From e085ac7a6ddbd746966083c5e13aa290c3e9a253 Mon Sep 17 00:00:00 2001 From: Xie XiuQi Date: Mon, 4 Dec 2017 17:54:37 +0100 Subject: x86/MCE: Extend table to report action optional errors through CMCI too According to the Intel SDM Volume 3B (253669-063US, July 2017), action optional (SRAO) errors can be reported either via MCE or CMC: In cases when SRAO is signaled via CMCI the error signature is indicated via UC=1, PCC=0, S=0. Type(*1) UC EN PCC S AR Signaling --------------------------------------------------------------- UC 1 1 1 x x MCE SRAR 1 1 0 1 1 MCE SRAO 1 x(*2) 0 x(*2) 0 MCE/CMC UCNA 1 x 0 0 0 CMC CE 0 x x x x CMC NOTES: 1. SRAR, SRAO and UCNA errors are supported by the processor only when IA32_MCG_CAP[24] (MCG_SER_P) is set. 2. EN=1, S=1 when signaled via MCE. EN=x, S=0 when signaled via CMC. And there is a description in 15.6.2 UCR Error Reporting and Logging, for bit S: S (Signaling) flag, bit 56 - Indicates (when set) that a machine check exception was generated for the UCR error reported in this MC bank... When the S flag in the IA32_MCi_STATUS register is clear, this UCR error was not signaled via a machine check exception and instead was reported as a corrected machine check (CMC). So merge the two cases and just remove the S=0 check for SRAO in mce_severity(). [ Borislav: Massage commit message.] Signed-off-by: Xie XiuQi Signed-off-by: Borislav Petkov Signed-off-by: Thomas Gleixner Reviewed-by: Tony Luck Tested-by: Chen Wei Cc: linux-edac Link: http://lkml.kernel.org/r/1511575548-41992-1-git-send-email-xiexiuqi@huawei.com --- arch/x86/kernel/cpu/mcheck/mce-severity.c | 26 +++++++++++++++++--------- 1 file changed, 17 insertions(+), 9 deletions(-) (limited to 'arch') diff --git a/arch/x86/kernel/cpu/mcheck/mce-severity.c b/arch/x86/kernel/cpu/mcheck/mce-severity.c index 4ca632a06e0b..5bbd06f38ff6 100644 --- a/arch/x86/kernel/cpu/mcheck/mce-severity.c +++ b/arch/x86/kernel/cpu/mcheck/mce-severity.c @@ -59,6 +59,7 @@ static struct severity { #define MCGMASK(x, y) .mcgmask = x, .mcgres = y #define MASK(x, y) .mask = x, .result = y #define MCI_UC_S (MCI_STATUS_UC|MCI_STATUS_S) +#define MCI_UC_AR (MCI_STATUS_UC|MCI_STATUS_AR) #define MCI_UC_SAR (MCI_STATUS_UC|MCI_STATUS_S|MCI_STATUS_AR) #define MCI_ADDR (MCI_STATUS_ADDRV|MCI_STATUS_MISCV) @@ -101,6 +102,22 @@ static struct severity { NOSER, BITCLR(MCI_STATUS_UC) ), + /* + * known AO MCACODs reported via MCE or CMC: + * + * SRAO could be signaled either via a machine check exception or + * CMCI with the corresponding bit S 1 or 0. So we don't need to + * check bit S for SRAO. + */ + MCESEV( + AO, "Action optional: memory scrubbing error", + SER, MASK(MCI_STATUS_OVER|MCI_UC_AR|MCACOD_SCRUBMSK, MCI_STATUS_UC|MCACOD_SCRUB) + ), + MCESEV( + AO, "Action optional: last level cache writeback error", + SER, MASK(MCI_STATUS_OVER|MCI_UC_AR|MCACOD, MCI_STATUS_UC|MCACOD_L3WB) + ), + /* ignore OVER for UCNA */ MCESEV( UCNA, "Uncorrected no action required", @@ -149,15 +166,6 @@ static struct severity { SER, MASK(MCI_STATUS_OVER|MCI_UC_SAR, MCI_UC_SAR) ), - /* known AO MCACODs: */ - MCESEV( - AO, "Action optional: memory scrubbing error", - SER, MASK(MCI_STATUS_OVER|MCI_UC_SAR|MCACOD_SCRUBMSK, MCI_UC_S|MCACOD_SCRUB) - ), - MCESEV( - AO, "Action optional: last level cache writeback error", - SER, MASK(MCI_STATUS_OVER|MCI_UC_SAR|MCACOD, MCI_UC_S|MCACOD_L3WB) - ), MCESEV( SOME, "Action optional: unknown MCACOD", SER, MASK(MCI_STATUS_OVER|MCI_UC_SAR, MCI_UC_S) -- cgit v1.2.3 From c8a4364c33ac7ed63278267b8f6d8c15810d5fd1 Mon Sep 17 00:00:00 2001 From: Yazen Ghannam Date: Mon, 4 Dec 2017 17:54:38 +0100 Subject: x86/mce/AMD: Don't set DEF_INT_TYPE in MSR_CU_DEF_ERR on SMCA systems The McaIntrCfg register (MSRC000_0410), previously known as CU_DEFER_ERR, is used on SMCA systems to set the LVT offset for the Threshold and Deferred error interrupts. This register was used on non-SMCA systems to also set the Deferred interrupt type in bits 2:1. However, these bits are reserved on SMCA systems. Only set MSRC000_0410[2:1] on non-SMCA systems. Signed-off-by: Yazen Ghannam Signed-off-by: Borislav Petkov Signed-off-by: Thomas Gleixner Cc: Tony Luck Cc: linux-edac Link: http://lkml.kernel.org/r/20171120162646.5210-1-Yazen.Ghannam@amd.com --- arch/x86/kernel/cpu/mcheck/mce_amd.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'arch') diff --git a/arch/x86/kernel/cpu/mcheck/mce_amd.c b/arch/x86/kernel/cpu/mcheck/mce_amd.c index 486f640b02ef..a38ab1fa53a2 100644 --- a/arch/x86/kernel/cpu/mcheck/mce_amd.c +++ b/arch/x86/kernel/cpu/mcheck/mce_amd.c @@ -407,7 +407,9 @@ static void deferred_error_interrupt_enable(struct cpuinfo_x86 *c) (deferred_error_int_vector != amd_deferred_error_interrupt)) deferred_error_int_vector = amd_deferred_error_interrupt; - low = (low & ~MASK_DEF_INT_TYPE) | DEF_INT_TYPE_APIC; + if (!mce_flags.smca) + low = (low & ~MASK_DEF_INT_TYPE) | DEF_INT_TYPE_APIC; + wrmsr(MSR_CU_DEF_ERR, low, high); } -- cgit v1.2.3 From 11cf887728a3d1de77cc12ce247b64ef32608891 Mon Sep 17 00:00:00 2001 From: Yazen Ghannam Date: Mon, 18 Dec 2017 12:37:12 +0100 Subject: x86/MCE/AMD: Define a function to get SMCA bank type Scalable MCA systems have various types of banks. The bank's type can determine how we handle errors from it. For example, if a bank represents a UMC (Unified Memory Controller) then we will need to convert its address from a normalized address to a system physical address before handling the error. [ bp: Verify m->bank is within range and use bank pointer. ] Signed-off-by: Yazen Ghannam Signed-off-by: Borislav Petkov Signed-off-by: Thomas Gleixner Link: http://lkml.kernel.org/r/20171207203955.118171-1-Yazen.Ghannam@amd.com --- arch/x86/kernel/cpu/mcheck/mce_amd.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+) (limited to 'arch') diff --git a/arch/x86/kernel/cpu/mcheck/mce_amd.c b/arch/x86/kernel/cpu/mcheck/mce_amd.c index a38ab1fa53a2..661c4738be27 100644 --- a/arch/x86/kernel/cpu/mcheck/mce_amd.c +++ b/arch/x86/kernel/cpu/mcheck/mce_amd.c @@ -110,6 +110,20 @@ const char *smca_get_long_name(enum smca_bank_types t) } EXPORT_SYMBOL_GPL(smca_get_long_name); +static enum smca_bank_types smca_get_bank_type(struct mce *m) +{ + struct smca_bank *b; + + if (m->bank >= N_SMCA_BANK_TYPES) + return N_SMCA_BANK_TYPES; + + b = &smca_banks[m->bank]; + if (!b->hwid) + return N_SMCA_BANK_TYPES; + + return b->hwid->bank_type; +} + static struct smca_hwid smca_hwid_mcatypes[] = { /* { bank_type, hwid_mcatype, xec_bitmap } */ -- cgit v1.2.3 From c6708d50f166bea2d763c96485d31fdbc50204f1 Mon Sep 17 00:00:00 2001 From: Yazen Ghannam Date: Mon, 18 Dec 2017 12:37:13 +0100 Subject: x86/MCE: Report only DRAM ECC as memory errors on AMD systems The MCA_STATUS[ErrorCodeExt] field is very bank type specific. We currently check if the ErrorCodeExt value is 0x0 or 0x8 in mce_is_memory_error(), but we don't check the bank number. This means that we could flag non-memory errors as memory errors. We know that we want to flag DRAM ECC errors as memory errors, so let's do those cases first. We can add more cases later when needed. Define a wrapper function in mce_amd.c so we can use SMCA enums. [ bp: Remove brackets around return statements. ] Signed-off-by: Yazen Ghannam Signed-off-by: Borislav Petkov Signed-off-by: Thomas Gleixner Link: http://lkml.kernel.org/r/20171207203955.118171-2-Yazen.Ghannam@amd.com --- arch/x86/include/asm/mce.h | 2 ++ arch/x86/kernel/cpu/mcheck/mce.c | 4 +--- arch/x86/kernel/cpu/mcheck/mce_amd.c | 11 +++++++++++ 3 files changed, 14 insertions(+), 3 deletions(-) (limited to 'arch') diff --git a/arch/x86/include/asm/mce.h b/arch/x86/include/asm/mce.h index b1e8d8db921f..96ea4b5ba658 100644 --- a/arch/x86/include/asm/mce.h +++ b/arch/x86/include/asm/mce.h @@ -376,6 +376,7 @@ struct smca_bank { extern struct smca_bank smca_banks[MAX_NR_BANKS]; extern const char *smca_get_long_name(enum smca_bank_types t); +extern bool amd_mce_is_memory_error(struct mce *m); extern int mce_threshold_create_device(unsigned int cpu); extern int mce_threshold_remove_device(unsigned int cpu); @@ -384,6 +385,7 @@ extern int mce_threshold_remove_device(unsigned int cpu); static inline int mce_threshold_create_device(unsigned int cpu) { return 0; }; static inline int mce_threshold_remove_device(unsigned int cpu) { return 0; }; +static inline bool amd_mce_is_memory_error(struct mce *m) { return false; }; #endif diff --git a/arch/x86/kernel/cpu/mcheck/mce.c b/arch/x86/kernel/cpu/mcheck/mce.c index b1d616d08eee..321c7a80be66 100644 --- a/arch/x86/kernel/cpu/mcheck/mce.c +++ b/arch/x86/kernel/cpu/mcheck/mce.c @@ -503,10 +503,8 @@ static int mce_usable_address(struct mce *m) bool mce_is_memory_error(struct mce *m) { if (m->cpuvendor == X86_VENDOR_AMD) { - /* ErrCodeExt[20:16] */ - u8 xec = (m->status >> 16) & 0x1f; + return amd_mce_is_memory_error(m); - return (xec == 0x0 || xec == 0x8); } else if (m->cpuvendor == X86_VENDOR_INTEL) { /* * Intel SDM Volume 3B - 15.9.2 Compound Error Codes diff --git a/arch/x86/kernel/cpu/mcheck/mce_amd.c b/arch/x86/kernel/cpu/mcheck/mce_amd.c index 661c4738be27..0f32ad242324 100644 --- a/arch/x86/kernel/cpu/mcheck/mce_amd.c +++ b/arch/x86/kernel/cpu/mcheck/mce_amd.c @@ -754,6 +754,17 @@ out_err: } EXPORT_SYMBOL_GPL(umc_normaddr_to_sysaddr); +bool amd_mce_is_memory_error(struct mce *m) +{ + /* ErrCodeExt[20:16] */ + u8 xec = (m->status >> 16) & 0x1f; + + if (mce_flags.smca) + return smca_get_bank_type(m) == SMCA_UMC && xec == 0x0; + + return m->bank == 4 && xec == 0x8; +} + static void __log_error(unsigned int bank, u64 status, u64 addr, u64 misc) { struct mce m; -- cgit v1.2.3 From 179eb850ac57c06edaed67fc744ba9d902172f96 Mon Sep 17 00:00:00 2001 From: Yazen Ghannam Date: Mon, 18 Dec 2017 12:37:14 +0100 Subject: x86/MCE: Make correctable error detection look at the Deferred bit AMD systems may log Deferred errors. These are errors that are uncorrected but which do not need immediate action. The MCA_STATUS[UC] bit may not be set for Deferred errors. Flag the error as not correctable when MCA_STATUS[Deferred] is set and do not feed it into the Correctable Errors Collector. [ bp: Massage commit message. ] Signed-off-by: Yazen Ghannam Signed-off-by: Borislav Petkov Signed-off-by: Thomas Gleixner Link: http://lkml.kernel.org/r/20171212165143.27475-1-Yazen.Ghannam@amd.com --- arch/x86/kernel/cpu/mcheck/mce.c | 13 ++++++++++++- 1 file changed, 12 insertions(+), 1 deletion(-) (limited to 'arch') diff --git a/arch/x86/kernel/cpu/mcheck/mce.c b/arch/x86/kernel/cpu/mcheck/mce.c index 321c7a80be66..1b2c11473376 100644 --- a/arch/x86/kernel/cpu/mcheck/mce.c +++ b/arch/x86/kernel/cpu/mcheck/mce.c @@ -528,6 +528,17 @@ bool mce_is_memory_error(struct mce *m) } EXPORT_SYMBOL_GPL(mce_is_memory_error); +static bool mce_is_correctable(struct mce *m) +{ + if (m->cpuvendor == X86_VENDOR_AMD && m->status & MCI_STATUS_DEFERRED) + return false; + + if (m->status & MCI_STATUS_UC) + return false; + + return true; +} + static bool cec_add_mce(struct mce *m) { if (!m) @@ -535,7 +546,7 @@ static bool cec_add_mce(struct mce *m) /* We eat only correctable DRAM errors with usable addresses. */ if (mce_is_memory_error(m) && - !(m->status & MCI_STATUS_UC) && + mce_is_correctable(m) && mce_usable_address(m)) if (!cec_add_elem(m->addr >> PAGE_SHIFT)) return true; -- cgit v1.2.3