linux_dsm_epyc7002/drivers/edac/e7xxx_edac.c

605 lines
16 KiB
C
Raw Normal View History

/*
* Intel e7xxx Memory Controller kernel module
* (C) 2003 Linux Networx (http://lnxi.com)
* This file may be distributed under the terms of the
* GNU General Public License.
*
* See "enum e7xxx_chips" below for supported chipsets
*
* Written by Thayne Harbaugh
* 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)
* Jim Garlick (Lawrence Livermore National Labs)
* Dave Peterson (Lawrence Livermore National Labs)
* That One Guy (Some other place)
* Wang Zhenyu (intel.com)
*
* $Id: edac_e7xxx.c,v 1.5.2.9 2005/10/05 00:43:44 dsp_llnl Exp $
*
*/
#include <linux/module.h>
#include <linux/init.h>
#include <linux/pci.h>
#include <linux/pci_ids.h>
#include <linux/edac.h>
#include "edac_module.h"
#define EDAC_MOD_STR "e7xxx_edac"
#define e7xxx_printk(level, 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)
#ifndef PCI_DEVICE_ID_INTEL_7205_0
#define PCI_DEVICE_ID_INTEL_7205_0 0x255d
#endif /* PCI_DEVICE_ID_INTEL_7205_0 */
#ifndef PCI_DEVICE_ID_INTEL_7205_1_ERR
#define PCI_DEVICE_ID_INTEL_7205_1_ERR 0x2551
#endif /* PCI_DEVICE_ID_INTEL_7205_1_ERR */
#ifndef PCI_DEVICE_ID_INTEL_7500_0
#define PCI_DEVICE_ID_INTEL_7500_0 0x2540
#endif /* PCI_DEVICE_ID_INTEL_7500_0 */
#ifndef PCI_DEVICE_ID_INTEL_7500_1_ERR
#define PCI_DEVICE_ID_INTEL_7500_1_ERR 0x2541
#endif /* PCI_DEVICE_ID_INTEL_7500_1_ERR */
#ifndef PCI_DEVICE_ID_INTEL_7501_0
#define PCI_DEVICE_ID_INTEL_7501_0 0x254c
#endif /* PCI_DEVICE_ID_INTEL_7501_0 */
#ifndef PCI_DEVICE_ID_INTEL_7501_1_ERR
#define PCI_DEVICE_ID_INTEL_7501_1_ERR 0x2541
#endif /* PCI_DEVICE_ID_INTEL_7501_1_ERR */
#ifndef PCI_DEVICE_ID_INTEL_7505_0
#define PCI_DEVICE_ID_INTEL_7505_0 0x2550
#endif /* PCI_DEVICE_ID_INTEL_7505_0 */
#ifndef PCI_DEVICE_ID_INTEL_7505_1_ERR
#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 /* 2 channels, 4 dimms/channel */
/* 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) */
/*
* 31 Device width row 7 0=x8 1=x4
* 27 Device width row 6
* 23 Device width row 5
* 19 Device width row 4
* 15 Device width row 3
* 11 Device width row 2
* 7 Device width row 1
* 3 Device width row 0
*/
#define E7XXX_DRC 0x7C /* DRAM controller mode reg (32b) */
/*
* 22 Number channels 0=1,1=2
* 19:18 DRB Granularity 32/64MB
*/
#define E7XXX_TOLM 0xC4 /* DRAM top of low memory reg (16b) */
#define E7XXX_REMAPBASE 0xC6 /* DRAM remap base address reg (16b) */
#define E7XXX_REMAPLIMIT 0xC8 /* DRAM remap limit address reg (16b) */
/* E7XXX register addresses - device 0 function 1 */
#define E7XXX_DRAM_FERR 0x80 /* DRAM first error register (8b) */
#define E7XXX_DRAM_NERR 0x82 /* DRAM next error register (8b) */
#define E7XXX_DRAM_CELOG_ADD 0xA0 /* DRAM first correctable memory */
/* error address register (32b) */
/*
* 31:28 Reserved
* 27:6 CE address (4k block 33:12)
* 5:0 Reserved
*/
#define E7XXX_DRAM_UELOG_ADD 0xB0 /* DRAM first uncorrectable memory */
/* error address register (32b) */
/*
* 31:28 Reserved
* 27:6 CE address (4k block 33:12)
* 5:0 Reserved
*/
#define E7XXX_DRAM_CELOG_SYNDROME 0xD0 /* DRAM first correctable memory */
/* error syndrome register (16b) */
enum e7xxx_chips {
E7500 = 0,
E7501,
E7505,
E7205,
};
struct e7xxx_pvt {
struct pci_dev *bridge_ck;
u32 tolm;
u32 remapbase;
u32 remaplimit;
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_celog_add;
u16 dram_celog_syndrome;
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"},
[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"},
[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)
{
edac_dbg(3, "\n");
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)
{
u32 remap;
struct e7xxx_pvt *pvt = (struct e7xxx_pvt *)mci->pvt_info;
edac_dbg(3, "\n");
if ((page < pvt->tolm) ||
((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;
u16 syndrome;
int row;
int channel;
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 */
/* 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_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)
{
edac_dbg(3, "\n");
edac_mc_handle_error(HW_EVENT_ERR_CORRECTED, 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)
{
u32 error_2b, block_page;
int row;
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 */
row = edac_mc_find_csrow_by_page(mci, block_page);
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)
{
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)
{
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);
if ((info->dram_ferr & 1) || (info->dram_nerr & 1)) {
pci_read_config_dword(pvt->bridge_ck, E7XXX_DRAM_CELOG_ADD,
&info->dram_celog_add);
pci_read_config_word(pvt->bridge_ck,
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);
if (info->dram_ferr & 3)
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);
}
static int e7xxx_process_error_info(struct mem_ctl_info *mci,
struct e7xxx_error_info *info,
int handle_errors)
{
int error_found;
error_found = 0;
/* decode and report errors */
if (info->dram_ferr & 1) { /* check first error correctable */
error_found = 1;
if (handle_errors)
process_ce(mci, info);
}
if (info->dram_ferr & 2) { /* check first error uncorrectable */
error_found = 1;
if (handle_errors)
process_ue(mci, info);
}
if (info->dram_nerr & 1) { /* check next error correctable */
error_found = 1;
if (handle_errors) {
if (info->dram_ferr & 1)
process_ce_no_info(mci);
else
process_ce(mci, info);
}
}
if (info->dram_nerr & 2) { /* check next error uncorrectable */
error_found = 1;
if (handle_errors) {
if (info->dram_ferr & 2)
process_ue_no_info(mci);
else
process_ue(mci, info);
}
}
return error_found;
}
static void e7xxx_check(struct mem_ctl_info *mci)
{
struct e7xxx_error_info info;
edac_dbg(3, "\n");
e7xxx_get_error_info(mci, &info);
e7xxx_process_error_info(mci, &info, 1);
}
/* Return 1 if dual channel mode is active. Else return 0. */
static inline int dual_channel_active(u32 drc, int dev_idx)
{
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 */
return (dev_idx == E7501) ? ((drc >> 18) & 0x3) : 1;
}
static void e7xxx_init_csrows(struct mem_ctl_info *mci, struct pci_dev *pdev,
int dev_idx, u32 drc)
{
unsigned long last_cumul_size;
edac: move dimm properties to struct dimm_info On systems based on chip select rows, all channels need to use memories with the same properties, otherwise the memories on channels A and B won't be recognized. However, such assumption is not true for all types of memory controllers. Controllers for FB-DIMM's don't have such requirements. Also, modern Intel controllers seem to be capable of handling such differences. So, we need to get rid of storing the DIMM information into a per-csrow data, storing it, instead at the right place. The first step is to move grain, mtype, dtype and edac_mode to the per-dimm struct. Reviewed-by: Aristeu Rozanski <arozansk@redhat.com> Reviewed-by: Borislav Petkov <borislav.petkov@amd.com> Acked-by: Chris Metcalf <cmetcalf@tilera.com> Cc: Doug Thompson <norsk5@yahoo.com> Cc: Borislav Petkov <borislav.petkov@amd.com> Cc: Mark Gross <mark.gross@intel.com> Cc: Jason Uhlenkott <juhlenko@akamai.com> Cc: Tim Small <tim@buttersideup.com> Cc: Ranganathan Desikan <ravi@jetztechnologies.com> Cc: "Arvind R." <arvino55@gmail.com> Cc: Olof Johansson <olof@lixom.net> Cc: Egor Martovetsky <egor@pasemi.com> Cc: Michal Marek <mmarek@suse.cz> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Joe Perches <joe@perches.com> Cc: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Hitoshi Mitake <h.mitake@gmail.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: James Bottomley <James.Bottomley@parallels.com> Cc: "Niklas Söderlund" <niklas.soderlund@ericsson.com> Cc: Shaohui Xie <Shaohui.Xie@freescale.com> Cc: Josh Boyer <jwboyer@gmail.com> Cc: Mike Williams <mike@mikebwilliams.com> Cc: linuxppc-dev@lists.ozlabs.org Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-01-28 04:38:08 +07:00
int index, j;
u8 value;
u32 dra, cumul_size, nr_pages;
int drc_chan, drc_drbg, drc_ddim, mem_dev;
struct csrow_info *csrow;
edac: move dimm properties to struct dimm_info On systems based on chip select rows, all channels need to use memories with the same properties, otherwise the memories on channels A and B won't be recognized. However, such assumption is not true for all types of memory controllers. Controllers for FB-DIMM's don't have such requirements. Also, modern Intel controllers seem to be capable of handling such differences. So, we need to get rid of storing the DIMM information into a per-csrow data, storing it, instead at the right place. The first step is to move grain, mtype, dtype and edac_mode to the per-dimm struct. Reviewed-by: Aristeu Rozanski <arozansk@redhat.com> Reviewed-by: Borislav Petkov <borislav.petkov@amd.com> Acked-by: Chris Metcalf <cmetcalf@tilera.com> Cc: Doug Thompson <norsk5@yahoo.com> Cc: Borislav Petkov <borislav.petkov@amd.com> Cc: Mark Gross <mark.gross@intel.com> Cc: Jason Uhlenkott <juhlenko@akamai.com> Cc: Tim Small <tim@buttersideup.com> Cc: Ranganathan Desikan <ravi@jetztechnologies.com> Cc: "Arvind R." <arvino55@gmail.com> Cc: Olof Johansson <olof@lixom.net> Cc: Egor Martovetsky <egor@pasemi.com> Cc: Michal Marek <mmarek@suse.cz> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Joe Perches <joe@perches.com> Cc: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Hitoshi Mitake <h.mitake@gmail.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: James Bottomley <James.Bottomley@parallels.com> Cc: "Niklas Söderlund" <niklas.soderlund@ericsson.com> Cc: Shaohui Xie <Shaohui.Xie@freescale.com> Cc: Josh Boyer <jwboyer@gmail.com> Cc: Mike Williams <mike@mikebwilliams.com> Cc: linuxppc-dev@lists.ozlabs.org Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-01-28 04:38:08 +07:00
struct dimm_info *dimm;
enum edac_type edac_mode;
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
* 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 (index = 0; index < mci->nr_csrows; index++) {
/* mem_dev 0=x8, 1=x4 */
mem_dev = (dra >> (index * 4 + 3)) & 0x1;
csrow = mci->csrows[index];
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);
edac_dbg(3, "(%d) cumul_size 0x%x\n", index, cumul_size);
if (cumul_size == last_cumul_size)
continue; /* not populated */
csrow->first_page = last_cumul_size;
csrow->last_page = cumul_size - 1;
nr_pages = cumul_size - last_cumul_size;
last_cumul_size = cumul_size;
edac: move dimm properties to struct dimm_info On systems based on chip select rows, all channels need to use memories with the same properties, otherwise the memories on channels A and B won't be recognized. However, such assumption is not true for all types of memory controllers. Controllers for FB-DIMM's don't have such requirements. Also, modern Intel controllers seem to be capable of handling such differences. So, we need to get rid of storing the DIMM information into a per-csrow data, storing it, instead at the right place. The first step is to move grain, mtype, dtype and edac_mode to the per-dimm struct. Reviewed-by: Aristeu Rozanski <arozansk@redhat.com> Reviewed-by: Borislav Petkov <borislav.petkov@amd.com> Acked-by: Chris Metcalf <cmetcalf@tilera.com> Cc: Doug Thompson <norsk5@yahoo.com> Cc: Borislav Petkov <borislav.petkov@amd.com> Cc: Mark Gross <mark.gross@intel.com> Cc: Jason Uhlenkott <juhlenko@akamai.com> Cc: Tim Small <tim@buttersideup.com> Cc: Ranganathan Desikan <ravi@jetztechnologies.com> Cc: "Arvind R." <arvino55@gmail.com> Cc: Olof Johansson <olof@lixom.net> Cc: Egor Martovetsky <egor@pasemi.com> Cc: Michal Marek <mmarek@suse.cz> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Joe Perches <joe@perches.com> Cc: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Hitoshi Mitake <h.mitake@gmail.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: James Bottomley <James.Bottomley@parallels.com> Cc: "Niklas Söderlund" <niklas.soderlund@ericsson.com> Cc: Shaohui Xie <Shaohui.Xie@freescale.com> Cc: Josh Boyer <jwboyer@gmail.com> Cc: Mike Williams <mike@mikebwilliams.com> Cc: linuxppc-dev@lists.ozlabs.org Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-01-28 04:38:08 +07:00
/*
* if single channel or x8 devices then SECDED
* if dual channel and x4 then S4ECD4ED
*/
if (drc_ddim) {
if (drc_chan && mem_dev) {
edac_mode = EDAC_S4ECD4ED;
mci->edac_cap |= EDAC_FLAG_S4ECD4ED;
} else {
edac_mode = EDAC_SECDED;
mci->edac_cap |= EDAC_FLAG_SECDED;
}
} else
edac_mode = EDAC_NONE;
edac: move dimm properties to struct dimm_info On systems based on chip select rows, all channels need to use memories with the same properties, otherwise the memories on channels A and B won't be recognized. However, such assumption is not true for all types of memory controllers. Controllers for FB-DIMM's don't have such requirements. Also, modern Intel controllers seem to be capable of handling such differences. So, we need to get rid of storing the DIMM information into a per-csrow data, storing it, instead at the right place. The first step is to move grain, mtype, dtype and edac_mode to the per-dimm struct. Reviewed-by: Aristeu Rozanski <arozansk@redhat.com> Reviewed-by: Borislav Petkov <borislav.petkov@amd.com> Acked-by: Chris Metcalf <cmetcalf@tilera.com> Cc: Doug Thompson <norsk5@yahoo.com> Cc: Borislav Petkov <borislav.petkov@amd.com> Cc: Mark Gross <mark.gross@intel.com> Cc: Jason Uhlenkott <juhlenko@akamai.com> Cc: Tim Small <tim@buttersideup.com> Cc: Ranganathan Desikan <ravi@jetztechnologies.com> Cc: "Arvind R." <arvino55@gmail.com> Cc: Olof Johansson <olof@lixom.net> Cc: Egor Martovetsky <egor@pasemi.com> Cc: Michal Marek <mmarek@suse.cz> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Joe Perches <joe@perches.com> Cc: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Hitoshi Mitake <h.mitake@gmail.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: James Bottomley <James.Bottomley@parallels.com> Cc: "Niklas Söderlund" <niklas.soderlund@ericsson.com> Cc: Shaohui Xie <Shaohui.Xie@freescale.com> Cc: Josh Boyer <jwboyer@gmail.com> Cc: Mike Williams <mike@mikebwilliams.com> Cc: linuxppc-dev@lists.ozlabs.org Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-01-28 04:38:08 +07:00
for (j = 0; j < drc_chan + 1; j++) {
dimm = csrow->channels[j]->dimm;
edac: move dimm properties to struct dimm_info On systems based on chip select rows, all channels need to use memories with the same properties, otherwise the memories on channels A and B won't be recognized. However, such assumption is not true for all types of memory controllers. Controllers for FB-DIMM's don't have such requirements. Also, modern Intel controllers seem to be capable of handling such differences. So, we need to get rid of storing the DIMM information into a per-csrow data, storing it, instead at the right place. The first step is to move grain, mtype, dtype and edac_mode to the per-dimm struct. Reviewed-by: Aristeu Rozanski <arozansk@redhat.com> Reviewed-by: Borislav Petkov <borislav.petkov@amd.com> Acked-by: Chris Metcalf <cmetcalf@tilera.com> Cc: Doug Thompson <norsk5@yahoo.com> Cc: Borislav Petkov <borislav.petkov@amd.com> Cc: Mark Gross <mark.gross@intel.com> Cc: Jason Uhlenkott <juhlenko@akamai.com> Cc: Tim Small <tim@buttersideup.com> Cc: Ranganathan Desikan <ravi@jetztechnologies.com> Cc: "Arvind R." <arvino55@gmail.com> Cc: Olof Johansson <olof@lixom.net> Cc: Egor Martovetsky <egor@pasemi.com> Cc: Michal Marek <mmarek@suse.cz> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Joe Perches <joe@perches.com> Cc: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Hitoshi Mitake <h.mitake@gmail.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: James Bottomley <James.Bottomley@parallels.com> Cc: "Niklas Söderlund" <niklas.soderlund@ericsson.com> Cc: Shaohui Xie <Shaohui.Xie@freescale.com> Cc: Josh Boyer <jwboyer@gmail.com> Cc: Mike Williams <mike@mikebwilliams.com> Cc: linuxppc-dev@lists.ozlabs.org Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-01-28 04:38:08 +07:00
dimm->nr_pages = nr_pages / (drc_chan + 1);
edac: move dimm properties to struct dimm_info On systems based on chip select rows, all channels need to use memories with the same properties, otherwise the memories on channels A and B won't be recognized. However, such assumption is not true for all types of memory controllers. Controllers for FB-DIMM's don't have such requirements. Also, modern Intel controllers seem to be capable of handling such differences. So, we need to get rid of storing the DIMM information into a per-csrow data, storing it, instead at the right place. The first step is to move grain, mtype, dtype and edac_mode to the per-dimm struct. Reviewed-by: Aristeu Rozanski <arozansk@redhat.com> Reviewed-by: Borislav Petkov <borislav.petkov@amd.com> Acked-by: Chris Metcalf <cmetcalf@tilera.com> Cc: Doug Thompson <norsk5@yahoo.com> Cc: Borislav Petkov <borislav.petkov@amd.com> Cc: Mark Gross <mark.gross@intel.com> Cc: Jason Uhlenkott <juhlenko@akamai.com> Cc: Tim Small <tim@buttersideup.com> Cc: Ranganathan Desikan <ravi@jetztechnologies.com> Cc: "Arvind R." <arvino55@gmail.com> Cc: Olof Johansson <olof@lixom.net> Cc: Egor Martovetsky <egor@pasemi.com> Cc: Michal Marek <mmarek@suse.cz> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Joe Perches <joe@perches.com> Cc: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Hitoshi Mitake <h.mitake@gmail.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: James Bottomley <James.Bottomley@parallels.com> Cc: "Niklas Söderlund" <niklas.soderlund@ericsson.com> Cc: Shaohui Xie <Shaohui.Xie@freescale.com> Cc: Josh Boyer <jwboyer@gmail.com> Cc: Mike Williams <mike@mikebwilliams.com> Cc: linuxppc-dev@lists.ozlabs.org Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-01-28 04:38:08 +07:00
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;
edac: move dimm properties to struct dimm_info On systems based on chip select rows, all channels need to use memories with the same properties, otherwise the memories on channels A and B won't be recognized. However, such assumption is not true for all types of memory controllers. Controllers for FB-DIMM's don't have such requirements. Also, modern Intel controllers seem to be capable of handling such differences. So, we need to get rid of storing the DIMM information into a per-csrow data, storing it, instead at the right place. The first step is to move grain, mtype, dtype and edac_mode to the per-dimm struct. Reviewed-by: Aristeu Rozanski <arozansk@redhat.com> Reviewed-by: Borislav Petkov <borislav.petkov@amd.com> Acked-by: Chris Metcalf <cmetcalf@tilera.com> Cc: Doug Thompson <norsk5@yahoo.com> Cc: Borislav Petkov <borislav.petkov@amd.com> Cc: Mark Gross <mark.gross@intel.com> Cc: Jason Uhlenkott <juhlenko@akamai.com> Cc: Tim Small <tim@buttersideup.com> Cc: Ranganathan Desikan <ravi@jetztechnologies.com> Cc: "Arvind R." <arvino55@gmail.com> Cc: Olof Johansson <olof@lixom.net> Cc: Egor Martovetsky <egor@pasemi.com> Cc: Michal Marek <mmarek@suse.cz> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Joe Perches <joe@perches.com> Cc: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Hitoshi Mitake <h.mitake@gmail.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: James Bottomley <James.Bottomley@parallels.com> Cc: "Niklas Söderlund" <niklas.soderlund@ericsson.com> Cc: Shaohui Xie <Shaohui.Xie@freescale.com> Cc: Josh Boyer <jwboyer@gmail.com> Cc: Mike Williams <mike@mikebwilliams.com> Cc: linuxppc-dev@lists.ozlabs.org Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-01-28 04:38:08 +07:00
}
}
}
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->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);
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(pdev, E7XXX_TOLM, &pci_data);
pvt->tolm = ((u32) pci_data) << 4;
pci_read_config_word(pdev, E7XXX_REMAPBASE, &pci_data);
pvt->remapbase = ((u32) pci_data) << 14;
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);
/* clear any pending errors, or initial state bits */
e7xxx_get_error_info(mci, &discard);
/* 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 */
edac_dbg(3, "success\n");
return 0;
fail1:
pci_dev_put(pvt->bridge_ck);
fail0:
edac_mc_free(mci);
return -ENODEV;
}
/* returns count (>= 0), or negative on error */
static int e7xxx_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
{
edac_dbg(0, "\n");
/* wake up and enable device */
return pci_enable_device(pdev) ?
-EIO : e7xxx_probe1(pdev, ent->driver_data);
}
static void e7xxx_remove_one(struct pci_dev *pdev)
{
struct mem_ctl_info *mci;
struct e7xxx_pvt *pvt;
edac_dbg(0, "\n");
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[] = {
{
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, 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},
{
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,
.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);
}
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");
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");