| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <20130903190829.GA312@gmail.com>
Date: Tue, 3 Sep 2013 21:08:29 +0200
From: Ingo Molnar <mingo@...nel.org>
To: Linus Torvalds <torvalds@...ux-foundation.org>
Cc: linux-kernel@...r.kernel.org, Borislav Petkov <bp@...en8.de>,
"H. Peter Anvin" <hpa@...or.com>,
Thomas Gleixner <tglx@...utronix.de>,
Andrew Morton <akpm@...ux-foundation.org>
Subject: [GIT PULL] x86/RAS changes for v3.12
Linus,
Please pull the latest x86-ras-for-linus git tree from:
git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git x86-ras-for-linus
HEAD: ead6fa95b7e9d38b4526503403ba1c029b03dd72 Merge tag 'edac_fixes_for_3.12' of git://git.kernel.org/pub/scm/linux/kernel/git/bp/bp into x86/ras
[ The reason for drivers/ updates is that Boris asked for the
drivers/edac/ changes to go via x86/ras in this cycle. ]
Main changes:
- AMD CPUs:
. Add ECC event decoding support for new F15h models
. Various erratum fixes
. Fix single-channel on dual-channel-controllers bug.
- Intel CPUs:
. UC uncorrectable memory error parsing fix
. Add support for CMC (Corrected Machine Check) 'FF' (Firmware First)
flag in the APEI HEST
- Various cleanups and fixes.
out-of-topic modifications in x86-ras-for-linus:
------------------------------------------------
drivers/acpi/apei/erst.c # cb82a2e: APEI/ERST: Fix error message form
drivers/acpi/apei/ghes.c # cf870c7: mce: acpi/apei: Soft-offline a pa
drivers/acpi/apei/hest.c # 7781544: x86/mce, acpi/apei: Only disable
# 9ad9587: mce: acpi/apei: Add a boot option
# c3d1fb5: mce: acpi/apei: Honour Firmware F
drivers/edac/amd64_edac.c # 4fc06b3: amd64_edac: Fix incorrect wraparo
# 3f0aba4: amd64_edac: Correct erratum 505 r
# a4b4bed: amd64_edac: Get rid of boot_cpu_d
# 18b94f6: amd64_edac: Add ECC decoding supp
# f0a56c4: amd64_edac: Fix single-channel se
drivers/edac/amd64_edac.h # a4b4bed: amd64_edac: Get rid of boot_cpu_d
# 18b94f6: amd64_edac: Add ECC decoding supp
drivers/edac/cpc925_edac.c # 75a9551: cpc925_edac: Use proper array ter
drivers/edac/edac_mc_sysfs.c # c542b53: EDAC: Replace strict_strtol() wit
drivers/edac/i3200_edac.c # 166e933: i3200_edac: Make a local function
drivers/edac/x38_edac.c # e0d391a: x38_edac: Make a local function s
include/linux/mm.h # cf870c7: mce: acpi/apei: Soft-offline a pa
include/linux/pci_ids.h # 6bdaa63: pci_ids: Add PCI device ID functi
mm/memory-failure.c # cf870c7: mce: acpi/apei: Soft-offline a pa
Thanks,
Ingo
------------------>
Aravind Gopalakrishnan (4):
pci_ids: Add PCI device ID functions 3 and 4 for newer F15h models.
x86, amd_nb: Clarify F15h, model 30h GART and L3 support
amd64_edac: Add ECC decoding support for newer F15h models
amd64_edac: Fix incorrect wraparounds
Borislav Petkov (4):
amd64_edac: Fix single-channel setups
APEI/ERST: Fix error message formatting
amd64_edac: Get rid of boot_cpu_data accesses
amd64_edac: Correct erratum 505 range
Jingoo Han (4):
EDAC: Replace strict_strtol() with kstrtol()
i3200_edac: Make a local function static
x38_edac: Make a local function static
cpc925_edac: Use proper array termination
Naveen N. Rao (4):
mce: acpi/apei: Honour Firmware First for MCA banks listed in APEI HEST CMC
mce: acpi/apei: Add a boot option to disable ff mode for corrected errors
mce: acpi/apei: Soft-offline a page on firmware GHES notification
x86/mce, acpi/apei: Only disable banks listed in HEST if mce is configured
Tony Luck (1):
x86/mce: Pay no attention to 'F' bit in MCACOD when parsing 'UC' errors
Documentation/x86/x86_64/boot-options.txt | 5 +
arch/x86/include/asm/acpi.h | 2 +
arch/x86/include/asm/mce.h | 16 +-
arch/x86/kernel/acpi/boot.c | 5 +
arch/x86/kernel/amd_nb.c | 13 +-
arch/x86/kernel/cpu/mcheck/mce-internal.h | 3 +
arch/x86/kernel/cpu/mcheck/mce.c | 28 +++
arch/x86/kernel/cpu/mcheck/mce_intel.c | 42 +++-
drivers/acpi/apei/erst.c | 51 ++---
drivers/acpi/apei/ghes.c | 38 +++-
drivers/acpi/apei/hest.c | 39 ++++
drivers/edac/amd64_edac.c | 334 +++++++++++++++++++++++-------
drivers/edac/amd64_edac.h | 60 +++++-
drivers/edac/cpc925_edac.c | 2 +-
drivers/edac/edac_mc_sysfs.c | 6 +-
drivers/edac/i3200_edac.c | 3 +-
drivers/edac/x38_edac.c | 3 +-
include/linux/mm.h | 1 +
include/linux/pci_ids.h | 2 +
mm/memory-failure.c | 5 +-
20 files changed, 523 insertions(+), 135 deletions(-)
diff --git a/Documentation/x86/x86_64/boot-options.txt b/Documentation/x86/x86_64/boot-options.txt
index e9e8ddb..1228b22 100644
--- a/Documentation/x86/x86_64/boot-options.txt
+++ b/Documentation/x86/x86_64/boot-options.txt
@@ -176,6 +176,11 @@ ACPI
acpi=noirq Don't route interrupts
+ acpi=nocmcff Disable firmware first mode for corrected errors. This
+ disables parsing the HEST CMC error source to check if
+ firmware has set the FF flag. This may result in
+ duplicate corrected error reports.
+
PCI
pci=off Don't use PCI
diff --git a/arch/x86/include/asm/acpi.h b/arch/x86/include/asm/acpi.h
index 2dfac58..b1977ba 100644
--- a/arch/x86/include/asm/acpi.h
+++ b/arch/x86/include/asm/acpi.h
@@ -86,6 +86,7 @@ extern int acpi_pci_disabled;
extern int acpi_skip_timer_override;
extern int acpi_use_timer_override;
extern int acpi_fix_pin2_polarity;
+extern int acpi_disable_cmcff;
extern u8 acpi_sci_flags;
extern int acpi_sci_override_gsi;
@@ -168,6 +169,7 @@ static inline void arch_acpi_set_pdc_bits(u32 *buf)
#define acpi_lapic 0
#define acpi_ioapic 0
+#define acpi_disable_cmcff 0
static inline void acpi_noirq_set(void) { }
static inline void acpi_disable_pci(void) { }
static inline void disable_acpi(void) { }
diff --git a/arch/x86/include/asm/mce.h b/arch/x86/include/asm/mce.h
index 29e3093..cbe6b9e 100644
--- a/arch/x86/include/asm/mce.h
+++ b/arch/x86/include/asm/mce.h
@@ -32,11 +32,20 @@
#define MCI_STATUS_PCC (1ULL<<57) /* processor context corrupt */
#define MCI_STATUS_S (1ULL<<56) /* Signaled machine check */
#define MCI_STATUS_AR (1ULL<<55) /* Action required */
-#define MCACOD 0xffff /* MCA Error Code */
+
+/*
+ * Note that the full MCACOD field of IA32_MCi_STATUS MSR is
+ * bits 15:0. But bit 12 is the 'F' bit, defined for corrected
+ * errors to indicate that errors are being filtered by hardware.
+ * We should mask out bit 12 when looking for specific signatures
+ * of uncorrected errors - so the F bit is deliberately skipped
+ * in this #define.
+ */
+#define MCACOD 0xefff /* MCA Error Code */
/* Architecturally defined codes from SDM Vol. 3B Chapter 15 */
#define MCACOD_SCRUB 0x00C0 /* 0xC0-0xCF Memory Scrubbing */
-#define MCACOD_SCRUBMSK 0xfff0
+#define MCACOD_SCRUBMSK 0xeff0 /* Skip bit 12 ('F' bit) */
#define MCACOD_L3WB 0x017A /* L3 Explicit Writeback */
#define MCACOD_DATA 0x0134 /* Data Load */
#define MCACOD_INSTR 0x0150 /* Instruction Fetch */
@@ -188,6 +197,9 @@ extern void register_mce_write_callback(ssize_t (*)(struct file *filp,
const char __user *ubuf,
size_t usize, loff_t *off));
+/* Disable CMCI/polling for MCA bank claimed by firmware */
+extern void mce_disable_bank(int bank);
+
/*
* Exception handler
*/
diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index 2627a81..fead832 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -67,6 +67,7 @@ EXPORT_SYMBOL(acpi_pci_disabled);
int acpi_lapic;
int acpi_ioapic;
int acpi_strict;
+int acpi_disable_cmcff;
u8 acpi_sci_flags __initdata;
int acpi_sci_override_gsi __initdata;
@@ -1626,6 +1627,10 @@ static int __init parse_acpi(char *arg)
/* "acpi=copy_dsdt" copys DSDT */
else if (strcmp(arg, "copy_dsdt") == 0) {
acpi_gbl_copy_dsdt_locally = 1;
+ }
+ /* "acpi=nocmcff" disables FF mode for corrected errors */
+ else if (strcmp(arg, "nocmcff") == 0) {
+ acpi_disable_cmcff = 1;
} else {
/* Core will printk when we return error. */
return -EINVAL;
diff --git a/arch/x86/kernel/amd_nb.c b/arch/x86/kernel/amd_nb.c
index 3048ded..59554dc 100644
--- a/arch/x86/kernel/amd_nb.c
+++ b/arch/x86/kernel/amd_nb.c
@@ -20,6 +20,7 @@ const struct pci_device_id amd_nb_misc_ids[] = {
{ PCI_DEVICE(PCI_VENDOR_ID_AMD, PCI_DEVICE_ID_AMD_10H_NB_MISC) },
{ PCI_DEVICE(PCI_VENDOR_ID_AMD, PCI_DEVICE_ID_AMD_15H_NB_F3) },
{ PCI_DEVICE(PCI_VENDOR_ID_AMD, PCI_DEVICE_ID_AMD_15H_M10H_F3) },
+ { PCI_DEVICE(PCI_VENDOR_ID_AMD, PCI_DEVICE_ID_AMD_15H_M30H_NB_F3) },
{ PCI_DEVICE(PCI_VENDOR_ID_AMD, PCI_DEVICE_ID_AMD_16H_NB_F3) },
{}
};
@@ -27,6 +28,7 @@ EXPORT_SYMBOL(amd_nb_misc_ids);
static const struct pci_device_id amd_nb_link_ids[] = {
{ PCI_DEVICE(PCI_VENDOR_ID_AMD, PCI_DEVICE_ID_AMD_15H_NB_F4) },
+ { PCI_DEVICE(PCI_VENDOR_ID_AMD, PCI_DEVICE_ID_AMD_15H_M30H_NB_F4) },
{ PCI_DEVICE(PCI_VENDOR_ID_AMD, PCI_DEVICE_ID_AMD_16H_NB_F4) },
{}
};
@@ -81,13 +83,20 @@ int amd_cache_northbridges(void)
next_northbridge(misc, amd_nb_misc_ids);
node_to_amd_nb(i)->link = link =
next_northbridge(link, amd_nb_link_ids);
- }
+ }
+ /* GART present only on Fam15h upto model 0fh */
if (boot_cpu_data.x86 == 0xf || boot_cpu_data.x86 == 0x10 ||
- boot_cpu_data.x86 == 0x15)
+ (boot_cpu_data.x86 == 0x15 && boot_cpu_data.x86_model < 0x10))
amd_northbridges.flags |= AMD_NB_GART;
/*
+ * Check for L3 cache presence.
+ */
+ if (!cpuid_edx(0x80000006))
+ return 0;
+
+ /*
* Some CPU families support L3 Cache Index Disable. There are some
* limitations because of E382 and E388 on family 0x10.
*/
diff --git a/arch/x86/kernel/cpu/mcheck/mce-internal.h b/arch/x86/kernel/cpu/mcheck/mce-internal.h
index 5b7d4fa..09edd0b 100644
--- a/arch/x86/kernel/cpu/mcheck/mce-internal.h
+++ b/arch/x86/kernel/cpu/mcheck/mce-internal.h
@@ -25,15 +25,18 @@ int mce_severity(struct mce *a, int tolerant, char **msg);
struct dentry *mce_get_debugfs_dir(void);
extern struct mce_bank *mce_banks;
+extern mce_banks_t mce_banks_ce_disabled;
#ifdef CONFIG_X86_MCE_INTEL
unsigned long mce_intel_adjust_timer(unsigned long interval);
void mce_intel_cmci_poll(void);
void mce_intel_hcpu_update(unsigned long cpu);
+void cmci_disable_bank(int bank);
#else
# define mce_intel_adjust_timer mce_adjust_timer_default
static inline void mce_intel_cmci_poll(void) { }
static inline void mce_intel_hcpu_update(unsigned long cpu) { }
+static inline void cmci_disable_bank(int bank) { }
#endif
void mce_timer_kick(unsigned long interval);
diff --git a/arch/x86/kernel/cpu/mcheck/mce.c b/arch/x86/kernel/cpu/mcheck/mce.c
index 87a65c9..b3218cd 100644
--- a/arch/x86/kernel/cpu/mcheck/mce.c
+++ b/arch/x86/kernel/cpu/mcheck/mce.c
@@ -97,6 +97,15 @@ DEFINE_PER_CPU(mce_banks_t, mce_poll_banks) = {
[0 ... BITS_TO_LONGS(MAX_NR_BANKS)-1] = ~0UL
};
+/*
+ * MCA banks controlled through firmware first for corrected errors.
+ * This is a global list of banks for which we won't enable CMCI and we
+ * won't poll. Firmware controls these banks and is responsible for
+ * reporting corrected errors through GHES. Uncorrected/recoverable
+ * errors are still notified through a machine check.
+ */
+mce_banks_t mce_banks_ce_disabled;
+
static DEFINE_PER_CPU(struct work_struct, mce_work);
static void (*quirk_no_way_out)(int bank, struct mce *m, struct pt_regs *regs);
@@ -1935,6 +1944,25 @@ static struct miscdevice mce_chrdev_device = {
&mce_chrdev_ops,
};
+static void __mce_disable_bank(void *arg)
+{
+ int bank = *((int *)arg);
+ __clear_bit(bank, __get_cpu_var(mce_poll_banks));
+ cmci_disable_bank(bank);
+}
+
+void mce_disable_bank(int bank)
+{
+ if (bank >= mca_cfg.banks) {
+ pr_warn(FW_BUG
+ "Ignoring request to disable invalid MCA bank %d.\n",
+ bank);
+ return;
+ }
+ set_bit(bank, mce_banks_ce_disabled);
+ on_each_cpu(__mce_disable_bank, &bank, 1);
+}
+
/*
* mce=off Disables machine check
* mce=no_cmci Disables CMCI
diff --git a/arch/x86/kernel/cpu/mcheck/mce_intel.c b/arch/x86/kernel/cpu/mcheck/mce_intel.c
index d5640530..4cfe045 100644
--- a/arch/x86/kernel/cpu/mcheck/mce_intel.c
+++ b/arch/x86/kernel/cpu/mcheck/mce_intel.c
@@ -203,6 +203,10 @@ static void cmci_discover(int banks)
if (test_bit(i, owned))
continue;
+ /* Skip banks in firmware first mode */
+ if (test_bit(i, mce_banks_ce_disabled))
+ continue;
+
rdmsrl(MSR_IA32_MCx_CTL2(i), val);
/* Already owned by someone else? */
@@ -271,6 +275,19 @@ void cmci_recheck(void)
local_irq_restore(flags);
}
+/* Caller must hold the lock on cmci_discover_lock */
+static void __cmci_disable_bank(int bank)
+{
+ u64 val;
+
+ if (!test_bit(bank, __get_cpu_var(mce_banks_owned)))
+ return;
+ rdmsrl(MSR_IA32_MCx_CTL2(bank), val);
+ val &= ~MCI_CTL2_CMCI_EN;
+ wrmsrl(MSR_IA32_MCx_CTL2(bank), val);
+ __clear_bit(bank, __get_cpu_var(mce_banks_owned));
+}
+
/*
* Disable CMCI on this CPU for all banks it owns when it goes down.
* This allows other CPUs to claim the banks on rediscovery.
@@ -280,20 +297,12 @@ void cmci_clear(void)
unsigned long flags;
int i;
int banks;
- u64 val;
if (!cmci_supported(&banks))
return;
raw_spin_lock_irqsave(&cmci_discover_lock, flags);
- for (i = 0; i < banks; i++) {
- if (!test_bit(i, __get_cpu_var(mce_banks_owned)))
- continue;
- /* Disable CMCI */
- rdmsrl(MSR_IA32_MCx_CTL2(i), val);
- val &= ~MCI_CTL2_CMCI_EN;
- wrmsrl(MSR_IA32_MCx_CTL2(i), val);
- __clear_bit(i, __get_cpu_var(mce_banks_owned));
- }
+ for (i = 0; i < banks; i++)
+ __cmci_disable_bank(i);
raw_spin_unlock_irqrestore(&cmci_discover_lock, flags);
}
@@ -327,6 +336,19 @@ void cmci_reenable(void)
cmci_discover(banks);
}
+void cmci_disable_bank(int bank)
+{
+ int banks;
+ unsigned long flags;
+
+ if (!cmci_supported(&banks))
+ return;
+
+ raw_spin_lock_irqsave(&cmci_discover_lock, flags);
+ __cmci_disable_bank(bank);
+ raw_spin_unlock_irqrestore(&cmci_discover_lock, flags);
+}
+
static void intel_init_cmci(void)
{
int banks;
diff --git a/drivers/acpi/apei/erst.c b/drivers/acpi/apei/erst.c
index 88d0b0f..40ffcda 100644
--- a/drivers/acpi/apei/erst.c
+++ b/drivers/acpi/apei/erst.c
@@ -39,7 +39,8 @@
#include "apei-internal.h"
-#define ERST_PFX "ERST: "
+#undef pr_fmt
+#define pr_fmt(fmt) "ERST: " fmt
/* ERST command status */
#define ERST_STATUS_SUCCESS 0x0
@@ -109,8 +110,7 @@ static inline int erst_errno(int command_status)
static int erst_timedout(u64 *t, u64 spin_unit)
{
if ((s64)*t < spin_unit) {
- pr_warning(FW_WARN ERST_PFX
- "Firmware does not respond in time\n");
+ pr_warn(FW_WARN "Firmware does not respond in time.\n");
return 1;
}
*t -= spin_unit;
@@ -186,8 +186,8 @@ static int erst_exec_stall(struct apei_exec_context *ctx,
if (ctx->value > FIRMWARE_MAX_STALL) {
if (!in_nmi())
- pr_warning(FW_WARN ERST_PFX
- "Too long stall time for stall instruction: %llx.\n",
+ pr_warn(FW_WARN
+ "Too long stall time for stall instruction: 0x%llx.\n",
ctx->value);
stall_time = FIRMWARE_MAX_STALL;
} else
@@ -206,8 +206,8 @@ static int erst_exec_stall_while_true(struct apei_exec_context *ctx,
if (ctx->var1 > FIRMWARE_MAX_STALL) {
if (!in_nmi())
- pr_warning(FW_WARN ERST_PFX
- "Too long stall time for stall while true instruction: %llx.\n",
+ pr_warn(FW_WARN
+ "Too long stall time for stall while true instruction: 0x%llx.\n",
ctx->var1);
stall_time = FIRMWARE_MAX_STALL;
} else
@@ -271,8 +271,7 @@ static int erst_exec_move_data(struct apei_exec_context *ctx,
/* ioremap does not work in interrupt context */
if (in_interrupt()) {
- pr_warning(ERST_PFX
- "MOVE_DATA can not be used in interrupt context");
+ pr_warn("MOVE_DATA can not be used in interrupt context.\n");
return -EBUSY;
}
@@ -522,8 +521,7 @@ retry:
ERST_RECORD_ID_CACHE_SIZE_MAX);
if (new_size <= erst_record_id_cache.size) {
if (printk_ratelimit())
- pr_warning(FW_WARN ERST_PFX
- "too many record ID!\n");
+ pr_warn(FW_WARN "too many record IDs!\n");
return 0;
}
alloc_size = new_size * sizeof(entries[0]);
@@ -759,8 +757,7 @@ static int __erst_clear_from_storage(u64 record_id)
static void pr_unimpl_nvram(void)
{
if (printk_ratelimit())
- pr_warning(ERST_PFX
- "NVRAM ERST Log Address Range is not implemented yet\n");
+ pr_warn("NVRAM ERST Log Address Range not implemented yet.\n");
}
static int __erst_write_to_nvram(const struct cper_record_header *record)
@@ -1120,7 +1117,7 @@ static int __init erst_init(void)
goto err;
if (erst_disable) {
- pr_info(ERST_PFX
+ pr_info(
"Error Record Serialization Table (ERST) support is disabled.\n");
goto err;
}
@@ -1131,14 +1128,14 @@ static int __init erst_init(void)
goto err;
else if (ACPI_FAILURE(status)) {
const char *msg = acpi_format_exception(status);
- pr_err(ERST_PFX "Failed to get table, %s\n", msg);
+ pr_err("Failed to get table, %s\n", msg);
rc = -EINVAL;
goto err;
}
rc = erst_check_table(erst_tab);
if (rc) {
- pr_err(FW_BUG ERST_PFX "ERST table is invalid\n");
+ pr_err(FW_BUG "ERST table is invalid.\n");
goto err;
}
@@ -1156,21 +1153,19 @@ static int __init erst_init(void)
rc = erst_get_erange(&erst_erange);
if (rc) {
if (rc == -ENODEV)
- pr_info(ERST_PFX
+ pr_info(
"The corresponding hardware device or firmware implementation "
"is not available.\n");
else
- pr_err(ERST_PFX
- "Failed to get Error Log Address Range.\n");
+ pr_err("Failed to get Error Log Address Range.\n");
goto err_unmap_reg;
}
r = request_mem_region(erst_erange.base, erst_erange.size, "APEI ERST");
if (!r) {
- pr_err(ERST_PFX
- "Can not request iomem region <0x%16llx-0x%16llx> for ERST.\n",
- (unsigned long long)erst_erange.base,
- (unsigned long long)erst_erange.base + erst_erange.size);
+ pr_err("Can not request [mem %#010llx-%#010llx] for ERST.\n",
+ (unsigned long long)erst_erange.base,
+ (unsigned long long)erst_erange.base + erst_erange.size - 1);
rc = -EIO;
goto err_unmap_reg;
}
@@ -1180,7 +1175,7 @@ static int __init erst_init(void)
if (!erst_erange.vaddr)
goto err_release_erange;
- pr_info(ERST_PFX
+ pr_info(
"Error Record Serialization Table (ERST) support is initialized.\n");
buf = kmalloc(erst_erange.size, GFP_KERNEL);
@@ -1192,15 +1187,15 @@ static int __init erst_init(void)
rc = pstore_register(&erst_info);
if (rc) {
if (rc != -EPERM)
- pr_info(ERST_PFX
- "Could not register with persistent store\n");
+ pr_info(
+ "Could not register with persistent store.\n");
erst_info.buf = NULL;
erst_info.bufsize = 0;
kfree(buf);
}
} else
- pr_err(ERST_PFX
- "Failed to allocate %lld bytes for persistent store error log\n",
+ pr_err(
+ "Failed to allocate %lld bytes for persistent store error log.\n",
erst_erange.size);
return 0;
diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index ec9b57d..8ec37bb 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -409,6 +409,34 @@ static void ghes_clear_estatus(struct ghes *ghes)
ghes->flags &= ~GHES_TO_CLEAR;
}
+static void ghes_handle_memory_failure(struct acpi_hest_generic_data *gdata, int sev)
+{
+#ifdef CONFIG_ACPI_APEI_MEMORY_FAILURE
+ unsigned long pfn;
+ int sec_sev = ghes_severity(gdata->error_severity);
+ struct cper_sec_mem_err *mem_err;
+ mem_err = (struct cper_sec_mem_err *)(gdata + 1);
+
+ if (sec_sev == GHES_SEV_CORRECTED &&
+ (gdata->flags & CPER_SEC_ERROR_THRESHOLD_EXCEEDED) &&
+ (mem_err->validation_bits & CPER_MEM_VALID_PHYSICAL_ADDRESS)) {
+ pfn = mem_err->physical_addr >> PAGE_SHIFT;
+ if (pfn_valid(pfn))
+ memory_failure_queue(pfn, 0, MF_SOFT_OFFLINE);
+ else if (printk_ratelimit())
+ pr_warn(FW_WARN GHES_PFX
+ "Invalid address in generic error data: %#llx\n",
+ mem_err->physical_addr);
+ }
+ if (sev == GHES_SEV_RECOVERABLE &&
+ sec_sev == GHES_SEV_RECOVERABLE &&
+ mem_err->validation_bits & CPER_MEM_VALID_PHYSICAL_ADDRESS) {
+ pfn = mem_err->physical_addr >> PAGE_SHIFT;
+ memory_failure_queue(pfn, 0, 0);
+ }
+#endif
+}
+
static void ghes_do_proc(struct ghes *ghes,
const struct acpi_hest_generic_status *estatus)
{
@@ -428,15 +456,7 @@ static void ghes_do_proc(struct ghes *ghes,
apei_mce_report_mem_error(sev == GHES_SEV_CORRECTED,
mem_err);
#endif
-#ifdef CONFIG_ACPI_APEI_MEMORY_FAILURE
- if (sev == GHES_SEV_RECOVERABLE &&
- sec_sev == GHES_SEV_RECOVERABLE &&
- mem_err->validation_bits & CPER_MEM_VALID_PHYSICAL_ADDRESS) {
- unsigned long pfn;
- pfn = mem_err->physical_addr >> PAGE_SHIFT;
- memory_failure_queue(pfn, 0, 0);
- }
-#endif
+ ghes_handle_memory_failure(gdata, sev);
}
#ifdef CONFIG_ACPI_APEI_PCIEAER
else if (!uuid_le_cmp(*(uuid_le *)gdata->section_type,
diff --git a/drivers/acpi/apei/hest.c b/drivers/acpi/apei/hest.c
index f5ef5d5..f5e37f3 100644
--- a/drivers/acpi/apei/hest.c
+++ b/drivers/acpi/apei/hest.c
@@ -36,6 +36,7 @@
#include <linux/io.h>
#include <linux/platform_device.h>
#include <acpi/apei.h>
+#include <asm/mce.h>
#include "apei-internal.h"
@@ -121,6 +122,41 @@ int apei_hest_parse(apei_hest_func_t func, void *data)
}
EXPORT_SYMBOL_GPL(apei_hest_parse);
+/*
+ * Check if firmware advertises firmware first mode. We need FF bit to be set
+ * along with a set of MC banks which work in FF mode.
+ */
+static int __init hest_parse_cmc(struct acpi_hest_header *hest_hdr, void *data)
+{
+#ifdef CONFIG_X86_MCE
+ int i;
+ struct acpi_hest_ia_corrected *cmc;
+ struct acpi_hest_ia_error_bank *mc_bank;
+
+ if (hest_hdr->type != ACPI_HEST_TYPE_IA32_CORRECTED_CHECK)
+ return 0;
+
+ cmc = (struct acpi_hest_ia_corrected *)hest_hdr;
+ if (!cmc->enabled)
+ return 0;
+
+ /*
+ * We expect HEST to provide a list of MC banks that report errors
+ * in firmware first mode. Otherwise, return non-zero value to
+ * indicate that we are done parsing HEST.
+ */
+ if (!(cmc->flags & ACPI_HEST_FIRMWARE_FIRST) || !cmc->num_hardware_banks)
+ return 1;
+
+ pr_info(HEST_PFX "Enabling Firmware First mode for corrected errors.\n");
+
+ mc_bank = (struct acpi_hest_ia_error_bank *)(cmc + 1);
+ for (i = 0; i < cmc->num_hardware_banks; i++, mc_bank++)
+ mce_disable_bank(mc_bank->bank_number);
+#endif
+ return 1;
+}
+
struct ghes_arr {
struct platform_device **ghes_devs;
unsigned int count;
@@ -227,6 +263,9 @@ void __init acpi_hest_init(void)
goto err;
}
+ if (!acpi_disable_cmcff)
+ apei_hest_parse(hest_parse_cmc, NULL);
+
if (!ghes_disable) {
rc = apei_hest_parse(hest_parse_ghes_count, &ghes_count);
if (rc)
diff --git a/drivers/edac/amd64_edac.c b/drivers/edac/amd64_edac.c
index 8b6a034..3c9e4e9 100644
--- a/drivers/edac/amd64_edac.c
+++ b/drivers/edac/amd64_edac.c
@@ -123,7 +123,7 @@ static void f15h_select_dct(struct amd64_pvt *pvt, u8 dct)
u32 reg = 0;
amd64_read_pci_cfg(pvt->F1, DCT_CFG_SEL, ®);
- reg &= 0xfffffffe;
+ reg &= (pvt->model >= 0x30) ? ~3 : ~1;
reg |= dct;
amd64_write_pci_cfg(pvt->F1, DCT_CFG_SEL, reg);
}
@@ -133,8 +133,9 @@ static int f15_read_dct_pci_cfg(struct amd64_pvt *pvt, int addr, u32 *val,
{
u8 dct = 0;
+ /* For F15 M30h, the second dct is DCT 3, refer to BKDG Section 2.10 */
if (addr >= 0x140 && addr <= 0x1a0) {
- dct = 1;
+ dct = (pvt->model >= 0x30) ? 3 : 1;
addr -= 0x100;
}
@@ -202,11 +203,11 @@ static int amd64_set_scrub_rate(struct mem_ctl_info *mci, u32 bw)
struct amd64_pvt *pvt = mci->pvt_info;
u32 min_scrubrate = 0x5;
- if (boot_cpu_data.x86 == 0xf)
+ if (pvt->fam == 0xf)
min_scrubrate = 0x0;
- /* F15h Erratum #505 */
- if (boot_cpu_data.x86 == 0x15)
+ /* Erratum #505 */
+ if (pvt->fam == 0x15 && pvt->model < 0x10)
f15h_select_dct(pvt, 0);
return __amd64_set_scrub_rate(pvt->F3, bw, min_scrubrate);
@@ -218,8 +219,8 @@ static int amd64_get_scrub_rate(struct mem_ctl_info *mci)
u32 scrubval = 0;
int i, retval = -EINVAL;
- /* F15h Erratum #505 */
- if (boot_cpu_data.x86 == 0x15)
+ /* Erratum #505 */
+ if (pvt->fam == 0x15 && pvt->model < 0x10)
f15h_select_dct(pvt, 0);
amd64_read_pci_cfg(pvt->F3, SCRCTRL, &scrubval);
@@ -335,7 +336,7 @@ static void get_cs_base_and_mask(struct amd64_pvt *pvt, int csrow, u8 dct,
u64 csbase, csmask, base_bits, mask_bits;
u8 addr_shift;
- if (boot_cpu_data.x86 == 0xf && pvt->ext_model < K8_REV_F) {
+ if (pvt->fam == 0xf && pvt->ext_model < K8_REV_F) {
csbase = pvt->csels[dct].csbases[csrow];
csmask = pvt->csels[dct].csmasks[csrow];
base_bits = GENMASK(21, 31) | GENMASK(9, 15);
@@ -343,10 +344,11 @@ static void get_cs_base_and_mask(struct amd64_pvt *pvt, int csrow, u8 dct,
addr_shift = 4;
/*
- * F16h needs two addr_shift values: 8 for high and 6 for low
- * (cf. F16h BKDG).
- */
- } else if (boot_cpu_data.x86 == 0x16) {
+ * F16h and F15h, models 30h and later need two addr_shift values:
+ * 8 for high and 6 for low (cf. F16h BKDG).
+ */
+ } else if (pvt->fam == 0x16 ||
+ (pvt->fam == 0x15 && pvt->model >= 0x30)) {
csbase = pvt->csels[dct].csbases[csrow];
csmask = pvt->csels[dct].csmasks[csrow >> 1];
@@ -367,7 +369,7 @@ static void get_cs_base_and_mask(struct amd64_pvt *pvt, int csrow, u8 dct,
csmask = pvt->csels[dct].csmasks[csrow >> 1];
addr_shift = 8;
- if (boot_cpu_data.x86 == 0x15)
+ if (pvt->fam == 0x15)
base_bits = mask_bits = GENMASK(19,30) | GENMASK(5,13);
else
base_bits = mask_bits = GENMASK(19,28) | GENMASK(5,13);
@@ -447,14 +449,14 @@ int amd64_get_dram_hole_info(struct mem_ctl_info *mci, u64 *hole_base,
struct amd64_pvt *pvt = mci->pvt_info;
/* only revE and later have the DRAM Hole Address Register */
- if (boot_cpu_data.x86 == 0xf && pvt->ext_model < K8_REV_E) {
+ if (pvt->fam == 0xf && pvt->ext_model < K8_REV_E) {
edac_dbg(1, " revision %d for node %d does not support DHAR\n",
pvt->ext_model, pvt->mc_node_id);
return 1;
}
/* valid for Fam10h and above */
- if (boot_cpu_data.x86 >= 0x10 && !dhar_mem_hoist_valid(pvt)) {
+ if (pvt->fam >= 0x10 && !dhar_mem_hoist_valid(pvt)) {
edac_dbg(1, " Dram Memory Hoisting is DISABLED on this system\n");
return 1;
}
@@ -486,10 +488,8 @@ int amd64_get_dram_hole_info(struct mem_ctl_info *mci, u64 *hole_base,
*hole_base = dhar_base(pvt);
*hole_size = (1ULL << 32) - *hole_base;
- if (boot_cpu_data.x86 > 0xf)
- *hole_offset = f10_dhar_offset(pvt);
- else
- *hole_offset = k8_dhar_offset(pvt);
+ *hole_offset = (pvt->fam > 0xf) ? f10_dhar_offset(pvt)
+ : k8_dhar_offset(pvt);
edac_dbg(1, " DHAR info for node %d base 0x%lx offset 0x%lx size 0x%lx\n",
pvt->mc_node_id, (unsigned long)*hole_base,
@@ -663,7 +663,7 @@ static unsigned long amd64_determine_edac_cap(struct amd64_pvt *pvt)
u8 bit;
unsigned long edac_cap = EDAC_FLAG_NONE;
- bit = (boot_cpu_data.x86 > 0xf || pvt->ext_model >= K8_REV_F)
+ bit = (pvt->fam > 0xf || pvt->ext_model >= K8_REV_F)
? 19
: 17;
@@ -675,7 +675,7 @@ static unsigned long amd64_determine_edac_cap(struct amd64_pvt *pvt)
static void amd64_debug_display_dimm_sizes(struct amd64_pvt *, u8);
-static void amd64_dump_dramcfg_low(u32 dclr, int chan)
+static void amd64_dump_dramcfg_low(struct amd64_pvt *pvt, u32 dclr, int chan)
{
edac_dbg(1, "F2x%d90 (DRAM Cfg Low): 0x%08x\n", chan, dclr);
@@ -686,7 +686,7 @@ static void amd64_dump_dramcfg_low(u32 dclr, int chan)
edac_dbg(1, " PAR/ERR parity: %s\n",
(dclr & BIT(8)) ? "enabled" : "disabled");
- if (boot_cpu_data.x86 == 0x10)
+ if (pvt->fam == 0x10)
edac_dbg(1, " DCT 128bit mode width: %s\n",
(dclr & BIT(11)) ? "128b" : "64b");
@@ -709,21 +709,21 @@ static void dump_misc_regs(struct amd64_pvt *pvt)
(pvt->nbcap & NBCAP_SECDED) ? "yes" : "no",
(pvt->nbcap & NBCAP_CHIPKILL) ? "yes" : "no");
- amd64_dump_dramcfg_low(pvt->dclr0, 0);
+ amd64_dump_dramcfg_low(pvt, pvt->dclr0, 0);
edac_dbg(1, "F3xB0 (Online Spare): 0x%08x\n", pvt->online_spare);
edac_dbg(1, "F1xF0 (DRAM Hole Address): 0x%08x, base: 0x%08x, offset: 0x%08x\n",
pvt->dhar, dhar_base(pvt),
- (boot_cpu_data.x86 == 0xf) ? k8_dhar_offset(pvt)
- : f10_dhar_offset(pvt));
+ (pvt->fam == 0xf) ? k8_dhar_offset(pvt)
+ : f10_dhar_offset(pvt));
edac_dbg(1, " DramHoleValid: %s\n", dhar_valid(pvt) ? "yes" : "no");
amd64_debug_display_dimm_sizes(pvt, 0);
/* everything below this point is Fam10h and above */
- if (boot_cpu_data.x86 == 0xf)
+ if (pvt->fam == 0xf)
return;
amd64_debug_display_dimm_sizes(pvt, 1);
@@ -732,17 +732,20 @@ static void dump_misc_regs(struct amd64_pvt *pvt)
/* Only if NOT ganged does dclr1 have valid info */
if (!dct_ganging_enabled(pvt))
- amd64_dump_dramcfg_low(pvt->dclr1, 1);
+ amd64_dump_dramcfg_low(pvt, pvt->dclr1, 1);
}
/*
- * see BKDG, F2x[1,0][5C:40], F2[1,0][6C:60]
+ * See BKDG, F2x[1,0][5C:40], F2[1,0][6C:60]
*/
static void prep_chip_selects(struct amd64_pvt *pvt)
{
- if (boot_cpu_data.x86 == 0xf && pvt->ext_model < K8_REV_F) {
+ if (pvt->fam == 0xf && pvt->ext_model < K8_REV_F) {
pvt->csels[0].b_cnt = pvt->csels[1].b_cnt = 8;
pvt->csels[0].m_cnt = pvt->csels[1].m_cnt = 8;
+ } else if (pvt->fam == 0x15 && pvt->model >= 0x30) {
+ pvt->csels[0].b_cnt = pvt->csels[1].b_cnt = 4;
+ pvt->csels[0].m_cnt = pvt->csels[1].m_cnt = 2;
} else {
pvt->csels[0].b_cnt = pvt->csels[1].b_cnt = 8;
pvt->csels[0].m_cnt = pvt->csels[1].m_cnt = 4;
@@ -768,7 +771,7 @@ static void read_dct_base_mask(struct amd64_pvt *pvt)
edac_dbg(0, " DCSB0[%d]=0x%08x reg: F2x%x\n",
cs, *base0, reg0);
- if (boot_cpu_data.x86 == 0xf || dct_ganging_enabled(pvt))
+ if (pvt->fam == 0xf || dct_ganging_enabled(pvt))
continue;
if (!amd64_read_dct_pci_cfg(pvt, reg1, base1))
@@ -786,7 +789,7 @@ static void read_dct_base_mask(struct amd64_pvt *pvt)
edac_dbg(0, " DCSM0[%d]=0x%08x reg: F2x%x\n",
cs, *mask0, reg0);
- if (boot_cpu_data.x86 == 0xf || dct_ganging_enabled(pvt))
+ if (pvt->fam == 0xf || dct_ganging_enabled(pvt))
continue;
if (!amd64_read_dct_pci_cfg(pvt, reg1, mask1))
@@ -800,9 +803,9 @@ static enum mem_type amd64_determine_memory_type(struct amd64_pvt *pvt, int cs)
enum mem_type type;
/* F15h supports only DDR3 */
- if (boot_cpu_data.x86 >= 0x15)
+ if (pvt->fam >= 0x15)
type = (pvt->dclr0 & BIT(16)) ? MEM_DDR3 : MEM_RDDR3;
- else if (boot_cpu_data.x86 == 0x10 || pvt->ext_model >= K8_REV_F) {
+ else if (pvt->fam == 0x10 || pvt->ext_model >= K8_REV_F) {
if (pvt->dchr0 & DDR3_MODE)
type = (pvt->dclr0 & BIT(16)) ? MEM_DDR3 : MEM_RDDR3;
else
@@ -835,14 +838,13 @@ static int k8_early_channel_count(struct amd64_pvt *pvt)
}
/* On F10h and later ErrAddr is MC4_ADDR[47:1] */
-static u64 get_error_address(struct mce *m)
+static u64 get_error_address(struct amd64_pvt *pvt, struct mce *m)
{
- struct cpuinfo_x86 *c = &boot_cpu_data;
u64 addr;
u8 start_bit = 1;
u8 end_bit = 47;
- if (c->x86 == 0xf) {
+ if (pvt->fam == 0xf) {
start_bit = 3;
end_bit = 39;
}
@@ -852,7 +854,7 @@ static u64 get_error_address(struct mce *m)
/*
* Erratum 637 workaround
*/
- if (c->x86 == 0x15) {
+ if (pvt->fam == 0x15) {
struct amd64_pvt *pvt;
u64 cc6_base, tmp_addr;
u32 tmp;
@@ -916,15 +918,15 @@ static struct pci_dev *pci_get_related_function(unsigned int vendor,
static void read_dram_base_limit_regs(struct amd64_pvt *pvt, unsigned range)
{
struct amd_northbridge *nb;
- struct pci_dev *misc, *f1 = NULL;
- struct cpuinfo_x86 *c = &boot_cpu_data;
+ struct pci_dev *f1 = NULL;
+ unsigned int pci_func;
int off = range << 3;
u32 llim;
amd64_read_pci_cfg(pvt->F1, DRAM_BASE_LO + off, &pvt->ranges[range].base.lo);
amd64_read_pci_cfg(pvt->F1, DRAM_LIMIT_LO + off, &pvt->ranges[range].lim.lo);
- if (c->x86 == 0xf)
+ if (pvt->fam == 0xf)
return;
if (!dram_rw(pvt, range))
@@ -934,15 +936,17 @@ static void read_dram_base_limit_regs(struct amd64_pvt *pvt, unsigned range)
amd64_read_pci_cfg(pvt->F1, DRAM_LIMIT_HI + off, &pvt->ranges[range].lim.hi);
/* F15h: factor in CC6 save area by reading dst node's limit reg */
- if (c->x86 != 0x15)
+ if (pvt->fam != 0x15)
return;
nb = node_to_amd_nb(dram_dst_node(pvt, range));
if (WARN_ON(!nb))
return;
- misc = nb->misc;
- f1 = pci_get_related_function(misc->vendor, PCI_DEVICE_ID_AMD_15H_NB_F1, misc);
+ pci_func = (pvt->model == 0x30) ? PCI_DEVICE_ID_AMD_15H_M30H_NB_F1
+ : PCI_DEVICE_ID_AMD_15H_NB_F1;
+
+ f1 = pci_get_related_function(nb->misc->vendor, pci_func, nb->misc);
if (WARN_ON(!f1))
return;
@@ -1089,7 +1093,7 @@ static int f1x_early_channel_count(struct amd64_pvt *pvt)
int i, j, channels = 0;
/* On F10h, if we are in 128 bit mode, then we are using 2 channels */
- if (boot_cpu_data.x86 == 0x10 && (pvt->dclr0 & WIDTH_128))
+ if (pvt->fam == 0x10 && (pvt->dclr0 & WIDTH_128))
return 2;
/*
@@ -1173,7 +1177,7 @@ static int f15_dbam_to_chip_select(struct amd64_pvt *pvt, u8 dct,
}
/*
- * F16h has only limited cs_modes
+ * F16h and F15h model 30h have only limited cs_modes.
*/
static int f16_dbam_to_chip_select(struct amd64_pvt *pvt, u8 dct,
unsigned cs_mode)
@@ -1190,7 +1194,7 @@ static int f16_dbam_to_chip_select(struct amd64_pvt *pvt, u8 dct,
static void read_dram_ctl_register(struct amd64_pvt *pvt)
{
- if (boot_cpu_data.x86 == 0xf)
+ if (pvt->fam == 0xf)
return;
if (!amd64_read_dct_pci_cfg(pvt, DCT_SEL_LO, &pvt->dct_sel_lo)) {
@@ -1218,6 +1222,29 @@ static void read_dram_ctl_register(struct amd64_pvt *pvt)
}
/*
+ * Determine channel (DCT) based on the interleaving mode (see F15h M30h BKDG,
+ * 2.10.12 Memory Interleaving Modes).
+ */
+static u8 f15_m30h_determine_channel(struct amd64_pvt *pvt, u64 sys_addr,
+ u8 intlv_en, int num_dcts_intlv,
+ u32 dct_sel)
+{
+ u8 channel = 0;
+ u8 select;
+
+ if (!(intlv_en))
+ return (u8)(dct_sel);
+
+ 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;
+
+ return channel;
+}
+
+/*
* Determine channel (DCT) based on the interleaving mode: F10h BKDG, 2.8.9 Memory
* Interleaving Modes.
*/
@@ -1366,6 +1393,10 @@ static int f1x_lookup_addr_in_dct(u64 in_addr, u8 nid, u8 dct)
(in_addr & cs_mask), (cs_base & cs_mask));
if ((in_addr & cs_mask) == (cs_base & cs_mask)) {
+ if (pvt->fam == 0x15 && pvt->model >= 0x30) {
+ cs_found = csrow;
+ break;
+ }
cs_found = f10_process_possible_spare(pvt, dct, csrow);
edac_dbg(1, " MATCH csrow=%d\n", cs_found);
@@ -1384,11 +1415,9 @@ static u64 f1x_swap_interleaved_region(struct amd64_pvt *pvt, u64 sys_addr)
{
u32 swap_reg, swap_base, swap_limit, rgn_size, tmp_addr;
- if (boot_cpu_data.x86 == 0x10) {
+ if (pvt->fam == 0x10) {
/* only revC3 and revE have that feature */
- if (boot_cpu_data.x86_model < 4 ||
- (boot_cpu_data.x86_model < 0xa &&
- boot_cpu_data.x86_mask < 3))
+ if (pvt->model < 4 || (pvt->model < 0xa && pvt->stepping < 3))
return sys_addr;
}
@@ -1492,20 +1521,143 @@ static int f1x_match_to_this_node(struct amd64_pvt *pvt, unsigned range,
return cs_found;
}
-static int f1x_translate_sysaddr_to_cs(struct amd64_pvt *pvt, u64 sys_addr,
- int *chan_sel)
+static int f15_m30h_match_to_this_node(struct amd64_pvt *pvt, unsigned range,
+ u64 sys_addr, int *chan_sel)
+{
+ int cs_found = -EINVAL;
+ int num_dcts_intlv = 0;
+ u64 chan_addr, chan_offset;
+ u64 dct_base, dct_limit;
+ u32 dct_cont_base_reg, dct_cont_limit_reg, tmp;
+ u8 channel, alias_channel, leg_mmio_hole, dct_sel, dct_offset_en;
+
+ u64 dhar_offset = f10_dhar_offset(pvt);
+ u8 intlv_addr = dct_sel_interleave_addr(pvt);
+ u8 node_id = dram_dst_node(pvt, range);
+ u8 intlv_en = dram_intlv_en(pvt, range);
+
+ amd64_read_pci_cfg(pvt->F1, DRAM_CONT_BASE, &dct_cont_base_reg);
+ amd64_read_pci_cfg(pvt->F1, DRAM_CONT_LIMIT, &dct_cont_limit_reg);
+
+ dct_offset_en = (u8) ((dct_cont_base_reg >> 3) & BIT(0));
+ dct_sel = (u8) ((dct_cont_base_reg >> 4) & 0x7);
+
+ edac_dbg(1, "(range %d) SystemAddr= 0x%llx Limit=0x%llx\n",
+ range, sys_addr, get_dram_limit(pvt, range));
+
+ if (!(get_dram_base(pvt, range) <= sys_addr) &&
+ !(get_dram_limit(pvt, range) >= sys_addr))
+ return -EINVAL;
+
+ if (dhar_valid(pvt) &&
+ dhar_base(pvt) <= sys_addr &&
+ sys_addr < BIT_64(32)) {
+ amd64_warn("Huh? Address is in the MMIO hole: 0x%016llx\n",
+ sys_addr);
+ return -EINVAL;
+ }
+
+ /* Verify sys_addr is within DCT Range. */
+ dct_base = (u64) dct_sel_baseaddr(pvt);
+ dct_limit = (dct_cont_limit_reg >> 11) & 0x1FFF;
+
+ if (!(dct_cont_base_reg & BIT(0)) &&
+ !(dct_base <= (sys_addr >> 27) &&
+ dct_limit >= (sys_addr >> 27)))
+ return -EINVAL;
+
+ /* Verify number of dct's that participate in channel interleaving. */
+ num_dcts_intlv = (int) hweight8(intlv_en);
+
+ if (!(num_dcts_intlv % 2 == 0) || (num_dcts_intlv > 4))
+ return -EINVAL;
+
+ channel = f15_m30h_determine_channel(pvt, sys_addr, intlv_en,
+ num_dcts_intlv, dct_sel);
+
+ /* Verify we stay within the MAX number of channels allowed */
+ if (channel > 4 || channel < 0)
+ return -EINVAL;
+
+ leg_mmio_hole = (u8) (dct_cont_base_reg >> 1 & BIT(0));
+
+ /* Get normalized DCT addr */
+ if (leg_mmio_hole && (sys_addr >= BIT_64(32)))
+ chan_offset = dhar_offset;
+ else
+ chan_offset = dct_base << 27;
+
+ chan_addr = sys_addr - chan_offset;
+
+ /* remove channel interleave */
+ if (num_dcts_intlv == 2) {
+ if (intlv_addr == 0x4)
+ chan_addr = ((chan_addr >> 9) << 8) |
+ (chan_addr & 0xff);
+ else if (intlv_addr == 0x5)
+ chan_addr = ((chan_addr >> 10) << 9) |
+ (chan_addr & 0x1ff);
+ else
+ return -EINVAL;
+
+ } else if (num_dcts_intlv == 4) {
+ if (intlv_addr == 0x4)
+ chan_addr = ((chan_addr >> 10) << 8) |
+ (chan_addr & 0xff);
+ else if (intlv_addr == 0x5)
+ chan_addr = ((chan_addr >> 11) << 9) |
+ (chan_addr & 0x1ff);
+ else
+ return -EINVAL;
+ }
+
+ if (dct_offset_en) {
+ amd64_read_pci_cfg(pvt->F1,
+ DRAM_CONT_HIGH_OFF + (int) channel * 4,
+ &tmp);
+ chan_addr += (u64) ((tmp >> 11) & 0xfff) << 27;
+ }
+
+ f15h_select_dct(pvt, channel);
+
+ edac_dbg(1, " Normalized DCT addr: 0x%llx\n", chan_addr);
+
+ /*
+ * Find Chip select:
+ * if channel = 3, then alias it to 1. This is because, in F15 M30h,
+ * there is support for 4 DCT's, but only 2 are currently functional.
+ * They are DCT0 and DCT3. But we have read all registers of DCT3 into
+ * pvt->csels[1]. So we need to use '1' here to get correct info.
+ * Refer F15 M30h BKDG Section 2.10 and 2.10.3 for clarifications.
+ */
+ alias_channel = (channel == 3) ? 1 : channel;
+
+ cs_found = f1x_lookup_addr_in_dct(chan_addr, node_id, alias_channel);
+
+ if (cs_found >= 0)
+ *chan_sel = alias_channel;
+
+ return cs_found;
+}
+
+static int f1x_translate_sysaddr_to_cs(struct amd64_pvt *pvt,
+ u64 sys_addr,
+ int *chan_sel)
{
int cs_found = -EINVAL;
unsigned range;
for (range = 0; range < DRAM_RANGES; range++) {
-
if (!dram_rw(pvt, range))
continue;
- if ((get_dram_base(pvt, range) <= sys_addr) &&
- (get_dram_limit(pvt, range) >= sys_addr)) {
+ if (pvt->fam == 0x15 && pvt->model >= 0x30)
+ cs_found = f15_m30h_match_to_this_node(pvt, range,
+ sys_addr,
+ chan_sel);
+ else if ((get_dram_base(pvt, range) <= sys_addr) &&
+ (get_dram_limit(pvt, range) >= sys_addr)) {
cs_found = f1x_match_to_this_node(pvt, range,
sys_addr, chan_sel);
if (cs_found >= 0)
@@ -1554,7 +1706,7 @@ static void amd64_debug_display_dimm_sizes(struct amd64_pvt *pvt, u8 ctrl)
u32 *dcsb = ctrl ? pvt->csels[1].csbases : pvt->csels[0].csbases;
u32 dbam = ctrl ? pvt->dbam1 : pvt->dbam0;
- if (boot_cpu_data.x86 == 0xf) {
+ if (pvt->fam == 0xf) {
/* K8 families < revF not supported yet */
if (pvt->ext_model < K8_REV_F)
return;
@@ -1624,6 +1776,17 @@ static struct amd64_family_type amd64_family_types[] = {
.read_dct_pci_cfg = f15_read_dct_pci_cfg,
}
},
+ [F15_M30H_CPUS] = {
+ .ctl_name = "F15h_M30h",
+ .f1_id = PCI_DEVICE_ID_AMD_15H_M30H_NB_F1,
+ .f3_id = PCI_DEVICE_ID_AMD_15H_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 = f15_read_dct_pci_cfg,
+ }
+ },
[F16_CPUS] = {
.ctl_name = "F16h",
.f1_id = PCI_DEVICE_ID_AMD_16H_NB_F1,
@@ -1860,7 +2023,7 @@ static inline void __amd64_decode_bus_error(struct mem_ctl_info *mci,
memset(&err, 0, sizeof(err));
- sys_addr = get_error_address(m);
+ sys_addr = get_error_address(pvt, m);
if (ecc_type == 2)
err.syndrome = extract_syndrome(m->status);
@@ -1921,10 +2084,9 @@ static void free_mc_sibling_devs(struct amd64_pvt *pvt)
*/
static void read_mc_regs(struct amd64_pvt *pvt)
{
- struct cpuinfo_x86 *c = &boot_cpu_data;
+ unsigned range;
u64 msr_val;
u32 tmp;
- unsigned range;
/*
* Retrieve TOP_MEM and TOP_MEM2; no masking off of reserved bits since
@@ -1985,14 +2147,14 @@ static void read_mc_regs(struct amd64_pvt *pvt)
pvt->ecc_sym_sz = 4;
- if (c->x86 >= 0x10) {
+ if (pvt->fam >= 0x10) {
amd64_read_pci_cfg(pvt->F3, EXT_NB_MCA_CFG, &tmp);
- if (c->x86 != 0x16)
+ if (pvt->fam != 0x16)
/* F16h has only DCT0 */
amd64_read_dct_pci_cfg(pvt, DBAM1, &pvt->dbam1);
/* F10h, revD and later can do x8 ECC too */
- if ((c->x86 > 0x10 || c->x86_model > 7) && tmp & BIT(25))
+ if ((pvt->fam > 0x10 || pvt->model > 7) && tmp & BIT(25))
pvt->ecc_sym_sz = 8;
}
dump_misc_regs(pvt);
@@ -2086,7 +2248,7 @@ static int init_csrows(struct mem_ctl_info *mci)
bool row_dct0 = !!csrow_enabled(i, 0, pvt);
bool row_dct1 = false;
- if (boot_cpu_data.x86 != 0xf)
+ if (pvt->fam != 0xf)
row_dct1 = !!csrow_enabled(i, 1, pvt);
if (!row_dct0 && !row_dct1)
@@ -2104,7 +2266,7 @@ static int init_csrows(struct mem_ctl_info *mci)
}
/* K8 has only one DCT */
- if (boot_cpu_data.x86 != 0xf && row_dct1) {
+ if (pvt->fam != 0xf && row_dct1) {
int row_dct1_pages = amd64_csrow_nr_pages(pvt, 1, i);
csrow->channels[1]->dimm->nr_pages = row_dct1_pages;
@@ -2333,13 +2495,14 @@ static bool ecc_enabled(struct pci_dev *F3, u16 nid)
static int set_mc_sysfs_attrs(struct mem_ctl_info *mci)
{
+ struct amd64_pvt *pvt = mci->pvt_info;
int rc;
rc = amd64_create_sysfs_dbg_files(mci);
if (rc < 0)
return rc;
- if (boot_cpu_data.x86 >= 0x10) {
+ if (pvt->fam >= 0x10) {
rc = amd64_create_sysfs_inject_files(mci);
if (rc < 0)
return rc;
@@ -2350,9 +2513,11 @@ static int set_mc_sysfs_attrs(struct mem_ctl_info *mci)
static void del_mc_sysfs_attrs(struct mem_ctl_info *mci)
{
+ struct amd64_pvt *pvt = mci->pvt_info;
+
amd64_remove_sysfs_dbg_files(mci);
- if (boot_cpu_data.x86 >= 0x10)
+ if (pvt->fam >= 0x10)
amd64_remove_sysfs_inject_files(mci);
}
@@ -2387,10 +2552,14 @@ static void setup_mci_misc_attrs(struct mem_ctl_info *mci,
*/
static struct amd64_family_type *amd64_per_family_init(struct amd64_pvt *pvt)
{
- u8 fam = boot_cpu_data.x86;
struct amd64_family_type *fam_type = NULL;
- switch (fam) {
+ pvt->ext_model = boot_cpu_data.x86_model >> 4;
+ pvt->stepping = boot_cpu_data.x86_mask;
+ pvt->model = boot_cpu_data.x86_model;
+ pvt->fam = boot_cpu_data.x86;
+
+ switch (pvt->fam) {
case 0xf:
fam_type = &amd64_family_types[K8_CPUS];
pvt->ops = &amd64_family_types[K8_CPUS].ops;
@@ -2402,6 +2571,12 @@ static struct amd64_family_type *amd64_per_family_init(struct amd64_pvt *pvt)
break;
case 0x15:
+ if (pvt->model == 0x30) {
+ fam_type = &amd64_family_types[F15_M30H_CPUS];
+ pvt->ops = &amd64_family_types[F15_M30H_CPUS].ops;
+ break;
+ }
+
fam_type = &amd64_family_types[F15_CPUS];
pvt->ops = &amd64_family_types[F15_CPUS].ops;
break;
@@ -2416,10 +2591,8 @@ static struct amd64_family_type *amd64_per_family_init(struct amd64_pvt *pvt)
return NULL;
}
- pvt->ext_model = boot_cpu_data.x86_model >> 4;
-
amd64_info("%s %sdetected (node %d).\n", fam_type->ctl_name,
- (fam == 0xf ?
+ (pvt->fam == 0xf ?
(pvt->ext_model >= K8_REV_F ? "revF or later "
: "revE or earlier ")
: ""), pvt->mc_node_id);
@@ -2470,8 +2643,15 @@ static int amd64_init_one_instance(struct pci_dev *F2)
layers[0].size = pvt->csels[0].b_cnt;
layers[0].is_virt_csrow = true;
layers[1].type = EDAC_MC_LAYER_CHANNEL;
- layers[1].size = pvt->channel_count;
+
+ /*
+ * Always allocate two channels since we can have setups with DIMMs on
+ * only one channel. Also, this simplifies handling later for the price
+ * of a couple of KBs tops.
+ */
+ layers[1].size = 2;
layers[1].is_virt_csrow = false;
+
mci = edac_mc_alloc(nid, ARRAY_SIZE(layers), layers, 0);
if (!mci)
goto err_siblings;
@@ -2579,6 +2759,8 @@ static void amd64_remove_one_instance(struct pci_dev *pdev)
struct ecc_settings *s = ecc_stngs[nid];
mci = find_mci_by_dev(&pdev->dev);
+ WARN_ON(!mci);
+
del_mc_sysfs_attrs(mci);
/* Remove from EDAC CORE tracking list */
mci = edac_mc_del_mc(&pdev->dev);
@@ -2638,6 +2820,14 @@ static DEFINE_PCI_DEVICE_TABLE(amd64_pci_table) = {
},
{
.vendor = PCI_VENDOR_ID_AMD,
+ .device = PCI_DEVICE_ID_AMD_15H_M30H_NB_F2,
+ .subvendor = PCI_ANY_ID,
+ .subdevice = PCI_ANY_ID,
+ .class = 0,
+ .class_mask = 0,
+ },
+ {
+ .vendor = PCI_VENDOR_ID_AMD,
.device = PCI_DEVICE_ID_AMD_16H_NB_F2,
.subvendor = PCI_ANY_ID,
.subdevice = PCI_ANY_ID,
diff --git a/drivers/edac/amd64_edac.h b/drivers/edac/amd64_edac.h
index 2c6f113..d2443cf 100644
--- a/drivers/edac/amd64_edac.h
+++ b/drivers/edac/amd64_edac.h
@@ -170,6 +170,8 @@
/*
* PCI-defined configuration space registers
*/
+#define PCI_DEVICE_ID_AMD_15H_M30H_NB_F1 0x141b
+#define PCI_DEVICE_ID_AMD_15H_M30H_NB_F2 0x141c
#define PCI_DEVICE_ID_AMD_15H_NB_F1 0x1601
#define PCI_DEVICE_ID_AMD_15H_NB_F2 0x1602
#define PCI_DEVICE_ID_AMD_16H_NB_F1 0x1531
@@ -181,13 +183,22 @@
#define DRAM_BASE_LO 0x40
#define DRAM_LIMIT_LO 0x44
-#define dram_intlv_en(pvt, i) ((u8)((pvt->ranges[i].base.lo >> 8) & 0x7))
+/*
+ * F15 M30h D18F1x2[1C:00]
+ */
+#define DRAM_CONT_BASE 0x200
+#define DRAM_CONT_LIMIT 0x204
+
+/*
+ * F15 M30h D18F1x2[4C:40]
+ */
+#define DRAM_CONT_HIGH_OFF 0x240
+
#define dram_rw(pvt, i) ((u8)(pvt->ranges[i].base.lo & 0x3))
#define dram_intlv_sel(pvt, i) ((u8)((pvt->ranges[i].lim.lo >> 8) & 0x7))
#define dram_dst_node(pvt, i) ((u8)(pvt->ranges[i].lim.lo & 0x7))
#define DHAR 0xf0
-#define dhar_valid(pvt) ((pvt)->dhar & BIT(0))
#define dhar_mem_hoist_valid(pvt) ((pvt)->dhar & BIT(1))
#define dhar_base(pvt) ((pvt)->dhar & 0xff000000)
#define k8_dhar_offset(pvt) (((pvt)->dhar & 0x0000ff00) << 16)
@@ -234,8 +245,6 @@
#define DDR3_MODE BIT(8)
#define DCT_SEL_LO 0x110
-#define dct_sel_baseaddr(pvt) ((pvt)->dct_sel_lo & 0xFFFFF800)
-#define dct_sel_interleave_addr(pvt) (((pvt)->dct_sel_lo >> 6) & 0x3)
#define dct_high_range_enabled(pvt) ((pvt)->dct_sel_lo & BIT(0))
#define dct_interleave_enabled(pvt) ((pvt)->dct_sel_lo & BIT(2))
@@ -297,6 +306,7 @@ enum amd_families {
K8_CPUS = 0,
F10_CPUS,
F15_CPUS,
+ F15_M30H_CPUS,
F16_CPUS,
NUM_FAMILIES,
};
@@ -337,6 +347,10 @@ struct amd64_pvt {
struct pci_dev *F1, *F2, *F3;
u16 mc_node_id; /* MC index of this MC node */
+ u8 fam; /* CPU family */
+ u8 model; /* ... model */
+ u8 stepping; /* ... stepping */
+
int ext_model; /* extended model value of this node */
int channel_count;
@@ -414,6 +428,14 @@ static inline u16 extract_syndrome(u64 status)
return ((status >> 47) & 0xff) | ((status >> 16) & 0xff00);
}
+static inline u8 dct_sel_interleave_addr(struct amd64_pvt *pvt)
+{
+ if (pvt->fam == 0x15 && pvt->model >= 0x30)
+ return (((pvt->dct_sel_hi >> 9) & 0x1) << 2) |
+ ((pvt->dct_sel_lo >> 6) & 0x3);
+
+ return ((pvt)->dct_sel_lo >> 6) & 0x3;
+}
/*
* per-node ECC settings descriptor
*/
@@ -504,3 +526,33 @@ static inline void enable_caches(void *dummy)
{
write_cr0(read_cr0() & ~X86_CR0_CD);
}
+
+static inline u8 dram_intlv_en(struct amd64_pvt *pvt, unsigned int i)
+{
+ if (pvt->fam == 0x15 && pvt->model >= 0x30) {
+ u32 tmp;
+ amd64_read_pci_cfg(pvt->F1, DRAM_CONT_LIMIT, &tmp);
+ return (u8) tmp & 0xF;
+ }
+ return (u8) (pvt->ranges[i].base.lo >> 8) & 0x7;
+}
+
+static inline u8 dhar_valid(struct amd64_pvt *pvt)
+{
+ if (pvt->fam == 0x15 && pvt->model >= 0x30) {
+ u32 tmp;
+ amd64_read_pci_cfg(pvt->F1, DRAM_CONT_BASE, &tmp);
+ return (tmp >> 1) & BIT(0);
+ }
+ return (pvt)->dhar & BIT(0);
+}
+
+static inline u32 dct_sel_baseaddr(struct amd64_pvt *pvt)
+{
+ if (pvt->fam == 0x15 && pvt->model >= 0x30) {
+ u32 tmp;
+ amd64_read_pci_cfg(pvt->F1, DRAM_CONT_BASE, &tmp);
+ return (tmp >> 11) & 0x1FFF;
+ }
+ return (pvt)->dct_sel_lo & 0xFFFFF800;
+}
diff --git a/drivers/edac/cpc925_edac.c b/drivers/edac/cpc925_edac.c
index 7f3c571..df6575f 100644
--- a/drivers/edac/cpc925_edac.c
+++ b/drivers/edac/cpc925_edac.c
@@ -789,7 +789,7 @@ static struct cpc925_dev_info cpc925_devs[] = {
.exit = cpc925_htlink_exit,
.check = cpc925_htlink_check,
},
- {0}, /* Terminated by NULL */
+ { }
};
/*
diff --git a/drivers/edac/edac_mc_sysfs.c b/drivers/edac/edac_mc_sysfs.c
index e7c32c4..9f7e0e60 100644
--- a/drivers/edac/edac_mc_sysfs.c
+++ b/drivers/edac/edac_mc_sysfs.c
@@ -58,8 +58,10 @@ static int edac_set_poll_msec(const char *val, struct kernel_param *kp)
if (!val)
return -EINVAL;
- ret = strict_strtol(val, 0, &l);
- if (ret == -EINVAL || ((int)l != l))
+ ret = kstrtol(val, 0, &l);
+ if (ret)
+ return ret;
+ if ((int)l != l)
return -EINVAL;
*((int *)kp->arg) = l;
diff --git a/drivers/edac/i3200_edac.c b/drivers/edac/i3200_edac.c
index aa44c17..be10a74 100644
--- a/drivers/edac/i3200_edac.c
+++ b/drivers/edac/i3200_edac.c
@@ -260,8 +260,7 @@ static void i3200_check(struct mem_ctl_info *mci)
i3200_process_error_info(mci, &info);
}
-
-void __iomem *i3200_map_mchbar(struct pci_dev *pdev)
+static void __iomem *i3200_map_mchbar(struct pci_dev *pdev)
{
union {
u64 mchbar;
diff --git a/drivers/edac/x38_edac.c b/drivers/edac/x38_edac.c
index c9db24d..1a4df82 100644
--- a/drivers/edac/x38_edac.c
+++ b/drivers/edac/x38_edac.c
@@ -248,8 +248,7 @@ static void x38_check(struct mem_ctl_info *mci)
x38_process_error_info(mci, &info);
}
-
-void __iomem *x38_map_mchbar(struct pci_dev *pdev)
+static void __iomem *x38_map_mchbar(struct pci_dev *pdev)
{
union {
u64 mchbar;
diff --git a/include/linux/mm.h b/include/linux/mm.h
index f022460..d2d59b4 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -1798,6 +1798,7 @@ enum mf_flags {
MF_COUNT_INCREASED = 1 << 0,
MF_ACTION_REQUIRED = 1 << 1,
MF_MUST_KILL = 1 << 2,
+ MF_SOFT_OFFLINE = 1 << 3,
};
extern int memory_failure(unsigned long pfn, int trapno, int flags);
extern void memory_failure_queue(unsigned long pfn, int trapno, int flags);
diff --git a/include/linux/pci_ids.h b/include/linux/pci_ids.h
index 3bed2e8..d1fe5d0 100644
--- a/include/linux/pci_ids.h
+++ b/include/linux/pci_ids.h
@@ -518,6 +518,8 @@
#define PCI_DEVICE_ID_AMD_11H_NB_MISC 0x1303
#define PCI_DEVICE_ID_AMD_11H_NB_LINK 0x1304
#define PCI_DEVICE_ID_AMD_15H_M10H_F3 0x1403
+#define PCI_DEVICE_ID_AMD_15H_M30H_NB_F3 0x141d
+#define PCI_DEVICE_ID_AMD_15H_M30H_NB_F4 0x141e
#define PCI_DEVICE_ID_AMD_15H_NB_F0 0x1600
#define PCI_DEVICE_ID_AMD_15H_NB_F1 0x1601
#define PCI_DEVICE_ID_AMD_15H_NB_F2 0x1602
diff --git a/mm/memory-failure.c b/mm/memory-failure.c
index 2c13aa7..55d7c80 100644
--- a/mm/memory-failure.c
+++ b/mm/memory-failure.c
@@ -1286,7 +1286,10 @@ static void memory_failure_work_func(struct work_struct *work)
spin_unlock_irqrestore(&mf_cpu->lock, proc_flags);
if (!gotten)
break;
- memory_failure(entry.pfn, entry.trapno, entry.flags);
+ if (entry.flags & MF_SOFT_OFFLINE)
+ soft_offline_page(pfn_to_page(entry.pfn), entry.flags);
+ else
+ memory_failure(entry.pfn, entry.trapno, entry.flags);
}
}
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists