* http://www.anime.net/~goemon/linux-ecc/
*
* Contributors:
- * Eric Biederman (Linux Networx)
- * Tom Zimmerman (Linux Networx)
- * Jim Garlick (Lawrence Livermore National Labs)
+ * Eric Biederman (Linux Networx)
+ * Tom Zimmerman (Linux Networx)
+ * Jim Garlick (Lawrence Livermore National Labs)
* Dave Peterson (Lawrence Livermore National Labs)
* That One Guy (Some other place)
* Wang Zhenyu (intel.com)
*
*/
-
-#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 " __DATE__
+#define EDAC_MOD_STR "e7xxx_edac"
#define e7xxx_printk(level, fmt, arg...) \
- edac_printk(level, "e7xxx", fmt, ##arg)
-
+ edac_printk(level, "e7xxx", fmt, ##arg)
#define e7xxx_mc_printk(mci, level, fmt, arg...) \
- edac_mc_chipset_printk(mci, level, "e7xxx", fmt, ##arg)
-
+ edac_mc_chipset_printk(mci, level, "e7xxx", fmt, ##arg)
#ifndef PCI_DEVICE_ID_INTEL_7205_0
#define PCI_DEVICE_ID_INTEL_7205_0 0x255d
#define PCI_DEVICE_ID_INTEL_7505_1_ERR 0x2551
#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? */
-
/* E7XXX register addresses - device 0 function 0 */
#define E7XXX_DRB 0x60 /* DRAM row boundary register (8b) */
#define E7XXX_DRA 0x70 /* DRAM row attribute register (8b) */
E7205,
};
-
struct e7xxx_pvt {
struct pci_dev *bridge_ck;
u32 tolm;
const struct e7xxx_dev_info *dev_info;
};
-
struct e7xxx_dev_info {
u16 err_dev;
const char *ctl_name;
};
-
struct e7xxx_error_info {
u8 dram_ferr;
u8 dram_nerr;
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"},
+ .err_dev = PCI_DEVICE_ID_INTEL_7500_1_ERR,
+ .ctl_name = "E7500"},
[E7501] = {
- .err_dev = PCI_DEVICE_ID_INTEL_7501_1_ERR,
- .ctl_name = "E7501"},
+ .err_dev = PCI_DEVICE_ID_INTEL_7501_1_ERR,
+ .ctl_name = "E7501"},
[E7505] = {
- .err_dev = PCI_DEVICE_ID_INTEL_7505_1_ERR,
- .ctl_name = "E7505"},
+ .err_dev = PCI_DEVICE_ID_INTEL_7505_1_ERR,
+ .ctl_name = "E7505"},
[E7205] = {
- .err_dev = PCI_DEVICE_ID_INTEL_7205_1_ERR,
- .ctl_name = "E7205"},
+ .err_dev = PCI_DEVICE_ID_INTEL_7205_1_ERR,
+ .ctl_name = "E7205"},
};
-
/* FIXME - is this valid for both SECDED and S4ECD4ED? */
static inline int e7xxx_find_channel(u16 syndrome)
{
if ((syndrome & 0xff00) == 0)
return 0;
+
if ((syndrome & 0x00ff) == 0)
return 1;
+
if ((syndrome & 0xf000) == 0 || (syndrome & 0x0f00) == 0)
return 0;
+
return 1;
}
-
-static unsigned long
-ctl_page_to_phys(struct mem_ctl_info *mci, unsigned long page)
+static unsigned long ctl_page_to_phys(struct mem_ctl_info *mci,
+ 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__);
if ((page < pvt->tolm) ||
- ((page >= 0x100000) && (page < pvt->remapbase)))
+ ((page >= 0x100000) && (page < pvt->remapbase)))
return page;
+
remap = (page - pvt->tolm) + pvt->remapbase;
+
if (remap < pvt->remaplimit)
return remap;
+
e7xxx_printk(KERN_ERR, "Invalid page %lx - out of range\n", page);
return pvt->tolm - 1;
}
-
static void process_ce(struct mem_ctl_info *mci, struct e7xxx_error_info *info)
{
u32 error_1b, page;
int channel;
debugf3("%s()\n", __func__);
-
/* read the error address */
error_1b = info->dram_celog_add;
/* FIXME - should use PAGE_SHIFT */
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_ce(mci, page, 0, syndrome, row, channel, "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");
}
-
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__);
-
/* read the error address */
error_2b = info->dram_uelog_add;
/* FIXME - should use PAGE_SHIFT */
edac_mc_handle_ue(mci, block_page, 0, row, "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");
}
-
-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,
- &info->dram_celog_add);
+ &info->dram_celog_add);
pci_read_config_word(pvt->bridge_ck,
- E7XXX_DRAM_CELOG_SYNDROME, &info->dram_celog_syndrome);
+ E7XXX_DRAM_CELOG_SYNDROME,
+ &info->dram_celog_syndrome);
}
if ((info->dram_ferr & 2) || (info->dram_nerr & 2))
pci_read_config_dword(pvt->bridge_ck, E7XXX_DRAM_UELOG_ADD,
- &info->dram_uelog_add);
+ &info->dram_uelog_add);
if (info->dram_ferr & 3)
- pci_write_bits8(pvt->bridge_ck, E7XXX_DRAM_FERR, 0x03,
- 0x03);
+ pci_write_bits8(pvt->bridge_ck, E7XXX_DRAM_FERR, 0x03, 0x03);
if (info->dram_nerr & 3)
- pci_write_bits8(pvt->bridge_ck, E7XXX_DRAM_NERR, 0x03,
- 0x03);
+ 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;
return error_found;
}
-
static void e7xxx_check(struct mem_ctl_info *mci)
{
struct e7xxx_error_info 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__);
+ return (dev_idx == E7501) ? ((drc >> 22) & 0x1) : 1;
+}
- /* need to find out the number of channels */
- pci_read_config_dword(pdev, E7XXX_DRC, &drc);
+/* 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;
- }
-
- debugf3("%s(): more mci init\n", __func__);
- mci->ctl_name = pvt->dev_info->ctl_name;
+ return (dev_idx == E7501) ? ((drc >> 18) & 0x3) : 1;
+}
- 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;
+ u8 value;
+ u32 dra, cumul_size;
+ int drc_chan, drc_drbg, drc_ddim, mem_dev;
+ struct csrow_info *csrow;
- /* 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,
} else
csrow->edac_mode = EDAC_NONE;
}
+}
- 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 e7xxx_pvt *pvt = NULL;
+ u32 drc;
+ int drc_chan;
+ struct e7xxx_error_info discard;
+
+ debugf0("%s(): mci\n", __func__);
+
+ /* make sure error reporting method is sane */
+ switch (edac_op_state) {
+ case EDAC_OPSTATE_POLL:
+ case EDAC_OPSTATE_NMI:
+ break;
+ default:
+ edac_op_state = EDAC_OPSTATE_POLL;
+ break;
+ }
+
+ pci_read_config_dword(pdev, E7XXX_DRC, &drc);
+
+ drc_chan = dual_channel_active(drc, dev_idx);
+ mci = edac_mc_alloc(sizeof(*pvt), E7XXX_NR_CSROWS, drc_chan + 1, 0);
+
+ if (mci == NULL)
+ return -ENOMEM;
+
+ 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 = E7XXX_REVISION;
+ mci->dev = &pdev->dev;
+ 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 fail0;
+ }
+
+ debugf3("%s(): more mci init\n", __func__);
+ 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;
debugf3("%s(): tolm, remapbase, remaplimit\n", __func__);
/* 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, pvt->remapbase, pvt->remaplimit);
+ "tolm = %x, remapbase = %x, remaplimit = %x\n", pvt->tolm,
+ pvt->remapbase, pvt->remaplimit);
/* clear any pending errors, or initial state bits */
e7xxx_get_error_info(mci, &discard);
- if (edac_mc_add_mc(mci) != 0) {
+ /* 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)) {
debugf3("%s(): failed edac_mc_add_mc()\n", __func__);
- goto fail;
+ 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__);
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);
- return rc;
+fail0:
+ edac_mc_free(mci);
+
+ 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 __devinit e7xxx_init_one(struct pci_dev *pdev,
+ const struct pci_device_id *ent)
{
debugf0("%s()\n", __func__);
/* wake up and enable device */
return pci_enable_device(pdev) ?
- -EIO : e7xxx_probe1(pdev, ent->driver_data);
+ -EIO : e7xxx_probe1(pdev, ent->driver_data);
}
-
static void __devexit e7xxx_remove_one(struct pci_dev *pdev)
{
struct mem_ctl_info *mci;
debugf0("%s()\n", __func__);
- if (((mci = edac_mc_find_mci_by_pdev(pdev)) != 0) &&
- !edac_mc_del_mc(mci)) {
- pvt = (struct e7xxx_pvt *) mci->pvt_info;
- pci_dev_put(pvt->bridge_ck);
- edac_mc_free(mci);
- }
-}
+ 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;
+ pci_dev_put(pvt->bridge_ck);
+ edac_mc_free(mci);
+}
static const struct pci_device_id e7xxx_pci_tbl[] __devinitdata = {
- {PCI_VEND_DEV(INTEL, 7205_0), PCI_ANY_ID, PCI_ANY_ID, 0, 0,
+ {
+ 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,
+ {
+ 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,
+ {
+ 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,
+ {
+ 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);
-
static struct pci_driver e7xxx_driver = {
.name = EDAC_MOD_STR,
.probe = e7xxx_init_one,
.id_table = e7xxx_pci_tbl,
};
-
static int __init e7xxx_init(void)
{
return pci_register_driver(&e7xxx_driver);
}
-
static void __exit e7xxx_exit(void)
{
pci_unregister_driver(&e7xxx_driver);
module_init(e7xxx_init);
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");
git.openpandora.org / pandora-kernel.git / blobdiff