diff options
Diffstat (limited to 'arch/x86/kernel/cpu/mcheck')
| -rw-r--r-- | arch/x86/kernel/cpu/mcheck/mce-severity.c | 16 | ||||
| -rw-r--r-- | arch/x86/kernel/cpu/mcheck/mce.c | 11 | ||||
| -rw-r--r-- | arch/x86/kernel/cpu/mcheck/mce_amd.c | 60 | ||||
| -rw-r--r-- | arch/x86/kernel/cpu/mcheck/therm_throt.c | 29 |
4 files changed, 70 insertions, 46 deletions
diff --git a/arch/x86/kernel/cpu/mcheck/mce-severity.c b/arch/x86/kernel/cpu/mcheck/mce-severity.c index 1e8d66c..362190b 100644 --- a/arch/x86/kernel/cpu/mcheck/mce-severity.c +++ b/arch/x86/kernel/cpu/mcheck/mce-severity.c @@ -101,15 +101,19 @@ static struct severity { }; /* - * If the EIPV bit is set, it means the saved IP is the - * instruction which caused the MCE. + * If mcgstatus indicated that ip/cs on the stack were + * no good, then "m->cs" will be zero and we will have + * to assume the worst case (IN_KERNEL) as we actually + * have no idea what we were executing when the machine + * check hit. + * If we do have a good "m->cs" (or a faked one in the + * case we were executing in VM86 mode) we can use it to + * distinguish an exception taken in user from from one + * taken in the kernel. */ static int error_context(struct mce *m) { - if (m->mcgstatus & MCG_STATUS_EIPV) - return (m->ip && (m->cs & 3) == 3) ? IN_USER : IN_KERNEL; - /* Unknown, assume kernel */ - return IN_KERNEL; + return ((m->cs & 3) == 3) ? IN_USER : IN_KERNEL; } int mce_severity(struct mce *a, int tolerant, char **msg) diff --git a/arch/x86/kernel/cpu/mcheck/mce.c b/arch/x86/kernel/cpu/mcheck/mce.c index ff1ae9b..1396edf 100644 --- a/arch/x86/kernel/cpu/mcheck/mce.c +++ b/arch/x86/kernel/cpu/mcheck/mce.c @@ -122,9 +122,7 @@ void mce_setup(struct mce *m) m->time = get_seconds(); m->cpuvendor = boot_cpu_data.x86_vendor; m->cpuid = cpuid_eax(1); -#ifdef CONFIG_SMP m->socketid = cpu_data(m->extcpu).phys_proc_id; -#endif m->apicid = cpu_data(m->extcpu).initial_apicid; rdmsrl(MSR_IA32_MCG_CAP, m->mcgcap); } @@ -453,6 +451,13 @@ static inline void mce_get_rip(struct mce *m, struct pt_regs *regs) if (regs && (m->mcgstatus & (MCG_STATUS_RIPV|MCG_STATUS_EIPV))) { m->ip = regs->ip; m->cs = regs->cs; + /* + * When in VM86 mode make the cs look like ring 3 + * always. This is a lie, but it's better than passing + * the additional vm86 bit around everywhere. + */ + if (v8086_mode(regs)) + m->cs |= 3; } else { m->ip = 0; m->cs = 0; @@ -990,6 +995,7 @@ void do_machine_check(struct pt_regs *regs, long error_code) */ add_taint(TAINT_MACHINE_CHECK); + mce_get_rip(&m, regs); severity = mce_severity(&m, tolerant, NULL); /* @@ -1028,7 +1034,6 @@ void do_machine_check(struct pt_regs *regs, long error_code) if (severity == MCE_AO_SEVERITY && mce_usable_address(&m)) mce_ring_add(m.addr >> PAGE_SHIFT); - mce_get_rip(&m, regs); mce_log(&m); if (severity > worst) { diff --git a/arch/x86/kernel/cpu/mcheck/mce_amd.c b/arch/x86/kernel/cpu/mcheck/mce_amd.c index bb0adad..b97aa72 100644 --- a/arch/x86/kernel/cpu/mcheck/mce_amd.c +++ b/arch/x86/kernel/cpu/mcheck/mce_amd.c @@ -52,6 +52,7 @@ struct threshold_block { unsigned int cpu; u32 address; u16 interrupt_enable; + bool interrupt_capable; u16 threshold_limit; struct kobject kobj; struct list_head miscj; @@ -64,11 +65,9 @@ struct threshold_bank { }; static DEFINE_PER_CPU(struct threshold_bank * [NR_BANKS], threshold_banks); -#ifdef CONFIG_SMP static unsigned char shared_bank[NR_BANKS] = { 0, 0, 0, 0, 1 }; -#endif static DEFINE_PER_CPU(unsigned char, bank_map); /* see which banks are on */ @@ -86,6 +85,21 @@ struct thresh_restart { u16 old_limit; }; +static bool lvt_interrupt_supported(unsigned int bank, u32 msr_high_bits) +{ + /* + * bank 4 supports APIC LVT interrupts implicitly since forever. + */ + if (bank == 4) + return true; + + /* + * IntP: interrupt present; if this bit is set, the thresholding + * bank can generate APIC LVT interrupts + */ + return msr_high_bits & BIT(28); +} + static int lvt_off_valid(struct threshold_block *b, int apic, u32 lo, u32 hi) { int msr = (hi & MASK_LVTOFF_HI) >> 20; @@ -107,8 +121,10 @@ static int lvt_off_valid(struct threshold_block *b, int apic, u32 lo, u32 hi) return 1; }; -/* must be called with correct cpu affinity */ -/* Called via smp_call_function_single() */ +/* + * Called via smp_call_function_single(), must be called with correct + * cpu affinity. + */ static void threshold_restart_bank(void *_tr) { struct thresh_restart *tr = _tr; @@ -131,6 +147,12 @@ static void threshold_restart_bank(void *_tr) (new_count & THRESHOLD_MAX); } + /* clear IntType */ + hi &= ~MASK_INT_TYPE_HI; + + if (!tr->b->interrupt_capable) + goto done; + if (tr->set_lvt_off) { if (lvt_off_valid(tr->b, tr->lvt_off, lo, hi)) { /* set new lvt offset */ @@ -139,9 +161,10 @@ static void threshold_restart_bank(void *_tr) } } - tr->b->interrupt_enable ? - (hi = (hi & ~MASK_INT_TYPE_HI) | INT_TYPE_APIC) : - (hi &= ~MASK_INT_TYPE_HI); + if (tr->b->interrupt_enable) + hi |= INT_TYPE_APIC; + + done: hi |= MASK_COUNT_EN_HI; wrmsr(tr->b->address, lo, hi); @@ -202,18 +225,21 @@ void mce_amd_feature_init(struct cpuinfo_x86 *c) if (!block) per_cpu(bank_map, cpu) |= (1 << bank); -#ifdef CONFIG_SMP + if (shared_bank[bank] && c->cpu_core_id) break; -#endif - offset = setup_APIC_mce(offset, - (high & MASK_LVTOFF_HI) >> 20); memset(&b, 0, sizeof(b)); - b.cpu = cpu; - b.bank = bank; - b.block = block; - b.address = address; + b.cpu = cpu; + b.bank = bank; + b.block = block; + b.address = address; + b.interrupt_capable = lvt_interrupt_supported(bank, high); + + if (b.interrupt_capable) { + int new = (high & MASK_LVTOFF_HI) >> 20; + offset = setup_APIC_mce(offset, new); + } mce_threshold_block_init(&b, offset); mce_threshold_vector = amd_threshold_interrupt; @@ -313,6 +339,9 @@ store_interrupt_enable(struct threshold_block *b, const char *buf, size_t size) struct thresh_restart tr; unsigned long new; + if (!b->interrupt_capable) + return -EINVAL; + if (strict_strtoul(buf, 0, &new) < 0) return -EINVAL; @@ -471,6 +500,7 @@ static __cpuinit int allocate_threshold_blocks(unsigned int cpu, b->cpu = cpu; b->address = address; b->interrupt_enable = 0; + b->interrupt_capable = lvt_interrupt_supported(bank, high); b->threshold_limit = THRESHOLD_MAX; INIT_LIST_HEAD(&b->miscj); diff --git a/arch/x86/kernel/cpu/mcheck/therm_throt.c b/arch/x86/kernel/cpu/mcheck/therm_throt.c index 27c6251..99cd9d2 100644 --- a/arch/x86/kernel/cpu/mcheck/therm_throt.c +++ b/arch/x86/kernel/cpu/mcheck/therm_throt.c @@ -322,17 +322,6 @@ device_initcall(thermal_throttle_init_device); #endif /* CONFIG_SYSFS */ -/* - * Set up the most two significant bit to notify mce log that this thermal - * event type. - * This is a temp solution. May be changed in the future with mce log - * infrasture. - */ -#define CORE_THROTTLED (0) -#define CORE_POWER_LIMIT ((__u64)1 << 62) -#define PACKAGE_THROTTLED ((__u64)2 << 62) -#define PACKAGE_POWER_LIMIT ((__u64)3 << 62) - static void notify_thresholds(__u64 msr_val) { /* check whether the interrupt handler is defined; @@ -362,27 +351,23 @@ static void intel_thermal_interrupt(void) if (therm_throt_process(msr_val & THERM_STATUS_PROCHOT, THERMAL_THROTTLING_EVENT, CORE_LEVEL) != 0) - mce_log_therm_throt_event(CORE_THROTTLED | msr_val); + mce_log_therm_throt_event(msr_val); if (this_cpu_has(X86_FEATURE_PLN)) - if (therm_throt_process(msr_val & THERM_STATUS_POWER_LIMIT, + therm_throt_process(msr_val & THERM_STATUS_POWER_LIMIT, POWER_LIMIT_EVENT, - CORE_LEVEL) != 0) - mce_log_therm_throt_event(CORE_POWER_LIMIT | msr_val); + CORE_LEVEL); if (this_cpu_has(X86_FEATURE_PTS)) { rdmsrl(MSR_IA32_PACKAGE_THERM_STATUS, msr_val); - if (therm_throt_process(msr_val & PACKAGE_THERM_STATUS_PROCHOT, + therm_throt_process(msr_val & PACKAGE_THERM_STATUS_PROCHOT, THERMAL_THROTTLING_EVENT, - PACKAGE_LEVEL) != 0) - mce_log_therm_throt_event(PACKAGE_THROTTLED | msr_val); + PACKAGE_LEVEL); if (this_cpu_has(X86_FEATURE_PLN)) - if (therm_throt_process(msr_val & + therm_throt_process(msr_val & PACKAGE_THERM_STATUS_POWER_LIMIT, POWER_LIMIT_EVENT, - PACKAGE_LEVEL) != 0) - mce_log_therm_throt_event(PACKAGE_POWER_LIMIT - | msr_val); + PACKAGE_LEVEL); } } |