apfs/linux-apfs
0
0
Fork 0
mirror of https://github.com/linux-apfs/linux-apfs.git synced 2026-05-01 15:00:59 -07:00
Code Issues Packages Projects Releases Wiki Activity

Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/bp/bp

Browse Source 
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/bp/bp: (26 commits)
  amd64_edac: add MAINTAINERS entry
  EDAC: do not enable modules by default
  amd64_edac: do not enable module by default
  amd64_edac: add module registration routines
  amd64_edac: add ECC reporting initializers
  amd64_edac: add EDAC core-related initializers
  amd64_edac: add error decoding logic
  amd64_edac: add ECC chipkill syndrome mapping table
  amd64_edac: add per-family descriptors
  amd64_edac: add F10h-and-later methods-p3
  amd64_edac: add F10h-and-later methods-p2
  amd64_edac: add F10h-and-later methods-p1
  amd64_edac: add k8-specific methods
  amd64_edac: assign DRAM chip select base and mask in a family-specific way
  amd64_edac: add helper to dump relevant registers
  amd64_edac: add DRAM address type conversion facilities
  amd64_edac: add functionality to compute the DRAM hole
  amd64_edac: add sys addr to memory controller mapping helpers
  amd64_edac: add memory scrubber interface
  amd64_edac: add MCA error types
  ...
This commit is contained in:
Linus Torvalds
2009-06-11 10:33:36 -07:00
parent d3d07d941f c476c23b45
commit c29f5ec022
13 changed files with 4853 additions and 103 deletions
Download Patch File Download Diff File Expand all files Collapse all files
MAINTAINERS
+10
View File
@@ -1979,6 +1979,16 @@ F: Documentation/edac.txt
F: drivers/edac/edac_*
F: include/linux/edac.h
EDAC-AMD64
P: Doug Thompson
M: dougthompson@xmission.com
P: Borislav Petkov
M: borislav.petkov@amd.com
L: bluesmoke-devel@lists.sourceforge.net (moderated for non-subscribers)
W: bluesmoke.sourceforge.net
S: Supported
F: drivers/edac/amd64_edac*
EDAC-E752X
P: Mark Gross
M: mark.gross@intel.com
arch/x86/include/asm/msr.h
+23
View File
@@ -12,6 +12,17 @@
#include <asm/asm.h>
#include <asm/errno.h>
#include <asm/cpumask.h>
struct msr {
union {
struct {
u32 l;
u32 h;
};
u64 q;
};
};
static inline unsigned long long native_read_tscp(unsigned int *aux)
{
@@ -216,6 +227,8 @@ do { \
#ifdef CONFIG_SMP
int rdmsr_on_cpu(unsigned int cpu, u32 msr_no, u32 *l, u32 *h);
int wrmsr_on_cpu(unsigned int cpu, u32 msr_no, u32 l, u32 h);
void rdmsr_on_cpus(const cpumask_t *mask, u32 msr_no, struct msr *msrs);
void wrmsr_on_cpus(const cpumask_t *mask, u32 msr_no, struct msr *msrs);
int rdmsr_safe_on_cpu(unsigned int cpu, u32 msr_no, u32 *l, u32 *h);
int wrmsr_safe_on_cpu(unsigned int cpu, u32 msr_no, u32 l, u32 h);
#else /* CONFIG_SMP */
@@ -229,6 +242,16 @@ static inline int wrmsr_on_cpu(unsigned int cpu, u32 msr_no, u32 l, u32 h)
wrmsr(msr_no, l, h);
return 0;
}
static inline void rdmsr_on_cpus(const cpumask_t *m, u32 msr_no,
struct msr *msrs)
{
rdmsr_on_cpu(0, msr_no, &(msrs[0].l), &(msrs[0].h));
}
static inline void wrmsr_on_cpus(const cpumask_t *m, u32 msr_no,
struct msr *msrs)
{
wrmsr_on_cpu(0, msr_no, msrs[0].l, msrs[0].h);
}
static inline int rdmsr_safe_on_cpu(unsigned int cpu, u32 msr_no,
u32 *l, u32 *h)
{
arch/x86/lib/Makefile
+1 -1
View File
@@ -2,7 +2,7 @@
# Makefile for x86 specific library files.
#
obj-$(CONFIG_SMP) := msr-on-cpu.o
obj-$(CONFIG_SMP) := msr.o
lib-y := delay.o
lib-y += thunk_$(BITS).o
arch/x86/lib/msr-on-cpu.c
-97
View File
@@ -1,97 +0,0 @@
#include <linux/module.h>
#include <linux/preempt.h>
#include <linux/smp.h>
#include <asm/msr.h>
struct msr_info {
u32 msr_no;
u32 l, h;
int err;
};
static void __rdmsr_on_cpu(void *info)
{
struct msr_info *rv = info;
rdmsr(rv->msr_no, rv->l, rv->h);
}
static void __wrmsr_on_cpu(void *info)
{
struct msr_info *rv = info;
wrmsr(rv->msr_no, rv->l, rv->h);
}
int rdmsr_on_cpu(unsigned int cpu, u32 msr_no, u32 *l, u32 *h)
{
int err;
struct msr_info rv;
rv.msr_no = msr_no;
err = smp_call_function_single(cpu, __rdmsr_on_cpu, &rv, 1);
*l = rv.l;
*h = rv.h;
return err;
}
int wrmsr_on_cpu(unsigned int cpu, u32 msr_no, u32 l, u32 h)
{
int err;
struct msr_info rv;
rv.msr_no = msr_no;
rv.l = l;
rv.h = h;
err = smp_call_function_single(cpu, __wrmsr_on_cpu, &rv, 1);
return err;
}
/* These "safe" variants are slower and should be used when the target MSR
may not actually exist. */
static void __rdmsr_safe_on_cpu(void *info)
{
struct msr_info *rv = info;
rv->err = rdmsr_safe(rv->msr_no, &rv->l, &rv->h);
}
static void __wrmsr_safe_on_cpu(void *info)
{
struct msr_info *rv = info;
rv->err = wrmsr_safe(rv->msr_no, rv->l, rv->h);
}
int rdmsr_safe_on_cpu(unsigned int cpu, u32 msr_no, u32 *l, u32 *h)
{
int err;
struct msr_info rv;
rv.msr_no = msr_no;
err = smp_call_function_single(cpu, __rdmsr_safe_on_cpu, &rv, 1);
*l = rv.l;
*h = rv.h;
return err ? err : rv.err;
}
int wrmsr_safe_on_cpu(unsigned int cpu, u32 msr_no, u32 l, u32 h)
{
int err;
struct msr_info rv;
rv.msr_no = msr_no;
rv.l = l;
rv.h = h;
err = smp_call_function_single(cpu, __wrmsr_safe_on_cpu, &rv, 1);
return err ? err : rv.err;
}
EXPORT_SYMBOL(rdmsr_on_cpu);
EXPORT_SYMBOL(wrmsr_on_cpu);
EXPORT_SYMBOL(rdmsr_safe_on_cpu);
EXPORT_SYMBOL(wrmsr_safe_on_cpu);
arch/x86/lib/msr.c
+183
View File
@@ -0,0 +1,183 @@
#include <linux/module.h>
#include <linux/preempt.h>
#include <linux/smp.h>
#include <asm/msr.h>
struct msr_info {
u32 msr_no;
struct msr reg;
struct msr *msrs;
int off;
int err;
};
static void __rdmsr_on_cpu(void *info)
{
struct msr_info *rv = info;
struct msr *reg;
int this_cpu = raw_smp_processor_id();
if (rv->msrs)
reg = &rv->msrs[this_cpu - rv->off];
else
reg = &rv->reg;
rdmsr(rv->msr_no, reg->l, reg->h);
}
static void __wrmsr_on_cpu(void *info)
{
struct msr_info *rv = info;
struct msr *reg;
int this_cpu = raw_smp_processor_id();
if (rv->msrs)
reg = &rv->msrs[this_cpu - rv->off];
else
reg = &rv->reg;
wrmsr(rv->msr_no, reg->l, reg->h);
}
int rdmsr_on_cpu(unsigned int cpu, u32 msr_no, u32 *l, u32 *h)
{
int err;
struct msr_info rv;
memset(&rv, 0, sizeof(rv));
rv.msr_no = msr_no;
err = smp_call_function_single(cpu, __rdmsr_on_cpu, &rv, 1);
*l = rv.reg.l;
*h = rv.reg.h;
return err;
}
EXPORT_SYMBOL(rdmsr_on_cpu);
int wrmsr_on_cpu(unsigned int cpu, u32 msr_no, u32 l, u32 h)
{
int err;
struct msr_info rv;
memset(&rv, 0, sizeof(rv));
rv.msr_no = msr_no;
rv.reg.l = l;
rv.reg.h = h;
err = smp_call_function_single(cpu, __wrmsr_on_cpu, &rv, 1);
return err;
}
EXPORT_SYMBOL(wrmsr_on_cpu);
/* rdmsr on a bunch of CPUs
*
* @mask: which CPUs
* @msr_no: which MSR
* @msrs: array of MSR values
*
*/
void rdmsr_on_cpus(const cpumask_t *mask, u32 msr_no, struct msr *msrs)
{
struct msr_info rv;
int this_cpu;
memset(&rv, 0, sizeof(rv));
rv.off = cpumask_first(mask);
rv.msrs = msrs;
rv.msr_no = msr_no;
preempt_disable();
/*
* FIXME: handle the CPU we're executing on separately for now until
* smp_call_function_many has been fixed to not skip it.
*/
this_cpu = raw_smp_processor_id();
smp_call_function_single(this_cpu, __rdmsr_on_cpu, &rv, 1);
smp_call_function_many(mask, __rdmsr_on_cpu, &rv, 1);
preempt_enable();
}
EXPORT_SYMBOL(rdmsr_on_cpus);
/*
* wrmsr on a bunch of CPUs
*
* @mask: which CPUs
* @msr_no: which MSR
* @msrs: array of MSR values
*
*/
void wrmsr_on_cpus(const cpumask_t *mask, u32 msr_no, struct msr *msrs)
{
struct msr_info rv;
int this_cpu;
memset(&rv, 0, sizeof(rv));
rv.off = cpumask_first(mask);
rv.msrs = msrs;
rv.msr_no = msr_no;
preempt_disable();
/*
* FIXME: handle the CPU we're executing on separately for now until
* smp_call_function_many has been fixed to not skip it.
*/
this_cpu = raw_smp_processor_id();
smp_call_function_single(this_cpu, __wrmsr_on_cpu, &rv, 1);
smp_call_function_many(mask, __wrmsr_on_cpu, &rv, 1);
preempt_enable();
}
EXPORT_SYMBOL(wrmsr_on_cpus);
/* These "safe" variants are slower and should be used when the target MSR
may not actually exist. */
static void __rdmsr_safe_on_cpu(void *info)
{
struct msr_info *rv = info;
rv->err = rdmsr_safe(rv->msr_no, &rv->reg.l, &rv->reg.h);
}
static void __wrmsr_safe_on_cpu(void *info)
{
struct msr_info *rv = info;
rv->err = wrmsr_safe(rv->msr_no, rv->reg.l, rv->reg.h);
}
int rdmsr_safe_on_cpu(unsigned int cpu, u32 msr_no, u32 *l, u32 *h)
{
int err;
struct msr_info rv;
memset(&rv, 0, sizeof(rv));
rv.msr_no = msr_no;
err = smp_call_function_single(cpu, __rdmsr_safe_on_cpu, &rv, 1);
*l = rv.reg.l;
*h = rv.reg.h;
return err ? err : rv.err;
}
EXPORT_SYMBOL(rdmsr_safe_on_cpu);
int wrmsr_safe_on_cpu(unsigned int cpu, u32 msr_no, u32 l, u32 h)
{
int err;
struct msr_info rv;
memset(&rv, 0, sizeof(rv));
rv.msr_no = msr_no;
rv.reg.l = l;
rv.reg.h = h;
err = smp_call_function_single(cpu, __wrmsr_safe_on_cpu, &rv, 1);
return err ? err : rv.err;
}
EXPORT_SYMBOL(wrmsr_safe_on_cpu);
drivers/edac/Kconfig
+25 -1
View File
@@ -49,7 +49,6 @@ config EDAC_DEBUG_VERBOSE
config EDAC_MM_EDAC
tristate "Main Memory EDAC (Error Detection And Correction) reporting"
default y
help
Some systems are able to detect and correct errors in main
memory. EDAC can report statistics on memory error
@@ -58,6 +57,31 @@ config EDAC_MM_EDAC
occurred so that a particular failing memory module can be
replaced. If unsure, select 'Y'.
config EDAC_AMD64
tristate "AMD64 (Opteron, Athlon64) K8, F10h, F11h"
depends on EDAC_MM_EDAC && K8_NB && X86_64 && PCI
help
Support for error detection and correction on the AMD 64
Families of Memory Controllers (K8, F10h and F11h)
config EDAC_AMD64_ERROR_INJECTION
bool "Sysfs Error Injection facilities"
depends on EDAC_AMD64
help
Recent Opterons (Family 10h and later) provide for Memory Error
Injection into the ECC detection circuits. The amd64_edac module
allows the operator/user to inject Uncorrectable and Correctable
errors into DRAM.
When enabled, in each of the respective memory controller directories
(/sys/devices/system/edac/mc/mcX), there are 3 input files:
- inject_section (0..3, 16-byte section of 64-byte cacheline),
- inject_word (0..8, 16-bit word of 16-byte section),
- inject_ecc_vector (hex ecc vector: select bits of inject word)
In addition, there are two control files, inject_read and inject_write,
which trigger the DRAM ECC Read and Write respectively.
config EDAC_AMD76X
tristate "AMD 76x (760, 762, 768)"
drivers/edac/Makefile
+7
View File
@@ -30,6 +30,13 @@ obj-$(CONFIG_EDAC_I3000) += i3000_edac.o
obj-$(CONFIG_EDAC_X38) += x38_edac.o
obj-$(CONFIG_EDAC_I82860) += i82860_edac.o
obj-$(CONFIG_EDAC_R82600) += r82600_edac.o
amd64_edac_mod-y := amd64_edac_err_types.o amd64_edac.o
amd64_edac_mod-$(CONFIG_EDAC_DEBUG) += amd64_edac_dbg.o
amd64_edac_mod-$(CONFIG_EDAC_AMD64_ERROR_INJECTION) += amd64_edac_inj.o
obj-$(CONFIG_EDAC_AMD64) += amd64_edac_mod.o
obj-$(CONFIG_EDAC_PASEMI) += pasemi_edac.o
obj-$(CONFIG_EDAC_MPC85XX) += mpc85xx_edac.o
obj-$(CONFIG_EDAC_MV64X60) += mv64x60_edac.o
drivers/edac/amd64_edac.c
+3354
View File
File diff suppressed because it is too large Load Diff
drivers/edac/amd64_edac.h
+644
View File
File diff suppressed because it is too large Load Diff
drivers/edac/amd64_edac_dbg.c
+255
View File
@@ -0,0 +1,255 @@
#include "amd64_edac.h"
/*
* accept a hex value and store it into the virtual error register file, field:
* nbeal and nbeah. Assume virtual error values have already been set for: NBSL,
* NBSH and NBCFG. Then proceed to map the error values to a MC, CSROW and
* CHANNEL
*/
static ssize_t amd64_nbea_store(struct mem_ctl_info *mci, const char *data,
size_t count)
{
struct amd64_pvt *pvt = mci->pvt_info;
unsigned long long value;
int ret = 0;
ret = strict_strtoull(data, 16, &value);
if (ret != -EINVAL) {
debugf0("received NBEA= 0x%llx\n", value);
/* place the value into the virtual error packet */
pvt->ctl_error_info.nbeal = (u32) value;
value >>= 32;
pvt->ctl_error_info.nbeah = (u32) value;
/* Process the Mapping request */
/* TODO: Add race prevention */
amd64_process_error_info(mci, &pvt->ctl_error_info, 1);
return count;
}
return ret;
}
/* display back what the last NBEA (MCA NB Address (MC4_ADDR)) was written */
static ssize_t amd64_nbea_show(struct mem_ctl_info *mci, char *data)
{
struct amd64_pvt *pvt = mci->pvt_info;
u64 value;
value = pvt->ctl_error_info.nbeah;
value <<= 32;
value |= pvt->ctl_error_info.nbeal;
return sprintf(data, "%llx\n", value);
}
/* store the NBSL (MCA NB Status Low (MC4_STATUS)) value user desires */
static ssize_t amd64_nbsl_store(struct mem_ctl_info *mci, const char *data,
size_t count)
{
struct amd64_pvt *pvt = mci->pvt_info;
unsigned long value;
int ret = 0;
ret = strict_strtoul(data, 16, &value);
if (ret != -EINVAL) {
debugf0("received NBSL= 0x%lx\n", value);
pvt->ctl_error_info.nbsl = (u32) value;
return count;
}
return ret;
}
/* display back what the last NBSL value written */
static ssize_t amd64_nbsl_show(struct mem_ctl_info *mci, char *data)
{
struct amd64_pvt *pvt = mci->pvt_info;
u32 value;
value = pvt->ctl_error_info.nbsl;
return sprintf(data, "%x\n", value);
}
/* store the NBSH (MCA NB Status High) value user desires */
static ssize_t amd64_nbsh_store(struct mem_ctl_info *mci, const char *data,
size_t count)
{
struct amd64_pvt *pvt = mci->pvt_info;
unsigned long value;
int ret = 0;
ret = strict_strtoul(data, 16, &value);
if (ret != -EINVAL) {
debugf0("received NBSH= 0x%lx\n", value);
pvt->ctl_error_info.nbsh = (u32) value;
return count;
}
return ret;
}
/* display back what the last NBSH value written */
static ssize_t amd64_nbsh_show(struct mem_ctl_info *mci, char *data)
{
struct amd64_pvt *pvt = mci->pvt_info;
u32 value;
value = pvt->ctl_error_info.nbsh;
return sprintf(data, "%x\n", value);
}
/* accept and store the NBCFG (MCA NB Configuration) value user desires */
static ssize_t amd64_nbcfg_store(struct mem_ctl_info *mci,
const char *data, size_t count)
{
struct amd64_pvt *pvt = mci->pvt_info;
unsigned long value;
int ret = 0;
ret = strict_strtoul(data, 16, &value);
if (ret != -EINVAL) {
debugf0("received NBCFG= 0x%lx\n", value);
pvt->ctl_error_info.nbcfg = (u32) value;
return count;
}
return ret;
}
/* various show routines for the controls of a MCI */
static ssize_t amd64_nbcfg_show(struct mem_ctl_info *mci, char *data)
{
struct amd64_pvt *pvt = mci->pvt_info;
return sprintf(data, "%x\n", pvt->ctl_error_info.nbcfg);
}
static ssize_t amd64_dhar_show(struct mem_ctl_info *mci, char *data)
{
struct amd64_pvt *pvt = mci->pvt_info;
return sprintf(data, "%x\n", pvt->dhar);
}
static ssize_t amd64_dbam_show(struct mem_ctl_info *mci, char *data)
{
struct amd64_pvt *pvt = mci->pvt_info;
return sprintf(data, "%x\n", pvt->dbam0);
}
static ssize_t amd64_topmem_show(struct mem_ctl_info *mci, char *data)
{
struct amd64_pvt *pvt = mci->pvt_info;
return sprintf(data, "%llx\n", pvt->top_mem);
}
static ssize_t amd64_topmem2_show(struct mem_ctl_info *mci, char *data)
{
struct amd64_pvt *pvt = mci->pvt_info;
return sprintf(data, "%llx\n", pvt->top_mem2);
}
static ssize_t amd64_hole_show(struct mem_ctl_info *mci, char *data)
{
u64 hole_base = 0;
u64 hole_offset = 0;
u64 hole_size = 0;
amd64_get_dram_hole_info(mci, &hole_base, &hole_offset, &hole_size);
return sprintf(data, "%llx %llx %llx\n", hole_base, hole_offset,
hole_size);
}
/*
* update NUM_DBG_ATTRS in case you add new members
*/
struct mcidev_sysfs_attribute amd64_dbg_attrs[] = {
{
.attr = {
.name = "nbea_ctl",
.mode = (S_IRUGO | S_IWUSR)
},
.show = amd64_nbea_show,
.store = amd64_nbea_store,
},
{
.attr = {
.name = "nbsl_ctl",
.mode = (S_IRUGO | S_IWUSR)
},
.show = amd64_nbsl_show,
.store = amd64_nbsl_store,
},
{
.attr = {
.name = "nbsh_ctl",
.mode = (S_IRUGO | S_IWUSR)
},
.show = amd64_nbsh_show,
.store = amd64_nbsh_store,
},
{
.attr = {
.name = "nbcfg_ctl",
.mode = (S_IRUGO | S_IWUSR)
},
.show = amd64_nbcfg_show,
.store = amd64_nbcfg_store,
},
{
.attr = {
.name = "dhar",
.mode = (S_IRUGO)
},
.show = amd64_dhar_show,
.store = NULL,
},
{
.attr = {
.name = "dbam",
.mode = (S_IRUGO)
},
.show = amd64_dbam_show,
.store = NULL,
},
{
.attr = {
.name = "topmem",
.mode = (S_IRUGO)
},
.show = amd64_topmem_show,
.store = NULL,
},
{
.attr = {
.name = "topmem2",
.mode = (S_IRUGO)
},
.show = amd64_topmem2_show,
.store = NULL,
},
{
.attr = {
.name = "dram_hole",
.mode = (S_IRUGO)
},
.show = amd64_hole_show,
.store = NULL,
},
};
drivers/edac/amd64_edac_err_types.c
+161
View File
@@ -0,0 +1,161 @@
#include "amd64_edac.h"
/*
* See F2x80 for K8 and F2x[1,0]80 for Fam10 and later. The table below is only
* for DDR2 DRAM mapping.
*/
u32 revf_quad_ddr2_shift[] = {
0, /* 0000b NULL DIMM (128mb) */
28, /* 0001b 256mb */
29, /* 0010b 512mb */
29, /* 0011b 512mb */
29, /* 0100b 512mb */
30, /* 0101b 1gb */
30, /* 0110b 1gb */
31, /* 0111b 2gb */
31, /* 1000b 2gb */
32, /* 1001b 4gb */
32, /* 1010b 4gb */
33, /* 1011b 8gb */
0, /* 1100b future */
0, /* 1101b future */
0, /* 1110b future */
0 /* 1111b future */
};
/*
* Valid scrub rates for the K8 hardware memory scrubber. We map the scrubbing
* bandwidth to a valid bit pattern. The 'set' operation finds the 'matching-
* or higher value'.
*
*FIXME: Produce a better mapping/linearisation.
*/
struct scrubrate scrubrates[] = {
{ 0x01, 1600000000UL},
{ 0x02, 800000000UL},
{ 0x03, 400000000UL},
{ 0x04, 200000000UL},
{ 0x05, 100000000UL},
{ 0x06, 50000000UL},
{ 0x07, 25000000UL},
{ 0x08, 12284069UL},
{ 0x09, 6274509UL},
{ 0x0A, 3121951UL},
{ 0x0B, 1560975UL},
{ 0x0C, 781440UL},
{ 0x0D, 390720UL},
{ 0x0E, 195300UL},
{ 0x0F, 97650UL},
{ 0x10, 48854UL},
{ 0x11, 24427UL},
{ 0x12, 12213UL},
{ 0x13, 6101UL},
{ 0x14, 3051UL},
{ 0x15, 1523UL},
{ 0x16, 761UL},
{ 0x00, 0UL}, /* scrubbing off */
};
/*
* string representation for the different MCA reported error types, see F3x48
* or MSR0000_0411.
*/
const char *tt_msgs[] = { /* transaction type */
"instruction",
"data",
"generic",
"reserved"
};
const char *ll_msgs[] = { /* cache level */
"L0",
"L1",
"L2",
"L3/generic"
};
const char *rrrr_msgs[] = {
"generic",
"generic read",
"generic write",
"data read",
"data write",
"inst fetch",
"prefetch",
"evict",
"snoop",
"reserved RRRR= 9",
"reserved RRRR= 10",
"reserved RRRR= 11",
"reserved RRRR= 12",
"reserved RRRR= 13",
"reserved RRRR= 14",
"reserved RRRR= 15"
};
const char *pp_msgs[] = { /* participating processor */
"local node originated (SRC)",
"local node responded to request (RES)",
"local node observed as 3rd party (OBS)",
"generic"
};
const char *to_msgs[] = {
"no timeout",
"timed out"
};
const char *ii_msgs[] = { /* memory or i/o */
"mem access",
"reserved",
"i/o access",
"generic"
};
/* Map the 5 bits of Extended Error code to the string table. */
const char *ext_msgs[] = { /* extended error */
"K8 ECC error/F10 reserved", /* 0_0000b */
"CRC error", /* 0_0001b */
"sync error", /* 0_0010b */
"mst abort", /* 0_0011b */
"tgt abort", /* 0_0100b */
"GART error", /* 0_0101b */
"RMW error", /* 0_0110b */
"Wdog timer error", /* 0_0111b */
"F10-ECC/K8-Chipkill error", /* 0_1000b */
"DEV Error", /* 0_1001b */
"Link Data error", /* 0_1010b */
"Link or L3 Protocol error", /* 0_1011b */
"NB Array error", /* 0_1100b */
"DRAM Parity error", /* 0_1101b */
"Link Retry/GART Table Walk/DEV Table Walk error", /* 0_1110b */
"Res 0x0ff error", /* 0_1111b */
"Res 0x100 error", /* 1_0000b */
"Res 0x101 error", /* 1_0001b */
"Res 0x102 error", /* 1_0010b */
"Res 0x103 error", /* 1_0011b */
"Res 0x104 error", /* 1_0100b */
"Res 0x105 error", /* 1_0101b */
"Res 0x106 error", /* 1_0110b */
"Res 0x107 error", /* 1_0111b */
"Res 0x108 error", /* 1_1000b */
"Res 0x109 error", /* 1_1001b */
"Res 0x10A error", /* 1_1010b */
"Res 0x10B error", /* 1_1011b */
"L3 Cache Data error", /* 1_1100b */
"L3 CacheTag error", /* 1_1101b */
"L3 Cache LRU error", /* 1_1110b */
"Res 0x1FF error" /* 1_1111b */
};
const char *htlink_msgs[] = {
"none",
"1",
"2",
"1 2",
"3",
"1 3",
"2 3",
"1 2 3"
};
drivers/edac/amd64_edac_inj.c
+185
View File
@@ -0,0 +1,185 @@
#include "amd64_edac.h"
/*
* store error injection section value which refers to one of 4 16-byte sections
* within a 64-byte cacheline
*
* range: 0..3
*/
static ssize_t amd64_inject_section_store(struct mem_ctl_info *mci,
const char *data, size_t count)
{
struct amd64_pvt *pvt = mci->pvt_info;
unsigned long value;
int ret = 0;
ret = strict_strtoul(data, 10, &value);
if (ret != -EINVAL) {
pvt->injection.section = (u32) value;
return count;
}
return ret;
}
/*
* store error injection word value which refers to one of 9 16-bit word of the
* 16-byte (128-bit + ECC bits) section
*
* range: 0..8
*/
static ssize_t amd64_inject_word_store(struct mem_ctl_info *mci,
const char *data, size_t count)
{
struct amd64_pvt *pvt = mci->pvt_info;
unsigned long value;
int ret = 0;
ret = strict_strtoul(data, 10, &value);
if (ret != -EINVAL) {
value = (value <= 8) ? value : 0;
pvt->injection.word = (u32) value;
return count;
}
return ret;
}
/*
* store 16 bit error injection vector which enables injecting errors to the
* corresponding bit within the error injection word above. When used during a
* DRAM ECC read, it holds the contents of the of the DRAM ECC bits.
*/
static ssize_t amd64_inject_ecc_vector_store(struct mem_ctl_info *mci,
const char *data, size_t count)
{
struct amd64_pvt *pvt = mci->pvt_info;
unsigned long value;
int ret = 0;
ret = strict_strtoul(data, 16, &value);
if (ret != -EINVAL) {
pvt->injection.bit_map = (u32) value & 0xFFFF;
return count;
}
return ret;
}
/*
* Do a DRAM ECC read. Assemble staged values in the pvt area, format into
* fields needed by the injection registers and read the NB Array Data Port.
*/
static ssize_t amd64_inject_read_store(struct mem_ctl_info *mci,
const char *data, size_t count)
{
struct amd64_pvt *pvt = mci->pvt_info;
unsigned long value;
u32 section, word_bits;
int ret = 0;
ret = strict_strtoul(data, 10, &value);
if (ret != -EINVAL) {
/* Form value to choose 16-byte section of cacheline */
section = F10_NB_ARRAY_DRAM_ECC |
SET_NB_ARRAY_ADDRESS(pvt->injection.section);
pci_write_config_dword(pvt->misc_f3_ctl,
F10_NB_ARRAY_ADDR, section);
word_bits = SET_NB_DRAM_INJECTION_READ(pvt->injection.word,
pvt->injection.bit_map);
/* Issue 'word' and 'bit' along with the READ request */
pci_write_config_dword(pvt->misc_f3_ctl,
F10_NB_ARRAY_DATA, word_bits);
debugf0("section=0x%x word_bits=0x%x\n", section, word_bits);
return count;
}
return ret;
}
/*
* Do a DRAM ECC write. Assemble staged values in the pvt area and format into
* fields needed by the injection registers.
*/
static ssize_t amd64_inject_write_store(struct mem_ctl_info *mci,
const char *data, size_t count)
{
struct amd64_pvt *pvt = mci->pvt_info;
unsigned long value;
u32 section, word_bits;
int ret = 0;
ret = strict_strtoul(data, 10, &value);
if (ret != -EINVAL) {
/* Form value to choose 16-byte section of cacheline */
section = F10_NB_ARRAY_DRAM_ECC |
SET_NB_ARRAY_ADDRESS(pvt->injection.section);
pci_write_config_dword(pvt->misc_f3_ctl,
F10_NB_ARRAY_ADDR, section);
word_bits = SET_NB_DRAM_INJECTION_WRITE(pvt->injection.word,
pvt->injection.bit_map);
/* Issue 'word' and 'bit' along with the READ request */
pci_write_config_dword(pvt->misc_f3_ctl,
F10_NB_ARRAY_DATA, word_bits);
debugf0("section=0x%x word_bits=0x%x\n", section, word_bits);
return count;
}
return ret;
}
/*
* update NUM_INJ_ATTRS in case you add new members
*/
struct mcidev_sysfs_attribute amd64_inj_attrs[] = {
{
.attr = {
.name = "inject_section",
.mode = (S_IRUGO | S_IWUSR)
},
.show = NULL,
.store = amd64_inject_section_store,
},
{
.attr = {
.name = "inject_word",
.mode = (S_IRUGO | S_IWUSR)
},
.show = NULL,
.store = amd64_inject_word_store,
},
{
.attr = {
.name = "inject_ecc_vector",
.mode = (S_IRUGO | S_IWUSR)
},
.show = NULL,
.store = amd64_inject_ecc_vector_store,
},
{
.attr = {
.name = "inject_write",
.mode = (S_IRUGO | S_IWUSR)
},
.show = NULL,
.store = amd64_inject_write_store,
},
{
.attr = {
.name = "inject_read",
.mode = (S_IRUGO | S_IWUSR)
},
.show = NULL,
.store = amd64_inject_read_store,
},
};
drivers/edac/edac_core.h
+5 -4
View File
@@ -76,10 +76,11 @@
extern int edac_debug_level;
#ifndef CONFIG_EDAC_DEBUG_VERBOSE
#define edac_debug_printk(level, fmt, arg...) \
do { \
if (level <= edac_debug_level) \
edac_printk(KERN_DEBUG, EDAC_DEBUG, fmt, ##arg); \
#define edac_debug_printk(level, fmt, arg...) \
do { \
if (level <= edac_debug_level) \
edac_printk(KERN_DEBUG, EDAC_DEBUG, \
"%s: " fmt, __func__, ##arg); \
} while (0)
#else /* CONFIG_EDAC_DEBUG_VERBOSE */
#define edac_debug_printk(level, fmt, arg...) \