1 files changed, 110 insertions, 64 deletions

diff --git a/arch/x86/kernel/cpu/mcheck/mce.c b/arch/x86/kernel/cpu/mcheck/mce.c
index 4b2af86e3e8d..3ab9c886b613 100644
--- a/arch/x86/kernel/cpu/mcheck/mce.c
+++ b/arch/x86/kernel/cpu/mcheck/mce.c
@@ -46,6 +46,16 @@
 
 #include "mce-internal.h"
 
+static DEFINE_MUTEX(mce_read_mutex);
+
+#define rcu_dereference_check_mce(p) \
+	rcu_dereference_check((p), \
+			      rcu_read_lock_sched_held() || \
+			      lockdep_is_held(&mce_read_mutex))
+
+#define CREATE_TRACE_POINTS
+#include <trace/events/mce.h>
+
 int mce_disabled __read_mostly;
 
 #define MISC_MCELOG_MINOR	227
@@ -85,6 +95,26 @@ static DECLARE_WAIT_QUEUE_HEAD(mce_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);
+
+static int default_decode_mce(struct notifier_block *nb, unsigned long val,
+			       void *data)
+{
+	pr_emerg("No human readable MCE decoding support on this CPU type.\n");
+	pr_emerg("Run the message through 'mcelog --ascii' to decode.\n");
+
+	return NOTIFY_STOP;
+}
+
+static struct notifier_block mce_dec_nb = {
+	.notifier_call = default_decode_mce,
+	.priority      = -1,
+};
 
 /* MCA banks polled by the period polling timer for corrected events */
 DEFINE_PER_CPU(mce_banks_t, mce_poll_banks) = {
@@ -129,10 +159,13 @@ void mce_log(struct mce *mce)
 {
 	unsigned next, entry;
 
+	/* Emit the trace record: */
+	trace_mce_record(mce);
+
 	mce->finished = 0;
 	wmb();
 	for (;;) {
-		entry = rcu_dereference(mcelog.next);
+		entry = rcu_dereference_check_mce(mcelog.next);
 		for (;;) {
 			/*
 			 * When the buffer fills up discard new entries.
@@ -165,49 +198,46 @@ void mce_log(struct mce *mce)
 	set_bit(0, &mce_need_notify);
 }
 
-void __weak decode_mce(struct mce *m)
-{
-	return;
-}
-
 static void print_mce(struct mce *m)
 {
-	printk(KERN_EMERG
-	       "CPU %d: Machine Check Exception: %16Lx Bank %d: %016Lx\n",
+	pr_emerg("CPU %d: Machine Check Exception: %16Lx Bank %d: %016Lx\n",
 	       m->extcpu, m->mcgstatus, m->bank, m->status);
+
 	if (m->ip) {
-		printk(KERN_EMERG "RIP%s %02x:<%016Lx> ",
-		       !(m->mcgstatus & MCG_STATUS_EIPV) ? " !INEXACT!" : "",
-		       m->cs, m->ip);
+		pr_emerg("RIP%s %02x:<%016Lx> ",
+			!(m->mcgstatus & MCG_STATUS_EIPV) ? " !INEXACT!" : "",
+				m->cs, m->ip);
+
 		if (m->cs == __KERNEL_CS)
 			print_symbol("{%s}", m->ip);
-		printk(KERN_CONT "\n");
+		pr_cont("\n");
 	}
-	printk(KERN_EMERG "TSC %llx ", m->tsc);
+
+	pr_emerg("TSC %llx ", m->tsc);
 	if (m->addr)
-		printk(KERN_CONT "ADDR %llx ", m->addr);
+		pr_cont("ADDR %llx ", m->addr);
 	if (m->misc)
-		printk(KERN_CONT "MISC %llx ", m->misc);
-	printk(KERN_CONT "\n");
-	printk(KERN_EMERG "PROCESSOR %u:%x TIME %llu SOCKET %u APIC %x\n",
-			m->cpuvendor, m->cpuid, m->time, m->socketid,
-			m->apicid);
+		pr_cont("MISC %llx ", m->misc);
+
+	pr_cont("\n");
+	pr_emerg("PROCESSOR %u:%x TIME %llu SOCKET %u APIC %x\n",
+		m->cpuvendor, m->cpuid, m->time, m->socketid, m->apicid);
 
-	decode_mce(m);
+	/*
+	 * Print out human-readable details about the MCE error,
+	 * (if the CPU has an implementation for that)
+	 */
+	atomic_notifier_call_chain(&x86_mce_decoder_chain, 0, m);
 }
 
 static void print_mce_head(void)
 {
-	printk(KERN_EMERG "\nHARDWARE ERROR\n");
+	pr_emerg("\nHARDWARE ERROR\n");
 }
 
 static void print_mce_tail(void)
 {
-	printk(KERN_EMERG "This is not a software problem!\n"
-#if (!defined(CONFIG_EDAC) || !defined(CONFIG_CPU_SUP_AMD))
-	       "Run through mcelog --ascii to decode and contact your hardware vendor\n"
-#endif
-	       );
+	pr_emerg("This is not a software problem!\n");
 }
 
 #define PANIC_TIMEOUT 5 /* 5 seconds */
@@ -221,6 +251,7 @@ static atomic_t mce_fake_paniced;
 static void wait_for_panic(void)
 {
 	long timeout = PANIC_TIMEOUT*USEC_PER_SEC;
+
 	preempt_disable();
 	local_irq_enable();
 	while (timeout-- > 0)
@@ -288,6 +319,7 @@ static void mce_panic(char *msg, struct mce *final, char *exp)
 static int msr_to_offset(u32 msr)
 {
 	unsigned bank = __get_cpu_var(injectm.bank);
+
 	if (msr == rip_msr)
 		return offsetof(struct mce, ip);
 	if (msr == MSR_IA32_MCx_STATUS(bank))
@@ -1111,7 +1143,7 @@ static int check_interval = 5 * 60; /* 5 minutes */
 static DEFINE_PER_CPU(int, mce_next_interval); /* in jiffies */
 static DEFINE_PER_CPU(struct timer_list, mce_timer);
 
-static void mcheck_timer(unsigned long data)
+static void mce_start_timer(unsigned long data)
 {
 	struct timer_list *t = &per_cpu(mce_timer, data);
 	int *n;
@@ -1176,7 +1208,7 @@ int mce_notify_irq(void)
 }
 EXPORT_SYMBOL_GPL(mce_notify_irq);
 
-static int mce_banks_init(void)
+static int __cpuinit __mcheck_cpu_mce_banks_init(void)
 {
 	int i;
 
@@ -1195,7 +1227,7 @@ static int mce_banks_init(void)
 /*
  * Initialize Machine Checks for a CPU.
  */
-static int __cpuinit mce_cap_init(void)
+static int __cpuinit __mcheck_cpu_cap_init(void)
 {
 	unsigned b;
 	u64 cap;
@@ -1203,7 +1235,8 @@ static int __cpuinit mce_cap_init(void)
 	rdmsrl(MSR_IA32_MCG_CAP, cap);
 
 	b = cap & MCG_BANKCNT_MASK;
-	printk(KERN_INFO "mce: CPU supports %d MCE banks\n", b);
+	if (!banks)
+		printk(KERN_INFO "mce: CPU supports %d MCE banks\n", b);
 
 	if (b > MAX_NR_BANKS) {
 		printk(KERN_WARNING
@@ -1216,7 +1249,7 @@ static int __cpuinit mce_cap_init(void)
 	WARN_ON(banks != 0 && b != banks);
 	banks = b;
 	if (!mce_banks) {
-		int err = mce_banks_init();
+		int err = __mcheck_cpu_mce_banks_init();
 
 		if (err)
 			return err;
@@ -1232,7 +1265,7 @@ static int __cpuinit mce_cap_init(void)
 	return 0;
 }
 
-static void mce_init(void)
+static void __mcheck_cpu_init_generic(void)
 {
 	mce_banks_t all_banks;
 	u64 cap;
@@ -1261,7 +1294,7 @@ static void mce_init(void)
 }
 
 /* Add per CPU specific workarounds here */
-static int __cpuinit mce_cpu_quirks(struct cpuinfo_x86 *c)
+static int __cpuinit __mcheck_cpu_apply_quirks(struct cpuinfo_x86 *c)
 {
 	if (c->x86_vendor == X86_VENDOR_UNKNOWN) {
 		pr_info("MCE: unknown CPU type - not enabling MCE support.\n");
@@ -1329,7 +1362,7 @@ static int __cpuinit mce_cpu_quirks(struct cpuinfo_x86 *c)
 	return 0;
 }
 
-static void __cpuinit mce_ancient_init(struct cpuinfo_x86 *c)
+static void __cpuinit __mcheck_cpu_ancient_init(struct cpuinfo_x86 *c)
 {
 	if (c->x86 != 5)
 		return;
@@ -1343,7 +1376,7 @@ static void __cpuinit mce_ancient_init(struct cpuinfo_x86 *c)
 	}
 }
 
-static void mce_cpu_features(struct cpuinfo_x86 *c)
+static void __mcheck_cpu_init_vendor(struct cpuinfo_x86 *c)
 {
 	switch (c->x86_vendor) {
 	case X86_VENDOR_INTEL:
@@ -1357,18 +1390,19 @@ static void mce_cpu_features(struct cpuinfo_x86 *c)
 	}
 }
 
-static void mce_init_timer(void)
+static void __mcheck_cpu_init_timer(void)
 {
 	struct timer_list *t = &__get_cpu_var(mce_timer);
 	int *n = &__get_cpu_var(mce_next_interval);
 
+	setup_timer(t, mce_start_timer, smp_processor_id());
+
 	if (mce_ignore_ce)
 		return;
 
 	*n = check_interval * HZ;
 	if (!*n)
 		return;
-	setup_timer(t, mcheck_timer, smp_processor_id());
 	t->expires = round_jiffies(jiffies + *n);
 	add_timer_on(t, smp_processor_id());
 }
@@ -1388,27 +1422,28 @@ void (*machine_check_vector)(struct pt_regs *, long error_code) =
  * Called for each booted CPU to set up machine checks.
  * Must be called with preempt off:
  */
-void __cpuinit mcheck_init(struct cpuinfo_x86 *c)
+void __cpuinit mcheck_cpu_init(struct cpuinfo_x86 *c)
 {
 	if (mce_disabled)
 		return;
 
-	mce_ancient_init(c);
+	__mcheck_cpu_ancient_init(c);
 
 	if (!mce_available(c))
 		return;
 
-	if (mce_cap_init() < 0 || mce_cpu_quirks(c) < 0) {
+	if (__mcheck_cpu_cap_init() < 0 || __mcheck_cpu_apply_quirks(c) < 0) {
 		mce_disabled = 1;
 		return;
 	}
 
 	machine_check_vector = do_machine_check;
 
-	mce_init();
-	mce_cpu_features(c);
-	mce_init_timer();
+	__mcheck_cpu_init_generic();
+	__mcheck_cpu_init_vendor(c);
+	__mcheck_cpu_init_timer();
 	INIT_WORK(&__get_cpu_var(mce_work), mce_process_work);
+
 }
 
 /*
@@ -1457,8 +1492,6 @@ static void collect_tscs(void *data)
 	rdtscll(cpu_tsc[smp_processor_id()]);
 }
 
-static DEFINE_MUTEX(mce_read_mutex);
-
 static ssize_t mce_read(struct file *filp, char __user *ubuf, size_t usize,
 			loff_t *off)
 {
@@ -1472,7 +1505,7 @@ static ssize_t mce_read(struct file *filp, char __user *ubuf, size_t usize,
 		return -ENOMEM;
 
 	mutex_lock(&mce_read_mutex);
-	next = rcu_dereference(mcelog.next);
+	next = rcu_dereference_check_mce(mcelog.next);
 
 	/* Only supports full reads right now */
 	if (*off != 0 || usize < MCE_LOG_LEN*sizeof(struct mce)) {
@@ -1537,7 +1570,7 @@ timeout:
 static unsigned int mce_poll(struct file *file, poll_table *wait)
 {
 	poll_wait(file, &mce_wait, wait);
-	if (rcu_dereference(mcelog.next))
+	if (rcu_dereference_check_mce(mcelog.next))
 		return POLLIN | POLLRDNORM;
 	return 0;
 }
@@ -1628,6 +1661,15 @@ static int __init mcheck_enable(char *str)
 }
 __setup("mce", mcheck_enable);
 
+int __init mcheck_init(void)
+{
+	atomic_notifier_chain_register(&x86_mce_decoder_chain, &mce_dec_nb);
+
+	mcheck_intel_therm_init();
+
+	return 0;
+}
+
 /*
  * Sysfs support
  */
@@ -1636,7 +1678,7 @@ __setup("mce", mcheck_enable);
  * Disable machine checks on suspend and shutdown. We can't really handle
  * them later.
  */
-static int mce_disable(void)
+static int mce_disable_error_reporting(void)
 {
 	int i;
 
@@ -1651,12 +1693,12 @@ static int mce_disable(void)
 
 static int mce_suspend(struct sys_device *dev, pm_message_t state)
 {
-	return mce_disable();
+	return mce_disable_error_reporting();
 }
 
 static int mce_shutdown(struct sys_device *dev)
 {
-	return mce_disable();
+	return mce_disable_error_reporting();
 }
 
 /*
@@ -1666,8 +1708,8 @@ static int mce_shutdown(struct sys_device *dev)
  */
 static int mce_resume(struct sys_device *dev)
 {
-	mce_init();
-	mce_cpu_features(&current_cpu_data);
+	__mcheck_cpu_init_generic();
+	__mcheck_cpu_init_vendor(&current_cpu_data);
 
 	return 0;
 }
@@ -1677,8 +1719,8 @@ static void mce_cpu_restart(void *data)
 	del_timer_sync(&__get_cpu_var(mce_timer));
 	if (!mce_available(&current_cpu_data))
 		return;
-	mce_init();
-	mce_init_timer();
+	__mcheck_cpu_init_generic();
+	__mcheck_cpu_init_timer();
 }
 
 /* Reinit MCEs after user configuration changes */
@@ -1704,7 +1746,7 @@ static void mce_enable_ce(void *all)
 	cmci_reenable();
 	cmci_recheck();
 	if (all)
-		mce_init_timer();
+		__mcheck_cpu_init_timer();
 }
 
 static struct sysdev_class mce_sysclass = {
@@ -1892,7 +1934,7 @@ error2:
 		sysdev_remove_file(&per_cpu(mce_dev, cpu), &mce_banks[j].attr);
 error:
 	while (--i >= 0)
-		sysdev_remove_file(&per_cpu(mce_dev, cpu), &mce_banks[i].attr);
+		sysdev_remove_file(&per_cpu(mce_dev, cpu), mce_attrs[i]);
 
 	sysdev_unregister(&per_cpu(mce_dev, cpu));
 
@@ -1917,13 +1959,14 @@ static __cpuinit void mce_remove_device(unsigned int cpu)
 }
 
 /* Make sure there are no machine checks on offlined CPUs. */
-static void mce_disable_cpu(void *h)
+static void __cpuinit mce_disable_cpu(void *h)
 {
 	unsigned long action = *(unsigned long *)h;
 	int i;
 
 	if (!mce_available(&current_cpu_data))
 		return;
+
 	if (!(action & CPU_TASKS_FROZEN))
 		cmci_clear();
 	for (i = 0; i < banks; i++) {
@@ -1934,7 +1977,7 @@ static void mce_disable_cpu(void *h)
 	}
 }
 
-static void mce_reenable_cpu(void *h)
+static void __cpuinit mce_reenable_cpu(void *h)
 {
 	unsigned long action = *(unsigned long *)h;
 	int i;
@@ -1979,9 +2022,11 @@ mce_cpu_callback(struct notifier_block *nfb, unsigned long action, void *hcpu)
 		break;
 	case CPU_DOWN_FAILED:
 	case CPU_DOWN_FAILED_FROZEN:
-		t->expires = round_jiffies(jiffies +
+		if (!mce_ignore_ce && check_interval) {
+			t->expires = round_jiffies(jiffies +
 					   __get_cpu_var(mce_next_interval));
-		add_timer_on(t, cpu);
+			add_timer_on(t, cpu);
+		}
 		smp_call_function_single(cpu, mce_reenable_cpu, &action, 1);
 		break;
 	case CPU_POST_DEAD:
@@ -2004,6 +2049,7 @@ static __init void mce_init_banks(void)
 		struct mce_bank *b = &mce_banks[i];
 		struct sysdev_attribute *a = &b->attr;
 
+		sysfs_attr_init(&a->attr);
 		a->attr.name	= b->attrname;
 		snprintf(b->attrname, ATTR_LEN, "bank%d", i);
 
@@ -2013,7 +2059,7 @@ static __init void mce_init_banks(void)
 	}
 }
 
-static __init int mce_init_device(void)
+static __init int mcheck_init_device(void)
 {
 	int err;
 	int i = 0;
@@ -2041,7 +2087,7 @@ static __init int mce_init_device(void)
 	return err;
 }
 
-device_initcall(mce_init_device);
+device_initcall(mcheck_init_device);
 
 /*
  * Old style boot options parsing. Only for compatibility.
@@ -2089,7 +2135,7 @@ static int fake_panic_set(void *data, u64 val)
 DEFINE_SIMPLE_ATTRIBUTE(fake_panic_fops, fake_panic_get,
 			fake_panic_set, "%llu\n");
 
-static int __init mce_debugfs_init(void)
+static int __init mcheck_debugfs_init(void)
 {
 	struct dentry *dmce, *ffake_panic;
 
@@ -2103,5 +2149,5 @@ static int __init mce_debugfs_init(void)
 
 	return 0;
 }
-late_initcall(mce_debugfs_init);
+late_initcall(mcheck_debugfs_init);
 #endif