Merge branch 'x86/mce2' into x86/core

arch/x86/Kconfig

@@ -783,6 +783,11 @@ config X86_MCE_AMD
 	   Additional support for AMD specific MCE features such as
 	   the DRAM Error Threshold.
 
+config X86_MCE_THRESHOLD
+	depends on X86_MCE_AMD || X86_MCE_INTEL
+	bool
+	default y
+
 config X86_MCE_NONFATAL
 	tristate "Check for non-fatal errors on AMD Athlon/Duron / Intel Pentium 4"
 	depends on X86_32 && X86_MCE

arch/x86/include/asm/apicdef.h

@@ -53,6 +53,7 @@
 #define		APIC_ESR_SENDILL	0x00020
 #define		APIC_ESR_RECVILL	0x00040
 #define		APIC_ESR_ILLREGA	0x00080
+#define 	APIC_LVTCMCI	0x2f0
 #define	APIC_ICR	0x300
 #define		APIC_DEST_SELF		0x40000
 #define		APIC_DEST_ALLINC	0x80000

arch/x86/include/asm/mce.h

@@ -11,6 +11,8 @@
  */
 
 #define MCG_CTL_P	 (1UL<<8)   /* MCG_CAP register available */
+#define MCG_EXT_P	 (1ULL<<9)   /* Extended registers available */
+#define MCG_CMCI_P	 (1ULL<<10)  /* CMCI supported */
 
 #define MCG_STATUS_RIPV  (1UL<<0)   /* restart ip valid */
 #define MCG_STATUS_EIPV  (1UL<<1)   /* ip points to correct instruction */
@@ -90,14 +92,29 @@ extern int mce_disabled;
 
 #include <asm/atomic.h>
 
+void mce_setup(struct mce *m);
 void mce_log(struct mce *m);
 DECLARE_PER_CPU(struct sys_device, device_mce);
 extern void (*threshold_cpu_callback)(unsigned long action, unsigned int cpu);
 
+/*
+ * To support more than 128 would need to escape the predefined
+ * Linux defined extended banks first.
+ */
+#define MAX_NR_BANKS (MCE_EXTENDED_BANK - 1)
+
 #ifdef CONFIG_X86_MCE_INTEL
 void mce_intel_feature_init(struct cpuinfo_x86 *c);
+void cmci_clear(void);
+void cmci_reenable(void);
+void cmci_rediscover(int dying);
+void cmci_recheck(void);
 #else
 static inline void mce_intel_feature_init(struct cpuinfo_x86 *c) { }
+static inline void cmci_clear(void) {}
+static inline void cmci_reenable(void) {}
+static inline void cmci_rediscover(int dying) {}
+static inline void cmci_recheck(void) {}
 #endif
 
 #ifdef CONFIG_X86_MCE_AMD
@@ -106,11 +123,23 @@ void mce_amd_feature_init(struct cpuinfo_x86 *c);
 static inline void mce_amd_feature_init(struct cpuinfo_x86 *c) { }
 #endif
 
-void mce_log_therm_throt_event(unsigned int cpu, __u64 status);
+extern int mce_available(struct cpuinfo_x86 *c);
+
+void mce_log_therm_throt_event(__u64 status);
 
 extern atomic_t mce_entry;
 
 extern void do_machine_check(struct pt_regs *, long);
+
+typedef DECLARE_BITMAP(mce_banks_t, MAX_NR_BANKS);
+DECLARE_PER_CPU(mce_banks_t, mce_poll_banks);
+
+enum mcp_flags {
+	MCP_TIMESTAMP = (1 << 0),	/* log time stamp */
+	MCP_UC = (1 << 1),		/* log uncorrected errors */
+};
+extern void machine_check_poll(enum mcp_flags flags, mce_banks_t *b);
+
 extern int mce_notify_user(void);
 
 #endif /* !CONFIG_X86_32 */
@@ -120,8 +149,8 @@ extern void mcheck_init(struct cpuinfo_x86 *c);
 #else
 #define mcheck_init(c) do { } while (0)
 #endif
-extern void stop_mce(void);
-extern void restart_mce(void);
+
+extern void (*mce_threshold_vector)(void);
 
 #endif /* __KERNEL__ */
 #endif /* _ASM_X86_MCE_H */

arch/x86/include/asm/msr-index.h

@@ -77,6 +77,11 @@
 #define MSR_IA32_MC0_ADDR		0x00000402
 #define MSR_IA32_MC0_MISC		0x00000403
 
+/* These are consecutive and not in the normal 4er MCE bank block */
+#define MSR_IA32_MC0_CTL2		0x00000280
+#define CMCI_EN			(1ULL << 30)
+#define CMCI_THRESHOLD_MASK		0xffffULL
+
 #define MSR_P6_PERFCTR0			0x000000c1
 #define MSR_P6_PERFCTR1			0x000000c2
 #define MSR_P6_EVNTSEL0			0x00000186

arch/x86/kernel/alternative.c

@@ -414,9 +414,17 @@ void __init alternative_instructions(void)
 	   that might execute the to be patched code.
 	   Other CPUs are not running. */
 	stop_nmi();
-#ifdef CONFIG_X86_MCE
-	stop_mce();
-#endif
+
+	/*
+	 * Don't stop machine check exceptions while patching.
+	 * MCEs only happen when something got corrupted and in this
+	 * case we must do something about the corruption.
+	 * Ignoring it is worse than a unlikely patching race.
+	 * Also machine checks tend to be broadcast and if one CPU
+	 * goes into machine check the others follow quickly, so we don't
+	 * expect a machine check to cause undue problems during to code
+	 * patching.
+	 */
 
 	apply_alternatives(__alt_instructions, __alt_instructions_end);
 
@@ -456,9 +464,6 @@ void __init alternative_instructions(void)
 				(unsigned long)__smp_locks_end);
 
 	restart_nmi();
-#ifdef CONFIG_X86_MCE
-	restart_mce();
-#endif
 }
 
 /**

arch/x86/kernel/apic/apic.c

@@ -46,6 +46,7 @@
 #include <asm/idle.h>
 #include <asm/mtrr.h>
 #include <asm/smp.h>
+#include <asm/mce.h>
 
 unsigned int num_processors;
 
@@ -842,6 +843,14 @@ void clear_local_APIC(void)
 		apic_write(APIC_LVTTHMR, v | APIC_LVT_MASKED);
 	}
 #endif
+#ifdef CONFIG_X86_MCE_INTEL
+	if (maxlvt >= 6) {
+		v = apic_read(APIC_LVTCMCI);
+		if (!(v & APIC_LVT_MASKED))
+			apic_write(APIC_LVTCMCI, v | APIC_LVT_MASKED);
+	}
+#endif
+
 	/*
 	 * Clean APIC state for other OSs:
 	 */
@@ -1241,6 +1250,12 @@ void __cpuinit setup_local_APIC(void)
 	apic_write(APIC_LVT1, value);
 
 	preempt_enable();
+
+#ifdef CONFIG_X86_MCE_INTEL
+	/* Recheck CMCI information after local APIC is up on CPU #0 */
+	if (smp_processor_id() == 0)
+		cmci_recheck();
+#endif
 }
 
 void __cpuinit end_local_APIC_setup(void)

arch/x86/kernel/cpu/mcheck/Makefile

@@ -4,3 +4,4 @@ obj-$(CONFIG_X86_32)		+= k7.o p4.o p5.o p6.o winchip.o
 obj-$(CONFIG_X86_MCE_INTEL)	+= mce_intel_64.o
 obj-$(CONFIG_X86_MCE_AMD)	+= mce_amd_64.o
 obj-$(CONFIG_X86_MCE_NONFATAL)	+= non-fatal.o
+obj-$(CONFIG_X86_MCE_THRESHOLD) += threshold.o

arch/x86/kernel/cpu/mcheck/mce_32.c

@@ -60,20 +60,6 @@ void mcheck_init(struct cpuinfo_x86 *c)
 	}
 }
 
-static unsigned long old_cr4 __initdata;
-
-void __init stop_mce(void)
-{
-	old_cr4 = read_cr4();
-	clear_in_cr4(X86_CR4_MCE);
-}
-
-void __init restart_mce(void)
-{
-	if (old_cr4 & X86_CR4_MCE)
-		set_in_cr4(X86_CR4_MCE);
-}
-
 static int __init mcheck_disable(char *str)
 {
 	mce_disabled = 1;