]> git.proxmox.com Git - mirror_ubuntu-artful-kernel.git/blame - arch/x86/include/asm/mce.h
x86/mce: Add Local MCE definitions
[mirror_ubuntu-artful-kernel.git] / arch / x86 / include / asm / mce.h
CommitLineData
1965aae3
PA
1#ifndef _ASM_X86_MCE_H
2#define _ASM_X86_MCE_H
e2f43029 3
af170c50 4#include <uapi/asm/mce.h>
e2f43029 5
f51bde6f
BP
6/*
7 * Machine Check support for x86
8 */
9
10/* MCG_CAP register defines */
11#define MCG_BANKCNT_MASK 0xff /* Number of Banks */
12#define MCG_CTL_P (1ULL<<8) /* MCG_CTL register available */
13#define MCG_EXT_P (1ULL<<9) /* Extended registers available */
14#define MCG_CMCI_P (1ULL<<10) /* CMCI supported */
15#define MCG_EXT_CNT_MASK 0xff0000 /* Number of Extended registers */
16#define MCG_EXT_CNT_SHIFT 16
17#define MCG_EXT_CNT(c) (((c) & MCG_EXT_CNT_MASK) >> MCG_EXT_CNT_SHIFT)
18#define MCG_SER_P (1ULL<<24) /* MCA recovery/new status bits */
4b3db708 19#define MCG_ELOG_P (1ULL<<26) /* Extended error log supported */
bc12edb8 20#define MCG_LMCE_P (1ULL<<27) /* Local machine check supported */
f51bde6f
BP
21
22/* MCG_STATUS register defines */
23#define MCG_STATUS_RIPV (1ULL<<0) /* restart ip valid */
24#define MCG_STATUS_EIPV (1ULL<<1) /* ip points to correct instruction */
25#define MCG_STATUS_MCIP (1ULL<<2) /* machine check in progress */
bc12edb8
AR
26#define MCG_STATUS_LMCES (1ULL<<3) /* LMCE signaled */
27
28/* MCG_EXT_CTL register defines */
29#define MCG_EXT_CTL_LMCE_EN (1ULL<<0) /* Enable LMCE */
f51bde6f
BP
30
31/* MCi_STATUS register defines */
32#define MCI_STATUS_VAL (1ULL<<63) /* valid error */
33#define MCI_STATUS_OVER (1ULL<<62) /* previous errors lost */
34#define MCI_STATUS_UC (1ULL<<61) /* uncorrected error */
35#define MCI_STATUS_EN (1ULL<<60) /* error enabled */
36#define MCI_STATUS_MISCV (1ULL<<59) /* misc error reg. valid */
37#define MCI_STATUS_ADDRV (1ULL<<58) /* addr reg. valid */
38#define MCI_STATUS_PCC (1ULL<<57) /* processor context corrupt */
39#define MCI_STATUS_S (1ULL<<56) /* Signaled machine check */
40#define MCI_STATUS_AR (1ULL<<55) /* Action required */
0ca06c08 41
e3480271
CY
42/* AMD-specific bits */
43#define MCI_STATUS_DEFERRED (1ULL<<44) /* declare an uncorrected error */
44#define MCI_STATUS_POISON (1ULL<<43) /* access poisonous data */
45
0ca06c08
TL
46/*
47 * Note that the full MCACOD field of IA32_MCi_STATUS MSR is
48 * bits 15:0. But bit 12 is the 'F' bit, defined for corrected
49 * errors to indicate that errors are being filtered by hardware.
50 * We should mask out bit 12 when looking for specific signatures
51 * of uncorrected errors - so the F bit is deliberately skipped
52 * in this #define.
53 */
54#define MCACOD 0xefff /* MCA Error Code */
f51bde6f
BP
55
56/* Architecturally defined codes from SDM Vol. 3B Chapter 15 */
57#define MCACOD_SCRUB 0x00C0 /* 0xC0-0xCF Memory Scrubbing */
0ca06c08 58#define MCACOD_SCRUBMSK 0xeff0 /* Skip bit 12 ('F' bit) */
f51bde6f
BP
59#define MCACOD_L3WB 0x017A /* L3 Explicit Writeback */
60#define MCACOD_DATA 0x0134 /* Data Load */
61#define MCACOD_INSTR 0x0150 /* Instruction Fetch */
62
63/* MCi_MISC register defines */
64#define MCI_MISC_ADDR_LSB(m) ((m) & 0x3f)
65#define MCI_MISC_ADDR_MODE(m) (((m) >> 6) & 7)
66#define MCI_MISC_ADDR_SEGOFF 0 /* segment offset */
67#define MCI_MISC_ADDR_LINEAR 1 /* linear address */
68#define MCI_MISC_ADDR_PHYS 2 /* physical address */
69#define MCI_MISC_ADDR_MEM 3 /* memory address */
70#define MCI_MISC_ADDR_GENERIC 7 /* generic */
71
72/* CTL2 register defines */
73#define MCI_CTL2_CMCI_EN (1ULL << 30)
74#define MCI_CTL2_CMCI_THRESHOLD_MASK 0x7fffULL
75
76#define MCJ_CTX_MASK 3
77#define MCJ_CTX(flags) ((flags) & MCJ_CTX_MASK)
78#define MCJ_CTX_RANDOM 0 /* inject context: random */
79#define MCJ_CTX_PROCESS 0x1 /* inject context: process */
80#define MCJ_CTX_IRQ 0x2 /* inject context: IRQ */
81#define MCJ_NMI_BROADCAST 0x4 /* do NMI broadcasting */
82#define MCJ_EXCEPTION 0x8 /* raise as exception */
a9093684 83#define MCJ_IRQ_BROADCAST 0x10 /* do IRQ broadcasting */
f51bde6f
BP
84
85#define MCE_OVERFLOW 0 /* bit 0 in flags means overflow */
86
87/* Software defined banks */
88#define MCE_EXTENDED_BANK 128
89#define MCE_THERMAL_BANK (MCE_EXTENDED_BANK + 0)
f51bde6f
BP
90
91#define MCE_LOG_LEN 32
92#define MCE_LOG_SIGNATURE "MACHINECHECK"
93
94/*
95 * This structure contains all data related to the MCE log. Also
96 * carries a signature to make it easier to find from external
97 * debugging tools. Each entry is only valid when its finished flag
98 * is set.
99 */
100struct mce_log {
101 char signature[12]; /* "MACHINECHECK" */
102 unsigned len; /* = MCE_LOG_LEN */
103 unsigned next;
104 unsigned flags;
105 unsigned recordlen; /* length of struct mce */
106 struct mce entry[MCE_LOG_LEN];
107};
d203f0b8
BP
108
109struct mca_config {
110 bool dont_log_ce;
7af19e4a
BP
111 bool cmci_disabled;
112 bool ignore_ce;
1462594b
BP
113 bool disabled;
114 bool ser;
115 bool bios_cmci_threshold;
d203f0b8 116 u8 banks;
84c2559d 117 s8 bootlog;
d203f0b8 118 int tolerant;
84c2559d 119 int monarch_timeout;
7af19e4a 120 int panic_timeout;
84c2559d 121 u32 rip_msr;
d203f0b8
BP
122};
123
bf80bbd7 124struct mce_vendor_flags {
7559e13f
AG
125 /*
126 * overflow recovery cpuid bit indicates that overflow
127 * conditions are not fatal
128 */
129 __u64 overflow_recov : 1,
130
131 /*
132 * SUCCOR stands for S/W UnCorrectable error COntainment
133 * and Recovery. It indicates support for data poisoning
134 * in HW and deferred error interrupts.
135 */
136 succor : 1,
137 __reserved_0 : 62;
bf80bbd7
AG
138};
139extern struct mce_vendor_flags mce_flags;
140
7af19e4a 141extern struct mca_config mca_cfg;
3653ada5
BP
142extern void mce_register_decode_chain(struct notifier_block *nb);
143extern void mce_unregister_decode_chain(struct notifier_block *nb);
df39a2e4 144
9e55e44e 145#include <linux/percpu.h>
60063497 146#include <linux/atomic.h>
9e55e44e 147
c6978369 148extern int mce_p5_enabled;
e2f43029 149
58995d2d 150#ifdef CONFIG_X86_MCE
a2202aa2 151int mcheck_init(void);
5e09954a 152void mcheck_cpu_init(struct cpuinfo_x86 *c);
43eaa2a1 153void mcheck_vendor_init_severity(void);
58995d2d 154#else
a2202aa2 155static inline int mcheck_init(void) { return 0; }
5e09954a 156static inline void mcheck_cpu_init(struct cpuinfo_x86 *c) {}
43eaa2a1 157static inline void mcheck_vendor_init_severity(void) {}
58995d2d
HS
158#endif
159
9e55e44e
HS
160#ifdef CONFIG_X86_ANCIENT_MCE
161void intel_p5_mcheck_init(struct cpuinfo_x86 *c);
162void winchip_mcheck_init(struct cpuinfo_x86 *c);
c6978369 163static inline void enable_p5_mce(void) { mce_p5_enabled = 1; }
9e55e44e
HS
164#else
165static inline void intel_p5_mcheck_init(struct cpuinfo_x86 *c) {}
166static inline void winchip_mcheck_init(struct cpuinfo_x86 *c) {}
c6978369 167static inline void enable_p5_mce(void) {}
9e55e44e
HS
168#endif
169
b5f2fa4e 170void mce_setup(struct mce *m);
e2f43029 171void mce_log(struct mce *m);
d6126ef5 172DECLARE_PER_CPU(struct device *, mce_device);
e2f43029 173
41fdff32 174/*
3ccdccfa
AK
175 * Maximum banks number.
176 * This is the limit of the current register layout on
177 * Intel CPUs.
41fdff32 178 */
3ccdccfa 179#define MAX_NR_BANKS 32
41fdff32 180
e2f43029
TG
181#ifdef CONFIG_X86_MCE_INTEL
182void mce_intel_feature_init(struct cpuinfo_x86 *c);
88ccbedd
AK
183void cmci_clear(void);
184void cmci_reenable(void);
7a0c819d 185void cmci_rediscover(void);
88ccbedd 186void cmci_recheck(void);
e2f43029
TG
187#else
188static inline void mce_intel_feature_init(struct cpuinfo_x86 *c) { }
88ccbedd
AK
189static inline void cmci_clear(void) {}
190static inline void cmci_reenable(void) {}
7a0c819d 191static inline void cmci_rediscover(void) {}
88ccbedd 192static inline void cmci_recheck(void) {}
e2f43029
TG
193#endif
194
195#ifdef CONFIG_X86_MCE_AMD
196void mce_amd_feature_init(struct cpuinfo_x86 *c);
197#else
198static inline void mce_amd_feature_init(struct cpuinfo_x86 *c) { }
199#endif
200
38736072 201int mce_available(struct cpuinfo_x86 *c);
88ccbedd 202
01ca79f1 203DECLARE_PER_CPU(unsigned, mce_exception_count);
ca84f696 204DECLARE_PER_CPU(unsigned, mce_poll_count);
01ca79f1 205
ee031c31
AK
206typedef DECLARE_BITMAP(mce_banks_t, MAX_NR_BANKS);
207DECLARE_PER_CPU(mce_banks_t, mce_poll_banks);
208
b79109c3 209enum mcp_flags {
3f2f0680
BP
210 MCP_TIMESTAMP = BIT(0), /* log time stamp */
211 MCP_UC = BIT(1), /* log uncorrected errors */
212 MCP_DONTLOG = BIT(2), /* only clear, don't log */
b79109c3 213};
3f2f0680 214bool machine_check_poll(enum mcp_flags flags, mce_banks_t *b);
b79109c3 215
9ff36ee9 216int mce_notify_irq(void);
e2f43029 217
ea149b36 218DECLARE_PER_CPU(struct mce, injectm);
66f5ddf3
TL
219
220extern void register_mce_write_callback(ssize_t (*)(struct file *filp,
221 const char __user *ubuf,
222 size_t usize, loff_t *off));
ea149b36 223
c3d1fb56
NR
224/* Disable CMCI/polling for MCA bank claimed by firmware */
225extern void mce_disable_bank(int bank);
226
58995d2d
HS
227/*
228 * Exception handler
229 */
230
231/* Call the installed machine check handler for this CPU setup. */
232extern void (*machine_check_vector)(struct pt_regs *, long error_code);
233void do_machine_check(struct pt_regs *, long);
234
235/*
236 * Threshold handler
237 */
e2f43029 238
b2762686 239extern void (*mce_threshold_vector)(void);
58995d2d 240extern void (*threshold_cpu_callback)(unsigned long action, unsigned int cpu);
b2762686 241
24fd78a8
AG
242/* Deferred error interrupt handler */
243extern void (*deferred_error_int_vector)(void);
244
e8ce2c5e
HS
245/*
246 * Thermal handler
247 */
248
e8ce2c5e
HS
249void intel_init_thermal(struct cpuinfo_x86 *c);
250
e8ce2c5e 251void mce_log_therm_throt_event(__u64 status);
a2202aa2 252
9e76a97e
D
253/* Interrupt Handler for core thermal thresholds */
254extern int (*platform_thermal_notify)(__u64 msr_val);
255
25cdce17
SP
256/* Interrupt Handler for package thermal thresholds */
257extern int (*platform_thermal_package_notify)(__u64 msr_val);
258
259/* Callback support of rate control, return true, if
260 * callback has rate control */
261extern bool (*platform_thermal_package_rate_control)(void);
262
a2202aa2
YW
263#ifdef CONFIG_X86_THERMAL_VECTOR
264extern void mcheck_intel_therm_init(void);
265#else
266static inline void mcheck_intel_therm_init(void) { }
267#endif
268
d334a491
HY
269/*
270 * Used by APEI to report memory error via /dev/mcelog
271 */
272
273struct cper_sec_mem_err;
274extern void apei_mce_report_mem_error(int corrected,
275 struct cper_sec_mem_err *mem_err);
276
1965aae3 277#endif /* _ASM_X86_MCE_H */