diff options
Diffstat (limited to 'drivers/edac/e7xxx_edac.c')
| -rw-r--r-- | drivers/edac/e7xxx_edac.c | 372 |
1 files changed, 211 insertions, 161 deletions
diff --git a/drivers/edac/e7xxx_edac.c b/drivers/edac/e7xxx_edac.c index a9518d3e4be..3cda79bc8b0 100644 --- a/drivers/edac/e7xxx_edac.c +++ b/drivers/edac/e7xxx_edac.c @@ -10,6 +10,9 @@ * Based on work by Dan Hollis <goemon at anime dot net> and others. * http://www.anime.net/~goemon/linux-ecc/ * + * Datasheet: + * http://www.intel.com/content/www/us/en/chipsets/e7501-chipset-memory-controller-hub-datasheet.html + * * Contributors: * Eric Biederman (Linux Networx) * Tom Zimmerman (Linux Networx) @@ -22,13 +25,15 @@ * */ -#include <linux/config.h> #include <linux/module.h> #include <linux/init.h> #include <linux/pci.h> #include <linux/pci_ids.h> -#include <linux/slab.h> -#include "edac_mc.h" +#include <linux/edac.h> +#include "edac_core.h" + +#define E7XXX_REVISION " Ver: 2.0.2" +#define EDAC_MOD_STR "e7xxx_edac" #define e7xxx_printk(level, fmt, arg...) \ edac_printk(level, "e7xxx", fmt, ##arg) @@ -69,7 +74,7 @@ #endif /* PCI_DEVICE_ID_INTEL_7505_1_ERR */ #define E7XXX_NR_CSROWS 8 /* number of csrows */ -#define E7XXX_NR_DIMMS 8 /* FIXME - is this correct? */ +#define E7XXX_NR_DIMMS 8 /* 2 channels, 4 dimms/channel */ /* E7XXX register addresses - device 0 function 0 */ #define E7XXX_DRB 0x60 /* DRAM row boundary register (8b) */ @@ -141,29 +146,27 @@ struct e7xxx_error_info { u32 dram_uelog_add; }; +static struct edac_pci_ctl_info *e7xxx_pci; + static const struct e7xxx_dev_info e7xxx_devs[] = { [E7500] = { .err_dev = PCI_DEVICE_ID_INTEL_7500_1_ERR, - .ctl_name = "E7500" - }, + .ctl_name = "E7500"}, [E7501] = { .err_dev = PCI_DEVICE_ID_INTEL_7501_1_ERR, - .ctl_name = "E7501" - }, + .ctl_name = "E7501"}, [E7505] = { .err_dev = PCI_DEVICE_ID_INTEL_7505_1_ERR, - .ctl_name = "E7505" - }, + .ctl_name = "E7505"}, [E7205] = { .err_dev = PCI_DEVICE_ID_INTEL_7205_1_ERR, - .ctl_name = "E7205" - }, + .ctl_name = "E7205"}, }; /* FIXME - is this valid for both SECDED and S4ECD4ED? */ static inline int e7xxx_find_channel(u16 syndrome) { - debugf3("%s()\n", __func__); + edac_dbg(3, "\n"); if ((syndrome & 0xff00) == 0) return 0; @@ -178,15 +181,15 @@ static inline int e7xxx_find_channel(u16 syndrome) } static unsigned long ctl_page_to_phys(struct mem_ctl_info *mci, - unsigned long page) + unsigned long page) { u32 remap; - struct e7xxx_pvt *pvt = (struct e7xxx_pvt *) mci->pvt_info; + struct e7xxx_pvt *pvt = (struct e7xxx_pvt *)mci->pvt_info; - debugf3("%s()\n", __func__); + edac_dbg(3, "\n"); if ((page < pvt->tolm) || - ((page >= 0x100000) && (page < pvt->remapbase))) + ((page >= 0x100000) && (page < pvt->remapbase))) return page; remap = (page - pvt->tolm) + pvt->remapbase; @@ -198,65 +201,67 @@ static unsigned long ctl_page_to_phys(struct mem_ctl_info *mci, return pvt->tolm - 1; } -static void process_ce(struct mem_ctl_info *mci, - struct e7xxx_error_info *info) +static void process_ce(struct mem_ctl_info *mci, struct e7xxx_error_info *info) { u32 error_1b, page; u16 syndrome; int row; int channel; - debugf3("%s()\n", __func__); + edac_dbg(3, "\n"); /* read the error address */ error_1b = info->dram_celog_add; /* FIXME - should use PAGE_SHIFT */ - page = error_1b >> 6; /* convert the address to 4k page */ + page = error_1b >> 6; /* convert the address to 4k page */ /* read the syndrome */ syndrome = info->dram_celog_syndrome; /* FIXME - check for -1 */ row = edac_mc_find_csrow_by_page(mci, page); /* convert syndrome to channel */ channel = e7xxx_find_channel(syndrome); - edac_mc_handle_ce(mci, page, 0, syndrome, row, channel, "e7xxx CE"); + edac_mc_handle_error(HW_EVENT_ERR_CORRECTED, mci, 1, page, 0, syndrome, + row, channel, -1, "e7xxx CE", ""); } static void process_ce_no_info(struct mem_ctl_info *mci) { - debugf3("%s()\n", __func__); - edac_mc_handle_ce_no_info(mci, "e7xxx CE log register overflow"); + edac_dbg(3, "\n"); + edac_mc_handle_error(HW_EVENT_ERR_UNCORRECTED, mci, 1, 0, 0, 0, -1, -1, -1, + "e7xxx CE log register overflow", ""); } -static void process_ue(struct mem_ctl_info *mci, - struct e7xxx_error_info *info) +static void process_ue(struct mem_ctl_info *mci, struct e7xxx_error_info *info) { u32 error_2b, block_page; int row; - debugf3("%s()\n", __func__); + edac_dbg(3, "\n"); /* read the error address */ error_2b = info->dram_uelog_add; /* FIXME - should use PAGE_SHIFT */ - block_page = error_2b >> 6; /* convert to 4k address */ + block_page = error_2b >> 6; /* convert to 4k address */ row = edac_mc_find_csrow_by_page(mci, block_page); - edac_mc_handle_ue(mci, block_page, 0, row, "e7xxx UE"); + + edac_mc_handle_error(HW_EVENT_ERR_UNCORRECTED, mci, 1, block_page, 0, 0, + row, -1, -1, "e7xxx UE", ""); } static void process_ue_no_info(struct mem_ctl_info *mci) { - debugf3("%s()\n", __func__); - edac_mc_handle_ue_no_info(mci, "e7xxx UE log register overflow"); + edac_dbg(3, "\n"); + + edac_mc_handle_error(HW_EVENT_ERR_UNCORRECTED, mci, 1, 0, 0, 0, -1, -1, -1, + "e7xxx UE log register overflow", ""); } -static void e7xxx_get_error_info (struct mem_ctl_info *mci, - struct e7xxx_error_info *info) +static void e7xxx_get_error_info(struct mem_ctl_info *mci, + struct e7xxx_error_info *info) { struct e7xxx_pvt *pvt; - pvt = (struct e7xxx_pvt *) mci->pvt_info; - pci_read_config_byte(pvt->bridge_ck, E7XXX_DRAM_FERR, - &info->dram_ferr); - pci_read_config_byte(pvt->bridge_ck, E7XXX_DRAM_NERR, - &info->dram_nerr); + pvt = (struct e7xxx_pvt *)mci->pvt_info; + pci_read_config_byte(pvt->bridge_ck, E7XXX_DRAM_FERR, &info->dram_ferr); + pci_read_config_byte(pvt->bridge_ck, E7XXX_DRAM_NERR, &info->dram_nerr); if ((info->dram_ferr & 1) || (info->dram_nerr & 1)) { pci_read_config_dword(pvt->bridge_ck, E7XXX_DRAM_CELOG_ADD, @@ -277,8 +282,9 @@ static void e7xxx_get_error_info (struct mem_ctl_info *mci, pci_write_bits8(pvt->bridge_ck, E7XXX_DRAM_NERR, 0x03, 0x03); } -static int e7xxx_process_error_info (struct mem_ctl_info *mci, - struct e7xxx_error_info *info, int handle_errors) +static int e7xxx_process_error_info(struct mem_ctl_info *mci, + struct e7xxx_error_info *info, + int handle_errors) { int error_found; @@ -328,131 +334,159 @@ static void e7xxx_check(struct mem_ctl_info *mci) { struct e7xxx_error_info info; - debugf3("%s()\n", __func__); + edac_dbg(3, "\n"); e7xxx_get_error_info(mci, &info); e7xxx_process_error_info(mci, &info, 1); } -static int e7xxx_probe1(struct pci_dev *pdev, int dev_idx) +/* Return 1 if dual channel mode is active. Else return 0. */ +static inline int dual_channel_active(u32 drc, int dev_idx) { - int rc = -ENODEV; - int index; - u16 pci_data; - struct mem_ctl_info *mci = NULL; - struct e7xxx_pvt *pvt = NULL; - u32 drc; - int drc_chan = 1; /* Number of channels 0=1chan,1=2chan */ - int drc_drbg = 1; /* DRB granularity 0=32mb,1=64mb */ - int drc_ddim; /* DRAM Data Integrity Mode 0=none,2=edac */ - u32 dra; - unsigned long last_cumul_size; - struct e7xxx_error_info discard; - - debugf0("%s(): mci\n", __func__); - - /* need to find out the number of channels */ - pci_read_config_dword(pdev, E7XXX_DRC, &drc); + return (dev_idx == E7501) ? ((drc >> 22) & 0x1) : 1; +} +/* Return DRB granularity (0=32mb, 1=64mb). */ +static inline int drb_granularity(u32 drc, int dev_idx) +{ /* only e7501 can be single channel */ - if (dev_idx == E7501) { - drc_chan = ((drc >> 22) & 0x1); - drc_drbg = (drc >> 18) & 0x3; - } - - drc_ddim = (drc >> 20) & 0x3; - mci = edac_mc_alloc(sizeof(*pvt), E7XXX_NR_CSROWS, drc_chan + 1); - - if (mci == NULL) { - rc = -ENOMEM; - goto fail; - } - - debugf3("%s(): init mci\n", __func__); - mci->mtype_cap = MEM_FLAG_RDDR; - mci->edac_ctl_cap = EDAC_FLAG_NONE | EDAC_FLAG_SECDED | - EDAC_FLAG_S4ECD4ED; - /* FIXME - what if different memory types are in different csrows? */ - mci->mod_name = EDAC_MOD_STR; - mci->mod_ver = "$Revision: 1.5.2.9 $"; - mci->pdev = pdev; - - debugf3("%s(): init pvt\n", __func__); - pvt = (struct e7xxx_pvt *) mci->pvt_info; - pvt->dev_info = &e7xxx_devs[dev_idx]; - pvt->bridge_ck = pci_get_device(PCI_VENDOR_ID_INTEL, - pvt->dev_info->err_dev, - pvt->bridge_ck); - - if (!pvt->bridge_ck) { - e7xxx_printk(KERN_ERR, "error reporting device not found:" - "vendor %x device 0x%x (broken BIOS?)\n", - PCI_VENDOR_ID_INTEL, e7xxx_devs[dev_idx].err_dev); - goto fail; - } + return (dev_idx == E7501) ? ((drc >> 18) & 0x3) : 1; +} - debugf3("%s(): more mci init\n", __func__); - mci->ctl_name = pvt->dev_info->ctl_name; - mci->edac_check = e7xxx_check; - mci->ctl_page_to_phys = ctl_page_to_phys; +static void e7xxx_init_csrows(struct mem_ctl_info *mci, struct pci_dev *pdev, + int dev_idx, u32 drc) +{ + unsigned long last_cumul_size; + int index, j; + u8 value; + u32 dra, cumul_size, nr_pages; + int drc_chan, drc_drbg, drc_ddim, mem_dev; + struct csrow_info *csrow; + struct dimm_info *dimm; + enum edac_type edac_mode; - /* find out the device types */ pci_read_config_dword(pdev, E7XXX_DRA, &dra); + drc_chan = dual_channel_active(drc, dev_idx); + drc_drbg = drb_granularity(drc, dev_idx); + drc_ddim = (drc >> 20) & 0x3; + last_cumul_size = 0; - /* - * The dram row boundary (DRB) reg values are boundary address + /* The dram row boundary (DRB) reg values are boundary address * for each DRAM row with a granularity of 32 or 64MB (single/dual * channel operation). DRB regs are cumulative; therefore DRB7 will * contain the total memory contained in all eight rows. */ - for (last_cumul_size = index = 0; index < mci->nr_csrows; index++) { - u8 value; - u32 cumul_size; + for (index = 0; index < mci->nr_csrows; index++) { /* mem_dev 0=x8, 1=x4 */ - int mem_dev = (dra >> (index * 4 + 3)) & 0x1; - struct csrow_info *csrow = &mci->csrows[index]; + mem_dev = (dra >> (index * 4 + 3)) & 0x1; + csrow = mci->csrows[index]; - pci_read_config_byte(mci->pdev, E7XXX_DRB + index, &value); + pci_read_config_byte(pdev, E7XXX_DRB + index, &value); /* convert a 64 or 32 MiB DRB to a page size. */ cumul_size = value << (25 + drc_drbg - PAGE_SHIFT); - debugf3("%s(): (%d) cumul_size 0x%x\n", __func__, index, - cumul_size); - + edac_dbg(3, "(%d) cumul_size 0x%x\n", index, cumul_size); if (cumul_size == last_cumul_size) - continue; /* not populated */ + continue; /* not populated */ csrow->first_page = last_cumul_size; csrow->last_page = cumul_size - 1; - csrow->nr_pages = cumul_size - last_cumul_size; + nr_pages = cumul_size - last_cumul_size; last_cumul_size = cumul_size; - csrow->grain = 1 << 12; /* 4KiB - resolution of CELOG */ - csrow->mtype = MEM_RDDR; /* only one type supported */ - csrow->dtype = mem_dev ? DEV_X4 : DEV_X8; /* - * if single channel or x8 devices then SECDED - * if dual channel and x4 then S4ECD4ED - */ + * if single channel or x8 devices then SECDED + * if dual channel and x4 then S4ECD4ED + */ if (drc_ddim) { if (drc_chan && mem_dev) { - csrow->edac_mode = EDAC_S4ECD4ED; + edac_mode = EDAC_S4ECD4ED; mci->edac_cap |= EDAC_FLAG_S4ECD4ED; } else { - csrow->edac_mode = EDAC_SECDED; + edac_mode = EDAC_SECDED; mci->edac_cap |= EDAC_FLAG_SECDED; } } else - csrow->edac_mode = EDAC_NONE; + edac_mode = EDAC_NONE; + + for (j = 0; j < drc_chan + 1; j++) { + dimm = csrow->channels[j]->dimm; + + dimm->nr_pages = nr_pages / (drc_chan + 1); + dimm->grain = 1 << 12; /* 4KiB - resolution of CELOG */ + dimm->mtype = MEM_RDDR; /* only one type supported */ + dimm->dtype = mem_dev ? DEV_X4 : DEV_X8; + dimm->edac_mode = edac_mode; + } } +} - mci->edac_cap |= EDAC_FLAG_NONE; +static int e7xxx_probe1(struct pci_dev *pdev, int dev_idx) +{ + u16 pci_data; + struct mem_ctl_info *mci = NULL; + struct edac_mc_layer layers[2]; + struct e7xxx_pvt *pvt = NULL; + u32 drc; + int drc_chan; + struct e7xxx_error_info discard; + + edac_dbg(0, "mci\n"); + + pci_read_config_dword(pdev, E7XXX_DRC, &drc); + + drc_chan = dual_channel_active(drc, dev_idx); + /* + * According with the datasheet, this device has a maximum of + * 4 DIMMS per channel, either single-rank or dual-rank. So, the + * total amount of dimms is 8 (E7XXX_NR_DIMMS). + * That means that the DIMM is mapped as CSROWs, and the channel + * will map the rank. So, an error to either channel should be + * attributed to the same dimm. + */ + layers[0].type = EDAC_MC_LAYER_CHIP_SELECT; + layers[0].size = E7XXX_NR_CSROWS; + layers[0].is_virt_csrow = true; + layers[1].type = EDAC_MC_LAYER_CHANNEL; + layers[1].size = drc_chan + 1; + layers[1].is_virt_csrow = false; + mci = edac_mc_alloc(0, ARRAY_SIZE(layers), layers, sizeof(*pvt)); + if (mci == NULL) + return -ENOMEM; + + edac_dbg(3, "init mci\n"); + mci->mtype_cap = MEM_FLAG_RDDR; + mci->edac_ctl_cap = EDAC_FLAG_NONE | EDAC_FLAG_SECDED | + EDAC_FLAG_S4ECD4ED; + /* FIXME - what if different memory types are in different csrows? */ + mci->mod_name = EDAC_MOD_STR; + mci->mod_ver = E7XXX_REVISION; + mci->pdev = &pdev->dev; + edac_dbg(3, "init pvt\n"); + pvt = (struct e7xxx_pvt *)mci->pvt_info; + pvt->dev_info = &e7xxx_devs[dev_idx]; + pvt->bridge_ck = pci_get_device(PCI_VENDOR_ID_INTEL, + pvt->dev_info->err_dev, pvt->bridge_ck); - debugf3("%s(): tolm, remapbase, remaplimit\n", __func__); + if (!pvt->bridge_ck) { + e7xxx_printk(KERN_ERR, "error reporting device not found:" + "vendor %x device 0x%x (broken BIOS?)\n", + PCI_VENDOR_ID_INTEL, e7xxx_devs[dev_idx].err_dev); + goto fail0; + } + + edac_dbg(3, "more mci init\n"); + mci->ctl_name = pvt->dev_info->ctl_name; + mci->dev_name = pci_name(pdev); + mci->edac_check = e7xxx_check; + mci->ctl_page_to_phys = ctl_page_to_phys; + e7xxx_init_csrows(mci, pdev, dev_idx, drc); + mci->edac_cap |= EDAC_FLAG_NONE; + edac_dbg(3, "tolm, remapbase, remaplimit\n"); /* load the top of low memory, remap base, and remap limit vars */ - pci_read_config_word(mci->pdev, E7XXX_TOLM, &pci_data); + pci_read_config_word(pdev, E7XXX_TOLM, &pci_data); pvt->tolm = ((u32) pci_data) << 4; - pci_read_config_word(mci->pdev, E7XXX_REMAPBASE, &pci_data); + pci_read_config_word(pdev, E7XXX_REMAPBASE, &pci_data); pvt->remapbase = ((u32) pci_data) << 14; - pci_read_config_word(mci->pdev, E7XXX_REMAPLIMIT, &pci_data); + pci_read_config_word(pdev, E7XXX_REMAPLIMIT, &pci_data); pvt->remaplimit = ((u32) pci_data) << 14; e7xxx_printk(KERN_INFO, "tolm = %x, remapbase = %x, remaplimit = %x\n", pvt->tolm, @@ -461,71 +495,82 @@ static int e7xxx_probe1(struct pci_dev *pdev, int dev_idx) /* clear any pending errors, or initial state bits */ e7xxx_get_error_info(mci, &discard); - if (edac_mc_add_mc(mci) != 0) { - debugf3("%s(): failed edac_mc_add_mc()\n", __func__); - goto fail; + /* Here we assume that we will never see multiple instances of this + * type of memory controller. The ID is therefore hardcoded to 0. + */ + if (edac_mc_add_mc(mci)) { + edac_dbg(3, "failed edac_mc_add_mc()\n"); + goto fail1; + } + + /* allocating generic PCI control info */ + e7xxx_pci = edac_pci_create_generic_ctl(&pdev->dev, EDAC_MOD_STR); + if (!e7xxx_pci) { + printk(KERN_WARNING + "%s(): Unable to create PCI control\n", + __func__); + printk(KERN_WARNING + "%s(): PCI error report via EDAC not setup\n", + __func__); } /* get this far and it's successful */ - debugf3("%s(): success\n", __func__); + edac_dbg(3, "success\n"); return 0; -fail: - if (mci != NULL) { - if(pvt != NULL && pvt->bridge_ck) - pci_dev_put(pvt->bridge_ck); - edac_mc_free(mci); - } +fail1: + pci_dev_put(pvt->bridge_ck); + +fail0: + edac_mc_free(mci); - return rc; + return -ENODEV; } /* returns count (>= 0), or negative on error */ -static int __devinit e7xxx_init_one(struct pci_dev *pdev, - const struct pci_device_id *ent) +static int e7xxx_init_one(struct pci_dev *pdev, const struct pci_device_id *ent) { - debugf0("%s()\n", __func__); + edac_dbg(0, "\n"); /* wake up and enable device */ return pci_enable_device(pdev) ? -EIO : e7xxx_probe1(pdev, ent->driver_data); } -static void __devexit e7xxx_remove_one(struct pci_dev *pdev) +static void e7xxx_remove_one(struct pci_dev *pdev) { struct mem_ctl_info *mci; struct e7xxx_pvt *pvt; - debugf0("%s()\n", __func__); + edac_dbg(0, "\n"); - if ((mci = edac_mc_del_mc(pdev)) == NULL) + if (e7xxx_pci) + edac_pci_release_generic_ctl(e7xxx_pci); + + if ((mci = edac_mc_del_mc(&pdev->dev)) == NULL) return; - pvt = (struct e7xxx_pvt *) mci->pvt_info; + pvt = (struct e7xxx_pvt *)mci->pvt_info; pci_dev_put(pvt->bridge_ck); edac_mc_free(mci); } -static const struct pci_device_id e7xxx_pci_tbl[] __devinitdata = { +static const struct pci_device_id e7xxx_pci_tbl[] = { { - PCI_VEND_DEV(INTEL, 7205_0), PCI_ANY_ID, PCI_ANY_ID, 0, 0, - E7205 - }, + PCI_VEND_DEV(INTEL, 7205_0), PCI_ANY_ID, PCI_ANY_ID, 0, 0, + E7205}, { - PCI_VEND_DEV(INTEL, 7500_0), PCI_ANY_ID, PCI_ANY_ID, 0, 0, - E7500 - }, + PCI_VEND_DEV(INTEL, 7500_0), PCI_ANY_ID, PCI_ANY_ID, 0, 0, + E7500}, { - PCI_VEND_DEV(INTEL, 7501_0), PCI_ANY_ID, PCI_ANY_ID, 0, 0, - E7501 - }, + PCI_VEND_DEV(INTEL, 7501_0), PCI_ANY_ID, PCI_ANY_ID, 0, 0, + E7501}, { - PCI_VEND_DEV(INTEL, 7505_0), PCI_ANY_ID, PCI_ANY_ID, 0, 0, - E7505 - }, + PCI_VEND_DEV(INTEL, 7505_0), PCI_ANY_ID, PCI_ANY_ID, 0, 0, + E7505}, { - 0, - } /* 0 terminated list. */ + 0, + } /* 0 terminated list. */ }; MODULE_DEVICE_TABLE(pci, e7xxx_pci_tbl); @@ -533,12 +578,15 @@ MODULE_DEVICE_TABLE(pci, e7xxx_pci_tbl); static struct pci_driver e7xxx_driver = { .name = EDAC_MOD_STR, .probe = e7xxx_init_one, - .remove = __devexit_p(e7xxx_remove_one), + .remove = e7xxx_remove_one, .id_table = e7xxx_pci_tbl, }; static int __init e7xxx_init(void) { + /* Ensure that the OPSTATE is set correctly for POLL or NMI */ + opstate_init(); + return pci_register_driver(&e7xxx_driver); } @@ -552,5 +600,7 @@ module_exit(e7xxx_exit); MODULE_LICENSE("GPL"); MODULE_AUTHOR("Linux Networx (http://lnxi.com) Thayne Harbaugh et al\n" - "Based on.work by Dan Hollis et al"); + "Based on.work by Dan Hollis et al"); MODULE_DESCRIPTION("MC support for Intel e7xxx memory controllers"); +module_param(edac_op_state, int, 0444); +MODULE_PARM_DESC(edac_op_state, "EDAC Error Reporting state: 0=Poll,1=NMI"); |
