summaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/radeon/r100.c
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/gpu/drm/radeon/r100.c')
-rw-r--r--drivers/gpu/drm/radeon/r100.c87
1 files changed, 61 insertions, 26 deletions
diff --git a/drivers/gpu/drm/radeon/r100.c b/drivers/gpu/drm/radeon/r100.c
index 5f15820..e372f9e 100644
--- a/drivers/gpu/drm/radeon/r100.c
+++ b/drivers/gpu/drm/radeon/r100.c
@@ -70,23 +70,6 @@ MODULE_FIRMWARE(FIRMWARE_R520);
void r100_pre_page_flip(struct radeon_device *rdev, int crtc)
{
- struct radeon_crtc *radeon_crtc = rdev->mode_info.crtcs[crtc];
- u32 tmp;
-
- /* make sure flip is at vb rather than hb */
- tmp = RREG32(RADEON_CRTC_OFFSET_CNTL + radeon_crtc->crtc_offset);
- tmp &= ~RADEON_CRTC_OFFSET_FLIP_CNTL;
- /* make sure pending bit is asserted */
- tmp |= RADEON_CRTC_GUI_TRIG_OFFSET_LEFT_EN;
- WREG32(RADEON_CRTC_OFFSET_CNTL + radeon_crtc->crtc_offset, tmp);
-
- /* set pageflip to happen as late as possible in the vblank interval.
- * same field for crtc1/2
- */
- tmp = RREG32(RADEON_CRTC_GEN_CNTL);
- tmp &= ~RADEON_CRTC_VSTAT_MODE_MASK;
- WREG32(RADEON_CRTC_GEN_CNTL, tmp);
-
/* enable the pflip int */
radeon_irq_kms_pflip_irq_get(rdev, crtc);
}
@@ -1041,7 +1024,7 @@ int r100_cp_init(struct radeon_device *rdev, unsigned ring_size)
return r;
}
rdev->cp.ready = true;
- rdev->mc.active_vram_size = rdev->mc.real_vram_size;
+ radeon_ttm_set_active_vram_size(rdev, rdev->mc.real_vram_size);
return 0;
}
@@ -1059,7 +1042,7 @@ void r100_cp_fini(struct radeon_device *rdev)
void r100_cp_disable(struct radeon_device *rdev)
{
/* Disable ring */
- rdev->mc.active_vram_size = rdev->mc.visible_vram_size;
+ radeon_ttm_set_active_vram_size(rdev, rdev->mc.visible_vram_size);
rdev->cp.ready = false;
WREG32(RADEON_CP_CSQ_MODE, 0);
WREG32(RADEON_CP_CSQ_CNTL, 0);
@@ -1427,6 +1410,7 @@ static int r100_packet0_check(struct radeon_cs_parser *p,
}
track->zb.robj = reloc->robj;
track->zb.offset = idx_value;
+ track->zb_dirty = true;
ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset);
break;
case RADEON_RB3D_COLOROFFSET:
@@ -1439,6 +1423,7 @@ static int r100_packet0_check(struct radeon_cs_parser *p,
}
track->cb[0].robj = reloc->robj;
track->cb[0].offset = idx_value;
+ track->cb_dirty = true;
ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset);
break;
case RADEON_PP_TXOFFSET_0:
@@ -1454,6 +1439,7 @@ static int r100_packet0_check(struct radeon_cs_parser *p,
}
ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset);
track->textures[i].robj = reloc->robj;
+ track->tex_dirty = true;
break;
case RADEON_PP_CUBIC_OFFSET_T0_0:
case RADEON_PP_CUBIC_OFFSET_T0_1:
@@ -1471,6 +1457,7 @@ static int r100_packet0_check(struct radeon_cs_parser *p,
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;
+ track->tex_dirty = true;
break;
case RADEON_PP_CUBIC_OFFSET_T1_0:
case RADEON_PP_CUBIC_OFFSET_T1_1:
@@ -1488,6 +1475,7 @@ static int r100_packet0_check(struct radeon_cs_parser *p,
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;
+ track->tex_dirty = true;
break;
case RADEON_PP_CUBIC_OFFSET_T2_0:
case RADEON_PP_CUBIC_OFFSET_T2_1:
@@ -1505,9 +1493,12 @@ static int r100_packet0_check(struct radeon_cs_parser *p,
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;
+ track->tex_dirty = true;
break;
case RADEON_RE_WIDTH_HEIGHT:
track->maxy = ((idx_value >> 16) & 0x7FF);
+ track->cb_dirty = true;
+ track->zb_dirty = true;
break;
case RADEON_RB3D_COLORPITCH:
r = r100_cs_packet_next_reloc(p, &reloc);
@@ -1528,9 +1519,11 @@ static int r100_packet0_check(struct radeon_cs_parser *p,
ib[idx] = tmp;
track->cb[0].pitch = idx_value & RADEON_COLORPITCH_MASK;
+ track->cb_dirty = true;
break;
case RADEON_RB3D_DEPTHPITCH:
track->zb.pitch = idx_value & RADEON_DEPTHPITCH_MASK;
+ track->zb_dirty = true;
break;
case RADEON_RB3D_CNTL:
switch ((idx_value >> RADEON_RB3D_COLOR_FORMAT_SHIFT) & 0x1f) {
@@ -1555,6 +1548,8 @@ static int r100_packet0_check(struct radeon_cs_parser *p,
return -EINVAL;
}
track->z_enabled = !!(idx_value & RADEON_Z_ENABLE);
+ track->cb_dirty = true;
+ track->zb_dirty = true;
break;
case RADEON_RB3D_ZSTENCILCNTL:
switch (idx_value & 0xf) {
@@ -1572,6 +1567,7 @@ static int r100_packet0_check(struct radeon_cs_parser *p,
default:
break;
}
+ track->zb_dirty = true;
break;
case RADEON_RB3D_ZPASS_ADDR:
r = r100_cs_packet_next_reloc(p, &reloc);
@@ -1588,6 +1584,7 @@ static int r100_packet0_check(struct radeon_cs_parser *p,
uint32_t temp = idx_value >> 4;
for (i = 0; i < track->num_texture; i++)
track->textures[i].enabled = !!(temp & (1 << i));
+ track->tex_dirty = true;
}
break;
case RADEON_SE_VF_CNTL:
@@ -1602,12 +1599,14 @@ static int r100_packet0_check(struct radeon_cs_parser *p,
i = (reg - RADEON_PP_TEX_SIZE_0) / 8;
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;
+ track->tex_dirty = true;
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 = idx_value + 32;
+ track->tex_dirty = true;
break;
case RADEON_PP_TXFILTER_0:
case RADEON_PP_TXFILTER_1:
@@ -1621,6 +1620,7 @@ static int r100_packet0_check(struct radeon_cs_parser *p,
tmp = (idx_value >> 27) & 0x7;
if (tmp == 2 || tmp == 6)
track->textures[i].roundup_h = false;
+ track->tex_dirty = true;
break;
case RADEON_PP_TXFORMAT_0:
case RADEON_PP_TXFORMAT_1:
@@ -1673,6 +1673,7 @@ static int r100_packet0_check(struct radeon_cs_parser *p,
}
track->textures[i].cube_info[4].width = 1 << ((idx_value >> 16) & 0xf);
track->textures[i].cube_info[4].height = 1 << ((idx_value >> 20) & 0xf);
+ track->tex_dirty = true;
break;
case RADEON_PP_CUBIC_FACES_0:
case RADEON_PP_CUBIC_FACES_1:
@@ -1683,6 +1684,7 @@ static int r100_packet0_check(struct radeon_cs_parser *p,
track->textures[i].cube_info[face].width = 1 << ((tmp >> (face * 8)) & 0xf);
track->textures[i].cube_info[face].height = 1 << ((tmp >> ((face * 8) + 4)) & 0xf);
}
+ track->tex_dirty = true;
break;
default:
printk(KERN_ERR "Forbidden register 0x%04X in cs at %d\n",
@@ -2310,7 +2312,6 @@ void r100_vram_init_sizes(struct radeon_device *rdev)
/* FIXME we don't use the second aperture yet when we could use it */
if (rdev->mc.visible_vram_size > rdev->mc.aper_size)
rdev->mc.visible_vram_size = rdev->mc.aper_size;
- rdev->mc.active_vram_size = rdev->mc.visible_vram_size;
config_aper_size = RREG32(RADEON_CONFIG_APER_SIZE);
if (rdev->flags & RADEON_IS_IGP) {
uint32_t tom;
@@ -3318,9 +3319,9 @@ int r100_cs_track_check(struct radeon_device *rdev, struct r100_cs_track *track)
unsigned long size;
unsigned prim_walk;
unsigned nverts;
- unsigned num_cb = track->num_cb;
+ unsigned num_cb = track->cb_dirty ? track->num_cb : 0;
- if (!track->zb_cb_clear && !track->color_channel_mask &&
+ if (num_cb && !track->zb_cb_clear && !track->color_channel_mask &&
!track->blend_read_enable)
num_cb = 0;
@@ -3341,7 +3342,9 @@ int r100_cs_track_check(struct radeon_device *rdev, struct r100_cs_track *track)
return -EINVAL;
}
}
- if (track->z_enabled) {
+ track->cb_dirty = false;
+
+ if (track->zb_dirty && track->z_enabled) {
if (track->zb.robj == NULL) {
DRM_ERROR("[drm] No buffer for z buffer !\n");
return -EINVAL;
@@ -3358,6 +3361,28 @@ int r100_cs_track_check(struct radeon_device *rdev, struct r100_cs_track *track)
return -EINVAL;
}
}
+ track->zb_dirty = false;
+
+ if (track->aa_dirty && track->aaresolve) {
+ if (track->aa.robj == NULL) {
+ DRM_ERROR("[drm] No buffer for AA resolve buffer %d !\n", i);
+ return -EINVAL;
+ }
+ /* I believe the format comes from colorbuffer0. */
+ size = track->aa.pitch * track->cb[0].cpp * track->maxy;
+ size += track->aa.offset;
+ if (size > radeon_bo_size(track->aa.robj)) {
+ DRM_ERROR("[drm] Buffer too small for AA resolve buffer %d "
+ "(need %lu have %lu) !\n", i, size,
+ radeon_bo_size(track->aa.robj));
+ DRM_ERROR("[drm] AA resolve buffer %d (%u %u %u %u)\n",
+ i, track->aa.pitch, track->cb[0].cpp,
+ track->aa.offset, track->maxy);
+ return -EINVAL;
+ }
+ }
+ track->aa_dirty = false;
+
prim_walk = (track->vap_vf_cntl >> 4) & 0x3;
if (track->vap_vf_cntl & (1 << 14)) {
nverts = track->vap_alt_nverts;
@@ -3417,13 +3442,23 @@ int r100_cs_track_check(struct radeon_device *rdev, struct r100_cs_track *track)
prim_walk);
return -EINVAL;
}
- return r100_cs_track_texture_check(rdev, track);
+
+ if (track->tex_dirty) {
+ track->tex_dirty = false;
+ return r100_cs_track_texture_check(rdev, track);
+ }
+ return 0;
}
void r100_cs_track_clear(struct radeon_device *rdev, struct r100_cs_track *track)
{
unsigned i, face;
+ track->cb_dirty = true;
+ track->zb_dirty = true;
+ track->tex_dirty = true;
+ track->aa_dirty = true;
+
if (rdev->family < CHIP_R300) {
track->num_cb = 1;
if (rdev->family <= CHIP_RS200)
@@ -3437,6 +3472,8 @@ void r100_cs_track_clear(struct radeon_device *rdev, struct r100_cs_track *track
track->num_texture = 16;
track->maxy = 4096;
track->separate_cube = 0;
+ track->aaresolve = false;
+ track->aa.robj = NULL;
}
for (i = 0; i < track->num_cb; i++) {
@@ -3746,8 +3783,6 @@ static int r100_startup(struct radeon_device *rdev)
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) */
r100_enable_bm(rdev);
generated by cgit v0.10.2 at 2024-07-19 01:22:00 (GMT)