV350, - CHIP_RV380, - CHIP_R420, - CHIP_R423, - CHIP_RV410, - CHIP_RS400, - CHIP_RS480, - CHIP_RS600, - CHIP_RS690, - CHIP_RS740, - CHIP_RV515, - CHIP_R520, - CHIP_RV530, - CHIP_RV560, - CHIP_RV570, - CHIP_R580, - CHIP_R600, - CHIP_RV610, - CHIP_RV630, - CHIP_RV620, - CHIP_RV635, - CHIP_RV670, - CHIP_RS780, - CHIP_RS880, - CHIP_RV770, - CHIP_RV730, - CHIP_RV710, - CHIP_RV740, - CHIP_LAST, -}; - enum radeon_cp_microcode_version { UCODE_R100, UCODE_R200, UCODE_R300, }; -/* - * Chip flags - */ -enum radeon_chip_flags { - RADEON_FAMILY_MASK = 0x0000ffffUL, - RADEON_FLAGS_MASK = 0xffff0000UL, - RADEON_IS_MOBILITY = 0x00010000UL, - RADEON_IS_IGP = 0x00020000UL, - RADEON_SINGLE_CRTC = 0x00040000UL, - RADEON_IS_AGP = 0x00080000UL, - RADEON_HAS_HIERZ = 0x00100000UL, - RADEON_IS_PCIE = 0x00200000UL, - RADEON_NEW_MEMMAP = 0x00400000UL, - RADEON_IS_PCI = 0x00800000UL, - RADEON_IS_IGPGART = 0x01000000UL, -}; - typedef struct drm_radeon_freelist { unsigned int age; struct drm_buf *buf; diff --git a/drivers/gpu/drm/radeon/radeon_encoders.c b/drivers/gpu/drm/radeon/radeon_encoders.c index 6216467..d42bc51 100644 --- a/drivers/gpu/drm/radeon/radeon_encoders.c +++ b/drivers/gpu/drm/radeon/radeon_encoders.c @@ -31,6 +31,10 @@ extern int atom_debug; +/* evil but including atombios.h is much worse */ +bool radeon_atom_get_tv_timings(struct radeon_device *rdev, int index, + struct drm_display_mode *mode); + uint32_t radeon_get_encoder_id(struct drm_device *dev, uint32_t supported_device, uint8_t dac) { @@ -167,49 +171,17 @@ void radeon_rmx_mode_fixup(struct drm_encoder *encoder, struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); struct drm_device *dev = encoder->dev; struct radeon_device *rdev = dev->dev_private; - struct radeon_native_mode *native_mode = &radeon_encoder->native_mode; - - if (mode->hdisplay < native_mode->panel_xres || - mode->vdisplay < native_mode->panel_yres) { - if (ASIC_IS_AVIVO(rdev)) { - adjusted_mode->hdisplay = native_mode->panel_xres; - adjusted_mode->vdisplay = native_mode->panel_yres; - adjusted_mode->htotal = native_mode->panel_xres + native_mode->hblank; - adjusted_mode->hsync_start = native_mode->panel_xres + native_mode->hoverplus; - adjusted_mode->hsync_end = adjusted_mode->hsync_start + native_mode->hsync_width; - adjusted_mode->vtotal = native_mode->panel_yres + native_mode->vblank; - adjusted_mode->vsync_start = native_mode->panel_yres + native_mode->voverplus; - adjusted_mode->vsync_end = adjusted_mode->vsync_start + native_mode->vsync_width; - /* update crtc values */ - drm_mode_set_crtcinfo(adjusted_mode, CRTC_INTERLACE_HALVE_V); - /* adjust crtc values */ - adjusted_mode->crtc_hdisplay = native_mode->panel_xres; - adjusted_mode->crtc_vdisplay = native_mode->panel_yres; - adjusted_mode->crtc_htotal = adjusted_mode->crtc_hdisplay + native_mode->hblank; - adjusted_mode->crtc_hsync_start = adjusted_mode->crtc_hdisplay + native_mode->hoverplus; - adjusted_mode->crtc_hsync_end = adjusted_mode->crtc_hsync_start + native_mode->hsync_width; - adjusted_mode->crtc_vtotal = adjusted_mode->crtc_vdisplay + native_mode->vblank; - adjusted_mode->crtc_vsync_start = adjusted_mode->crtc_vdisplay + native_mode->voverplus; - adjusted_mode->crtc_vsync_end = adjusted_mode->crtc_vsync_start + native_mode->vsync_width; - } else { - adjusted_mode->htotal = native_mode->panel_xres + native_mode->hblank; - adjusted_mode->hsync_start = native_mode->panel_xres + native_mode->hoverplus; - adjusted_mode->hsync_end = adjusted_mode->hsync_start + native_mode->hsync_width; - adjusted_mode->vtotal = native_mode->panel_yres + native_mode->vblank; - adjusted_mode->vsync_start = native_mode->panel_yres + native_mode->voverplus; - adjusted_mode->vsync_end = adjusted_mode->vsync_start + native_mode->vsync_width; - /* update crtc values */ - drm_mode_set_crtcinfo(adjusted_mode, CRTC_INTERLACE_HALVE_V); - /* adjust crtc values */ - adjusted_mode->crtc_htotal = adjusted_mode->crtc_hdisplay + native_mode->hblank; - adjusted_mode->crtc_hsync_start = adjusted_mode->crtc_hdisplay + native_mode->hoverplus; - adjusted_mode->crtc_hsync_end = adjusted_mode->crtc_hsync_start + native_mode->hsync_width; - adjusted_mode->crtc_vtotal = adjusted_mode->crtc_vdisplay + native_mode->vblank; - adjusted_mode->crtc_vsync_start = adjusted_mode->crtc_vdisplay + native_mode->voverplus; - adjusted_mode->crtc_vsync_end = adjusted_mode->crtc_vsync_start + native_mode->vsync_width; + struct drm_display_mode *native_mode = &radeon_encoder->native_mode; + + if (mode->hdisplay < native_mode->hdisplay || + mode->vdisplay < native_mode->vdisplay) { + int mode_id = adjusted_mode->base.id; + *adjusted_mode = *native_mode; + if (!ASIC_IS_AVIVO(rdev)) { + adjusted_mode->hdisplay = mode->hdisplay; + adjusted_mode->vdisplay = mode->vdisplay; } - adjusted_mode->flags = native_mode->flags; - adjusted_mode->clock = native_mode->dotclock; + adjusted_mode->base.id = mode_id; } } @@ -219,7 +191,11 @@ static bool radeon_atom_mode_fixup(struct drm_encoder *encoder, struct drm_display_mode *adjusted_mode) { struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + /* set the active encoder to connector routing */ + radeon_encoder_set_active_device(encoder); drm_mode_set_crtcinfo(adjusted_mode, 0); if (radeon_encoder->rmx_type != RMX_OFF) @@ -230,6 +206,18 @@ static bool radeon_atom_mode_fixup(struct drm_encoder *encoder, && (mode->crtc_vsync_start < (mode->crtc_vdisplay + 2))) adjusted_mode->crtc_vsync_start = adjusted_mode->crtc_vdisplay + 2; + if (radeon_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT)) { + struct radeon_encoder_atom_dac *tv_dac = radeon_encoder->enc_priv; + if (tv_dac) { + if (tv_dac->tv_std == TV_STD_NTSC || + tv_dac->tv_std == TV_STD_NTSC_J || + tv_dac->tv_std == TV_STD_PAL_M) + radeon_atom_get_tv_timings(rdev, 0, adjusted_mode); + else + radeon_atom_get_tv_timings(rdev, 1, adjusted_mode); + } + } + return true; } @@ -461,7 +449,7 @@ atombios_digital_setup(struct drm_encoder *encoder, int action) case 1: args.v1.ucMisc = 0; args.v1.ucAction = action; - if (drm_detect_hdmi_monitor((struct edid *)connector->edid_blob_ptr)) + if (drm_detect_hdmi_monitor(radeon_connector->edid)) args.v1.ucMisc |= PANEL_ENCODER_MISC_HDMI_TYPE; args.v1.usPixelClock = cpu_to_le16(radeon_encoder->pixel_clock / 10); if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) { @@ -486,7 +474,7 @@ atombios_digital_setup(struct drm_encoder *encoder, int action) if (dig->coherent_mode) args.v2.ucMisc |= PANEL_ENCODER_MISC_COHERENT; } - if (drm_detect_hdmi_monitor((struct edid *)connector->edid_blob_ptr)) + if (drm_detect_hdmi_monitor(radeon_connector->edid)) args.v2.ucMisc |= PANEL_ENCODER_MISC_HDMI_TYPE; args.v2.usPixelClock = cpu_to_le16(radeon_encoder->pixel_clock / 10); args.v2.ucTruncate = 0; @@ -544,7 +532,7 @@ atombios_get_encoder_mode(struct drm_encoder *encoder) switch (connector->connector_type) { case DRM_MODE_CONNECTOR_DVII: case DRM_MODE_CONNECTOR_HDMIB: /* HDMI-B is basically DL-DVI; analog works fine */ - if (drm_detect_hdmi_monitor((struct edid *)connector->edid_blob_ptr)) + if (drm_detect_hdmi_monitor(radeon_connector->edid)) return ATOM_ENCODER_MODE_HDMI; else if (radeon_connector->use_digital) return ATOM_ENCODER_MODE_DVI; @@ -554,7 +542,7 @@ atombios_get_encoder_mode(struct drm_encoder *encoder) case DRM_MODE_CONNECTOR_DVID: case DRM_MODE_CONNECTOR_HDMIA: default: - if (drm_detect_hdmi_monitor((struct edid *)connector->edid_blob_ptr)) + if (drm_detect_hdmi_monitor(radeon_connector->edid)) return ATOM_ENCODER_MODE_HDMI; else return ATOM_ENCODER_MODE_DVI; @@ -566,7 +554,7 @@ atombios_get_encoder_mode(struct drm_encoder *encoder) /*if (radeon_output->MonType == MT_DP) return ATOM_ENCODER_MODE_DP; else*/ - if (drm_detect_hdmi_monitor((struct edid *)connector->edid_blob_ptr)) + if (drm_detect_hdmi_monitor(radeon_connector->edid)) return ATOM_ENCODER_MODE_HDMI; else return ATOM_ENCODER_MODE_DVI; @@ -734,14 +722,17 @@ atombios_dig_transmitter_setup(struct drm_encoder *encoder, int action) atom_parse_cmd_header(rdev->mode_info.atom_context, index, &frev, &crev); args.v1.ucAction = action; - + if (action == ATOM_TRANSMITTER_ACTION_INIT) { + args.v1.usInitInfo = radeon_connector->connector_object_id; + } else { + if (radeon_encoder->pixel_clock > 165000) + args.v1.usPixelClock = cpu_to_le16((radeon_encoder->pixel_clock / 2) / 10); + else + args.v1.usPixelClock = cpu_to_le16(radeon_encoder->pixel_clock / 10); + } if (ASIC_IS_DCE32(rdev)) { - if (radeon_encoder->pixel_clock > 165000) { - args.v2.usPixelClock = cpu_to_le16((radeon_encoder->pixel_clock * 10 * 2) / 100); - args.v2.acConfig.fDualLinkConnector = 1; - } else { - args.v2.usPixelClock = cpu_to_le16((radeon_encoder->pixel_clock * 10 * 4) / 100); - } + if (radeon_encoder->pixel_clock > 165000) + args.v2.usPixelClock = cpu_to_le16((radeon_encoder->pixel_clock / 2) / 10); if (dig->dig_block) args.v2.acConfig.ucEncoderSel = 1; @@ -766,7 +757,6 @@ atombios_dig_transmitter_setup(struct drm_encoder *encoder, int action) } } else { args.v1.ucConfig = ATOM_TRANSMITTER_CONFIG_CLKSRC_PPLL; - args.v1.usPixelClock = cpu_to_le16((radeon_encoder->pixel_clock) / 10); switch (radeon_encoder->encoder_id) { case ENCODER_OBJECT_ID_INTERNAL_UNIPHY: @@ -874,16 +864,9 @@ radeon_atom_encoder_dpms(struct drm_encoder *encoder, int mode) DISPLAY_DEVICE_OUTPUT_CONTROL_PS_ALLOCATION args; int index = 0; bool is_dig = false; - int devices; memset(&args, 0, sizeof(args)); - /* on DPMS off we have no idea if active device is meaningful */ - if (mode != DRM_MODE_DPMS_ON && !radeon_encoder->active_device) - devices = radeon_encoder->devices; - else - devices = radeon_encoder->active_device; - DRM_DEBUG("encoder dpms %d to mode %d, devices %08x, active_devices %08x\n", radeon_encoder->encoder_id, mode, radeon_encoder->devices, radeon_encoder->active_device); @@ -914,18 +897,18 @@ radeon_atom_encoder_dpms(struct drm_encoder *encoder, int mode) break; case ENCODER_OBJECT_ID_INTERNAL_DAC1: case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1: - if (devices & (ATOM_DEVICE_TV_SUPPORT)) + if (radeon_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT)) index = GetIndexIntoMasterTable(COMMAND, TV1OutputControl); - else if (devices & (ATOM_DEVICE_CV_SUPPORT)) + else if (radeon_encoder->active_device & (ATOM_DEVICE_CV_SUPPORT)) index = GetIndexIntoMasterTable(COMMAND, CV1OutputControl); else index = GetIndexIntoMasterTable(COMMAND, DAC1OutputControl); break; case ENCODER_OBJECT_ID_INTERNAL_DAC2: case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2: - if (devices & (ATOM_DEVICE_TV_SUPPORT)) + if (radeon_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT)) index = GetIndexIntoMasterTable(COMMAND, TV1OutputControl); - else if (devices & (ATOM_DEVICE_CV_SUPPORT)) + else if (radeon_encoder->active_device & (ATOM_DEVICE_CV_SUPPORT)) index = GetIndexIntoMasterTable(COMMAND, CV1OutputControl); else index = GetIndexIntoMasterTable(COMMAND, DAC2OutputControl); @@ -1104,8 +1087,11 @@ atombios_apply_encoder_quirks(struct drm_encoder *encoder, } /* set scaler clears this on some chips */ - if (ASIC_IS_AVIVO(rdev) && (mode->flags & DRM_MODE_FLAG_INTERLACE)) - WREG32(AVIVO_D1MODE_DATA_FORMAT + radeon_crtc->crtc_offset, AVIVO_D1MODE_INTERLEAVE_EN); + if (!(radeon_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT))) { + if (ASIC_IS_AVIVO(rdev) && (mode->flags & DRM_MODE_FLAG_INTERLACE)) + WREG32(AVIVO_D1MODE_DATA_FORMAT + radeon_crtc->crtc_offset, + AVIVO_D1MODE_INTERLEAVE_EN); + } } static void @@ -1153,6 +1139,7 @@ radeon_atom_encoder_mode_set(struct drm_encoder *encoder, /* setup and enable the encoder and transmitter */ atombios_dig_encoder_setup(encoder, ATOM_ENABLE); + atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_INIT); atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_SETUP); atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_ENABLE); break; @@ -1268,8 +1255,6 @@ static void radeon_atom_encoder_prepare(struct drm_encoder *encoder) { radeon_atom_output_lock(encoder, true); radeon_atom_encoder_dpms(encoder, DRM_MODE_DPMS_OFF); - - radeon_encoder_set_active_device(encoder); } static void radeon_atom_encoder_commit(struct drm_encoder *encoder) @@ -1345,6 +1330,7 @@ radeon_atombios_set_dig_info(struct radeon_encoder *radeon_encoder) void radeon_add_atom_encoder(struct drm_device *dev, uint32_t encoder_id, uint32_t supported_device) { + struct radeon_device *rdev = dev->dev_private; struct drm_encoder *encoder; struct radeon_encoder *radeon_encoder; @@ -1364,7 +1350,10 @@ radeon_add_atom_encoder(struct drm_device *dev, uint32_t encoder_id, uint32_t su return; encoder = &radeon_encoder->base; - encoder->possible_crtcs = 0x3; + if (rdev->flags & RADEON_SINGLE_CRTC) + encoder->possible_crtcs = 0x1; + else + encoder->possible_crtcs = 0x3; encoder->possible_clones = 0; radeon_encoder->enc_priv = NULL; diff --git a/drivers/gpu/drm/radeon/radeon_family.h b/drivers/gpu/drm/radeon/radeon_family.h new file mode 100644 index 0000000..797972e --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_family.h @@ -0,0 +1,97 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ + +/* this file defines the CHIP_ and family flags used in the pciids, + * its is common between kms and non-kms because duplicating it and + * changing one place is fail. + */ +#ifndef RADEON_FAMILY_H +#define RADEON_FAMILY_H +/* + * Radeon chip families + */ +enum radeon_family { + CHIP_R100, + CHIP_RV100, + CHIP_RS100, + CHIP_RV200, + CHIP_RS200, + CHIP_R200, + CHIP_RV250, + CHIP_RS300, + CHIP_RV280, + CHIP_R300, + CHIP_R350, + CHIP_RV350, + CHIP_RV380, + CHIP_R420, + CHIP_R423, + CHIP_RV410, + CHIP_RS400, + CHIP_RS480, + CHIP_RS600, + CHIP_RS690, + CHIP_RS740, + CHIP_RV515, + CHIP_R520, + CHIP_RV530, + CHIP_RV560, + CHIP_RV570, + CHIP_R580, + CHIP_R600, + CHIP_RV610, + CHIP_RV630, + CHIP_RV670, + CHIP_RV620, + CHIP_RV635, + CHIP_RS780, + CHIP_RS880, + CHIP_RV770, + CHIP_RV730, + CHIP_RV710, + CHIP_RV740, + CHIP_LAST, +}; + +/* + * Chip flags + */ +enum radeon_chip_flags { + RADEON_FAMILY_MASK = 0x0000ffffUL, + RADEON_FLAGS_MASK = 0xffff0000UL, + RADEON_IS_MOBILITY = 0x00010000UL, + RADEON_IS_IGP = 0x00020000UL, + RADEON_SINGLE_CRTC = 0x00040000UL, + RADEON_IS_AGP = 0x00080000UL, + RADEON_HAS_HIERZ = 0x00100000UL, + RADEON_IS_PCIE = 0x00200000UL, + RADEON_NEW_MEMMAP = 0x00400000UL, + RADEON_IS_PCI = 0x00800000UL, + RADEON_IS_IGPGART = 0x01000000UL, +}; +#endif diff --git a/drivers/gpu/drm/radeon/radeon_fb.c b/drivers/gpu/drm/radeon/radeon_fb.c index 944e4fa..b38c4c8 100644 --- a/drivers/gpu/drm/radeon/radeon_fb.c +++ b/drivers/gpu/drm/radeon/radeon_fb.c @@ -55,6 +55,7 @@ static struct fb_ops radeonfb_ops = { .fb_imageblit = cfb_imageblit, .fb_pan_display = drm_fb_helper_pan_display, .fb_blank = drm_fb_helper_blank, + .fb_setcmap = drm_fb_helper_setcmap, }; /** @@ -123,11 +124,13 @@ static int radeon_align_pitch(struct radeon_device *rdev, int width, int bpp, bo static struct drm_fb_helper_funcs radeon_fb_helper_funcs = { .gamma_set = radeon_crtc_fb_gamma_set, + .gamma_get = radeon_crtc_fb_gamma_get, }; int radeonfb_create(struct drm_device *dev, uint32_t fb_width, uint32_t fb_height, uint32_t surface_width, uint32_t surface_height, + uint32_t surface_depth, uint32_t surface_bpp, struct drm_framebuffer **fb_p) { struct radeon_device *rdev = dev->dev_private; @@ -145,13 +148,19 @@ int radeonfb_create(struct drm_device *dev, unsigned long tmp; bool fb_tiled = false; /* useful for testing */ u32 tiling_flags = 0; + int crtc_count; mode_cmd.width = surface_width; mode_cmd.height = surface_height; - mode_cmd.bpp = 32; + + /* avivo can't scanout real 24bpp */ + if ((surface_bpp == 24) && ASIC_IS_AVIVO(rdev)) + surface_bpp = 32; + + mode_cmd.bpp = surface_bpp; /* need to align pitch with crtc limits */ mode_cmd.pitch = radeon_align_pitch(rdev, mode_cmd.width, mode_cmd.bpp, fb_tiled) * ((mode_cmd.bpp + 1) / 8); - mode_cmd.depth = 24; + mode_cmd.depth = surface_depth; size = mode_cmd.pitch * mode_cmd.height; aligned_size = ALIGN(size, PAGE_SIZE); @@ -216,7 +225,11 @@ int radeonfb_create(struct drm_device *dev, rfbdev = info->par; rfbdev->helper.funcs = &radeon_fb_helper_funcs; rfbdev->helper.dev = dev; - ret = drm_fb_helper_init_crtc_count(&rfbdev->helper, 2, + if (rdev->flags & RADEON_SINGLE_CRTC) + crtc_count = 1; + else + crtc_count = 2; + ret = drm_fb_helper_init_crtc_count(&rfbdev->helper, crtc_count, RADEONFB_CONN_LIMIT); if (ret) goto out_unref; @@ -233,7 +246,7 @@ int radeonfb_create(struct drm_device *dev, strcpy(info->fix.id, "radeondrmfb"); - drm_fb_helper_fill_fix(info, fb->pitch); + drm_fb_helper_fill_fix(info, fb->pitch, fb->depth); info->flags = FBINFO_DEFAULT; info->fbops = &radeonfb_ops; @@ -290,13 +303,26 @@ out: return ret; } +static char *mode_option; +int radeon_parse_options(char *options) +{ + char *this_opt; + + if (!options || !*options) + return 0; + + while ((this_opt = strsep(&options, ",")) != NULL) { + if (!*this_opt) + continue; + mode_option = this_opt; + } + return 0; +} + int radeonfb_probe(struct drm_device *dev) { - int ret; - ret = drm_fb_helper_single_fb_probe(dev, &radeonfb_create); - return ret; + return drm_fb_helper_single_fb_probe(dev, 32, &radeonfb_create); } -EXPORT_SYMBOL(radeonfb_probe); int radeonfb_remove(struct drm_device *dev, struct drm_framebuffer *fb) { diff --git a/drivers/gpu/drm/radeon/radeon_gart.c b/drivers/gpu/drm/radeon/radeon_gart.c index a931af0..a68d756 100644 --- a/drivers/gpu/drm/radeon/radeon_gart.c +++ b/drivers/gpu/drm/radeon/radeon_gart.c @@ -140,15 +140,15 @@ void radeon_gart_unbind(struct radeon_device *rdev, unsigned offset, WARN(1, "trying to unbind memory to unitialized GART !\n"); return; } - t = offset / 4096; - p = t / (PAGE_SIZE / 4096); + t = offset / RADEON_GPU_PAGE_SIZE; + p = t / (PAGE_SIZE / RADEON_GPU_PAGE_SIZE); for (i = 0; i < pages; i++, p++) { if (rdev->gart.pages[p]) { pci_unmap_page(rdev->pdev, rdev->gart.pages_addr[p], PAGE_SIZE, PCI_DMA_BIDIRECTIONAL); rdev->gart.pages[p] = NULL; rdev->gart.pages_addr[p] = 0; - for (j = 0; j < (PAGE_SIZE / 4096); j++, t++) { + for (j = 0; j < (PAGE_SIZE / RADEON_GPU_PAGE_SIZE); j++, t++) { radeon_gart_set_page(rdev, t, 0); } } @@ -169,8 +169,8 @@ int radeon_gart_bind(struct radeon_device *rdev, unsigned offset, DRM_ERROR("trying to bind memory to unitialized GART !\n"); return -EINVAL; } - t = offset / 4096; - p = t / (PAGE_SIZE / 4096); + t = offset / RADEON_GPU_PAGE_SIZE; + p = t / (PAGE_SIZE / RADEON_GPU_PAGE_SIZE); for (i = 0; i < pages; i++, p++) { /* we need to support large memory configurations */ @@ -185,9 +185,9 @@ int radeon_gart_bind(struct radeon_device *rdev, unsigned offset, } rdev->gart.pages[p] = pagelist[i]; page_base = rdev->gart.pages_addr[p]; - for (j = 0; j < (PAGE_SIZE / 4096); j++, t++) { + for (j = 0; j < (PAGE_SIZE / RADEON_GPU_PAGE_SIZE); j++, t++) { radeon_gart_set_page(rdev, t, page_base); - page_base += 4096; + page_base += RADEON_GPU_PAGE_SIZE; } } mb(); @@ -200,14 +200,14 @@ int radeon_gart_init(struct radeon_device *rdev) if (rdev->gart.pages) { return 0; } - /* We need PAGE_SIZE >= 4096 */ - if (PAGE_SIZE < 4096) { + /* We need PAGE_SIZE >= RADEON_GPU_PAGE_SIZE */ + if (PAGE_SIZE < RADEON_GPU_PAGE_SIZE) { DRM_ERROR("Page size is smaller than GPU page size!\n"); return -EINVAL; } /* Compute table size */ rdev->gart.num_cpu_pages = rdev->mc.gtt_size / PAGE_SIZE; - rdev->gart.num_gpu_pages = rdev->mc.gtt_size / 4096; + rdev->gart.num_gpu_pages = rdev->mc.gtt_size / RADEON_GPU_PAGE_SIZE; DRM_INFO("GART: num cpu pages %u, num gpu pages %u\n", rdev->gart.num_cpu_pages, rdev->gart.num_gpu_pages); /* Allocate pages table */ diff --git a/drivers/gpu/drm/radeon/radeon_irq_kms.c b/drivers/gpu/drm/radeon/radeon_irq_kms.c index 1841145..a0fe623 100644 --- a/drivers/gpu/drm/radeon/radeon_irq_kms.c +++ b/drivers/gpu/drm/radeon/radeon_irq_kms.c @@ -83,11 +83,22 @@ void radeon_driver_irq_uninstall_kms(struct drm_device *dev) int radeon_irq_kms_init(struct radeon_device *rdev) { int r = 0; + int num_crtc = 2; - r = drm_vblank_init(rdev->ddev, 2); + if (rdev->flags & RADEON_SINGLE_CRTC) + num_crtc = 1; + + r = drm_vblank_init(rdev->ddev, num_crtc); if (r) { return r; } + /* enable msi */ + rdev->msi_enabled = 0; + if (rdev->family >= CHIP_RV380) { + int ret = pci_enable_msi(rdev->pdev); + if (!ret) + rdev->msi_enabled = 1; + } drm_irq_install(rdev->ddev); rdev->irq.installed = true; DRM_INFO("radeon: irq initialized.\n"); @@ -99,5 +110,7 @@ void radeon_irq_kms_fini(struct radeon_device *rdev) if (rdev->irq.installed) { rdev->irq.installed = false; drm_irq_uninstall(rdev->ddev); + if (rdev->msi_enabled) + pci_disable_msi(rdev->pdev); } } diff --git a/drivers/gpu/drm/radeon/radeon_kms.c b/drivers/gpu/drm/radeon/radeon_kms.c index 709bd89..ba12862 100644 --- a/drivers/gpu/drm/radeon/radeon_kms.c +++ b/drivers/gpu/drm/radeon/radeon_kms.c @@ -201,55 +201,6 @@ void radeon_disable_vblank_kms(struct drm_device *dev, int crtc) /* - * For multiple master (like multiple X). - */ -struct drm_radeon_master_private { - drm_local_map_t *sarea; - drm_radeon_sarea_t *sarea_priv; -}; - -int radeon_master_create_kms(struct drm_device *dev, struct drm_master *master) -{ - struct drm_radeon_master_private *master_priv; - unsigned long sareapage; - int ret; - - master_priv = kzalloc(sizeof(*master_priv), GFP_KERNEL); - if (master_priv == NULL) { - return -ENOMEM; - } - /* prebuild the SAREA */ - sareapage = max_t(unsigned long, SAREA_MAX, PAGE_SIZE); - ret = drm_addmap(dev, 0, sareapage, _DRM_SHM, - _DRM_CONTAINS_LOCK, - &master_priv->sarea); - if (ret) { - DRM_ERROR("SAREA setup failed\n"); - return ret; - } - master_priv->sarea_priv = master_priv->sarea->handle + sizeof(struct drm_sarea); - master_priv->sarea_priv->pfCurrentPage = 0; - master->driver_priv = master_priv; - return 0; -} - -void radeon_master_destroy_kms(struct drm_device *dev, - struct drm_master *master) -{ - struct drm_radeon_master_private *master_priv = master->driver_priv; - - if (master_priv == NULL) { - return; - } - if (master_priv->sarea) { - drm_rmmap_locked(dev, master_priv->sarea); - } - kfree(master_priv); - master->driver_priv = NULL; -} - - -/* * IOCTL. */ int radeon_dma_ioctl_kms(struct drm_device *dev, void *data, diff --git a/drivers/gpu/drm/radeon/radeon_legacy_crtc.c b/drivers/gpu/drm/radeon/radeon_legacy_crtc.c index 2b997a1..8d0b7aa 100644 --- a/drivers/gpu/drm/radeon/radeon_legacy_crtc.c +++ b/drivers/gpu/drm/radeon/radeon_legacy_crtc.c @@ -48,7 +48,7 @@ static void radeon_legacy_rmx_mode_set(struct drm_crtc *crtc, u32 fp_horz_stretch, fp_vert_stretch, fp_horz_vert_active; u32 fp_h_sync_strt_wid, fp_crtc_h_total_disp; u32 fp_v_sync_strt_wid, fp_crtc_v_total_disp; - struct radeon_native_mode *native_mode = &radeon_crtc->native_mode; + struct drm_display_mode *native_mode = &radeon_crtc->native_mode; fp_vert_stretch = RREG32(RADEON_FP_VERT_STRETCH) & (RADEON_VERT_STRETCH_RESERVED | @@ -95,19 +95,19 @@ static void radeon_legacy_rmx_mode_set(struct drm_crtc *crtc, fp_horz_vert_active = 0; - if (native_mode->panel_xres == 0 || - native_mode->panel_yres == 0) { + if (native_mode->hdisplay == 0 || + native_mode->vdisplay == 0) { hscale = false; vscale = false; } else { - if (xres > native_mode->panel_xres) - xres = native_mode->panel_xres; - if (yres > native_mode->panel_yres) - yres = native_mode->panel_yres; + if (xres > native_mode->hdisplay) + xres = native_mode->hdisplay; + if (yres > native_mode->vdisplay) + yres = native_mode->vdisplay; - if (xres == native_mode->panel_xres) + if (xres == native_mode->hdisplay) hscale = false; - if (yres == native_mode->panel_yres) + if (yres == native_mode->vdisplay) vscale = false; } @@ -119,11 +119,11 @@ static void radeon_legacy_rmx_mode_set(struct drm_crtc *crtc, else { inc = (fp_horz_stretch & RADEON_HORZ_AUTO_RATIO_INC) ? 1 : 0; scale = ((xres + inc) * RADEON_HORZ_STRETCH_RATIO_MAX) - / native_mode->panel_xres + 1; + / native_mode->hdisplay + 1; fp_horz_stretch |= (((scale) & RADEON_HORZ_STRETCH_RATIO_MASK) | RADEON_HORZ_STRETCH_BLEND | RADEON_HORZ_STRETCH_ENABLE | - ((native_mode->panel_xres/8-1) << 16)); + ((native_mode->hdisplay/8-1) << 16)); } if (!vscale) @@ -131,11 +131,11 @@ static void radeon_legacy_rmx_mode_set(struct drm_crtc *crtc, else { inc = (fp_vert_stretch & RADEON_VERT_AUTO_RATIO_INC) ? 1 : 0; scale = ((yres + inc) * RADEON_VERT_STRETCH_RATIO_MAX) - / native_mode->panel_yres + 1; + / native_mode->vdisplay + 1; fp_vert_stretch |= (((scale) & RADEON_VERT_STRETCH_RATIO_MASK) | RADEON_VERT_STRETCH_ENABLE | RADEON_VERT_STRETCH_BLEND | - ((native_mode->panel_yres-1) << 12)); + ((native_mode->vdisplay-1) << 12)); } break; case RMX_CENTER: @@ -175,8 +175,8 @@ static void radeon_legacy_rmx_mode_set(struct drm_crtc *crtc, ? RADEON_CRTC_V_SYNC_POL : 0))); - fp_horz_vert_active = (((native_mode->panel_yres) & 0xfff) | - (((native_mode->panel_xres / 8) & 0x1ff) << 16)); + fp_horz_vert_active = (((native_mode->vdisplay) & 0xfff) | + (((native_mode->hdisplay / 8) & 0x1ff) << 16)); break; case RMX_OFF: default: @@ -532,6 +532,10 @@ int radeon_crtc_set_base(struct drm_crtc *crtc, int x, int y, radeon_fb = to_radeon_framebuffer(old_fb); radeon_gem_object_unpin(radeon_fb->obj); } + + /* Bytes per pixel may have changed */ + radeon_bandwidth_update(rdev); + return 0; } @@ -664,6 +668,9 @@ static bool radeon_set_crtc_timing(struct drm_crtc *crtc, struct drm_display_mod WREG32(RADEON_DISP2_MERGE_CNTL, disp2_merge_cntl); WREG32(RADEON_CRTC2_GEN_CNTL, crtc2_gen_cntl); + + WREG32(RADEON_FP_H2_SYNC_STRT_WID, crtc_h_sync_strt_wid); + WREG32(RADEON_FP_V2_SYNC_STRT_WID, crtc_v_sync_strt_wid); } else { uint32_t crtc_gen_cntl; uint32_t crtc_ext_cntl; @@ -1015,14 +1022,11 @@ static int radeon_crtc_mode_set(struct drm_crtc *crtc, int x, int y, struct drm_framebuffer *old_fb) { struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); - struct drm_device *dev = crtc->dev; - struct radeon_device *rdev = dev->dev_private; /* TODO TV */ radeon_crtc_set_base(crtc, x, y, old_fb); radeon_set_crtc_timing(crtc, adjusted_mode); radeon_set_pll(crtc, adjusted_mode); - radeon_bandwidth_update(rdev); if (radeon_crtc->crtc_id == 0) { radeon_legacy_rmx_mode_set(crtc, mode, adjusted_mode); } else { @@ -1053,6 +1057,7 @@ static const struct drm_crtc_helper_funcs legacy_helper_funcs = { .mode_set_base = radeon_crtc_set_base, .prepare = radeon_crtc_prepare, .commit = radeon_crtc_commit, + .load_lut = radeon_crtc_load_lut, }; diff --git a/drivers/gpu/drm/radeon/radeon_legacy_encoders.c b/drivers/gpu/drm/radeon/radeon_legacy_encoders.c index b1547f7..0038212 100644 --- a/drivers/gpu/drm/radeon/radeon_legacy_encoders.c +++ b/drivers/gpu/drm/radeon/radeon_legacy_encoders.c @@ -107,8 +107,6 @@ static void radeon_legacy_lvds_prepare(struct drm_encoder *encoder) else radeon_combios_output_lock(encoder, true); radeon_legacy_lvds_dpms(encoder, DRM_MODE_DPMS_OFF); - - radeon_encoder_set_active_device(encoder); } static void radeon_legacy_lvds_commit(struct drm_encoder *encoder) @@ -192,6 +190,8 @@ static bool radeon_legacy_lvds_mode_fixup(struct drm_encoder *encoder, { struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + /* set the active encoder to connector routing */ + radeon_encoder_set_active_device(encoder); drm_mode_set_crtcinfo(adjusted_mode, 0); if (radeon_encoder->rmx_type != RMX_OFF) @@ -218,7 +218,8 @@ static bool radeon_legacy_primary_dac_mode_fixup(struct drm_encoder *encoder, struct drm_display_mode *mode, struct drm_display_mode *adjusted_mode) { - + /* set the active encoder to connector routing */ + radeon_encoder_set_active_device(encoder); drm_mode_set_crtcinfo(adjusted_mode, 0); return true; @@ -272,7 +273,6 @@ static void radeon_legacy_primary_dac_prepare(struct drm_encoder *encoder) else radeon_combios_output_lock(encoder, true); radeon_legacy_primary_dac_dpms(encoder, DRM_MODE_DPMS_OFF); - radeon_encoder_set_active_device(encoder); } static void radeon_legacy_primary_dac_commit(struct drm_encoder *encoder) @@ -468,7 +468,6 @@ static void radeon_legacy_tmds_int_prepare(struct drm_encoder *encoder) else radeon_combios_output_lock(encoder, true); radeon_legacy_tmds_int_dpms(encoder, DRM_MODE_DPMS_OFF); - radeon_encoder_set_active_device(encoder); } static void radeon_legacy_tmds_int_commit(struct drm_encoder *encoder) @@ -543,6 +542,14 @@ static void radeon_legacy_tmds_int_mode_set(struct drm_encoder *encoder, fp_gen_cntl &= ~(RADEON_FP_FPON | RADEON_FP_TMDS_EN); + fp_gen_cntl &= ~(RADEON_FP_RMX_HVSYNC_CONTROL_EN | + RADEON_FP_DFP_SYNC_SEL | + RADEON_FP_CRT_SYNC_SEL | + RADEON_FP_CRTC_LOCK_8DOT | + RADEON_FP_USE_SHADOW_EN | + RADEON_FP_CRTC_USE_SHADOW_VEND | + RADEON_FP_CRT_SYNC_ALT); + if (1) /* FIXME rgbBits == 8 */ fp_gen_cntl |= RADEON_FP_PANEL_FORMAT; /* 24 bit format */ else @@ -556,7 +563,7 @@ static void radeon_legacy_tmds_int_mode_set(struct drm_encoder *encoder, else fp_gen_cntl |= R200_FP_SOURCE_SEL_CRTC1; } else - fp_gen_cntl |= RADEON_FP_SEL_CRTC1; + fp_gen_cntl &= ~RADEON_FP_SEL_CRTC2; } else { if (ASIC_IS_R300(rdev) || rdev->family == CHIP_R200) { fp_gen_cntl &= ~R200_FP_SOURCE_SEL_MASK; @@ -593,7 +600,8 @@ static bool radeon_legacy_tmds_ext_mode_fixup(struct drm_encoder *encoder, struct drm_display_mode *mode, struct drm_display_mode *adjusted_mode) { - + /* set the active encoder to connector routing */ + radeon_encoder_set_active_device(encoder); drm_mode_set_crtcinfo(adjusted_mode, 0); return true; @@ -636,7 +644,6 @@ static void radeon_legacy_tmds_ext_prepare(struct drm_encoder *encoder) else radeon_combios_output_lock(encoder, true); radeon_legacy_tmds_ext_dpms(encoder, DRM_MODE_DPMS_OFF); - radeon_encoder_set_active_device(encoder); } static void radeon_legacy_tmds_ext_commit(struct drm_encoder *encoder) @@ -735,7 +742,8 @@ static bool radeon_legacy_tv_dac_mode_fixup(struct drm_encoder *encoder, struct drm_display_mode *mode, struct drm_display_mode *adjusted_mode) { - + /* set the active encoder to connector routing */ + radeon_encoder_set_active_device(encoder); drm_mode_set_crtcinfo(adjusted_mode, 0); return true; @@ -839,7 +847,6 @@ static void radeon_legacy_tv_dac_prepare(struct drm_encoder *encoder) else radeon_combios_output_lock(encoder, true); radeon_legacy_tv_dac_dpms(encoder, DRM_MODE_DPMS_OFF); - radeon_encoder_set_active_device(encoder); } static void radeon_legacy_tv_dac_commit(struct drm_encoder *encoder) @@ -881,7 +888,7 @@ static void radeon_legacy_tv_dac_mode_set(struct drm_encoder *encoder, R420_TV_DAC_DACADJ_MASK | R420_TV_DAC_RDACPD | R420_TV_DAC_GDACPD | - R420_TV_DAC_GDACPD | + R420_TV_DAC_BDACPD | R420_TV_DAC_TVENABLE); } else { tv_dac_cntl &= ~(RADEON_TV_DAC_STD_MASK | @@ -889,7 +896,7 @@ static void radeon_legacy_tv_dac_mode_set(struct drm_encoder *encoder, RADEON_TV_DAC_DACADJ_MASK | RADEON_TV_DAC_RDACPD | RADEON_TV_DAC_GDACPD | - RADEON_TV_DAC_GDACPD); + RADEON_TV_DAC_BDACPD); } /* FIXME TV */ @@ -1318,7 +1325,10 @@ radeon_add_legacy_encoder(struct drm_device *dev, uint32_t encoder_id, uint32_t return; encoder = &radeon_encoder->base; - encoder->possible_crtcs = 0x3; + if (rdev->flags & RADEON_SINGLE_CRTC) + encoder->possible_crtcs = 0x1; + else + encoder->possible_crtcs = 0x3; encoder->possible_clones = 0; radeon_encoder->enc_priv = NULL; diff --git a/drivers/gpu/drm/radeon/radeon_mode.h b/drivers/gpu/drm/radeon/radeon_mode.h index 570a587..ace726a 100644 --- a/drivers/gpu/drm/radeon/radeon_mode.h +++ b/drivers/gpu/drm/radeon/radeon_mode.h @@ -172,6 +172,7 @@ enum radeon_connector_table { struct radeon_mode_info { struct atom_context *atom_context; + struct card_info *atom_card_info; enum radeon_connector_table connector_table; bool mode_config_initialized; struct radeon_crtc *crtcs[2]; @@ -186,17 +187,6 @@ struct radeon_mode_info { }; -struct radeon_native_mode { - /* preferred mode */ - uint32_t panel_xres, panel_yres; - uint32_t hoverplus, hsync_width; - uint32_t hblank; - uint32_t voverplus, vsync_width; - uint32_t vblank; - uint32_t dotclock; - uint32_t flags; -}; - #define MAX_H_CODE_TIMING_LEN 32 #define MAX_V_CODE_TIMING_LEN 32 @@ -228,7 +218,7 @@ struct radeon_crtc { enum radeon_rmx_type rmx_type; fixed20_12 vsc; fixed20_12 hsc; - struct radeon_native_mode native_mode; + struct drm_display_mode native_mode; }; struct radeon_encoder_primary_dac { @@ -248,7 +238,7 @@ struct radeon_encoder_lvds { bool use_bios_dividers; uint32_t lvds_gen_cntl; /* panel mode */ - struct radeon_native_mode native_mode; + struct drm_display_mode native_mode; }; struct radeon_encoder_tv_dac { @@ -271,6 +261,16 @@ struct radeon_encoder_int_tmds { struct radeon_tmds_pll tmds_pll[4]; }; +/* spread spectrum */ +struct radeon_atom_ss { + uint16_t percentage; + uint8_t type; + uint8_t step; + uint8_t delay; + uint8_t range; + uint8_t refdiv; +}; + struct radeon_encoder_atom_dig { /* atom dig */ bool coherent_mode; @@ -278,8 +278,9 @@ struct radeon_encoder_atom_dig { /* atom lvds */ uint32_t lvds_misc; uint16_t panel_pwr_delay; + struct radeon_atom_ss *ss; /* panel mode */ - struct radeon_native_mode native_mode; + struct drm_display_mode native_mode; }; struct radeon_encoder_atom_dac { @@ -294,7 +295,7 @@ struct radeon_encoder { uint32_t flags; uint32_t pixel_clock; enum radeon_rmx_type rmx_type; - struct radeon_native_mode native_mode; + struct drm_display_mode native_mode; void *enc_priv; }; @@ -308,12 +309,15 @@ struct radeon_connector { uint32_t connector_id; uint32_t devices; struct radeon_i2c_chan *ddc_bus; + /* some systems have a an hdmi and vga port with a shared ddc line */ + bool shared_ddc; bool use_digital; /* we need to mind the EDID between detect and get modes due to analog/digital/tvencoder */ struct edid *edid; void *con_priv; bool dac_load_detect; + uint16_t connector_object_id; }; struct radeon_framebuffer { @@ -407,6 +411,8 @@ extern void radeon_combios_encoder_dpms_scratch_regs(struct drm_encoder *encoder, bool on); extern void radeon_crtc_fb_gamma_set(struct drm_crtc *crtc, u16 red, u16 green, u16 blue, int regno); +extern void radeon_crtc_fb_gamma_get(struct drm_crtc *crtc, u16 *red, u16 *green, + u16 *blue, int regno); struct drm_framebuffer *radeon_framebuffer_create(struct drm_device *dev, struct drm_mode_fb_cmd *mode_cmd, struct drm_gem_object *obj); diff --git a/drivers/gpu/drm/radeon/radeon_object.c b/drivers/gpu/drm/radeon/radeon_object.c index 73af463..1f056da 100644 --- a/drivers/gpu/drm/radeon/radeon_object.c +++ b/drivers/gpu/drm/radeon/radeon_object.c @@ -400,11 +400,9 @@ void radeon_object_list_add_object(struct radeon_object_list *lobj, int radeon_object_list_reserve(struct list_head *head) { struct radeon_object_list *lobj; - struct list_head *i; int r; - list_for_each(i, head) { - lobj = list_entry(i, struct radeon_object_list, list); + list_for_each_entry(lobj, head, list){ if (!lobj->robj->pin_count) { r = radeon_object_reserve(lobj->robj, true); if (unlikely(r != 0)) { @@ -420,13 +418,10 @@ int radeon_object_list_reserve(struct list_head *head) void radeon_object_list_unreserve(struct list_head *head) { struct radeon_object_list *lobj; - struct list_head *i; - list_for_each(i, head) { - lobj = list_entry(i, struct radeon_object_list, list); + list_for_each_entry(lobj, head, list) { if (!lobj->robj->pin_count) { radeon_object_unreserve(lobj->robj); - } else { } } } @@ -436,7 +431,6 @@ int radeon_object_list_validate(struct list_head *head, void *fence) struct radeon_object_list *lobj; struct radeon_object *robj; struct radeon_fence *old_fence = NULL; - struct list_head *i; int r; r = radeon_object_list_reserve(head); @@ -444,8 +438,7 @@ int radeon_object_list_validate(struct list_head *head, void *fence) radeon_object_list_unreserve(head); return r; } - list_for_each(i, head) { - lobj = list_entry(i, struct radeon_object_list, list); + list_for_each_entry(lobj, head, list) { robj = lobj->robj; if (!robj->pin_count) { if (lobj->wdomain) { @@ -482,10 +475,8 @@ void radeon_object_list_unvalidate(struct list_head *head) { struct radeon_object_list *lobj; struct radeon_fence *old_fence = NULL; - struct list_head *i; - list_for_each(i, head) { - lobj = list_entry(i, struct radeon_object_list, list); + list_for_each_entry(lobj, head, list) { old_fence = (struct radeon_fence *)lobj->robj->tobj.sync_obj; lobj->robj->tobj.sync_obj = NULL; if (old_fence) { diff --git a/drivers/gpu/drm/radeon/radeon_pm.c b/drivers/gpu/drm/radeon/radeon_pm.c new file mode 100644 index 0000000..46146c6 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_pm.c @@ -0,0 +1,65 @@ +/* + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Rafał Miłecki <zajec5@gmail.com> + */ +#include "drmP.h" +#include "radeon.h" + +int radeon_debugfs_pm_init(struct radeon_device *rdev); + +int radeon_pm_init(struct radeon_device *rdev) +{ + if (radeon_debugfs_pm_init(rdev)) { + DRM_ERROR("Failed to register debugfs file for CP !\n"); + } + + return 0; +} + +/* + * Debugfs info + */ +#if defined(CONFIG_DEBUG_FS) + +static int radeon_debugfs_pm_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + struct radeon_device *rdev = dev->dev_private; + + seq_printf(m, "engine clock: %u0 Hz\n", radeon_get_engine_clock(rdev)); + seq_printf(m, "memory clock: %u0 Hz\n", radeon_get_memory_clock(rdev)); + + return 0; +} + +static struct drm_info_list radeon_pm_info_list[] = { + {"radeon_pm_info", radeon_debugfs_pm_info, 0, NULL}, +}; +#endif + +int radeon_debugfs_pm_init(struct radeon_device *rdev) +{ +#if defined(CONFIG_DEBUG_FS) + return radeon_debugfs_add_files(rdev, radeon_pm_info_list, ARRAY_SIZE(radeon_pm_info_list)); +#else + return 0; +#endif +} diff --git a/drivers/gpu/drm/radeon/radeon_reg.h b/drivers/gpu/drm/radeon/radeon_reg.h index 21da871..29ab759 100644 --- a/drivers/gpu/drm/radeon/radeon_reg.h +++ b/drivers/gpu/drm/radeon/radeon_reg.h @@ -290,6 +290,8 @@ #define RADEON_BUS_CNTL 0x0030 # define RADEON_BUS_MASTER_DIS (1 << 6) # define RADEON_BUS_BIOS_DIS_ROM (1 << 12) +# define RS600_BUS_MASTER_DIS (1 << 14) +# define RS600_MSI_REARM (1 << 20) /* rs600/rs690/rs740 */ # define RADEON_BUS_RD_DISCARD_EN (1 << 24) # define RADEON_BUS_RD_ABORT_EN (1 << 25) # define RADEON_BUS_MSTR_DISCONNECT_EN (1 << 28) @@ -297,6 +299,9 @@ # define RADEON_BUS_READ_BURST (1 << 30) #define RADEON_BUS_CNTL1 0x0034 # define RADEON_BUS_WAIT_ON_LOCK_EN (1 << 4) +/* rv370/rv380, rv410, r423/r430/r480, r5xx */ +#define RADEON_MSI_REARM_EN 0x0160 +# define RV370_MSI_REARM_EN (1 << 0) /* #define RADEON_PCIE_INDEX 0x0030 */ /* #define RADEON_PCIE_DATA 0x0034 */ @@ -3311,6 +3316,7 @@ #define RADEON_AIC_CNTL 0x01d0 # define RADEON_PCIGART_TRANSLATE_EN (1 << 0) # define RADEON_DIS_OUT_OF_PCI_GART_ACCESS (1 << 1) +# define RS400_MSI_REARM (1 << 3) /* rs400/rs480 */ #define RADEON_AIC_LO_ADDR 0x01dc #define RADEON_AIC_PT_BASE 0x01d8 #define RADEON_AIC_HI_ADDR 0x01e0 @@ -3333,6 +3339,7 @@ # define RADEON_CP_PACKET_MAX_DWORDS (1 << 12) # define RADEON_CP_PACKET0_REG_MASK 0x000007ff # define R300_CP_PACKET0_REG_MASK 0x00001fff +# define R600_CP_PACKET0_REG_MASK 0x0000ffff # define RADEON_CP_PACKET1_REG0_MASK 0x000007ff # define RADEON_CP_PACKET1_REG1_MASK 0x003ff800 diff --git a/drivers/gpu/drm/radeon/radeon_test.c b/drivers/gpu/drm/radeon/radeon_test.c index 03c33cf..f8a465d 100644 --- a/drivers/gpu/drm/radeon/radeon_test.c +++ b/drivers/gpu/drm/radeon/radeon_test.c @@ -42,7 +42,7 @@ void radeon_test_moves(struct radeon_device *rdev) /* Number of tests = * (Total GTT - IB pool - writeback page - ring buffer) / test size */ - n = (rdev->mc.gtt_size - RADEON_IB_POOL_SIZE*64*1024 - 4096 - + n = (rdev->mc.gtt_size - RADEON_IB_POOL_SIZE*64*1024 - RADEON_GPU_PAGE_SIZE - rdev->cp.ring_size) / size; gtt_obj = kzalloc(n * sizeof(*gtt_obj), GFP_KERNEL); @@ -102,7 +102,7 @@ void radeon_test_moves(struct radeon_device *rdev) goto out_cleanup; } - r = radeon_copy(rdev, gtt_addr, vram_addr, size / 4096, fence); + r = radeon_copy(rdev, gtt_addr, vram_addr, size / RADEON_GPU_PAGE_SIZE, fence); if (r) { DRM_ERROR("Failed GTT->VRAM copy %d\n", i); goto out_cleanup; @@ -145,7 +145,7 @@ void radeon_test_moves(struct radeon_device *rdev) goto out_cleanup; } - r = radeon_copy(rdev, vram_addr, gtt_addr, size / 4096, fence); + r = radeon_copy(rdev, vram_addr, gtt_addr, size / RADEON_GPU_PAGE_SIZE, fence); if (r) { DRM_ERROR("Failed VRAM->GTT copy %d\n", i); goto out_cleanup; diff --git a/drivers/gpu/drm/radeon/radeon_ttm.c b/drivers/gpu/drm/radeon/radeon_ttm.c index acd889c..1381e06 100644 --- a/drivers/gpu/drm/radeon/radeon_ttm.c +++ b/drivers/gpu/drm/radeon/radeon_ttm.c @@ -295,6 +295,12 @@ static int radeon_move_vram_ram(struct ttm_buffer_object *bo, if (unlikely(r)) { return r; } + + r = ttm_tt_set_placement_caching(bo->ttm, tmp_mem.placement); + if (unlikely(r)) { + goto out_cleanup; + } + r = ttm_tt_bind(bo->ttm, &tmp_mem); if (unlikely(r)) { goto out_cleanup; @@ -530,7 +536,7 @@ void radeon_ttm_fini(struct radeon_device *rdev) } static struct vm_operations_struct radeon_ttm_vm_ops; -static struct vm_operations_struct *ttm_vm_ops = NULL; +static const struct vm_operations_struct *ttm_vm_ops = NULL; static int radeon_ttm_fault(struct vm_area_struct *vma, struct vm_fault *vmf) { @@ -689,9 +695,6 @@ struct ttm_backend *radeon_ttm_backend_create(struct radeon_device *rdev) #define RADEON_DEBUGFS_MEM_TYPES 2 -static struct drm_info_list radeon_mem_types_list[RADEON_DEBUGFS_MEM_TYPES]; -static char radeon_mem_types_names[RADEON_DEBUGFS_MEM_TYPES][32]; - #if defined(CONFIG_DEBUG_FS) static int radeon_mm_dump_table(struct seq_file *m, void *data) { @@ -711,9 +714,11 @@ static int radeon_mm_dump_table(struct seq_file *m, void *data) static int radeon_ttm_debugfs_init(struct radeon_device *rdev) { +#if defined(CONFIG_DEBUG_FS) + static struct drm_info_list radeon_mem_types_list[RADEON_DEBUGFS_MEM_TYPES]; + static char radeon_mem_types_names[RADEON_DEBUGFS_MEM_TYPES][32]; unsigned i; -#if defined(CONFIG_DEBUG_FS) for (i = 0; i < RADEON_DEBUGFS_MEM_TYPES; i++) { if (i == 0) sprintf(radeon_mem_types_names[i], "radeon_vram_mm"); diff --git a/drivers/gpu/drm/radeon/rs100d.h b/drivers/gpu/drm/radeon/rs100d.h new file mode 100644 index 0000000..48a913a --- /dev/null +++ b/drivers/gpu/drm/radeon/rs100d.h @@ -0,0 +1,40 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef __RS100D_H__ +#define __RS100D_H__ + +/* Registers */ +#define R_00015C_NB_TOM 0x00015C +#define S_00015C_MC_FB_START(x) (((x) & 0xFFFF) << 0) +#define G_00015C_MC_FB_START(x) (((x) >> 0) & 0xFFFF) +#define C_00015C_MC_FB_START 0xFFFF0000 +#define S_00015C_MC_FB_TOP(x) (((x) & 0xFFFF) << 16) +#define G_00015C_MC_FB_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_00015C_MC_FB_TOP 0x0000FFFF + +#endif diff --git a/drivers/gpu/drm/radeon/rs400.c b/drivers/gpu/drm/radeon/rs400.c index a3fbdad..ca03716 100644 --- a/drivers/gpu/drm/radeon/rs400.c +++ b/drivers/gpu/drm/radeon/rs400.c @@ -27,27 +27,12 @@ */ #include <linux/seq_file.h> #include <drm/drmP.h> -#include "radeon_reg.h" #include "radeon.h" +#include "rs400d.h" -/* rs400,rs480 depends on : */ -void r100_hdp_reset(struct radeon_device *rdev); -void r100_mc_disable_clients(struct radeon_device *rdev); -int r300_mc_wait_for_idle(struct radeon_device *rdev); -void r420_pipes_init(struct radeon_device *rdev); +/* This files gather functions specifics to : rs400,rs480 */ +static int rs400_debugfs_pcie_gart_info_init(struct radeon_device *rdev); -/* This files gather functions specifics to : - * rs400,rs480 - * - * Some of these functions might be used by newer ASICs. - */ -void rs400_gpu_init(struct radeon_device *rdev); -int rs400_debugfs_pcie_gart_info_init(struct radeon_device *rdev); - - -/* - * GART functions. - */ void rs400_gart_adjust_size(struct radeon_device *rdev) { /* Check gart size */ @@ -238,61 +223,6 @@ int rs400_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr) return 0; } - -/* - * MC functions. - */ -int rs400_mc_init(struct radeon_device *rdev) -{ - uint32_t tmp; - int r; - - if (r100_debugfs_rbbm_init(rdev)) { - DRM_ERROR("Failed to register debugfs file for RBBM !\n"); - } - - rs400_gpu_init(rdev); - rs400_gart_disable(rdev); - rdev->mc.gtt_location = rdev->mc.mc_vram_size; - rdev->mc.gtt_location += (rdev->mc.gtt_size - 1); - rdev->mc.gtt_location &= ~(rdev->mc.gtt_size - 1); - r = radeon_mc_setup(rdev); - if (r) { - return r; - } - - r100_mc_disable_clients(rdev); - if (r300_mc_wait_for_idle(rdev)) { - printk(KERN_WARNING "Failed to wait MC idle while " - "programming pipes. Bad things might happen.\n"); - } - - tmp = rdev->mc.vram_location + rdev->mc.mc_vram_size - 1; - tmp = REG_SET(RADEON_MC_FB_TOP, tmp >> 16); - tmp |= REG_SET(RADEON_MC_FB_START, rdev->mc.vram_location >> 16); - WREG32(RADEON_MC_FB_LOCATION, tmp); - tmp = RREG32(RADEON_HOST_PATH_CNTL) | RADEON_HP_LIN_RD_CACHE_DIS; - WREG32(RADEON_HOST_PATH_CNTL, tmp | RADEON_HDP_SOFT_RESET | RADEON_HDP_READ_BUFFER_INVALIDATE); - (void)RREG32(RADEON_HOST_PATH_CNTL); - WREG32(RADEON_HOST_PATH_CNTL, tmp); - (void)RREG32(RADEON_HOST_PATH_CNTL); - - return 0; -} - -void rs400_mc_fini(struct radeon_device *rdev) -{ -} - - -/* - * Global GPU functions - */ -void rs400_errata(struct radeon_device *rdev) -{ - rdev->pll_errata = 0; -} - void rs400_gpu_init(struct radeon_device *rdev) { /* FIXME: HDP same place on rs400 ? */ @@ -305,10 +235,6 @@ void rs400_gpu_init(struct radeon_device *rdev) } } - -/* - * VRAM info. - */ void rs400_vram_info(struct radeon_device *rdev) { rs400_gart_adjust_size(rdev); @@ -319,10 +245,6 @@ void rs400_vram_info(struct radeon_device *rdev) r100_vram_init_sizes(rdev); } - -/* - * Indirect registers accessor - */ uint32_t rs400_mc_rreg(struct radeon_device *rdev, uint32_t reg) { uint32_t r; @@ -340,10 +262,6 @@ void rs400_mc_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v) WREG32(RS480_NB_MC_INDEX, 0xff); } - -/* - * Debugfs info - */ #if defined(CONFIG_DEBUG_FS) static int rs400_debugfs_gart_info(struct seq_file *m, void *data) { @@ -419,7 +337,7 @@ static struct drm_info_list rs400_gart_info_list[] = { }; #endif -int rs400_debugfs_pcie_gart_info_init(struct radeon_device *rdev) +static int rs400_debugfs_pcie_gart_info_init(struct radeon_device *rdev) { #if defined(CONFIG_DEBUG_FS) return radeon_debugfs_add_files(rdev, rs400_gart_info_list, 1); @@ -427,3 +345,190 @@ int rs400_debugfs_pcie_gart_info_init(struct radeon_device *rdev) return 0; #endif } + +static int rs400_mc_init(struct radeon_device *rdev) +{ + int r; + u32 tmp; + + /* Setup GPU memory space */ + tmp = G_00015C_MC_FB_START(RREG32(R_00015C_NB_TOM)); + rdev->mc.vram_location = G_00015C_MC_FB_START(tmp) << 16; + rdev->mc.gtt_location = 0xFFFFFFFFUL; + r = radeon_mc_setup(rdev); + if (r) + return r; + return 0; +} + +void rs400_mc_program(struct radeon_device *rdev) +{ + struct r100_mc_save save; + + /* Stops all mc clients */ + r100_mc_stop(rdev, &save); + + /* Wait for mc idle */ + if (r300_mc_wait_for_idle(rdev)) + dev_warn(rdev->dev, "Wait MC idle timeout before updating MC.\n"); + WREG32(R_000148_MC_FB_LOCATION, + S_000148_MC_FB_START(rdev->mc.vram_start >> 16) | + S_000148_MC_FB_TOP(rdev->mc.vram_end >> 16)); + + r100_mc_resume(rdev, &save); +} + +static int rs400_startup(struct radeon_device *rdev) +{ + int r; + + rs400_mc_program(rdev); + /* Resume clock */ + r300_clock_startup(rdev); + /* Initialize GPU configuration (# pipes, ...) */ + rs400_gpu_init(rdev); + /* Initialize GART (initialize after TTM so we can allocate + * memory through TTM but finalize after TTM) */ + r = rs400_gart_enable(rdev); + if (r) + return r; + /* Enable IRQ */ + rdev->irq.sw_int = true; + r100_irq_set(rdev); + /* 1M ring buffer */ + r = r100_cp_init(rdev, 1024 * 1024); + if (r) { + dev_err(rdev->dev, "failled initializing CP (%d).\n", r); + return r; + } + r = r100_wb_init(rdev); + if (r) + dev_err(rdev->dev, "failled initializing WB (%d).\n", r); + r = r100_ib_init(rdev); + if (r) { + dev_err(rdev->dev, "failled initializing IB (%d).\n", r); + return r; + } + return 0; +} + +int rs400_resume(struct radeon_device *rdev) +{ + /* Make sur GART are not working */ + rs400_gart_disable(rdev); + /* Resume clock before doing reset */ + r300_clock_startup(rdev); + /* setup MC before calling post tables */ + rs400_mc_program(rdev); + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_gpu_reset(rdev)) { + dev_warn(rdev->dev, "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* post */ + radeon_combios_asic_init(rdev->ddev); + /* Resume clock after posting */ + r300_clock_startup(rdev); + return rs400_startup(rdev); +} + +int rs400_suspend(struct radeon_device *rdev) +{ + r100_cp_disable(rdev); + r100_wb_disable(rdev); + r100_irq_disable(rdev); + rs400_gart_disable(rdev); + return 0; +} + +void rs400_fini(struct radeon_device *rdev) +{ + rs400_suspend(rdev); + r100_cp_fini(rdev); + r100_wb_fini(rdev); + r100_ib_fini(rdev); + radeon_gem_fini(rdev); + rs400_gart_fini(rdev); + radeon_irq_kms_fini(rdev); + radeon_fence_driver_fini(rdev); + radeon_object_fini(rdev); + radeon_atombios_fini(rdev); + kfree(rdev->bios); + rdev->bios = NULL; +} + +int rs400_init(struct radeon_device *rdev) +{ + int r; + + /* Disable VGA */ + r100_vga_render_disable(rdev); + /* Initialize scratch registers */ + radeon_scratch_init(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + /* TODO: disable VGA need to use VGA request */ + /* BIOS*/ + if (!radeon_get_bios(rdev)) { + if (ASIC_IS_AVIVO(rdev)) + return -EINVAL; + } + if (rdev->is_atom_bios) { + dev_err(rdev->dev, "Expecting combios for RS400/RS480 GPU\n"); + return -EINVAL; + } else { + r = radeon_combios_init(rdev); + if (r) + return r; + } + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_gpu_reset(rdev)) { + dev_warn(rdev->dev, + "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* check if cards are posted or not */ + if (!radeon_card_posted(rdev) && rdev->bios) { + DRM_INFO("GPU not posted. posting now...\n"); + radeon_combios_asic_init(rdev->ddev); + } + /* Initialize clocks */ + radeon_get_clock_info(rdev->ddev); + /* Get vram informations */ + rs400_vram_info(rdev); + /* Initialize memory controller (also test AGP) */ + r = rs400_mc_init(rdev); + if (r) + return r; + /* Fence driver */ + r = radeon_fence_driver_init(rdev); + if (r) + return r; + r = radeon_irq_kms_init(rdev); + if (r) + return r; + /* Memory manager */ + r = radeon_object_init(rdev); + if (r) + return r; + r = rs400_gart_init(rdev); + if (r) + return r; + r300_set_reg_safe(rdev); + rdev->accel_working = true; + r = rs400_startup(rdev); + if (r) { + /* Somethings want wront with the accel init stop accel */ + dev_err(rdev->dev, "Disabling GPU acceleration\n"); + rs400_suspend(rdev); + r100_cp_fini(rdev); + r100_wb_fini(rdev); + r100_ib_fini(rdev); + rs400_gart_fini(rdev); + radeon_irq_kms_fini(rdev); + rdev->accel_working = false; + } + return 0; +} diff --git a/drivers/gpu/drm/radeon/rs400d.h b/drivers/gpu/drm/radeon/rs400d.h new file mode 100644 index 0000000..6d8bac5 --- /dev/null +++ b/drivers/gpu/drm/radeon/rs400d.h @@ -0,0 +1,160 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef __RS400D_H__ +#define __RS400D_H__ + +/* Registers */ +#define R_000148_MC_FB_LOCATION 0x000148 +#define S_000148_MC_FB_START(x) (((x) & 0xFFFF) << 0) +#define G_000148_MC_FB_START(x) (((x) >> 0) & 0xFFFF) +#define C_000148_MC_FB_START 0xFFFF0000 +#define S_000148_MC_FB_TOP(x) (((x) & 0xFFFF) << 16) +#define G_000148_MC_FB_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_000148_MC_FB_TOP 0x0000FFFF +#define R_00015C_NB_TOM 0x00015C +#define S_00015C_MC_FB_START(x) (((x) & 0xFFFF) << 0) +#define G_00015C_MC_FB_START(x) (((x) >> 0) & 0xFFFF) +#define C_00015C_MC_FB_START 0xFFFF0000 +#define S_00015C_MC_FB_TOP(x) (((x) & 0xFFFF) << 16) +#define G_00015C_MC_FB_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_00015C_MC_FB_TOP 0x0000FFFF +#define R_0007C0_CP_STAT 0x0007C0 +#define S_0007C0_MRU_BUSY(x) (((x) & 0x1) << 0) +#define G_0007C0_MRU_BUSY(x) (((x) >> 0) & 0x1) +#define C_0007C0_MRU_BUSY 0xFFFFFFFE +#define S_0007C0_MWU_BUSY(x) (((x) & 0x1) << 1) +#define G_0007C0_MWU_BUSY(x) (((x) >> 1) & 0x1) +#define C_0007C0_MWU_BUSY 0xFFFFFFFD +#define S_0007C0_RSIU_BUSY(x) (((x) & 0x1) << 2) +#define G_0007C0_RSIU_BUSY(x) (((x) >> 2) & 0x1) +#define C_0007C0_RSIU_BUSY 0xFFFFFFFB +#define S_0007C0_RCIU_BUSY(x) (((x) & 0x1) << 3) +#define G_0007C0_RCIU_BUSY(x) (((x) >> 3) & 0x1) +#define C_0007C0_RCIU_BUSY 0xFFFFFFF7 +#define S_0007C0_CSF_PRIMARY_BUSY(x) (((x) & 0x1) << 9) +#define G_0007C0_CSF_PRIMARY_BUSY(x) (((x) >> 9) & 0x1) +#define C_0007C0_CSF_PRIMARY_BUSY 0xFFFFFDFF +#define S_0007C0_CSF_INDIRECT_BUSY(x) (((x) & 0x1) << 10) +#define G_0007C0_CSF_INDIRECT_BUSY(x) (((x) >> 10) & 0x1) +#define C_0007C0_CSF_INDIRECT_BUSY 0xFFFFFBFF +#define S_0007C0_CSQ_PRIMARY_BUSY(x) (((x) & 0x1) << 11) +#define G_0007C0_CSQ_PRIMARY_BUSY(x) (((x) >> 11) & 0x1) +#define C_0007C0_CSQ_PRIMARY_BUSY 0xFFFFF7FF +#define S_0007C0_CSQ_INDIRECT_BUSY(x) (((x) & 0x1) << 12) +#define G_0007C0_CSQ_INDIRECT_BUSY(x) (((x) >> 12) & 0x1) +#define C_0007C0_CSQ_INDIRECT_BUSY 0xFFFFEFFF +#define S_0007C0_CSI_BUSY(x) (((x) & 0x1) << 13) +#define G_0007C0_CSI_BUSY(x) (((x) >> 13) & 0x1) +#define C_0007C0_CSI_BUSY 0xFFFFDFFF +#define S_0007C0_CSF_INDIRECT2_BUSY(x) (((x) & 0x1) << 14) +#define G_0007C0_CSF_INDIRECT2_BUSY(x) (((x) >> 14) & 0x1) +#define C_0007C0_CSF_INDIRECT2_BUSY 0xFFFFBFFF +#define S_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) & 0x1) << 15) +#define G_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) >> 15) & 0x1) +#define C_0007C0_CSQ_INDIRECT2_BUSY 0xFFFF7FFF +#define S_0007C0_GUIDMA_BUSY(x) (((x) & 0x1) << 28) +#define G_0007C0_GUIDMA_BUSY(x) (((x) >> 28) & 0x1) +#define C_0007C0_GUIDMA_BUSY 0xEFFFFFFF +#define S_0007C0_VIDDMA_BUSY(x) (((x) & 0x1) << 29) +#define G_0007C0_VIDDMA_BUSY(x) (((x) >> 29) & 0x1) +#define C_0007C0_VIDDMA_BUSY 0xDFFFFFFF +#define S_0007C0_CMDSTRM_BUSY(x) (((x) & 0x1) << 30) +#define G_0007C0_CMDSTRM_BUSY(x) (((x) >> 30) & 0x1) +#define C_0007C0_CMDSTRM_BUSY 0xBFFFFFFF +#define S_0007C0_CP_BUSY(x) (((x) & 0x1) << 31) +#define G_0007C0_CP_BUSY(x) (((x) >> 31) & 0x1) +#define C_0007C0_CP_BUSY 0x7FFFFFFF +#define R_000E40_RBBM_STATUS 0x000E40 +#define S_000E40_CMDFIFO_AVAIL(x) (((x) & 0x7F) << 0) +#define G_000E40_CMDFIFO_AVAIL(x) (((x) >> 0) & 0x7F) +#define C_000E40_CMDFIFO_AVAIL 0xFFFFFF80 +#define S_000E40_HIRQ_ON_RBB(x) (((x) & 0x1) << 8) +#define G_000E40_HIRQ_ON_RBB(x) (((x) >> 8) & 0x1) +#define C_000E40_HIRQ_ON_RBB 0xFFFFFEFF +#define S_000E40_CPRQ_ON_RBB(x) (((x) & 0x1) << 9) +#define G_000E40_CPRQ_ON_RBB(x) (((x) >> 9) & 0x1) +#define C_000E40_CPRQ_ON_RBB 0xFFFFFDFF +#define S_000E40_CFRQ_ON_RBB(x) (((x) & 0x1) << 10) +#define G_000E40_CFRQ_ON_RBB(x) (((x) >> 10) & 0x1) +#define C_000E40_CFRQ_ON_RBB 0xFFFFFBFF +#define S_000E40_HIRQ_IN_RTBUF(x) (((x) & 0x1) << 11) +#define G_000E40_HIRQ_IN_RTBUF(x) (((x) >> 11) & 0x1) +#define C_000E40_HIRQ_IN_RTBUF 0xFFFFF7FF +#define S_000E40_CPRQ_IN_RTBUF(x) (((x) & 0x1) << 12) +#define G_000E40_CPRQ_IN_RTBUF(x) (((x) >> 12) & 0x1) +#define C_000E40_CPRQ_IN_RTBUF 0xFFFFEFFF +#define S_000E40_CFRQ_IN_RTBUF(x) (((x) & 0x1) << 13) +#define G_000E40_CFRQ_IN_RTBUF(x) (((x) >> 13) & 0x1) +#define C_000E40_CFRQ_IN_RTBUF 0xFFFFDFFF +#define S_000E40_CF_PIPE_BUSY(x) (((x) & 0x1) << 14) +#define G_000E40_CF_PIPE_BUSY(x) (((x) >> 14) & 0x1) +#define C_000E40_CF_PIPE_BUSY 0xFFFFBFFF +#define S_000E40_ENG_EV_BUSY(x) (((x) & 0x1) << 15) +#define G_000E40_ENG_EV_BUSY(x) (((x) >> 15) & 0x1) +#define C_000E40_ENG_EV_BUSY 0xFFFF7FFF +#define S_000E40_CP_CMDSTRM_BUSY(x) (((x) & 0x1) << 16) +#define G_000E40_CP_CMDSTRM_BUSY(x) (((x) >> 16) & 0x1) +#define C_000E40_CP_CMDSTRM_BUSY 0xFFFEFFFF +#define S_000E40_E2_BUSY(x) (((x) & 0x1) << 17) +#define G_000E40_E2_BUSY(x) (((x) >> 17) & 0x1) +#define C_000E40_E2_BUSY 0xFFFDFFFF +#define S_000E40_RB2D_BUSY(x) (((x) & 0x1) << 18) +#define G_000E40_RB2D_BUSY(x) (((x) >> 18) & 0x1) +#define C_000E40_RB2D_BUSY 0xFFFBFFFF +#define S_000E40_RB3D_BUSY(x) (((x) & 0x1) << 19) +#define G_000E40_RB3D_BUSY(x) (((x) >> 19) & 0x1) +#define C_000E40_RB3D_BUSY 0xFFF7FFFF +#define S_000E40_VAP_BUSY(x) (((x) & 0x1) << 20) +#define G_000E40_VAP_BUSY(x) (((x) >> 20) & 0x1) +#define C_000E40_VAP_BUSY 0xFFEFFFFF +#define S_000E40_RE_BUSY(x) (((x) & 0x1) << 21) +#define G_000E40_RE_BUSY(x) (((x) >> 21) & 0x1) +#define C_000E40_RE_BUSY 0xFFDFFFFF +#define S_000E40_TAM_BUSY(x) (((x) & 0x1) << 22) +#define G_000E40_TAM_BUSY(x) (((x) >> 22) & 0x1) +#define C_000E40_TAM_BUSY 0xFFBFFFFF +#define S_000E40_TDM_BUSY(x) (((x) & 0x1) << 23) +#define G_000E40_TDM_BUSY(x) (((x) >> 23) & 0x1) +#define C_000E40_TDM_BUSY 0xFF7FFFFF +#define S_000E40_PB_BUSY(x) (((x) & 0x1) << 24) +#define G_000E40_PB_BUSY(x) (((x) >> 24) & 0x1) +#define C_000E40_PB_BUSY 0xFEFFFFFF +#define S_000E40_TIM_BUSY(x) (((x) & 0x1) << 25) +#define G_000E40_TIM_BUSY(x) (((x) >> 25) & 0x1) +#define C_000E40_TIM_BUSY 0xFDFFFFFF +#define S_000E40_GA_BUSY(x) (((x) & 0x1) << 26) +#define G_000E40_GA_BUSY(x) (((x) >> 26) & 0x1) +#define C_000E40_GA_BUSY 0xFBFFFFFF +#define S_000E40_CBA2D_BUSY(x) (((x) & 0x1) << 27) +#define G_000E40_CBA2D_BUSY(x) (((x) >> 27) & 0x1) +#define C_000E40_CBA2D_BUSY 0xF7FFFFFF +#define S_000E40_GUI_ACTIVE(x) (((x) & 0x1) << 31) +#define G_000E40_GUI_ACTIVE(x) (((x) >> 31) & 0x1) +#define C_000E40_GUI_ACTIVE 0x7FFFFFFF + +#endif diff --git a/drivers/gpu/drm/radeon/rs600.c b/drivers/gpu/drm/radeon/rs600.c index 0e791e26..5f117cd 100644 --- a/drivers/gpu/drm/radeon/rs600.c +++ b/drivers/gpu/drm/radeon/rs600.c @@ -25,28 +25,25 @@ * Alex Deucher * Jerome Glisse */ +/* RS600 / Radeon X1250/X1270 integrated GPU + * + * This file gather function specific to RS600 which is the IGP of + * the X1250/X1270 family supporting intel CPU (while RS690/RS740 + * is the X1250/X1270 supporting AMD CPU). The display engine are + * the avivo one, bios is an atombios, 3D block are the one of the + * R4XX family. The GART is different from the RS400 one and is very + * close to the one of the R600 family (R600 likely being an evolution + * of the RS600 GART block). + */ #include "drmP.h" -#include "radeon_reg.h" #include "radeon.h" -#include "avivod.h" +#include "atom.h" +#include "rs600d.h" #include "rs600_reg_safe.h" -/* rs600 depends on : */ -void r100_hdp_reset(struct radeon_device *rdev); -int r100_gui_wait_for_idle(struct radeon_device *rdev); -int r300_mc_wait_for_idle(struct radeon_device *rdev); -void r420_pipes_init(struct radeon_device *rdev); - -/* This files gather functions specifics to : - * rs600 - * - * Some of these functions might be used by newer ASICs. - */ void rs600_gpu_init(struct radeon_device *rdev); int rs600_mc_wait_for_idle(struct radeon_device *rdev); -void rs600_disable_vga(struct radeon_device *rdev); - /* * GART. @@ -55,18 +52,18 @@ void rs600_gart_tlb_flush(struct radeon_device *rdev) { uint32_t tmp; - tmp = RREG32_MC(RS600_MC_PT0_CNTL); - tmp &= ~(RS600_INVALIDATE_ALL_L1_TLBS | RS600_INVALIDATE_L2_CACHE); - WREG32_MC(RS600_MC_PT0_CNTL, tmp); + tmp = RREG32_MC(R_000100_MC_PT0_CNTL); + tmp &= C_000100_INVALIDATE_ALL_L1_TLBS & C_000100_INVALIDATE_L2_CACHE; + WREG32_MC(R_000100_MC_PT0_CNTL, tmp); - tmp = RREG32_MC(RS600_MC_PT0_CNTL); - tmp |= RS600_INVALIDATE_ALL_L1_TLBS | RS600_INVALIDATE_L2_CACHE; - WREG32_MC(RS600_MC_PT0_CNTL, tmp); + tmp = RREG32_MC(R_000100_MC_PT0_CNTL); + tmp |= S_000100_INVALIDATE_ALL_L1_TLBS(1) & S_000100_INVALIDATE_L2_CACHE(1); + WREG32_MC(R_000100_MC_PT0_CNTL, tmp); - tmp = RREG32_MC(RS600_MC_PT0_CNTL); - tmp &= ~(RS600_INVALIDATE_ALL_L1_TLBS | RS600_INVALIDATE_L2_CACHE); - WREG32_MC(RS600_MC_PT0_CNTL, tmp); - tmp = RREG32_MC(RS600_MC_PT0_CNTL); + tmp = RREG32_MC(R_000100_MC_PT0_CNTL); + tmp &= C_000100_INVALIDATE_ALL_L1_TLBS & C_000100_INVALIDATE_L2_CACHE; + WREG32_MC(R_000100_MC_PT0_CNTL, tmp); + tmp = RREG32_MC(R_000100_MC_PT0_CNTL); } int rs600_gart_init(struct radeon_device *rdev) @@ -88,7 +85,7 @@ int rs600_gart_init(struct radeon_device *rdev) int rs600_gart_enable(struct radeon_device *rdev) { - uint32_t tmp; + u32 tmp; int r, i; if (rdev->gart.table.vram.robj == NULL) { @@ -98,46 +95,50 @@ int rs600_gart_enable(struct radeon_device *rdev) r = radeon_gart_table_vram_pin(rdev); if (r) return r; + /* Enable bus master */ + tmp = RREG32(R_00004C_BUS_CNTL) & C_00004C_BUS_MASTER_DIS; + WREG32(R_00004C_BUS_CNTL, tmp); /* FIXME: setup default page */ - WREG32_MC(RS600_MC_PT0_CNTL, - (RS600_EFFECTIVE_L2_CACHE_SIZE(6) | - RS600_EFFECTIVE_L2_QUEUE_SIZE(6))); + WREG32_MC(R_000100_MC_PT0_CNTL, + (S_000100_EFFECTIVE_L2_CACHE_SIZE(6) | + S_000100_EFFECTIVE_L2_QUEUE_SIZE(6))); for (i = 0; i < 19; i++) { - WREG32_MC(RS600_MC_PT0_CLIENT0_CNTL + i, - (RS600_ENABLE_TRANSLATION_MODE_OVERRIDE | - RS600_SYSTEM_ACCESS_MODE_IN_SYS | - RS600_SYSTEM_APERTURE_UNMAPPED_ACCESS_DEFAULT_PAGE | - RS600_EFFECTIVE_L1_CACHE_SIZE(3) | - RS600_ENABLE_FRAGMENT_PROCESSING | - RS600_EFFECTIVE_L1_QUEUE_SIZE(3))); + WREG32_MC(R_00016C_MC_PT0_CLIENT0_CNTL + i, + S_00016C_ENABLE_TRANSLATION_MODE_OVERRIDE(1) | + S_00016C_SYSTEM_ACCESS_MODE_MASK( + V_00016C_SYSTEM_ACCESS_MODE_IN_SYS) | + S_00016C_SYSTEM_APERTURE_UNMAPPED_ACCESS( + V_00016C_SYSTEM_APERTURE_UNMAPPED_DEFAULT_PAGE) | + S_00016C_EFFECTIVE_L1_CACHE_SIZE(1) | + S_00016C_ENABLE_FRAGMENT_PROCESSING(1) | + S_00016C_EFFECTIVE_L1_QUEUE_SIZE(1)); } /* System context map to GART space */ - WREG32_MC(RS600_MC_PT0_SYSTEM_APERTURE_LOW_ADDR, rdev->mc.gtt_location); - tmp = rdev->mc.gtt_location + rdev->mc.gtt_size - 1; - WREG32_MC(RS600_MC_PT0_SYSTEM_APERTURE_HIGH_ADDR, tmp); + WREG32_MC(R_000112_MC_PT0_SYSTEM_APERTURE_LOW_ADDR, rdev->mc.gtt_start); + WREG32_MC(R_000114_MC_PT0_SYSTEM_APERTURE_HIGH_ADDR, rdev->mc.gtt_end); /* enable first context */ - WREG32_MC(RS600_MC_PT0_CONTEXT0_FLAT_START_ADDR, rdev->mc.gtt_location); - tmp = rdev->mc.gtt_location + rdev->mc.gtt_size - 1; - WREG32_MC(RS600_MC_PT0_CONTEXT0_FLAT_END_ADDR, tmp); - WREG32_MC(RS600_MC_PT0_CONTEXT0_CNTL, - (RS600_ENABLE_PAGE_TABLE | RS600_PAGE_TABLE_TYPE_FLAT)); + WREG32_MC(R_00013C_MC_PT0_CONTEXT0_FLAT_START_ADDR, rdev->mc.gtt_start); + WREG32_MC(R_00014C_MC_PT0_CONTEXT0_FLAT_END_ADDR, rdev->mc.gtt_end); + WREG32_MC(R_000102_MC_PT0_CONTEXT0_CNTL, + S_000102_ENABLE_PAGE_TABLE(1) | + S_000102_PAGE_TABLE_DEPTH(V_000102_PAGE_TABLE_FLAT)); /* disable all other contexts */ for (i = 1; i < 8; i++) { - WREG32_MC(RS600_MC_PT0_CONTEXT0_CNTL + i, 0); + WREG32_MC(R_000102_MC_PT0_CONTEXT0_CNTL + i, 0); } /* setup the page table */ - WREG32_MC(RS600_MC_PT0_CONTEXT0_FLAT_BASE_ADDR, - rdev->gart.table_addr); - WREG32_MC(RS600_MC_PT0_CONTEXT0_DEFAULT_READ_ADDR, 0); + WREG32_MC(R_00012C_MC_PT0_CONTEXT0_FLAT_BASE_ADDR, + rdev->gart.table_addr); + WREG32_MC(R_00011C_MC_PT0_CONTEXT0_DEFAULT_READ_ADDR, 0); /* enable page tables */ - tmp = RREG32_MC(RS600_MC_PT0_CNTL); - WREG32_MC(RS600_MC_PT0_CNTL, (tmp | RS600_ENABLE_PT)); - tmp = RREG32_MC(RS600_MC_CNTL1); - WREG32_MC(RS600_MC_CNTL1, (tmp | RS600_ENABLE_PAGE_TABLES)); + tmp = RREG32_MC(R_000100_MC_PT0_CNTL); + WREG32_MC(R_000100_MC_PT0_CNTL, (tmp | S_000100_ENABLE_PT(1))); + tmp = RREG32_MC(R_000009_MC_CNTL1); + WREG32_MC(R_000009_MC_CNTL1, (tmp | S_000009_ENABLE_PAGE_TABLES(1))); rs600_gart_tlb_flush(rdev); rdev->gart.ready = true; return 0; @@ -148,10 +149,9 @@ void rs600_gart_disable(struct radeon_device *rdev) uint32_t tmp; /* FIXME: disable out of gart access */ - WREG32_MC(RS600_MC_PT0_CNTL, 0); - tmp = RREG32_MC(RS600_MC_CNTL1); - tmp &= ~RS600_ENABLE_PAGE_TABLES; - WREG32_MC(RS600_MC_CNTL1, tmp); + WREG32_MC(R_000100_MC_PT0_CNTL, 0); + tmp = RREG32_MC(R_000009_MC_CNTL1); + WREG32_MC(R_000009_MC_CNTL1, tmp & C_000009_ENABLE_PAGE_TABLES); if (rdev->gart.table.vram.robj) { radeon_object_kunmap(rdev->gart.table.vram.robj); radeon_object_unpin(rdev->gart.table.vram.robj); @@ -185,132 +185,64 @@ int rs600_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr) return 0; } - -/* - * MC. - */ -void rs600_mc_disable_clients(struct radeon_device *rdev) -{ - unsigned tmp; - - if (r100_gui_wait_for_idle(rdev)) { - printk(KERN_WARNING "Failed to wait GUI idle while " - "programming pipes. Bad things might happen.\n"); - } - - radeon_avivo_vga_render_disable(rdev); - - tmp = RREG32(AVIVO_D1VGA_CONTROL); - WREG32(AVIVO_D1VGA_CONTROL, tmp & ~AVIVO_DVGA_CONTROL_MODE_ENABLE); - tmp = RREG32(AVIVO_D2VGA_CONTROL); - WREG32(AVIVO_D2VGA_CONTROL, tmp & ~AVIVO_DVGA_CONTROL_MODE_ENABLE); - - tmp = RREG32(AVIVO_D1CRTC_CONTROL); - WREG32(AVIVO_D1CRTC_CONTROL, tmp & ~AVIVO_CRTC_EN); - tmp = RREG32(AVIVO_D2CRTC_CONTROL); - WREG32(AVIVO_D2CRTC_CONTROL, tmp & ~AVIVO_CRTC_EN); - - /* make sure all previous write got through */ - tmp = RREG32(AVIVO_D2CRTC_CONTROL); - - mdelay(1); -} - -int rs600_mc_init(struct radeon_device *rdev) -{ - uint32_t tmp; - int r; - - if (r100_debugfs_rbbm_init(rdev)) { - DRM_ERROR("Failed to register debugfs file for RBBM !\n"); - } - - rs600_gpu_init(rdev); - rs600_gart_disable(rdev); - - /* Setup GPU memory space */ - rdev->mc.vram_location = 0xFFFFFFFFUL; - rdev->mc.gtt_location = 0xFFFFFFFFUL; - r = radeon_mc_setup(rdev); - if (r) { - return r; - } - - /* Program GPU memory space */ - /* Enable bus master */ - tmp = RREG32(RADEON_BUS_CNTL) & ~RS600_BUS_MASTER_DIS; - WREG32(RADEON_BUS_CNTL, tmp); - /* FIXME: What does AGP means for such chipset ? */ - WREG32_MC(RS600_MC_AGP_LOCATION, 0x0FFFFFFF); - /* FIXME: are this AGP reg in indirect MC range ? */ - WREG32_MC(RS600_MC_AGP_BASE, 0); - WREG32_MC(RS600_MC_AGP_BASE_2, 0); - rs600_mc_disable_clients(rdev); - if (rs600_mc_wait_for_idle(rdev)) { - printk(KERN_WARNING "Failed to wait MC idle while " - "programming pipes. Bad things might happen.\n"); - } - tmp = rdev->mc.vram_location + rdev->mc.mc_vram_size - 1; - tmp = REG_SET(RS600_MC_FB_TOP, tmp >> 16); - tmp |= REG_SET(RS600_MC_FB_START, rdev->mc.vram_location >> 16); - WREG32_MC(RS600_MC_FB_LOCATION, tmp); - WREG32(RS690_HDP_FB_LOCATION, rdev->mc.vram_location >> 16); - return 0; -} - -void rs600_mc_fini(struct radeon_device *rdev) -{ -} - - -/* - * Interrupts - */ int rs600_irq_set(struct radeon_device *rdev) { uint32_t tmp = 0; uint32_t mode_int = 0; if (rdev->irq.sw_int) { - tmp |= RADEON_SW_INT_ENABLE; + tmp |= S_000040_SW_INT_EN(1); } if (rdev->irq.crtc_vblank_int[0]) { - mode_int |= AVIVO_D1MODE_INT_MASK; + mode_int |= S_006540_D1MODE_VBLANK_INT_MASK(1); } if (rdev->irq.crtc_vblank_int[1]) { - mode_int |= AVIVO_D2MODE_INT_MASK; + mode_int |= S_006540_D2MODE_VBLANK_INT_MASK(1); } - WREG32(RADEON_GEN_INT_CNTL, tmp); - WREG32(AVIVO_DxMODE_INT_MASK, mode_int); + WREG32(R_000040_GEN_INT_CNTL, tmp); + WREG32(R_006540_DxMODE_INT_MASK, mode_int); return 0; } static inline uint32_t rs600_irq_ack(struct radeon_device *rdev, u32 *r500_disp_int) { - uint32_t irqs = RREG32(RADEON_GEN_INT_STATUS); - uint32_t irq_mask = RADEON_SW_INT_TEST; - - if (irqs & AVIVO_DISPLAY_INT_STATUS) { - *r500_disp_int = RREG32(AVIVO_DISP_INTERRUPT_STATUS); - if (*r500_disp_int & AVIVO_D1_VBLANK_INTERRUPT) { - WREG32(AVIVO_D1MODE_VBLANK_STATUS, AVIVO_VBLANK_ACK); + uint32_t irqs = RREG32(R_000044_GEN_INT_STATUS); + uint32_t irq_mask = ~C_000044_SW_INT; + + if (G_000044_DISPLAY_INT_STAT(irqs)) { + *r500_disp_int = RREG32(R_007EDC_DISP_INTERRUPT_STATUS); + if (G_007EDC_LB_D1_VBLANK_INTERRUPT(*r500_disp_int)) { + WREG32(R_006534_D1MODE_VBLANK_STATUS, + S_006534_D1MODE_VBLANK_ACK(1)); } - if (*r500_disp_int & AVIVO_D2_VBLANK_INTERRUPT) { - WREG32(AVIVO_D2MODE_VBLANK_STATUS, AVIVO_VBLANK_ACK); + if (G_007EDC_LB_D2_VBLANK_INTERRUPT(*r500_disp_int)) { + WREG32(R_006D34_D2MODE_VBLANK_STATUS, + S_006D34_D2MODE_VBLANK_ACK(1)); } } else { *r500_disp_int = 0; } if (irqs) { - WREG32(RADEON_GEN_INT_STATUS, irqs); + WREG32(R_000044_GEN_INT_STATUS, irqs); } return irqs & irq_mask; } +void rs600_irq_disable(struct radeon_device *rdev) +{ + u32 tmp; + + WREG32(R_000040_GEN_INT_CNTL, 0); + WREG32(R_006540_DxMODE_INT_MASK, 0); + /* Wait and acknowledge irq */ + mdelay(1); + rs600_irq_ack(rdev, &tmp); +} + int rs600_irq_process(struct radeon_device *rdev) { - uint32_t status; + uint32_t status, msi_rearm; uint32_t r500_disp_int; status = rs600_irq_ack(rdev, &r500_disp_int); @@ -319,85 +251,65 @@ int rs600_irq_process(struct radeon_device *rdev) } while (status || r500_disp_int) { /* SW interrupt */ - if (status & RADEON_SW_INT_TEST) { + if (G_000040_SW_INT_EN(status)) radeon_fence_process(rdev); - } /* Vertical blank interrupts */ - if (r500_disp_int & AVIVO_D1_VBLANK_INTERRUPT) { + if (G_007EDC_LB_D1_VBLANK_INTERRUPT(r500_disp_int)) drm_handle_vblank(rdev->ddev, 0); - } - if (r500_disp_int & AVIVO_D2_VBLANK_INTERRUPT) { + if (G_007EDC_LB_D2_VBLANK_INTERRUPT(r500_disp_int)) drm_handle_vblank(rdev->ddev, 1); - } status = rs600_irq_ack(rdev, &r500_disp_int); } + if (rdev->msi_enabled) { + switch (rdev->family) { + case CHIP_RS600: + case CHIP_RS690: + case CHIP_RS740: + msi_rearm = RREG32(RADEON_BUS_CNTL) & ~RS600_MSI_REARM; + WREG32(RADEON_BUS_CNTL, msi_rearm); + WREG32(RADEON_BUS_CNTL, msi_rearm | RS600_MSI_REARM); + break; + default: + msi_rearm = RREG32(RADEON_MSI_REARM_EN) & ~RV370_MSI_REARM_EN; + WREG32(RADEON_MSI_REARM_EN, msi_rearm); + WREG32(RADEON_MSI_REARM_EN, msi_rearm | RV370_MSI_REARM_EN); + break; + } + } return IRQ_HANDLED; } u32 rs600_get_vblank_counter(struct radeon_device *rdev, int crtc) { if (crtc == 0) - return RREG32(AVIVO_D1CRTC_FRAME_COUNT); + return RREG32(R_0060A4_D1CRTC_STATUS_FRAME_COUNT); else - return RREG32(AVIVO_D2CRTC_FRAME_COUNT); -} - - -/* - * Global GPU functions - */ -void rs600_disable_vga(struct radeon_device *rdev) -{ - unsigned tmp; - - WREG32(0x330, 0); - WREG32(0x338, 0); - tmp = RREG32(0x300); - tmp &= ~(3 << 16); - WREG32(0x300, tmp); - WREG32(0x308, (1 << 8)); - WREG32(0x310, rdev->mc.vram_location); - WREG32(0x594, 0); + return RREG32(R_0068A4_D2CRTC_STATUS_FRAME_COUNT); } int rs600_mc_wait_for_idle(struct radeon_device *rdev) { unsigned i; - uint32_t tmp; for (i = 0; i < rdev->usec_timeout; i++) { - /* read MC_STATUS */ - tmp = RREG32_MC(RS600_MC_STATUS); - if (tmp & RS600_MC_STATUS_IDLE) { + if (G_000000_MC_IDLE(RREG32_MC(R_000000_MC_STATUS))) return 0; - } - DRM_UDELAY(1); + udelay(1); } return -1; } -void rs600_errata(struct radeon_device *rdev) -{ - rdev->pll_errata = 0; -} - void rs600_gpu_init(struct radeon_device *rdev) { /* FIXME: HDP same place on rs600 ? */ r100_hdp_reset(rdev); - rs600_disable_vga(rdev); /* FIXME: is this correct ? */ r420_pipes_init(rdev); - if (rs600_mc_wait_for_idle(rdev)) { - printk(KERN_WARNING "Failed to wait MC idle while " - "programming pipes. Bad things might happen.\n"); - } + /* Wait for mc idle */ + if (rs600_mc_wait_for_idle(rdev)) + dev_warn(rdev->dev, "Wait MC idle timeout before updating MC.\n"); } - -/* - * VRAM info. - */ void rs600_vram_info(struct radeon_device *rdev) { /* FIXME: to do or is these values sane ? */ @@ -410,31 +322,208 @@ void rs600_bandwidth_update(struct radeon_device *rdev) /* FIXME: implement, should this be like rs690 ? */ } - -/* - * Indirect registers accessor - */ uint32_t rs600_mc_rreg(struct radeon_device *rdev, uint32_t reg) { - uint32_t r; - - WREG32(RS600_MC_INDEX, - ((reg & RS600_MC_ADDR_MASK) | RS600_MC_IND_CITF_ARB0)); - r = RREG32(RS600_MC_DATA); - return r; + WREG32(R_000070_MC_IND_INDEX, S_000070_MC_IND_ADDR(reg) | + S_000070_MC_IND_CITF_ARB0(1)); + return RREG32(R_000074_MC_IND_DATA); } void rs600_mc_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v) { - WREG32(RS600_MC_INDEX, - RS600_MC_IND_WR_EN | RS600_MC_IND_CITF_ARB0 | - ((reg) & RS600_MC_ADDR_MASK)); - WREG32(RS600_MC_DATA, v); + WREG32(R_000070_MC_IND_INDEX, S_000070_MC_IND_ADDR(reg) | + S_000070_MC_IND_CITF_ARB0(1) | S_000070_MC_IND_WR_EN(1)); + WREG32(R_000074_MC_IND_DATA, v); } -int rs600_init(struct radeon_device *rdev) +void rs600_debugfs(struct radeon_device *rdev) +{ + if (r100_debugfs_rbbm_init(rdev)) + DRM_ERROR("Failed to register debugfs file for RBBM !\n"); +} + +void rs600_set_safe_registers(struct radeon_device *rdev) { rdev->config.r300.reg_safe_bm = rs600_reg_safe_bm; rdev->config.r300.reg_safe_bm_size = ARRAY_SIZE(rs600_reg_safe_bm); +} + +static void rs600_mc_program(struct radeon_device *rdev) +{ + struct rv515_mc_save save; + + /* Stops all mc clients */ + rv515_mc_stop(rdev, &save); + + /* Wait for mc idle */ + if (rs600_mc_wait_for_idle(rdev)) + dev_warn(rdev->dev, "Wait MC idle timeout before updating MC.\n"); + + /* FIXME: What does AGP means for such chipset ? */ + WREG32_MC(R_000005_MC_AGP_LOCATION, 0x0FFFFFFF); + WREG32_MC(R_000006_AGP_BASE, 0); + WREG32_MC(R_000007_AGP_BASE_2, 0); + /* Program MC */ + WREG32_MC(R_000004_MC_FB_LOCATION, + S_000004_MC_FB_START(rdev->mc.vram_start >> 16) | + S_000004_MC_FB_TOP(rdev->mc.vram_end >> 16)); + WREG32(R_000134_HDP_FB_LOCATION, + S_000134_HDP_FB_START(rdev->mc.vram_start >> 16)); + + rv515_mc_resume(rdev, &save); +} + +static int rs600_startup(struct radeon_device *rdev) +{ + int r; + + rs600_mc_program(rdev); + /* Resume clock */ + rv515_clock_startup(rdev); + /* Initialize GPU configuration (# pipes, ...) */ + rs600_gpu_init(rdev); + /* Initialize GART (initialize after TTM so we can allocate + * memory through TTM but finalize after TTM) */ + r = rs600_gart_enable(rdev); + if (r) + return r; + /* Enable IRQ */ + rdev->irq.sw_int = true; + rs600_irq_set(rdev); + /* 1M ring buffer */ + r = r100_cp_init(rdev, 1024 * 1024); + if (r) { + dev_err(rdev->dev, "failled initializing CP (%d).\n", r); + return r; + } + r = r100_wb_init(rdev); + if (r) + dev_err(rdev->dev, "failled initializing WB (%d).\n", r); + r = r100_ib_init(rdev); + if (r) { + dev_err(rdev->dev, "failled initializing IB (%d).\n", r); + return r; + } + return 0; +} + +int rs600_resume(struct radeon_device *rdev) +{ + /* Make sur GART are not working */ + rs600_gart_disable(rdev); + /* Resume clock before doing reset */ + rv515_clock_startup(rdev); + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_gpu_reset(rdev)) { + dev_warn(rdev->dev, "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* post */ + atom_asic_init(rdev->mode_info.atom_context); + /* Resume clock after posting */ + rv515_clock_startup(rdev); + return rs600_startup(rdev); +} + +int rs600_suspend(struct radeon_device *rdev) +{ + r100_cp_disable(rdev); + r100_wb_disable(rdev); + rs600_irq_disable(rdev); + rs600_gart_disable(rdev); + return 0; +} + +void rs600_fini(struct radeon_device *rdev) +{ + rs600_suspend(rdev); + r100_cp_fini(rdev); + r100_wb_fini(rdev); + r100_ib_fini(rdev); + radeon_gem_fini(rdev); + rs600_gart_fini(rdev); + radeon_irq_kms_fini(rdev); + radeon_fence_driver_fini(rdev); + radeon_object_fini(rdev); + radeon_atombios_fini(rdev); + kfree(rdev->bios); + rdev->bios = NULL; +} + +int rs600_init(struct radeon_device *rdev) +{ + int r; + + /* Disable VGA */ + rv515_vga_render_disable(rdev); + /* Initialize scratch registers */ + radeon_scratch_init(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + /* BIOS */ + if (!radeon_get_bios(rdev)) { + if (ASIC_IS_AVIVO(rdev)) + return -EINVAL; + } + if (rdev->is_atom_bios) { + r = radeon_atombios_init(rdev); + if (r) + return r; + } else { + dev_err(rdev->dev, "Expecting atombios for RS600 GPU\n"); + return -EINVAL; + } + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_gpu_reset(rdev)) { + dev_warn(rdev->dev, + "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* check if cards are posted or not */ + if (!radeon_card_posted(rdev) && rdev->bios) { + DRM_INFO("GPU not posted. posting now...\n"); + atom_asic_init(rdev->mode_info.atom_context); + } + /* Initialize clocks */ + radeon_get_clock_info(rdev->ddev); + /* Initialize power management */ + radeon_pm_init(rdev); + /* Get vram informations */ + rs600_vram_info(rdev); + /* Initialize memory controller (also test AGP) */ + r = r420_mc_init(rdev); + if (r) + return r; + rs600_debugfs(rdev); + /* Fence driver */ + r = radeon_fence_driver_init(rdev); + if (r) + return r; + r = radeon_irq_kms_init(rdev); + if (r) + return r; + /* Memory manager */ + r = radeon_object_init(rdev); + if (r) + return r; + r = rs600_gart_init(rdev); + if (r) + return r; + rs600_set_safe_registers(rdev); + rdev->accel_working = true; + r = rs600_startup(rdev); + if (r) { + /* Somethings want wront with the accel init stop accel */ + dev_err(rdev->dev, "Disabling GPU acceleration\n"); + rs600_suspend(rdev); + r100_cp_fini(rdev); + r100_wb_fini(rdev); + r100_ib_fini(rdev); + rs600_gart_fini(rdev); + radeon_irq_kms_fini(rdev); + rdev->accel_working = false; + } return 0; } diff --git a/drivers/gpu/drm/radeon/rs600d.h b/drivers/gpu/drm/radeon/rs600d.h new file mode 100644 index 0000000..8130892 --- /dev/null +++ b/drivers/gpu/drm/radeon/rs600d.h @@ -0,0 +1,470 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef __RS600D_H__ +#define __RS600D_H__ + +/* Registers */ +#define R_000040_GEN_INT_CNTL 0x000040 +#define S_000040_DISPLAY_INT_STATUS(x) (((x) & 0x1) << 0) +#define G_000040_DISPLAY_INT_STATUS(x) (((x) >> 0) & 0x1) +#define C_000040_DISPLAY_INT_STATUS 0xFFFFFFFE +#define S_000040_DMA_VIPH0_INT_EN(x) (((x) & 0x1) << 12) +#define G_000040_DMA_VIPH0_INT_EN(x) (((x) >> 12) & 0x1) +#define C_000040_DMA_VIPH0_INT_EN 0xFFFFEFFF +#define S_000040_CRTC2_VSYNC(x) (((x) & 0x1) << 6) +#define G_000040_CRTC2_VSYNC(x) (((x) >> 6) & 0x1) +#define C_000040_CRTC2_VSYNC 0xFFFFFFBF +#define S_000040_SNAPSHOT2(x) (((x) & 0x1) << 7) +#define G_000040_SNAPSHOT2(x) (((x) >> 7) & 0x1) +#define C_000040_SNAPSHOT2 0xFFFFFF7F +#define S_000040_CRTC2_VBLANK(x) (((x) & 0x1) << 9) +#define G_000040_CRTC2_VBLANK(x) (((x) >> 9) & 0x1) +#define C_000040_CRTC2_VBLANK 0xFFFFFDFF +#define S_000040_FP2_DETECT(x) (((x) & 0x1) << 10) +#define G_000040_FP2_DETECT(x) (((x) >> 10) & 0x1) +#define C_000040_FP2_DETECT 0xFFFFFBFF +#define S_000040_VSYNC_DIFF_OVER_LIMIT(x) (((x) & 0x1) << 11) +#define G_000040_VSYNC_DIFF_OVER_LIMIT(x) (((x) >> 11) & 0x1) +#define C_000040_VSYNC_DIFF_OVER_LIMIT 0xFFFFF7FF +#define S_000040_DMA_VIPH1_INT_EN(x) (((x) & 0x1) << 13) +#define G_000040_DMA_VIPH1_INT_EN(x) (((x) >> 13) & 0x1) +#define C_000040_DMA_VIPH1_INT_EN 0xFFFFDFFF +#define S_000040_DMA_VIPH2_INT_EN(x) (((x) & 0x1) << 14) +#define G_000040_DMA_VIPH2_INT_EN(x) (((x) >> 14) & 0x1) +#define C_000040_DMA_VIPH2_INT_EN 0xFFFFBFFF +#define S_000040_DMA_VIPH3_INT_EN(x) (((x) & 0x1) << 15) +#define G_000040_DMA_VIPH3_INT_EN(x) (((x) >> 15) & 0x1) +#define C_000040_DMA_VIPH3_INT_EN 0xFFFF7FFF +#define S_000040_I2C_INT_EN(x) (((x) & 0x1) << 17) +#define G_000040_I2C_INT_EN(x) (((x) >> 17) & 0x1) +#define C_000040_I2C_INT_EN 0xFFFDFFFF +#define S_000040_GUI_IDLE(x) (((x) & 0x1) << 19) +#define G_000040_GUI_IDLE(x) (((x) >> 19) & 0x1) +#define C_000040_GUI_IDLE 0xFFF7FFFF +#define S_000040_VIPH_INT_EN(x) (((x) & 0x1) << 24) +#define G_000040_VIPH_INT_EN(x) (((x) >> 24) & 0x1) +#define C_000040_VIPH_INT_EN 0xFEFFFFFF +#define S_000040_SW_INT_EN(x) (((x) & 0x1) << 25) +#define G_000040_SW_INT_EN(x) (((x) >> 25) & 0x1) +#define C_000040_SW_INT_EN 0xFDFFFFFF +#define S_000040_GEYSERVILLE(x) (((x) & 0x1) << 27) +#define G_000040_GEYSERVILLE(x) (((x) >> 27) & 0x1) +#define C_000040_GEYSERVILLE 0xF7FFFFFF +#define S_000040_HDCP_AUTHORIZED_INT(x) (((x) & 0x1) << 28) +#define G_000040_HDCP_AUTHORIZED_INT(x) (((x) >> 28) & 0x1) +#define C_000040_HDCP_AUTHORIZED_INT 0xEFFFFFFF +#define S_000040_DVI_I2C_INT(x) (((x) & 0x1) << 29) +#define G_000040_DVI_I2C_INT(x) (((x) >> 29) & 0x1) +#define C_000040_DVI_I2C_INT 0xDFFFFFFF +#define S_000040_GUIDMA(x) (((x) & 0x1) << 30) +#define G_000040_GUIDMA(x) (((x) >> 30) & 0x1) +#define C_000040_GUIDMA 0xBFFFFFFF +#define S_000040_VIDDMA(x) (((x) & 0x1) << 31) +#define G_000040_VIDDMA(x) (((x) >> 31) & 0x1) +#define C_000040_VIDDMA 0x7FFFFFFF +#define R_000044_GEN_INT_STATUS 0x000044 +#define S_000044_DISPLAY_INT_STAT(x) (((x) & 0x1) << 0) +#define G_000044_DISPLAY_INT_STAT(x) (((x) >> 0) & 0x1) +#define C_000044_DISPLAY_INT_STAT 0xFFFFFFFE +#define S_000044_VGA_INT_STAT(x) (((x) & 0x1) << 1) +#define G_000044_VGA_INT_STAT(x) (((x) >> 1) & 0x1) +#define C_000044_VGA_INT_STAT 0xFFFFFFFD +#define S_000044_CAP0_INT_ACTIVE(x) (((x) & 0x1) << 8) +#define G_000044_CAP0_INT_ACTIVE(x) (((x) >> 8) & 0x1) +#define C_000044_CAP0_INT_ACTIVE 0xFFFFFEFF +#define S_000044_DMA_VIPH0_INT(x) (((x) & 0x1) << 12) +#define G_000044_DMA_VIPH0_INT(x) (((x) >> 12) & 0x1) +#define C_000044_DMA_VIPH0_INT 0xFFFFEFFF +#define S_000044_DMA_VIPH1_INT(x) (((x) & 0x1) << 13) +#define G_000044_DMA_VIPH1_INT(x) (((x) >> 13) & 0x1) +#define C_000044_DMA_VIPH1_INT 0xFFFFDFFF +#define S_000044_DMA_VIPH2_INT(x) (((x) & 0x1) << 14) +#define G_000044_DMA_VIPH2_INT(x) (((x) >> 14) & 0x1) +#define C_000044_DMA_VIPH2_INT 0xFFFFBFFF +#define S_000044_DMA_VIPH3_INT(x) (((x) & 0x1) << 15) +#define G_000044_DMA_VIPH3_INT(x) (((x) >> 15) & 0x1) +#define C_000044_DMA_VIPH3_INT 0xFFFF7FFF +#define S_000044_MC_PROBE_FAULT_STAT(x) (((x) & 0x1) << 16) +#define G_000044_MC_PROBE_FAULT_STAT(x) (((x) >> 16) & 0x1) +#define C_000044_MC_PROBE_FAULT_STAT 0xFFFEFFFF +#define S_000044_I2C_INT(x) (((x) & 0x1) << 17) +#define G_000044_I2C_INT(x) (((x) >> 17) & 0x1) +#define C_000044_I2C_INT 0xFFFDFFFF +#define S_000044_SCRATCH_INT_STAT(x) (((x) & 0x1) << 18) +#define G_000044_SCRATCH_INT_STAT(x) (((x) >> 18) & 0x1) +#define C_000044_SCRATCH_INT_STAT 0xFFFBFFFF +#define S_000044_GUI_IDLE_STAT(x) (((x) & 0x1) << 19) +#define G_000044_GUI_IDLE_STAT(x) (((x) >> 19) & 0x1) +#define C_000044_GUI_IDLE_STAT 0xFFF7FFFF +#define S_000044_ATI_OVERDRIVE_INT_STAT(x) (((x) & 0x1) << 20) +#define G_000044_ATI_OVERDRIVE_INT_STAT(x) (((x) >> 20) & 0x1) +#define C_000044_ATI_OVERDRIVE_INT_STAT 0xFFEFFFFF +#define S_000044_MC_PROTECTION_FAULT_STAT(x) (((x) & 0x1) << 21) +#define G_000044_MC_PROTECTION_FAULT_STAT(x) (((x) >> 21) & 0x1) +#define C_000044_MC_PROTECTION_FAULT_STAT 0xFFDFFFFF +#define S_000044_RBBM_READ_INT_STAT(x) (((x) & 0x1) << 22) +#define G_000044_RBBM_READ_INT_STAT(x) (((x) >> 22) & 0x1) +#define C_000044_RBBM_READ_INT_STAT 0xFFBFFFFF +#define S_000044_CB_CONTEXT_SWITCH_STAT(x) (((x) & 0x1) << 23) +#define G_000044_CB_CONTEXT_SWITCH_STAT(x) (((x) >> 23) & 0x1) +#define C_000044_CB_CONTEXT_SWITCH_STAT 0xFF7FFFFF +#define S_000044_VIPH_INT(x) (((x) & 0x1) << 24) +#define G_000044_VIPH_INT(x) (((x) >> 24) & 0x1) +#define C_000044_VIPH_INT 0xFEFFFFFF +#define S_000044_SW_INT(x) (((x) & 0x1) << 25) +#define G_000044_SW_INT(x) (((x) >> 25) & 0x1) +#define C_000044_SW_INT 0xFDFFFFFF +#define S_000044_SW_INT_SET(x) (((x) & 0x1) << 26) +#define G_000044_SW_INT_SET(x) (((x) >> 26) & 0x1) +#define C_000044_SW_INT_SET 0xFBFFFFFF +#define S_000044_IDCT_INT_STAT(x) (((x) & 0x1) << 27) +#define G_000044_IDCT_INT_STAT(x) (((x) >> 27) & 0x1) +#define C_000044_IDCT_INT_STAT 0xF7FFFFFF +#define S_000044_GUIDMA_STAT(x) (((x) & 0x1) << 30) +#define G_000044_GUIDMA_STAT(x) (((x) >> 30) & 0x1) +#define C_000044_GUIDMA_STAT 0xBFFFFFFF +#define S_000044_VIDDMA_STAT(x) (((x) & 0x1) << 31) +#define G_000044_VIDDMA_STAT(x) (((x) >> 31) & 0x1) +#define C_000044_VIDDMA_STAT 0x7FFFFFFF +#define R_00004C_BUS_CNTL 0x00004C +#define S_00004C_BUS_MASTER_DIS(x) (((x) & 0x1) << 14) +#define G_00004C_BUS_MASTER_DIS(x) (((x) >> 14) & 0x1) +#define C_00004C_BUS_MASTER_DIS 0xFFFFBFFF +#define S_00004C_BUS_MSI_REARM(x) (((x) & 0x1) << 20) +#define G_00004C_BUS_MSI_REARM(x) (((x) >> 20) & 0x1) +#define C_00004C_BUS_MSI_REARM 0xFFEFFFFF +#define R_000070_MC_IND_INDEX 0x000070 +#define S_000070_MC_IND_ADDR(x) (((x) & 0xFFFF) << 0) +#define G_000070_MC_IND_ADDR(x) (((x) >> 0) & 0xFFFF) +#define C_000070_MC_IND_ADDR 0xFFFF0000 +#define S_000070_MC_IND_SEQ_RBS_0(x) (((x) & 0x1) << 16) +#define G_000070_MC_IND_SEQ_RBS_0(x) (((x) >> 16) & 0x1) +#define C_000070_MC_IND_SEQ_RBS_0 0xFFFEFFFF +#define S_000070_MC_IND_SEQ_RBS_1(x) (((x) & 0x1) << 17) +#define G_000070_MC_IND_SEQ_RBS_1(x) (((x) >> 17) & 0x1) +#define C_000070_MC_IND_SEQ_RBS_1 0xFFFDFFFF +#define S_000070_MC_IND_SEQ_RBS_2(x) (((x) & 0x1) << 18) +#define G_000070_MC_IND_SEQ_RBS_2(x) (((x) >> 18) & 0x1) +#define C_000070_MC_IND_SEQ_RBS_2 0xFFFBFFFF +#define S_000070_MC_IND_SEQ_RBS_3(x) (((x) & 0x1) << 19) +#define G_000070_MC_IND_SEQ_RBS_3(x) (((x) >> 19) & 0x1) +#define C_000070_MC_IND_SEQ_RBS_3 0xFFF7FFFF +#define S_000070_MC_IND_AIC_RBS(x) (((x) & 0x1) << 20) +#define G_000070_MC_IND_AIC_RBS(x) (((x) >> 20) & 0x1) +#define C_000070_MC_IND_AIC_RBS 0xFFEFFFFF +#define S_000070_MC_IND_CITF_ARB0(x) (((x) & 0x1) << 21) +#define G_000070_MC_IND_CITF_ARB0(x) (((x) >> 21) & 0x1) +#define C_000070_MC_IND_CITF_ARB0 0xFFDFFFFF +#define S_000070_MC_IND_CITF_ARB1(x) (((x) & 0x1) << 22) +#define G_000070_MC_IND_CITF_ARB1(x) (((x) >> 22) & 0x1) +#define C_000070_MC_IND_CITF_ARB1 0xFFBFFFFF +#define S_000070_MC_IND_WR_EN(x) (((x) & 0x1) << 23) +#define G_000070_MC_IND_WR_EN(x) (((x) >> 23) & 0x1) +#define C_000070_MC_IND_WR_EN 0xFF7FFFFF +#define S_000070_MC_IND_RD_INV(x) (((x) & 0x1) << 24) +#define G_000070_MC_IND_RD_INV(x) (((x) >> 24) & 0x1) +#define C_000070_MC_IND_RD_INV 0xFEFFFFFF +#define R_000074_MC_IND_DATA 0x000074 +#define S_000074_MC_IND_DATA(x) (((x) & 0xFFFFFFFF) << 0) +#define G_000074_MC_IND_DATA(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_000074_MC_IND_DATA 0x00000000 +#define R_000134_HDP_FB_LOCATION 0x000134 +#define S_000134_HDP_FB_START(x) (((x) & 0xFFFF) << 0) +#define G_000134_HDP_FB_START(x) (((x) >> 0) & 0xFFFF) +#define C_000134_HDP_FB_START 0xFFFF0000 +#define R_0007C0_CP_STAT 0x0007C0 +#define S_0007C0_MRU_BUSY(x) (((x) & 0x1) << 0) +#define G_0007C0_MRU_BUSY(x) (((x) >> 0) & 0x1) +#define C_0007C0_MRU_BUSY 0xFFFFFFFE +#define S_0007C0_MWU_BUSY(x) (((x) & 0x1) << 1) +#define G_0007C0_MWU_BUSY(x) (((x) >> 1) & 0x1) +#define C_0007C0_MWU_BUSY 0xFFFFFFFD +#define S_0007C0_RSIU_BUSY(x) (((x) & 0x1) << 2) +#define G_0007C0_RSIU_BUSY(x) (((x) >> 2) & 0x1) +#define C_0007C0_RSIU_BUSY 0xFFFFFFFB +#define S_0007C0_RCIU_BUSY(x) (((x) & 0x1) << 3) +#define G_0007C0_RCIU_BUSY(x) (((x) >> 3) & 0x1) +#define C_0007C0_RCIU_BUSY 0xFFFFFFF7 +#define S_0007C0_CSF_PRIMARY_BUSY(x) (((x) & 0x1) << 9) +#define G_0007C0_CSF_PRIMARY_BUSY(x) (((x) >> 9) & 0x1) +#define C_0007C0_CSF_PRIMARY_BUSY 0xFFFFFDFF +#define S_0007C0_CSF_INDIRECT_BUSY(x) (((x) & 0x1) << 10) +#define G_0007C0_CSF_INDIRECT_BUSY(x) (((x) >> 10) & 0x1) +#define C_0007C0_CSF_INDIRECT_BUSY 0xFFFFFBFF +#define S_0007C0_CSQ_PRIMARY_BUSY(x) (((x) & 0x1) << 11) +#define G_0007C0_CSQ_PRIMARY_BUSY(x) (((x) >> 11) & 0x1) +#define C_0007C0_CSQ_PRIMARY_BUSY 0xFFFFF7FF +#define S_0007C0_CSQ_INDIRECT_BUSY(x) (((x) & 0x1) << 12) +#define G_0007C0_CSQ_INDIRECT_BUSY(x) (((x) >> 12) & 0x1) +#define C_0007C0_CSQ_INDIRECT_BUSY 0xFFFFEFFF +#define S_0007C0_CSI_BUSY(x) (((x) & 0x1) << 13) +#define G_0007C0_CSI_BUSY(x) (((x) >> 13) & 0x1) +#define C_0007C0_CSI_BUSY 0xFFFFDFFF +#define S_0007C0_CSF_INDIRECT2_BUSY(x) (((x) & 0x1) << 14) +#define G_0007C0_CSF_INDIRECT2_BUSY(x) (((x) >> 14) & 0x1) +#define C_0007C0_CSF_INDIRECT2_BUSY 0xFFFFBFFF +#define S_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) & 0x1) << 15) +#define G_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) >> 15) & 0x1) +#define C_0007C0_CSQ_INDIRECT2_BUSY 0xFFFF7FFF +#define S_0007C0_GUIDMA_BUSY(x) (((x) & 0x1) << 28) +#define G_0007C0_GUIDMA_BUSY(x) (((x) >> 28) & 0x1) +#define C_0007C0_GUIDMA_BUSY 0xEFFFFFFF +#define S_0007C0_VIDDMA_BUSY(x) (((x) & 0x1) << 29) +#define G_0007C0_VIDDMA_BUSY(x) (((x) >> 29) & 0x1) +#define C_0007C0_VIDDMA_BUSY 0xDFFFFFFF +#define S_0007C0_CMDSTRM_BUSY(x) (((x) & 0x1) << 30) +#define G_0007C0_CMDSTRM_BUSY(x) (((x) >> 30) & 0x1) +#define C_0007C0_CMDSTRM_BUSY 0xBFFFFFFF +#define S_0007C0_CP_BUSY(x) (((x) & 0x1) << 31) +#define G_0007C0_CP_BUSY(x) (((x) >> 31) & 0x1) +#define C_0007C0_CP_BUSY 0x7FFFFFFF +#define R_000E40_RBBM_STATUS 0x000E40 +#define S_000E40_CMDFIFO_AVAIL(x) (((x) & 0x7F) << 0) +#define G_000E40_CMDFIFO_AVAIL(x) (((x) >> 0) & 0x7F) +#define C_000E40_CMDFIFO_AVAIL 0xFFFFFF80 +#define S_000E40_HIRQ_ON_RBB(x) (((x) & 0x1) << 8) +#define G_000E40_HIRQ_ON_RBB(x) (((x) >> 8) & 0x1) +#define C_000E40_HIRQ_ON_RBB 0xFFFFFEFF +#define S_000E40_CPRQ_ON_RBB(x) (((x) & 0x1) << 9) +#define G_000E40_CPRQ_ON_RBB(x) (((x) >> 9) & 0x1) +#define C_000E40_CPRQ_ON_RBB 0xFFFFFDFF +#define S_000E40_CFRQ_ON_RBB(x) (((x) & 0x1) << 10) +#define G_000E40_CFRQ_ON_RBB(x) (((x) >> 10) & 0x1) +#define C_000E40_CFRQ_ON_RBB 0xFFFFFBFF +#define S_000E40_HIRQ_IN_RTBUF(x) (((x) & 0x1) << 11) +#define G_000E40_HIRQ_IN_RTBUF(x) (((x) >> 11) & 0x1) +#define C_000E40_HIRQ_IN_RTBUF 0xFFFFF7FF +#define S_000E40_CPRQ_IN_RTBUF(x) (((x) & 0x1) << 12) +#define G_000E40_CPRQ_IN_RTBUF(x) (((x) >> 12) & 0x1) +#define C_000E40_CPRQ_IN_RTBUF 0xFFFFEFFF +#define S_000E40_CFRQ_IN_RTBUF(x) (((x) & 0x1) << 13) +#define G_000E40_CFRQ_IN_RTBUF(x) (((x) >> 13) & 0x1) +#define C_000E40_CFRQ_IN_RTBUF 0xFFFFDFFF +#define S_000E40_CF_PIPE_BUSY(x) (((x) & 0x1) << 14) +#define G_000E40_CF_PIPE_BUSY(x) (((x) >> 14) & 0x1) +#define C_000E40_CF_PIPE_BUSY 0xFFFFBFFF +#define S_000E40_ENG_EV_BUSY(x) (((x) & 0x1) << 15) +#define G_000E40_ENG_EV_BUSY(x) (((x) >> 15) & 0x1) +#define C_000E40_ENG_EV_BUSY 0xFFFF7FFF +#define S_000E40_CP_CMDSTRM_BUSY(x) (((x) & 0x1) << 16) +#define G_000E40_CP_CMDSTRM_BUSY(x) (((x) >> 16) & 0x1) +#define C_000E40_CP_CMDSTRM_BUSY 0xFFFEFFFF +#define S_000E40_E2_BUSY(x) (((x) & 0x1) << 17) +#define G_000E40_E2_BUSY(x) (((x) >> 17) & 0x1) +#define C_000E40_E2_BUSY 0xFFFDFFFF +#define S_000E40_RB2D_BUSY(x) (((x) & 0x1) << 18) +#define G_000E40_RB2D_BUSY(x) (((x) >> 18) & 0x1) +#define C_000E40_RB2D_BUSY 0xFFFBFFFF +#define S_000E40_RB3D_BUSY(x) (((x) & 0x1) << 19) +#define G_000E40_RB3D_BUSY(x) (((x) >> 19) & 0x1) +#define C_000E40_RB3D_BUSY 0xFFF7FFFF +#define S_000E40_VAP_BUSY(x) (((x) & 0x1) << 20) +#define G_000E40_VAP_BUSY(x) (((x) >> 20) & 0x1) +#define C_000E40_VAP_BUSY 0xFFEFFFFF +#define S_000E40_RE_BUSY(x) (((x) & 0x1) << 21) +#define G_000E40_RE_BUSY(x) (((x) >> 21) & 0x1) +#define C_000E40_RE_BUSY 0xFFDFFFFF +#define S_000E40_TAM_BUSY(x) (((x) & 0x1) << 22) +#define G_000E40_TAM_BUSY(x) (((x) >> 22) & 0x1) +#define C_000E40_TAM_BUSY 0xFFBFFFFF +#define S_000E40_TDM_BUSY(x) (((x) & 0x1) << 23) +#define G_000E40_TDM_BUSY(x) (((x) >> 23) & 0x1) +#define C_000E40_TDM_BUSY 0xFF7FFFFF +#define S_000E40_PB_BUSY(x) (((x) & 0x1) << 24) +#define G_000E40_PB_BUSY(x) (((x) >> 24) & 0x1) +#define C_000E40_PB_BUSY 0xFEFFFFFF +#define S_000E40_TIM_BUSY(x) (((x) & 0x1) << 25) +#define G_000E40_TIM_BUSY(x) (((x) >> 25) & 0x1) +#define C_000E40_TIM_BUSY 0xFDFFFFFF +#define S_000E40_GA_BUSY(x) (((x) & 0x1) << 26) +#define G_000E40_GA_BUSY(x) (((x) >> 26) & 0x1) +#define C_000E40_GA_BUSY 0xFBFFFFFF +#define S_000E40_CBA2D_BUSY(x) (((x) & 0x1) << 27) +#define G_000E40_CBA2D_BUSY(x) (((x) >> 27) & 0x1) +#define C_000E40_CBA2D_BUSY 0xF7FFFFFF +#define S_000E40_GUI_ACTIVE(x) (((x) & 0x1) << 31) +#define G_000E40_GUI_ACTIVE(x) (((x) >> 31) & 0x1) +#define C_000E40_GUI_ACTIVE 0x7FFFFFFF +#define R_0060A4_D1CRTC_STATUS_FRAME_COUNT 0x0060A4 +#define S_0060A4_D1CRTC_FRAME_COUNT(x) (((x) & 0xFFFFFF) << 0) +#define G_0060A4_D1CRTC_FRAME_COUNT(x) (((x) >> 0) & 0xFFFFFF) +#define C_0060A4_D1CRTC_FRAME_COUNT 0xFF000000 +#define R_006534_D1MODE_VBLANK_STATUS 0x006534 +#define S_006534_D1MODE_VBLANK_OCCURRED(x) (((x) & 0x1) << 0) +#define G_006534_D1MODE_VBLANK_OCCURRED(x) (((x) >> 0) & 0x1) +#define C_006534_D1MODE_VBLANK_OCCURRED 0xFFFFFFFE +#define S_006534_D1MODE_VBLANK_ACK(x) (((x) & 0x1) << 4) +#define G_006534_D1MODE_VBLANK_ACK(x) (((x) >> 4) & 0x1) +#define C_006534_D1MODE_VBLANK_ACK 0xFFFFFFEF +#define S_006534_D1MODE_VBLANK_STAT(x) (((x) & 0x1) << 12) +#define G_006534_D1MODE_VBLANK_STAT(x) (((x) >> 12) & 0x1) +#define C_006534_D1MODE_VBLANK_STAT 0xFFFFEFFF +#define S_006534_D1MODE_VBLANK_INTERRUPT(x) (((x) & 0x1) << 16) +#define G_006534_D1MODE_VBLANK_INTERRUPT(x) (((x) >> 16) & 0x1) +#define C_006534_D1MODE_VBLANK_INTERRUPT 0xFFFEFFFF +#define R_006540_DxMODE_INT_MASK 0x006540 +#define S_006540_D1MODE_VBLANK_INT_MASK(x) (((x) & 0x1) << 0) +#define G_006540_D1MODE_VBLANK_INT_MASK(x) (((x) >> 0) & 0x1) +#define C_006540_D1MODE_VBLANK_INT_MASK 0xFFFFFFFE +#define S_006540_D1MODE_VLINE_INT_MASK(x) (((x) & 0x1) << 4) +#define G_006540_D1MODE_VLINE_INT_MASK(x) (((x) >> 4) & 0x1) +#define C_006540_D1MODE_VLINE_INT_MASK 0xFFFFFFEF +#define S_006540_D2MODE_VBLANK_INT_MASK(x) (((x) & 0x1) << 8) +#define G_006540_D2MODE_VBLANK_INT_MASK(x) (((x) >> 8) & 0x1) +#define C_006540_D2MODE_VBLANK_INT_MASK 0xFFFFFEFF +#define S_006540_D2MODE_VLINE_INT_MASK(x) (((x) & 0x1) << 12) +#define G_006540_D2MODE_VLINE_INT_MASK(x) (((x) >> 12) & 0x1) +#define C_006540_D2MODE_VLINE_INT_MASK 0xFFFFEFFF +#define S_006540_D1MODE_VBLANK_CP_SEL(x) (((x) & 0x1) << 30) +#define G_006540_D1MODE_VBLANK_CP_SEL(x) (((x) >> 30) & 0x1) +#define C_006540_D1MODE_VBLANK_CP_SEL 0xBFFFFFFF +#define S_006540_D2MODE_VBLANK_CP_SEL(x) (((x) & 0x1) << 31) +#define G_006540_D2MODE_VBLANK_CP_SEL(x) (((x) >> 31) & 0x1) +#define C_006540_D2MODE_VBLANK_CP_SEL 0x7FFFFFFF +#define R_0068A4_D2CRTC_STATUS_FRAME_COUNT 0x0068A4 +#define S_0068A4_D2CRTC_FRAME_COUNT(x) (((x) & 0xFFFFFF) << 0) +#define G_0068A4_D2CRTC_FRAME_COUNT(x) (((x) >> 0) & 0xFFFFFF) +#define C_0068A4_D2CRTC_FRAME_COUNT 0xFF000000 +#define R_006D34_D2MODE_VBLANK_STATUS 0x006D34 +#define S_006D34_D2MODE_VBLANK_OCCURRED(x) (((x) & 0x1) << 0) +#define G_006D34_D2MODE_VBLANK_OCCURRED(x) (((x) >> 0) & 0x1) +#define C_006D34_D2MODE_VBLANK_OCCURRED 0xFFFFFFFE +#define S_006D34_D2MODE_VBLANK_ACK(x) (((x) & 0x1) << 4) +#define G_006D34_D2MODE_VBLANK_ACK(x) (((x) >> 4) & 0x1) +#define C_006D34_D2MODE_VBLANK_ACK 0xFFFFFFEF +#define S_006D34_D2MODE_VBLANK_STAT(x) (((x) & 0x1) << 12) +#define G_006D34_D2MODE_VBLANK_STAT(x) (((x) >> 12) & 0x1) +#define C_006D34_D2MODE_VBLANK_STAT 0xFFFFEFFF +#define S_006D34_D2MODE_VBLANK_INTERRUPT(x) (((x) & 0x1) << 16) +#define G_006D34_D2MODE_VBLANK_INTERRUPT(x) (((x) >> 16) & 0x1) +#define C_006D34_D2MODE_VBLANK_INTERRUPT 0xFFFEFFFF +#define R_007EDC_DISP_INTERRUPT_STATUS 0x007EDC +#define S_007EDC_LB_D1_VBLANK_INTERRUPT(x) (((x) & 0x1) << 4) +#define G_007EDC_LB_D1_VBLANK_INTERRUPT(x) (((x) >> 4) & 0x1) +#define C_007EDC_LB_D1_VBLANK_INTERRUPT 0xFFFFFFEF +#define S_007EDC_LB_D2_VBLANK_INTERRUPT(x) (((x) & 0x1) << 5) +#define G_007EDC_LB_D2_VBLANK_INTERRUPT(x) (((x) >> 5) & 0x1) +#define C_007EDC_LB_D2_VBLANK_INTERRUPT 0xFFFFFFDF + + +/* MC registers */ +#define R_000000_MC_STATUS 0x000000 +#define S_000000_MC_IDLE(x) (((x) & 0x1) << 0) +#define G_000000_MC_IDLE(x) (((x) >> 0) & 0x1) +#define C_000000_MC_IDLE 0xFFFFFFFE +#define R_000004_MC_FB_LOCATION 0x000004 +#define S_000004_MC_FB_START(x) (((x) & 0xFFFF) << 0) +#define G_000004_MC_FB_START(x) (((x) >> 0) & 0xFFFF) +#define C_000004_MC_FB_START 0xFFFF0000 +#define S_000004_MC_FB_TOP(x) (((x) & 0xFFFF) << 16) +#define G_000004_MC_FB_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_000004_MC_FB_TOP 0x0000FFFF +#define R_000005_MC_AGP_LOCATION 0x000005 +#define S_000005_MC_AGP_START(x) (((x) & 0xFFFF) << 0) +#define G_000005_MC_AGP_START(x) (((x) >> 0) & 0xFFFF) +#define C_000005_MC_AGP_START 0xFFFF0000 +#define S_000005_MC_AGP_TOP(x) (((x) & 0xFFFF) << 16) +#define G_000005_MC_AGP_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_000005_MC_AGP_TOP 0x0000FFFF +#define R_000006_AGP_BASE 0x000006 +#define S_000006_AGP_BASE_ADDR(x) (((x) & 0xFFFFFFFF) << 0) +#define G_000006_AGP_BASE_ADDR(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_000006_AGP_BASE_ADDR 0x00000000 +#define R_000007_AGP_BASE_2 0x000007 +#define S_000007_AGP_BASE_ADDR_2(x) (((x) & 0xF) << 0) +#define G_000007_AGP_BASE_ADDR_2(x) (((x) >> 0) & 0xF) +#define C_000007_AGP_BASE_ADDR_2 0xFFFFFFF0 +#define R_000009_MC_CNTL1 0x000009 +#define S_000009_ENABLE_PAGE_TABLES(x) (((x) & 0x1) << 26) +#define G_000009_ENABLE_PAGE_TABLES(x) (((x) >> 26) & 0x1) +#define C_000009_ENABLE_PAGE_TABLES 0xFBFFFFFF +/* FIXME don't know the various field size need feedback from AMD */ +#define R_000100_MC_PT0_CNTL 0x000100 +#define S_000100_ENABLE_PT(x) (((x) & 0x1) << 0) +#define G_000100_ENABLE_PT(x) (((x) >> 0) & 0x1) +#define C_000100_ENABLE_PT 0xFFFFFFFE +#define S_000100_EFFECTIVE_L2_CACHE_SIZE(x) (((x) & 0x7) << 15) +#define G_000100_EFFECTIVE_L2_CACHE_SIZE(x) (((x) >> 15) & 0x7) +#define C_000100_EFFECTIVE_L2_CACHE_SIZE 0xFFFC7FFF +#define S_000100_EFFECTIVE_L2_QUEUE_SIZE(x) (((x) & 0x7) << 21) +#define G_000100_EFFECTIVE_L2_QUEUE_SIZE(x) (((x) >> 21) & 0x7) +#define C_000100_EFFECTIVE_L2_QUEUE_SIZE 0xFF1FFFFF +#define S_000100_INVALIDATE_ALL_L1_TLBS(x) (((x) & 0x1) << 28) +#define G_000100_INVALIDATE_ALL_L1_TLBS(x) (((x) >> 28) & 0x1) +#define C_000100_INVALIDATE_ALL_L1_TLBS 0xEFFFFFFF +#define S_000100_INVALIDATE_L2_CACHE(x) (((x) & 0x1) << 29) +#define G_000100_INVALIDATE_L2_CACHE(x) (((x) >> 29) & 0x1) +#define C_000100_INVALIDATE_L2_CACHE 0xDFFFFFFF +#define R_000102_MC_PT0_CONTEXT0_CNTL 0x000102 +#define S_000102_ENABLE_PAGE_TABLE(x) (((x) & 0x1) << 0) +#define G_000102_ENABLE_PAGE_TABLE(x) (((x) >> 0) & 0x1) +#define C_000102_ENABLE_PAGE_TABLE 0xFFFFFFFE +#define S_000102_PAGE_TABLE_DEPTH(x) (((x) & 0x3) << 1) +#define G_000102_PAGE_TABLE_DEPTH(x) (((x) >> 1) & 0x3) +#define C_000102_PAGE_TABLE_DEPTH 0xFFFFFFF9 +#define V_000102_PAGE_TABLE_FLAT 0 +/* R600 documentation suggest that this should be a number of pages */ +#define R_000112_MC_PT0_SYSTEM_APERTURE_LOW_ADDR 0x000112 +#define R_000114_MC_PT0_SYSTEM_APERTURE_HIGH_ADDR 0x000114 +#define R_00011C_MC_PT0_CONTEXT0_DEFAULT_READ_ADDR 0x00011C +#define R_00012C_MC_PT0_CONTEXT0_FLAT_BASE_ADDR 0x00012C +#define R_00013C_MC_PT0_CONTEXT0_FLAT_START_ADDR 0x00013C +#define R_00014C_MC_PT0_CONTEXT0_FLAT_END_ADDR 0x00014C +#define R_00016C_MC_PT0_CLIENT0_CNTL 0x00016C +#define S_00016C_ENABLE_TRANSLATION_MODE_OVERRIDE(x) (((x) & 0x1) << 0) +#define G_00016C_ENABLE_TRANSLATION_MODE_OVERRIDE(x) (((x) >> 0) & 0x1) +#define C_00016C_ENABLE_TRANSLATION_MODE_OVERRIDE 0xFFFFFFFE +#define S_00016C_TRANSLATION_MODE_OVERRIDE(x) (((x) & 0x1) << 1) +#define G_00016C_TRANSLATION_MODE_OVERRIDE(x) (((x) >> 1) & 0x1) +#define C_00016C_TRANSLATION_MODE_OVERRIDE 0xFFFFFFFD +#define S_00016C_SYSTEM_ACCESS_MODE_MASK(x) (((x) & 0x3) << 8) +#define G_00016C_SYSTEM_ACCESS_MODE_MASK(x) (((x) >> 8) & 0x3) +#define C_00016C_SYSTEM_ACCESS_MODE_MASK 0xFFFFFCFF +#define V_00016C_SYSTEM_ACCESS_MODE_PA_ONLY 0 +#define V_00016C_SYSTEM_ACCESS_MODE_USE_SYS_MAP 1 +#define V_00016C_SYSTEM_ACCESS_MODE_IN_SYS 2 +#define V_00016C_SYSTEM_ACCESS_MODE_NOT_IN_SYS 3 +#define S_00016C_SYSTEM_APERTURE_UNMAPPED_ACCESS(x) (((x) & 0x1) << 10) +#define G_00016C_SYSTEM_APERTURE_UNMAPPED_ACCESS(x) (((x) >> 10) & 0x1) +#define C_00016C_SYSTEM_APERTURE_UNMAPPED_ACCESS 0xFFFFFBFF +#define V_00016C_SYSTEM_APERTURE_UNMAPPED_PASSTHROUGH 0 +#define V_00016C_SYSTEM_APERTURE_UNMAPPED_DEFAULT_PAGE 1 +#define S_00016C_EFFECTIVE_L1_CACHE_SIZE(x) (((x) & 0x7) << 11) +#define G_00016C_EFFECTIVE_L1_CACHE_SIZE(x) (((x) >> 11) & 0x7) +#define C_00016C_EFFECTIVE_L1_CACHE_SIZE 0xFFFFC7FF +#define S_00016C_ENABLE_FRAGMENT_PROCESSING(x) (((x) & 0x1) << 14) +#define G_00016C_ENABLE_FRAGMENT_PROCESSING(x) (((x) >> 14) & 0x1) +#define C_00016C_ENABLE_FRAGMENT_PROCESSING 0xFFFFBFFF +#define S_00016C_EFFECTIVE_L1_QUEUE_SIZE(x) (((x) & 0x7) << 15) +#define G_00016C_EFFECTIVE_L1_QUEUE_SIZE(x) (((x) >> 15) & 0x7) +#define C_00016C_EFFECTIVE_L1_QUEUE_SIZE 0xFFFC7FFF +#define S_00016C_INVALIDATE_L1_TLB(x) (((x) & 0x1) << 20) +#define G_00016C_INVALIDATE_L1_TLB(x) (((x) >> 20) & 0x1) +#define C_00016C_INVALIDATE_L1_TLB 0xFFEFFFFF + +#endif diff --git a/drivers/gpu/drm/radeon/rs690.c b/drivers/gpu/drm/radeon/rs690.c index 0f585ca..2754717 100644 --- a/drivers/gpu/drm/radeon/rs690.c +++ b/drivers/gpu/drm/radeon/rs690.c @@ -26,106 +26,29 @@ * Jerome Glisse */ #include "drmP.h" -#include "radeon_reg.h" #include "radeon.h" -#include "rs690r.h" #include "atom.h" -#include "atom-bits.h" - -/* rs690,rs740 depends on : */ -void r100_hdp_reset(struct radeon_device *rdev); -int r300_mc_wait_for_idle(struct radeon_device *rdev); -void r420_pipes_init(struct radeon_device *rdev); -void rs400_gart_disable(struct radeon_device *rdev); -int rs400_gart_enable(struct radeon_device *rdev); -void rs400_gart_adjust_size(struct radeon_device *rdev); -void rs600_mc_disable_clients(struct radeon_device *rdev); -void rs600_disable_vga(struct radeon_device *rdev); - -/* This files gather functions specifics to : - * rs690,rs740 - * - * Some of these functions might be used by newer ASICs. - */ -void rs690_gpu_init(struct radeon_device *rdev); -int rs690_mc_wait_for_idle(struct radeon_device *rdev); - +#include "rs690d.h" -/* - * MC functions. - */ -int rs690_mc_init(struct radeon_device *rdev) -{ - uint32_t tmp; - int r; - - if (r100_debugfs_rbbm_init(rdev)) { - DRM_ERROR("Failed to register debugfs file for RBBM !\n"); - } - - rs690_gpu_init(rdev); - rs400_gart_disable(rdev); - - /* Setup GPU memory space */ - rdev->mc.gtt_location = rdev->mc.mc_vram_size; - rdev->mc.gtt_location += (rdev->mc.gtt_size - 1); - rdev->mc.gtt_location &= ~(rdev->mc.gtt_size - 1); - rdev->mc.vram_location = 0xFFFFFFFFUL; - r = radeon_mc_setup(rdev); - if (r) { - return r; - } - - /* Program GPU memory space */ - rs600_mc_disable_clients(rdev); - if (rs690_mc_wait_for_idle(rdev)) { - printk(KERN_WARNING "Failed to wait MC idle while " - "programming pipes. Bad things might happen.\n"); - } - tmp = rdev->mc.vram_location + rdev->mc.mc_vram_size - 1; - tmp = REG_SET(RS690_MC_FB_TOP, tmp >> 16); - tmp |= REG_SET(RS690_MC_FB_START, rdev->mc.vram_location >> 16); - WREG32_MC(RS690_MCCFG_FB_LOCATION, tmp); - /* FIXME: Does this reg exist on RS480,RS740 ? */ - WREG32(0x310, rdev->mc.vram_location); - WREG32(RS690_HDP_FB_LOCATION, rdev->mc.vram_location >> 16); - return 0; -} - -void rs690_mc_fini(struct radeon_device *rdev) -{ -} - - -/* - * Global GPU functions - */ -int rs690_mc_wait_for_idle(struct radeon_device *rdev) +static int rs690_mc_wait_for_idle(struct radeon_device *rdev) { unsigned i; uint32_t tmp; for (i = 0; i < rdev->usec_timeout; i++) { /* read MC_STATUS */ - tmp = RREG32_MC(RS690_MC_STATUS); - if (tmp & RS690_MC_STATUS_IDLE) { + tmp = RREG32_MC(R_000090_MC_SYSTEM_STATUS); + if (G_000090_MC_SYSTEM_IDLE(tmp)) return 0; - } - DRM_UDELAY(1); + udelay(1); } return -1; } -void rs690_errata(struct radeon_device *rdev) -{ - rdev->pll_errata = 0; -} - -void rs690_gpu_init(struct radeon_device *rdev) +static void rs690_gpu_init(struct radeon_device *rdev) { /* FIXME: HDP same place on rs690 ? */ r100_hdp_reset(rdev); - rs600_disable_vga(rdev); /* FIXME: is this correct ? */ r420_pipes_init(rdev); if (rs690_mc_wait_for_idle(rdev)) { @@ -134,10 +57,6 @@ void rs690_gpu_init(struct radeon_device *rdev) } } - -/* - * VRAM info. - */ void rs690_pm_info(struct radeon_device *rdev) { int index = GetIndexIntoMasterTable(DATA, IntegratedSystemInfo); @@ -251,39 +170,39 @@ void rs690_line_buffer_adjust(struct radeon_device *rdev, /* * Line Buffer Setup * There is a single line buffer shared by both display controllers. - * DC_LB_MEMORY_SPLIT controls how that line buffer is shared between + * R_006520_DC_LB_MEMORY_SPLIT controls how that line buffer is shared between * the display controllers. The paritioning can either be done * manually or via one of four preset allocations specified in bits 1:0: * 0 - line buffer is divided in half and shared between crtc * 1 - D1 gets 3/4 of the line buffer, D2 gets 1/4 * 2 - D1 gets the whole buffer * 3 - D1 gets 1/4 of the line buffer, D2 gets 3/4 - * Setting bit 2 of DC_LB_MEMORY_SPLIT controls switches to manual + * Setting bit 2 of R_006520_DC_LB_MEMORY_SPLIT controls switches to manual * allocation mode. In manual allocation mode, D1 always starts at 0, * D1 end/2 is specified in bits 14:4; D2 allocation follows D1. */ - tmp = RREG32(DC_LB_MEMORY_SPLIT) & ~DC_LB_MEMORY_SPLIT_MASK; - tmp &= ~DC_LB_MEMORY_SPLIT_SHIFT_MODE; + tmp = RREG32(R_006520_DC_LB_MEMORY_SPLIT) & C_006520_DC_LB_MEMORY_SPLIT; + tmp &= ~C_006520_DC_LB_MEMORY_SPLIT_MODE; /* auto */ if (mode1 && mode2) { if (mode1->hdisplay > mode2->hdisplay) { if (mode1->hdisplay > 2560) - tmp |= DC_LB_MEMORY_SPLIT_D1_3Q_D2_1Q; + tmp |= V_006520_DC_LB_MEMORY_SPLIT_D1_3Q_D2_1Q; else - tmp |= DC_LB_MEMORY_SPLIT_D1HALF_D2HALF; + tmp |= V_006520_DC_LB_MEMORY_SPLIT_D1HALF_D2HALF; } else if (mode2->hdisplay > mode1->hdisplay) { if (mode2->hdisplay > 2560) - tmp |= DC_LB_MEMORY_SPLIT_D1_1Q_D2_3Q; + tmp |= V_006520_DC_LB_MEMORY_SPLIT_D1_1Q_D2_3Q; else - tmp |= DC_LB_MEMORY_SPLIT_D1HALF_D2HALF; + tmp |= V_006520_DC_LB_MEMORY_SPLIT_D1HALF_D2HALF; } else - tmp |= AVIVO_DC_LB_MEMORY_SPLIT_D1HALF_D2HALF; + tmp |= V_006520_DC_LB_MEMORY_SPLIT_D1HALF_D2HALF; } else if (mode1) { - tmp |= DC_LB_MEMORY_SPLIT_D1_ONLY; + tmp |= V_006520_DC_LB_MEMORY_SPLIT_D1_ONLY; } else if (mode2) { - tmp |= DC_LB_MEMORY_SPLIT_D1_1Q_D2_3Q; + tmp |= V_006520_DC_LB_MEMORY_SPLIT_D1_1Q_D2_3Q; } - WREG32(DC_LB_MEMORY_SPLIT, tmp); + WREG32(R_006520_DC_LB_MEMORY_SPLIT, tmp); } struct rs690_watermark { @@ -488,28 +407,28 @@ void rs690_bandwidth_update(struct radeon_device *rdev) * option. */ if (rdev->disp_priority == 2) { - tmp = RREG32_MC(MC_INIT_MISC_LAT_TIMER); - tmp &= ~MC_DISP1R_INIT_LAT_MASK; - tmp &= ~MC_DISP0R_INIT_LAT_MASK; - if (mode1) - tmp |= (1 << MC_DISP1R_INIT_LAT_SHIFT); + tmp = RREG32_MC(R_000104_MC_INIT_MISC_LAT_TIMER); + tmp &= C_000104_MC_DISP0R_INIT_LAT; + tmp &= C_000104_MC_DISP1R_INIT_LAT; if (mode0) - tmp |= (1 << MC_DISP0R_INIT_LAT_SHIFT); - WREG32_MC(MC_INIT_MISC_LAT_TIMER, tmp); + tmp |= S_000104_MC_DISP0R_INIT_LAT(1); + if (mode1) + tmp |= S_000104_MC_DISP1R_INIT_LAT(1); + WREG32_MC(R_000104_MC_INIT_MISC_LAT_TIMER, tmp); } rs690_line_buffer_adjust(rdev, mode0, mode1); if ((rdev->family == CHIP_RS690) || (rdev->family == CHIP_RS740)) - WREG32(DCP_CONTROL, 0); + WREG32(R_006C9C_DCP_CONTROL, 0); if ((rdev->family == CHIP_RS780) || (rdev->family == CHIP_RS880)) - WREG32(DCP_CONTROL, 2); + WREG32(R_006C9C_DCP_CONTROL, 2); rs690_crtc_bandwidth_compute(rdev, rdev->mode_info.crtcs[0], &wm0); rs690_crtc_bandwidth_compute(rdev, rdev->mode_info.crtcs[1], &wm1); tmp = (wm0.lb_request_fifo_depth - 1); tmp |= (wm1.lb_request_fifo_depth - 1) << 16; - WREG32(LB_MAX_REQ_OUTSTANDING, tmp); + WREG32(R_006D58_LB_MAX_REQ_OUTSTANDING, tmp); if (mode0 && mode1) { if (rfixed_trunc(wm0.dbpp) > 64) @@ -562,10 +481,10 @@ void rs690_bandwidth_update(struct radeon_device *rdev) priority_mark12.full = 0; if (wm1.priority_mark_max.full > priority_mark12.full) priority_mark12.full = wm1.priority_mark_max.full; - WREG32(D1MODE_PRIORITY_A_CNT, rfixed_trunc(priority_mark02)); - WREG32(D1MODE_PRIORITY_B_CNT, rfixed_trunc(priority_mark02)); - WREG32(D2MODE_PRIORITY_A_CNT, rfixed_trunc(priority_mark12)); - WREG32(D2MODE_PRIORITY_B_CNT, rfixed_trunc(priority_mark12)); + WREG32(R_006548_D1MODE_PRIORITY_A_CNT, rfixed_trunc(priority_mark02)); + WREG32(R_00654C_D1MODE_PRIORITY_B_CNT, rfixed_trunc(priority_mark02)); + WREG32(R_006D48_D2MODE_PRIORITY_A_CNT, rfixed_trunc(priority_mark12)); + WREG32(R_006D4C_D2MODE_PRIORITY_B_CNT, rfixed_trunc(priority_mark12)); } else if (mode0) { if (rfixed_trunc(wm0.dbpp) > 64) a.full = rfixed_mul(wm0.dbpp, wm0.num_line_pair); @@ -592,10 +511,12 @@ void rs690_bandwidth_update(struct radeon_device *rdev) priority_mark02.full = 0; if (wm0.priority_mark_max.full > priority_mark02.full) priority_mark02.full = wm0.priority_mark_max.full; - WREG32(D1MODE_PRIORITY_A_CNT, rfixed_trunc(priority_mark02)); - WREG32(D1MODE_PRIORITY_B_CNT, rfixed_trunc(priority_mark02)); - WREG32(D2MODE_PRIORITY_A_CNT, MODE_PRIORITY_OFF); - WREG32(D2MODE_PRIORITY_B_CNT, MODE_PRIORITY_OFF); + WREG32(R_006548_D1MODE_PRIORITY_A_CNT, rfixed_trunc(priority_mark02)); + WREG32(R_00654C_D1MODE_PRIORITY_B_CNT, rfixed_trunc(priority_mark02)); + WREG32(R_006D48_D2MODE_PRIORITY_A_CNT, + S_006D48_D2MODE_PRIORITY_A_OFF(1)); + WREG32(R_006D4C_D2MODE_PRIORITY_B_CNT, + S_006D4C_D2MODE_PRIORITY_B_OFF(1)); } else { if (rfixed_trunc(wm1.dbpp) > 64) a.full = rfixed_mul(wm1.dbpp, wm1.num_line_pair); @@ -622,30 +543,205 @@ void rs690_bandwidth_update(struct radeon_device *rdev) priority_mark12.full = 0; if (wm1.priority_mark_max.full > priority_mark12.full) priority_mark12.full = wm1.priority_mark_max.full; - WREG32(D1MODE_PRIORITY_A_CNT, MODE_PRIORITY_OFF); - WREG32(D1MODE_PRIORITY_B_CNT, MODE_PRIORITY_OFF); - WREG32(D2MODE_PRIORITY_A_CNT, rfixed_trunc(priority_mark12)); - WREG32(D2MODE_PRIORITY_B_CNT, rfixed_trunc(priority_mark12)); + WREG32(R_006548_D1MODE_PRIORITY_A_CNT, + S_006548_D1MODE_PRIORITY_A_OFF(1)); + WREG32(R_00654C_D1MODE_PRIORITY_B_CNT, + S_00654C_D1MODE_PRIORITY_B_OFF(1)); + WREG32(R_006D48_D2MODE_PRIORITY_A_CNT, rfixed_trunc(priority_mark12)); + WREG32(R_006D4C_D2MODE_PRIORITY_B_CNT, rfixed_trunc(priority_mark12)); } } -/* - * Indirect registers accessor - */ uint32_t rs690_mc_rreg(struct radeon_device *rdev, uint32_t reg) { uint32_t r; - WREG32(RS690_MC_INDEX, (reg & RS690_MC_INDEX_MASK)); - r = RREG32(RS690_MC_DATA); - WREG32(RS690_MC_INDEX, RS690_MC_INDEX_MASK); + WREG32(R_000078_MC_INDEX, S_000078_MC_IND_ADDR(reg)); + r = RREG32(R_00007C_MC_DATA); + WREG32(R_000078_MC_INDEX, ~C_000078_MC_IND_ADDR); return r; } void rs690_mc_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v) { - WREG32(RS690_MC_INDEX, - RS690_MC_INDEX_WR_EN | ((reg) & RS690_MC_INDEX_MASK)); - WREG32(RS690_MC_DATA, v); - WREG32(RS690_MC_INDEX, RS690_MC_INDEX_WR_ACK); + WREG32(R_000078_MC_INDEX, S_000078_MC_IND_ADDR(reg) | + S_000078_MC_IND_WR_EN(1)); + WREG32(R_00007C_MC_DATA, v); + WREG32(R_000078_MC_INDEX, 0x7F); +} + +void rs690_mc_program(struct radeon_device *rdev) +{ + struct rv515_mc_save save; + + /* Stops all mc clients */ + rv515_mc_stop(rdev, &save); + + /* Wait for mc idle */ + if (rs690_mc_wait_for_idle(rdev)) + dev_warn(rdev->dev, "Wait MC idle timeout before updating MC.\n"); + /* Program MC, should be a 32bits limited address space */ + WREG32_MC(R_000100_MCCFG_FB_LOCATION, + S_000100_MC_FB_START(rdev->mc.vram_start >> 16) | + S_000100_MC_FB_TOP(rdev->mc.vram_end >> 16)); + WREG32(R_000134_HDP_FB_LOCATION, + S_000134_HDP_FB_START(rdev->mc.vram_start >> 16)); + + rv515_mc_resume(rdev, &save); +} + +static int rs690_startup(struct radeon_device *rdev) +{ + int r; + + rs690_mc_program(rdev); + /* Resume clock */ + rv515_clock_startup(rdev); + /* Initialize GPU configuration (# pipes, ...) */ + rs690_gpu_init(rdev); + /* Initialize GART (initialize after TTM so we can allocate + * memory through TTM but finalize after TTM) */ + r = rs400_gart_enable(rdev); + if (r) + return r; + /* Enable IRQ */ + rdev->irq.sw_int = true; + rs600_irq_set(rdev); + /* 1M ring buffer */ + r = r100_cp_init(rdev, 1024 * 1024); + if (r) { + dev_err(rdev->dev, "failled initializing CP (%d).\n", r); + return r; + } + r = r100_wb_init(rdev); + if (r) + dev_err(rdev->dev, "failled initializing WB (%d).\n", r); + r = r100_ib_init(rdev); + if (r) { + dev_err(rdev->dev, "failled initializing IB (%d).\n", r); + return r; + } + return 0; +} + +int rs690_resume(struct radeon_device *rdev) +{ + /* Make sur GART are not working */ + rs400_gart_disable(rdev); + /* Resume clock before doing reset */ + rv515_clock_startup(rdev); + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_gpu_reset(rdev)) { + dev_warn(rdev->dev, "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* post */ + atom_asic_init(rdev->mode_info.atom_context); + /* Resume clock after posting */ + rv515_clock_startup(rdev); + return rs690_startup(rdev); +} + +int rs690_suspend(struct radeon_device *rdev) +{ + r100_cp_disable(rdev); + r100_wb_disable(rdev); + rs600_irq_disable(rdev); + rs400_gart_disable(rdev); + return 0; +} + +void rs690_fini(struct radeon_device *rdev) +{ + rs690_suspend(rdev); + r100_cp_fini(rdev); + r100_wb_fini(rdev); + r100_ib_fini(rdev); + radeon_gem_fini(rdev); + rs400_gart_fini(rdev); + radeon_irq_kms_fini(rdev); + radeon_fence_driver_fini(rdev); + radeon_object_fini(rdev); + radeon_atombios_fini(rdev); + kfree(rdev->bios); + rdev->bios = NULL; +} + +int rs690_init(struct radeon_device *rdev) +{ + int r; + + /* Disable VGA */ + rv515_vga_render_disable(rdev); + /* Initialize scratch registers */ + radeon_scratch_init(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + /* TODO: disable VGA need to use VGA request */ + /* BIOS*/ + if (!radeon_get_bios(rdev)) { + if (ASIC_IS_AVIVO(rdev)) + return -EINVAL; + } + if (rdev->is_atom_bios) { + r = radeon_atombios_init(rdev); + if (r) + return r; + } else { + dev_err(rdev->dev, "Expecting atombios for RV515 GPU\n"); + return -EINVAL; + } + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_gpu_reset(rdev)) { + dev_warn(rdev->dev, + "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* check if cards are posted or not */ + if (!radeon_card_posted(rdev) && rdev->bios) { + DRM_INFO("GPU not posted. posting now...\n"); + atom_asic_init(rdev->mode_info.atom_context); + } + /* Initialize clocks */ + radeon_get_clock_info(rdev->ddev); + /* Initialize power management */ + radeon_pm_init(rdev); + /* Get vram informations */ + rs690_vram_info(rdev); + /* Initialize memory controller (also test AGP) */ + r = r420_mc_init(rdev); + if (r) + return r; + rv515_debugfs(rdev); + /* Fence driver */ + r = radeon_fence_driver_init(rdev); + if (r) + return r; + r = radeon_irq_kms_init(rdev); + if (r) + return r; + /* Memory manager */ + r = radeon_object_init(rdev); + if (r) + return r; + r = rs400_gart_init(rdev); + if (r) + return r; + rs600_set_safe_registers(rdev); + rdev->accel_working = true; + r = rs690_startup(rdev); + if (r) { + /* Somethings want wront with the accel init stop accel */ + dev_err(rdev->dev, "Disabling GPU acceleration\n"); + rs690_suspend(rdev); + r100_cp_fini(rdev); + r100_wb_fini(rdev); + r100_ib_fini(rdev); + rs400_gart_fini(rdev); + radeon_irq_kms_fini(rdev); + rdev->accel_working = false; + } + return 0; } diff --git a/drivers/gpu/drm/radeon/rs690d.h b/drivers/gpu/drm/radeon/rs690d.h new file mode 100644 index 0000000..62d31e7 --- /dev/null +++ b/drivers/gpu/drm/radeon/rs690d.h @@ -0,0 +1,307 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef __RS690D_H__ +#define __RS690D_H__ + +/* Registers */ +#define R_000078_MC_INDEX 0x000078 +#define S_000078_MC_IND_ADDR(x) (((x) & 0x1FF) << 0) +#define G_000078_MC_IND_ADDR(x) (((x) >> 0) & 0x1FF) +#define C_000078_MC_IND_ADDR 0xFFFFFE00 +#define S_000078_MC_IND_WR_EN(x) (((x) & 0x1) << 9) +#define G_000078_MC_IND_WR_EN(x) (((x) >> 9) & 0x1) +#define C_000078_MC_IND_WR_EN 0xFFFFFDFF +#define R_00007C_MC_DATA 0x00007C +#define S_00007C_MC_DATA(x) (((x) & 0xFFFFFFFF) << 0) +#define G_00007C_MC_DATA(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_00007C_MC_DATA 0x00000000 +#define R_0000F8_CONFIG_MEMSIZE 0x0000F8 +#define S_0000F8_CONFIG_MEMSIZE(x) (((x) & 0xFFFFFFFF) << 0) +#define G_0000F8_CONFIG_MEMSIZE(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_0000F8_CONFIG_MEMSIZE 0x00000000 +#define R_000134_HDP_FB_LOCATION 0x000134 +#define S_000134_HDP_FB_START(x) (((x) & 0xFFFF) << 0) +#define G_000134_HDP_FB_START(x) (((x) >> 0) & 0xFFFF) +#define C_000134_HDP_FB_START 0xFFFF0000 +#define R_0007C0_CP_STAT 0x0007C0 +#define S_0007C0_MRU_BUSY(x) (((x) & 0x1) << 0) +#define G_0007C0_MRU_BUSY(x) (((x) >> 0) & 0x1) +#define C_0007C0_MRU_BUSY 0xFFFFFFFE +#define S_0007C0_MWU_BUSY(x) (((x) & 0x1) << 1) +#define G_0007C0_MWU_BUSY(x) (((x) >> 1) & 0x1) +#define C_0007C0_MWU_BUSY 0xFFFFFFFD +#define S_0007C0_RSIU_BUSY(x) (((x) & 0x1) << 2) +#define G_0007C0_RSIU_BUSY(x) (((x) >> 2) & 0x1) +#define C_0007C0_RSIU_BUSY 0xFFFFFFFB +#define S_0007C0_RCIU_BUSY(x) (((x) & 0x1) << 3) +#define G_0007C0_RCIU_BUSY(x) (((x) >> 3) & 0x1) +#define C_0007C0_RCIU_BUSY 0xFFFFFFF7 +#define S_0007C0_CSF_PRIMARY_BUSY(x) (((x) & 0x1) << 9) +#define G_0007C0_CSF_PRIMARY_BUSY(x) (((x) >> 9) & 0x1) +#define C_0007C0_CSF_PRIMARY_BUSY 0xFFFFFDFF +#define S_0007C0_CSF_INDIRECT_BUSY(x) (((x) & 0x1) << 10) +#define G_0007C0_CSF_INDIRECT_BUSY(x) (((x) >> 10) & 0x1) +#define C_0007C0_CSF_INDIRECT_BUSY 0xFFFFFBFF +#define S_0007C0_CSQ_PRIMARY_BUSY(x) (((x) & 0x1) << 11) +#define G_0007C0_CSQ_PRIMARY_BUSY(x) (((x) >> 11) & 0x1) +#define C_0007C0_CSQ_PRIMARY_BUSY 0xFFFFF7FF +#define S_0007C0_CSQ_INDIRECT_BUSY(x) (((x) & 0x1) << 12) +#define G_0007C0_CSQ_INDIRECT_BUSY(x) (((x) >> 12) & 0x1) +#define C_0007C0_CSQ_INDIRECT_BUSY 0xFFFFEFFF +#define S_0007C0_CSI_BUSY(x) (((x) & 0x1) << 13) +#define G_0007C0_CSI_BUSY(x) (((x) >> 13) & 0x1) +#define C_0007C0_CSI_BUSY 0xFFFFDFFF +#define S_0007C0_CSF_INDIRECT2_BUSY(x) (((x) & 0x1) << 14) +#define G_0007C0_CSF_INDIRECT2_BUSY(x) (((x) >> 14) & 0x1) +#define C_0007C0_CSF_INDIRECT2_BUSY 0xFFFFBFFF +#define S_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) & 0x1) << 15) +#define G_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) >> 15) & 0x1) +#define C_0007C0_CSQ_INDIRECT2_BUSY 0xFFFF7FFF +#define S_0007C0_GUIDMA_BUSY(x) (((x) & 0x1) << 28) +#define G_0007C0_GUIDMA_BUSY(x) (((x) >> 28) & 0x1) +#define C_0007C0_GUIDMA_BUSY 0xEFFFFFFF +#define S_0007C0_VIDDMA_BUSY(x) (((x) & 0x1) << 29) +#define G_0007C0_VIDDMA_BUSY(x) (((x) >> 29) & 0x1) +#define C_0007C0_VIDDMA_BUSY 0xDFFFFFFF +#define S_0007C0_CMDSTRM_BUSY(x) (((x) & 0x1) << 30) +#define G_0007C0_CMDSTRM_BUSY(x) (((x) >> 30) & 0x1) +#define C_0007C0_CMDSTRM_BUSY 0xBFFFFFFF +#define S_0007C0_CP_BUSY(x) (((x) & 0x1) << 31) +#define G_0007C0_CP_BUSY(x) (((x) >> 31) & 0x1) +#define C_0007C0_CP_BUSY 0x7FFFFFFF +#define R_000E40_RBBM_STATUS 0x000E40 +#define S_000E40_CMDFIFO_AVAIL(x) (((x) & 0x7F) << 0) +#define G_000E40_CMDFIFO_AVAIL(x) (((x) >> 0) & 0x7F) +#define C_000E40_CMDFIFO_AVAIL 0xFFFFFF80 +#define S_000E40_HIRQ_ON_RBB(x) (((x) & 0x1) << 8) +#define G_000E40_HIRQ_ON_RBB(x) (((x) >> 8) & 0x1) +#define C_000E40_HIRQ_ON_RBB 0xFFFFFEFF +#define S_000E40_CPRQ_ON_RBB(x) (((x) & 0x1) << 9) +#define G_000E40_CPRQ_ON_RBB(x) (((x) >> 9) & 0x1) +#define C_000E40_CPRQ_ON_RBB 0xFFFFFDFF +#define S_000E40_CFRQ_ON_RBB(x) (((x) & 0x1) << 10) +#define G_000E40_CFRQ_ON_RBB(x) (((x) >> 10) & 0x1) +#define C_000E40_CFRQ_ON_RBB 0xFFFFFBFF +#define S_000E40_HIRQ_IN_RTBUF(x) (((x) & 0x1) << 11) +#define G_000E40_HIRQ_IN_RTBUF(x) (((x) >> 11) & 0x1) +#define C_000E40_HIRQ_IN_RTBUF 0xFFFFF7FF +#define S_000E40_CPRQ_IN_RTBUF(x) (((x) & 0x1) << 12) +#define G_000E40_CPRQ_IN_RTBUF(x) (((x) >> 12) & 0x1) +#define C_000E40_CPRQ_IN_RTBUF 0xFFFFEFFF +#define S_000E40_CFRQ_IN_RTBUF(x) (((x) & 0x1) << 13) +#define G_000E40_CFRQ_IN_RTBUF(x) (((x) >> 13) & 0x1) +#define C_000E40_CFRQ_IN_RTBUF 0xFFFFDFFF +#define S_000E40_CF_PIPE_BUSY(x) (((x) & 0x1) << 14) +#define G_000E40_CF_PIPE_BUSY(x) (((x) >> 14) & 0x1) +#define C_000E40_CF_PIPE_BUSY 0xFFFFBFFF +#define S_000E40_ENG_EV_BUSY(x) (((x) & 0x1) << 15) +#define G_000E40_ENG_EV_BUSY(x) (((x) >> 15) & 0x1) +#define C_000E40_ENG_EV_BUSY 0xFFFF7FFF +#define S_000E40_CP_CMDSTRM_BUSY(x) (((x) & 0x1) << 16) +#define G_000E40_CP_CMDSTRM_BUSY(x) (((x) >> 16) & 0x1) +#define C_000E40_CP_CMDSTRM_BUSY 0xFFFEFFFF +#define S_000E40_E2_BUSY(x) (((x) & 0x1) << 17) +#define G_000E40_E2_BUSY(x) (((x) >> 17) & 0x1) +#define C_000E40_E2_BUSY 0xFFFDFFFF +#define S_000E40_RB2D_BUSY(x) (((x) & 0x1) << 18) +#define G_000E40_RB2D_BUSY(x) (((x) >> 18) & 0x1) +#define C_000E40_RB2D_BUSY 0xFFFBFFFF +#define S_000E40_RB3D_BUSY(x) (((x) & 0x1) << 19) +#define G_000E40_RB3D_BUSY(x) (((x) >> 19) & 0x1) +#define C_000E40_RB3D_BUSY 0xFFF7FFFF +#define S_000E40_VAP_BUSY(x) (((x) & 0x1) << 20) +#define G_000E40_VAP_BUSY(x) (((x) >> 20) & 0x1) +#define C_000E40_VAP_BUSY 0xFFEFFFFF +#define S_000E40_RE_BUSY(x) (((x) & 0x1) << 21) +#define G_000E40_RE_BUSY(x) (((x) >> 21) & 0x1) +#define C_000E40_RE_BUSY 0xFFDFFFFF +#define S_000E40_TAM_BUSY(x) (((x) & 0x1) << 22) +#define G_000E40_TAM_BUSY(x) (((x) >> 22) & 0x1) +#define C_000E40_TAM_BUSY 0xFFBFFFFF +#define S_000E40_TDM_BUSY(x) (((x) & 0x1) << 23) +#define G_000E40_TDM_BUSY(x) (((x) >> 23) & 0x1) +#define C_000E40_TDM_BUSY 0xFF7FFFFF +#define S_000E40_PB_BUSY(x) (((x) & 0x1) << 24) +#define G_000E40_PB_BUSY(x) (((x) >> 24) & 0x1) +#define C_000E40_PB_BUSY 0xFEFFFFFF +#define S_000E40_TIM_BUSY(x) (((x) & 0x1) << 25) +#define G_000E40_TIM_BUSY(x) (((x) >> 25) & 0x1) +#define C_000E40_TIM_BUSY 0xFDFFFFFF +#define S_000E40_GA_BUSY(x) (((x) & 0x1) << 26) +#define G_000E40_GA_BUSY(x) (((x) >> 26) & 0x1) +#define C_000E40_GA_BUSY 0xFBFFFFFF +#define S_000E40_CBA2D_BUSY(x) (((x) & 0x1) << 27) +#define G_000E40_CBA2D_BUSY(x) (((x) >> 27) & 0x1) +#define C_000E40_CBA2D_BUSY 0xF7FFFFFF +#define S_000E40_GUI_ACTIVE(x) (((x) & 0x1) << 31) +#define G_000E40_GUI_ACTIVE(x) (((x) >> 31) & 0x1) +#define C_000E40_GUI_ACTIVE 0x7FFFFFFF +#define R_006520_DC_LB_MEMORY_SPLIT 0x006520 +#define S_006520_DC_LB_MEMORY_SPLIT(x) (((x) & 0x3) << 0) +#define G_006520_DC_LB_MEMORY_SPLIT(x) (((x) >> 0) & 0x3) +#define C_006520_DC_LB_MEMORY_SPLIT 0xFFFFFFFC +#define S_006520_DC_LB_MEMORY_SPLIT_MODE(x) (((x) & 0x1) << 2) +#define G_006520_DC_LB_MEMORY_SPLIT_MODE(x) (((x) >> 2) & 0x1) +#define C_006520_DC_LB_MEMORY_SPLIT_MODE 0xFFFFFFFB +#define V_006520_DC_LB_MEMORY_SPLIT_D1HALF_D2HALF 0 +#define V_006520_DC_LB_MEMORY_SPLIT_D1_3Q_D2_1Q 1 +#define V_006520_DC_LB_MEMORY_SPLIT_D1_ONLY 2 +#define V_006520_DC_LB_MEMORY_SPLIT_D1_1Q_D2_3Q 3 +#define S_006520_DC_LB_DISP1_END_ADR(x) (((x) & 0x7FF) << 4) +#define G_006520_DC_LB_DISP1_END_ADR(x) (((x) >> 4) & 0x7FF) +#define C_006520_DC_LB_DISP1_END_ADR 0xFFFF800F +#define R_006548_D1MODE_PRIORITY_A_CNT 0x006548 +#define S_006548_D1MODE_PRIORITY_MARK_A(x) (((x) & 0x7FFF) << 0) +#define G_006548_D1MODE_PRIORITY_MARK_A(x) (((x) >> 0) & 0x7FFF) +#define C_006548_D1MODE_PRIORITY_MARK_A 0xFFFF8000 +#define S_006548_D1MODE_PRIORITY_A_OFF(x) (((x) & 0x1) << 16) +#define G_006548_D1MODE_PRIORITY_A_OFF(x) (((x) >> 16) & 0x1) +#define C_006548_D1MODE_PRIORITY_A_OFF 0xFFFEFFFF +#define S_006548_D1MODE_PRIORITY_A_FORCE_MASK(x) (((x) & 0x1) << 24) +#define G_006548_D1MODE_PRIORITY_A_FORCE_MASK(x) (((x) >> 24) & 0x1) +#define C_006548_D1MODE_PRIORITY_A_FORCE_MASK 0xFEFFFFFF +#define R_00654C_D1MODE_PRIORITY_B_CNT 0x00654C +#define S_00654C_D1MODE_PRIORITY_MARK_B(x) (((x) & 0x7FFF) << 0) +#define G_00654C_D1MODE_PRIORITY_MARK_B(x) (((x) >> 0) & 0x7FFF) +#define C_00654C_D1MODE_PRIORITY_MARK_B 0xFFFF8000 +#define S_00654C_D1MODE_PRIORITY_B_OFF(x) (((x) & 0x1) << 16) +#define G_00654C_D1MODE_PRIORITY_B_OFF(x) (((x) >> 16) & 0x1) +#define C_00654C_D1MODE_PRIORITY_B_OFF 0xFFFEFFFF +#define S_00654C_D1MODE_PRIORITY_B_ALWAYS_ON(x) (((x) & 0x1) << 20) +#define G_00654C_D1MODE_PRIORITY_B_ALWAYS_ON(x) (((x) >> 20) & 0x1) +#define C_00654C_D1MODE_PRIORITY_B_ALWAYS_ON 0xFFEFFFFF +#define S_00654C_D1MODE_PRIORITY_B_FORCE_MASK(x) (((x) & 0x1) << 24) +#define G_00654C_D1MODE_PRIORITY_B_FORCE_MASK(x) (((x) >> 24) & 0x1) +#define C_00654C_D1MODE_PRIORITY_B_FORCE_MASK 0xFEFFFFFF +#define R_006C9C_DCP_CONTROL 0x006C9C +#define R_006D48_D2MODE_PRIORITY_A_CNT 0x006D48 +#define S_006D48_D2MODE_PRIORITY_MARK_A(x) (((x) & 0x7FFF) << 0) +#define G_006D48_D2MODE_PRIORITY_MARK_A(x) (((x) >> 0) & 0x7FFF) +#define C_006D48_D2MODE_PRIORITY_MARK_A 0xFFFF8000 +#define S_006D48_D2MODE_PRIORITY_A_OFF(x) (((x) & 0x1) << 16) +#define G_006D48_D2MODE_PRIORITY_A_OFF(x) (((x) >> 16) & 0x1) +#define C_006D48_D2MODE_PRIORITY_A_OFF 0xFFFEFFFF +#define S_006D48_D2MODE_PRIORITY_A_ALWAYS_ON(x) (((x) & 0x1) << 20) +#define G_006D48_D2MODE_PRIORITY_A_ALWAYS_ON(x) (((x) >> 20) & 0x1) +#define C_006D48_D2MODE_PRIORITY_A_ALWAYS_ON 0xFFEFFFFF +#define S_006D48_D2MODE_PRIORITY_A_FORCE_MASK(x) (((x) & 0x1) << 24) +#define G_006D48_D2MODE_PRIORITY_A_FORCE_MASK(x) (((x) >> 24) & 0x1) +#define C_006D48_D2MODE_PRIORITY_A_FORCE_MASK 0xFEFFFFFF +#define R_006D4C_D2MODE_PRIORITY_B_CNT 0x006D4C +#define S_006D4C_D2MODE_PRIORITY_MARK_B(x) (((x) & 0x7FFF) << 0) +#define G_006D4C_D2MODE_PRIORITY_MARK_B(x) (((x) >> 0) & 0x7FFF) +#define C_006D4C_D2MODE_PRIORITY_MARK_B 0xFFFF8000 +#define S_006D4C_D2MODE_PRIORITY_B_OFF(x) (((x) & 0x1) << 16) +#define G_006D4C_D2MODE_PRIORITY_B_OFF(x) (((x) >> 16) & 0x1) +#define C_006D4C_D2MODE_PRIORITY_B_OFF 0xFFFEFFFF +#define S_006D4C_D2MODE_PRIORITY_B_ALWAYS_ON(x) (((x) & 0x1) << 20) +#define G_006D4C_D2MODE_PRIORITY_B_ALWAYS_ON(x) (((x) >> 20) & 0x1) +#define C_006D4C_D2MODE_PRIORITY_B_ALWAYS_ON 0xFFEFFFFF +#define S_006D4C_D2MODE_PRIORITY_B_FORCE_MASK(x) (((x) & 0x1) << 24) +#define G_006D4C_D2MODE_PRIORITY_B_FORCE_MASK(x) (((x) >> 24) & 0x1) +#define C_006D4C_D2MODE_PRIORITY_B_FORCE_MASK 0xFEFFFFFF +#define R_006D58_LB_MAX_REQ_OUTSTANDING 0x006D58 +#define S_006D58_LB_D1_MAX_REQ_OUTSTANDING(x) (((x) & 0xF) << 0) +#define G_006D58_LB_D1_MAX_REQ_OUTSTANDING(x) (((x) >> 0) & 0xF) +#define C_006D58_LB_D1_MAX_REQ_OUTSTANDING 0xFFFFFFF0 +#define S_006D58_LB_D2_MAX_REQ_OUTSTANDING(x) (((x) & 0xF) << 16) +#define G_006D58_LB_D2_MAX_REQ_OUTSTANDING(x) (((x) >> 16) & 0xF) +#define C_006D58_LB_D2_MAX_REQ_OUTSTANDING 0xFFF0FFFF + + +#define R_000090_MC_SYSTEM_STATUS 0x000090 +#define S_000090_MC_SYSTEM_IDLE(x) (((x) & 0x1) << 0) +#define G_000090_MC_SYSTEM_IDLE(x) (((x) >> 0) & 0x1) +#define C_000090_MC_SYSTEM_IDLE 0xFFFFFFFE +#define S_000090_MC_SEQUENCER_IDLE(x) (((x) & 0x1) << 1) +#define G_000090_MC_SEQUENCER_IDLE(x) (((x) >> 1) & 0x1) +#define C_000090_MC_SEQUENCER_IDLE 0xFFFFFFFD +#define S_000090_MC_ARBITER_IDLE(x) (((x) & 0x1) << 2) +#define G_000090_MC_ARBITER_IDLE(x) (((x) >> 2) & 0x1) +#define C_000090_MC_ARBITER_IDLE 0xFFFFFFFB +#define S_000090_MC_SELECT_PM(x) (((x) & 0x1) << 3) +#define G_000090_MC_SELECT_PM(x) (((x) >> 3) & 0x1) +#define C_000090_MC_SELECT_PM 0xFFFFFFF7 +#define S_000090_RESERVED4(x) (((x) & 0xF) << 4) +#define G_000090_RESERVED4(x) (((x) >> 4) & 0xF) +#define C_000090_RESERVED4 0xFFFFFF0F +#define S_000090_RESERVED8(x) (((x) & 0xF) << 8) +#define G_000090_RESERVED8(x) (((x) >> 8) & 0xF) +#define C_000090_RESERVED8 0xFFFFF0FF +#define S_000090_RESERVED12(x) (((x) & 0xF) << 12) +#define G_000090_RESERVED12(x) (((x) >> 12) & 0xF) +#define C_000090_RESERVED12 0xFFFF0FFF +#define S_000090_MCA_INIT_EXECUTED(x) (((x) & 0x1) << 16) +#define G_000090_MCA_INIT_EXECUTED(x) (((x) >> 16) & 0x1) +#define C_000090_MCA_INIT_EXECUTED 0xFFFEFFFF +#define S_000090_MCA_IDLE(x) (((x) & 0x1) << 17) +#define G_000090_MCA_IDLE(x) (((x) >> 17) & 0x1) +#define C_000090_MCA_IDLE 0xFFFDFFFF +#define S_000090_MCA_SEQ_IDLE(x) (((x) & 0x1) << 18) +#define G_000090_MCA_SEQ_IDLE(x) (((x) >> 18) & 0x1) +#define C_000090_MCA_SEQ_IDLE 0xFFFBFFFF +#define S_000090_MCA_ARB_IDLE(x) (((x) & 0x1) << 19) +#define G_000090_MCA_ARB_IDLE(x) (((x) >> 19) & 0x1) +#define C_000090_MCA_ARB_IDLE 0xFFF7FFFF +#define S_000090_RESERVED20(x) (((x) & 0xFFF) << 20) +#define G_000090_RESERVED20(x) (((x) >> 20) & 0xFFF) +#define C_000090_RESERVED20 0x000FFFFF +#define R_000100_MCCFG_FB_LOCATION 0x000100 +#define S_000100_MC_FB_START(x) (((x) & 0xFFFF) << 0) +#define G_000100_MC_FB_START(x) (((x) >> 0) & 0xFFFF) +#define C_000100_MC_FB_START 0xFFFF0000 +#define S_000100_MC_FB_TOP(x) (((x) & 0xFFFF) << 16) +#define G_000100_MC_FB_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_000100_MC_FB_TOP 0x0000FFFF +#define R_000104_MC_INIT_MISC_LAT_TIMER 0x000104 +#define S_000104_MC_CPR_INIT_LAT(x) (((x) & 0xF) << 0) +#define G_000104_MC_CPR_INIT_LAT(x) (((x) >> 0) & 0xF) +#define C_000104_MC_CPR_INIT_LAT 0xFFFFFFF0 +#define S_000104_MC_VF_INIT_LAT(x) (((x) & 0xF) << 4) +#define G_000104_MC_VF_INIT_LAT(x) (((x) >> 4) & 0xF) +#define C_000104_MC_VF_INIT_LAT 0xFFFFFF0F +#define S_000104_MC_DISP0R_INIT_LAT(x) (((x) & 0xF) << 8) +#define G_000104_MC_DISP0R_INIT_LAT(x) (((x) >> 8) & 0xF) +#define C_000104_MC_DISP0R_INIT_LAT 0xFFFFF0FF +#define S_000104_MC_DISP1R_INIT_LAT(x) (((x) & 0xF) << 12) +#define G_000104_MC_DISP1R_INIT_LAT(x) (((x) >> 12) & 0xF) +#define C_000104_MC_DISP1R_INIT_LAT 0xFFFF0FFF +#define S_000104_MC_FIXED_INIT_LAT(x) (((x) & 0xF) << 16) +#define G_000104_MC_FIXED_INIT_LAT(x) (((x) >> 16) & 0xF) +#define C_000104_MC_FIXED_INIT_LAT 0xFFF0FFFF +#define S_000104_MC_E2R_INIT_LAT(x) (((x) & 0xF) << 20) +#define G_000104_MC_E2R_INIT_LAT(x) (((x) >> 20) & 0xF) +#define C_000104_MC_E2R_INIT_LAT 0xFF0FFFFF +#define S_000104_SAME_PAGE_PRIO(x) (((x) & 0xF) << 24) +#define G_000104_SAME_PAGE_PRIO(x) (((x) >> 24) & 0xF) +#define C_000104_SAME_PAGE_PRIO 0xF0FFFFFF +#define S_000104_MC_GLOBW_INIT_LAT(x) (((x) & 0xF) << 28) +#define G_000104_MC_GLOBW_INIT_LAT(x) (((x) >> 28) & 0xF) +#define C_000104_MC_GLOBW_INIT_LAT 0x0FFFFFFF + +#endif diff --git a/drivers/gpu/drm/radeon/rs690r.h b/drivers/gpu/drm/radeon/rs690r.h deleted file mode 100644 index c0d9faa..0000000 --- a/drivers/gpu/drm/radeon/rs690r.h +++ /dev/null @@ -1,99 +0,0 @@ -/* - * Copyright 2008 Advanced Micro Devices, Inc. - * Copyright 2008 Red Hat Inc. - * Copyright 2009 Jerome Glisse. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: Dave Airlie - * Alex Deucher - * Jerome Glisse - */ -#ifndef RS690R_H -#define RS690R_H - -/* RS690/RS740 registers */ -#define MC_INDEX 0x0078 -# define MC_INDEX_MASK 0x1FF -# define MC_INDEX_WR_EN (1 << 9) -# define MC_INDEX_WR_ACK 0x7F -#define MC_DATA 0x007C -#define HDP_FB_LOCATION 0x0134 -#define DC_LB_MEMORY_SPLIT 0x6520 -#define DC_LB_MEMORY_SPLIT_MASK 0x00000003 -#define DC_LB_MEMORY_SPLIT_SHIFT 0 -#define DC_LB_MEMORY_SPLIT_D1HALF_D2HALF 0 -#define DC_LB_MEMORY_SPLIT_D1_3Q_D2_1Q 1 -#define DC_LB_MEMORY_SPLIT_D1_ONLY 2 -#define DC_LB_MEMORY_SPLIT_D1_1Q_D2_3Q 3 -#define DC_LB_MEMORY_SPLIT_SHIFT_MODE (1 << 2) -#define DC_LB_DISP1_END_ADR_SHIFT 4 -#define DC_LB_DISP1_END_ADR_MASK 0x00007FF0 -#define D1MODE_PRIORITY_A_CNT 0x6548 -#define MODE_PRIORITY_MARK_MASK 0x00007FFF -#define MODE_PRIORITY_OFF (1 << 16) -#define MODE_PRIORITY_ALWAYS_ON (1 << 20) -#define MODE_PRIORITY_FORCE_MASK (1 << 24) -#define D1MODE_PRIORITY_B_CNT 0x654C -#define LB_MAX_REQ_OUTSTANDING 0x6D58 -#define LB_D1_MAX_REQ_OUTSTANDING_MASK 0x0000000F -#define LB_D1_MAX_REQ_OUTSTANDING_SHIFT 0 -#define LB_D2_MAX_REQ_OUTSTANDING_MASK 0x000F0000 -#define LB_D2_MAX_REQ_OUTSTANDING_SHIFT 16 -#define DCP_CONTROL 0x6C9C -#define D2MODE_PRIORITY_A_CNT 0x6D48 -#define D2MODE_PRIORITY_B_CNT 0x6D4C - -/* MC indirect registers */ -#define MC_STATUS_IDLE (1 << 0) -#define MC_MISC_CNTL 0x18 -#define DISABLE_GTW (1 << 1) -#define GART_INDEX_REG_EN (1 << 12) -#define BLOCK_GFX_D3_EN (1 << 14) -#define GART_FEATURE_ID 0x2B -#define HANG_EN (1 << 11) -#define TLB_ENABLE (1 << 18) -#define P2P_ENABLE (1 << 19) -#define GTW_LAC_EN (1 << 25) -#define LEVEL2_GART (0 << 30) -#define LEVEL1_GART (1 << 30) -#define PDC_EN (1 << 31) -#define GART_BASE 0x2C -#define GART_CACHE_CNTRL 0x2E -# define GART_CACHE_INVALIDATE (1 << 0) -#define MC_STATUS 0x90 -#define MCCFG_FB_LOCATION 0x100 -#define MC_FB_START_MASK 0x0000FFFF -#define MC_FB_START_SHIFT 0 -#define MC_FB_TOP_MASK 0xFFFF0000 -#define MC_FB_TOP_SHIFT 16 -#define MCCFG_AGP_LOCATION 0x101 -#define MC_AGP_START_MASK 0x0000FFFF -#define MC_AGP_START_SHIFT 0 -#define MC_AGP_TOP_MASK 0xFFFF0000 -#define MC_AGP_TOP_SHIFT 16 -#define MCCFG_AGP_BASE 0x102 -#define MCCFG_AGP_BASE_2 0x103 -#define MC_INIT_MISC_LAT_TIMER 0x104 -#define MC_DISP0R_INIT_LAT_SHIFT 8 -#define MC_DISP0R_INIT_LAT_MASK 0x00000F00 -#define MC_DISP1R_INIT_LAT_SHIFT 12 -#define MC_DISP1R_INIT_LAT_MASK 0x0000F000 - -#endif diff --git a/drivers/gpu/drm/radeon/rv200d.h b/drivers/gpu/drm/radeon/rv200d.h new file mode 100644 index 0000000..c5b3983 --- /dev/null +++ b/drivers/gpu/drm/radeon/rv200d.h @@ -0,0 +1,36 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef __RV200D_H__ +#define __RV200D_H__ + +#define R_00015C_AGP_BASE_2 0x00015C +#define S_00015C_AGP_BASE_ADDR_2(x) (((x) & 0xF) << 0) +#define G_00015C_AGP_BASE_ADDR_2(x) (((x) >> 0) & 0xF) +#define C_00015C_AGP_BASE_ADDR_2 0xFFFFFFF0 + +#endif diff --git a/drivers/gpu/drm/radeon/rv250d.h b/drivers/gpu/drm/radeon/rv250d.h new file mode 100644 index 0000000..e5a70b0 --- /dev/null +++ b/drivers/gpu/drm/radeon/rv250d.h @@ -0,0 +1,123 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef __RV250D_H__ +#define __RV250D_H__ + +#define R_00000D_SCLK_CNTL_M6 0x00000D +#define S_00000D_SCLK_SRC_SEL(x) (((x) & 0x7) << 0) +#define G_00000D_SCLK_SRC_SEL(x) (((x) >> 0) & 0x7) +#define C_00000D_SCLK_SRC_SEL 0xFFFFFFF8 +#define S_00000D_CP_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 3) +#define G_00000D_CP_MAX_DYN_STOP_LAT(x) (((x) >> 3) & 0x1) +#define C_00000D_CP_MAX_DYN_STOP_LAT 0xFFFFFFF7 +#define S_00000D_HDP_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 4) +#define G_00000D_HDP_MAX_DYN_STOP_LAT(x) (((x) >> 4) & 0x1) +#define C_00000D_HDP_MAX_DYN_STOP_LAT 0xFFFFFFEF +#define S_00000D_TV_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 5) +#define G_00000D_TV_MAX_DYN_STOP_LAT(x) (((x) >> 5) & 0x1) +#define C_00000D_TV_MAX_DYN_STOP_LAT 0xFFFFFFDF +#define S_00000D_E2_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 6) +#define G_00000D_E2_MAX_DYN_STOP_LAT(x) (((x) >> 6) & 0x1) +#define C_00000D_E2_MAX_DYN_STOP_LAT 0xFFFFFFBF +#define S_00000D_SE_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 7) +#define G_00000D_SE_MAX_DYN_STOP_LAT(x) (((x) >> 7) & 0x1) +#define C_00000D_SE_MAX_DYN_STOP_LAT 0xFFFFFF7F +#define S_00000D_IDCT_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 8) +#define G_00000D_IDCT_MAX_DYN_STOP_LAT(x) (((x) >> 8) & 0x1) +#define C_00000D_IDCT_MAX_DYN_STOP_LAT 0xFFFFFEFF +#define S_00000D_VIP_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 9) +#define G_00000D_VIP_MAX_DYN_STOP_LAT(x) (((x) >> 9) & 0x1) +#define C_00000D_VIP_MAX_DYN_STOP_LAT 0xFFFFFDFF +#define S_00000D_RE_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 10) +#define G_00000D_RE_MAX_DYN_STOP_LAT(x) (((x) >> 10) & 0x1) +#define C_00000D_RE_MAX_DYN_STOP_LAT 0xFFFFFBFF +#define S_00000D_PB_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 11) +#define G_00000D_PB_MAX_DYN_STOP_LAT(x) (((x) >> 11) & 0x1) +#define C_00000D_PB_MAX_DYN_STOP_LAT 0xFFFFF7FF +#define S_00000D_TAM_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 12) +#define G_00000D_TAM_MAX_DYN_STOP_LAT(x) (((x) >> 12) & 0x1) +#define C_00000D_TAM_MAX_DYN_STOP_LAT 0xFFFFEFFF +#define S_00000D_TDM_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 13) +#define G_00000D_TDM_MAX_DYN_STOP_LAT(x) (((x) >> 13) & 0x1) +#define C_00000D_TDM_MAX_DYN_STOP_LAT 0xFFFFDFFF +#define S_00000D_RB_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 14) +#define G_00000D_RB_MAX_DYN_STOP_LAT(x) (((x) >> 14) & 0x1) +#define C_00000D_RB_MAX_DYN_STOP_LAT 0xFFFFBFFF +#define S_00000D_FORCE_DISP2(x) (((x) & 0x1) << 15) +#define G_00000D_FORCE_DISP2(x) (((x) >> 15) & 0x1) +#define C_00000D_FORCE_DISP2 0xFFFF7FFF +#define S_00000D_FORCE_CP(x) (((x) & 0x1) << 16) +#define G_00000D_FORCE_CP(x) (((x) >> 16) & 0x1) +#define C_00000D_FORCE_CP 0xFFFEFFFF +#define S_00000D_FORCE_HDP(x) (((x) & 0x1) << 17) +#define G_00000D_FORCE_HDP(x) (((x) >> 17) & 0x1) +#define C_00000D_FORCE_HDP 0xFFFDFFFF +#define S_00000D_FORCE_DISP1(x) (((x) & 0x1) << 18) +#define G_00000D_FORCE_DISP1(x) (((x) >> 18) & 0x1) +#define C_00000D_FORCE_DISP1 0xFFFBFFFF +#define S_00000D_FORCE_TOP(x) (((x) & 0x1) << 19) +#define G_00000D_FORCE_TOP(x) (((x) >> 19) & 0x1) +#define C_00000D_FORCE_TOP 0xFFF7FFFF +#define S_00000D_FORCE_E2(x) (((x) & 0x1) << 20) +#define G_00000D_FORCE_E2(x) (((x) >> 20) & 0x1) +#define C_00000D_FORCE_E2 0xFFEFFFFF +#define S_00000D_FORCE_SE(x) (((x) & 0x1) << 21) +#define G_00000D_FORCE_SE(x) (((x) >> 21) & 0x1) +#define C_00000D_FORCE_SE 0xFFDFFFFF +#define S_00000D_FORCE_IDCT(x) (((x) & 0x1) << 22) +#define G_00000D_FORCE_IDCT(x) (((x) >> 22) & 0x1) +#define C_00000D_FORCE_IDCT 0xFFBFFFFF +#define S_00000D_FORCE_VIP(x) (((x) & 0x1) << 23) +#define G_00000D_FORCE_VIP(x) (((x) >> 23) & 0x1) +#define C_00000D_FORCE_VIP 0xFF7FFFFF +#define S_00000D_FORCE_RE(x) (((x) & 0x1) << 24) +#define G_00000D_FORCE_RE(x) (((x) >> 24) & 0x1) +#define C_00000D_FORCE_RE 0xFEFFFFFF +#define S_00000D_FORCE_PB(x) (((x) & 0x1) << 25) +#define G_00000D_FORCE_PB(x) (((x) >> 25) & 0x1) +#define C_00000D_FORCE_PB 0xFDFFFFFF +#define S_00000D_FORCE_TAM(x) (((x) & 0x1) << 26) +#define G_00000D_FORCE_TAM(x) (((x) >> 26) & 0x1) +#define C_00000D_FORCE_TAM 0xFBFFFFFF +#define S_00000D_FORCE_TDM(x) (((x) & 0x1) << 27) +#define G_00000D_FORCE_TDM(x) (((x) >> 27) & 0x1) +#define C_00000D_FORCE_TDM 0xF7FFFFFF +#define S_00000D_FORCE_RB(x) (((x) & 0x1) << 28) +#define G_00000D_FORCE_RB(x) (((x) >> 28) & 0x1) +#define C_00000D_FORCE_RB 0xEFFFFFFF +#define S_00000D_FORCE_TV_SCLK(x) (((x) & 0x1) << 29) +#define G_00000D_FORCE_TV_SCLK(x) (((x) >> 29) & 0x1) +#define C_00000D_FORCE_TV_SCLK 0xDFFFFFFF +#define S_00000D_FORCE_SUBPIC(x) (((x) & 0x1) << 30) +#define G_00000D_FORCE_SUBPIC(x) (((x) >> 30) & 0x1) +#define C_00000D_FORCE_SUBPIC 0xBFFFFFFF +#define S_00000D_FORCE_OV0(x) (((x) & 0x1) << 31) +#define G_00000D_FORCE_OV0(x) (((x) >> 31) & 0x1) +#define C_00000D_FORCE_OV0 0x7FFFFFFF + +#endif diff --git a/drivers/gpu/drm/radeon/rv350d.h b/drivers/gpu/drm/radeon/rv350d.h new file mode 100644 index 0000000..c75c5ed --- /dev/null +++ b/drivers/gpu/drm/radeon/rv350d.h @@ -0,0 +1,52 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#ifndef __RV350D_H__ +#define __RV350D_H__ + +/* RV350, RV380 registers */ +/* #define R_00000D_SCLK_CNTL 0x00000D */ +#define S_00000D_FORCE_VAP(x) (((x) & 0x1) << 21) +#define G_00000D_FORCE_VAP(x) (((x) >> 21) & 0x1) +#define C_00000D_FORCE_VAP 0xFFDFFFFF +#define S_00000D_FORCE_SR(x) (((x) & 0x1) << 25) +#define G_00000D_FORCE_SR(x) (((x) >> 25) & 0x1) +#define C_00000D_FORCE_SR 0xFDFFFFFF +#define S_00000D_FORCE_PX(x) (((x) & 0x1) << 26) +#define G_00000D_FORCE_PX(x) (((x) >> 26) & 0x1) +#define C_00000D_FORCE_PX 0xFBFFFFFF +#define S_00000D_FORCE_TX(x) (((x) & 0x1) << 27) +#define G_00000D_FORCE_TX(x) (((x) >> 27) & 0x1) +#define C_00000D_FORCE_TX 0xF7FFFFFF +#define S_00000D_FORCE_US(x) (((x) & 0x1) << 28) +#define G_00000D_FORCE_US(x) (((x) >> 28) & 0x1) +#define C_00000D_FORCE_US 0xEFFFFFFF +#define S_00000D_FORCE_SU(x) (((x) & 0x1) << 30) +#define G_00000D_FORCE_SU(x) (((x) >> 30) & 0x1) +#define C_00000D_FORCE_SU 0xBFFFFFFF + +#endif diff --git a/drivers/gpu/drm/radeon/rv515.c b/drivers/gpu/drm/radeon/rv515.c index fd79974..ba68c9f 100644 --- a/drivers/gpu/drm/radeon/rv515.c +++ b/drivers/gpu/drm/radeon/rv515.c @@ -29,37 +29,17 @@ #include "drmP.h" #include "rv515d.h" #include "radeon.h" - +#include "atom.h" #include "rv515_reg_safe.h" -/* rv515 depends on : */ -void r100_hdp_reset(struct radeon_device *rdev); -int r100_cp_reset(struct radeon_device *rdev); -int r100_rb2d_reset(struct radeon_device *rdev); -int r100_gui_wait_for_idle(struct radeon_device *rdev); -int r100_cp_init(struct radeon_device *rdev, unsigned ring_size); -void r420_pipes_init(struct radeon_device *rdev); -void rs600_mc_disable_clients(struct radeon_device *rdev); -void rs600_disable_vga(struct radeon_device *rdev); - -/* This files gather functions specifics to: - * rv515 - * - * Some of these functions might be used by newer ASICs. - */ + +/* This files gather functions specifics to: rv515 */ int rv515_debugfs_pipes_info_init(struct radeon_device *rdev); int rv515_debugfs_ga_info_init(struct radeon_device *rdev); void rv515_gpu_init(struct radeon_device *rdev); int rv515_mc_wait_for_idle(struct radeon_device *rdev); - -/* - * MC - */ -int rv515_mc_init(struct radeon_device *rdev) +void rv515_debugfs(struct radeon_device *rdev) { - uint32_t tmp; - int r; - if (r100_debugfs_rbbm_init(rdev)) { DRM_ERROR("Failed to register debugfs file for RBBM !\n"); } @@ -69,67 +49,8 @@ int rv515_mc_init(struct radeon_device *rdev) if (rv515_debugfs_ga_info_init(rdev)) { DRM_ERROR("Failed to register debugfs file for pipes !\n"); } - - rv515_gpu_init(rdev); - rv370_pcie_gart_disable(rdev); - - /* Setup GPU memory space */ - rdev->mc.vram_location = 0xFFFFFFFFUL; - rdev->mc.gtt_location = 0xFFFFFFFFUL; - if (rdev->flags & RADEON_IS_AGP) { - r = radeon_agp_init(rdev); - if (r) { - printk(KERN_WARNING "[drm] Disabling AGP\n"); - rdev->flags &= ~RADEON_IS_AGP; - rdev->mc.gtt_size = radeon_gart_size * 1024 * 1024; - } else { - rdev->mc.gtt_location = rdev->mc.agp_base; - } - } - r = radeon_mc_setup(rdev); - if (r) { - return r; - } - - /* Program GPU memory space */ - rs600_mc_disable_clients(rdev); - if (rv515_mc_wait_for_idle(rdev)) { - printk(KERN_WARNING "Failed to wait MC idle while " - "programming pipes. Bad things might happen.\n"); - } - /* Write VRAM size in case we are limiting it */ - WREG32(RADEON_CONFIG_MEMSIZE, rdev->mc.real_vram_size); - tmp = REG_SET(MC_FB_START, rdev->mc.vram_location >> 16); - WREG32(0x134, tmp); - tmp = rdev->mc.vram_location + rdev->mc.mc_vram_size - 1; - tmp = REG_SET(MC_FB_TOP, tmp >> 16); - tmp |= REG_SET(MC_FB_START, rdev->mc.vram_location >> 16); - WREG32_MC(MC_FB_LOCATION, tmp); - WREG32(HDP_FB_LOCATION, rdev->mc.vram_location >> 16); - WREG32(0x310, rdev->mc.vram_location); - if (rdev->flags & RADEON_IS_AGP) { - tmp = rdev->mc.gtt_location + rdev->mc.gtt_size - 1; - tmp = REG_SET(MC_AGP_TOP, tmp >> 16); - tmp |= REG_SET(MC_AGP_START, rdev->mc.gtt_location >> 16); - WREG32_MC(MC_AGP_LOCATION, tmp); - WREG32_MC(MC_AGP_BASE, rdev->mc.agp_base); - WREG32_MC(MC_AGP_BASE_2, 0); - } else { - WREG32_MC(MC_AGP_LOCATION, 0x0FFFFFFF); - WREG32_MC(MC_AGP_BASE, 0); - WREG32_MC(MC_AGP_BASE_2, 0); - } - return 0; -} - -void rv515_mc_fini(struct radeon_device *rdev) -{ } - -/* - * Global GPU functions - */ void rv515_ring_start(struct radeon_device *rdev) { int r; @@ -198,11 +119,6 @@ void rv515_ring_start(struct radeon_device *rdev) radeon_ring_unlock_commit(rdev); } -void rv515_errata(struct radeon_device *rdev) -{ - rdev->pll_errata = 0; -} - int rv515_mc_wait_for_idle(struct radeon_device *rdev) { unsigned i; @@ -219,6 +135,12 @@ int rv515_mc_wait_for_idle(struct radeon_device *rdev) return -1; } +void rv515_vga_render_disable(struct radeon_device *rdev) +{ + WREG32(R_000300_VGA_RENDER_CONTROL, + RREG32(R_000300_VGA_RENDER_CONTROL) & C_000300_VGA_VSTATUS_CNTL); +} + void rv515_gpu_init(struct radeon_device *rdev) { unsigned pipe_select_current, gb_pipe_select, tmp; @@ -231,7 +153,7 @@ void rv515_gpu_init(struct radeon_device *rdev) "reseting GPU. Bad things might happen.\n"); } - rs600_disable_vga(rdev); + rv515_vga_render_disable(rdev); r420_pipes_init(rdev); gb_pipe_select = RREG32(0x402C); @@ -335,10 +257,6 @@ int rv515_gpu_reset(struct radeon_device *rdev) return 0; } - -/* - * VRAM info - */ static void rv515_vram_get_type(struct radeon_device *rdev) { uint32_t tmp; @@ -374,10 +292,6 @@ void rv515_vram_info(struct radeon_device *rdev) rdev->pm.sclk.full = rfixed_div(rdev->pm.sclk, a); } - -/* - * Indirect registers accessor - */ uint32_t rv515_mc_rreg(struct radeon_device *rdev, uint32_t reg) { uint32_t r; @@ -395,9 +309,6 @@ void rv515_mc_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v) WREG32(MC_IND_INDEX, 0); } -/* - * Debugfs info - */ #if defined(CONFIG_DEBUG_FS) static int rv515_debugfs_pipes_info(struct seq_file *m, void *data) { @@ -459,13 +370,260 @@ int rv515_debugfs_ga_info_init(struct radeon_device *rdev) #endif } -/* - * Asic initialization - */ -int rv515_init(struct radeon_device *rdev) +void rv515_mc_stop(struct radeon_device *rdev, struct rv515_mc_save *save) +{ + save->d1vga_control = RREG32(R_000330_D1VGA_CONTROL); + save->d2vga_control = RREG32(R_000338_D2VGA_CONTROL); + save->vga_render_control = RREG32(R_000300_VGA_RENDER_CONTROL); + save->vga_hdp_control = RREG32(R_000328_VGA_HDP_CONTROL); + save->d1crtc_control = RREG32(R_006080_D1CRTC_CONTROL); + save->d2crtc_control = RREG32(R_006880_D2CRTC_CONTROL); + + /* Stop all video */ + WREG32(R_0068E8_D2CRTC_UPDATE_LOCK, 0); + WREG32(R_000300_VGA_RENDER_CONTROL, 0); + WREG32(R_0060E8_D1CRTC_UPDATE_LOCK, 1); + WREG32(R_0068E8_D2CRTC_UPDATE_LOCK, 1); + WREG32(R_006080_D1CRTC_CONTROL, 0); + WREG32(R_006880_D2CRTC_CONTROL, 0); + WREG32(R_0060E8_D1CRTC_UPDATE_LOCK, 0); + WREG32(R_0068E8_D2CRTC_UPDATE_LOCK, 0); + WREG32(R_000330_D1VGA_CONTROL, 0); + WREG32(R_000338_D2VGA_CONTROL, 0); +} + +void rv515_mc_resume(struct radeon_device *rdev, struct rv515_mc_save *save) +{ + WREG32(R_006110_D1GRPH_PRIMARY_SURFACE_ADDRESS, rdev->mc.vram_start); + WREG32(R_006118_D1GRPH_SECONDARY_SURFACE_ADDRESS, rdev->mc.vram_start); + WREG32(R_006910_D2GRPH_PRIMARY_SURFACE_ADDRESS, rdev->mc.vram_start); + WREG32(R_006918_D2GRPH_SECONDARY_SURFACE_ADDRESS, rdev->mc.vram_start); + WREG32(R_000310_VGA_MEMORY_BASE_ADDRESS, rdev->mc.vram_start); + /* Unlock host access */ + WREG32(R_000328_VGA_HDP_CONTROL, save->vga_hdp_control); + mdelay(1); + /* Restore video state */ + WREG32(R_000330_D1VGA_CONTROL, save->d1vga_control); + WREG32(R_000338_D2VGA_CONTROL, save->d2vga_control); + WREG32(R_0060E8_D1CRTC_UPDATE_LOCK, 1); + WREG32(R_0068E8_D2CRTC_UPDATE_LOCK, 1); + WREG32(R_006080_D1CRTC_CONTROL, save->d1crtc_control); + WREG32(R_006880_D2CRTC_CONTROL, save->d2crtc_control); + WREG32(R_0060E8_D1CRTC_UPDATE_LOCK, 0); + WREG32(R_0068E8_D2CRTC_UPDATE_LOCK, 0); + WREG32(R_000300_VGA_RENDER_CONTROL, save->vga_render_control); +} + +void rv515_mc_program(struct radeon_device *rdev) +{ + struct rv515_mc_save save; + + /* Stops all mc clients */ + rv515_mc_stop(rdev, &save); + + /* Wait for mc idle */ + if (rv515_mc_wait_for_idle(rdev)) + dev_warn(rdev->dev, "Wait MC idle timeout before updating MC.\n"); + /* Write VRAM size in case we are limiting it */ + WREG32(R_0000F8_CONFIG_MEMSIZE, rdev->mc.real_vram_size); + /* Program MC, should be a 32bits limited address space */ + WREG32_MC(R_000001_MC_FB_LOCATION, + S_000001_MC_FB_START(rdev->mc.vram_start >> 16) | + S_000001_MC_FB_TOP(rdev->mc.vram_end >> 16)); + WREG32(R_000134_HDP_FB_LOCATION, + S_000134_HDP_FB_START(rdev->mc.vram_start >> 16)); + if (rdev->flags & RADEON_IS_AGP) { + WREG32_MC(R_000002_MC_AGP_LOCATION, + S_000002_MC_AGP_START(rdev->mc.gtt_start >> 16) | + S_000002_MC_AGP_TOP(rdev->mc.gtt_end >> 16)); + WREG32_MC(R_000003_MC_AGP_BASE, lower_32_bits(rdev->mc.agp_base)); + WREG32_MC(R_000004_MC_AGP_BASE_2, + S_000004_AGP_BASE_ADDR_2(upper_32_bits(rdev->mc.agp_base))); + } else { + WREG32_MC(R_000002_MC_AGP_LOCATION, 0xFFFFFFFF); + WREG32_MC(R_000003_MC_AGP_BASE, 0); + WREG32_MC(R_000004_MC_AGP_BASE_2, 0); + } + + rv515_mc_resume(rdev, &save); +} + +void rv515_clock_startup(struct radeon_device *rdev) +{ + if (radeon_dynclks != -1 && radeon_dynclks) + radeon_atom_set_clock_gating(rdev, 1); + /* We need to force on some of the block */ + WREG32_PLL(R_00000F_CP_DYN_CNTL, + RREG32_PLL(R_00000F_CP_DYN_CNTL) | S_00000F_CP_FORCEON(1)); + WREG32_PLL(R_000011_E2_DYN_CNTL, + RREG32_PLL(R_000011_E2_DYN_CNTL) | S_000011_E2_FORCEON(1)); + WREG32_PLL(R_000013_IDCT_DYN_CNTL, + RREG32_PLL(R_000013_IDCT_DYN_CNTL) | S_000013_IDCT_FORCEON(1)); +} + +static int rv515_startup(struct radeon_device *rdev) +{ + int r; + + rv515_mc_program(rdev); + /* Resume clock */ + rv515_clock_startup(rdev); + /* Initialize GPU configuration (# pipes, ...) */ + rv515_gpu_init(rdev); + /* Initialize GART (initialize after TTM so we can allocate + * memory through TTM but finalize after TTM) */ + if (rdev->flags & RADEON_IS_PCIE) { + r = rv370_pcie_gart_enable(rdev); + if (r) + return r; + } + /* Enable IRQ */ + rdev->irq.sw_int = true; + rs600_irq_set(rdev); + /* 1M ring buffer */ + r = r100_cp_init(rdev, 1024 * 1024); + if (r) { + dev_err(rdev->dev, "failled initializing CP (%d).\n", r); + return r; + } + r = r100_wb_init(rdev); + if (r) + dev_err(rdev->dev, "failled initializing WB (%d).\n", r); + r = r100_ib_init(rdev); + if (r) { + dev_err(rdev->dev, "failled initializing IB (%d).\n", r); + return r; + } + return 0; +} + +int rv515_resume(struct radeon_device *rdev) +{ + /* Make sur GART are not working */ + if (rdev->flags & RADEON_IS_PCIE) + rv370_pcie_gart_disable(rdev); + /* Resume clock before doing reset */ + rv515_clock_startup(rdev); + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_gpu_reset(rdev)) { + dev_warn(rdev->dev, "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* post */ + atom_asic_init(rdev->mode_info.atom_context); + /* Resume clock after posting */ + rv515_clock_startup(rdev); + return rv515_startup(rdev); +} + +int rv515_suspend(struct radeon_device *rdev) +{ + r100_cp_disable(rdev); + r100_wb_disable(rdev); + rs600_irq_disable(rdev); + if (rdev->flags & RADEON_IS_PCIE) + rv370_pcie_gart_disable(rdev); + return 0; +} + +void rv515_set_safe_registers(struct radeon_device *rdev) { rdev->config.r300.reg_safe_bm = rv515_reg_safe_bm; rdev->config.r300.reg_safe_bm_size = ARRAY_SIZE(rv515_reg_safe_bm); +} + +void rv515_fini(struct radeon_device *rdev) +{ + rv515_suspend(rdev); + r100_cp_fini(rdev); + r100_wb_fini(rdev); + r100_ib_fini(rdev); + radeon_gem_fini(rdev); + rv370_pcie_gart_fini(rdev); + radeon_agp_fini(rdev); + radeon_irq_kms_fini(rdev); + radeon_fence_driver_fini(rdev); + radeon_object_fini(rdev); + radeon_atombios_fini(rdev); + kfree(rdev->bios); + rdev->bios = NULL; +} + +int rv515_init(struct radeon_device *rdev) +{ + int r; + + /* Initialize scratch registers */ + radeon_scratch_init(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + /* TODO: disable VGA need to use VGA request */ + /* BIOS*/ + if (!radeon_get_bios(rdev)) { + if (ASIC_IS_AVIVO(rdev)) + return -EINVAL; + } + if (rdev->is_atom_bios) { + r = radeon_atombios_init(rdev); + if (r) + return r; + } else { + dev_err(rdev->dev, "Expecting atombios for RV515 GPU\n"); + return -EINVAL; + } + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_gpu_reset(rdev)) { + dev_warn(rdev->dev, + "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* check if cards are posted or not */ + if (!radeon_card_posted(rdev) && rdev->bios) { + DRM_INFO("GPU not posted. posting now...\n"); + atom_asic_init(rdev->mode_info.atom_context); + } + /* Initialize clocks */ + radeon_get_clock_info(rdev->ddev); + /* Initialize power management */ + radeon_pm_init(rdev); + /* Get vram informations */ + rv515_vram_info(rdev); + /* Initialize memory controller (also test AGP) */ + r = r420_mc_init(rdev); + if (r) + return r; + rv515_debugfs(rdev); + /* Fence driver */ + r = radeon_fence_driver_init(rdev); + if (r) + return r; + r = radeon_irq_kms_init(rdev); + if (r) + return r; + /* Memory manager */ + r = radeon_object_init(rdev); + if (r) + return r; + r = rv370_pcie_gart_init(rdev); + if (r) + return r; + rv515_set_safe_registers(rdev); + rdev->accel_working = true; + r = rv515_startup(rdev); + if (r) { + /* Somethings want wront with the accel init stop accel */ + dev_err(rdev->dev, "Disabling GPU acceleration\n"); + rv515_suspend(rdev); + r100_cp_fini(rdev); + r100_wb_fini(rdev); + r100_ib_fini(rdev); + rv370_pcie_gart_fini(rdev); + radeon_agp_fini(rdev); + radeon_irq_kms_fini(rdev); + rdev->accel_working = false; + } return 0; } diff --git a/drivers/gpu/drm/radeon/rv515d.h b/drivers/gpu/drm/radeon/rv515d.h index a65e17e..fc216e4 100644 --- a/drivers/gpu/drm/radeon/rv515d.h +++ b/drivers/gpu/drm/radeon/rv515d.h @@ -216,5 +216,388 @@ #define CP_PACKET0_GET_ONE_REG_WR(h) (((h) >> 15) & 1) #define CP_PACKET3_GET_OPCODE(h) (((h) >> 8) & 0xFF) -#endif +/* Registers */ +#define R_0000F8_CONFIG_MEMSIZE 0x0000F8 +#define S_0000F8_CONFIG_MEMSIZE(x) (((x) & 0xFFFFFFFF) << 0) +#define G_0000F8_CONFIG_MEMSIZE(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_0000F8_CONFIG_MEMSIZE 0x00000000 +#define R_000134_HDP_FB_LOCATION 0x000134 +#define S_000134_HDP_FB_START(x) (((x) & 0xFFFF) << 0) +#define G_000134_HDP_FB_START(x) (((x) >> 0) & 0xFFFF) +#define C_000134_HDP_FB_START 0xFFFF0000 +#define R_000300_VGA_RENDER_CONTROL 0x000300 +#define S_000300_VGA_BLINK_RATE(x) (((x) & 0x1F) << 0) +#define G_000300_VGA_BLINK_RATE(x) (((x) >> 0) & 0x1F) +#define C_000300_VGA_BLINK_RATE 0xFFFFFFE0 +#define S_000300_VGA_BLINK_MODE(x) (((x) & 0x3) << 5) +#define G_000300_VGA_BLINK_MODE(x) (((x) >> 5) & 0x3) +#define C_000300_VGA_BLINK_MODE 0xFFFFFF9F +#define S_000300_VGA_CURSOR_BLINK_INVERT(x) (((x) & 0x1) << 7) +#define G_000300_VGA_CURSOR_BLINK_INVERT(x) (((x) >> 7) & 0x1) +#define C_000300_VGA_CURSOR_BLINK_INVERT 0xFFFFFF7F +#define S_000300_VGA_EXTD_ADDR_COUNT_ENABLE(x) (((x) & 0x1) << 8) +#define G_000300_VGA_EXTD_ADDR_COUNT_ENABLE(x) (((x) >> 8) & 0x1) +#define C_000300_VGA_EXTD_ADDR_COUNT_ENABLE 0xFFFFFEFF +#define S_000300_VGA_VSTATUS_CNTL(x) (((x) & 0x3) << 16) +#define G_000300_VGA_VSTATUS_CNTL(x) (((x) >> 16) & 0x3) +#define C_000300_VGA_VSTATUS_CNTL 0xFFFCFFFF +#define S_000300_VGA_LOCK_8DOT(x) (((x) & 0x1) << 24) +#define G_000300_VGA_LOCK_8DOT(x) (((x) >> 24) & 0x1) +#define C_000300_VGA_LOCK_8DOT 0xFEFFFFFF +#define S_000300_VGAREG_LINECMP_COMPATIBILITY_SEL(x) (((x) & 0x1) << 25) +#define G_000300_VGAREG_LINECMP_COMPATIBILITY_SEL(x) (((x) >> 25) & 0x1) +#define C_000300_VGAREG_LINECMP_COMPATIBILITY_SEL 0xFDFFFFFF +#define R_000310_VGA_MEMORY_BASE_ADDRESS 0x000310 +#define S_000310_VGA_MEMORY_BASE_ADDRESS(x) (((x) & 0xFFFFFFFF) << 0) +#define G_000310_VGA_MEMORY_BASE_ADDRESS(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_000310_VGA_MEMORY_BASE_ADDRESS 0x00000000 +#define R_000328_VGA_HDP_CONTROL 0x000328 +#define S_000328_VGA_MEM_PAGE_SELECT_EN(x) (((x) & 0x1) << 0) +#define G_000328_VGA_MEM_PAGE_SELECT_EN(x) (((x) >> 0) & 0x1) +#define C_000328_VGA_MEM_PAGE_SELECT_EN 0xFFFFFFFE +#define S_000328_VGA_RBBM_LOCK_DISABLE(x) (((x) & 0x1) << 8) +#define G_000328_VGA_RBBM_LOCK_DISABLE(x) (((x) >> 8) & 0x1) +#define C_000328_VGA_RBBM_LOCK_DISABLE 0xFFFFFEFF +#define S_000328_VGA_SOFT_RESET(x) (((x) & 0x1) << 16) +#define G_000328_VGA_SOFT_RESET(x) (((x) >> 16) & 0x1) +#define C_000328_VGA_SOFT_RESET 0xFFFEFFFF +#define S_000328_VGA_TEST_RESET_CONTROL(x) (((x) & 0x1) << 24) +#define G_000328_VGA_TEST_RESET_CONTROL(x) (((x) >> 24) & 0x1) +#define C_000328_VGA_TEST_RESET_CONTROL 0xFEFFFFFF +#define R_000330_D1VGA_CONTROL 0x000330 +#define S_000330_D1VGA_MODE_ENABLE(x) (((x) & 0x1) << 0) +#define G_000330_D1VGA_MODE_ENABLE(x) (((x) >> 0) & 0x1) +#define C_000330_D1VGA_MODE_ENABLE 0xFFFFFFFE +#define S_000330_D1VGA_TIMING_SELECT(x) (((x) & 0x1) << 8) +#define G_000330_D1VGA_TIMING_SELECT(x) (((x) >> 8) & 0x1) +#define C_000330_D1VGA_TIMING_SELECT 0xFFFFFEFF +#define S_000330_D1VGA_SYNC_POLARITY_SELECT(x) (((x) & 0x1) << 9) +#define G_000330_D1VGA_SYNC_POLARITY_SELECT(x) (((x) >> 9) & 0x1) +#define C_000330_D1VGA_SYNC_POLARITY_SELECT 0xFFFFFDFF +#define S_000330_D1VGA_OVERSCAN_TIMING_SELECT(x) (((x) & 0x1) << 10) +#define G_000330_D1VGA_OVERSCAN_TIMING_SELECT(x) (((x) >> 10) & 0x1) +#define C_000330_D1VGA_OVERSCAN_TIMING_SELECT 0xFFFFFBFF +#define S_000330_D1VGA_OVERSCAN_COLOR_EN(x) (((x) & 0x1) << 16) +#define G_000330_D1VGA_OVERSCAN_COLOR_EN(x) (((x) >> 16) & 0x1) +#define C_000330_D1VGA_OVERSCAN_COLOR_EN 0xFFFEFFFF +#define S_000330_D1VGA_ROTATE(x) (((x) & 0x3) << 24) +#define G_000330_D1VGA_ROTATE(x) (((x) >> 24) & 0x3) +#define C_000330_D1VGA_ROTATE 0xFCFFFFFF +#define R_000338_D2VGA_CONTROL 0x000338 +#define S_000338_D2VGA_MODE_ENABLE(x) (((x) & 0x1) << 0) +#define G_000338_D2VGA_MODE_ENABLE(x) (((x) >> 0) & 0x1) +#define C_000338_D2VGA_MODE_ENABLE 0xFFFFFFFE +#define S_000338_D2VGA_TIMING_SELECT(x) (((x) & 0x1) << 8) +#define G_000338_D2VGA_TIMING_SELECT(x) (((x) >> 8) & 0x1) +#define C_000338_D2VGA_TIMING_SELECT 0xFFFFFEFF +#define S_000338_D2VGA_SYNC_POLARITY_SELECT(x) (((x) & 0x1) << 9) +#define G_000338_D2VGA_SYNC_POLARITY_SELECT(x) (((x) >> 9) & 0x1) +#define C_000338_D2VGA_SYNC_POLARITY_SELECT 0xFFFFFDFF +#define S_000338_D2VGA_OVERSCAN_TIMING_SELECT(x) (((x) & 0x1) << 10) +#define G_000338_D2VGA_OVERSCAN_TIMING_SELECT(x) (((x) >> 10) & 0x1) +#define C_000338_D2VGA_OVERSCAN_TIMING_SELECT 0xFFFFFBFF +#define S_000338_D2VGA_OVERSCAN_COLOR_EN(x) (((x) & 0x1) << 16) +#define G_000338_D2VGA_OVERSCAN_COLOR_EN(x) (((x) >> 16) & 0x1) +#define C_000338_D2VGA_OVERSCAN_COLOR_EN 0xFFFEFFFF +#define S_000338_D2VGA_ROTATE(x) (((x) & 0x3) << 24) +#define G_000338_D2VGA_ROTATE(x) (((x) >> 24) & 0x3) +#define C_000338_D2VGA_ROTATE 0xFCFFFFFF +#define R_0007C0_CP_STAT 0x0007C0 +#define S_0007C0_MRU_BUSY(x) (((x) & 0x1) << 0) +#define G_0007C0_MRU_BUSY(x) (((x) >> 0) & 0x1) +#define C_0007C0_MRU_BUSY 0xFFFFFFFE +#define S_0007C0_MWU_BUSY(x) (((x) & 0x1) << 1) +#define G_0007C0_MWU_BUSY(x) (((x) >> 1) & 0x1) +#define C_0007C0_MWU_BUSY 0xFFFFFFFD +#define S_0007C0_RSIU_BUSY(x) (((x) & 0x1) << 2) +#define G_0007C0_RSIU_BUSY(x) (((x) >> 2) & 0x1) +#define C_0007C0_RSIU_BUSY 0xFFFFFFFB +#define S_0007C0_RCIU_BUSY(x) (((x) & 0x1) << 3) +#define G_0007C0_RCIU_BUSY(x) (((x) >> 3) & 0x1) +#define C_0007C0_RCIU_BUSY 0xFFFFFFF7 +#define S_0007C0_CSF_PRIMARY_BUSY(x) (((x) & 0x1) << 9) +#define G_0007C0_CSF_PRIMARY_BUSY(x) (((x) >> 9) & 0x1) +#define C_0007C0_CSF_PRIMARY_BUSY 0xFFFFFDFF +#define S_0007C0_CSF_INDIRECT_BUSY(x) (((x) & 0x1) << 10) +#define G_0007C0_CSF_INDIRECT_BUSY(x) (((x) >> 10) & 0x1) +#define C_0007C0_CSF_INDIRECT_BUSY 0xFFFFFBFF +#define S_0007C0_CSQ_PRIMARY_BUSY(x) (((x) & 0x1) << 11) +#define G_0007C0_CSQ_PRIMARY_BUSY(x) (((x) >> 11) & 0x1) +#define C_0007C0_CSQ_PRIMARY_BUSY 0xFFFFF7FF +#define S_0007C0_CSQ_INDIRECT_BUSY(x) (((x) & 0x1) << 12) +#define G_0007C0_CSQ_INDIRECT_BUSY(x) (((x) >> 12) & 0x1) +#define C_0007C0_CSQ_INDIRECT_BUSY 0xFFFFEFFF +#define S_0007C0_CSI_BUSY(x) (((x) & 0x1) << 13) +#define G_0007C0_CSI_BUSY(x) (((x) >> 13) & 0x1) +#define C_0007C0_CSI_BUSY 0xFFFFDFFF +#define S_0007C0_CSF_INDIRECT2_BUSY(x) (((x) & 0x1) << 14) +#define G_0007C0_CSF_INDIRECT2_BUSY(x) (((x) >> 14) & 0x1) +#define C_0007C0_CSF_INDIRECT2_BUSY 0xFFFFBFFF +#define S_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) & 0x1) << 15) +#define G_0007C0_CSQ_INDIRECT2_BUSY(x) (((x) >> 15) & 0x1) +#define C_0007C0_CSQ_INDIRECT2_BUSY 0xFFFF7FFF +#define S_0007C0_GUIDMA_BUSY(x) (((x) & 0x1) << 28) +#define G_0007C0_GUIDMA_BUSY(x) (((x) >> 28) & 0x1) +#define C_0007C0_GUIDMA_BUSY 0xEFFFFFFF +#define S_0007C0_VIDDMA_BUSY(x) (((x) & 0x1) << 29) +#define G_0007C0_VIDDMA_BUSY(x) (((x) >> 29) & 0x1) +#define C_0007C0_VIDDMA_BUSY 0xDFFFFFFF +#define S_0007C0_CMDSTRM_BUSY(x) (((x) & 0x1) << 30) +#define G_0007C0_CMDSTRM_BUSY(x) (((x) >> 30) & 0x1) +#define C_0007C0_CMDSTRM_BUSY 0xBFFFFFFF +#define S_0007C0_CP_BUSY(x) (((x) & 0x1) << 31) +#define G_0007C0_CP_BUSY(x) (((x) >> 31) & 0x1) +#define C_0007C0_CP_BUSY 0x7FFFFFFF +#define R_000E40_RBBM_STATUS 0x000E40 +#define S_000E40_CMDFIFO_AVAIL(x) (((x) & 0x7F) << 0) +#define G_000E40_CMDFIFO_AVAIL(x) (((x) >> 0) & 0x7F) +#define C_000E40_CMDFIFO_AVAIL 0xFFFFFF80 +#define S_000E40_HIRQ_ON_RBB(x) (((x) & 0x1) << 8) +#define G_000E40_HIRQ_ON_RBB(x) (((x) >> 8) & 0x1) +#define C_000E40_HIRQ_ON_RBB 0xFFFFFEFF +#define S_000E40_CPRQ_ON_RBB(x) (((x) & 0x1) << 9) +#define G_000E40_CPRQ_ON_RBB(x) (((x) >> 9) & 0x1) +#define C_000E40_CPRQ_ON_RBB 0xFFFFFDFF +#define S_000E40_CFRQ_ON_RBB(x) (((x) & 0x1) << 10) +#define G_000E40_CFRQ_ON_RBB(x) (((x) >> 10) & 0x1) +#define C_000E40_CFRQ_ON_RBB 0xFFFFFBFF +#define S_000E40_HIRQ_IN_RTBUF(x) (((x) & 0x1) << 11) +#define G_000E40_HIRQ_IN_RTBUF(x) (((x) >> 11) & 0x1) +#define C_000E40_HIRQ_IN_RTBUF 0xFFFFF7FF +#define S_000E40_CPRQ_IN_RTBUF(x) (((x) & 0x1) << 12) +#define G_000E40_CPRQ_IN_RTBUF(x) (((x) >> 12) & 0x1) +#define C_000E40_CPRQ_IN_RTBUF 0xFFFFEFFF +#define S_000E40_CFRQ_IN_RTBUF(x) (((x) & 0x1) << 13) +#define G_000E40_CFRQ_IN_RTBUF(x) (((x) >> 13) & 0x1) +#define C_000E40_CFRQ_IN_RTBUF 0xFFFFDFFF +#define S_000E40_CF_PIPE_BUSY(x) (((x) & 0x1) << 14) +#define G_000E40_CF_PIPE_BUSY(x) (((x) >> 14) & 0x1) +#define C_000E40_CF_PIPE_BUSY 0xFFFFBFFF +#define S_000E40_ENG_EV_BUSY(x) (((x) & 0x1) << 15) +#define G_000E40_ENG_EV_BUSY(x) (((x) >> 15) & 0x1) +#define C_000E40_ENG_EV_BUSY 0xFFFF7FFF +#define S_000E40_CP_CMDSTRM_BUSY(x) (((x) & 0x1) << 16) +#define G_000E40_CP_CMDSTRM_BUSY(x) (((x) >> 16) & 0x1) +#define C_000E40_CP_CMDSTRM_BUSY 0xFFFEFFFF +#define S_000E40_E2_BUSY(x) (((x) & 0x1) << 17) +#define G_000E40_E2_BUSY(x) (((x) >> 17) & 0x1) +#define C_000E40_E2_BUSY 0xFFFDFFFF +#define S_000E40_RB2D_BUSY(x) (((x) & 0x1) << 18) +#define G_000E40_RB2D_BUSY(x) (((x) >> 18) & 0x1) +#define C_000E40_RB2D_BUSY 0xFFFBFFFF +#define S_000E40_RB3D_BUSY(x) (((x) & 0x1) << 19) +#define G_000E40_RB3D_BUSY(x) (((x) >> 19) & 0x1) +#define C_000E40_RB3D_BUSY 0xFFF7FFFF +#define S_000E40_VAP_BUSY(x) (((x) & 0x1) << 20) +#define G_000E40_VAP_BUSY(x) (((x) >> 20) & 0x1) +#define C_000E40_VAP_BUSY 0xFFEFFFFF +#define S_000E40_RE_BUSY(x) (((x) & 0x1) << 21) +#define G_000E40_RE_BUSY(x) (((x) >> 21) & 0x1) +#define C_000E40_RE_BUSY 0xFFDFFFFF +#define S_000E40_TAM_BUSY(x) (((x) & 0x1) << 22) +#define G_000E40_TAM_BUSY(x) (((x) >> 22) & 0x1) +#define C_000E40_TAM_BUSY 0xFFBFFFFF +#define S_000E40_TDM_BUSY(x) (((x) & 0x1) << 23) +#define G_000E40_TDM_BUSY(x) (((x) >> 23) & 0x1) +#define C_000E40_TDM_BUSY 0xFF7FFFFF +#define S_000E40_PB_BUSY(x) (((x) & 0x1) << 24) +#define G_000E40_PB_BUSY(x) (((x) >> 24) & 0x1) +#define C_000E40_PB_BUSY 0xFEFFFFFF +#define S_000E40_TIM_BUSY(x) (((x) & 0x1) << 25) +#define G_000E40_TIM_BUSY(x) (((x) >> 25) & 0x1) +#define C_000E40_TIM_BUSY 0xFDFFFFFF +#define S_000E40_GA_BUSY(x) (((x) & 0x1) << 26) +#define G_000E40_GA_BUSY(x) (((x) >> 26) & 0x1) +#define C_000E40_GA_BUSY 0xFBFFFFFF +#define S_000E40_CBA2D_BUSY(x) (((x) & 0x1) << 27) +#define G_000E40_CBA2D_BUSY(x) (((x) >> 27) & 0x1) +#define C_000E40_CBA2D_BUSY 0xF7FFFFFF +#define S_000E40_RBBM_HIBUSY(x) (((x) & 0x1) << 28) +#define G_000E40_RBBM_HIBUSY(x) (((x) >> 28) & 0x1) +#define C_000E40_RBBM_HIBUSY 0xEFFFFFFF +#define S_000E40_SKID_CFBUSY(x) (((x) & 0x1) << 29) +#define G_000E40_SKID_CFBUSY(x) (((x) >> 29) & 0x1) +#define C_000E40_SKID_CFBUSY 0xDFFFFFFF +#define S_000E40_VAP_VF_BUSY(x) (((x) & 0x1) << 30) +#define G_000E40_VAP_VF_BUSY(x) (((x) >> 30) & 0x1) +#define C_000E40_VAP_VF_BUSY 0xBFFFFFFF +#define S_000E40_GUI_ACTIVE(x) (((x) & 0x1) << 31) +#define G_000E40_GUI_ACTIVE(x) (((x) >> 31) & 0x1) +#define C_000E40_GUI_ACTIVE 0x7FFFFFFF +#define R_006080_D1CRTC_CONTROL 0x006080 +#define S_006080_D1CRTC_MASTER_EN(x) (((x) & 0x1) << 0) +#define G_006080_D1CRTC_MASTER_EN(x) (((x) >> 0) & 0x1) +#define C_006080_D1CRTC_MASTER_EN 0xFFFFFFFE +#define S_006080_D1CRTC_SYNC_RESET_SEL(x) (((x) & 0x1) << 4) +#define G_006080_D1CRTC_SYNC_RESET_SEL(x) (((x) >> 4) & 0x1) +#define C_006080_D1CRTC_SYNC_RESET_SEL 0xFFFFFFEF +#define S_006080_D1CRTC_DISABLE_POINT_CNTL(x) (((x) & 0x3) << 8) +#define G_006080_D1CRTC_DISABLE_POINT_CNTL(x) (((x) >> 8) & 0x3) +#define C_006080_D1CRTC_DISABLE_POINT_CNTL 0xFFFFFCFF +#define S_006080_D1CRTC_CURRENT_MASTER_EN_STATE(x) (((x) & 0x1) << 16) +#define G_006080_D1CRTC_CURRENT_MASTER_EN_STATE(x) (((x) >> 16) & 0x1) +#define C_006080_D1CRTC_CURRENT_MASTER_EN_STATE 0xFFFEFFFF +#define S_006080_D1CRTC_DISP_READ_REQUEST_DISABLE(x) (((x) & 0x1) << 24) +#define G_006080_D1CRTC_DISP_READ_REQUEST_DISABLE(x) (((x) >> 24) & 0x1) +#define C_006080_D1CRTC_DISP_READ_REQUEST_DISABLE 0xFEFFFFFF +#define R_0060E8_D1CRTC_UPDATE_LOCK 0x0060E8 +#define S_0060E8_D1CRTC_UPDATE_LOCK(x) (((x) & 0x1) << 0) +#define G_0060E8_D1CRTC_UPDATE_LOCK(x) (((x) >> 0) & 0x1) +#define C_0060E8_D1CRTC_UPDATE_LOCK 0xFFFFFFFE +#define R_006110_D1GRPH_PRIMARY_SURFACE_ADDRESS 0x006110 +#define S_006110_D1GRPH_PRIMARY_SURFACE_ADDRESS(x) (((x) & 0xFFFFFFFF) << 0) +#define G_006110_D1GRPH_PRIMARY_SURFACE_ADDRESS(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_006110_D1GRPH_PRIMARY_SURFACE_ADDRESS 0x00000000 +#define R_006118_D1GRPH_SECONDARY_SURFACE_ADDRESS 0x006118 +#define S_006118_D1GRPH_SECONDARY_SURFACE_ADDRESS(x) (((x) & 0xFFFFFFFF) << 0) +#define G_006118_D1GRPH_SECONDARY_SURFACE_ADDRESS(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_006118_D1GRPH_SECONDARY_SURFACE_ADDRESS 0x00000000 +#define R_006880_D2CRTC_CONTROL 0x006880 +#define S_006880_D2CRTC_MASTER_EN(x) (((x) & 0x1) << 0) +#define G_006880_D2CRTC_MASTER_EN(x) (((x) >> 0) & 0x1) +#define C_006880_D2CRTC_MASTER_EN 0xFFFFFFFE +#define S_006880_D2CRTC_SYNC_RESET_SEL(x) (((x) & 0x1) << 4) +#define G_006880_D2CRTC_SYNC_RESET_SEL(x) (((x) >> 4) & 0x1) +#define C_006880_D2CRTC_SYNC_RESET_SEL 0xFFFFFFEF +#define S_006880_D2CRTC_DISABLE_POINT_CNTL(x) (((x) & 0x3) << 8) +#define G_006880_D2CRTC_DISABLE_POINT_CNTL(x) (((x) >> 8) & 0x3) +#define C_006880_D2CRTC_DISABLE_POINT_CNTL 0xFFFFFCFF +#define S_006880_D2CRTC_CURRENT_MASTER_EN_STATE(x) (((x) & 0x1) << 16) +#define G_006880_D2CRTC_CURRENT_MASTER_EN_STATE(x) (((x) >> 16) & 0x1) +#define C_006880_D2CRTC_CURRENT_MASTER_EN_STATE 0xFFFEFFFF +#define S_006880_D2CRTC_DISP_READ_REQUEST_DISABLE(x) (((x) & 0x1) << 24) +#define G_006880_D2CRTC_DISP_READ_REQUEST_DISABLE(x) (((x) >> 24) & 0x1) +#define C_006880_D2CRTC_DISP_READ_REQUEST_DISABLE 0xFEFFFFFF +#define R_0068E8_D2CRTC_UPDATE_LOCK 0x0068E8 +#define S_0068E8_D2CRTC_UPDATE_LOCK(x) (((x) & 0x1) << 0) +#define G_0068E8_D2CRTC_UPDATE_LOCK(x) (((x) >> 0) & 0x1) +#define C_0068E8_D2CRTC_UPDATE_LOCK 0xFFFFFFFE +#define R_006910_D2GRPH_PRIMARY_SURFACE_ADDRESS 0x006910 +#define S_006910_D2GRPH_PRIMARY_SURFACE_ADDRESS(x) (((x) & 0xFFFFFFFF) << 0) +#define G_006910_D2GRPH_PRIMARY_SURFACE_ADDRESS(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_006910_D2GRPH_PRIMARY_SURFACE_ADDRESS 0x00000000 +#define R_006918_D2GRPH_SECONDARY_SURFACE_ADDRESS 0x006918 +#define S_006918_D2GRPH_SECONDARY_SURFACE_ADDRESS(x) (((x) & 0xFFFFFFFF) << 0) +#define G_006918_D2GRPH_SECONDARY_SURFACE_ADDRESS(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_006918_D2GRPH_SECONDARY_SURFACE_ADDRESS 0x00000000 + + +#define R_000001_MC_FB_LOCATION 0x000001 +#define S_000001_MC_FB_START(x) (((x) & 0xFFFF) << 0) +#define G_000001_MC_FB_START(x) (((x) >> 0) & 0xFFFF) +#define C_000001_MC_FB_START 0xFFFF0000 +#define S_000001_MC_FB_TOP(x) (((x) & 0xFFFF) << 16) +#define G_000001_MC_FB_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_000001_MC_FB_TOP 0x0000FFFF +#define R_000002_MC_AGP_LOCATION 0x000002 +#define S_000002_MC_AGP_START(x) (((x) & 0xFFFF) << 0) +#define G_000002_MC_AGP_START(x) (((x) >> 0) & 0xFFFF) +#define C_000002_MC_AGP_START 0xFFFF0000 +#define S_000002_MC_AGP_TOP(x) (((x) & 0xFFFF) << 16) +#define G_000002_MC_AGP_TOP(x) (((x) >> 16) & 0xFFFF) +#define C_000002_MC_AGP_TOP 0x0000FFFF +#define R_000003_MC_AGP_BASE 0x000003 +#define S_000003_AGP_BASE_ADDR(x) (((x) & 0xFFFFFFFF) << 0) +#define G_000003_AGP_BASE_ADDR(x) (((x) >> 0) & 0xFFFFFFFF) +#define C_000003_AGP_BASE_ADDR 0x00000000 +#define R_000004_MC_AGP_BASE_2 0x000004 +#define S_000004_AGP_BASE_ADDR_2(x) (((x) & 0xF) << 0) +#define G_000004_AGP_BASE_ADDR_2(x) (((x) >> 0) & 0xF) +#define C_000004_AGP_BASE_ADDR_2 0xFFFFFFF0 + +#define R_00000F_CP_DYN_CNTL 0x00000F +#define S_00000F_CP_FORCEON(x) (((x) & 0x1) << 0) +#define G_00000F_CP_FORCEON(x) (((x) >> 0) & 0x1) +#define C_00000F_CP_FORCEON 0xFFFFFFFE +#define S_00000F_CP_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 1) +#define G_00000F_CP_MAX_DYN_STOP_LAT(x) (((x) >> 1) & 0x1) +#define C_00000F_CP_MAX_DYN_STOP_LAT 0xFFFFFFFD +#define S_00000F_CP_CLOCK_STATUS(x) (((x) & 0x1) << 2) +#define G_00000F_CP_CLOCK_STATUS(x) (((x) >> 2) & 0x1) +#define C_00000F_CP_CLOCK_STATUS 0xFFFFFFFB +#define S_00000F_CP_PROG_SHUTOFF(x) (((x) & 0x1) << 3) +#define G_00000F_CP_PROG_SHUTOFF(x) (((x) >> 3) & 0x1) +#define C_00000F_CP_PROG_SHUTOFF 0xFFFFFFF7 +#define S_00000F_CP_PROG_DELAY_VALUE(x) (((x) & 0xFF) << 4) +#define G_00000F_CP_PROG_DELAY_VALUE(x) (((x) >> 4) & 0xFF) +#define C_00000F_CP_PROG_DELAY_VALUE 0xFFFFF00F +#define S_00000F_CP_LOWER_POWER_IDLE(x) (((x) & 0xFF) << 12) +#define G_00000F_CP_LOWER_POWER_IDLE(x) (((x) >> 12) & 0xFF) +#define C_00000F_CP_LOWER_POWER_IDLE 0xFFF00FFF +#define S_00000F_CP_LOWER_POWER_IGNORE(x) (((x) & 0x1) << 20) +#define G_00000F_CP_LOWER_POWER_IGNORE(x) (((x) >> 20) & 0x1) +#define C_00000F_CP_LOWER_POWER_IGNORE 0xFFEFFFFF +#define S_00000F_CP_NORMAL_POWER_IGNORE(x) (((x) & 0x1) << 21) +#define G_00000F_CP_NORMAL_POWER_IGNORE(x) (((x) >> 21) & 0x1) +#define C_00000F_CP_NORMAL_POWER_IGNORE 0xFFDFFFFF +#define S_00000F_SPARE(x) (((x) & 0x3) << 22) +#define G_00000F_SPARE(x) (((x) >> 22) & 0x3) +#define C_00000F_SPARE 0xFF3FFFFF +#define S_00000F_CP_NORMAL_POWER_BUSY(x) (((x) & 0xFF) << 24) +#define G_00000F_CP_NORMAL_POWER_BUSY(x) (((x) >> 24) & 0xFF) +#define C_00000F_CP_NORMAL_POWER_BUSY 0x00FFFFFF +#define R_000011_E2_DYN_CNTL 0x000011 +#define S_000011_E2_FORCEON(x) (((x) & 0x1) << 0) +#define G_000011_E2_FORCEON(x) (((x) >> 0) & 0x1) +#define C_000011_E2_FORCEON 0xFFFFFFFE +#define S_000011_E2_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 1) +#define G_000011_E2_MAX_DYN_STOP_LAT(x) (((x) >> 1) & 0x1) +#define C_000011_E2_MAX_DYN_STOP_LAT 0xFFFFFFFD +#define S_000011_E2_CLOCK_STATUS(x) (((x) & 0x1) << 2) +#define G_000011_E2_CLOCK_STATUS(x) (((x) >> 2) & 0x1) +#define C_000011_E2_CLOCK_STATUS 0xFFFFFFFB +#define S_000011_E2_PROG_SHUTOFF(x) (((x) & 0x1) << 3) +#define G_000011_E2_PROG_SHUTOFF(x) (((x) >> 3) & 0x1) +#define C_000011_E2_PROG_SHUTOFF 0xFFFFFFF7 +#define S_000011_E2_PROG_DELAY_VALUE(x) (((x) & 0xFF) << 4) +#define G_000011_E2_PROG_DELAY_VALUE(x) (((x) >> 4) & 0xFF) +#define C_000011_E2_PROG_DELAY_VALUE 0xFFFFF00F +#define S_000011_E2_LOWER_POWER_IDLE(x) (((x) & 0xFF) << 12) +#define G_000011_E2_LOWER_POWER_IDLE(x) (((x) >> 12) & 0xFF) +#define C_000011_E2_LOWER_POWER_IDLE 0xFFF00FFF +#define S_000011_E2_LOWER_POWER_IGNORE(x) (((x) & 0x1) << 20) +#define G_000011_E2_LOWER_POWER_IGNORE(x) (((x) >> 20) & 0x1) +#define C_000011_E2_LOWER_POWER_IGNORE 0xFFEFFFFF +#define S_000011_E2_NORMAL_POWER_IGNORE(x) (((x) & 0x1) << 21) +#define G_000011_E2_NORMAL_POWER_IGNORE(x) (((x) >> 21) & 0x1) +#define C_000011_E2_NORMAL_POWER_IGNORE 0xFFDFFFFF +#define S_000011_SPARE(x) (((x) & 0x3) << 22) +#define G_000011_SPARE(x) (((x) >> 22) & 0x3) +#define C_000011_SPARE 0xFF3FFFFF +#define S_000011_E2_NORMAL_POWER_BUSY(x) (((x) & 0xFF) << 24) +#define G_000011_E2_NORMAL_POWER_BUSY(x) (((x) >> 24) & 0xFF) +#define C_000011_E2_NORMAL_POWER_BUSY 0x00FFFFFF +#define R_000013_IDCT_DYN_CNTL 0x000013 +#define S_000013_IDCT_FORCEON(x) (((x) & 0x1) << 0) +#define G_000013_IDCT_FORCEON(x) (((x) >> 0) & 0x1) +#define C_000013_IDCT_FORCEON 0xFFFFFFFE +#define S_000013_IDCT_MAX_DYN_STOP_LAT(x) (((x) & 0x1) << 1) +#define G_000013_IDCT_MAX_DYN_STOP_LAT(x) (((x) >> 1) & 0x1) +#define C_000013_IDCT_MAX_DYN_STOP_LAT 0xFFFFFFFD +#define S_000013_IDCT_CLOCK_STATUS(x) (((x) & 0x1) << 2) +#define G_000013_IDCT_CLOCK_STATUS(x) (((x) >> 2) & 0x1) +#define C_000013_IDCT_CLOCK_STATUS 0xFFFFFFFB +#define S_000013_IDCT_PROG_SHUTOFF(x) (((x) & 0x1) << 3) +#define G_000013_IDCT_PROG_SHUTOFF(x) (((x) >> 3) & 0x1) +#define C_000013_IDCT_PROG_SHUTOFF 0xFFFFFFF7 +#define S_000013_IDCT_PROG_DELAY_VALUE(x) (((x) & 0xFF) << 4) +#define G_000013_IDCT_PROG_DELAY_VALUE(x) (((x) >> 4) & 0xFF) +#define C_000013_IDCT_PROG_DELAY_VALUE 0xFFFFF00F +#define S_000013_IDCT_LOWER_POWER_IDLE(x) (((x) & 0xFF) << 12) +#define G_000013_IDCT_LOWER_POWER_IDLE(x) (((x) >> 12) & 0xFF) +#define C_000013_IDCT_LOWER_POWER_IDLE 0xFFF00FFF +#define S_000013_IDCT_LOWER_POWER_IGNORE(x) (((x) & 0x1) << 20) +#define G_000013_IDCT_LOWER_POWER_IGNORE(x) (((x) >> 20) & 0x1) +#define C_000013_IDCT_LOWER_POWER_IGNORE 0xFFEFFFFF +#define S_000013_IDCT_NORMAL_POWER_IGNORE(x) (((x) & 0x1) << 21) +#define G_000013_IDCT_NORMAL_POWER_IGNORE(x) (((x) >> 21) & 0x1) +#define C_000013_IDCT_NORMAL_POWER_IGNORE 0xFFDFFFFF +#define S_000013_SPARE(x) (((x) & 0x3) << 22) +#define G_000013_SPARE(x) (((x) >> 22) & 0x3) +#define C_000013_SPARE 0xFF3FFFFF +#define S_000013_IDCT_NORMAL_POWER_BUSY(x) (((x) & 0xFF) << 24) +#define G_000013_IDCT_NORMAL_POWER_BUSY(x) (((x) >> 24) & 0xFF) +#define C_000013_IDCT_NORMAL_POWER_BUSY 0x00FFFFFF + +#endif diff --git a/drivers/gpu/drm/radeon/rv770.c b/drivers/gpu/drm/radeon/rv770.c index b574c73..b0efd0d 100644 --- a/drivers/gpu/drm/radeon/rv770.c +++ b/drivers/gpu/drm/radeon/rv770.c @@ -31,8 +31,8 @@ #include "radeon.h" #include "radeon_drm.h" #include "rv770d.h" -#include "avivod.h" #include "atom.h" +#include "avivod.h" #define R700_PFP_UCODE_SIZE 848 #define R700_PM4_UCODE_SIZE 1360 @@ -75,7 +75,7 @@ int rv770_pcie_gart_enable(struct radeon_device *rdev) WREG32(MC_VM_MB_L1_TLB2_CNTL, tmp); WREG32(MC_VM_MB_L1_TLB3_CNTL, tmp); WREG32(VM_CONTEXT0_PAGE_TABLE_START_ADDR, rdev->mc.gtt_start >> 12); - WREG32(VM_CONTEXT0_PAGE_TABLE_END_ADDR, (rdev->mc.gtt_end - 1) >> 12); + WREG32(VM_CONTEXT0_PAGE_TABLE_END_ADDR, rdev->mc.gtt_end >> 12); WREG32(VM_CONTEXT0_PAGE_TABLE_BASE_ADDR, rdev->gart.table_addr >> 12); WREG32(VM_CONTEXT0_CNTL, ENABLE_CONTEXT | PAGE_TABLE_DEPTH(0) | RANGE_PROTECTION_FAULT_ENABLE_DEFAULT); @@ -126,17 +126,36 @@ void rv770_pcie_gart_fini(struct radeon_device *rdev) } -/* - * MC - */ -static void rv770_mc_resume(struct radeon_device *rdev) +void rv770_agp_enable(struct radeon_device *rdev) +{ + u32 tmp; + int i; + + /* Setup L2 cache */ + WREG32(VM_L2_CNTL, ENABLE_L2_CACHE | ENABLE_L2_FRAGMENT_PROCESSING | + ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE | + EFFECTIVE_L2_QUEUE_SIZE(7)); + WREG32(VM_L2_CNTL2, 0); + WREG32(VM_L2_CNTL3, BANK_SELECT(0) | CACHE_UPDATE_MODE(2)); + /* Setup TLB control */ + tmp = ENABLE_L1_TLB | ENABLE_L1_FRAGMENT_PROCESSING | + SYSTEM_ACCESS_MODE_NOT_IN_SYS | + SYSTEM_APERTURE_UNMAPPED_ACCESS_PASS_THRU | + EFFECTIVE_L1_TLB_SIZE(5) | EFFECTIVE_L1_QUEUE_SIZE(5); + WREG32(MC_VM_MD_L1_TLB0_CNTL, tmp); + WREG32(MC_VM_MD_L1_TLB1_CNTL, tmp); + WREG32(MC_VM_MD_L1_TLB2_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB0_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB1_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB2_CNTL, tmp); + WREG32(MC_VM_MB_L1_TLB3_CNTL, tmp); + for (i = 0; i < 7; i++) + WREG32(VM_CONTEXT0_CNTL + (i * 4), 0); +} + +static void rv770_mc_program(struct radeon_device *rdev) { - u32 d1vga_control, d2vga_control; - u32 vga_render_control, vga_hdp_control; - u32 d1crtc_control, d2crtc_control; - u32 new_d1grph_primary, new_d1grph_secondary; - u32 new_d2grph_primary, new_d2grph_secondary; - u64 old_vram_start; + struct rv515_mc_save save; u32 tmp; int i, j; @@ -150,53 +169,42 @@ static void rv770_mc_resume(struct radeon_device *rdev) } WREG32(HDP_REG_COHERENCY_FLUSH_CNTL, 0); - d1vga_control = RREG32(D1VGA_CONTROL); - d2vga_control = RREG32(D2VGA_CONTROL); - vga_render_control = RREG32(VGA_RENDER_CONTROL); - vga_hdp_control = RREG32(VGA_HDP_CONTROL); - d1crtc_control = RREG32(D1CRTC_CONTROL); - d2crtc_control = RREG32(D2CRTC_CONTROL); - old_vram_start = (u64)(RREG32(MC_VM_FB_LOCATION) & 0xFFFF) << 24; - new_d1grph_primary = RREG32(D1GRPH_PRIMARY_SURFACE_ADDRESS); - new_d1grph_secondary = RREG32(D1GRPH_SECONDARY_SURFACE_ADDRESS); - new_d1grph_primary += rdev->mc.vram_start - old_vram_start; - new_d1grph_secondary += rdev->mc.vram_start - old_vram_start; - new_d2grph_primary = RREG32(D2GRPH_PRIMARY_SURFACE_ADDRESS); - new_d2grph_secondary = RREG32(D2GRPH_SECONDARY_SURFACE_ADDRESS); - new_d2grph_primary += rdev->mc.vram_start - old_vram_start; - new_d2grph_secondary += rdev->mc.vram_start - old_vram_start; - - /* Stop all video */ - WREG32(D1VGA_CONTROL, 0); - WREG32(D2VGA_CONTROL, 0); - WREG32(VGA_RENDER_CONTROL, 0); - WREG32(D1CRTC_UPDATE_LOCK, 1); - WREG32(D2CRTC_UPDATE_LOCK, 1); - WREG32(D1CRTC_CONTROL, 0); - WREG32(D2CRTC_CONTROL, 0); - WREG32(D1CRTC_UPDATE_LOCK, 0); - WREG32(D2CRTC_UPDATE_LOCK, 0); - - mdelay(1); + rv515_mc_stop(rdev, &save); if (r600_mc_wait_for_idle(rdev)) { - printk(KERN_WARNING "[drm] MC not idle !\n"); + dev_warn(rdev->dev, "Wait for MC idle timedout !\n"); } - /* Lockout access through VGA aperture*/ WREG32(VGA_HDP_CONTROL, VGA_MEMORY_DISABLE); - /* Update configuration */ - WREG32(MC_VM_SYSTEM_APERTURE_LOW_ADDR, rdev->mc.vram_start >> 12); - WREG32(MC_VM_SYSTEM_APERTURE_HIGH_ADDR, (rdev->mc.vram_end - 1) >> 12); + if (rdev->flags & RADEON_IS_AGP) { + if (rdev->mc.vram_start < rdev->mc.gtt_start) { + /* VRAM before AGP */ + WREG32(MC_VM_SYSTEM_APERTURE_LOW_ADDR, + rdev->mc.vram_start >> 12); + WREG32(MC_VM_SYSTEM_APERTURE_HIGH_ADDR, + rdev->mc.gtt_end >> 12); + } else { + /* VRAM after AGP */ + WREG32(MC_VM_SYSTEM_APERTURE_LOW_ADDR, + rdev->mc.gtt_start >> 12); + WREG32(MC_VM_SYSTEM_APERTURE_HIGH_ADDR, + rdev->mc.vram_end >> 12); + } + } else { + WREG32(MC_VM_SYSTEM_APERTURE_LOW_ADDR, + rdev->mc.vram_start >> 12); + WREG32(MC_VM_SYSTEM_APERTURE_HIGH_ADDR, + rdev->mc.vram_end >> 12); + } WREG32(MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR, 0); - tmp = (((rdev->mc.vram_end - 1) >> 24) & 0xFFFF) << 16; + tmp = ((rdev->mc.vram_end >> 24) & 0xFFFF) << 16; tmp |= ((rdev->mc.vram_start >> 24) & 0xFFFF); WREG32(MC_VM_FB_LOCATION, tmp); WREG32(HDP_NONSURFACE_BASE, (rdev->mc.vram_start >> 8)); WREG32(HDP_NONSURFACE_INFO, (2 << 7)); WREG32(HDP_NONSURFACE_SIZE, (rdev->mc.mc_vram_size - 1) | 0x3FF); if (rdev->flags & RADEON_IS_AGP) { - WREG32(MC_VM_AGP_TOP, (rdev->mc.gtt_end - 1) >> 16); + WREG32(MC_VM_AGP_TOP, rdev->mc.gtt_end >> 16); WREG32(MC_VM_AGP_BOT, rdev->mc.gtt_start >> 16); WREG32(MC_VM_AGP_BASE, rdev->mc.agp_base >> 22); } else { @@ -204,34 +212,13 @@ static void rv770_mc_resume(struct radeon_device *rdev) WREG32(MC_VM_AGP_TOP, 0x0FFFFFFF); WREG32(MC_VM_AGP_BOT, 0x0FFFFFFF); } - WREG32(D1GRPH_PRIMARY_SURFACE_ADDRESS, new_d1grph_primary); - WREG32(D1GRPH_SECONDARY_SURFACE_ADDRESS, new_d1grph_secondary); - WREG32(D2GRPH_PRIMARY_SURFACE_ADDRESS, new_d2grph_primary); - WREG32(D2GRPH_SECONDARY_SURFACE_ADDRESS, new_d2grph_secondary); - WREG32(VGA_MEMORY_BASE_ADDRESS, rdev->mc.vram_start); - - /* Unlock host access */ - WREG32(VGA_HDP_CONTROL, vga_hdp_control); - - mdelay(1); if (r600_mc_wait_for_idle(rdev)) { - printk(KERN_WARNING "[drm] MC not idle !\n"); + dev_warn(rdev->dev, "Wait for MC idle timedout !\n"); } - - /* Restore video state */ - WREG32(D1CRTC_UPDATE_LOCK, 1); - WREG32(D2CRTC_UPDATE_LOCK, 1); - WREG32(D1CRTC_CONTROL, d1crtc_control); - WREG32(D2CRTC_CONTROL, d2crtc_control); - WREG32(D1CRTC_UPDATE_LOCK, 0); - WREG32(D2CRTC_UPDATE_LOCK, 0); - WREG32(D1VGA_CONTROL, d1vga_control); - WREG32(D2VGA_CONTROL, d2vga_control); - WREG32(VGA_RENDER_CONTROL, vga_render_control); - + rv515_mc_resume(rdev, &save); /* we need to own VRAM, so turn off the VGA renderer here * to stop it overwriting our objects */ - radeon_avivo_vga_render_disable(rdev); + rv515_vga_render_disable(rdev); } @@ -542,11 +529,11 @@ static void rv770_gpu_init(struct radeon_device *rdev) if (rdev->family == CHIP_RV770) gb_tiling_config |= BANK_TILING(1); else - gb_tiling_config |= BANK_TILING((mc_arb_ramcfg & NOOFBANK_SHIFT) >> NOOFBANK_MASK); + gb_tiling_config |= BANK_TILING((mc_arb_ramcfg & NOOFBANK_MASK) >> NOOFBANK_SHIFT); gb_tiling_config |= GROUP_SIZE(0); - if (((mc_arb_ramcfg & NOOFROWS_MASK) & NOOFROWS_SHIFT) > 3) { + if (((mc_arb_ramcfg & NOOFROWS_MASK) >> NOOFROWS_SHIFT) > 3) { gb_tiling_config |= ROW_TILING(3); gb_tiling_config |= SAMPLE_SPLIT(3); } else { @@ -592,14 +579,14 @@ static void rv770_gpu_init(struct radeon_device *rdev) /* set HW defaults for 3D engine */ WREG32(CP_QUEUE_THRESHOLDS, (ROQ_IB1_START(0x16) | - ROQ_IB2_START(0x2b))); + ROQ_IB2_START(0x2b))); WREG32(CP_MEQ_THRESHOLDS, STQ_SPLIT(0x30)); WREG32(TA_CNTL_AUX, (DISABLE_CUBE_ANISO | - SYNC_GRADIENT | - SYNC_WALKER | - SYNC_ALIGNER)); + SYNC_GRADIENT | + SYNC_WALKER | + SYNC_ALIGNER)); sx_debug_1 = RREG32(SX_DEBUG_1); sx_debug_1 |= ENABLE_NEW_SMX_ADDRESS; @@ -611,9 +598,9 @@ static void rv770_gpu_init(struct radeon_device *rdev) WREG32(SMX_DC_CTL0, smx_dc_ctl0); WREG32(SMX_EVENT_CTL, (ES_FLUSH_CTL(4) | - GS_FLUSH_CTL(4) | - ACK_FLUSH_CTL(3) | - SYNC_FLUSH_CTL)); + GS_FLUSH_CTL(4) | + ACK_FLUSH_CTL(3) | + SYNC_FLUSH_CTL)); if (rdev->family == CHIP_RV770) WREG32(DB_DEBUG3, DB_CLK_OFF_DELAY(0x1f)); @@ -624,12 +611,12 @@ static void rv770_gpu_init(struct radeon_device *rdev) } WREG32(SX_EXPORT_BUFFER_SIZES, (COLOR_BUFFER_SIZE((rdev->config.rv770.sx_max_export_size / 4) - 1) | - POSITION_BUFFER_SIZE((rdev->config.rv770.sx_max_export_pos_size / 4) - 1) | - SMX_BUFFER_SIZE((rdev->config.rv770.sx_max_export_smx_size / 4) - 1))); + POSITION_BUFFER_SIZE((rdev->config.rv770.sx_max_export_pos_size / 4) - 1) | + SMX_BUFFER_SIZE((rdev->config.rv770.sx_max_export_smx_size / 4) - 1))); WREG32(PA_SC_FIFO_SIZE, (SC_PRIM_FIFO_SIZE(rdev->config.rv770.sc_prim_fifo_size) | - SC_HIZ_TILE_FIFO_SIZE(rdev->config.rv770.sc_hiz_tile_fifo_size) | - SC_EARLYZ_TILE_FIFO_SIZE(rdev->config.rv770.sc_earlyz_tile_fifo_fize))); + SC_HIZ_TILE_FIFO_SIZE(rdev->config.rv770.sc_hiz_tile_fifo_size) | + SC_EARLYZ_TILE_FIFO_SIZE(rdev->config.rv770.sc_earlyz_tile_fifo_fize))); WREG32(PA_SC_MULTI_CHIP_CNTL, 0); @@ -787,20 +774,49 @@ int rv770_mc_init(struct radeon_device *rdev) { fixed20_12 a; u32 tmp; + int chansize, numchan; int r; /* Get VRAM informations */ - /* FIXME: Don't know how to determine vram width, need to check - * vram_width usage - */ - rdev->mc.vram_width = 128; rdev->mc.vram_is_ddr = true; + tmp = RREG32(MC_ARB_RAMCFG); + if (tmp & CHANSIZE_OVERRIDE) { + chansize = 16; + } else if (tmp & CHANSIZE_MASK) { + chansize = 64; + } else { + chansize = 32; + } + tmp = RREG32(MC_SHARED_CHMAP); + switch ((tmp & NOOFCHAN_MASK) >> NOOFCHAN_SHIFT) { + case 0: + default: + numchan = 1; + break; + case 1: + numchan = 2; + break; + case 2: + numchan = 4; + break; + case 3: + numchan = 8; + break; + } + rdev->mc.vram_width = numchan * chansize; /* Could aper size report 0 ? */ rdev->mc.aper_base = drm_get_resource_start(rdev->ddev, 0); rdev->mc.aper_size = drm_get_resource_len(rdev->ddev, 0); /* Setup GPU memory space */ rdev->mc.mc_vram_size = RREG32(CONFIG_MEMSIZE); rdev->mc.real_vram_size = RREG32(CONFIG_MEMSIZE); + + if (rdev->mc.mc_vram_size > rdev->mc.aper_size) + rdev->mc.mc_vram_size = rdev->mc.aper_size; + + if (rdev->mc.real_vram_size > rdev->mc.aper_size) + rdev->mc.real_vram_size = rdev->mc.aper_size; + if (rdev->flags & RADEON_IS_AGP) { r = radeon_agp_init(rdev); if (r) @@ -833,9 +849,9 @@ int rv770_mc_init(struct radeon_device *rdev) rdev->mc.gtt_size = radeon_gart_size * 1024 * 1024; } rdev->mc.vram_start = rdev->mc.vram_location; - rdev->mc.vram_end = rdev->mc.vram_location + rdev->mc.mc_vram_size; + rdev->mc.vram_end = rdev->mc.vram_location + rdev->mc.mc_vram_size - 1; rdev->mc.gtt_start = rdev->mc.gtt_location; - rdev->mc.gtt_end = rdev->mc.gtt_location + rdev->mc.gtt_size; + rdev->mc.gtt_end = rdev->mc.gtt_location + rdev->mc.gtt_size - 1; /* FIXME: we should enforce default clock in case GPU is not in * default setup */ @@ -854,11 +870,14 @@ static int rv770_startup(struct radeon_device *rdev) { int r; - radeon_gpu_reset(rdev); - rv770_mc_resume(rdev); - r = rv770_pcie_gart_enable(rdev); - if (r) - return r; + rv770_mc_program(rdev); + if (rdev->flags & RADEON_IS_AGP) { + rv770_agp_enable(rdev); + } else { + r = rv770_pcie_gart_enable(rdev); + if (r) + return r; + } rv770_gpu_init(rdev); r = radeon_object_pin(rdev->r600_blit.shader_obj, RADEON_GEM_DOMAIN_VRAM, @@ -877,9 +896,8 @@ static int rv770_startup(struct radeon_device *rdev) r = r600_cp_resume(rdev); if (r) return r; - r = r600_wb_init(rdev); - if (r) - return r; + /* write back buffer are not vital so don't worry about failure */ + r600_wb_enable(rdev); return 0; } @@ -887,15 +905,12 @@ int rv770_resume(struct radeon_device *rdev) { int r; - if (radeon_gpu_reset(rdev)) { - /* FIXME: what do we want to do here ? */ - } + /* Do not reset GPU before posting, on rv770 hw unlike on r500 hw, + * posting will perform necessary task to bring back GPU into good + * shape. + */ /* post card */ - if (rdev->is_atom_bios) { - atom_asic_init(rdev->mode_info.atom_context); - } else { - radeon_combios_asic_init(rdev->ddev); - } + atom_asic_init(rdev->mode_info.atom_context); /* Initialize clocks */ r = radeon_clocks_init(rdev); if (r) { @@ -908,7 +923,7 @@ int rv770_resume(struct radeon_device *rdev) return r; } - r = radeon_ib_test(rdev); + r = r600_ib_test(rdev); if (r) { DRM_ERROR("radeon: failled testing IB (%d).\n", r); return r; @@ -922,8 +937,8 @@ int rv770_suspend(struct radeon_device *rdev) /* FIXME: we should wait for ring to be empty */ r700_cp_stop(rdev); rdev->cp.ready = false; + r600_wb_disable(rdev); rv770_pcie_gart_disable(rdev); - /* unpin shaders bo */ radeon_object_unpin(rdev->r600_blit.shader_obj); return 0; @@ -939,7 +954,6 @@ int rv770_init(struct radeon_device *rdev) { int r; - rdev->new_init_path = true; r = radeon_dummy_page_init(rdev); if (r) return r; @@ -953,8 +967,10 @@ int rv770_init(struct radeon_device *rdev) return -EINVAL; } /* Must be an ATOMBIOS */ - if (!rdev->is_atom_bios) + if (!rdev->is_atom_bios) { + dev_err(rdev->dev, "Expecting atombios for R600 GPU\n"); return -EINVAL; + } r = radeon_atombios_init(rdev); if (r) return r; @@ -967,24 +983,20 @@ int rv770_init(struct radeon_device *rdev) r600_scratch_init(rdev); /* Initialize surface registers */ radeon_surface_init(rdev); + /* Initialize clocks */ radeon_get_clock_info(rdev->ddev); r = radeon_clocks_init(rdev); if (r) return r; + /* Initialize power management */ + radeon_pm_init(rdev); /* Fence driver */ r = radeon_fence_driver_init(rdev); if (r) return r; r = rv770_mc_init(rdev); - if (r) { - if (rdev->flags & RADEON_IS_AGP) { - /* Retry with disabling AGP */ - rv770_fini(rdev); - rdev->flags &= ~RADEON_IS_AGP; - return rv770_init(rdev); - } + if (r) return r; - } /* Memory manager */ r = radeon_object_init(rdev); if (r) @@ -1013,12 +1025,10 @@ int rv770_init(struct radeon_device *rdev) r = rv770_startup(rdev); if (r) { - if (rdev->flags & RADEON_IS_AGP) { - /* Retry with disabling AGP */ - rv770_fini(rdev); - rdev->flags &= ~RADEON_IS_AGP; - return rv770_init(rdev); - } + rv770_suspend(rdev); + r600_wb_fini(rdev); + radeon_ring_fini(rdev); + rv770_pcie_gart_fini(rdev); rdev->accel_working = false; } if (rdev->accel_working) { @@ -1027,7 +1037,7 @@ int rv770_init(struct radeon_device *rdev) DRM_ERROR("radeon: failled initializing IB pool (%d).\n", r); rdev->accel_working = false; } - r = radeon_ib_test(rdev); + r = r600_ib_test(rdev); if (r) { DRM_ERROR("radeon: failled testing IB (%d).\n", r); rdev->accel_working = false; @@ -1042,20 +1052,15 @@ void rv770_fini(struct radeon_device *rdev) r600_blit_fini(rdev); radeon_ring_fini(rdev); + r600_wb_fini(rdev); rv770_pcie_gart_fini(rdev); radeon_gem_fini(rdev); radeon_fence_driver_fini(rdev); radeon_clocks_fini(rdev); -#if __OS_HAS_AGP if (rdev->flags & RADEON_IS_AGP) radeon_agp_fini(rdev); -#endif radeon_object_fini(rdev); - if (rdev->is_atom_bios) { - radeon_atombios_fini(rdev); - } else { - radeon_combios_fini(rdev); - } + radeon_atombios_fini(rdev); kfree(rdev->bios); rdev->bios = NULL; radeon_dummy_page_fini(rdev); diff --git a/drivers/gpu/drm/radeon/rv770d.h b/drivers/gpu/drm/radeon/rv770d.h index 4b9c3d6..a1367ab 100644 --- a/drivers/gpu/drm/radeon/rv770d.h +++ b/drivers/gpu/drm/radeon/rv770d.h @@ -129,6 +129,10 @@ #define HDP_REG_COHERENCY_FLUSH_CNTL 0x54A0 #define HDP_TILING_CONFIG 0x2F3C +#define MC_SHARED_CHMAP 0x2004 +#define NOOFCHAN_SHIFT 12 +#define NOOFCHAN_MASK 0x00003000 + #define MC_ARB_RAMCFG 0x2760 #define NOOFBANK_SHIFT 0 #define NOOFBANK_MASK 0x00000003 @@ -142,6 +146,7 @@ #define CHANSIZE_MASK 0x00000100 #define BURSTLENGTH_SHIFT 9 #define BURSTLENGTH_MASK 0x00000200 +#define CHANSIZE_OVERRIDE (1 << 11) #define MC_VM_AGP_TOP 0x2028 #define MC_VM_AGP_BOT 0x202C #define MC_VM_AGP_BASE 0x2030 diff --git a/drivers/gpu/drm/ttm/ttm_bo_vm.c b/drivers/gpu/drm/ttm/ttm_bo_vm.c index 33de763..1c040d0 100644 --- a/drivers/gpu/drm/ttm/ttm_bo_vm.c +++ b/drivers/gpu/drm/ttm/ttm_bo_vm.c @@ -228,7 +228,7 @@ static void ttm_bo_vm_close(struct vm_area_struct *vma) vma->vm_private_data = NULL; } -static struct vm_operations_struct ttm_bo_vm_ops = { +static const struct vm_operations_struct ttm_bo_vm_ops = { .fault = ttm_bo_vm_fault, .open = ttm_bo_vm_open, .close = ttm_bo_vm_close diff --git a/drivers/gpu/drm/ttm/ttm_global.c b/drivers/gpu/drm/ttm/ttm_global.c index 541744d..b170071 100644 --- a/drivers/gpu/drm/ttm/ttm_global.c +++ b/drivers/gpu/drm/ttm/ttm_global.c @@ -82,8 +82,8 @@ int ttm_global_item_ref(struct ttm_global_reference *ref) if (unlikely(ret != 0)) goto out_err; - ++item->refcount; } + ++item->refcount; ref->object = item->object; object = item->object; mutex_unlock(&item->mutex); diff --git a/drivers/gpu/drm/ttm/ttm_tt.c b/drivers/gpu/drm/ttm/ttm_tt.c index a55ee1a..7bcb89f 100644 --- a/drivers/gpu/drm/ttm/ttm_tt.c +++ b/drivers/gpu/drm/ttm/ttm_tt.c @@ -279,6 +279,7 @@ int ttm_tt_set_placement_caching(struct ttm_tt *ttm, uint32_t placement) return ttm_tt_set_caching(ttm, state); } +EXPORT_SYMBOL(ttm_tt_set_placement_caching); static void ttm_tt_free_alloced_pages(struct ttm_tt *ttm) { |