diff options
Diffstat (limited to 'drivers/edac')
-rw-r--r-- | drivers/edac/Kconfig | 7 | ||||
-rw-r--r-- | drivers/edac/Makefile | 1 | ||||
-rw-r--r-- | drivers/edac/cell_edac.c | 5 | ||||
-rw-r--r-- | drivers/edac/e752x_edac.c | 59 | ||||
-rw-r--r-- | drivers/edac/edac_core.h | 2 | ||||
-rw-r--r-- | drivers/edac/edac_device.c | 6 | ||||
-rw-r--r-- | drivers/edac/edac_mc.c | 6 | ||||
-rw-r--r-- | drivers/edac/edac_mc_sysfs.c | 158 | ||||
-rw-r--r-- | drivers/edac/edac_pci.c | 6 | ||||
-rw-r--r-- | drivers/edac/edac_pci_sysfs.c | 30 | ||||
-rw-r--r-- | drivers/edac/i5100_edac.c | 981 | ||||
-rw-r--r-- | drivers/edac/mpc85xx_edac.c | 70 | ||||
-rw-r--r-- | drivers/edac/mv64x60_edac.c | 37 |
13 files changed, 1176 insertions, 192 deletions
diff --git a/drivers/edac/Kconfig b/drivers/edac/Kconfig index 6e6c3c4..5a11e3c 100644 --- a/drivers/edac/Kconfig +++ b/drivers/edac/Kconfig @@ -123,6 +123,13 @@ config EDAC_I5000 Support for error detection and correction the Intel Greekcreek/Blackford chipsets. +config EDAC_I5100 + tristate "Intel San Clemente MCH" + depends on EDAC_MM_EDAC && X86 && PCI + help + Support for error detection and correction the Intel + San Clemente MCH. + config EDAC_MPC85XX tristate "Freescale MPC85xx" depends on EDAC_MM_EDAC && FSL_SOC && MPC85xx diff --git a/drivers/edac/Makefile b/drivers/edac/Makefile index 8380773..e5e9104 100644 --- a/drivers/edac/Makefile +++ b/drivers/edac/Makefile @@ -19,6 +19,7 @@ endif obj-$(CONFIG_EDAC_AMD76X) += amd76x_edac.o obj-$(CONFIG_EDAC_I5000) += i5000_edac.o +obj-$(CONFIG_EDAC_I5100) += i5100_edac.o obj-$(CONFIG_EDAC_E7XXX) += e7xxx_edac.o obj-$(CONFIG_EDAC_E752X) += e752x_edac.o obj-$(CONFIG_EDAC_I82443BXGX) += i82443bxgx_edac.o diff --git a/drivers/edac/cell_edac.c b/drivers/edac/cell_edac.c index b54112f..0e024fe 100644 --- a/drivers/edac/cell_edac.c +++ b/drivers/edac/cell_edac.c @@ -33,7 +33,7 @@ static void cell_edac_count_ce(struct mem_ctl_info *mci, int chan, u64 ar) { struct cell_edac_priv *priv = mci->pvt_info; struct csrow_info *csrow = &mci->csrows[0]; - unsigned long address, pfn, offset; + unsigned long address, pfn, offset, syndrome; dev_dbg(mci->dev, "ECC CE err on node %d, channel %d, ar = 0x%016lx\n", priv->node, chan, ar); @@ -44,10 +44,11 @@ static void cell_edac_count_ce(struct mem_ctl_info *mci, int chan, u64 ar) address = (address << 1) | chan; pfn = address >> PAGE_SHIFT; offset = address & ~PAGE_MASK; + syndrome = (ar & 0x000000001fe00000ul) >> 21; /* TODO: Decoding of the error addresss */ edac_mc_handle_ce(mci, csrow->first_page + pfn, offset, - 0, 0, chan, ""); + syndrome, 0, chan, ""); } static void cell_edac_count_ue(struct mem_ctl_info *mci, int chan, u64 ar) diff --git a/drivers/edac/e752x_edac.c b/drivers/edac/e752x_edac.c index c94a0eb..facfdb1 100644 --- a/drivers/edac/e752x_edac.c +++ b/drivers/edac/e752x_edac.c @@ -28,6 +28,7 @@ #define E752X_REVISION " Ver: 2.0.2 " __DATE__ #define EDAC_MOD_STR "e752x_edac" +static int report_non_memory_errors; static int force_function_unhide; static int sysbus_parity = -1; @@ -117,7 +118,7 @@ static struct edac_pci_ctl_info *e752x_pci; #define E752X_BUF_FERR 0x70 /* Memory buffer first error reg (8b) */ #define E752X_BUF_NERR 0x72 /* Memory buffer next error reg (8b) */ #define E752X_BUF_ERRMASK 0x74 /* Memory buffer error mask reg (8b) */ -#define E752X_BUF_SMICMD 0x7A /* Memory buffer SMI command reg (8b) */ +#define E752X_BUF_SMICMD 0x7A /* Memory buffer SMI cmd reg (8b) */ #define E752X_DRAM_FERR 0x80 /* DRAM first error register (16b) */ #define E752X_DRAM_NERR 0x82 /* DRAM next error register (16b) */ #define E752X_DRAM_ERRMASK 0x84 /* DRAM error mask register (8b) */ @@ -127,7 +128,7 @@ static struct edac_pci_ctl_info *e752x_pci; /* error address register (32b) */ /* * 31 Reserved - * 30:2 CE address (64 byte block 34:6) + * 30:2 CE address (64 byte block 34:6 * 1 Reserved * 0 HiLoCS */ @@ -147,11 +148,11 @@ static struct edac_pci_ctl_info *e752x_pci; * 1 Reserved * 0 HiLoCS */ -#define E752X_DRAM_SCRB_ADD 0xA8 /* DRAM first uncorrectable scrub memory */ +#define E752X_DRAM_SCRB_ADD 0xA8 /* DRAM 1st uncorrectable scrub mem */ /* error address register (32b) */ /* * 31 Reserved - * 30:2 CE address (64 byte block 34:6) + * 30:2 CE address (64 byte block 34:6 * 1 Reserved * 0 HiLoCS */ @@ -394,9 +395,12 @@ static void do_process_ded_retry(struct mem_ctl_info *mci, u16 error, struct e752x_pvt *pvt = (struct e752x_pvt *)mci->pvt_info; error_1b = retry_add; - page = error_1b >> (PAGE_SHIFT - 4); /* convert the addr to 4k page */ - row = pvt->mc_symmetric ? ((page >> 1) & 3) : /* chip select are bits 14 & 13 */ + page = error_1b >> (PAGE_SHIFT - 4); /* convert the addr to 4k page */ + + /* chip select are bits 14 & 13 */ + row = pvt->mc_symmetric ? ((page >> 1) & 3) : edac_mc_find_csrow_by_page(mci, page); + e752x_mc_printk(mci, KERN_WARNING, "CE page 0x%lx, row %d : Memory read retry\n", (long unsigned int)page, row); @@ -422,12 +426,21 @@ static inline void process_threshold_ce(struct mem_ctl_info *mci, u16 error, } static char *global_message[11] = { - "PCI Express C1", "PCI Express C", "PCI Express B1", - "PCI Express B", "PCI Express A1", "PCI Express A", - "DMA Controler", "HUB or NS Interface", "System Bus", - "DRAM Controler", "Internal Buffer" + "PCI Express C1", + "PCI Express C", + "PCI Express B1", + "PCI Express B", + "PCI Express A1", + "PCI Express A", + "DMA Controller", + "HUB or NS Interface", + "System Bus", + "DRAM Controller", /* 9th entry */ + "Internal Buffer" }; +#define DRAM_ENTRY 9 + static char *fatal_message[2] = { "Non-Fatal ", "Fatal " }; static void do_global_error(int fatal, u32 errors) @@ -435,9 +448,16 @@ static void do_global_error(int fatal, u32 errors) int i; for (i = 0; i < 11; i++) { - if (errors & (1 << i)) - e752x_printk(KERN_WARNING, "%sError %s\n", - fatal_message[fatal], global_message[i]); + if (errors & (1 << i)) { + /* If the error is from DRAM Controller OR + * we are to report ALL errors, then + * report the error + */ + if ((i == DRAM_ENTRY) || report_non_memory_errors) + e752x_printk(KERN_WARNING, "%sError %s\n", + fatal_message[fatal], + global_message[i]); + } } } @@ -1021,7 +1041,7 @@ static int e752x_get_devs(struct pci_dev *pdev, int dev_idx, struct pci_dev *dev; pvt->bridge_ck = pci_get_device(PCI_VENDOR_ID_INTEL, - pvt->dev_info->err_dev, pvt->bridge_ck); + pvt->dev_info->err_dev, pvt->bridge_ck); if (pvt->bridge_ck == NULL) pvt->bridge_ck = pci_scan_single_device(pdev->bus, @@ -1034,8 +1054,9 @@ static int e752x_get_devs(struct pci_dev *pdev, int dev_idx, return 1; } - dev = pci_get_device(PCI_VENDOR_ID_INTEL, e752x_devs[dev_idx].ctl_dev, - NULL); + dev = pci_get_device(PCI_VENDOR_ID_INTEL, + e752x_devs[dev_idx].ctl_dev, + NULL); if (dev == NULL) goto fail; @@ -1316,7 +1337,8 @@ MODULE_DESCRIPTION("MC support for Intel e752x/3100 memory controllers"); module_param(force_function_unhide, int, 0444); MODULE_PARM_DESC(force_function_unhide, "if BIOS sets Dev0:Fun1 up as hidden:" - " 1=force unhide and hope BIOS doesn't fight driver for Dev0:Fun1 access"); + " 1=force unhide and hope BIOS doesn't fight driver for " + "Dev0:Fun1 access"); module_param(edac_op_state, int, 0444); MODULE_PARM_DESC(edac_op_state, "EDAC Error Reporting state: 0=Poll,1=NMI"); @@ -1324,3 +1346,6 @@ MODULE_PARM_DESC(edac_op_state, "EDAC Error Reporting state: 0=Poll,1=NMI"); module_param(sysbus_parity, int, 0444); MODULE_PARM_DESC(sysbus_parity, "0=disable system bus parity checking," " 1=enable system bus parity checking, default=auto-detect"); +module_param(report_non_memory_errors, int, 0644); +MODULE_PARM_DESC(report_non_memory_errors, "0=disable non-memory error " + "reporting, 1=enable non-memory error reporting"); diff --git a/drivers/edac/edac_core.h b/drivers/edac/edac_core.h index a9aa845..b27b13c 100644 --- a/drivers/edac/edac_core.h +++ b/drivers/edac/edac_core.h @@ -97,7 +97,7 @@ extern int edac_debug_level; #define PCI_VEND_DEV(vend, dev) PCI_VENDOR_ID_ ## vend, \ PCI_DEVICE_ID_ ## vend ## _ ## dev -#define dev_name(dev) (dev)->dev_name +#define edac_dev_name(dev) (dev)->dev_name /* memory devices */ enum dev_type { diff --git a/drivers/edac/edac_device.c b/drivers/edac/edac_device.c index 63372fa..5fcd3d8 100644 --- a/drivers/edac/edac_device.c +++ b/drivers/edac/edac_device.c @@ -333,7 +333,7 @@ static int add_edac_dev_to_global_list(struct edac_device_ctl_info *edac_dev) fail0: edac_printk(KERN_WARNING, EDAC_MC, "%s (%s) %s %s already assigned %d\n", - rover->dev->bus_id, dev_name(rover), + rover->dev->bus_id, edac_dev_name(rover), rover->mod_name, rover->ctl_name, rover->dev_idx); return 1; @@ -538,7 +538,7 @@ int edac_device_add_device(struct edac_device_ctl_info *edac_dev) "'%s': DEV '%s' (%s)\n", edac_dev->mod_name, edac_dev->ctl_name, - dev_name(edac_dev), + edac_dev_name(edac_dev), edac_op_state_to_string(edac_dev->op_state)); mutex_unlock(&device_ctls_mutex); @@ -599,7 +599,7 @@ struct edac_device_ctl_info *edac_device_del_device(struct device *dev) edac_printk(KERN_INFO, EDAC_MC, "Removed device %d for %s %s: DEV %s\n", edac_dev->dev_idx, - edac_dev->mod_name, edac_dev->ctl_name, dev_name(edac_dev)); + edac_dev->mod_name, edac_dev->ctl_name, edac_dev_name(edac_dev)); return edac_dev; } diff --git a/drivers/edac/edac_mc.c b/drivers/edac/edac_mc.c index a4cf164..d110392 100644 --- a/drivers/edac/edac_mc.c +++ b/drivers/edac/edac_mc.c @@ -402,7 +402,7 @@ static int add_mc_to_global_list(struct mem_ctl_info *mci) fail0: edac_printk(KERN_WARNING, EDAC_MC, "%s (%s) %s %s already assigned %d\n", p->dev->bus_id, - dev_name(mci), p->mod_name, p->ctl_name, p->mc_idx); + edac_dev_name(mci), p->mod_name, p->ctl_name, p->mc_idx); return 1; fail1: @@ -517,7 +517,7 @@ int edac_mc_add_mc(struct mem_ctl_info *mci) /* Report action taken */ edac_mc_printk(mci, KERN_INFO, "Giving out device to '%s' '%s':" - " DEV %s\n", mci->mod_name, mci->ctl_name, dev_name(mci)); + " DEV %s\n", mci->mod_name, mci->ctl_name, edac_dev_name(mci)); mutex_unlock(&mem_ctls_mutex); return 0; @@ -565,7 +565,7 @@ struct mem_ctl_info *edac_mc_del_mc(struct device *dev) edac_printk(KERN_INFO, EDAC_MC, "Removed device %d for %s %s: DEV %s\n", mci->mc_idx, - mci->mod_name, mci->ctl_name, dev_name(mci)); + mci->mod_name, mci->ctl_name, edac_dev_name(mci)); return mci; } diff --git a/drivers/edac/edac_mc_sysfs.c b/drivers/edac/edac_mc_sysfs.c index 021d187..ad218fe 100644 --- a/drivers/edac/edac_mc_sysfs.c +++ b/drivers/edac/edac_mc_sysfs.c @@ -44,6 +44,25 @@ int edac_mc_get_poll_msec(void) return edac_mc_poll_msec; } +static int edac_set_poll_msec(const char *val, struct kernel_param *kp) +{ + long l; + int ret; + + if (!val) + return -EINVAL; + + ret = strict_strtol(val, 0, &l); + if (ret == -EINVAL || ((int)l != l)) + return -EINVAL; + *((int *)kp->arg) = l; + + /* notify edac_mc engine to reset the poll period */ + edac_mc_reset_delay_period(l); + + return 0; +} + /* Parameter declarations for above */ module_param(edac_mc_panic_on_ue, int, 0644); MODULE_PARM_DESC(edac_mc_panic_on_ue, "Panic on uncorrected error: 0=off 1=on"); @@ -53,7 +72,8 @@ MODULE_PARM_DESC(edac_mc_log_ue, module_param(edac_mc_log_ce, int, 0644); MODULE_PARM_DESC(edac_mc_log_ce, "Log correctable error to console: 0=off 1=on"); -module_param(edac_mc_poll_msec, int, 0644); +module_param_call(edac_mc_poll_msec, edac_set_poll_msec, param_get_int, + &edac_mc_poll_msec, 0644); MODULE_PARM_DESC(edac_mc_poll_msec, "Polling period in milliseconds"); /* @@ -103,16 +123,6 @@ static const char *edac_caps[] = { -/* - * /sys/devices/system/edac/mc; - * data structures and methods - */ -static ssize_t memctrl_int_show(void *ptr, char *buffer) -{ - int *value = (int *)ptr; - return sprintf(buffer, "%u\n", *value); -} - static ssize_t memctrl_int_store(void *ptr, const char *buffer, size_t count) { int *value = (int *)ptr; @@ -123,23 +133,6 @@ static ssize_t memctrl_int_store(void *ptr, const char *buffer, size_t count) return count; } -/* - * mc poll_msec time value - */ -static ssize_t poll_msec_int_store(void *ptr, const char *buffer, size_t count) -{ - int *value = (int *)ptr; - - if (isdigit(*buffer)) { - *value = simple_strtoul(buffer, NULL, 0); - - /* notify edac_mc engine to reset the poll period */ - edac_mc_reset_delay_period(*value); - } - - return count; -} - /* EDAC sysfs CSROW data structures and methods */ @@ -185,7 +178,11 @@ static ssize_t csrow_edac_mode_show(struct csrow_info *csrow, char *data, static ssize_t channel_dimm_label_show(struct csrow_info *csrow, char *data, int channel) { - return snprintf(data, EDAC_MC_LABEL_LEN, "%s", + /* if field has not been initialized, there is nothing to send */ + if (!csrow->channels[channel].label[0]) + return 0; + + return snprintf(data, EDAC_MC_LABEL_LEN, "%s\n", csrow->channels[channel].label); } @@ -649,98 +646,10 @@ static struct kobj_type ktype_mci = { .default_attrs = (struct attribute **)mci_attr, }; -/* show/store, tables, etc for the MC kset */ - - -struct memctrl_dev_attribute { - struct attribute attr; - void *value; - ssize_t(*show) (void *, char *); - ssize_t(*store) (void *, const char *, size_t); -}; - -/* Set of show/store abstract level functions for memory control object */ -static ssize_t memctrl_dev_show(struct kobject *kobj, - struct attribute *attr, char *buffer) -{ - struct memctrl_dev_attribute *memctrl_dev; - memctrl_dev = (struct memctrl_dev_attribute *)attr; - - if (memctrl_dev->show) - return memctrl_dev->show(memctrl_dev->value, buffer); - - return -EIO; -} - -static ssize_t memctrl_dev_store(struct kobject *kobj, struct attribute *attr, - const char *buffer, size_t count) -{ - struct memctrl_dev_attribute *memctrl_dev; - memctrl_dev = (struct memctrl_dev_attribute *)attr; - - if (memctrl_dev->store) - return memctrl_dev->store(memctrl_dev->value, buffer, count); - - return -EIO; -} - -static struct sysfs_ops memctrlfs_ops = { - .show = memctrl_dev_show, - .store = memctrl_dev_store -}; - -#define MEMCTRL_ATTR(_name, _mode, _show, _store) \ -static struct memctrl_dev_attribute attr_##_name = { \ - .attr = {.name = __stringify(_name), .mode = _mode }, \ - .value = &_name, \ - .show = _show, \ - .store = _store, \ -}; - -#define MEMCTRL_STRING_ATTR(_name, _data, _mode, _show, _store) \ -static struct memctrl_dev_attribute attr_##_name = { \ - .attr = {.name = __stringify(_name), .mode = _mode }, \ - .value = _data, \ - .show = _show, \ - .store = _store, \ -}; - -/* csrow<id> control files */ -MEMCTRL_ATTR(edac_mc_panic_on_ue, - S_IRUGO | S_IWUSR, memctrl_int_show, memctrl_int_store); - -MEMCTRL_ATTR(edac_mc_log_ue, - S_IRUGO | S_IWUSR, memctrl_int_show, memctrl_int_store); - -MEMCTRL_ATTR(edac_mc_log_ce, - S_IRUGO | S_IWUSR, memctrl_int_show, memctrl_int_store); - -MEMCTRL_ATTR(edac_mc_poll_msec, - S_IRUGO | S_IWUSR, memctrl_int_show, poll_msec_int_store); - -/* Base Attributes of the memory ECC object */ -static struct memctrl_dev_attribute *memctrl_attr[] = { - &attr_edac_mc_panic_on_ue, - &attr_edac_mc_log_ue, - &attr_edac_mc_log_ce, - &attr_edac_mc_poll_msec, - NULL, -}; - - -/* the ktype for the mc_kset internal kobj */ -static struct kobj_type ktype_mc_set_attribs = { - .sysfs_ops = &memctrlfs_ops, - .default_attrs = (struct attribute **)memctrl_attr, -}; - /* EDAC memory controller sysfs kset: * /sys/devices/system/edac/mc */ -static struct kset mc_kset = { - .kobj = {.ktype = &ktype_mc_set_attribs }, -}; - +static struct kset *mc_kset; /* * edac_mc_register_sysfs_main_kobj @@ -771,7 +680,7 @@ int edac_mc_register_sysfs_main_kobj(struct mem_ctl_info *mci) } /* this instance become part of the mc_kset */ - kobj_mci->kset = &mc_kset; + kobj_mci->kset = mc_kset; /* register the mc<id> kobject to the mc_kset */ err = kobject_init_and_add(kobj_mci, &ktype_mci, NULL, @@ -1001,12 +910,9 @@ int edac_sysfs_setup_mc_kset(void) } /* Init the MC's kobject */ - kobject_set_name(&mc_kset.kobj, "mc"); - mc_kset.kobj.parent = &edac_class->kset.kobj; - - /* register the mc_kset */ - err = kset_register(&mc_kset); - if (err) { + mc_kset = kset_create_and_add("mc", NULL, &edac_class->kset.kobj); + if (!mc_kset) { + err = -ENOMEM; debugf1("%s() Failed to register '.../edac/mc'\n", __func__); goto fail_out; } @@ -1028,6 +934,6 @@ fail_out: */ void edac_sysfs_teardown_mc_kset(void) { - kset_unregister(&mc_kset); + kset_unregister(mc_kset); } diff --git a/drivers/edac/edac_pci.c b/drivers/edac/edac_pci.c index 9b24340..22ec9d5 100644 --- a/drivers/edac/edac_pci.c +++ b/drivers/edac/edac_pci.c @@ -150,7 +150,7 @@ static int add_edac_pci_to_global_list(struct edac_pci_ctl_info *pci) fail0: edac_printk(KERN_WARNING, EDAC_PCI, "%s (%s) %s %s already assigned %d\n", - rover->dev->bus_id, dev_name(rover), + rover->dev->bus_id, edac_dev_name(rover), rover->mod_name, rover->ctl_name, rover->pci_idx); return 1; @@ -360,7 +360,7 @@ int edac_pci_add_device(struct edac_pci_ctl_info *pci, int edac_idx) " DEV '%s' (%s)\n", pci->mod_name, pci->ctl_name, - dev_name(pci), edac_op_state_to_string(pci->op_state)); + edac_dev_name(pci), edac_op_state_to_string(pci->op_state)); mutex_unlock(&edac_pci_ctls_mutex); return 0; @@ -415,7 +415,7 @@ struct edac_pci_ctl_info *edac_pci_del_device(struct device *dev) edac_printk(KERN_INFO, EDAC_PCI, "Removed device %d for %s %s: DEV %s\n", - pci->pci_idx, pci->mod_name, pci->ctl_name, dev_name(pci)); + pci->pci_idx, pci->mod_name, pci->ctl_name, edac_dev_name(pci)); return pci; } diff --git a/drivers/edac/edac_pci_sysfs.c b/drivers/edac/edac_pci_sysfs.c index 2c1fa1b..5c153dc 100644 --- a/drivers/edac/edac_pci_sysfs.c +++ b/drivers/edac/edac_pci_sysfs.c @@ -28,7 +28,7 @@ static int edac_pci_poll_msec = 1000; /* one second workq period */ static atomic_t pci_parity_count = ATOMIC_INIT(0); static atomic_t pci_nonparity_count = ATOMIC_INIT(0); -static struct kobject edac_pci_top_main_kobj; +static struct kobject *edac_pci_top_main_kobj; static atomic_t edac_pci_sysfs_refcount = ATOMIC_INIT(0); /* getter functions for the data variables */ @@ -83,7 +83,7 @@ static void edac_pci_instance_release(struct kobject *kobj) pci = to_instance(kobj); /* decrement reference count on top main kobj */ - kobject_put(&edac_pci_top_main_kobj); + kobject_put(edac_pci_top_main_kobj); kfree(pci); /* Free the control struct */ } @@ -166,7 +166,7 @@ static int edac_pci_create_instance_kobj(struct edac_pci_ctl_info *pci, int idx) * track the number of PCI instances we have, and thus nest * properly on keeping the module loaded */ - main_kobj = kobject_get(&edac_pci_top_main_kobj); + main_kobj = kobject_get(edac_pci_top_main_kobj); if (!main_kobj) { err = -ENODEV; goto error_out; @@ -174,11 +174,11 @@ static int edac_pci_create_instance_kobj(struct edac_pci_ctl_info *pci, int idx) /* And now register this new kobject under the main kobj */ err = kobject_init_and_add(&pci->kobj, &ktype_pci_instance, - &edac_pci_top_main_kobj, "pci%d", idx); + edac_pci_top_main_kobj, "pci%d", idx); if (err != 0) { debugf2("%s() failed to register instance pci%d\n", __func__, idx); - kobject_put(&edac_pci_top_main_kobj); + kobject_put(edac_pci_top_main_kobj); goto error_out; } @@ -316,9 +316,10 @@ static struct edac_pci_dev_attribute *edac_pci_attr[] = { */ static void edac_pci_release_main_kobj(struct kobject *kobj) { - debugf0("%s() here to module_put(THIS_MODULE)\n", __func__); + kfree(kobj); + /* last reference to top EDAC PCI kobject has been removed, * NOW release our ref count on the core module */ @@ -369,8 +370,16 @@ static int edac_pci_main_kobj_setup(void) goto decrement_count_fail; } + edac_pci_top_main_kobj = kzalloc(sizeof(struct kobject), GFP_KERNEL); + if (!edac_pci_top_main_kobj) { + debugf1("Failed to allocate\n"); + err = -ENOMEM; + goto kzalloc_fail; + } + /* Instanstiate the pci object */ - err = kobject_init_and_add(&edac_pci_top_main_kobj, &ktype_edac_pci_main_kobj, + err = kobject_init_and_add(edac_pci_top_main_kobj, + &ktype_edac_pci_main_kobj, &edac_class->kset.kobj, "pci"); if (err) { debugf1("Failed to register '.../edac/pci'\n"); @@ -381,13 +390,16 @@ static int edac_pci_main_kobj_setup(void) * for EDAC PCI, then edac_pci_main_kobj_teardown() * must be used, for resources to be cleaned up properly */ - kobject_uevent(&edac_pci_top_main_kobj, KOBJ_ADD); + kobject_uevent(edac_pci_top_main_kobj, KOBJ_ADD); debugf1("Registered '.../edac/pci' kobject\n"); return 0; /* Error unwind statck */ kobject_init_and_add_fail: + kfree(edac_pci_top_main_kobj); + +kzalloc_fail: module_put(THIS_MODULE); decrement_count_fail: @@ -414,7 +426,7 @@ static void edac_pci_main_kobj_teardown(void) if (atomic_dec_return(&edac_pci_sysfs_refcount) == 0) { debugf0("%s() called kobject_put on main kobj\n", __func__); - kobject_put(&edac_pci_top_main_kobj); + kobject_put(edac_pci_top_main_kobj); } } diff --git a/drivers/edac/i5100_edac.c b/drivers/edac/i5100_edac.c new file mode 100644 index 0000000..22db05a --- /dev/null +++ b/drivers/edac/i5100_edac.c @@ -0,0 +1,981 @@ +/* + * Intel 5100 Memory Controllers kernel module + * + * This file may be distributed under the terms of the + * GNU General Public License. + * + * This module is based on the following document: + * + * Intel 5100X Chipset Memory Controller Hub (MCH) - Datasheet + * http://download.intel.com/design/chipsets/datashts/318378.pdf + * + */ +#include <linux/module.h> +#include <linux/init.h> +#include <linux/pci.h> +#include <linux/pci_ids.h> +#include <linux/slab.h> +#include <linux/edac.h> +#include <linux/delay.h> +#include <linux/mmzone.h> + +#include "edac_core.h" + +/* register addresses */ + +/* device 16, func 1 */ +#define I5100_MC 0x40 /* Memory Control Register */ +#define I5100_MS 0x44 /* Memory Status Register */ +#define I5100_SPDDATA 0x48 /* Serial Presence Detect Status Reg */ +#define I5100_SPDCMD 0x4c /* Serial Presence Detect Command Reg */ +#define I5100_TOLM 0x6c /* Top of Low Memory */ +#define I5100_MIR0 0x80 /* Memory Interleave Range 0 */ +#define I5100_MIR1 0x84 /* Memory Interleave Range 1 */ +#define I5100_AMIR_0 0x8c /* Adjusted Memory Interleave Range 0 */ +#define I5100_AMIR_1 0x90 /* Adjusted Memory Interleave Range 1 */ +#define I5100_FERR_NF_MEM 0xa0 /* MC First Non Fatal Errors */ +#define I5100_FERR_NF_MEM_M16ERR_MASK (1 << 16) +#define I5100_FERR_NF_MEM_M15ERR_MASK (1 << 15) +#define I5100_FERR_NF_MEM_M14ERR_MASK (1 << 14) +#define I5100_FERR_NF_MEM_M12ERR_MASK (1 << 12) +#define I5100_FERR_NF_MEM_M11ERR_MASK (1 << 11) +#define I5100_FERR_NF_MEM_M10ERR_MASK (1 << 10) +#define I5100_FERR_NF_MEM_M6ERR_MASK (1 << 6) +#define I5100_FERR_NF_MEM_M5ERR_MASK (1 << 5) +#define I5100_FERR_NF_MEM_M4ERR_MASK (1 << 4) +#define I5100_FERR_NF_MEM_M1ERR_MASK 1 +#define I5100_FERR_NF_MEM_ANY_MASK \ + (I5100_FERR_NF_MEM_M16ERR_MASK | \ + I5100_FERR_NF_MEM_M15ERR_MASK | \ + I5100_FERR_NF_MEM_M14ERR_MASK | \ + I5100_FERR_NF_MEM_M12ERR_MASK | \ + I5100_FERR_NF_MEM_M11ERR_MASK | \ + I5100_FERR_NF_MEM_M10ERR_MASK | \ + I5100_FERR_NF_MEM_M6ERR_MASK | \ + I5100_FERR_NF_MEM_M5ERR_MASK | \ + I5100_FERR_NF_MEM_M4ERR_MASK | \ + I5100_FERR_NF_MEM_M1ERR_MASK) +#define I5100_NERR_NF_MEM 0xa4 /* MC Next Non-Fatal Errors */ +#define I5100_EMASK_MEM 0xa8 /* MC Error Mask Register */ + +/* device 21 and 22, func 0 */ +#define I5100_MTR_0 0x154 /* Memory Technology Registers 0-3 */ +#define I5100_DMIR 0x15c /* DIMM Interleave Range */ +#define I5100_VALIDLOG 0x18c /* Valid Log Markers */ +#define I5100_NRECMEMA 0x190 /* Non-Recoverable Memory Error Log Reg A */ +#define I5100_NRECMEMB 0x194 /* Non-Recoverable Memory Error Log Reg B */ +#define I5100_REDMEMA 0x198 /* Recoverable Memory Data Error Log Reg A */ +#define I5100_REDMEMB 0x19c /* Recoverable Memory Data Error Log Reg B */ +#define I5100_RECMEMA 0x1a0 /* Recoverable Memory Error Log Reg A */ +#define I5100_RECMEMB 0x1a4 /* Recoverable Memory Error Log Reg B */ +#define I5100_MTR_4 0x1b0 /* Memory Technology Registers 4,5 */ + +/* bit field accessors */ + +static inline u32 i5100_mc_errdeten(u32 mc) +{ + return mc >> 5 & 1; +} + +static inline u16 i5100_spddata_rdo(u16 a) +{ + return a >> 15 & 1; +} + +static inline u16 i5100_spddata_sbe(u16 a) +{ + return a >> 13 & 1; +} + +static inline u16 i5100_spddata_busy(u16 a) +{ + return a >> 12 & 1; +} + +static inline u16 i5100_spddata_data(u16 a) +{ + return a & ((1 << 8) - 1); +} + +static inline u32 i5100_spdcmd_create(u32 dti, u32 ckovrd, u32 sa, u32 ba, + u32 data, u32 cmd) +{ + return ((dti & ((1 << 4) - 1)) << 28) | + ((ckovrd & 1) << 27) | + ((sa & ((1 << 3) - 1)) << 24) | + ((ba & ((1 << 8) - 1)) << 16) | + ((data & ((1 << 8) - 1)) << 8) | + (cmd & 1); +} + +static inline u16 i5100_tolm_tolm(u16 a) +{ + return a >> 12 & ((1 << 4) - 1); +} + +static inline u16 i5100_mir_limit(u16 a) +{ + return a >> 4 & ((1 << 12) - 1); +} + +static inline u16 i5100_mir_way1(u16 a) +{ + return a >> 1 & 1; +} + +static inline u16 i5100_mir_way0(u16 a) +{ + return a & 1; +} + +static inline u32 i5100_ferr_nf_mem_chan_indx(u32 a) +{ + return a >> 28 & 1; +} + +static inline u32 i5100_ferr_nf_mem_any(u32 a) +{ + return a & I5100_FERR_NF_MEM_ANY_MASK; +} + +static inline u32 i5100_nerr_nf_mem_any(u32 a) +{ + return i5100_ferr_nf_mem_any(a); +} + +static inline u32 i5100_dmir_limit(u32 a) +{ + return a >> 16 & ((1 << 11) - 1); +} + +static inline u32 i5100_dmir_rank(u32 a, u32 i) +{ + return a >> (4 * i) & ((1 << 2) - 1); +} + +static inline u16 i5100_mtr_present(u16 a) +{ + return a >> 10 & 1; +} + +static inline u16 i5100_mtr_ethrottle(u16 a) +{ + return a >> 9 & 1; +} + +static inline u16 i5100_mtr_width(u16 a) +{ + return a >> 8 & 1; +} + +static inline u16 i5100_mtr_numbank(u16 a) +{ + return a >> 6 & 1; +} + +static inline u16 i5100_mtr_numrow(u16 a) +{ + return a >> 2 & ((1 << 2) - 1); +} + +static inline u16 i5100_mtr_numcol(u16 a) +{ + return a & ((1 << 2) - 1); +} + + +static inline u32 i5100_validlog_redmemvalid(u32 a) +{ + return a >> 2 & 1; +} + +static inline u32 i5100_validlog_recmemvalid(u32 a) +{ + return a >> 1 & 1; +} + +static inline u32 i5100_validlog_nrecmemvalid(u32 a) +{ + return a & 1; +} + +static inline u32 i5100_nrecmema_merr(u32 a) +{ + return a >> 15 & ((1 << 5) - 1); +} + +static inline u32 i5100_nrecmema_bank(u32 a) +{ + return a >> 12 & ((1 << 3) - 1); +} + +static inline u32 i5100_nrecmema_rank(u32 a) +{ + return a >> 8 & ((1 << 3) - 1); +} + +static inline u32 i5100_nrecmema_dm_buf_id(u32 a) +{ + return a & ((1 << 8) - 1); +} + +static inline u32 i5100_nrecmemb_cas(u32 a) +{ + return a >> 16 & ((1 << 13) - 1); +} + +static inline u32 i5100_nrecmemb_ras(u32 a) +{ + return a & ((1 << 16) - 1); +} + +static inline u32 i5100_redmemb_ecc_locator(u32 a) +{ + return a & ((1 << 18) - 1); +} + +static inline u32 i5100_recmema_merr(u32 a) +{ + return i5100_nrecmema_merr(a); +} + +static inline u32 i5100_recmema_bank(u32 a) +{ + return i5100_nrecmema_bank(a); +} + +static inline u32 i5100_recmema_rank(u32 a) +{ + return i5100_nrecmema_rank(a); +} + +static inline u32 i5100_recmema_dm_buf_id(u32 a) +{ + return i5100_nrecmema_dm_buf_id(a); +} + +static inline u32 i5100_recmemb_cas(u32 a) +{ + return i5100_nrecmemb_cas(a); +} + +static inline u32 i5100_recmemb_ras(u32 a) +{ + return i5100_nrecmemb_ras(a); +} + +/* some generic limits */ +#define I5100_MAX_RANKS_PER_CTLR 6 +#define I5100_MAX_CTLRS 2 +#define I5100_MAX_RANKS_PER_DIMM 4 +#define I5100_DIMM_ADDR_LINES (6 - 3) /* 64 bits / 8 bits per byte */ +#define I5100_MAX_DIMM_SLOTS_PER_CTLR 4 +#define I5100_MAX_RANK_INTERLEAVE 4 +#define I5100_MAX_DMIRS 5 + +struct i5100_priv { + /* ranks on each dimm -- 0 maps to not present -- obtained via SPD */ + int dimm_numrank[I5100_MAX_CTLRS][I5100_MAX_DIMM_SLOTS_PER_CTLR]; + + /* + * mainboard chip select map -- maps i5100 chip selects to + * DIMM slot chip selects. In the case of only 4 ranks per + * controller, the mapping is fairly obvious but not unique. + * we map -1 -> NC and assume both controllers use the same + * map... + * + */ + int dimm_csmap[I5100_MAX_DIMM_SLOTS_PER_CTLR][I5100_MAX_RANKS_PER_DIMM]; + + /* memory interleave range */ + struct { + u64 limit; + unsigned way[2]; + } mir[I5100_MAX_CTLRS]; + + /* adjusted memory interleave range register */ + unsigned amir[I5100_MAX_CTLRS]; + + /* dimm interleave range */ + struct { + unsigned rank[I5100_MAX_RANK_INTERLEAVE]; + u64 limit; + } dmir[I5100_MAX_CTLRS][I5100_MAX_DMIRS]; + + /* memory technology registers... */ + struct { + unsigned present; /* 0 or 1 */ + unsigned ethrottle; /* 0 or 1 */ + unsigned width; /* 4 or 8 bits */ + unsigned numbank; /* 2 or 3 lines */ + unsigned numrow; /* 13 .. 16 lines */ + unsigned numcol; /* 11 .. 12 lines */ + } mtr[I5100_MAX_CTLRS][I5100_MAX_RANKS_PER_CTLR]; + + u64 tolm; /* top of low memory in bytes */ + unsigned ranksperctlr; /* number of ranks per controller */ + + struct pci_dev *mc; /* device 16 func 1 */ + struct pci_dev *ch0mm; /* device 21 func 0 */ + struct pci_dev *ch1mm; /* device 22 func 0 */ +}; + +/* map a rank/ctlr to a slot number on the mainboard */ +static int i5100_rank_to_slot(const struct mem_ctl_info *mci, + int ctlr, int rank) +{ + const struct i5100_priv *priv = mci->pvt_info; + int i; + + for (i = 0; i < I5100_MAX_DIMM_SLOTS_PER_CTLR; i++) { + int j; + const int numrank = priv->dimm_numrank[ctlr][i]; + + for (j = 0; j < numrank; j++) + if (priv->dimm_csmap[i][j] == rank) + return i * 2 + ctlr; + } + + return -1; +} + +static const char *i5100_err_msg(unsigned err) +{ + static const char *merrs[] = { + "unknown", /* 0 */ + "uncorrectable data ECC on replay", /* 1 */ + "unknown", /* 2 */ + "unknown", /* 3 */ + "aliased uncorrectable demand data ECC", /* 4 */ + "aliased uncorrectable spare-copy data ECC", /* 5 */ + "aliased uncorrectable patrol data ECC", /* 6 */ + "unknown", /* 7 */ + "unknown", /* 8 */ + "unknown", /* 9 */ + "non-aliased uncorrectable demand data ECC", /* 10 */ + "non-aliased uncorrectable spare-copy data ECC", /* 11 */ + "non-aliased uncorrectable patrol data ECC", /* 12 */ + "unknown", /* 13 */ + "correctable demand data ECC", /* 14 */ + "correctable spare-copy data ECC", /* 15 */ + "correctable patrol data ECC", /* 16 */ + "unknown", /* 17 */ + "SPD protocol error", /* 18 */ + "unknown", /* 19 */ + "spare copy initiated", /* 20 */ + "spare copy completed", /* 21 */ + }; + unsigned i; + + for (i = 0; i < ARRAY_SIZE(merrs); i++) + if (1 << i & err) + return merrs[i]; + + return "none"; +} + +/* convert csrow index into a rank (per controller -- 0..5) */ +static int i5100_csrow_to_rank(const struct mem_ctl_info *mci, int csrow) +{ + const struct i5100_priv *priv = mci->pvt_info; + + return csrow % priv->ranksperctlr; +} + +/* convert csrow index into a controller (0..1) */ +static int i5100_csrow_to_cntlr(const struct mem_ctl_info *mci, int csrow) +{ + const struct i5100_priv *priv = mci->pvt_info; + + return csrow / priv->ranksperctlr; +} + +static unsigned i5100_rank_to_csrow(const struct mem_ctl_info *mci, + int ctlr, int rank) +{ + const struct i5100_priv *priv = mci->pvt_info; + + return ctlr * priv->ranksperctlr + rank; +} + +static void i5100_handle_ce(struct mem_ctl_info *mci, + int ctlr, + unsigned bank, + unsigned rank, + unsigned long syndrome, + unsigned cas, + unsigned ras, + const char *msg) +{ + const int csrow = i5100_rank_to_csrow(mci, ctlr, rank); + + printk(KERN_ERR + "CE ctlr %d, bank %u, rank %u, syndrome 0x%lx, " + "cas %u, ras %u, csrow %u, label \"%s\": %s\n", + ctlr, bank, rank, syndrome, cas, ras, + csrow, mci->csrows[csrow].channels[0].label, msg); + + mci->ce_count++; + mci->csrows[csrow].ce_count++; + mci->csrows[csrow].channels[0].ce_count++; +} + +static void i5100_handle_ue(struct mem_ctl_info *mci, + int ctlr, + unsigned bank, + unsigned rank, + unsigned long syndrome, + unsigned cas, + unsigned ras, + const char *msg) +{ + const int csrow = i5100_rank_to_csrow(mci, ctlr, rank); + + printk(KERN_ERR + "UE ctlr %d, bank %u, rank %u, syndrome 0x%lx, " + "cas %u, ras %u, csrow %u, label \"%s\": %s\n", + ctlr, bank, rank, syndrome, cas, ras, + csrow, mci->csrows[csrow].channels[0].label, msg); + + mci->ue_count++; + mci->csrows[csrow].ue_count++; +} + +static void i5100_read_log(struct mem_ctl_info *mci, int ctlr, + u32 ferr, u32 nerr) +{ + struct i5100_priv *priv = mci->pvt_info; + struct pci_dev *pdev = (ctlr) ? priv->ch1mm : priv->ch0mm; + u32 dw; + u32 dw2; + unsigned syndrome = 0; + unsigned ecc_loc = 0; + unsigned merr; + unsigned bank; + unsigned rank; + unsigned cas; + unsigned ras; + + pci_read_config_dword(pdev, I5100_VALIDLOG, &dw); + + if (i5100_validlog_redmemvalid(dw)) { + pci_read_config_dword(pdev, I5100_REDMEMA, &dw2); + syndrome = dw2; + pci_read_config_dword(pdev, I5100_REDMEMB, &dw2); + ecc_loc = i5100_redmemb_ecc_locator(dw2); + } + + if (i5100_validlog_recmemvalid(dw)) { + const char *msg; + + pci_read_config_dword(pdev, I5100_RECMEMA, &dw2); + merr = i5100_recmema_merr(dw2); + bank = i5100_recmema_bank(dw2); + rank = i5100_recmema_rank(dw2); + + pci_read_config_dword(pdev, I5100_RECMEMB, &dw2); + cas = i5100_recmemb_cas(dw2); + ras = i5100_recmemb_ras(dw2); + + /* FIXME: not really sure if this is what merr is... + */ + if (!merr) + msg = i5100_err_msg(ferr); + else + msg = i5100_err_msg(nerr); + + i5100_handle_ce(mci, ctlr, bank, rank, syndrome, cas, ras, msg); + } + + if (i5100_validlog_nrecmemvalid(dw)) { + const char *msg; + + pci_read_config_dword(pdev, I5100_NRECMEMA, &dw2); + merr = i5100_nrecmema_merr(dw2); + bank = i5100_nrecmema_bank(dw2); + rank = i5100_nrecmema_rank(dw2); + + pci_read_config_dword(pdev, I5100_NRECMEMB, &dw2); + cas = i5100_nrecmemb_cas(dw2); + ras = i5100_nrecmemb_ras(dw2); + + /* FIXME: not really sure if this is what merr is... + */ + if (!merr) + msg = i5100_err_msg(ferr); + else + msg = i5100_err_msg(nerr); + + i5100_handle_ue(mci, ctlr, bank, rank, syndrome, cas, ras, msg); + } + + pci_write_config_dword(pdev, I5100_VALIDLOG, dw); +} + +static void i5100_check_error(struct mem_ctl_info *mci) +{ + struct i5100_priv *priv = mci->pvt_info; + u32 dw; + + + pci_read_config_dword(priv->mc, I5100_FERR_NF_MEM, &dw); + if (i5100_ferr_nf_mem_any(dw)) { + u32 dw2; + + pci_read_config_dword(priv->mc, I5100_NERR_NF_MEM, &dw2); + if (dw2) + pci_write_config_dword(priv->mc, I5100_NERR_NF_MEM, + dw2); + pci_write_config_dword(priv->mc, I5100_FERR_NF_MEM, dw); + + i5100_read_log(mci, i5100_ferr_nf_mem_chan_indx(dw), + i5100_ferr_nf_mem_any(dw), + i5100_nerr_nf_mem_any(dw2)); + } +} + +static struct pci_dev *pci_get_device_func(unsigned vendor, + unsigned device, + unsigned func) +{ + struct pci_dev *ret = NULL; + + while (1) { + ret = pci_get_device(vendor, device, ret); + + if (!ret) + break; + + if (PCI_FUNC(ret->devfn) == func) + break; + } + + return ret; +} + +static unsigned long __devinit i5100_npages(struct mem_ctl_info *mci, + int csrow) +{ + struct i5100_priv *priv = mci->pvt_info; + const unsigned ctlr_rank = i5100_csrow_to_rank(mci, csrow); + const unsigned ctlr = i5100_csrow_to_cntlr(mci, csrow); + unsigned addr_lines; + + /* dimm present? */ + if (!priv->mtr[ctlr][ctlr_rank].present) + return 0ULL; + + addr_lines = + I5100_DIMM_ADDR_LINES + + priv->mtr[ctlr][ctlr_rank].numcol + + priv->mtr[ctlr][ctlr_rank].numrow + + priv->mtr[ctlr][ctlr_rank].numbank; + + return (unsigned long) + ((unsigned long long) (1ULL << addr_lines) / PAGE_SIZE); +} + +static void __devinit i5100_init_mtr(struct mem_ctl_info *mci) +{ + struct i5100_priv *priv = mci->pvt_info; + struct pci_dev *mms[2] = { priv->ch0mm, priv->ch1mm }; + int i; + + for (i = 0; i < I5100_MAX_CTLRS; i++) { + int j; + struct pci_dev *pdev = mms[i]; + + for (j = 0; j < I5100_MAX_RANKS_PER_CTLR; j++) { + const unsigned addr = + (j < 4) ? I5100_MTR_0 + j * 2 : + I5100_MTR_4 + (j - 4) * 2; + u16 w; + + pci_read_config_word(pdev, addr, &w); + + priv->mtr[i][j].present = i5100_mtr_present(w); + priv->mtr[i][j].ethrottle = i5100_mtr_ethrottle(w); + priv->mtr[i][j].width = 4 + 4 * i5100_mtr_width(w); + priv->mtr[i][j].numbank = 2 + i5100_mtr_numbank(w); + priv->mtr[i][j].numrow = 13 + i5100_mtr_numrow(w); + priv->mtr[i][j].numcol = 10 + i5100_mtr_numcol(w); + } + } +} + +/* + * FIXME: make this into a real i2c adapter (so that dimm-decode + * will work)? + */ +static int i5100_read_spd_byte(const struct mem_ctl_info *mci, + u8 ch, u8 slot, u8 addr, u8 *byte) +{ + struct i5100_priv *priv = mci->pvt_info; + u16 w; + unsigned long et; + + pci_read_config_word(priv->mc, I5100_SPDDATA, &w); + if (i5100_spddata_busy(w)) + return -1; + + pci_write_config_dword(priv->mc, I5100_SPDCMD, + i5100_spdcmd_create(0xa, 1, ch * 4 + slot, addr, + 0, 0)); + + /* wait up to 100ms */ + et = jiffies + HZ / 10; + udelay(100); + while (1) { + pci_read_config_word(priv->mc, I5100_SPDDATA, &w); + if (!i5100_spddata_busy(w)) + break; + udelay(100); + } + + if (!i5100_spddata_rdo(w) || i5100_spddata_sbe(w)) + return -1; + + *byte = i5100_spddata_data(w); + + return 0; +} + +/* + * fill dimm chip select map + * + * FIXME: + * o only valid for 4 ranks per controller + * o not the only way to may chip selects to dimm slots + * o investigate if there is some way to obtain this map from the bios + */ +static void __devinit i5100_init_dimm_csmap(struct mem_ctl_info *mci) +{ + struct i5100_priv *priv = mci->pvt_info; + int i; + + WARN_ON(priv->ranksperctlr != 4); + + for (i = 0; i < I5100_MAX_DIMM_SLOTS_PER_CTLR; i++) { + int j; + + for (j = 0; j < I5100_MAX_RANKS_PER_DIMM; j++) + priv->dimm_csmap[i][j] = -1; /* default NC */ + } + + /* only 2 chip selects per slot... */ + priv->dimm_csmap[0][0] = 0; + priv->dimm_csmap[0][1] = 3; + priv->dimm_csmap[1][0] = 1; + priv->dimm_csmap[1][1] = 2; + priv->dimm_csmap[2][0] = 2; + priv->dimm_csmap[3][0] = 3; +} + +static void __devinit i5100_init_dimm_layout(struct pci_dev *pdev, + struct mem_ctl_info *mci) +{ + struct i5100_priv *priv = mci->pvt_info; + int i; + + for (i = 0; i < I5100_MAX_CTLRS; i++) { + int j; + + for (j = 0; j < I5100_MAX_DIMM_SLOTS_PER_CTLR; j++) { + u8 rank; + + if (i5100_read_spd_byte(mci, i, j, 5, &rank) < 0) + priv->dimm_numrank[i][j] = 0; + else + priv->dimm_numrank[i][j] = (rank & 3) + 1; + } + } + + i5100_init_dimm_csmap(mci); +} + +static void __devinit i5100_init_interleaving(struct pci_dev *pdev, + struct mem_ctl_info *mci) +{ + u16 w; + u32 dw; + struct i5100_priv *priv = mci->pvt_info; + struct pci_dev *mms[2] = { priv->ch0mm, priv->ch1mm }; + int i; + + pci_read_config_word(pdev, I5100_TOLM, &w); + priv->tolm = (u64) i5100_tolm_tolm(w) * 256 * 1024 * 1024; + + pci_read_config_word(pdev, I5100_MIR0, &w); + priv->mir[0].limit = (u64) i5100_mir_limit(w) << 28; + priv->mir[0].way[1] = i5100_mir_way1(w); + priv->mir[0].way[0] = i5100_mir_way0(w); + + pci_read_config_word(pdev, I5100_MIR1, &w); + priv->mir[1].limit = (u64) i5100_mir_limit(w) << 28; + priv->mir[1].way[1] = i5100_mir_way1(w); + priv->mir[1].way[0] = i5100_mir_way0(w); + + pci_read_config_word(pdev, I5100_AMIR_0, &w); + priv->amir[0] = w; + pci_read_config_word(pdev, I5100_AMIR_1, &w); + priv->amir[1] = w; + + for (i = 0; i < I5100_MAX_CTLRS; i++) { + int j; + + for (j = 0; j < 5; j++) { + int k; + + pci_read_config_dword(mms[i], I5100_DMIR + j * 4, &dw); + + priv->dmir[i][j].limit = + (u64) i5100_dmir_limit(dw) << 28; + for (k = 0; k < I5100_MAX_RANKS_PER_DIMM; k++) + priv->dmir[i][j].rank[k] = + i5100_dmir_rank(dw, k); + } + } + + i5100_init_mtr(mci); +} + +static void __devinit i5100_init_csrows(struct mem_ctl_info *mci) +{ + int i; + unsigned long total_pages = 0UL; + struct i5100_priv *priv = mci->pvt_info; + + for (i = 0; i < mci->nr_csrows; i++) { + const unsigned long npages = i5100_npages(mci, i); + const unsigned cntlr = i5100_csrow_to_cntlr(mci, i); + const unsigned rank = i5100_csrow_to_rank(mci, i); + + if (!npages) + continue; + + /* + * FIXME: these two are totally bogus -- I don't see how to + * map them correctly to this structure... + */ + mci->csrows[i].first_page = total_pages; + mci->csrows[i].last_page = total_pages + npages - 1; + mci->csrows[i].page_mask = 0UL; + + mci->csrows[i].nr_pages = npages; + mci->csrows[i].grain = 32; + mci->csrows[i].csrow_idx = i; + mci->csrows[i].dtype = + (priv->mtr[cntlr][rank].width == 4) ? DEV_X4 : DEV_X8; + mci->csrows[i].ue_count = 0; + mci->csrows[i].ce_count = 0; + mci->csrows[i].mtype = MEM_RDDR2; + mci->csrows[i].edac_mode = EDAC_SECDED; + mci->csrows[i].mci = mci; + mci->csrows[i].nr_channels = 1; + mci->csrows[i].channels[0].chan_idx = 0; + mci->csrows[i].channels[0].ce_count = 0; + mci->csrows[i].channels[0].csrow = mci->csrows + i; + snprintf(mci->csrows[i].channels[0].label, + sizeof(mci->csrows[i].channels[0].label), + "DIMM%u", i5100_rank_to_slot(mci, cntlr, rank)); + + total_pages += npages; + } +} + +static int __devinit i5100_init_one(struct pci_dev *pdev, + const struct pci_device_id *id) +{ + int rc; + struct mem_ctl_info *mci; + struct i5100_priv *priv; + struct pci_dev *ch0mm, *ch1mm; + int ret = 0; + u32 dw; + int ranksperch; + + if (PCI_FUNC(pdev->devfn) != 1) + return -ENODEV; + + rc = pci_enable_device(pdev); + if (rc < 0) { + ret = rc; + goto bail; + } + + /* ECC enabled? */ + pci_read_config_dword(pdev, I5100_MC, &dw); + if (!i5100_mc_errdeten(dw)) { + printk(KERN_INFO "i5100_edac: ECC not enabled.\n"); + ret = -ENODEV; + goto bail_pdev; + } + + /* figure out how many ranks, from strapped state of 48GB_Mode input */ + pci_read_config_dword(pdev, I5100_MS, &dw); + ranksperch = !!(dw & (1 << 8)) * 2 + 4; + + if (ranksperch != 4) { + /* FIXME: get 6 ranks / controller to work - need hw... */ + printk(KERN_INFO "i5100_edac: unsupported configuration.\n"); + ret = -ENODEV; + goto bail_pdev; + } + + /* enable error reporting... */ + pci_read_config_dword(pdev, I5100_EMASK_MEM, &dw); + dw &= ~I5100_FERR_NF_MEM_ANY_MASK; + pci_write_config_dword(pdev, I5100_EMASK_MEM, dw); + + /* device 21, func 0, Channel 0 Memory Map, Error Flag/Mask, etc... */ + ch0mm = pci_get_device_func(PCI_VENDOR_ID_INTEL, + PCI_DEVICE_ID_INTEL_5100_21, 0); + if (!ch0mm) { + ret = -ENODEV; + goto bail_pdev; + } + + rc = pci_enable_device(ch0mm); + if (rc < 0) { + ret = rc; + goto bail_ch0; + } + + /* device 22, func 0, Channel 1 Memory Map, Error Flag/Mask, etc... */ + ch1mm = pci_get_device_func(PCI_VENDOR_ID_INTEL, + PCI_DEVICE_ID_INTEL_5100_22, 0); + if (!ch1mm) { + ret = -ENODEV; + goto bail_disable_ch0; + } + + rc = pci_enable_device(ch1mm); + if (rc < 0) { + ret = rc; + goto bail_ch1; + } + + mci = edac_mc_alloc(sizeof(*priv), ranksperch * 2, 1, 0); + if (!mci) { + ret = -ENOMEM; + goto bail_disable_ch1; + } + + mci->dev = &pdev->dev; + + priv = mci->pvt_info; + priv->ranksperctlr = ranksperch; + priv->mc = pdev; + priv->ch0mm = ch0mm; + priv->ch1mm = ch1mm; + + i5100_init_dimm_layout(pdev, mci); + i5100_init_interleaving(pdev, mci); + + mci->mtype_cap = MEM_FLAG_FB_DDR2; + mci->edac_ctl_cap = EDAC_FLAG_SECDED; + mci->edac_cap = EDAC_FLAG_SECDED; + mci->mod_name = "i5100_edac.c"; + mci->mod_ver = "not versioned"; + mci->ctl_name = "i5100"; + mci->dev_name = pci_name(pdev); + mci->ctl_page_to_phys = NULL; + + mci->edac_check = i5100_check_error; + + i5100_init_csrows(mci); + + /* this strange construction seems to be in every driver, dunno why */ + switch (edac_op_state) { + case EDAC_OPSTATE_POLL: + case EDAC_OPSTATE_NMI: + break; + default: + edac_op_state = EDAC_OPSTATE_POLL; + break; + } + + if (edac_mc_add_mc(mci)) { + ret = -ENODEV; + goto bail_mc; + } + + return ret; + +bail_mc: + edac_mc_free(mci); + +bail_disable_ch1: + pci_disable_device(ch1mm); + +bail_ch1: + pci_dev_put(ch1mm); + +bail_disable_ch0: + pci_disable_device(ch0mm); + +bail_ch0: + pci_dev_put(ch0mm); + +bail_pdev: + pci_disable_device(pdev); + +bail: + return ret; +} + +static void __devexit i5100_remove_one(struct pci_dev *pdev) +{ + struct mem_ctl_info *mci; + struct i5100_priv *priv; + + mci = edac_mc_del_mc(&pdev->dev); + + if (!mci) + return; + + priv = mci->pvt_info; + pci_disable_device(pdev); + pci_disable_device(priv->ch0mm); + pci_disable_device(priv->ch1mm); + pci_dev_put(priv->ch0mm); + pci_dev_put(priv->ch1mm); + + edac_mc_free(mci); +} + +static const struct pci_device_id i5100_pci_tbl[] __devinitdata = { + /* Device 16, Function 0, Channel 0 Memory Map, Error Flag/Mask, ... */ + { PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_5100_16) }, + { 0, } +}; +MODULE_DEVICE_TABLE(pci, i5100_pci_tbl); + +static struct pci_driver i5100_driver = { + .name = KBUILD_BASENAME, + .probe = i5100_init_one, + .remove = __devexit_p(i5100_remove_one), + .id_table = i5100_pci_tbl, +}; + +static int __init i5100_init(void) +{ + int pci_rc; + + pci_rc = pci_register_driver(&i5100_driver); + + return (pci_rc < 0) ? pci_rc : 0; +} + +static void __exit i5100_exit(void) +{ + pci_unregister_driver(&i5100_driver); +} + +module_init(i5100_init); +module_exit(i5100_exit); + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR + ("Arthur Jones <ajones@riverbed.com>"); +MODULE_DESCRIPTION("MC Driver for Intel I5100 memory controllers"); diff --git a/drivers/edac/mpc85xx_edac.c b/drivers/edac/mpc85xx_edac.c index 065732d..2265d9c 100644 --- a/drivers/edac/mpc85xx_edac.c +++ b/drivers/edac/mpc85xx_edac.c @@ -20,7 +20,6 @@ #include <linux/of_platform.h> #include <linux/of_device.h> -#include <asm/mpc85xx.h> #include "edac_module.h" #include "edac_core.h" #include "mpc85xx_edac.h" @@ -43,8 +42,6 @@ static u32 orig_pci_err_en; static u32 orig_l2_err_disable; static u32 orig_hid1; -static const char *mpc85xx_ctl_name = "MPC85xx"; - /************************ MC SYSFS parts ***********************************/ static ssize_t mpc85xx_mc_inject_data_hi_show(struct mem_ctl_info *mci, @@ -198,14 +195,15 @@ static irqreturn_t mpc85xx_pci_isr(int irq, void *dev_id) return IRQ_HANDLED; } -static int __devinit mpc85xx_pci_err_probe(struct platform_device *pdev) +static int __devinit mpc85xx_pci_err_probe(struct of_device *op, + const struct of_device_id *match) { struct edac_pci_ctl_info *pci; struct mpc85xx_pci_pdata *pdata; - struct resource *r; + struct resource r; int res = 0; - if (!devres_open_group(&pdev->dev, mpc85xx_pci_err_probe, GFP_KERNEL)) + if (!devres_open_group(&op->dev, mpc85xx_pci_err_probe, GFP_KERNEL)) return -ENOMEM; pci = edac_pci_alloc_ctl_info(sizeof(*pdata), "mpc85xx_pci_err"); @@ -215,34 +213,37 @@ static int __devinit mpc85xx_pci_err_probe(struct platform_device *pdev) pdata = pci->pvt_info; pdata->name = "mpc85xx_pci_err"; pdata->irq = NO_IRQ; - platform_set_drvdata(pdev, pci); - pci->dev = &pdev->dev; + dev_set_drvdata(&op->dev, pci); + pci->dev = &op->dev; pci->mod_name = EDAC_MOD_STR; pci->ctl_name = pdata->name; - pci->dev_name = pdev->dev.bus_id; + pci->dev_name = op->dev.bus_id; if (edac_op_state == EDAC_OPSTATE_POLL) pci->edac_check = mpc85xx_pci_check; pdata->edac_idx = edac_pci_idx++; - r = platform_get_resource(pdev, IORESOURCE_MEM, 0); - if (!r) { + res = of_address_to_resource(op->node, 0, &r); + if (res) { printk(KERN_ERR "%s: Unable to get resource for " "PCI err regs\n", __func__); goto err; } - if (!devm_request_mem_region(&pdev->dev, r->start, - r->end - r->start + 1, pdata->name)) { + /* we only need the error registers */ + r.start += 0xe00; + + if (!devm_request_mem_region(&op->dev, r.start, + r.end - r.start + 1, pdata->name)) { printk(KERN_ERR "%s: Error while requesting mem region\n", __func__); res = -EBUSY; goto err; } - pdata->pci_vbase = devm_ioremap(&pdev->dev, r->start, - r->end - r->start + 1); + pdata->pci_vbase = devm_ioremap(&op->dev, r.start, + r.end - r.start + 1); if (!pdata->pci_vbase) { printk(KERN_ERR "%s: Unable to setup PCI err regs\n", __func__); res = -ENOMEM; @@ -269,14 +270,15 @@ static int __devinit mpc85xx_pci_err_probe(struct platform_device *pdev) } if (edac_op_state == EDAC_OPSTATE_INT) { - pdata->irq = platform_get_irq(pdev, 0); - res = devm_request_irq(&pdev->dev, pdata->irq, + pdata->irq = irq_of_parse_and_map(op->node, 0); + res = devm_request_irq(&op->dev, pdata->irq, mpc85xx_pci_isr, IRQF_DISABLED, "[EDAC] PCI err", pci); if (res < 0) { printk(KERN_ERR "%s: Unable to requiest irq %d for " "MPC85xx PCI err\n", __func__, pdata->irq); + irq_dispose_mapping(pdata->irq); res = -ENODEV; goto err2; } @@ -285,23 +287,23 @@ static int __devinit mpc85xx_pci_err_probe(struct platform_device *pdev) pdata->irq); } - devres_remove_group(&pdev->dev, mpc85xx_pci_err_probe); + devres_remove_group(&op->dev, mpc85xx_pci_err_probe); debugf3("%s(): success\n", __func__); printk(KERN_INFO EDAC_MOD_STR " PCI err registered\n"); return 0; err2: - edac_pci_del_device(&pdev->dev); + edac_pci_del_device(&op->dev); err: edac_pci_free_ctl_info(pci); - devres_release_group(&pdev->dev, mpc85xx_pci_err_probe); + devres_release_group(&op->dev, mpc85xx_pci_err_probe); return res; } -static int mpc85xx_pci_err_remove(struct platform_device *pdev) +static int mpc85xx_pci_err_remove(struct of_device *op) { - struct edac_pci_ctl_info *pci = platform_get_drvdata(pdev); + struct edac_pci_ctl_info *pci = dev_get_drvdata(&op->dev); struct mpc85xx_pci_pdata *pdata = pci->pvt_info; debugf0("%s()\n", __func__); @@ -321,12 +323,26 @@ static int mpc85xx_pci_err_remove(struct platform_device *pdev) return 0; } -static struct platform_driver mpc85xx_pci_err_driver = { +static struct of_device_id mpc85xx_pci_err_of_match[] = { + { + .compatible = "fsl,mpc8540-pcix", + }, + { + .compatible = "fsl,mpc8540-pci", + }, + {}, +}; + +static struct of_platform_driver mpc85xx_pci_err_driver = { + .owner = THIS_MODULE, + .name = "mpc85xx_pci_err", + .match_table = mpc85xx_pci_err_of_match, .probe = mpc85xx_pci_err_probe, .remove = __devexit_p(mpc85xx_pci_err_remove), .driver = { - .name = "mpc85xx_pci_err", - } + .name = "mpc85xx_pci_err", + .owner = THIS_MODULE, + }, }; #endif /* CONFIG_PCI */ @@ -1005,7 +1021,7 @@ static int __init mpc85xx_mc_init(void) printk(KERN_WARNING EDAC_MOD_STR "L2 fails to register\n"); #ifdef CONFIG_PCI - res = platform_driver_register(&mpc85xx_pci_err_driver); + res = of_register_platform_driver(&mpc85xx_pci_err_driver); if (res) printk(KERN_WARNING EDAC_MOD_STR "PCI fails to register\n"); #endif @@ -1028,7 +1044,7 @@ static void __exit mpc85xx_mc_exit(void) { mtspr(SPRN_HID1, orig_hid1); #ifdef CONFIG_PCI - platform_driver_unregister(&mpc85xx_pci_err_driver); + of_unregister_platform_driver(&mpc85xx_pci_err_driver); #endif of_unregister_platform_driver(&mpc85xx_l2_err_driver); of_unregister_platform_driver(&mpc85xx_mc_err_driver); diff --git a/drivers/edac/mv64x60_edac.c b/drivers/edac/mv64x60_edac.c index bf071f1..083ce8d 100644 --- a/drivers/edac/mv64x60_edac.c +++ b/drivers/edac/mv64x60_edac.c @@ -71,6 +71,35 @@ static irqreturn_t mv64x60_pci_isr(int irq, void *dev_id) return IRQ_HANDLED; } +/* + * Bit 0 of MV64x60_PCIx_ERR_MASK does not exist on the 64360 and because of + * errata FEr-#11 and FEr-##16 for the 64460, it should be 0 on that chip as + * well. IOW, don't set bit 0. + */ + +/* Erratum FEr PCI-#16: clear bit 0 of PCI SERRn Mask reg. */ +static int __init mv64x60_pci_fixup(struct platform_device *pdev) +{ + struct resource *r; + void __iomem *pci_serr; + + r = platform_get_resource(pdev, IORESOURCE_MEM, 1); + if (!r) { + printk(KERN_ERR "%s: Unable to get resource for " + "PCI err regs\n", __func__); + return -ENOENT; + } + + pci_serr = ioremap(r->start, r->end - r->start + 1); + if (!pci_serr) + return -ENOMEM; + + out_le32(pci_serr, in_le32(pci_serr) & ~0x1); + iounmap(pci_serr); + + return 0; +} + static int __devinit mv64x60_pci_err_probe(struct platform_device *pdev) { struct edac_pci_ctl_info *pci; @@ -128,6 +157,12 @@ static int __devinit mv64x60_pci_err_probe(struct platform_device *pdev) goto err; } + res = mv64x60_pci_fixup(pdev); + if (res < 0) { + printk(KERN_ERR "%s: PCI fixup failed\n", __func__); + goto err; + } + out_le32(pdata->pci_vbase + MV64X60_PCI_ERROR_CAUSE, 0); out_le32(pdata->pci_vbase + MV64X60_PCI_ERROR_MASK, 0); out_le32(pdata->pci_vbase + MV64X60_PCI_ERROR_MASK, @@ -612,7 +647,7 @@ static void get_total_mem(struct mv64x60_mc_pdata *pdata) if (!np) return; - reg = get_property(np, "reg", NULL); + reg = of_get_property(np, "reg", NULL); pdata->total_mem = reg[1]; } |