From 3653ada5d3e173489b3a466305687cb5c44b2ab1 Mon Sep 17 00:00:00 2001 From: Borislav Petkov Date: Sun, 4 Dec 2011 15:12:09 +0100 Subject: [PATCH 1/4] x86, mce: Add wrappers for registering on the decode chain No functionality change, this is done so that in a follow-on patch all queued-up MCEs can be decoded after registering on the chain. Signed-off-by: Borislav Petkov --- arch/x86/include/asm/mce.h | 3 ++- arch/x86/kernel/cpu/mcheck/mce.c | 25 ++++++++++++++++++------- drivers/edac/i7core_edac.c | 4 ++-- drivers/edac/mce_amd.c | 4 ++-- drivers/edac/sb_edac.c | 6 ++---- 5 files changed, 26 insertions(+), 16 deletions(-) diff --git a/arch/x86/include/asm/mce.h b/arch/x86/include/asm/mce.h index 0e8ae57d3656..b7c47a468fde 100644 --- a/arch/x86/include/asm/mce.h +++ b/arch/x86/include/asm/mce.h @@ -120,7 +120,8 @@ struct mce_log { #ifdef __KERNEL__ -extern struct atomic_notifier_head x86_mce_decoder_chain; +extern void mce_register_decode_chain(struct notifier_block *nb); +extern void mce_unregister_decode_chain(struct notifier_block *nb); #include #include diff --git a/arch/x86/kernel/cpu/mcheck/mce.c b/arch/x86/kernel/cpu/mcheck/mce.c index 2af127d4c3d1..c3c66ac6ef74 100644 --- a/arch/x86/kernel/cpu/mcheck/mce.c +++ b/arch/x86/kernel/cpu/mcheck/mce.c @@ -95,13 +95,6 @@ static DECLARE_WAIT_QUEUE_HEAD(mce_chrdev_wait); static DEFINE_PER_CPU(struct mce, mces_seen); static int cpu_missing; -/* - * CPU/chipset specific EDAC code can register a notifier call here to print - * MCE errors in a human-readable form. - */ -ATOMIC_NOTIFIER_HEAD(x86_mce_decoder_chain); -EXPORT_SYMBOL_GPL(x86_mce_decoder_chain); - /* MCA banks polled by the period polling timer for corrected events */ DEFINE_PER_CPU(mce_banks_t, mce_poll_banks) = { [0 ... BITS_TO_LONGS(MAX_NR_BANKS)-1] = ~0UL @@ -109,6 +102,12 @@ DEFINE_PER_CPU(mce_banks_t, mce_poll_banks) = { static DEFINE_PER_CPU(struct work_struct, mce_work); +/* + * CPU/chipset specific EDAC code can register a notifier call here to print + * MCE errors in a human-readable form. + */ +ATOMIC_NOTIFIER_HEAD(x86_mce_decoder_chain); + /* Do initial initialization of a struct mce */ void mce_setup(struct mce *m) { @@ -190,6 +189,18 @@ void mce_log(struct mce *mce) set_bit(0, &mce_need_notify); } +void mce_register_decode_chain(struct notifier_block *nb) +{ + atomic_notifier_chain_register(&x86_mce_decoder_chain, nb); +} +EXPORT_SYMBOL_GPL(mce_register_decode_chain); + +void mce_unregister_decode_chain(struct notifier_block *nb) +{ + atomic_notifier_chain_unregister(&x86_mce_decoder_chain, nb); +} +EXPORT_SYMBOL_GPL(mce_unregister_decode_chain); + static void print_mce(struct mce *m) { int ret = 0; diff --git a/drivers/edac/i7core_edac.c b/drivers/edac/i7core_edac.c index 70ad8923f1d7..8568d9b61875 100644 --- a/drivers/edac/i7core_edac.c +++ b/drivers/edac/i7core_edac.c @@ -2234,7 +2234,7 @@ static void i7core_unregister_mci(struct i7core_dev *i7core_dev) if (pvt->enable_scrub) disable_sdram_scrub_setting(mci); - atomic_notifier_chain_unregister(&x86_mce_decoder_chain, &i7_mce_dec); + mce_unregister_decode_chain(&i7_mce_dec); /* Disable EDAC polling */ i7core_pci_ctl_release(pvt); @@ -2336,7 +2336,7 @@ static int i7core_register_mci(struct i7core_dev *i7core_dev) /* DCLK for scrub rate setting */ pvt->dclk_freq = get_dclk_freq(); - atomic_notifier_chain_register(&x86_mce_decoder_chain, &i7_mce_dec); + mce_register_decode_chain(&i7_mce_dec); return 0; diff --git a/drivers/edac/mce_amd.c b/drivers/edac/mce_amd.c index d0864d9c38ad..bd926ea2e00c 100644 --- a/drivers/edac/mce_amd.c +++ b/drivers/edac/mce_amd.c @@ -884,7 +884,7 @@ static int __init mce_amd_init(void) pr_info("MCE: In-kernel MCE decoding enabled.\n"); - atomic_notifier_chain_register(&x86_mce_decoder_chain, &amd_mce_dec_nb); + mce_register_decode_chain(&amd_mce_dec_nb); return 0; } @@ -893,7 +893,7 @@ early_initcall(mce_amd_init); #ifdef MODULE static void __exit mce_amd_exit(void) { - atomic_notifier_chain_unregister(&x86_mce_decoder_chain, &amd_mce_dec_nb); + mce_unregister_decode_chain(&amd_mce_dec_nb); kfree(fam_ops); } diff --git a/drivers/edac/sb_edac.c b/drivers/edac/sb_edac.c index 7a402bfbee7d..965bc0cc0cf6 100644 --- a/drivers/edac/sb_edac.c +++ b/drivers/edac/sb_edac.c @@ -1661,8 +1661,7 @@ static void sbridge_unregister_mci(struct sbridge_dev *sbridge_dev) debugf0("MC: " __FILE__ ": %s(): mci = %p, dev = %p\n", __func__, mci, &sbridge_dev->pdev[0]->dev); - atomic_notifier_chain_unregister(&x86_mce_decoder_chain, - &sbridge_mce_dec); + mce_unregister_decode_chain(&sbridge_mce_dec); /* Remove MC sysfs nodes */ edac_mc_del_mc(mci->dev); @@ -1731,8 +1730,7 @@ static int sbridge_register_mci(struct sbridge_dev *sbridge_dev) goto fail0; } - atomic_notifier_chain_register(&x86_mce_decoder_chain, - &sbridge_mce_dec); + mce_register_decode_chain(&sbridge_mce_dec); return 0; fail0: From 0937195715713b37ec843f28d99930dd7b1e8fbe Mon Sep 17 00:00:00 2001 From: Borislav Petkov Date: Thu, 8 Dec 2011 12:28:33 +0100 Subject: [PATCH 2/4] x86, MCE: Drain mcelog buffer Add a function which drains whatever MCEs were logged in already during boot and before the decoder chains were registered. Signed-off-by: Borislav Petkov --- arch/x86/kernel/cpu/mcheck/mce.c | 39 ++++++++++++++++++++++++++++++++ 1 file changed, 39 insertions(+) diff --git a/arch/x86/kernel/cpu/mcheck/mce.c b/arch/x86/kernel/cpu/mcheck/mce.c index c3c66ac6ef74..5be2464cce6a 100644 --- a/arch/x86/kernel/cpu/mcheck/mce.c +++ b/arch/x86/kernel/cpu/mcheck/mce.c @@ -189,9 +189,48 @@ void mce_log(struct mce *mce) set_bit(0, &mce_need_notify); } +static void drain_mcelog_buffer(void) +{ + unsigned int next, i, prev = 0; + + next = rcu_dereference_check_mce(mcelog.next); + + do { + struct mce *m; + + /* drain what was logged during boot */ + for (i = prev; i < next; i++) { + unsigned long start = jiffies; + unsigned retries = 1; + + m = &mcelog.entry[i]; + + while (!m->finished) { + if (time_after_eq(jiffies, start + 2*retries)) + retries++; + + cpu_relax(); + + if (!m->finished && retries >= 4) { + pr_err("MCE: skipping error being logged currently!\n"); + break; + } + } + smp_rmb(); + atomic_notifier_call_chain(&x86_mce_decoder_chain, 0, m); + } + + memset(mcelog.entry + prev, 0, (next - prev) * sizeof(*m)); + prev = next; + next = cmpxchg(&mcelog.next, prev, 0); + } while (next != prev); +} + + void mce_register_decode_chain(struct notifier_block *nb) { atomic_notifier_chain_register(&x86_mce_decoder_chain, nb); + drain_mcelog_buffer(); } EXPORT_SYMBOL_GPL(mce_register_decode_chain); From 29e9bf1841e4f9df13b4992a716fece7087dd237 Mon Sep 17 00:00:00 2001 From: Fenghua Yu Date: Fri, 4 Nov 2011 13:31:23 -0700 Subject: [PATCH 3/4] x86, mce, therm_throt: Don't report power limit and package level thermal throttle events in mcelog Thermal throttle and power limit events are not defined as MCE errors in x86 architecture and should not generate MCE errors in mcelog. Current kernel generates fake software defined MCE errors for these events. This may confuse users because they may think the machine has real MCE errors while actually only thermal throttle or power limit events happen. To make it worse, buggy firmware on some platforms may falsely generate the events. Therefore, kernel reports MCE errors which users think as real hardware errors. Although the firmware bugs should be fixed, on the other hand, kernel should not report MCE errors either. So mcelog is not a good mechanism to report these events. To report the events, we count them in respective counters (core_power_limit_count, package_power_limit_count, core_throttle_count, and package_throttle_count) in /sys/devices/system/cpu/cpu#/thermal_throttle/. Users can check the counters for each event on each CPU. Please note that all CPU's on one package report duplicate counters. It's user application's responsibity to retrieve a package level counter for one package. This patch doesn't report package level power limit, core level power limit, and package level thermal throttle events in mcelog. When the events happen, only report them in respective counters in sysfs. Since core level thermal throttle has been legacy code in kernel for a while and users accepted it as MCE error in mcelog, core level thermal throttle is still reported in mcelog. In the mean time, the event is counted in a counter in sysfs as well. Signed-off-by: Fenghua Yu Acked-by: Borislav Petkov Acked-by: Tony Luck Link: http://lkml.kernel.org/r/20111215001945.GA21009@linux-os.sc.intel.com Signed-off-by: H. Peter Anvin --- arch/x86/kernel/cpu/mcheck/therm_throt.c | 29 ++++++------------------ 1 file changed, 7 insertions(+), 22 deletions(-) diff --git a/arch/x86/kernel/cpu/mcheck/therm_throt.c b/arch/x86/kernel/cpu/mcheck/therm_throt.c index 787e06c84ea6..ce04b5804085 100644 --- a/arch/x86/kernel/cpu/mcheck/therm_throt.c +++ b/arch/x86/kernel/cpu/mcheck/therm_throt.c @@ -323,17 +323,6 @@ 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; @@ -363,27 +352,23 @@ 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); } } From 2c29d9dd577b74b44e580f957ea44d1df73af23a Mon Sep 17 00:00:00 2001 From: Chen Gong Date: Wed, 7 Dec 2011 09:21:37 -0800 Subject: [PATCH 4/4] x86: add IRQ context simulation in module mce-inject mce-inject provides a mechanism to simulate errors so that test scripts can check for correct operation of the kernel without requiring any specialized hardware to create rare events. The existing code can simulate events in normal process context and also in NMI context - but not in IRQ context. This patch fills that gap. Link: https://lkml.org/lkml/2011/12/7/537 Signed-off-by: Chen Gong Signed-off-by: Tony Luck --- arch/x86/include/asm/mce.h | 9 ++++--- arch/x86/kernel/cpu/mcheck/mce-inject.c | 34 ++++++++++++++++++++++--- 2 files changed, 35 insertions(+), 8 deletions(-) diff --git a/arch/x86/include/asm/mce.h b/arch/x86/include/asm/mce.h index 0e8ae57d3656..98165835dc68 100644 --- a/arch/x86/include/asm/mce.h +++ b/arch/x86/include/asm/mce.h @@ -50,10 +50,11 @@ #define MCJ_CTX_MASK 3 #define MCJ_CTX(flags) ((flags) & MCJ_CTX_MASK) #define MCJ_CTX_RANDOM 0 /* inject context: random */ -#define MCJ_CTX_PROCESS 1 /* inject context: process */ -#define MCJ_CTX_IRQ 2 /* inject context: IRQ */ -#define MCJ_NMI_BROADCAST 4 /* do NMI broadcasting */ -#define MCJ_EXCEPTION 8 /* raise as exception */ +#define MCJ_CTX_PROCESS 0x1 /* inject context: process */ +#define MCJ_CTX_IRQ 0x2 /* inject context: IRQ */ +#define MCJ_NMI_BROADCAST 0x4 /* do NMI broadcasting */ +#define MCJ_EXCEPTION 0x8 /* raise as exception */ +#define MCJ_IRQ_BRAODCAST 0x10 /* do IRQ broadcasting */ /* Fields are zero when not available */ struct mce { diff --git a/arch/x86/kernel/cpu/mcheck/mce-inject.c b/arch/x86/kernel/cpu/mcheck/mce-inject.c index 319882ef848d..fc4beb393577 100644 --- a/arch/x86/kernel/cpu/mcheck/mce-inject.c +++ b/arch/x86/kernel/cpu/mcheck/mce-inject.c @@ -17,6 +17,7 @@ #include #include #include +#include #include #include #include @@ -92,6 +93,18 @@ static int mce_raise_notify(unsigned int cmd, struct pt_regs *regs) return NMI_HANDLED; } +static void mce_irq_ipi(void *info) +{ + int cpu = smp_processor_id(); + struct mce *m = &__get_cpu_var(injectm); + + if (cpumask_test_cpu(cpu, mce_inject_cpumask) && + m->inject_flags & MCJ_EXCEPTION) { + cpumask_clear_cpu(cpu, mce_inject_cpumask); + raise_exception(m, NULL); + } +} + /* Inject mce on current CPU */ static int raise_local(void) { @@ -139,9 +152,10 @@ static void raise_mce(struct mce *m) return; #ifdef CONFIG_X86_LOCAL_APIC - if (m->inject_flags & MCJ_NMI_BROADCAST) { + if (m->inject_flags & (MCJ_IRQ_BRAODCAST | MCJ_NMI_BROADCAST)) { unsigned long start; int cpu; + get_online_cpus(); cpumask_copy(mce_inject_cpumask, cpu_online_mask); cpumask_clear_cpu(get_cpu(), mce_inject_cpumask); @@ -151,13 +165,25 @@ static void raise_mce(struct mce *m) MCJ_CTX(mcpu->inject_flags) != MCJ_CTX_RANDOM) cpumask_clear_cpu(cpu, mce_inject_cpumask); } - if (!cpumask_empty(mce_inject_cpumask)) - apic->send_IPI_mask(mce_inject_cpumask, NMI_VECTOR); + if (!cpumask_empty(mce_inject_cpumask)) { + if (m->inject_flags & MCJ_IRQ_BRAODCAST) { + /* + * don't wait because mce_irq_ipi is necessary + * to be sync with following raise_local + */ + preempt_disable(); + smp_call_function_many(mce_inject_cpumask, + mce_irq_ipi, NULL, 0); + preempt_enable(); + } else if (m->inject_flags & MCJ_NMI_BROADCAST) + apic->send_IPI_mask(mce_inject_cpumask, + NMI_VECTOR); + } start = jiffies; while (!cpumask_empty(mce_inject_cpumask)) { if (!time_before(jiffies, start + 2*HZ)) { printk(KERN_ERR - "Timeout waiting for mce inject NMI %lx\n", + "Timeout waiting for mce inject %lx\n", *cpumask_bits(mce_inject_cpumask)); break; }