diff options
Diffstat (limited to 'arch/x86/kernel/cpu')
24 files changed, 962 insertions, 874 deletions
diff --git a/arch/x86/kernel/cpu/Makefile b/arch/x86/kernel/cpu/Makefile index c1f253dac15..8dd30638fe4 100644 --- a/arch/x86/kernel/cpu/Makefile +++ b/arch/x86/kernel/cpu/Makefile @@ -13,7 +13,7 @@ CFLAGS_common.o := $(nostackp) obj-y := intel_cacheinfo.o addon_cpuid_features.o obj-y += proc.o capflags.o powerflags.o common.o -obj-y += vmware.o hypervisor.o +obj-y += vmware.o hypervisor.o sched.o obj-$(CONFIG_X86_32) += bugs.o cmpxchg.o obj-$(CONFIG_X86_64) += bugs_64.o diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c index 83b217c7225..f32fa71ccf9 100644 --- a/arch/x86/kernel/cpu/amd.c +++ b/arch/x86/kernel/cpu/amd.c @@ -253,6 +253,64 @@ static int __cpuinit nearby_node(int apicid) #endif /* + * Fixup core topology information for AMD multi-node processors. + * Assumption 1: Number of cores in each internal node is the same. + * Assumption 2: Mixed systems with both single-node and dual-node + * processors are not supported. + */ +#ifdef CONFIG_X86_HT +static void __cpuinit amd_fixup_dcm(struct cpuinfo_x86 *c) +{ +#ifdef CONFIG_PCI + u32 t, cpn; + u8 n, n_id; + int cpu = smp_processor_id(); + + /* fixup topology information only once for a core */ + if (cpu_has(c, X86_FEATURE_AMD_DCM)) + return; + + /* check for multi-node processor on boot cpu */ + t = read_pci_config(0, 24, 3, 0xe8); + if (!(t & (1 << 29))) + return; + + set_cpu_cap(c, X86_FEATURE_AMD_DCM); + + /* cores per node: each internal node has half the number of cores */ + cpn = c->x86_max_cores >> 1; + + /* even-numbered NB_id of this dual-node processor */ + n = c->phys_proc_id << 1; + + /* + * determine internal node id and assign cores fifty-fifty to + * each node of the dual-node processor + */ + t = read_pci_config(0, 24 + n, 3, 0xe8); + n = (t>>30) & 0x3; + if (n == 0) { + if (c->cpu_core_id < cpn) + n_id = 0; + else + n_id = 1; + } else { + if (c->cpu_core_id < cpn) + n_id = 1; + else + n_id = 0; + } + + /* compute entire NodeID, use llc_shared_map to store sibling info */ + per_cpu(cpu_llc_id, cpu) = (c->phys_proc_id << 1) + n_id; + + /* fixup core id to be in range from 0 to cpn */ + c->cpu_core_id = c->cpu_core_id % cpn; +#endif +} +#endif + +/* * On a AMD dual core setup the lower bits of the APIC id distingush the cores. * Assumes number of cores is a power of two. */ @@ -269,17 +327,31 @@ static void __cpuinit amd_detect_cmp(struct cpuinfo_x86 *c) c->phys_proc_id = c->initial_apicid >> bits; /* use socket ID also for last level cache */ per_cpu(cpu_llc_id, cpu) = c->phys_proc_id; + /* fixup topology information on multi-node processors */ + if ((c->x86 == 0x10) && (c->x86_model == 9)) + amd_fixup_dcm(c); #endif } +int amd_get_nb_id(int cpu) +{ + int id = 0; +#ifdef CONFIG_SMP + id = per_cpu(cpu_llc_id, cpu); +#endif + return id; +} +EXPORT_SYMBOL_GPL(amd_get_nb_id); + static void __cpuinit srat_detect_node(struct cpuinfo_x86 *c) { #if defined(CONFIG_NUMA) && defined(CONFIG_X86_64) int cpu = smp_processor_id(); int node; - unsigned apicid = cpu_has_apic ? hard_smp_processor_id() : c->apicid; + unsigned apicid = c->apicid; + + node = per_cpu(cpu_llc_id, cpu); - node = c->phys_proc_id; if (apicid_to_node[apicid] != NUMA_NO_NODE) node = apicid_to_node[apicid]; if (!node_online(node)) { @@ -406,12 +478,24 @@ static void __cpuinit init_amd(struct cpuinfo_x86 *c) /* * Some BIOSes incorrectly force this feature, but only K8 * revision D (model = 0x14) and later actually support it. + * (AMD Erratum #110, docId: 25759). */ - if (c->x86_model < 0x14) + if (c->x86_model < 0x14 && cpu_has(c, X86_FEATURE_LAHF_LM)) { + u64 val; + clear_cpu_cap(c, X86_FEATURE_LAHF_LM); + if (!rdmsrl_amd_safe(0xc001100d, &val)) { + val &= ~(1ULL << 32); + wrmsrl_amd_safe(0xc001100d, val); + } + } + } if (c->x86 == 0x10 || c->x86 == 0x11) set_cpu_cap(c, X86_FEATURE_REP_GOOD); + + /* get apicid instead of initial apic id from cpuid */ + c->apicid = hard_smp_processor_id(); #else /* diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c index 734eaad9365..2055fc2b2e6 100644 --- a/arch/x86/kernel/cpu/common.c +++ b/arch/x86/kernel/cpu/common.c @@ -94,45 +94,45 @@ DEFINE_PER_CPU_PAGE_ALIGNED(struct gdt_page, gdt_page) = { .gdt = { * TLS descriptors are currently at a different place compared to i386. * Hopefully nobody expects them at a fixed place (Wine?) */ - [GDT_ENTRY_KERNEL32_CS] = { { { 0x0000ffff, 0x00cf9b00 } } }, - [GDT_ENTRY_KERNEL_CS] = { { { 0x0000ffff, 0x00af9b00 } } }, - [GDT_ENTRY_KERNEL_DS] = { { { 0x0000ffff, 0x00cf9300 } } }, - [GDT_ENTRY_DEFAULT_USER32_CS] = { { { 0x0000ffff, 0x00cffb00 } } }, - [GDT_ENTRY_DEFAULT_USER_DS] = { { { 0x0000ffff, 0x00cff300 } } }, - [GDT_ENTRY_DEFAULT_USER_CS] = { { { 0x0000ffff, 0x00affb00 } } }, + [GDT_ENTRY_KERNEL32_CS] = GDT_ENTRY_INIT(0xc09b, 0, 0xfffff), + [GDT_ENTRY_KERNEL_CS] = GDT_ENTRY_INIT(0xa09b, 0, 0xfffff), + [GDT_ENTRY_KERNEL_DS] = GDT_ENTRY_INIT(0xc093, 0, 0xfffff), + [GDT_ENTRY_DEFAULT_USER32_CS] = GDT_ENTRY_INIT(0xc0fb, 0, 0xfffff), + [GDT_ENTRY_DEFAULT_USER_DS] = GDT_ENTRY_INIT(0xc0f3, 0, 0xfffff), + [GDT_ENTRY_DEFAULT_USER_CS] = GDT_ENTRY_INIT(0xa0fb, 0, 0xfffff), #else - [GDT_ENTRY_KERNEL_CS] = { { { 0x0000ffff, 0x00cf9a00 } } }, - [GDT_ENTRY_KERNEL_DS] = { { { 0x0000ffff, 0x00cf9200 } } }, - [GDT_ENTRY_DEFAULT_USER_CS] = { { { 0x0000ffff, 0x00cffa00 } } }, - [GDT_ENTRY_DEFAULT_USER_DS] = { { { 0x0000ffff, 0x00cff200 } } }, + [GDT_ENTRY_KERNEL_CS] = GDT_ENTRY_INIT(0xc09a, 0, 0xfffff), + [GDT_ENTRY_KERNEL_DS] = GDT_ENTRY_INIT(0xc092, 0, 0xfffff), + [GDT_ENTRY_DEFAULT_USER_CS] = GDT_ENTRY_INIT(0xc0fa, 0, 0xfffff), + [GDT_ENTRY_DEFAULT_USER_DS] = GDT_ENTRY_INIT(0xc0f2, 0, 0xfffff), /* * Segments used for calling PnP BIOS have byte granularity. * They code segments and data segments have fixed 64k limits, * the transfer segment sizes are set at run time. */ /* 32-bit code */ - [GDT_ENTRY_PNPBIOS_CS32] = { { { 0x0000ffff, 0x00409a00 } } }, + [GDT_ENTRY_PNPBIOS_CS32] = GDT_ENTRY_INIT(0x409a, 0, 0xffff), /* 16-bit code */ - [GDT_ENTRY_PNPBIOS_CS16] = { { { 0x0000ffff, 0x00009a00 } } }, + [GDT_ENTRY_PNPBIOS_CS16] = GDT_ENTRY_INIT(0x009a, 0, 0xffff), /* 16-bit data */ - [GDT_ENTRY_PNPBIOS_DS] = { { { 0x0000ffff, 0x00009200 } } }, + [GDT_ENTRY_PNPBIOS_DS] = GDT_ENTRY_INIT(0x0092, 0, 0xffff), /* 16-bit data */ - [GDT_ENTRY_PNPBIOS_TS1] = { { { 0x00000000, 0x00009200 } } }, + [GDT_ENTRY_PNPBIOS_TS1] = GDT_ENTRY_INIT(0x0092, 0, 0), /* 16-bit data */ - [GDT_ENTRY_PNPBIOS_TS2] = { { { 0x00000000, 0x00009200 } } }, + [GDT_ENTRY_PNPBIOS_TS2] = GDT_ENTRY_INIT(0x0092, 0, 0), /* * The APM segments have byte granularity and their bases * are set at run time. All have 64k limits. */ /* 32-bit code */ - [GDT_ENTRY_APMBIOS_BASE] = { { { 0x0000ffff, 0x00409a00 } } }, + [GDT_ENTRY_APMBIOS_BASE] = GDT_ENTRY_INIT(0x409a, 0, 0xffff), /* 16-bit code */ - [GDT_ENTRY_APMBIOS_BASE+1] = { { { 0x0000ffff, 0x00009a00 } } }, + [GDT_ENTRY_APMBIOS_BASE+1] = GDT_ENTRY_INIT(0x009a, 0, 0xffff), /* data */ - [GDT_ENTRY_APMBIOS_BASE+2] = { { { 0x0000ffff, 0x00409200 } } }, + [GDT_ENTRY_APMBIOS_BASE+2] = GDT_ENTRY_INIT(0x4092, 0, 0xffff), - [GDT_ENTRY_ESPFIX_SS] = { { { 0x0000ffff, 0x00cf9200 } } }, - [GDT_ENTRY_PERCPU] = { { { 0x0000ffff, 0x00cf9200 } } }, + [GDT_ENTRY_ESPFIX_SS] = GDT_ENTRY_INIT(0xc092, 0, 0xfffff), + [GDT_ENTRY_PERCPU] = GDT_ENTRY_INIT(0xc092, 0, 0xfffff), GDT_STACK_CANARY_INIT #endif } }; @@ -987,13 +987,21 @@ struct desc_ptr idt_descr = { NR_VECTORS * 16 - 1, (unsigned long) idt_table }; DEFINE_PER_CPU_FIRST(union irq_stack_union, irq_stack_union) __aligned(PAGE_SIZE); -DEFINE_PER_CPU(char *, irq_stack_ptr) = - init_per_cpu_var(irq_stack_union.irq_stack) + IRQ_STACK_SIZE - 64; +/* + * The following four percpu variables are hot. Align current_task to + * cacheline size such that all four fall in the same cacheline. + */ +DEFINE_PER_CPU(struct task_struct *, current_task) ____cacheline_aligned = + &init_task; +EXPORT_PER_CPU_SYMBOL(current_task); DEFINE_PER_CPU(unsigned long, kernel_stack) = (unsigned long)&init_thread_union - KERNEL_STACK_OFFSET + THREAD_SIZE; EXPORT_PER_CPU_SYMBOL(kernel_stack); +DEFINE_PER_CPU(char *, irq_stack_ptr) = + init_per_cpu_var(irq_stack_union.irq_stack) + IRQ_STACK_SIZE - 64; + DEFINE_PER_CPU(unsigned int, irq_count) = -1; /* @@ -1008,8 +1016,7 @@ static const unsigned int exception_stack_sizes[N_EXCEPTION_STACKS] = { }; static DEFINE_PER_CPU_PAGE_ALIGNED(char, exception_stacks - [(N_EXCEPTION_STACKS - 1) * EXCEPTION_STKSZ + DEBUG_STKSZ]) - __aligned(PAGE_SIZE); + [(N_EXCEPTION_STACKS - 1) * EXCEPTION_STKSZ + DEBUG_STKSZ]); /* May not be marked __init: used by software suspend */ void syscall_init(void) @@ -1042,8 +1049,11 @@ DEFINE_PER_CPU(struct orig_ist, orig_ist); #else /* CONFIG_X86_64 */ +DEFINE_PER_CPU(struct task_struct *, current_task) = &init_task; +EXPORT_PER_CPU_SYMBOL(current_task); + #ifdef CONFIG_CC_STACKPROTECTOR -DEFINE_PER_CPU(unsigned long, stack_canary); +DEFINE_PER_CPU_ALIGNED(struct stack_canary, stack_canary); #endif /* Make sure %fs and %gs are initialized properly in idle threads */ diff --git a/arch/x86/kernel/cpu/cpu_debug.c b/arch/x86/kernel/cpu/cpu_debug.c index 6b2a52dd040..dca325c0399 100644 --- a/arch/x86/kernel/cpu/cpu_debug.c +++ b/arch/x86/kernel/cpu/cpu_debug.c @@ -30,8 +30,8 @@ #include <asm/apic.h> #include <asm/desc.h> -static DEFINE_PER_CPU(struct cpu_cpuX_base, cpu_arr[CPU_REG_ALL_BIT]); -static DEFINE_PER_CPU(struct cpu_private *, priv_arr[MAX_CPU_FILES]); +static DEFINE_PER_CPU(struct cpu_cpuX_base [CPU_REG_ALL_BIT], cpu_arr); +static DEFINE_PER_CPU(struct cpu_private * [MAX_CPU_FILES], priv_arr); static DEFINE_PER_CPU(int, cpu_priv_count); static DEFINE_MUTEX(cpu_debug_lock); diff --git a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c index ae9b503220c..7bb676c533a 100644 --- a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c +++ b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c @@ -60,7 +60,6 @@ enum { }; #define INTEL_MSR_RANGE (0xffff) -#define CPUID_6_ECX_APERFMPERF_CAPABILITY (0x1) struct acpi_cpufreq_data { struct acpi_processor_performance *acpi_data; @@ -71,11 +70,7 @@ struct acpi_cpufreq_data { static DEFINE_PER_CPU(struct acpi_cpufreq_data *, drv_data); -struct acpi_msr_data { - u64 saved_aperf, saved_mperf; -}; - -static DEFINE_PER_CPU(struct acpi_msr_data, msr_data); +static DEFINE_PER_CPU(struct aperfmperf, old_perf); DEFINE_TRACE(power_mark); @@ -244,23 +239,12 @@ static u32 get_cur_val(const struct cpumask *mask) return cmd.val; } -struct perf_pair { - union { - struct { - u32 lo; - u32 hi; - } split; - u64 whole; - } aperf, mperf; -}; - /* Called via smp_call_function_single(), on the target CPU */ static void read_measured_perf_ctrs(void *_cur) { - struct perf_pair *cur = _cur; + struct aperfmperf *am = _cur; - rdmsr(MSR_IA32_APERF, cur->aperf.split.lo, cur->aperf.split.hi); - rdmsr(MSR_IA32_MPERF, cur->mperf.split.lo, cur->mperf.split.hi); + get_aperfmperf(am); } /* @@ -279,63 +263,17 @@ static void read_measured_perf_ctrs(void *_cur) static unsigned int get_measured_perf(struct cpufreq_policy *policy, unsigned int cpu) { - struct perf_pair readin, cur; - unsigned int perf_percent; + struct aperfmperf perf; + unsigned long ratio; unsigned int retval; - if (smp_call_function_single(cpu, read_measured_perf_ctrs, &readin, 1)) + if (smp_call_function_single(cpu, read_measured_perf_ctrs, &perf, 1)) return 0; - cur.aperf.whole = readin.aperf.whole - - per_cpu(msr_data, cpu).saved_aperf; - cur.mperf.whole = readin.mperf.whole - - per_cpu(msr_data, cpu).saved_mperf; - per_cpu(msr_data, cpu).saved_aperf = readin.aperf.whole; - per_cpu(msr_data, cpu).saved_mperf = readin.mperf.whole; - -#ifdef __i386__ - /* - * We dont want to do 64 bit divide with 32 bit kernel - * Get an approximate value. Return failure in case we cannot get - * an approximate value. - */ - if (unlikely(cur.aperf.split.hi || cur.mperf.split.hi)) { - int shift_count; - u32 h; - - h = max_t(u32, cur.aperf.split.hi, cur.mperf.split.hi); - shift_count = fls(h); - - cur.aperf.whole >>= shift_count; - cur.mperf.whole >>= shift_count; - } - - if (((unsigned long)(-1) / 100) < cur.aperf.split.lo) { - int shift_count = 7; - cur.aperf.split.lo >>= shift_count; - cur.mperf.split.lo >>= shift_count; - } - - if (cur.aperf.split.lo && cur.mperf.split.lo) - perf_percent = (cur.aperf.split.lo * 100) / cur.mperf.split.lo; - else - perf_percent = 0; + ratio = calc_aperfmperf_ratio(&per_cpu(old_perf, cpu), &perf); + per_cpu(old_perf, cpu) = perf; -#else - if (unlikely(((unsigned long)(-1) / 100) < cur.aperf.whole)) { - int shift_count = 7; - cur.aperf.whole >>= shift_count; - cur.mperf.whole >>= shift_count; - } - - if (cur.aperf.whole && cur.mperf.whole) - perf_percent = (cur.aperf.whole * 100) / cur.mperf.whole; - else - perf_percent = 0; - -#endif - - retval = (policy->cpuinfo.max_freq * perf_percent) / 100; + retval = (policy->cpuinfo.max_freq * ratio) >> APERFMPERF_SHIFT; return retval; } @@ -588,6 +526,21 @@ static const struct dmi_system_id sw_any_bug_dmi_table[] = { }, { } }; + +static int acpi_cpufreq_blacklist(struct cpuinfo_x86 *c) +{ + /* http://www.intel.com/Assets/PDF/specupdate/314554.pdf + * AL30: A Machine Check Exception (MCE) Occurring during an + * Enhanced Intel SpeedStep Technology Ratio Change May Cause + * Both Processor Cores to Lock Up when HT is enabled*/ + if (c->x86_vendor == X86_VENDOR_INTEL) { + if ((c->x86 == 15) && + (c->x86_model == 6) && + (c->x86_mask == 8) && smt_capable()) + return -ENODEV; + } + return 0; +} #endif static int acpi_cpufreq_cpu_init(struct cpufreq_policy *policy) @@ -602,6 +555,12 @@ static int acpi_cpufreq_cpu_init(struct cpufreq_policy *policy) dprintk("acpi_cpufreq_cpu_init\n"); +#ifdef CONFIG_SMP + result = acpi_cpufreq_blacklist(c); + if (result) + return result; +#endif + data = kzalloc(sizeof(struct acpi_cpufreq_data), GFP_KERNEL); if (!data) return -ENOMEM; @@ -731,12 +690,8 @@ static int acpi_cpufreq_cpu_init(struct cpufreq_policy *policy) acpi_processor_notify_smm(THIS_MODULE); /* Check for APERF/MPERF support in hardware */ - if (c->x86_vendor == X86_VENDOR_INTEL && c->cpuid_level >= 6) { - unsigned int ecx; - ecx = cpuid_ecx(6); - if (ecx & CPUID_6_ECX_APERFMPERF_CAPABILITY) - acpi_cpufreq_driver.getavg = get_measured_perf; - } + if (cpu_has(c, X86_FEATURE_APERFMPERF)) + acpi_cpufreq_driver.getavg = get_measured_perf; dprintk("CPU%u - ACPI performance management activated.\n", cpu); for (i = 0; i < perf->state_count; i++) diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c index 2a50ef89100..6394aa5c798 100644 --- a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c +++ b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c @@ -605,9 +605,10 @@ static int check_pst_table(struct powernow_k8_data *data, struct pst_s *pst, return 0; } -static void invalidate_entry(struct powernow_k8_data *data, unsigned int entry) +static void invalidate_entry(struct cpufreq_frequency_table *powernow_table, + unsigned int entry) { - data->powernow_table[entry].frequency = CPUFREQ_ENTRY_INVALID; + powernow_table[entry].frequency = CPUFREQ_ENTRY_INVALID; } static void print_basics(struct powernow_k8_data *data) @@ -854,6 +855,10 @@ static int powernow_k8_cpu_init_acpi(struct powernow_k8_data *data) goto err_out; } + /* fill in data */ + data->numps = data->acpi_data.state_count; + powernow_k8_acpi_pst_values(data, 0); + if (cpu_family == CPU_HW_PSTATE) ret_val = fill_powernow_table_pstate(data, powernow_table); else @@ -866,11 +871,8 @@ static int powernow_k8_cpu_init_acpi(struct powernow_k8_data *data) powernow_table[data->acpi_data.state_count].index = 0; data->powernow_table = powernow_table; - /* fill in data */ - data->numps = data->acpi_data.state_count; if (cpumask_first(cpu_core_mask(data->cpu)) == data->cpu) print_basics(data); - powernow_k8_acpi_pst_values(data, 0); /* notify BIOS that we exist */ acpi_processor_notify_smm(THIS_MODULE); @@ -914,13 +916,13 @@ static int fill_powernow_table_pstate(struct powernow_k8_data *data, "bad value %d.\n", i, index); printk(KERN_ERR PFX "Please report to BIOS " "manufacturer\n"); - invalidate_entry(data, i); + invalidate_entry(powernow_table, i); continue; } rdmsr(MSR_PSTATE_DEF_BASE + index, lo, hi); if (!(hi & HW_PSTATE_VALID_MASK)) { dprintk("invalid pstate %d, ignoring\n", index); - invalidate_entry(data, i); + invalidate_entry(powernow_table, i); continue; } @@ -941,7 +943,6 @@ static int fill_powernow_table_fidvid(struct powernow_k8_data *data, struct cpufreq_frequency_table *powernow_table) { int i; - int cntlofreq = 0; for (i = 0; i < data->acpi_data.state_count; i++) { u32 fid; @@ -970,7 +971,7 @@ static int fill_powernow_table_fidvid(struct powernow_k8_data *data, /* verify frequency is OK */ if ((freq > (MAX_FREQ * 1000)) || (freq < (MIN_FREQ * 1000))) { dprintk("invalid freq %u kHz, ignoring\n", freq); - invalidate_entry(data, i); + invalidate_entry(powernow_table, i); continue; } @@ -978,38 +979,17 @@ static int fill_powernow_table_fidvid(struct powernow_k8_data *data, * BIOSs are using "off" to indicate invalid */ if (vid == VID_OFF) { dprintk("invalid vid %u, ignoring\n", vid); - invalidate_entry(data, i); + invalidate_entry(powernow_table, i); continue; } - /* verify only 1 entry from the lo frequency table */ - if (fid < HI_FID_TABLE_BOTTOM) { - if (cntlofreq) { - /* if both entries are the same, - * ignore this one ... */ - if ((freq != powernow_table[cntlofreq].frequency) || - (index != powernow_table[cntlofreq].index)) { - printk(KERN_ERR PFX - "Too many lo freq table " - "entries\n"); - return 1; - } - - dprintk("double low frequency table entry, " - "ignoring it.\n"); - invalidate_entry(data, i); - continue; - } else - cntlofreq = i; - } - if (freq != (data->acpi_data.states[i].core_frequency * 1000)) { printk(KERN_INFO PFX "invalid freq entries " "%u kHz vs. %u kHz\n", freq, (unsigned int) (data->acpi_data.states[i].core_frequency * 1000)); - invalidate_entry(data, i); + invalidate_entry(powernow_table, i); continue; } } diff --git a/arch/x86/kernel/cpu/intel.c b/arch/x86/kernel/cpu/intel.c index 80a722a071b..40e1835b35e 100644 --- a/arch/x86/kernel/cpu/intel.c +++ b/arch/x86/kernel/cpu/intel.c @@ -350,6 +350,12 @@ static void __cpuinit init_intel(struct cpuinfo_x86 *c) set_cpu_cap(c, X86_FEATURE_ARCH_PERFMON); } + if (c->cpuid_level > 6) { + unsigned ecx = cpuid_ecx(6); + if (ecx & 0x01) + set_cpu_cap(c, X86_FEATURE_APERFMPERF); + } + if (cpu_has_xmm2) set_cpu_cap(c, X86_FEATURE_LFENCE_RDTSC); if (cpu_has_ds) { diff --git a/arch/x86/kernel/cpu/intel_cacheinfo.c b/arch/x86/kernel/cpu/intel_cacheinfo.c index 306bf0dca06..804c40e2bc3 100644 --- a/arch/x86/kernel/cpu/intel_cacheinfo.c +++ b/arch/x86/kernel/cpu/intel_cacheinfo.c @@ -241,7 +241,7 @@ amd_cpuid4(int leaf, union _cpuid4_leaf_eax *eax, case 0: if (!l1->val) return; - assoc = l1->assoc; + assoc = assocs[l1->assoc]; line_size = l1->line_size; lines_per_tag = l1->lines_per_tag; size_in_kb = l1->size_in_kb; @@ -249,7 +249,7 @@ amd_cpuid4(int leaf, union _cpuid4_leaf_eax *eax, case 2: if (!l2.val) return; - assoc = l2.assoc; + assoc = assocs[l2.assoc]; line_size = l2.line_size; lines_per_tag = l2.lines_per_tag; /* cpu_data has errata corrections for K7 applied */ @@ -258,10 +258,14 @@ amd_cpuid4(int leaf, union _cpuid4_leaf_eax *eax, case 3: if (!l3.val) return; - assoc = l3.assoc; + assoc = assocs[l3.assoc]; line_size = l3.line_size; lines_per_tag = l3.lines_per_tag; size_in_kb = l3.size_encoded * 512; + if (boot_cpu_has(X86_FEATURE_AMD_DCM)) { + size_in_kb = size_in_kb >> 1; + assoc = assoc >> 1; + } break; default: return; @@ -270,18 +274,14 @@ amd_cpuid4(int leaf, union _cpuid4_leaf_eax *eax, eax->split.is_self_initializing = 1; eax->split.type = types[leaf]; eax->split.level = levels[leaf]; - if (leaf == 3) - eax->split.num_threads_sharing = - current_cpu_data.x86_max_cores - 1; - else - eax->split.num_threads_sharing = 0; + eax->split.num_threads_sharing = 0; eax->split.num_cores_on_die = current_cpu_data.x86_max_cores - 1; - if (assoc == 0xf) + if (assoc == 0xffff) eax->split.is_fully_associative = 1; ebx->split.coherency_line_size = line_size - 1; - ebx->split.ways_of_associativity = assocs[assoc] - 1; + ebx->split.ways_of_associativity = assoc - 1; ebx->split.physical_line_partition = lines_per_tag - 1; ecx->split.number_of_sets = (size_in_kb * 1024) / line_size / (ebx->split.ways_of_associativity + 1) - 1; @@ -523,6 +523,18 @@ static void __cpuinit cache_shared_cpu_map_setup(unsigned int cpu, int index) int index_msb, i; struct cpuinfo_x86 *c = &cpu_data(cpu); + if ((index == 3) && (c->x86_vendor == X86_VENDOR_AMD)) { + struct cpuinfo_x86 *d; + for_each_online_cpu(i) { + if (!per_cpu(cpuid4_info, i)) + continue; + d = &cpu_data(i); + this_leaf = CPUID4_INFO_IDX(i, index); + cpumask_copy(to_cpumask(this_leaf->shared_cpu_map), + d->llc_shared_map); + } + return; + } this_leaf = CPUID4_INFO_IDX(cpu, index); num_threads_sharing = 1 + this_leaf->eax.split.num_threads_sharing; diff --git a/arch/x86/kernel/cpu/mcheck/Makefile b/arch/x86/kernel/cpu/mcheck/Makefile index 188a1ca5ad2..4ac6d48fe11 100644 --- a/arch/x86/kernel/cpu/mcheck/Makefile +++ b/arch/x86/kernel/cpu/mcheck/Makefile @@ -1,11 +1,8 @@ -obj-y = mce.o +obj-y = mce.o mce-severity.o -obj-$(CONFIG_X86_NEW_MCE) += mce-severity.o -obj-$(CONFIG_X86_OLD_MCE) += k7.o p4.o p6.o obj-$(CONFIG_X86_ANCIENT_MCE) += winchip.o p5.o obj-$(CONFIG_X86_MCE_INTEL) += mce_intel.o obj-$(CONFIG_X86_MCE_AMD) += mce_amd.o -obj-$(CONFIG_X86_MCE_NONFATAL) += non-fatal.o obj-$(CONFIG_X86_MCE_THRESHOLD) += threshold.o obj-$(CONFIG_X86_MCE_INJECT) += mce-inject.o diff --git a/arch/x86/kernel/cpu/mcheck/k7.c b/arch/x86/kernel/cpu/mcheck/k7.c deleted file mode 100644 index b945d5dbc60..00000000000 --- a/arch/x86/kernel/cpu/mcheck/k7.c +++ /dev/null @@ -1,116 +0,0 @@ -/* - * Athlon specific Machine Check Exception Reporting - * (C) Copyright 2002 Dave Jones <davej@redhat.com> - */ -#include <linux/interrupt.h> -#include <linux/kernel.h> -#include <linux/types.h> -#include <linux/init.h> -#include <linux/smp.h> - -#include <asm/processor.h> -#include <asm/system.h> -#include <asm/mce.h> -#include <asm/msr.h> - -/* Machine Check Handler For AMD Athlon/Duron: */ -static void k7_machine_check(struct pt_regs *regs, long error_code) -{ - u32 alow, ahigh, high, low; - u32 mcgstl, mcgsth; - int recover = 1; - int i; - - rdmsr(MSR_IA32_MCG_STATUS, mcgstl, mcgsth); - if (mcgstl & (1<<0)) /* Recoverable ? */ - recover = 0; - - printk(KERN_EMERG "CPU %d: Machine Check Exception: %08x%08x\n", - smp_processor_id(), mcgsth, mcgstl); - - for (i = 1; i < nr_mce_banks; i++) { - rdmsr(MSR_IA32_MC0_STATUS+i*4, low, high); - if (high & (1<<31)) { - char misc[20]; - char addr[24]; - - misc[0] = '\0'; - addr[0] = '\0'; - - if (high & (1<<29)) - recover |= 1; - if (high & (1<<25)) - recover |= 2; - high &= ~(1<<31); - - if (high & (1<<27)) { - rdmsr(MSR_IA32_MC0_MISC+i*4, alow, ahigh); - snprintf(misc, 20, "[%08x%08x]", ahigh, alow); - } - if (high & (1<<26)) { - rdmsr(MSR_IA32_MC0_ADDR+i*4, alow, ahigh); - snprintf(addr, 24, " at %08x%08x", ahigh, alow); - } - - printk(KERN_EMERG "CPU %d: Bank %d: %08x%08x%s%s\n", - smp_processor_id(), i, high, low, misc, addr); - - /* Clear it: */ - wrmsr(MSR_IA32_MC0_STATUS+i*4, 0UL, 0UL); - /* Serialize: */ - wmb(); - add_taint(TAINT_MACHINE_CHECK); - } - } - - if (recover & 2) - panic("CPU context corrupt"); - if (recover & 1) - panic("Unable to continue"); - - printk(KERN_EMERG "Attempting to continue.\n"); - - mcgstl &= ~(1<<2); - wrmsr(MSR_IA32_MCG_STATUS, mcgstl, mcgsth); -} - - -/* AMD K7 machine check is Intel like: */ -void amd_mcheck_init(struct cpuinfo_x86 *c) -{ - u32 l, h; - int i; - - if (!cpu_has(c, X86_FEATURE_MCE)) - return; - - machine_check_vector = k7_machine_check; - /* Make sure the vector pointer is visible before we enable MCEs: */ - wmb(); - - printk(KERN_INFO "Intel machine check architecture supported.\n"); - - rdmsr(MSR_IA32_MCG_CAP, l, h); - if (l & (1<<8)) /* Control register present ? */ - wrmsr(MSR_IA32_MCG_CTL, 0xffffffff, 0xffffffff); - nr_mce_banks = l & 0xff; - - /* - * Clear status for MC index 0 separately, we don't touch CTL, - * as some K7 Athlons cause spurious MCEs when its enabled: - */ - if (boot_cpu_data.x86 == 6) { - wrmsr(MSR_IA32_MC0_STATUS, 0x0, 0x0); - i = 1; - } else - i = 0; - - for (; i < nr_mce_banks; i++) { - wrmsr(MSR_IA32_MC0_CTL+4*i, 0xffffffff, 0xffffffff); - wrmsr(MSR_IA32_MC0_STATUS+4*i, 0x0, 0x0); - } - - set_in_cr4(X86_CR4_MCE); - printk(KERN_INFO "Intel machine check reporting enabled on CPU#%d.\n", - smp_processor_id()); -} diff --git a/arch/x86/kernel/cpu/mcheck/mce-inject.c b/arch/x86/kernel/cpu/mcheck/mce-inject.c index a3a235a53f0..7029f0e2aca 100644 --- a/arch/x86/kernel/cpu/mcheck/mce-inject.c +++ b/arch/x86/kernel/cpu/mcheck/mce-inject.c @@ -18,7 +18,12 @@ #include <linux/string.h> #include <linux/fs.h> #include <linux/smp.h> +#include <linux/notifier.h> +#include <linux/kdebug.h> +#include <linux/cpu.h> +#include <linux/sched.h> #include <asm/mce.h> +#include <asm/apic.h> /* Update fake mce registers on current CPU. */ static void inject_mce(struct mce *m) @@ -39,44 +44,141 @@ static void inject_mce(struct mce *m) i->finished = 1; } -struct delayed_mce { - struct timer_list timer; - struct mce m; -}; +static void raise_poll(struct mce *m) +{ + unsigned long flags; + mce_banks_t b; -/* Inject mce on current CPU */ -static void raise_mce(unsigned long data) + memset(&b, 0xff, sizeof(mce_banks_t)); + local_irq_save(flags); + machine_check_poll(0, &b); + local_irq_restore(flags); + m->finished = 0; +} + +static void raise_exception(struct mce *m, struct pt_regs *pregs) { - struct delayed_mce *dm = (struct delayed_mce *)data; - struct mce *m = &dm->m; - int cpu = m->extcpu; + struct pt_regs regs; + unsigned long flags; - inject_mce(m); - if (m->status & MCI_STATUS_UC) { - struct pt_regs regs; + if (!pregs) { memset(®s, 0, sizeof(struct pt_regs)); regs.ip = m->ip; regs.cs = m->cs; + pregs = ®s; + } + /* in mcheck exeception handler, irq will be disabled */ + local_irq_save(flags); + do_machine_check(pregs, 0); + local_irq_restore(flags); + m->finished = 0; +} + +static cpumask_t mce_inject_cpumask; + +static int mce_raise_notify(struct notifier_block *self, + unsigned long val, void *data) +{ + struct die_args *args = (struct die_args *)data; + int cpu = smp_processor_id(); + struct mce *m = &__get_cpu_var(injectm); + if (val != DIE_NMI_IPI || !cpu_isset(cpu, mce_inject_cpumask)) + return NOTIFY_DONE; + cpu_clear(cpu, mce_inject_cpumask); + if (m->inject_flags & MCJ_EXCEPTION) + raise_exception(m, args->regs); + else if (m->status) + raise_poll(m); + return NOTIFY_STOP; +} + +static struct notifier_block mce_raise_nb = { + .notifier_call = mce_raise_notify, + .priority = 1000, +}; + +/* Inject mce on current CPU */ +static int raise_local(struct mce *m) +{ + int context = MCJ_CTX(m->inject_flags); + int ret = 0; + int cpu = m->extcpu; + + if (m->inject_flags & MCJ_EXCEPTION) { printk(KERN_INFO "Triggering MCE exception on CPU %d\n", cpu); - do_machine_check(®s, 0); + switch (context) { + case MCJ_CTX_IRQ: + /* + * Could do more to fake interrupts like + * calling irq_enter, but the necessary + * machinery isn't exported currently. + */ + /*FALL THROUGH*/ + case MCJ_CTX_PROCESS: + raise_exception(m, NULL); + break; + default: + printk(KERN_INFO "Invalid MCE context\n"); + ret = -EINVAL; + } printk(KERN_INFO "MCE exception done on CPU %d\n", cpu); - } else { - mce_banks_t b; - memset(&b, 0xff, sizeof(mce_banks_t)); + } else if (m->status) { printk(KERN_INFO "Starting machine check poll CPU %d\n", cpu); - machine_check_poll(0, &b); + raise_poll(m); mce_notify_irq(); - printk(KERN_INFO "Finished machine check poll on CPU %d\n", - cpu); - } - kfree(dm); + printk(KERN_INFO "Machine check poll done on CPU %d\n", cpu); + } else + m->finished = 0; + + return ret; +} + +static void raise_mce(struct mce *m) +{ + int context = MCJ_CTX(m->inject_flags); + + inject_mce(m); + + if (context == MCJ_CTX_RANDOM) + return; + +#ifdef CONFIG_X86_LOCAL_APIC + if (m->inject_flags & MCJ_NMI_BROADCAST) { + unsigned long start; + int cpu; + get_online_cpus(); + mce_inject_cpumask = cpu_online_map; + cpu_clear(get_cpu(), mce_inject_cpumask); + for_each_online_cpu(cpu) { + struct mce *mcpu = &per_cpu(injectm, cpu); + if (!mcpu->finished || + MCJ_CTX(mcpu->inject_flags) != MCJ_CTX_RANDOM) + cpu_clear(cpu, mce_inject_cpumask); + } + if (!cpus_empty(mce_inject_cpumask)) + apic->send_IPI_mask(&mce_inject_cpumask, NMI_VECTOR); + start = jiffies; + while (!cpus_empty(mce_inject_cpumask)) { + if (!time_before(jiffies, start + 2*HZ)) { + printk(KERN_ERR + "Timeout waiting for mce inject NMI %lx\n", + *cpus_addr(mce_inject_cpumask)); + break; + } + cpu_relax(); + } + raise_local(m); + put_cpu(); + put_online_cpus(); + } else +#endif + raise_local(m); } /* Error injection interface */ static ssize_t mce_write(struct file *filp, const char __user *ubuf, size_t usize, loff_t *off) { - struct delayed_mce *dm; struct mce m; if (!capable(CAP_SYS_ADMIN)) @@ -96,19 +198,12 @@ static ssize_t mce_write(struct file *filp, const char __user *ubuf, if (m.extcpu >= num_possible_cpus() || !cpu_online(m.extcpu)) return -EINVAL; - dm = kmalloc(sizeof(struct delayed_mce), GFP_KERNEL); - if (!dm) - return -ENOMEM; - /* * Need to give user space some time to set everything up, * so do it a jiffie or two later everywhere. - * Should we use a hrtimer here for better synchronization? */ - memcpy(&dm->m, &m, sizeof(struct mce)); - setup_timer(&dm->timer, raise_mce, (unsigned long)dm); - dm->timer.expires = jiffies + 2; - add_timer_on(&dm->timer, m.extcpu); + schedule_timeout(2); + raise_mce(&m); return usize; } @@ -116,6 +211,7 @@ static int inject_init(void) { printk(KERN_INFO "Machine check injector initialized\n"); mce_chrdev_ops.write = mce_write; + register_die_notifier(&mce_raise_nb); return 0; } diff --git a/arch/x86/kernel/cpu/mcheck/mce-internal.h b/arch/x86/kernel/cpu/mcheck/mce-internal.h index 54dcb8ff12e..32996f9fab6 100644 --- a/arch/x86/kernel/cpu/mcheck/mce-internal.h +++ b/arch/x86/kernel/cpu/mcheck/mce-internal.h @@ -1,3 +1,4 @@ +#include <linux/sysdev.h> #include <asm/mce.h> enum severity_level { @@ -10,6 +11,20 @@ enum severity_level { MCE_PANIC_SEVERITY, }; +#define ATTR_LEN 16 + +/* One object for each MCE bank, shared by all CPUs */ +struct mce_bank { + u64 ctl; /* subevents to enable */ + unsigned char init; /* initialise bank? */ + struct sysdev_attribute attr; /* sysdev attribute */ + char attrname[ATTR_LEN]; /* attribute name */ +}; + int mce_severity(struct mce *a, int tolerant, char **msg); +struct dentry *mce_get_debugfs_dir(void); extern int mce_ser; + +extern struct mce_bank *mce_banks; + diff --git a/arch/x86/kernel/cpu/mcheck/mce-severity.c b/arch/x86/kernel/cpu/mcheck/mce-severity.c index ff0807f9705..8a85dd1b1aa 100644 --- a/arch/x86/kernel/cpu/mcheck/mce-severity.c +++ b/arch/x86/kernel/cpu/mcheck/mce-severity.c @@ -139,6 +139,7 @@ int mce_severity(struct mce *a, int tolerant, char **msg) } } +#ifdef CONFIG_DEBUG_FS static void *s_start(struct seq_file *f, loff_t *pos) { if (*pos >= ARRAY_SIZE(severities)) @@ -197,7 +198,7 @@ static int __init severities_debugfs_init(void) { struct dentry *dmce = NULL, *fseverities_coverage = NULL; - dmce = debugfs_create_dir("mce", NULL); + dmce = mce_get_debugfs_dir(); if (dmce == NULL) goto err_out; fseverities_coverage = debugfs_create_file("severities-coverage", @@ -209,10 +210,7 @@ static int __init severities_debugfs_init(void) return 0; err_out: - if (fseverities_coverage) - debugfs_remove(fseverities_coverage); - if (dmce) - debugfs_remove(dmce); return -ENOMEM; } late_initcall(severities_debugfs_init); +#endif diff --git a/arch/x86/kernel/cpu/mcheck/mce.c b/arch/x86/kernel/cpu/mcheck/mce.c index 01213048f62..2f5aab26320 100644 --- a/arch/x86/kernel/cpu/mcheck/mce.c +++ b/arch/x86/kernel/cpu/mcheck/mce.c @@ -34,6 +34,7 @@ #include <linux/smp.h> #include <linux/fs.h> #include <linux/mm.h> +#include <linux/debugfs.h> #include <asm/processor.h> #include <asm/hw_irq.h> @@ -45,21 +46,8 @@ #include "mce-internal.h" -/* Handle unconfigured int18 (should never happen) */ -static void unexpected_machine_check(struct pt_regs *regs, long error_code) -{ - printk(KERN_ERR "CPU#%d: Unexpected int18 (Machine Check).\n", - smp_processor_id()); -} - -/* Call the installed machine check handler for this CPU setup. */ -void (*machine_check_vector)(struct pt_regs *, long error_code) = - unexpected_machine_check; - int mce_disabled __read_mostly; -#ifdef CONFIG_X86_NEW_MCE - #define MISC_MCELOG_MINOR 227 #define SPINUNIT 100 /* 100ns */ @@ -77,7 +65,6 @@ DEFINE_PER_CPU(unsigned, mce_exception_count); */ static int tolerant __read_mostly = 1; static int banks __read_mostly; -static u64 *bank __read_mostly; static int rip_msr __read_mostly; static int mce_bootlog __read_mostly = -1; static int monarch_timeout __read_mostly = -1; @@ -87,13 +74,13 @@ int mce_cmci_disabled __read_mostly; int mce_ignore_ce __read_mostly; int mce_ser __read_mostly; +struct mce_bank *mce_banks __read_mostly; + /* User mode helper program triggered by machine check event */ static unsigned long mce_need_notify; static char mce_helper[128]; static char *mce_helper_argv[2] = { mce_helper, NULL }; -static unsigned long dont_init_banks; - static DECLARE_WAIT_QUEUE_HEAD(mce_wait); static DEFINE_PER_CPU(struct mce, mces_seen); static int cpu_missing; @@ -104,11 +91,6 @@ DEFINE_PER_CPU(mce_banks_t, mce_poll_banks) = { [0 ... BITS_TO_LONGS(MAX_NR_BANKS)-1] = ~0UL }; -static inline int skip_bank_init(int i) -{ - return i < BITS_PER_LONG && test_bit(i, &dont_init_banks); -} - static DEFINE_PER_CPU(struct work_struct, mce_work); /* Do initial initialization of a struct mce */ @@ -183,6 +165,11 @@ 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 @@ -205,6 +192,8 @@ static void print_mce(struct mce *m) printk(KERN_EMERG "PROCESSOR %u:%x TIME %llu SOCKET %u APIC %x\n", m->cpuvendor, m->cpuid, m->time, m->socketid, m->apicid); + + decode_mce(m); } static void print_mce_head(void) @@ -215,13 +204,19 @@ static void print_mce_head(void) static void print_mce_tail(void) { printk(KERN_EMERG "This is not a software problem!\n" - "Run through mcelog --ascii to decode and contact your hardware vendor\n"); +#if (!defined(CONFIG_EDAC) || !defined(CONFIG_CPU_SUP_AMD)) + "Run through mcelog --ascii to decode and contact your hardware vendor\n" +#endif + ); } #define PANIC_TIMEOUT 5 /* 5 seconds */ static atomic_t mce_paniced; +static int fake_panic; +static atomic_t mce_fake_paniced; + /* Panic in progress. Enable interrupts and wait for final IPI */ static void wait_for_panic(void) { @@ -239,15 +234,21 @@ static void mce_panic(char *msg, struct mce *final, char *exp) { int i; - /* - * Make sure only one CPU runs in machine check panic - */ - if (atomic_add_return(1, &mce_paniced) > 1) - wait_for_panic(); - barrier(); + if (!fake_panic) { + /* + * Make sure only one CPU runs in machine check panic + */ + if (atomic_inc_return(&mce_paniced) > 1) + wait_for_panic(); + barrier(); - bust_spinlocks(1); - console_verbose(); + bust_spinlocks(1); + console_verbose(); + } else { + /* Don't log too much for fake panic */ + if (atomic_inc_return(&mce_fake_paniced) > 1) + return; + } print_mce_head(); /* First print corrected ones that are still unlogged */ for (i = 0; i < MCE_LOG_LEN; i++) { @@ -274,9 +275,12 @@ static void mce_panic(char *msg, struct mce *final, char *exp) print_mce_tail(); if (exp) printk(KERN_EMERG "Machine check: %s\n", exp); - if (panic_timeout == 0) - panic_timeout = mce_panic_timeout; - panic(msg); + if (!fake_panic) { + if (panic_timeout == 0) + panic_timeout = mce_panic_timeout; + panic(msg); + } else + printk(KERN_EMERG "Fake kernel panic: %s\n", msg); } /* Support code for software error injection */ @@ -286,11 +290,11 @@ 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_MC0_STATUS + bank*4) + if (msr == MSR_IA32_MCx_STATUS(bank)) return offsetof(struct mce, status); - if (msr == MSR_IA32_MC0_ADDR + bank*4) + if (msr == MSR_IA32_MCx_ADDR(bank)) return offsetof(struct mce, addr); - if (msr == MSR_IA32_MC0_MISC + bank*4) + if (msr == MSR_IA32_MCx_MISC(bank)) return offsetof(struct mce, misc); if (msr == MSR_IA32_MCG_STATUS) return offsetof(struct mce, mcgstatus); @@ -495,7 +499,7 @@ void machine_check_poll(enum mcp_flags flags, mce_banks_t *b) m.mcgstatus = mce_rdmsrl(MSR_IA32_MCG_STATUS); for (i = 0; i < banks; i++) { - if (!bank[i] || !test_bit(i, *b)) + if (!mce_banks[i].ctl || !test_bit(i, *b)) continue; m.misc = 0; @@ -504,7 +508,7 @@ void machine_check_poll(enum mcp_flags flags, mce_banks_t *b) m.tsc = 0; barrier(); - m.status = mce_rdmsrl(MSR_IA32_MC0_STATUS + i*4); + m.status = mce_rdmsrl(MSR_IA32_MCx_STATUS(i)); if (!(m.status & MCI_STATUS_VAL)) continue; @@ -519,9 +523,9 @@ void machine_check_poll(enum mcp_flags flags, mce_banks_t *b) continue; if (m.status & MCI_STATUS_MISCV) - m.misc = mce_rdmsrl(MSR_IA32_MC0_MISC + i*4); + m.misc = mce_rdmsrl(MSR_IA32_MCx_MISC(i)); if (m.status & MCI_STATUS_ADDRV) - m.addr = mce_rdmsrl(MSR_IA32_MC0_ADDR + i*4); + m.addr = mce_rdmsrl(MSR_IA32_MCx_ADDR(i)); if (!(flags & MCP_TIMESTAMP)) m.tsc = 0; @@ -537,7 +541,7 @@ void machine_check_poll(enum mcp_flags flags, mce_banks_t *b) /* * Clear state for this bank. */ - mce_wrmsrl(MSR_IA32_MC0_STATUS+4*i, 0); + mce_wrmsrl(MSR_IA32_MCx_STATUS(i), 0); } /* @@ -558,7 +562,7 @@ static int mce_no_way_out(struct mce *m, char **msg) int i; for (i = 0; i < banks; i++) { - m->status = mce_rdmsrl(MSR_IA32_MC0_STATUS + i*4); + m->status = mce_rdmsrl(MSR_IA32_MCx_STATUS(i)); if (mce_severity(m, tolerant, msg) >= MCE_PANIC_SEVERITY) return 1; } @@ -618,7 +622,7 @@ out: * This way we prevent any potential data corruption in a unrecoverable case * and also makes sure always all CPU's errors are examined. * - * Also this detects the case of an machine check event coming from outer + * Also this detects the case of a machine check event coming from outer * space (not detected by any CPUs) In this case some external agent wants * us to shut down, so panic too. * @@ -671,7 +675,7 @@ static void mce_reign(void) * No machine check event found. Must be some external * source or one CPU is hung. Panic. */ - if (!m && tolerant < 3) + if (global_worst <= MCE_KEEP_SEVERITY && tolerant < 3) mce_panic("Machine check from unknown source", NULL, NULL); /* @@ -705,7 +709,7 @@ static int mce_start(int *no_way_out) * global_nwo should be updated before mce_callin */ smp_wmb(); - order = atomic_add_return(1, &mce_callin); + order = atomic_inc_return(&mce_callin); /* * Wait for everyone. @@ -842,7 +846,7 @@ static void mce_clear_state(unsigned long *toclear) for (i = 0; i < banks; i++) { if (test_bit(i, toclear)) - mce_wrmsrl(MSR_IA32_MC0_STATUS+4*i, 0); + mce_wrmsrl(MSR_IA32_MCx_STATUS(i), 0); } } @@ -895,11 +899,11 @@ void do_machine_check(struct pt_regs *regs, long error_code) mce_setup(&m); m.mcgstatus = mce_rdmsrl(MSR_IA32_MCG_STATUS); - no_way_out = mce_no_way_out(&m, &msg); - final = &__get_cpu_var(mces_seen); *final = m; + no_way_out = mce_no_way_out(&m, &msg); + barrier(); /* @@ -916,14 +920,14 @@ void do_machine_check(struct pt_regs *regs, long error_code) order = mce_start(&no_way_out); for (i = 0; i < banks; i++) { __clear_bit(i, toclear); - if (!bank[i]) + if (!mce_banks[i].ctl) continue; m.misc = 0; m.addr = 0; m.bank = i; - m.status = mce_rdmsrl(MSR_IA32_MC0_STATUS + i*4); + m.status = mce_rdmsrl(MSR_IA32_MCx_STATUS(i)); if ((m.status & MCI_STATUS_VAL) == 0) continue; @@ -964,9 +968,9 @@ void do_machine_check(struct pt_regs *regs, long error_code) kill_it = 1; if (m.status & MCI_STATUS_MISCV) - m.misc = mce_rdmsrl(MSR_IA32_MC0_MISC + i*4); + m.misc = mce_rdmsrl(MSR_IA32_MCx_MISC(i)); if (m.status & MCI_STATUS_ADDRV) - m.addr = mce_rdmsrl(MSR_IA32_MC0_ADDR + i*4); + m.addr = mce_rdmsrl(MSR_IA32_MCx_ADDR(i)); /* * Action optional error. Queue address for later processing. @@ -1091,7 +1095,7 @@ void mce_log_therm_throt_event(__u64 status) */ static int check_interval = 5 * 60; /* 5 minutes */ -static DEFINE_PER_CPU(int, next_interval); /* in jiffies */ +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) @@ -1110,7 +1114,7 @@ static void mcheck_timer(unsigned long data) * Alert userspace if needed. If we logged an MCE, reduce the * polling interval, otherwise increase the polling interval. */ - n = &__get_cpu_var(next_interval); + n = &__get_cpu_var(mce_next_interval); if (mce_notify_irq()) *n = max(*n/2, HZ/100); else @@ -1159,10 +1163,25 @@ int mce_notify_irq(void) } EXPORT_SYMBOL_GPL(mce_notify_irq); +static int mce_banks_init(void) +{ + int i; + + mce_banks = kzalloc(banks * sizeof(struct mce_bank), GFP_KERNEL); + if (!mce_banks) + return -ENOMEM; + for (i = 0; i < banks; i++) { + struct mce_bank *b = &mce_banks[i]; + b->ctl = -1ULL; + b->init = 1; + } + return 0; +} + /* * Initialize Machine Checks for a CPU. */ -static int mce_cap_init(void) +static int __cpuinit mce_cap_init(void) { unsigned b; u64 cap; @@ -1182,11 +1201,10 @@ static int mce_cap_init(void) /* Don't support asymmetric configurations today */ WARN_ON(banks != 0 && b != banks); banks = b; - if (!bank) { - bank = kmalloc(banks * sizeof(u64), GFP_KERNEL); - if (!bank) - return -ENOMEM; - memset(bank, 0xff, banks * sizeof(u64)); + if (!mce_banks) { + int err = mce_banks_init(); + if (err) + return err; } /* Use accurate RIP reporting if available. */ @@ -1218,15 +1236,16 @@ static void mce_init(void) wrmsr(MSR_IA32_MCG_CTL, 0xffffffff, 0xffffffff); for (i = 0; i < banks; i++) { - if (skip_bank_init(i)) + struct mce_bank *b = &mce_banks[i]; + if (!b->init) continue; - wrmsrl(MSR_IA32_MC0_CTL+4*i, bank[i]); - wrmsrl(MSR_IA32_MC0_STATUS+4*i, 0); + wrmsrl(MSR_IA32_MCx_CTL(i), b->ctl); + wrmsrl(MSR_IA32_MCx_STATUS(i), 0); } } /* Add per CPU specific workarounds here */ -static int mce_cpu_quirks(struct cpuinfo_x86 *c) +static int __cpuinit mce_cpu_quirks(struct cpuinfo_x86 *c) { if (c->x86_vendor == X86_VENDOR_UNKNOWN) { pr_info("MCE: unknown CPU type - not enabling MCE support.\n"); @@ -1241,7 +1260,7 @@ static int mce_cpu_quirks(struct cpuinfo_x86 *c) * trips off incorrectly with the IOMMU & 3ware * & Cerberus: */ - clear_bit(10, (unsigned long *)&bank[4]); + clear_bit(10, (unsigned long *)&mce_banks[4].ctl); } if (c->x86 <= 17 && mce_bootlog < 0) { /* @@ -1255,7 +1274,7 @@ static int mce_cpu_quirks(struct cpuinfo_x86 *c) * by default. */ if (c->x86 == 6 && banks > 0) - bank[0] = 0; + mce_banks[0].ctl = 0; } if (c->x86_vendor == X86_VENDOR_INTEL) { @@ -1268,8 +1287,8 @@ static int mce_cpu_quirks(struct cpuinfo_x86 *c) * valid event later, merely don't write CTL0. */ - if (c->x86 == 6 && c->x86_model < 0x1A) - __set_bit(0, &dont_init_banks); + if (c->x86 == 6 && c->x86_model < 0x1A && banks > 0) + mce_banks[0].init = 0; /* * All newer Intel systems support MCE broadcasting. Enable @@ -1325,7 +1344,7 @@ static void mce_cpu_features(struct cpuinfo_x86 *c) static void mce_init_timer(void) { struct timer_list *t = &__get_cpu_var(mce_timer); - int *n = &__get_cpu_var(next_interval); + int *n = &__get_cpu_var(mce_next_interval); if (mce_ignore_ce) return; @@ -1338,6 +1357,17 @@ static void mce_init_timer(void) add_timer_on(t, smp_processor_id()); } +/* Handle unconfigured int18 (should never happen) */ +static void unexpected_machine_check(struct pt_regs *regs, long error_code) +{ + printk(KERN_ERR "CPU#%d: Unexpected int18 (Machine Check).\n", + smp_processor_id()); +} + +/* Call the installed machine check handler for this CPU setup. */ +void (*machine_check_vector)(struct pt_regs *, long error_code) = + unexpected_machine_check; + /* * Called for each booted CPU to set up machine checks. * Must be called with preempt off: @@ -1551,8 +1581,10 @@ static struct miscdevice mce_log_device = { */ static int __init mcheck_enable(char *str) { - if (*str == 0) + if (*str == 0) { enable_p5_mce(); + return 1; + } if (*str == '=') str++; if (!strcmp(str, "off")) @@ -1593,8 +1625,9 @@ static int mce_disable(void) int i; for (i = 0; i < banks; i++) { - if (!skip_bank_init(i)) - wrmsrl(MSR_IA32_MC0_CTL + i*4, 0); + struct mce_bank *b = &mce_banks[i]; + if (b->init) + wrmsrl(MSR_IA32_MCx_CTL(i), 0); } return 0; } @@ -1669,14 +1702,15 @@ DEFINE_PER_CPU(struct sys_device, mce_dev); __cpuinitdata void (*threshold_cpu_callback)(unsigned long action, unsigned int cpu); -static struct sysdev_attribute *bank_attrs; +static inline struct mce_bank *attr_to_bank(struct sysdev_attribute *attr) +{ + return container_of(attr, struct mce_bank, attr); +} static ssize_t show_bank(struct sys_device *s, struct sysdev_attribute *attr, char *buf) { - u64 b = bank[attr - bank_attrs]; - - return sprintf(buf, "%llx\n", b); + return sprintf(buf, "%llx\n", attr_to_bank(attr)->ctl); } static ssize_t set_bank(struct sys_device *s, struct sysdev_attribute *attr, @@ -1687,7 +1721,7 @@ static ssize_t set_bank(struct sys_device *s, struct sysdev_attribute *attr, if (strict_strtoull(buf, 0, &new) < 0) return -EINVAL; - bank[attr - bank_attrs] = new; + attr_to_bank(attr)->ctl = new; mce_restart(); return size; @@ -1829,7 +1863,7 @@ static __cpuinit int mce_create_device(unsigned int cpu) } for (j = 0; j < banks; j++) { err = sysdev_create_file(&per_cpu(mce_dev, cpu), - &bank_attrs[j]); + &mce_banks[j].attr); if (err) goto error2; } @@ -1838,10 +1872,10 @@ static __cpuinit int mce_create_device(unsigned int cpu) return 0; error2: while (--j >= 0) - sysdev_remove_file(&per_cpu(mce_dev, cpu), &bank_attrs[j]); + 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_attrs[i]); + sysdev_remove_file(&per_cpu(mce_dev, cpu), &mce_banks[i].attr); sysdev_unregister(&per_cpu(mce_dev, cpu)); @@ -1859,7 +1893,7 @@ static __cpuinit void mce_remove_device(unsigned int cpu) sysdev_remove_file(&per_cpu(mce_dev, cpu), mce_attrs[i]); for (i = 0; i < banks; i++) - sysdev_remove_file(&per_cpu(mce_dev, cpu), &bank_attrs[i]); + sysdev_remove_file(&per_cpu(mce_dev, cpu), &mce_banks[i].attr); sysdev_unregister(&per_cpu(mce_dev, cpu)); cpumask_clear_cpu(cpu, mce_dev_initialized); @@ -1876,8 +1910,9 @@ static void mce_disable_cpu(void *h) if (!(action & CPU_TASKS_FROZEN)) cmci_clear(); for (i = 0; i < banks; i++) { - if (!skip_bank_init(i)) - wrmsrl(MSR_IA32_MC0_CTL + i*4, 0); + struct mce_bank *b = &mce_banks[i]; + if (b->init) + wrmsrl(MSR_IA32_MCx_CTL(i), 0); } } @@ -1892,8 +1927,9 @@ static void mce_reenable_cpu(void *h) if (!(action & CPU_TASKS_FROZEN)) cmci_reenable(); for (i = 0; i < banks; i++) { - if (!skip_bank_init(i)) - wrmsrl(MSR_IA32_MC0_CTL + i*4, bank[i]); + struct mce_bank *b = &mce_banks[i]; + if (b->init) + wrmsrl(MSR_IA32_MCx_CTL(i), b->ctl); } } @@ -1925,7 +1961,7 @@ mce_cpu_callback(struct notifier_block *nfb, unsigned long action, void *hcpu) case CPU_DOWN_FAILED: case CPU_DOWN_FAILED_FROZEN: t->expires = round_jiffies(jiffies + - __get_cpu_var(next_interval)); + __get_cpu_var(mce_next_interval)); add_timer_on(t, cpu); smp_call_function_single(cpu, mce_reenable_cpu, &action, 1); break; @@ -1941,35 +1977,21 @@ static struct notifier_block mce_cpu_notifier __cpuinitdata = { .notifier_call = mce_cpu_callback, }; -static __init int mce_init_banks(void) +static __init void mce_init_banks(void) { int i; - bank_attrs = kzalloc(sizeof(struct sysdev_attribute) * banks, - GFP_KERNEL); - if (!bank_attrs) - return -ENOMEM; - for (i = 0; i < banks; i++) { - struct sysdev_attribute *a = &bank_attrs[i]; + struct mce_bank *b = &mce_banks[i]; + struct sysdev_attribute *a = &b->attr; - a->attr.name = kasprintf(GFP_KERNEL, "bank%d", i); - if (!a->attr.name) - goto nomem; + a->attr.name = b->attrname; + snprintf(b->attrname, ATTR_LEN, "bank%d", i); a->attr.mode = 0644; a->show = show_bank; a->store = set_bank; } - return 0; - -nomem: - while (--i >= 0) - kfree(bank_attrs[i].attr.name); - kfree(bank_attrs); - bank_attrs = NULL; - - return -ENOMEM; } static __init int mce_init_device(void) @@ -1982,9 +2004,7 @@ static __init int mce_init_device(void) zalloc_cpumask_var(&mce_dev_initialized, GFP_KERNEL); - err = mce_init_banks(); - if (err) - return err; + mce_init_banks(); err = sysdev_class_register(&mce_sysclass); if (err) @@ -2004,57 +2024,65 @@ static __init int mce_init_device(void) device_initcall(mce_init_device); -#else /* CONFIG_X86_OLD_MCE: */ - -int nr_mce_banks; -EXPORT_SYMBOL_GPL(nr_mce_banks); /* non-fatal.o */ +/* + * Old style boot options parsing. Only for compatibility. + */ +static int __init mcheck_disable(char *str) +{ + mce_disabled = 1; + return 1; +} +__setup("nomce", mcheck_disable); -/* This has to be run for each processor */ -void mcheck_init(struct cpuinfo_x86 *c) +#ifdef CONFIG_DEBUG_FS +struct dentry *mce_get_debugfs_dir(void) { - if (mce_disabled) - return; + static struct dentry *dmce; - switch (c->x86_vendor) { - case X86_VENDOR_AMD: - amd_mcheck_init(c); - break; + if (!dmce) + dmce = debugfs_create_dir("mce", NULL); - case X86_VENDOR_INTEL: - if (c->x86 == 5) - intel_p5_mcheck_init(c); - if (c->x86 == 6) - intel_p6_mcheck_init(c); - if (c->x86 == 15) - intel_p4_mcheck_init(c); - break; + return dmce; +} - case X86_VENDOR_CENTAUR: - if (c->x86 == 5) - winchip_mcheck_init(c); - break; +static void mce_reset(void) +{ + cpu_missing = 0; + atomic_set(&mce_fake_paniced, 0); + atomic_set(&mce_executing, 0); + atomic_set(&mce_callin, 0); + atomic_set(&global_nwo, 0); +} - default: - break; - } - printk(KERN_INFO "mce: CPU supports %d MCE banks\n", nr_mce_banks); +static int fake_panic_get(void *data, u64 *val) +{ + *val = fake_panic; + return 0; } -static int __init mcheck_enable(char *str) +static int fake_panic_set(void *data, u64 val) { - mce_p5_enabled = 1; - return 1; + mce_reset(); + fake_panic = val; + return 0; } -__setup("mce", mcheck_enable); -#endif /* CONFIG_X86_OLD_MCE */ +DEFINE_SIMPLE_ATTRIBUTE(fake_panic_fops, fake_panic_get, + fake_panic_set, "%llu\n"); -/* - * Old style boot options parsing. Only for compatibility. - */ -static int __init mcheck_disable(char *str) +static int __init mce_debugfs_init(void) { - mce_disabled = 1; - return 1; + struct dentry *dmce, *ffake_panic; + + dmce = mce_get_debugfs_dir(); + if (!dmce) + return -ENOMEM; + ffake_panic = debugfs_create_file("fake_panic", 0444, dmce, NULL, + &fake_panic_fops); + if (!ffake_panic) + return -ENOMEM; + + return 0; } -__setup("nomce", mcheck_disable); +late_initcall(mce_debugfs_init); +#endif diff --git a/arch/x86/kernel/cpu/mcheck/mce_amd.c b/arch/x86/kernel/cpu/mcheck/mce_amd.c index ddae21620bd..8cd5224943b 100644 --- a/arch/x86/kernel/cpu/mcheck/mce_amd.c +++ b/arch/x86/kernel/cpu/mcheck/mce_amd.c @@ -69,7 +69,7 @@ struct threshold_bank { struct threshold_block *blocks; cpumask_var_t cpus; }; -static DEFINE_PER_CPU(struct threshold_bank *, threshold_banks[NR_BANKS]); +static DEFINE_PER_CPU(struct threshold_bank * [NR_BANKS], threshold_banks); #ifdef CONFIG_SMP static unsigned char shared_bank[NR_BANKS] = { @@ -489,12 +489,14 @@ static __cpuinit int threshold_create_bank(unsigned int cpu, unsigned int bank) int i, err = 0; struct threshold_bank *b = NULL; char name[32]; + struct cpuinfo_x86 *c = &cpu_data(cpu); + sprintf(name, "threshold_bank%i", bank); #ifdef CONFIG_SMP if (cpu_data(cpu).cpu_core_id && shared_bank[bank]) { /* symlink */ - i = cpumask_first(cpu_core_mask(cpu)); + i = cpumask_first(c->llc_shared_map); /* first core not up yet */ if (cpu_data(i).cpu_core_id) @@ -514,7 +516,7 @@ static __cpuinit int threshold_create_bank(unsigned int cpu, unsigned int bank) if (err) goto out; - cpumask_copy(b->cpus, cpu_core_mask(cpu)); + cpumask_copy(b->cpus, c->llc_shared_map); per_cpu(threshold_banks, cpu)[bank] = b; goto out; @@ -539,7 +541,7 @@ static __cpuinit int threshold_create_bank(unsigned int cpu, unsigned int bank) #ifndef CONFIG_SMP cpumask_setall(b->cpus); #else - cpumask_copy(b->cpus, cpu_core_mask(cpu)); + cpumask_copy(b->cpus, c->llc_shared_map); #endif per_cpu(threshold_banks, cpu)[bank] = b; diff --git a/arch/x86/kernel/cpu/mcheck/mce_intel.c b/arch/x86/kernel/cpu/mcheck/mce_intel.c index e1acec0f7a3..889f665fe93 100644 --- a/arch/x86/kernel/cpu/mcheck/mce_intel.c +++ b/arch/x86/kernel/cpu/mcheck/mce_intel.c @@ -90,7 +90,7 @@ static void cmci_discover(int banks, int boot) if (test_bit(i, owned)) continue; - rdmsrl(MSR_IA32_MC0_CTL2 + i, val); + rdmsrl(MSR_IA32_MCx_CTL2(i), val); /* Already owned by someone else? */ if (val & CMCI_EN) { @@ -101,8 +101,8 @@ static void cmci_discover(int banks, int boot) } val |= CMCI_EN | CMCI_THRESHOLD; - wrmsrl(MSR_IA32_MC0_CTL2 + i, val); - rdmsrl(MSR_IA32_MC0_CTL2 + i, val); + wrmsrl(MSR_IA32_MCx_CTL2(i), val); + rdmsrl(MSR_IA32_MCx_CTL2(i), val); /* Did the enable bit stick? -- the bank supports CMCI */ if (val & CMCI_EN) { @@ -152,9 +152,9 @@ void cmci_clear(void) if (!test_bit(i, __get_cpu_var(mce_banks_owned))) continue; /* Disable CMCI */ - rdmsrl(MSR_IA32_MC0_CTL2 + i, val); + rdmsrl(MSR_IA32_MCx_CTL2(i), val); val &= ~(CMCI_EN|CMCI_THRESHOLD_MASK); - wrmsrl(MSR_IA32_MC0_CTL2 + i, val); + wrmsrl(MSR_IA32_MCx_CTL2(i), val); __clear_bit(i, __get_cpu_var(mce_banks_owned)); } spin_unlock_irqrestore(&cmci_discover_lock, flags); diff --git a/arch/x86/kernel/cpu/mcheck/non-fatal.c b/arch/x86/kernel/cpu/mcheck/non-fatal.c deleted file mode 100644 index f5f2d6f71fb..00000000000 --- a/arch/x86/kernel/cpu/mcheck/non-fatal.c +++ /dev/null @@ -1,94 +0,0 @@ -/* - * Non Fatal Machine Check Exception Reporting - * - * (C) Copyright 2002 Dave Jones. <davej@redhat.com> - * - * This file contains routines to check for non-fatal MCEs every 15s - * - */ -#include <linux/interrupt.h> -#include <linux/workqueue.h> -#include <linux/jiffies.h> -#include <linux/kernel.h> -#include <linux/module.h> -#include <linux/types.h> -#include <linux/init.h> -#include <linux/smp.h> - -#include <asm/processor.h> -#include <asm/system.h> -#include <asm/mce.h> -#include <asm/msr.h> - -static int firstbank; - -#define MCE_RATE (15*HZ) /* timer rate is 15s */ - -static void mce_checkregs(void *info) -{ - u32 low, high; - int i; - - for (i = firstbank; i < nr_mce_banks; i++) { - rdmsr(MSR_IA32_MC0_STATUS+i*4, low, high); - - if (!(high & (1<<31))) - continue; - - printk(KERN_INFO "MCE: The hardware reports a non fatal, " - "correctable incident occurred on CPU %d.\n", - smp_processor_id()); - - printk(KERN_INFO "Bank %d: %08x%08x\n", i, high, low); - - /* - * Scrub the error so we don't pick it up in MCE_RATE - * seconds time: - */ - wrmsr(MSR_IA32_MC0_STATUS+i*4, 0UL, 0UL); - - /* Serialize: */ - wmb(); - add_taint(TAINT_MACHINE_CHECK); - } -} - -static void mce_work_fn(struct work_struct *work); -static DECLARE_DELAYED_WORK(mce_work, mce_work_fn); - -static void mce_work_fn(struct work_struct *work) -{ - on_each_cpu(mce_checkregs, NULL, 1); - schedule_delayed_work(&mce_work, round_jiffies_relative(MCE_RATE)); -} - -static int __init init_nonfatal_mce_checker(void) -{ - struct cpuinfo_x86 *c = &boot_cpu_data; - - /* Check for MCE support */ - if (!cpu_has(c, X86_FEATURE_MCE)) - return -ENODEV; - - /* Check for PPro style MCA */ - if (!cpu_has(c, X86_FEATURE_MCA)) - return -ENODEV; - - /* Some Athlons misbehave when we frob bank 0 */ - if (boot_cpu_data.x86_vendor == X86_VENDOR_AMD && - boot_cpu_data.x86 == 6) - firstbank = 1; - else - firstbank = 0; - - /* - * Check for non-fatal errors every MCE_RATE s - */ - schedule_delayed_work(&mce_work, round_jiffies_relative(MCE_RATE)); - printk(KERN_INFO "Machine check exception polling timer started.\n"); - - return 0; -} -module_init(init_nonfatal_mce_checker); - -MODULE_LICENSE("GPL"); diff --git a/arch/x86/kernel/cpu/mcheck/p4.c b/arch/x86/kernel/cpu/mcheck/p4.c deleted file mode 100644 index 4482aea9aa2..00000000000 --- a/arch/x86/kernel/cpu/mcheck/p4.c +++ /dev/null @@ -1,163 +0,0 @@ -/* - * P4 specific Machine Check Exception Reporting - */ -#include <linux/kernel.h> -#include <linux/types.h> -#include <linux/init.h> -#include <linux/smp.h> - -#include <asm/processor.h> -#include <asm/mce.h> -#include <asm/msr.h> - -/* as supported by the P4/Xeon family */ -struct intel_mce_extended_msrs { - u32 eax; - u32 ebx; - u32 ecx; - u32 edx; - u32 esi; - u32 edi; - u32 ebp; - u32 esp; - u32 eflags; - u32 eip; - /* u32 *reserved[]; */ -}; - -static int mce_num_extended_msrs; - -/* P4/Xeon Extended MCE MSR retrieval, return 0 if unsupported */ -static void intel_get_extended_msrs(struct intel_mce_extended_msrs *r) -{ - u32 h; - - rdmsr(MSR_IA32_MCG_EAX, r->eax, h); - rdmsr(MSR_IA32_MCG_EBX, r->ebx, h); - rdmsr(MSR_IA32_MCG_ECX, r->ecx, h); - rdmsr(MSR_IA32_MCG_EDX, r->edx, h); - rdmsr(MSR_IA32_MCG_ESI, r->esi, h); - rdmsr(MSR_IA32_MCG_EDI, r->edi, h); - rdmsr(MSR_IA32_MCG_EBP, r->ebp, h); - rdmsr(MSR_IA32_MCG_ESP, r->esp, h); - rdmsr(MSR_IA32_MCG_EFLAGS, r->eflags, h); - rdmsr(MSR_IA32_MCG_EIP, r->eip, h); -} - -static void intel_machine_check(struct pt_regs *regs, long error_code) -{ - u32 alow, ahigh, high, low; - u32 mcgstl, mcgsth; - int recover = 1; - int i; - - rdmsr(MSR_IA32_MCG_STATUS, mcgstl, mcgsth); - if (mcgstl & (1<<0)) /* Recoverable ? */ - recover = 0; - - printk(KERN_EMERG "CPU %d: Machine Check Exception: %08x%08x\n", - smp_processor_id(), mcgsth, mcgstl); - - if (mce_num_extended_msrs > 0) { - struct intel_mce_extended_msrs dbg; - - intel_get_extended_msrs(&dbg); - - printk(KERN_DEBUG "CPU %d: EIP: %08x EFLAGS: %08x\n" - "\teax: %08x ebx: %08x ecx: %08x edx: %08x\n" - "\tesi: %08x edi: %08x ebp: %08x esp: %08x\n", - smp_processor_id(), dbg.eip, dbg.eflags, - dbg.eax, dbg.ebx, dbg.ecx, dbg.edx, - dbg.esi, dbg.edi, dbg.ebp, dbg.esp); - } - - for (i = 0; i < nr_mce_banks; i++) { - rdmsr(MSR_IA32_MC0_STATUS+i*4, low, high); - if (high & (1<<31)) { - char misc[20]; - char addr[24]; - - misc[0] = addr[0] = '\0'; - if (high & (1<<29)) - recover |= 1; - if (high & (1<<25)) - recover |= 2; - high &= ~(1<<31); - if (high & (1<<27)) { - rdmsr(MSR_IA32_MC0_MISC+i*4, alow, ahigh); - snprintf(misc, 20, "[%08x%08x]", ahigh, alow); - } - if (high & (1<<26)) { - rdmsr(MSR_IA32_MC0_ADDR+i*4, alow, ahigh); - snprintf(addr, 24, " at %08x%08x", ahigh, alow); - } - printk(KERN_EMERG "CPU %d: Bank %d: %08x%08x%s%s\n", - smp_processor_id(), i, high, low, misc, addr); - } - } - - if (recover & 2) - panic("CPU context corrupt"); - if (recover & 1) - panic("Unable to continue"); - - printk(KERN_EMERG "Attempting to continue.\n"); - - /* - * Do not clear the MSR_IA32_MCi_STATUS if the error is not - * recoverable/continuable.This will allow BIOS to look at the MSRs - * for errors if the OS could not log the error. - */ - for (i = 0; i < nr_mce_banks; i++) { - u32 msr; - msr = MSR_IA32_MC0_STATUS+i*4; - rdmsr(msr, low, high); - if (high&(1<<31)) { - /* Clear it */ - wrmsr(msr, 0UL, 0UL); - /* Serialize */ - wmb(); - add_taint(TAINT_MACHINE_CHECK); - } - } - mcgstl &= ~(1<<2); - wrmsr(MSR_IA32_MCG_STATUS, mcgstl, mcgsth); -} - -void intel_p4_mcheck_init(struct cpuinfo_x86 *c) -{ - u32 l, h; - int i; - - machine_check_vector = intel_machine_check; - wmb(); - - printk(KERN_INFO "Intel machine check architecture supported.\n"); - rdmsr(MSR_IA32_MCG_CAP, l, h); - if (l & (1<<8)) /* Control register present ? */ - wrmsr(MSR_IA32_MCG_CTL, 0xffffffff, 0xffffffff); - nr_mce_banks = l & 0xff; - - for (i = 0; i < nr_mce_banks; i++) { - wrmsr(MSR_IA32_MC0_CTL+4*i, 0xffffffff, 0xffffffff); - wrmsr(MSR_IA32_MC0_STATUS+4*i, 0x0, 0x0); - } - - set_in_cr4(X86_CR4_MCE); - printk(KERN_INFO "Intel machine check reporting enabled on CPU#%d.\n", - smp_processor_id()); - - /* Check for P4/Xeon extended MCE MSRs */ - rdmsr(MSR_IA32_MCG_CAP, l, h); - if (l & (1<<9)) {/* MCG_EXT_P */ - mce_num_extended_msrs = (l >> 16) & 0xff; - printk(KERN_INFO "CPU%d: Intel P4/Xeon Extended MCE MSRs (%d)" - " available\n", - smp_processor_id(), mce_num_extended_msrs); - -#ifdef CONFIG_X86_MCE_P4THERMAL - /* Check for P4/Xeon Thermal monitor */ - intel_init_thermal(c); -#endif - } -} diff --git a/arch/x86/kernel/cpu/mcheck/p6.c b/arch/x86/kernel/cpu/mcheck/p6.c deleted file mode 100644 index 01e4f817818..00000000000 --- a/arch/x86/kernel/cpu/mcheck/p6.c +++ /dev/null @@ -1,127 +0,0 @@ -/* - * P6 specific Machine Check Exception Reporting - * (C) Copyright 2002 Alan Cox <alan@lxorguk.ukuu.org.uk> - */ -#include <linux/interrupt.h> -#include <linux/kernel.h> -#include <linux/types.h> -#include <linux/init.h> -#include <linux/smp.h> - -#include <asm/processor.h> -#include <asm/system.h> -#include <asm/mce.h> -#include <asm/msr.h> - -/* Machine Check Handler For PII/PIII */ -static void intel_machine_check(struct pt_regs *regs, long error_code) -{ - u32 alow, ahigh, high, low; - u32 mcgstl, mcgsth; - int recover = 1; - int i; - - rdmsr(MSR_IA32_MCG_STATUS, mcgstl, mcgsth); - if (mcgstl & (1<<0)) /* Recoverable ? */ - recover = 0; - - printk(KERN_EMERG "CPU %d: Machine Check Exception: %08x%08x\n", - smp_processor_id(), mcgsth, mcgstl); - - for (i = 0; i < nr_mce_banks; i++) { - rdmsr(MSR_IA32_MC0_STATUS+i*4, low, high); - if (high & (1<<31)) { - char misc[20]; - char addr[24]; - - misc[0] = '\0'; - addr[0] = '\0'; - - if (high & (1<<29)) - recover |= 1; - if (high & (1<<25)) - recover |= 2; - high &= ~(1<<31); - - if (high & (1<<27)) { - rdmsr(MSR_IA32_MC0_MISC+i*4, alow, ahigh); - snprintf(misc, 20, "[%08x%08x]", ahigh, alow); - } - if (high & (1<<26)) { - rdmsr(MSR_IA32_MC0_ADDR+i*4, alow, ahigh); - snprintf(addr, 24, " at %08x%08x", ahigh, alow); - } - - printk(KERN_EMERG "CPU %d: Bank %d: %08x%08x%s%s\n", - smp_processor_id(), i, high, low, misc, addr); - } - } - - if (recover & 2) - panic("CPU context corrupt"); - if (recover & 1) - panic("Unable to continue"); - - printk(KERN_EMERG "Attempting to continue.\n"); - /* - * Do not clear the MSR_IA32_MCi_STATUS if the error is not - * recoverable/continuable.This will allow BIOS to look at the MSRs - * for errors if the OS could not log the error: - */ - for (i = 0; i < nr_mce_banks; i++) { - unsigned int msr; - - msr = MSR_IA32_MC0_STATUS+i*4; - rdmsr(msr, low, high); - if (high & (1<<31)) { - /* Clear it: */ - wrmsr(msr, 0UL, 0UL); - /* Serialize: */ - wmb(); - add_taint(TAINT_MACHINE_CHECK); - } - } - mcgstl &= ~(1<<2); - wrmsr(MSR_IA32_MCG_STATUS, mcgstl, mcgsth); -} - -/* Set up machine check reporting for processors with Intel style MCE: */ -void intel_p6_mcheck_init(struct cpuinfo_x86 *c) -{ - u32 l, h; - int i; - - /* Check for MCE support */ - if (!cpu_has(c, X86_FEATURE_MCE)) - return; - - /* Check for PPro style MCA */ - if (!cpu_has(c, X86_FEATURE_MCA)) - return; - - /* Ok machine check is available */ - machine_check_vector = intel_machine_check; - /* Make sure the vector pointer is visible before we enable MCEs: */ - wmb(); - - printk(KERN_INFO "Intel machine check architecture supported.\n"); - rdmsr(MSR_IA32_MCG_CAP, l, h); - if (l & (1<<8)) /* Control register present ? */ - wrmsr(MSR_IA32_MCG_CTL, 0xffffffff, 0xffffffff); - nr_mce_banks = l & 0xff; - - /* - * Following the example in IA-32 SDM Vol 3: - * - MC0_CTL should not be written - * - Status registers on all banks should be cleared on reset - */ - for (i = 1; i < nr_mce_banks; i++) - wrmsr(MSR_IA32_MC0_CTL+4*i, 0xffffffff, 0xffffffff); - - for (i = 0; i < nr_mce_banks; i++) - wrmsr(MSR_IA32_MC0_STATUS+4*i, 0x0, 0x0); - - set_in_cr4(X86_CR4_MCE); - printk(KERN_INFO "Intel machine check reporting enabled on CPU#%d.\n", - smp_processor_id()); -} diff --git a/arch/x86/kernel/cpu/mcheck/therm_throt.c b/arch/x86/kernel/cpu/mcheck/therm_throt.c index 5957a93e517..63a56d147e4 100644 --- a/arch/x86/kernel/cpu/mcheck/therm_throt.c +++ b/arch/x86/kernel/cpu/mcheck/therm_throt.c @@ -260,9 +260,6 @@ void intel_init_thermal(struct cpuinfo_x86 *c) return; } - if (cpu_has(c, X86_FEATURE_TM2) && (l & MSR_IA32_MISC_ENABLE_TM2)) - tm2 = 1; - /* Check whether a vector already exists */ if (h & APIC_VECTOR_MASK) { printk(KERN_DEBUG @@ -271,6 +268,16 @@ void intel_init_thermal(struct cpuinfo_x86 *c) return; } + /* early Pentium M models use different method for enabling TM2 */ + if (cpu_has(c, X86_FEATURE_TM2)) { + if (c->x86 == 6 && (c->x86_model == 9 || c->x86_model == 13)) { + rdmsr(MSR_THERM2_CTL, l, h); + if (l & MSR_THERM2_CTL_TM_SELECT) + tm2 = 1; + } else if (l & MSR_IA32_MISC_ENABLE_TM2) + tm2 = 1; + } + /* We'll mask the thermal vector in the lapic till we're ready: */ h = THERMAL_APIC_VECTOR | APIC_DM_FIXED | APIC_LVT_MASKED; apic_write(APIC_LVTTHMR, h); diff --git a/arch/x86/kernel/cpu/mtrr/main.c b/arch/x86/kernel/cpu/mtrr/main.c index 7af0f88a416..84e83de5457 100644 --- a/arch/x86/kernel/cpu/mtrr/main.c +++ b/arch/x86/kernel/cpu/mtrr/main.c @@ -58,6 +58,7 @@ unsigned int mtrr_usage_table[MTRR_MAX_VAR_RANGES]; static DEFINE_MUTEX(mtrr_mutex); u64 size_or_mask, size_and_mask; +static bool mtrr_aps_delayed_init; static struct mtrr_ops *mtrr_ops[X86_VENDOR_NUM]; @@ -163,7 +164,10 @@ static void ipi_handler(void *info) if (data->smp_reg != ~0U) { mtrr_if->set(data->smp_reg, data->smp_base, data->smp_size, data->smp_type); - } else { + } else if (mtrr_aps_delayed_init) { + /* + * Initialize the MTRRs inaddition to the synchronisation. + */ mtrr_if->set_all(); } @@ -265,6 +269,8 @@ set_mtrr(unsigned int reg, unsigned long base, unsigned long size, mtrr_type typ */ if (reg != ~0U) mtrr_if->set(reg, base, size, type); + else if (!mtrr_aps_delayed_init) + mtrr_if->set_all(); /* Wait for the others */ while (atomic_read(&data.count)) @@ -721,9 +727,7 @@ void __init mtrr_bp_init(void) void mtrr_ap_init(void) { - unsigned long flags; - - if (!mtrr_if || !use_intel()) + if (!use_intel() || mtrr_aps_delayed_init) return; /* * Ideally we should hold mtrr_mutex here to avoid mtrr entries @@ -738,11 +742,7 @@ void mtrr_ap_init(void) * 2. cpu hotadd time. We let mtrr_add/del_page hold cpuhotplug * lock to prevent mtrr entry changes */ - local_irq_save(flags); - - mtrr_if->set_all(); - - local_irq_restore(flags); + set_mtrr(~0U, 0, 0, 0); } /** @@ -753,6 +753,34 @@ void mtrr_save_state(void) smp_call_function_single(0, mtrr_save_fixed_ranges, NULL, 1); } +void set_mtrr_aps_delayed_init(void) +{ + if (!use_intel()) + return; + + mtrr_aps_delayed_init = true; +} + +/* + * MTRR initialization for all AP's + */ +void mtrr_aps_init(void) +{ + if (!use_intel()) + return; + + set_mtrr(~0U, 0, 0, 0); + mtrr_aps_delayed_init = false; +} + +void mtrr_bp_restore(void) +{ + if (!use_intel()) + return; + + mtrr_if->set_all(); +} + static int __init mtrr_init_finialize(void) { if (!mtrr_if) diff --git a/arch/x86/kernel/cpu/perf_counter.c b/arch/x86/kernel/cpu/perf_counter.c index 900332b800f..2732e2c1e4d 100644 --- a/arch/x86/kernel/cpu/perf_counter.c +++ b/arch/x86/kernel/cpu/perf_counter.c @@ -6,6 +6,7 @@ * Copyright (C) 2009 Jaswinder Singh Rajput * Copyright (C) 2009 Advanced Micro Devices, Inc., Robert Richter * Copyright (C) 2008-2009 Red Hat, Inc., Peter Zijlstra <pzijlstr@redhat.com> + * Copyright (C) 2009 Intel Corporation, <markus.t.metzger@intel.com> * * For licencing details see kernel-base/COPYING */ @@ -20,6 +21,7 @@ #include <linux/sched.h> #include <linux/uaccess.h> #include <linux/highmem.h> +#include <linux/cpu.h> #include <asm/apic.h> #include <asm/stacktrace.h> @@ -27,12 +29,52 @@ static u64 perf_counter_mask __read_mostly; +/* The maximal number of PEBS counters: */ +#define MAX_PEBS_COUNTERS 4 + +/* The size of a BTS record in bytes: */ +#define BTS_RECORD_SIZE 24 + +/* The size of a per-cpu BTS buffer in bytes: */ +#define BTS_BUFFER_SIZE (BTS_RECORD_SIZE * 1024) + +/* The BTS overflow threshold in bytes from the end of the buffer: */ +#define BTS_OVFL_TH (BTS_RECORD_SIZE * 64) + + +/* + * Bits in the debugctlmsr controlling branch tracing. + */ +#define X86_DEBUGCTL_TR (1 << 6) +#define X86_DEBUGCTL_BTS (1 << 7) +#define X86_DEBUGCTL_BTINT (1 << 8) +#define X86_DEBUGCTL_BTS_OFF_OS (1 << 9) +#define X86_DEBUGCTL_BTS_OFF_USR (1 << 10) + +/* + * A debug store configuration. + * + * We only support architectures that use 64bit fields. + */ +struct debug_store { + u64 bts_buffer_base; + u64 bts_index; + u64 bts_absolute_maximum; + u64 bts_interrupt_threshold; + u64 pebs_buffer_base; + u64 pebs_index; + u64 pebs_absolute_maximum; + u64 pebs_interrupt_threshold; + u64 pebs_counter_reset[MAX_PEBS_COUNTERS]; +}; + struct cpu_hw_counters { struct perf_counter *counters[X86_PMC_IDX_MAX]; unsigned long used_mask[BITS_TO_LONGS(X86_PMC_IDX_MAX)]; unsigned long active_mask[BITS_TO_LONGS(X86_PMC_IDX_MAX)]; unsigned long interrupts; int enabled; + struct debug_store *ds; }; /* @@ -58,6 +100,8 @@ struct x86_pmu { int apic; u64 max_period; u64 intel_ctrl; + void (*enable_bts)(u64 config); + void (*disable_bts)(void); }; static struct x86_pmu x86_pmu __read_mostly; @@ -577,6 +621,9 @@ x86_perf_counter_update(struct perf_counter *counter, u64 prev_raw_count, new_raw_count; s64 delta; + if (idx == X86_PMC_IDX_FIXED_BTS) + return 0; + /* * Careful: an NMI might modify the previous counter value. * @@ -666,10 +713,110 @@ static void release_pmc_hardware(void) #endif } +static inline bool bts_available(void) +{ + return x86_pmu.enable_bts != NULL; +} + +static inline void init_debug_store_on_cpu(int cpu) +{ + struct debug_store *ds = per_cpu(cpu_hw_counters, cpu).ds; + + if (!ds) + return; + + wrmsr_on_cpu(cpu, MSR_IA32_DS_AREA, + (u32)((u64)(unsigned long)ds), + (u32)((u64)(unsigned long)ds >> 32)); +} + +static inline void fini_debug_store_on_cpu(int cpu) +{ + if (!per_cpu(cpu_hw_counters, cpu).ds) + return; + + wrmsr_on_cpu(cpu, MSR_IA32_DS_AREA, 0, 0); +} + +static void release_bts_hardware(void) +{ + int cpu; + + if (!bts_available()) + return; + + get_online_cpus(); + + for_each_online_cpu(cpu) + fini_debug_store_on_cpu(cpu); + + for_each_possible_cpu(cpu) { + struct debug_store *ds = per_cpu(cpu_hw_counters, cpu).ds; + + if (!ds) + continue; + + per_cpu(cpu_hw_counters, cpu).ds = NULL; + + kfree((void *)(unsigned long)ds->bts_buffer_base); + kfree(ds); + } + + put_online_cpus(); +} + +static int reserve_bts_hardware(void) +{ + int cpu, err = 0; + + if (!bts_available()) + return 0; + + get_online_cpus(); + + for_each_possible_cpu(cpu) { + struct debug_store *ds; + void *buffer; + + err = -ENOMEM; + buffer = kzalloc(BTS_BUFFER_SIZE, GFP_KERNEL); + if (unlikely(!buffer)) + break; + + ds = kzalloc(sizeof(*ds), GFP_KERNEL); + if (unlikely(!ds)) { + kfree(buffer); + break; + } + + ds->bts_buffer_base = (u64)(unsigned long)buffer; + ds->bts_index = ds->bts_buffer_base; + ds->bts_absolute_maximum = + ds->bts_buffer_base + BTS_BUFFER_SIZE; + ds->bts_interrupt_threshold = + ds->bts_absolute_maximum - BTS_OVFL_TH; + + per_cpu(cpu_hw_counters, cpu).ds = ds; + err = 0; + } + + if (err) + release_bts_hardware(); + else { + for_each_online_cpu(cpu) + init_debug_store_on_cpu(cpu); + } + + put_online_cpus(); + + return err; +} + static void hw_perf_counter_destroy(struct perf_counter *counter) { if (atomic_dec_and_mutex_lock(&active_counters, &pmc_reserve_mutex)) { release_pmc_hardware(); + release_bts_hardware(); mutex_unlock(&pmc_reserve_mutex); } } @@ -712,6 +859,42 @@ set_ext_hw_attr(struct hw_perf_counter *hwc, struct perf_counter_attr *attr) return 0; } +static void intel_pmu_enable_bts(u64 config) +{ + unsigned long debugctlmsr; + + debugctlmsr = get_debugctlmsr(); + + debugctlmsr |= X86_DEBUGCTL_TR; + debugctlmsr |= X86_DEBUGCTL_BTS; + debugctlmsr |= X86_DEBUGCTL_BTINT; + + if (!(config & ARCH_PERFMON_EVENTSEL_OS)) + debugctlmsr |= X86_DEBUGCTL_BTS_OFF_OS; + + if (!(config & ARCH_PERFMON_EVENTSEL_USR)) + debugctlmsr |= X86_DEBUGCTL_BTS_OFF_USR; + + update_debugctlmsr(debugctlmsr); +} + +static void intel_pmu_disable_bts(void) +{ + struct cpu_hw_counters *cpuc = &__get_cpu_var(cpu_hw_counters); + unsigned long debugctlmsr; + + if (!cpuc->ds) + return; + + debugctlmsr = get_debugctlmsr(); + + debugctlmsr &= + ~(X86_DEBUGCTL_TR | X86_DEBUGCTL_BTS | X86_DEBUGCTL_BTINT | + X86_DEBUGCTL_BTS_OFF_OS | X86_DEBUGCTL_BTS_OFF_USR); + + update_debugctlmsr(debugctlmsr); +} + /* * Setup the hardware configuration for a given attr_type */ @@ -728,9 +911,13 @@ static int __hw_perf_counter_init(struct perf_counter *counter) err = 0; if (!atomic_inc_not_zero(&active_counters)) { mutex_lock(&pmc_reserve_mutex); - if (atomic_read(&active_counters) == 0 && !reserve_pmc_hardware()) - err = -EBUSY; - else + if (atomic_read(&active_counters) == 0) { + if (!reserve_pmc_hardware()) + err = -EBUSY; + else + err = reserve_bts_hardware(); + } + if (!err) atomic_inc(&active_counters); mutex_unlock(&pmc_reserve_mutex); } @@ -793,6 +980,20 @@ static int __hw_perf_counter_init(struct perf_counter *counter) if (config == -1LL) return -EINVAL; + /* + * Branch tracing: + */ + if ((attr->config == PERF_COUNT_HW_BRANCH_INSTRUCTIONS) && + (hwc->sample_period == 1)) { + /* BTS is not supported by this architecture. */ + if (!bts_available()) + return -EOPNOTSUPP; + + /* BTS is currently only allowed for user-mode. */ + if (hwc->config & ARCH_PERFMON_EVENTSEL_OS) + return -EOPNOTSUPP; + } + hwc->config |= config; return 0; @@ -817,7 +1018,18 @@ static void p6_pmu_disable_all(void) static void intel_pmu_disable_all(void) { + struct cpu_hw_counters *cpuc = &__get_cpu_var(cpu_hw_counters); + + if (!cpuc->enabled) + return; + + cpuc->enabled = 0; + barrier(); + wrmsrl(MSR_CORE_PERF_GLOBAL_CTRL, 0); + + if (test_bit(X86_PMC_IDX_FIXED_BTS, cpuc->active_mask)) + intel_pmu_disable_bts(); } static void amd_pmu_disable_all(void) @@ -875,7 +1087,25 @@ static void p6_pmu_enable_all(void) static void intel_pmu_enable_all(void) { + struct cpu_hw_counters *cpuc = &__get_cpu_var(cpu_hw_counters); + + if (cpuc->enabled) + return; + + cpuc->enabled = 1; + barrier(); + wrmsrl(MSR_CORE_PERF_GLOBAL_CTRL, x86_pmu.intel_ctrl); + + if (test_bit(X86_PMC_IDX_FIXED_BTS, cpuc->active_mask)) { + struct perf_counter *counter = + cpuc->counters[X86_PMC_IDX_FIXED_BTS]; + + if (WARN_ON_ONCE(!counter)) + return; + + intel_pmu_enable_bts(counter->hw.config); + } } static void amd_pmu_enable_all(void) @@ -962,6 +1192,11 @@ p6_pmu_disable_counter(struct hw_perf_counter *hwc, int idx) static inline void intel_pmu_disable_counter(struct hw_perf_counter *hwc, int idx) { + if (unlikely(idx == X86_PMC_IDX_FIXED_BTS)) { + intel_pmu_disable_bts(); + return; + } + if (unlikely(hwc->config_base == MSR_ARCH_PERFMON_FIXED_CTR_CTRL)) { intel_pmu_disable_fixed(hwc, idx); return; @@ -976,7 +1211,7 @@ amd_pmu_disable_counter(struct hw_perf_counter *hwc, int idx) x86_pmu_disable_counter(hwc, idx); } -static DEFINE_PER_CPU(u64, prev_left[X86_PMC_IDX_MAX]); +static DEFINE_PER_CPU(u64 [X86_PMC_IDX_MAX], pmc_prev_left); /* * Set the next IRQ period, based on the hwc->period_left value. @@ -990,6 +1225,9 @@ x86_perf_counter_set_period(struct perf_counter *counter, s64 period = hwc->sample_period; int err, ret = 0; + if (idx == X86_PMC_IDX_FIXED_BTS) + return 0; + /* * If we are way outside a reasoable range then just skip forward: */ @@ -1015,7 +1253,7 @@ x86_perf_counter_set_period(struct perf_counter *counter, if (left > x86_pmu.max_period) left = x86_pmu.max_period; - per_cpu(prev_left[idx], smp_processor_id()) = left; + per_cpu(pmc_prev_left[idx], smp_processor_id()) = left; /* * The hw counter starts counting from this counter offset, @@ -1072,6 +1310,14 @@ static void p6_pmu_enable_counter(struct hw_perf_counter *hwc, int idx) static void intel_pmu_enable_counter(struct hw_perf_counter *hwc, int idx) { + if (unlikely(idx == X86_PMC_IDX_FIXED_BTS)) { + if (!__get_cpu_var(cpu_hw_counters).enabled) + return; + + intel_pmu_enable_bts(hwc->config); + return; + } + if (unlikely(hwc->config_base == MSR_ARCH_PERFMON_FIXED_CTR_CTRL)) { intel_pmu_enable_fixed(hwc, idx); return; @@ -1093,11 +1339,16 @@ fixed_mode_idx(struct perf_counter *counter, struct hw_perf_counter *hwc) { unsigned int event; + event = hwc->config & ARCH_PERFMON_EVENT_MASK; + + if (unlikely((event == + x86_pmu.event_map(PERF_COUNT_HW_BRANCH_INSTRUCTIONS)) && + (hwc->sample_period == 1))) + return X86_PMC_IDX_FIXED_BTS; + if (!x86_pmu.num_counters_fixed) return -1; - event = hwc->config & ARCH_PERFMON_EVENT_MASK; - if (unlikely(event == x86_pmu.event_map(PERF_COUNT_HW_INSTRUCTIONS))) return X86_PMC_IDX_FIXED_INSTRUCTIONS; if (unlikely(event == x86_pmu.event_map(PERF_COUNT_HW_CPU_CYCLES))) @@ -1118,7 +1369,15 @@ static int x86_pmu_enable(struct perf_counter *counter) int idx; idx = fixed_mode_idx(counter, hwc); - if (idx >= 0) { + if (idx == X86_PMC_IDX_FIXED_BTS) { + /* BTS is already occupied. */ + if (test_and_set_bit(idx, cpuc->used_mask)) + return -EAGAIN; + + hwc->config_base = 0; + hwc->counter_base = 0; + hwc->idx = idx; + } else if (idx >= 0) { /* * Try to get the fixed counter, if that is already taken * then try to get a generic counter: @@ -1211,7 +1470,7 @@ void perf_counter_print_debug(void) rdmsrl(x86_pmu.eventsel + idx, pmc_ctrl); rdmsrl(x86_pmu.perfctr + idx, pmc_count); - prev_left = per_cpu(prev_left[idx], cpu); + prev_left = per_cpu(pmc_prev_left[idx], cpu); pr_info("CPU#%d: gen-PMC%d ctrl: %016llx\n", cpu, idx, pmc_ctrl); @@ -1229,6 +1488,44 @@ void perf_counter_print_debug(void) local_irq_restore(flags); } +static void intel_pmu_drain_bts_buffer(struct cpu_hw_counters *cpuc, + struct perf_sample_data *data) +{ + struct debug_store *ds = cpuc->ds; + struct bts_record { + u64 from; + u64 to; + u64 flags; + }; + struct perf_counter *counter = cpuc->counters[X86_PMC_IDX_FIXED_BTS]; + unsigned long orig_ip = data->regs->ip; + struct bts_record *at, *top; + + if (!counter) + return; + + if (!ds) + return; + + at = (struct bts_record *)(unsigned long)ds->bts_buffer_base; + top = (struct bts_record *)(unsigned long)ds->bts_index; + + ds->bts_index = ds->bts_buffer_base; + + for (; at < top; at++) { + data->regs->ip = at->from; + data->addr = at->to; + + perf_counter_output(counter, 1, data); + } + + data->regs->ip = orig_ip; + data->addr = 0; + + /* There's new data available. */ + counter->pending_kill = POLL_IN; +} + static void x86_pmu_disable(struct perf_counter *counter) { struct cpu_hw_counters *cpuc = &__get_cpu_var(cpu_hw_counters); @@ -1253,6 +1550,15 @@ static void x86_pmu_disable(struct perf_counter *counter) * that we are disabling: */ x86_perf_counter_update(counter, hwc, idx); + + /* Drain the remaining BTS records. */ + if (unlikely(idx == X86_PMC_IDX_FIXED_BTS)) { + struct perf_sample_data data; + struct pt_regs regs; + + data.regs = ®s; + intel_pmu_drain_bts_buffer(cpuc, &data); + } cpuc->counters[idx] = NULL; clear_bit(idx, cpuc->used_mask); @@ -1280,6 +1586,7 @@ static int intel_pmu_save_and_restart(struct perf_counter *counter) static void intel_pmu_reset(void) { + struct debug_store *ds = __get_cpu_var(cpu_hw_counters).ds; unsigned long flags; int idx; @@ -1297,6 +1604,8 @@ static void intel_pmu_reset(void) for (idx = 0; idx < x86_pmu.num_counters_fixed; idx++) { checking_wrmsrl(MSR_ARCH_PERFMON_FIXED_CTR0 + idx, 0ull); } + if (ds) + ds->bts_index = ds->bts_buffer_base; local_irq_restore(flags); } @@ -1362,6 +1671,7 @@ static int intel_pmu_handle_irq(struct pt_regs *regs) cpuc = &__get_cpu_var(cpu_hw_counters); perf_disable(); + intel_pmu_drain_bts_buffer(cpuc, &data); status = intel_pmu_get_status(); if (!status) { perf_enable(); @@ -1571,6 +1881,8 @@ static struct x86_pmu intel_pmu = { * the generic counter period: */ .max_period = (1ULL << 31) - 1, + .enable_bts = intel_pmu_enable_bts, + .disable_bts = intel_pmu_disable_bts, }; static struct x86_pmu amd_pmu = { @@ -1798,8 +2110,8 @@ void callchain_store(struct perf_callchain_entry *entry, u64 ip) entry->ip[entry->nr++] = ip; } -static DEFINE_PER_CPU(struct perf_callchain_entry, irq_entry); -static DEFINE_PER_CPU(struct perf_callchain_entry, nmi_entry); +static DEFINE_PER_CPU(struct perf_callchain_entry, pmc_irq_entry); +static DEFINE_PER_CPU(struct perf_callchain_entry, pmc_nmi_entry); static DEFINE_PER_CPU(int, in_nmi_frame); @@ -1952,9 +2264,9 @@ struct perf_callchain_entry *perf_callchain(struct pt_regs *regs) struct perf_callchain_entry *entry; if (in_nmi()) - entry = &__get_cpu_var(nmi_entry); + entry = &__get_cpu_var(pmc_nmi_entry); else - entry = &__get_cpu_var(irq_entry); + entry = &__get_cpu_var(pmc_irq_entry); entry->nr = 0; @@ -1962,3 +2274,8 @@ struct perf_callchain_entry *perf_callchain(struct pt_regs *regs) return entry; } + +void hw_perf_counter_setup_online(int cpu) +{ + init_debug_store_on_cpu(cpu); +} diff --git a/arch/x86/kernel/cpu/proc.c b/arch/x86/kernel/cpu/proc.c index 1e904346bbf..62ac8cb6ba2 100644 --- a/arch/x86/kernel/cpu/proc.c +++ b/arch/x86/kernel/cpu/proc.c @@ -116,11 +116,9 @@ static int show_cpuinfo(struct seq_file *m, void *v) seq_printf(m, "TLB size\t: %d 4K pages\n", c->x86_tlbsize); #endif seq_printf(m, "clflush size\t: %u\n", c->x86_clflush_size); -#ifdef CONFIG_X86_64 seq_printf(m, "cache_alignment\t: %d\n", c->x86_cache_alignment); seq_printf(m, "address sizes\t: %u bits physical, %u bits virtual\n", c->x86_phys_bits, c->x86_virt_bits); -#endif seq_printf(m, "power management:"); for (i = 0; i < 32; i++) { diff --git a/arch/x86/kernel/cpu/sched.c b/arch/x86/kernel/cpu/sched.c new file mode 100644 index 00000000000..a640ae5ad20 --- /dev/null +++ b/arch/x86/kernel/cpu/sched.c @@ -0,0 +1,55 @@ +#include <linux/sched.h> +#include <linux/math64.h> +#include <linux/percpu.h> +#include <linux/irqflags.h> + +#include <asm/cpufeature.h> +#include <asm/processor.h> + +#ifdef CONFIG_SMP + +static DEFINE_PER_CPU(struct aperfmperf, old_perf_sched); + +static unsigned long scale_aperfmperf(void) +{ + struct aperfmperf val, *old = &__get_cpu_var(old_perf_sched); + unsigned long ratio, flags; + + local_irq_save(flags); + get_aperfmperf(&val); + local_irq_restore(flags); + + ratio = calc_aperfmperf_ratio(old, &val); + *old = val; + + return ratio; +} + +unsigned long arch_scale_freq_power(struct sched_domain *sd, int cpu) +{ + /* + * do aperf/mperf on the cpu level because it includes things + * like turbo mode, which are relevant to full cores. + */ + if (boot_cpu_has(X86_FEATURE_APERFMPERF)) + return scale_aperfmperf(); + + /* + * maybe have something cpufreq here + */ + + return default_scale_freq_power(sd, cpu); +} + +unsigned long arch_scale_smt_power(struct sched_domain *sd, int cpu) +{ + /* + * aperf/mperf already includes the smt gain + */ + if (boot_cpu_has(X86_FEATURE_APERFMPERF)) + return SCHED_LOAD_SCALE; + + return default_scale_smt_power(sd, cpu); +} + +#endif |