diff options
Diffstat (limited to 'drivers/edac')
-rw-r--r-- | drivers/edac/amd64_edac.c | 38 | ||||
-rw-r--r-- | drivers/edac/amd64_edac.h | 3 | ||||
-rw-r--r-- | drivers/edac/amd8111_edac.c | 44 | ||||
-rw-r--r-- | drivers/edac/e752x_edac.c | 30 | ||||
-rw-r--r-- | drivers/edac/i3200_edac.c | 2 | ||||
-rw-r--r-- | drivers/edac/i5100_edac.c | 17 | ||||
-rw-r--r-- | drivers/edac/i5400_edac.c | 2 | ||||
-rw-r--r-- | drivers/edac/i7core_edac.c | 10 | ||||
-rw-r--r-- | drivers/edac/i82875p_edac.c | 2 | ||||
-rw-r--r-- | drivers/edac/mce_amd.c | 65 | ||||
-rw-r--r-- | drivers/edac/mpc85xx_edac.c | 6 | ||||
-rw-r--r-- | drivers/edac/octeon_edac-lmc.c | 179 | ||||
-rw-r--r-- | drivers/edac/sb_edac.c | 27 |
13 files changed, 319 insertions, 106 deletions
diff --git a/drivers/edac/amd64_edac.c b/drivers/edac/amd64_edac.c index 98e14ee4833c..f8bf00010d45 100644 --- a/drivers/edac/amd64_edac.c +++ b/drivers/edac/amd64_edac.c @@ -1239,9 +1239,17 @@ static u8 f15_m30h_determine_channel(struct amd64_pvt *pvt, u64 sys_addr, if (num_dcts_intlv == 2) { select = (sys_addr >> 8) & 0x3; channel = select ? 0x3 : 0; - } else if (num_dcts_intlv == 4) - channel = (sys_addr >> 8) & 0x7; - + } else if (num_dcts_intlv == 4) { + u8 intlv_addr = dct_sel_interleave_addr(pvt); + switch (intlv_addr) { + case 0x4: + channel = (sys_addr >> 8) & 0x3; + break; + case 0x5: + channel = (sys_addr >> 9) & 0x3; + break; + } + } return channel; } @@ -1799,6 +1807,17 @@ static struct amd64_family_type family_types[] = { .read_dct_pci_cfg = f10_read_dct_pci_cfg, } }, + [F16_M30H_CPUS] = { + .ctl_name = "F16h_M30h", + .f1_id = PCI_DEVICE_ID_AMD_16H_M30H_NB_F1, + .f3_id = PCI_DEVICE_ID_AMD_16H_M30H_NB_F3, + .ops = { + .early_channel_count = f1x_early_channel_count, + .map_sysaddr_to_csrow = f1x_map_sysaddr_to_csrow, + .dbam_to_cs = f16_dbam_to_chip_select, + .read_dct_pci_cfg = f10_read_dct_pci_cfg, + } + }, }; /* @@ -2578,6 +2597,11 @@ static struct amd64_family_type *per_family_init(struct amd64_pvt *pvt) break; case 0x16: + if (pvt->model == 0x30) { + fam_type = &family_types[F16_M30H_CPUS]; + pvt->ops = &family_types[F16_M30H_CPUS].ops; + break; + } fam_type = &family_types[F16_CPUS]; pvt->ops = &family_types[F16_CPUS].ops; break; @@ -2830,6 +2854,14 @@ static const struct pci_device_id amd64_pci_table[] = { .class = 0, .class_mask = 0, }, + { + .vendor = PCI_VENDOR_ID_AMD, + .device = PCI_DEVICE_ID_AMD_16H_M30H_NB_F2, + .subvendor = PCI_ANY_ID, + .subdevice = PCI_ANY_ID, + .class = 0, + .class_mask = 0, + }, {0, } }; diff --git a/drivers/edac/amd64_edac.h b/drivers/edac/amd64_edac.h index 6dc1fcc25afb..d903e0c21144 100644 --- a/drivers/edac/amd64_edac.h +++ b/drivers/edac/amd64_edac.h @@ -168,6 +168,8 @@ #define PCI_DEVICE_ID_AMD_15H_NB_F2 0x1602 #define PCI_DEVICE_ID_AMD_16H_NB_F1 0x1531 #define PCI_DEVICE_ID_AMD_16H_NB_F2 0x1532 +#define PCI_DEVICE_ID_AMD_16H_M30H_NB_F1 0x1581 +#define PCI_DEVICE_ID_AMD_16H_M30H_NB_F2 0x1582 /* * Function 1 - Address Map @@ -300,6 +302,7 @@ enum amd_families { F15_CPUS, F15_M30H_CPUS, F16_CPUS, + F16_M30H_CPUS, NUM_FAMILIES, }; diff --git a/drivers/edac/amd8111_edac.c b/drivers/edac/amd8111_edac.c index ddd890052ce2..2b63f7c2d6d2 100644 --- a/drivers/edac/amd8111_edac.c +++ b/drivers/edac/amd8111_edac.c @@ -350,6 +350,7 @@ static int amd8111_dev_probe(struct pci_dev *dev, const struct pci_device_id *id) { struct amd8111_dev_info *dev_info = &amd8111_devices[id->driver_data]; + int ret = -ENODEV; dev_info->dev = pci_get_device(PCI_VENDOR_ID_AMD, dev_info->err_dev, NULL); @@ -359,16 +360,15 @@ static int amd8111_dev_probe(struct pci_dev *dev, "vendor %x, device %x, name %s\n", PCI_VENDOR_ID_AMD, dev_info->err_dev, dev_info->ctl_name); - return -ENODEV; + goto err; } if (pci_enable_device(dev_info->dev)) { - pci_dev_put(dev_info->dev); printk(KERN_ERR "failed to enable:" "vendor %x, device %x, name %s\n", PCI_VENDOR_ID_AMD, dev_info->err_dev, dev_info->ctl_name); - return -ENODEV; + goto err_dev_put; } /* @@ -381,8 +381,10 @@ static int amd8111_dev_probe(struct pci_dev *dev, edac_device_alloc_ctl_info(0, dev_info->ctl_name, 1, NULL, 0, 0, NULL, 0, dev_info->edac_idx); - if (!dev_info->edac_dev) - return -ENOMEM; + if (!dev_info->edac_dev) { + ret = -ENOMEM; + goto err_dev_put; + } dev_info->edac_dev->pvt_info = dev_info; dev_info->edac_dev->dev = &dev_info->dev->dev; @@ -399,8 +401,7 @@ static int amd8111_dev_probe(struct pci_dev *dev, if (edac_device_add_device(dev_info->edac_dev) > 0) { printk(KERN_ERR "failed to add edac_dev for %s\n", dev_info->ctl_name); - edac_device_free_ctl_info(dev_info->edac_dev); - return -ENODEV; + goto err_edac_free_ctl; } printk(KERN_INFO "added one edac_dev on AMD8111 " @@ -409,6 +410,13 @@ static int amd8111_dev_probe(struct pci_dev *dev, dev_info->ctl_name); return 0; + +err_edac_free_ctl: + edac_device_free_ctl_info(dev_info->edac_dev); +err_dev_put: + pci_dev_put(dev_info->dev); +err: + return ret; } static void amd8111_dev_remove(struct pci_dev *dev) @@ -437,6 +445,7 @@ static int amd8111_pci_probe(struct pci_dev *dev, const struct pci_device_id *id) { struct amd8111_pci_info *pci_info = &amd8111_pcis[id->driver_data]; + int ret = -ENODEV; pci_info->dev = pci_get_device(PCI_VENDOR_ID_AMD, pci_info->err_dev, NULL); @@ -446,16 +455,15 @@ static int amd8111_pci_probe(struct pci_dev *dev, "vendor %x, device %x, name %s\n", PCI_VENDOR_ID_AMD, pci_info->err_dev, pci_info->ctl_name); - return -ENODEV; + goto err; } if (pci_enable_device(pci_info->dev)) { - pci_dev_put(pci_info->dev); printk(KERN_ERR "failed to enable:" "vendor %x, device %x, name %s\n", PCI_VENDOR_ID_AMD, pci_info->err_dev, pci_info->ctl_name); - return -ENODEV; + goto err_dev_put; } /* @@ -465,8 +473,10 @@ static int amd8111_pci_probe(struct pci_dev *dev, */ pci_info->edac_idx = edac_pci_alloc_index(); pci_info->edac_dev = edac_pci_alloc_ctl_info(0, pci_info->ctl_name); - if (!pci_info->edac_dev) - return -ENOMEM; + if (!pci_info->edac_dev) { + ret = -ENOMEM; + goto err_dev_put; + } pci_info->edac_dev->pvt_info = pci_info; pci_info->edac_dev->dev = &pci_info->dev->dev; @@ -483,8 +493,7 @@ static int amd8111_pci_probe(struct pci_dev *dev, if (edac_pci_add_device(pci_info->edac_dev, pci_info->edac_idx) > 0) { printk(KERN_ERR "failed to add edac_pci for %s\n", pci_info->ctl_name); - edac_pci_free_ctl_info(pci_info->edac_dev); - return -ENODEV; + goto err_edac_free_ctl; } printk(KERN_INFO "added one edac_pci on AMD8111 " @@ -493,6 +502,13 @@ static int amd8111_pci_probe(struct pci_dev *dev, pci_info->ctl_name); return 0; + +err_edac_free_ctl: + edac_pci_free_ctl_info(pci_info->edac_dev); +err_dev_put: + pci_dev_put(pci_info->dev); +err: + return ret; } static void amd8111_pci_remove(struct pci_dev *dev) diff --git a/drivers/edac/e752x_edac.c b/drivers/edac/e752x_edac.c index 92d54fa65f93..b2d71388172b 100644 --- a/drivers/edac/e752x_edac.c +++ b/drivers/edac/e752x_edac.c @@ -209,7 +209,6 @@ enum e752x_chips { */ struct e752x_pvt { - struct pci_dev *bridge_ck; struct pci_dev *dev_d0f0; struct pci_dev *dev_d0f1; u32 tolm; @@ -891,7 +890,7 @@ static void e752x_get_error_info(struct mem_ctl_info *mci, info->buf_ferr); if (info->dram_ferr) - pci_write_bits16(pvt->bridge_ck, E752X_DRAM_FERR, + pci_write_bits16(pvt->dev_d0f1, E752X_DRAM_FERR, info->dram_ferr, info->dram_ferr); pci_write_config_dword(dev, E752X_FERR_GLOBAL, @@ -936,7 +935,7 @@ static void e752x_get_error_info(struct mem_ctl_info *mci, info->buf_nerr); if (info->dram_nerr) - pci_write_bits16(pvt->bridge_ck, E752X_DRAM_NERR, + pci_write_bits16(pvt->dev_d0f1, E752X_DRAM_NERR, info->dram_nerr, info->dram_nerr); pci_write_config_dword(dev, E752X_NERR_GLOBAL, @@ -1177,38 +1176,33 @@ static void e752x_init_mem_map_table(struct pci_dev *pdev, static int e752x_get_devs(struct pci_dev *pdev, int dev_idx, struct e752x_pvt *pvt) { - struct pci_dev *dev; - - pvt->bridge_ck = pci_get_device(PCI_VENDOR_ID_INTEL, - pvt->dev_info->err_dev, pvt->bridge_ck); + pvt->dev_d0f1 = pci_get_device(PCI_VENDOR_ID_INTEL, + pvt->dev_info->err_dev, NULL); - if (pvt->bridge_ck == NULL) { - pvt->bridge_ck = pci_scan_single_device(pdev->bus, + if (pvt->dev_d0f1 == NULL) { + pvt->dev_d0f1 = pci_scan_single_device(pdev->bus, PCI_DEVFN(0, 1)); - pci_dev_get(pvt->bridge_ck); + pci_dev_get(pvt->dev_d0f1); } - if (pvt->bridge_ck == NULL) { + if (pvt->dev_d0f1 == NULL) { e752x_printk(KERN_ERR, "error reporting device not found:" "vendor %x device 0x%x (broken BIOS?)\n", PCI_VENDOR_ID_INTEL, e752x_devs[dev_idx].err_dev); return 1; } - dev = pci_get_device(PCI_VENDOR_ID_INTEL, + pvt->dev_d0f0 = pci_get_device(PCI_VENDOR_ID_INTEL, e752x_devs[dev_idx].ctl_dev, NULL); - if (dev == NULL) + if (pvt->dev_d0f0 == NULL) goto fail; - pvt->dev_d0f0 = dev; - pvt->dev_d0f1 = pci_dev_get(pvt->bridge_ck); - return 0; fail: - pci_dev_put(pvt->bridge_ck); + pci_dev_put(pvt->dev_d0f1); return 1; } @@ -1385,7 +1379,6 @@ static int e752x_probe1(struct pci_dev *pdev, int dev_idx) fail: pci_dev_put(pvt->dev_d0f0); pci_dev_put(pvt->dev_d0f1); - pci_dev_put(pvt->bridge_ck); edac_mc_free(mci); return -ENODEV; @@ -1419,7 +1412,6 @@ static void e752x_remove_one(struct pci_dev *pdev) pvt = (struct e752x_pvt *)mci->pvt_info; pci_dev_put(pvt->dev_d0f0); pci_dev_put(pvt->dev_d0f1); - pci_dev_put(pvt->bridge_ck); edac_mc_free(mci); } diff --git a/drivers/edac/i3200_edac.c b/drivers/edac/i3200_edac.c index fa1326e5a4b0..022a70273ada 100644 --- a/drivers/edac/i3200_edac.c +++ b/drivers/edac/i3200_edac.c @@ -464,6 +464,8 @@ static void i3200_remove_one(struct pci_dev *pdev) iounmap(priv->window); edac_mc_free(mci); + + pci_disable_device(pdev); } static const struct pci_device_id i3200_pci_tbl[] = { diff --git a/drivers/edac/i5100_edac.c b/drivers/edac/i5100_edac.c index 36a38ee94fa8..6247d186177e 100644 --- a/drivers/edac/i5100_edac.c +++ b/drivers/edac/i5100_edac.c @@ -869,16 +869,13 @@ static void i5100_init_csrows(struct mem_ctl_info *mci) chan, rank, 0); dimm->nr_pages = npages; - if (npages) { - dimm->grain = 32; - dimm->dtype = (priv->mtr[chan][rank].width == 4) ? - DEV_X4 : DEV_X8; - dimm->mtype = MEM_RDDR2; - dimm->edac_mode = EDAC_SECDED; - snprintf(dimm->label, sizeof(dimm->label), - "DIMM%u", - i5100_rank_to_slot(mci, chan, rank)); - } + dimm->grain = 32; + dimm->dtype = (priv->mtr[chan][rank].width == 4) ? + DEV_X4 : DEV_X8; + dimm->mtype = MEM_RDDR2; + dimm->edac_mode = EDAC_SECDED; + snprintf(dimm->label, sizeof(dimm->label), "DIMM%u", + i5100_rank_to_slot(mci, chan, rank)); edac_dbg(2, "dimm channel %d, rank %d, size %ld\n", chan, rank, (long)PAGES_TO_MiB(npages)); diff --git a/drivers/edac/i5400_edac.c b/drivers/edac/i5400_edac.c index f189c333f406..6ef6ad1ba16e 100644 --- a/drivers/edac/i5400_edac.c +++ b/drivers/edac/i5400_edac.c @@ -1408,6 +1408,8 @@ static void i5400_remove_one(struct pci_dev *pdev) /* retrieve references to resources, and free those resources */ i5400_put_devices(mci); + pci_disable_device(pdev); + edac_mc_free(mci); } diff --git a/drivers/edac/i7core_edac.c b/drivers/edac/i7core_edac.c index ab127cf5c798..9cd0b301f81b 100644 --- a/drivers/edac/i7core_edac.c +++ b/drivers/edac/i7core_edac.c @@ -1708,7 +1708,7 @@ static void i7core_mce_output_error(struct mem_ctl_info *mci, const struct mce *m) { struct i7core_pvt *pvt = mci->pvt_info; - char *type, *optype, *err; + char *optype, *err; enum hw_event_mc_err_type tp_event; unsigned long error = m->status & 0x1ff0000l; bool uncorrected_error = m->mcgstatus & 1ll << 61; @@ -1721,15 +1721,11 @@ static void i7core_mce_output_error(struct mem_ctl_info *mci, u32 errnum = find_first_bit(&error, 32); if (uncorrected_error) { - if (ripv) { - type = "FATAL"; + if (ripv) tp_event = HW_EVENT_ERR_FATAL; - } else { - type = "NON_FATAL"; + else tp_event = HW_EVENT_ERR_UNCORRECTED; - } } else { - type = "CORRECTED"; tp_event = HW_EVENT_ERR_CORRECTED; } diff --git a/drivers/edac/i82875p_edac.c b/drivers/edac/i82875p_edac.c index 80573df0a4d7..8d0450b9b9af 100644 --- a/drivers/edac/i82875p_edac.c +++ b/drivers/edac/i82875p_edac.c @@ -406,8 +406,6 @@ static int i82875p_probe1(struct pci_dev *pdev, int dev_idx) edac_dbg(0, "\n"); - ovrfl_pdev = pci_get_device(PCI_VEND_DEV(INTEL, 82875_6), NULL); - if (i82875p_setup_overfl_dev(pdev, &ovrfl_pdev, &ovrfl_window)) return -ENODEV; drc = readl(ovrfl_window + I82875P_DRC); diff --git a/drivers/edac/mce_amd.c b/drivers/edac/mce_amd.c index 30f7309446a6..51b9caa0b024 100644 --- a/drivers/edac/mce_amd.c +++ b/drivers/edac/mce_amd.c @@ -741,6 +741,36 @@ int amd_decode_mce(struct notifier_block *nb, unsigned long val, void *data) if (amd_filter_mce(m)) return NOTIFY_STOP; + pr_emerg(HW_ERR "%s\n", decode_error_status(m)); + + pr_emerg(HW_ERR "CPU:%d (%x:%x:%x) MC%d_STATUS[%s|%s|%s|%s|%s", + m->extcpu, + c->x86, c->x86_model, c->x86_mask, + m->bank, + ((m->status & MCI_STATUS_OVER) ? "Over" : "-"), + ((m->status & MCI_STATUS_UC) ? "UE" : "CE"), + ((m->status & MCI_STATUS_MISCV) ? "MiscV" : "-"), + ((m->status & MCI_STATUS_PCC) ? "PCC" : "-"), + ((m->status & MCI_STATUS_ADDRV) ? "AddrV" : "-")); + + if (c->x86 == 0x15 || c->x86 == 0x16) + pr_cont("|%s|%s", + ((m->status & MCI_STATUS_DEFERRED) ? "Deferred" : "-"), + ((m->status & MCI_STATUS_POISON) ? "Poison" : "-")); + + /* do the two bits[14:13] together */ + ecc = (m->status >> 45) & 0x3; + if (ecc) + pr_cont("|%sECC", ((ecc == 2) ? "C" : "U")); + + pr_cont("]: 0x%016llx\n", m->status); + + if (m->status & MCI_STATUS_ADDRV) + pr_emerg(HW_ERR "MC%d_ADDR: 0x%016llx\n", m->bank, m->addr); + + if (!fam_ops) + goto err_code; + switch (m->bank) { case 0: decode_mc0_mce(m); @@ -774,33 +804,7 @@ int amd_decode_mce(struct notifier_block *nb, unsigned long val, void *data) break; } - pr_emerg(HW_ERR "Error Status: %s\n", decode_error_status(m)); - - pr_emerg(HW_ERR "CPU:%d (%x:%x:%x) MC%d_STATUS[%s|%s|%s|%s|%s", - m->extcpu, - c->x86, c->x86_model, c->x86_mask, - m->bank, - ((m->status & MCI_STATUS_OVER) ? "Over" : "-"), - ((m->status & MCI_STATUS_UC) ? "UE" : "CE"), - ((m->status & MCI_STATUS_MISCV) ? "MiscV" : "-"), - ((m->status & MCI_STATUS_PCC) ? "PCC" : "-"), - ((m->status & MCI_STATUS_ADDRV) ? "AddrV" : "-")); - - if (c->x86 == 0x15 || c->x86 == 0x16) - pr_cont("|%s|%s", - ((m->status & MCI_STATUS_DEFERRED) ? "Deferred" : "-"), - ((m->status & MCI_STATUS_POISON) ? "Poison" : "-")); - - /* do the two bits[14:13] together */ - ecc = (m->status >> 45) & 0x3; - if (ecc) - pr_cont("|%sECC", ((ecc == 2) ? "C" : "U")); - - pr_cont("]: 0x%016llx\n", m->status); - - if (m->status & MCI_STATUS_ADDRV) - pr_emerg(HW_ERR "MC%d_ADDR: 0x%016llx\n", m->bank, m->addr); - + err_code: amd_decode_err_code(m->status & 0xffff); return NOTIFY_STOP; @@ -816,10 +820,7 @@ static int __init mce_amd_init(void) struct cpuinfo_x86 *c = &boot_cpu_data; if (c->x86_vendor != X86_VENDOR_AMD) - return 0; - - if (c->x86 < 0xf || c->x86 > 0x16) - return 0; + return -ENODEV; fam_ops = kzalloc(sizeof(struct amd_decoder_ops), GFP_KERNEL); if (!fam_ops) @@ -874,7 +875,7 @@ static int __init mce_amd_init(void) default: printk(KERN_WARNING "Huh? What family is it: 0x%x?!\n", c->x86); kfree(fam_ops); - return -EINVAL; + fam_ops = NULL; } pr_info("MCE: In-kernel MCE decoding enabled.\n"); diff --git a/drivers/edac/mpc85xx_edac.c b/drivers/edac/mpc85xx_edac.c index 8f9182179a7c..f4aec2e6ef56 100644 --- a/drivers/edac/mpc85xx_edac.c +++ b/drivers/edac/mpc85xx_edac.c @@ -357,7 +357,7 @@ int mpc85xx_pci_err_probe(struct platform_device *op) pdata->irq = irq_of_parse_and_map(op->dev.of_node, 0); res = devm_request_irq(&op->dev, pdata->irq, mpc85xx_pci_isr, - IRQF_DISABLED | IRQF_SHARED, + IRQF_SHARED, "[EDAC] PCI err", pci); if (res < 0) { printk(KERN_ERR @@ -633,7 +633,7 @@ static int mpc85xx_l2_err_probe(struct platform_device *op) if (edac_op_state == EDAC_OPSTATE_INT) { pdata->irq = irq_of_parse_and_map(op->dev.of_node, 0); res = devm_request_irq(&op->dev, pdata->irq, - mpc85xx_l2_isr, IRQF_DISABLED, + mpc85xx_l2_isr, 0, "[EDAC] L2 err", edac_dev); if (res < 0) { printk(KERN_ERR @@ -1133,7 +1133,7 @@ static int mpc85xx_mc_err_probe(struct platform_device *op) pdata->irq = irq_of_parse_and_map(op->dev.of_node, 0); res = devm_request_irq(&op->dev, pdata->irq, mpc85xx_mc_isr, - IRQF_DISABLED | IRQF_SHARED, + IRQF_SHARED, "[EDAC] MC err", mci); if (res < 0) { printk(KERN_ERR "%s: Unable to request irq %d for " diff --git a/drivers/edac/octeon_edac-lmc.c b/drivers/edac/octeon_edac-lmc.c index 93412d6b3af1..4bd10f94f068 100644 --- a/drivers/edac/octeon_edac-lmc.c +++ b/drivers/edac/octeon_edac-lmc.c @@ -5,12 +5,16 @@ * * Copyright (C) 2009 Wind River Systems, * written by Ralf Baechle <ralf@linux-mips.org> + * + * Copyright (c) 2013 by Cisco Systems, Inc. + * All rights reserved. */ #include <linux/module.h> #include <linux/init.h> #include <linux/slab.h> #include <linux/io.h> #include <linux/edac.h> +#include <linux/ctype.h> #include <asm/octeon/octeon.h> #include <asm/octeon/cvmx-lmcx-defs.h> @@ -20,6 +24,18 @@ #define OCTEON_MAX_MC 4 +#define to_mci(k) container_of(k, struct mem_ctl_info, dev) + +struct octeon_lmc_pvt { + unsigned long inject; + unsigned long error_type; + unsigned long dimm; + unsigned long rank; + unsigned long bank; + unsigned long row; + unsigned long col; +}; + static void octeon_lmc_edac_poll(struct mem_ctl_info *mci) { union cvmx_lmcx_mem_cfg0 cfg0; @@ -55,14 +71,31 @@ static void octeon_lmc_edac_poll(struct mem_ctl_info *mci) static void octeon_lmc_edac_poll_o2(struct mem_ctl_info *mci) { + struct octeon_lmc_pvt *pvt = mci->pvt_info; union cvmx_lmcx_int int_reg; bool do_clear = false; char msg[64]; - int_reg.u64 = cvmx_read_csr(CVMX_LMCX_INT(mci->mc_idx)); + if (!pvt->inject) + int_reg.u64 = cvmx_read_csr(CVMX_LMCX_INT(mci->mc_idx)); + else { + if (pvt->error_type == 1) + int_reg.s.sec_err = 1; + if (pvt->error_type == 2) + int_reg.s.ded_err = 1; + } + if (int_reg.s.sec_err || int_reg.s.ded_err) { union cvmx_lmcx_fadr fadr; - fadr.u64 = cvmx_read_csr(CVMX_LMCX_FADR(mci->mc_idx)); + if (likely(!pvt->inject)) + fadr.u64 = cvmx_read_csr(CVMX_LMCX_FADR(mci->mc_idx)); + else { + fadr.cn61xx.fdimm = pvt->dimm; + fadr.cn61xx.fbunk = pvt->rank; + fadr.cn61xx.fbank = pvt->bank; + fadr.cn61xx.frow = pvt->row; + fadr.cn61xx.fcol = pvt->col; + } snprintf(msg, sizeof(msg), "DIMM %d rank %d bank %d row %d col %d", fadr.cn61xx.fdimm, fadr.cn61xx.fbunk, @@ -82,8 +115,128 @@ static void octeon_lmc_edac_poll_o2(struct mem_ctl_info *mci) int_reg.s.ded_err = -1; /* Done, re-arm */ do_clear = true; } - if (do_clear) - cvmx_write_csr(CVMX_LMCX_INT(mci->mc_idx), int_reg.u64); + + if (do_clear) { + if (likely(!pvt->inject)) + cvmx_write_csr(CVMX_LMCX_INT(mci->mc_idx), int_reg.u64); + else + pvt->inject = 0; + } +} + +/************************ MC SYSFS parts ***********************************/ + +/* Only a couple naming differences per template, so very similar */ +#define TEMPLATE_SHOW(reg) \ +static ssize_t octeon_mc_inject_##reg##_show(struct device *dev, \ + struct device_attribute *attr, \ + char *data) \ +{ \ + struct mem_ctl_info *mci = to_mci(dev); \ + struct octeon_lmc_pvt *pvt = mci->pvt_info; \ + return sprintf(data, "%016llu\n", (u64)pvt->reg); \ +} + +#define TEMPLATE_STORE(reg) \ +static ssize_t octeon_mc_inject_##reg##_store(struct device *dev, \ + struct device_attribute *attr, \ + const char *data, size_t count) \ +{ \ + struct mem_ctl_info *mci = to_mci(dev); \ + struct octeon_lmc_pvt *pvt = mci->pvt_info; \ + if (isdigit(*data)) { \ + if (!kstrtoul(data, 0, &pvt->reg)) \ + return count; \ + } \ + return 0; \ +} + +TEMPLATE_SHOW(inject); +TEMPLATE_STORE(inject); +TEMPLATE_SHOW(dimm); +TEMPLATE_STORE(dimm); +TEMPLATE_SHOW(bank); +TEMPLATE_STORE(bank); +TEMPLATE_SHOW(rank); +TEMPLATE_STORE(rank); +TEMPLATE_SHOW(row); +TEMPLATE_STORE(row); +TEMPLATE_SHOW(col); +TEMPLATE_STORE(col); + +static ssize_t octeon_mc_inject_error_type_store(struct device *dev, + struct device_attribute *attr, + const char *data, + size_t count) +{ + struct mem_ctl_info *mci = to_mci(dev); + struct octeon_lmc_pvt *pvt = mci->pvt_info; + + if (!strncmp(data, "single", 6)) + pvt->error_type = 1; + else if (!strncmp(data, "double", 6)) + pvt->error_type = 2; + + return count; +} + +static ssize_t octeon_mc_inject_error_type_show(struct device *dev, + struct device_attribute *attr, + char *data) +{ + struct mem_ctl_info *mci = to_mci(dev); + struct octeon_lmc_pvt *pvt = mci->pvt_info; + if (pvt->error_type == 1) + return sprintf(data, "single"); + else if (pvt->error_type == 2) + return sprintf(data, "double"); + + return 0; +} + +static DEVICE_ATTR(inject, S_IRUGO | S_IWUSR, + octeon_mc_inject_inject_show, octeon_mc_inject_inject_store); +static DEVICE_ATTR(error_type, S_IRUGO | S_IWUSR, + octeon_mc_inject_error_type_show, octeon_mc_inject_error_type_store); +static DEVICE_ATTR(dimm, S_IRUGO | S_IWUSR, + octeon_mc_inject_dimm_show, octeon_mc_inject_dimm_store); +static DEVICE_ATTR(rank, S_IRUGO | S_IWUSR, + octeon_mc_inject_rank_show, octeon_mc_inject_rank_store); +static DEVICE_ATTR(bank, S_IRUGO | S_IWUSR, + octeon_mc_inject_bank_show, octeon_mc_inject_bank_store); +static DEVICE_ATTR(row, S_IRUGO | S_IWUSR, + octeon_mc_inject_row_show, octeon_mc_inject_row_store); +static DEVICE_ATTR(col, S_IRUGO | S_IWUSR, + octeon_mc_inject_col_show, octeon_mc_inject_col_store); + + +static int octeon_set_mc_sysfs_attributes(struct mem_ctl_info *mci) +{ + int rc; + + rc = device_create_file(&mci->dev, &dev_attr_inject); + if (rc < 0) + return rc; + rc = device_create_file(&mci->dev, &dev_attr_error_type); + if (rc < 0) + return rc; + rc = device_create_file(&mci->dev, &dev_attr_dimm); + if (rc < 0) + return rc; + rc = device_create_file(&mci->dev, &dev_attr_rank); + if (rc < 0) + return rc; + rc = device_create_file(&mci->dev, &dev_attr_bank); + if (rc < 0) + return rc; + rc = device_create_file(&mci->dev, &dev_attr_row); + if (rc < 0) + return rc; + rc = device_create_file(&mci->dev, &dev_attr_col); + if (rc < 0) + return rc; + + return 0; } static int octeon_lmc_edac_probe(struct platform_device *pdev) @@ -92,6 +245,8 @@ static int octeon_lmc_edac_probe(struct platform_device *pdev) struct edac_mc_layer layers[1]; int mc = pdev->id; + opstate_init(); + layers[0].type = EDAC_MC_LAYER_CHANNEL; layers[0].size = 1; layers[0].is_virt_csrow = false; @@ -105,7 +260,7 @@ static int octeon_lmc_edac_probe(struct platform_device *pdev) return 0; } - mci = edac_mc_alloc(mc, ARRAY_SIZE(layers), layers, 0); + mci = edac_mc_alloc(mc, ARRAY_SIZE(layers), layers, sizeof(struct octeon_lmc_pvt)); if (!mci) return -ENXIO; @@ -122,6 +277,12 @@ static int octeon_lmc_edac_probe(struct platform_device *pdev) return -ENXIO; } + if (octeon_set_mc_sysfs_attributes(mci)) { + dev_err(&pdev->dev, "octeon_set_mc_sysfs_attributes() failed\n"); + return -ENXIO; + } + + cfg0.u64 = cvmx_read_csr(CVMX_LMCX_MEM_CFG0(mc)); cfg0.s.intr_ded_ena = 0; /* We poll */ cfg0.s.intr_sec_ena = 0; @@ -137,7 +298,7 @@ static int octeon_lmc_edac_probe(struct platform_device *pdev) return 0; } - mci = edac_mc_alloc(mc, ARRAY_SIZE(layers), layers, 0); + mci = edac_mc_alloc(mc, ARRAY_SIZE(layers), layers, sizeof(struct octeon_lmc_pvt)); if (!mci) return -ENXIO; @@ -154,6 +315,12 @@ static int octeon_lmc_edac_probe(struct platform_device *pdev) return -ENXIO; } + if (octeon_set_mc_sysfs_attributes(mci)) { + dev_err(&pdev->dev, "octeon_set_mc_sysfs_attributes() failed\n"); + return -ENXIO; + } + + en.u64 = cvmx_read_csr(CVMX_LMCX_MEM_CFG0(mc)); en.s.intr_ded_ena = 0; /* We poll */ en.s.intr_sec_ena = 0; diff --git a/drivers/edac/sb_edac.c b/drivers/edac/sb_edac.c index 3fa13dbf2859..deea0dc9999b 100644 --- a/drivers/edac/sb_edac.c +++ b/drivers/edac/sb_edac.c @@ -1263,7 +1263,7 @@ static int sbridge_get_onedevice(struct pci_dev **prev, struct pci_dev *pdev = NULL; u8 bus = 0; - sbridge_printk(KERN_INFO, + sbridge_printk(KERN_DEBUG, "Seeking for: dev %02x.%d PCI ID %04x:%04x\n", dev_descr->dev, dev_descr->func, PCI_VENDOR_ID_INTEL, dev_descr->dev_id); @@ -1828,6 +1828,7 @@ static int sbridge_mce_check_error(struct notifier_block *nb, unsigned long val, struct mce *mce = (struct mce *)data; struct mem_ctl_info *mci; struct sbridge_pvt *pvt; + char *type; if (get_edac_report_status() == EDAC_REPORTING_DISABLED) return NOTIFY_DONE; @@ -1846,17 +1847,23 @@ static int sbridge_mce_check_error(struct notifier_block *nb, unsigned long val, if ((mce->status & 0xefff) >> 7 != 1) return NOTIFY_DONE; - printk("sbridge: HANDLING MCE MEMORY ERROR\n"); + if (mce->mcgstatus & MCG_STATUS_MCIP) + type = "Exception"; + else + type = "Event"; + + sbridge_mc_printk(mci, KERN_DEBUG, "HANDLING MCE MEMORY ERROR\n"); - printk("CPU %d: Machine Check Exception: %Lx Bank %d: %016Lx\n", - mce->extcpu, mce->mcgstatus, mce->bank, mce->status); - printk("TSC %llx ", mce->tsc); - printk("ADDR %llx ", mce->addr); - printk("MISC %llx ", mce->misc); + sbridge_mc_printk(mci, KERN_DEBUG, "CPU %d: Machine Check %s: %Lx " + "Bank %d: %016Lx\n", mce->extcpu, type, + mce->mcgstatus, mce->bank, mce->status); + sbridge_mc_printk(mci, KERN_DEBUG, "TSC %llx ", mce->tsc); + sbridge_mc_printk(mci, KERN_DEBUG, "ADDR %llx ", mce->addr); + sbridge_mc_printk(mci, KERN_DEBUG, "MISC %llx ", mce->misc); - printk("PROCESSOR %u:%x TIME %llu SOCKET %u APIC %x\n", - mce->cpuvendor, mce->cpuid, mce->time, - mce->socketid, mce->apicid); + sbridge_mc_printk(mci, KERN_DEBUG, "PROCESSOR %u:%x TIME %llu SOCKET " + "%u APIC %x\n", mce->cpuvendor, mce->cpuid, + mce->time, mce->socketid, mce->apicid); /* Only handle if it is the right mc controller */ if (cpu_data(mce->cpu).phys_proc_id != pvt->sbridge_dev->mc) |