diff options
Diffstat (limited to 'arch')
-rw-r--r-- | arch/x86/include/asm/msr-index.h | 1 | ||||
-rw-r--r-- | arch/x86/include/asm/perf_event.h | 19 | ||||
-rw-r--r-- | arch/x86/kernel/cpu/perf_event.c | 21 | ||||
-rw-r--r-- | arch/x86/kernel/cpu/perf_event_intel_ds.c | 216 | ||||
-rw-r--r-- | arch/x86/oprofile/nmi_int.c | 6 | ||||
-rw-r--r-- | arch/x86/oprofile/op_model_amd.c | 146 |
6 files changed, 271 insertions, 138 deletions
diff --git a/arch/x86/include/asm/msr-index.h b/arch/x86/include/asm/msr-index.h index 83c4bb1d917d..3ea3dc487047 100644 --- a/arch/x86/include/asm/msr-index.h +++ b/arch/x86/include/asm/msr-index.h @@ -121,6 +121,7 @@ #define MSR_AMD64_IBSDCLINAD 0xc0011038 #define MSR_AMD64_IBSDCPHYSAD 0xc0011039 #define MSR_AMD64_IBSCTL 0xc001103a +#define MSR_AMD64_IBSBRTARGET 0xc001103b /* Fam 10h MSRs */ #define MSR_FAM10H_MMIO_CONF_BASE 0xc0010058 diff --git a/arch/x86/include/asm/perf_event.h b/arch/x86/include/asm/perf_event.h index 6e742cc4251b..550e26b1dbb3 100644 --- a/arch/x86/include/asm/perf_event.h +++ b/arch/x86/include/asm/perf_event.h @@ -111,17 +111,18 @@ union cpuid10_edx { #define X86_PMC_IDX_FIXED_BTS (X86_PMC_IDX_FIXED + 16) /* IbsFetchCtl bits/masks */ -#define IBS_FETCH_RAND_EN (1ULL<<57) -#define IBS_FETCH_VAL (1ULL<<49) -#define IBS_FETCH_ENABLE (1ULL<<48) -#define IBS_FETCH_CNT 0xFFFF0000ULL -#define IBS_FETCH_MAX_CNT 0x0000FFFFULL +#define IBS_FETCH_RAND_EN (1ULL<<57) +#define IBS_FETCH_VAL (1ULL<<49) +#define IBS_FETCH_ENABLE (1ULL<<48) +#define IBS_FETCH_CNT 0xFFFF0000ULL +#define IBS_FETCH_MAX_CNT 0x0000FFFFULL /* IbsOpCtl bits */ -#define IBS_OP_CNT_CTL (1ULL<<19) -#define IBS_OP_VAL (1ULL<<18) -#define IBS_OP_ENABLE (1ULL<<17) -#define IBS_OP_MAX_CNT 0x0000FFFFULL +#define IBS_OP_CNT_CTL (1ULL<<19) +#define IBS_OP_VAL (1ULL<<18) +#define IBS_OP_ENABLE (1ULL<<17) +#define IBS_OP_MAX_CNT 0x0000FFFFULL +#define IBS_OP_MAX_CNT_EXT 0x007FFFFFULL /* not a register bit mask */ #ifdef CONFIG_PERF_EVENTS extern void init_hw_perf_events(void); diff --git a/arch/x86/kernel/cpu/perf_event.c b/arch/x86/kernel/cpu/perf_event.c index c1e8c7a51164..ed6310183efb 100644 --- a/arch/x86/kernel/cpu/perf_event.c +++ b/arch/x86/kernel/cpu/perf_event.c @@ -237,6 +237,7 @@ struct x86_pmu { * Intel DebugStore bits */ int bts, pebs; + int bts_active, pebs_active; int pebs_record_size; void (*drain_pebs)(struct pt_regs *regs); struct event_constraint *pebs_constraints; @@ -380,7 +381,7 @@ static void release_pmc_hardware(void) {} #endif -static int reserve_ds_buffers(void); +static void reserve_ds_buffers(void); static void release_ds_buffers(void); static void hw_perf_event_destroy(struct perf_event *event) @@ -477,7 +478,7 @@ static int x86_setup_perfctr(struct perf_event *event) if ((attr->config == PERF_COUNT_HW_BRANCH_INSTRUCTIONS) && (hwc->sample_period == 1)) { /* BTS is not supported by this architecture. */ - if (!x86_pmu.bts) + if (!x86_pmu.bts_active) return -EOPNOTSUPP; /* BTS is currently only allowed for user-mode. */ @@ -496,12 +497,13 @@ static int x86_pmu_hw_config(struct perf_event *event) int precise = 0; /* Support for constant skid */ - if (x86_pmu.pebs) + if (x86_pmu.pebs_active) { precise++; - /* Support for IP fixup */ - if (x86_pmu.lbr_nr) - precise++; + /* Support for IP fixup */ + if (x86_pmu.lbr_nr) + precise++; + } if (event->attr.precise_ip > precise) return -EOPNOTSUPP; @@ -543,11 +545,8 @@ static int __x86_pmu_event_init(struct perf_event *event) if (atomic_read(&active_events) == 0) { if (!reserve_pmc_hardware()) err = -EBUSY; - else { - err = reserve_ds_buffers(); - if (err) - release_pmc_hardware(); - } + else + reserve_ds_buffers(); } if (!err) atomic_inc(&active_events); diff --git a/arch/x86/kernel/cpu/perf_event_intel_ds.c b/arch/x86/kernel/cpu/perf_event_intel_ds.c index 4977f9c400e5..b7dcd9f2b8a0 100644 --- a/arch/x86/kernel/cpu/perf_event_intel_ds.c +++ b/arch/x86/kernel/cpu/perf_event_intel_ds.c @@ -74,6 +74,107 @@ static void fini_debug_store_on_cpu(int cpu) wrmsr_on_cpu(cpu, MSR_IA32_DS_AREA, 0, 0); } +static int alloc_pebs_buffer(int cpu) +{ + struct debug_store *ds = per_cpu(cpu_hw_events, cpu).ds; + int node = cpu_to_node(cpu); + int max, thresh = 1; /* always use a single PEBS record */ + void *buffer; + + if (!x86_pmu.pebs) + return 0; + + buffer = kmalloc_node(PEBS_BUFFER_SIZE, GFP_KERNEL | __GFP_ZERO, node); + if (unlikely(!buffer)) + return -ENOMEM; + + max = PEBS_BUFFER_SIZE / x86_pmu.pebs_record_size; + + ds->pebs_buffer_base = (u64)(unsigned long)buffer; + ds->pebs_index = ds->pebs_buffer_base; + ds->pebs_absolute_maximum = ds->pebs_buffer_base + + max * x86_pmu.pebs_record_size; + + ds->pebs_interrupt_threshold = ds->pebs_buffer_base + + thresh * x86_pmu.pebs_record_size; + + return 0; +} + +static void release_pebs_buffer(int cpu) +{ + struct debug_store *ds = per_cpu(cpu_hw_events, cpu).ds; + + if (!ds || !x86_pmu.pebs) + return; + + kfree((void *)(unsigned long)ds->pebs_buffer_base); + ds->pebs_buffer_base = 0; +} + +static int alloc_bts_buffer(int cpu) +{ + struct debug_store *ds = per_cpu(cpu_hw_events, cpu).ds; + int node = cpu_to_node(cpu); + int max, thresh; + void *buffer; + + if (!x86_pmu.bts) + return 0; + + buffer = kmalloc_node(BTS_BUFFER_SIZE, GFP_KERNEL | __GFP_ZERO, node); + if (unlikely(!buffer)) + return -ENOMEM; + + max = BTS_BUFFER_SIZE / BTS_RECORD_SIZE; + thresh = max / 16; + + ds->bts_buffer_base = (u64)(unsigned long)buffer; + ds->bts_index = ds->bts_buffer_base; + ds->bts_absolute_maximum = ds->bts_buffer_base + + max * BTS_RECORD_SIZE; + ds->bts_interrupt_threshold = ds->bts_absolute_maximum - + thresh * BTS_RECORD_SIZE; + + return 0; +} + +static void release_bts_buffer(int cpu) +{ + struct debug_store *ds = per_cpu(cpu_hw_events, cpu).ds; + + if (!ds || !x86_pmu.bts) + return; + + kfree((void *)(unsigned long)ds->bts_buffer_base); + ds->bts_buffer_base = 0; +} + +static int alloc_ds_buffer(int cpu) +{ + int node = cpu_to_node(cpu); + struct debug_store *ds; + + ds = kmalloc_node(sizeof(*ds), GFP_KERNEL | __GFP_ZERO, node); + if (unlikely(!ds)) + return -ENOMEM; + + per_cpu(cpu_hw_events, cpu).ds = ds; + + return 0; +} + +static void release_ds_buffer(int cpu) +{ + struct debug_store *ds = per_cpu(cpu_hw_events, cpu).ds; + + if (!ds) + return; + + per_cpu(cpu_hw_events, cpu).ds = NULL; + kfree(ds); +} + static void release_ds_buffers(void) { int cpu; @@ -82,93 +183,77 @@ static void release_ds_buffers(void) 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_events, cpu).ds; - - if (!ds) - continue; - - per_cpu(cpu_hw_events, cpu).ds = NULL; - - kfree((void *)(unsigned long)ds->pebs_buffer_base); - kfree((void *)(unsigned long)ds->bts_buffer_base); - kfree(ds); + release_pebs_buffer(cpu); + release_bts_buffer(cpu); + release_ds_buffer(cpu); } - put_online_cpus(); } -static int reserve_ds_buffers(void) +static void reserve_ds_buffers(void) { - int cpu, err = 0; + int bts_err = 0, pebs_err = 0; + int cpu; + + x86_pmu.bts_active = 0; + x86_pmu.pebs_active = 0; if (!x86_pmu.bts && !x86_pmu.pebs) - return 0; + return; + + if (!x86_pmu.bts) + bts_err = 1; + + if (!x86_pmu.pebs) + pebs_err = 1; get_online_cpus(); for_each_possible_cpu(cpu) { - struct debug_store *ds; - void *buffer; - int max, thresh; + if (alloc_ds_buffer(cpu)) { + bts_err = 1; + pebs_err = 1; + } + + if (!bts_err && alloc_bts_buffer(cpu)) + bts_err = 1; - err = -ENOMEM; - ds = kzalloc(sizeof(*ds), GFP_KERNEL); - if (unlikely(!ds)) + if (!pebs_err && alloc_pebs_buffer(cpu)) + pebs_err = 1; + + if (bts_err && pebs_err) break; - per_cpu(cpu_hw_events, cpu).ds = ds; - - if (x86_pmu.bts) { - buffer = kzalloc(BTS_BUFFER_SIZE, GFP_KERNEL); - if (unlikely(!buffer)) - break; - - max = BTS_BUFFER_SIZE / BTS_RECORD_SIZE; - thresh = max / 16; - - ds->bts_buffer_base = (u64)(unsigned long)buffer; - ds->bts_index = ds->bts_buffer_base; - ds->bts_absolute_maximum = ds->bts_buffer_base + - max * BTS_RECORD_SIZE; - ds->bts_interrupt_threshold = ds->bts_absolute_maximum - - thresh * BTS_RECORD_SIZE; - } + } - if (x86_pmu.pebs) { - buffer = kzalloc(PEBS_BUFFER_SIZE, GFP_KERNEL); - if (unlikely(!buffer)) - break; - - max = PEBS_BUFFER_SIZE / x86_pmu.pebs_record_size; - - ds->pebs_buffer_base = (u64)(unsigned long)buffer; - ds->pebs_index = ds->pebs_buffer_base; - ds->pebs_absolute_maximum = ds->pebs_buffer_base + - max * x86_pmu.pebs_record_size; - /* - * Always use single record PEBS - */ - ds->pebs_interrupt_threshold = ds->pebs_buffer_base + - x86_pmu.pebs_record_size; - } + if (bts_err) { + for_each_possible_cpu(cpu) + release_bts_buffer(cpu); + } - err = 0; + if (pebs_err) { + for_each_possible_cpu(cpu) + release_pebs_buffer(cpu); } - if (err) - release_ds_buffers(); - else { + if (bts_err && pebs_err) { + for_each_possible_cpu(cpu) + release_ds_buffer(cpu); + } else { + if (x86_pmu.bts && !bts_err) + x86_pmu.bts_active = 1; + + if (x86_pmu.pebs && !pebs_err) + x86_pmu.pebs_active = 1; + for_each_online_cpu(cpu) init_debug_store_on_cpu(cpu); } put_online_cpus(); - - return err; } /* @@ -233,7 +318,7 @@ static int intel_pmu_drain_bts_buffer(void) if (!event) return 0; - if (!ds) + if (!x86_pmu.bts_active) return 0; at = (struct bts_record *)(unsigned long)ds->bts_buffer_base; @@ -503,7 +588,7 @@ static void intel_pmu_drain_pebs_core(struct pt_regs *iregs) struct pebs_record_core *at, *top; int n; - if (!ds || !x86_pmu.pebs) + if (!x86_pmu.pebs_active) return; at = (struct pebs_record_core *)(unsigned long)ds->pebs_buffer_base; @@ -545,7 +630,7 @@ static void intel_pmu_drain_pebs_nhm(struct pt_regs *iregs) u64 status = 0; int bit, n; - if (!ds || !x86_pmu.pebs) + if (!x86_pmu.pebs_active) return; at = (struct pebs_record_nhm *)(unsigned long)ds->pebs_buffer_base; @@ -630,9 +715,8 @@ static void intel_ds_init(void) #else /* CONFIG_CPU_SUP_INTEL */ -static int reserve_ds_buffers(void) +static void reserve_ds_buffers(void) { - return 0; } static void release_ds_buffers(void) diff --git a/arch/x86/oprofile/nmi_int.c b/arch/x86/oprofile/nmi_int.c index bd1489c3ce09..4e8baad36d37 100644 --- a/arch/x86/oprofile/nmi_int.c +++ b/arch/x86/oprofile/nmi_int.c @@ -726,6 +726,12 @@ int __init op_nmi_init(struct oprofile_operations *ops) case 0x11: cpu_type = "x86-64/family11h"; break; + case 0x12: + cpu_type = "x86-64/family12h"; + break; + case 0x14: + cpu_type = "x86-64/family14h"; + break; default: return -ENODEV; } diff --git a/arch/x86/oprofile/op_model_amd.c b/arch/x86/oprofile/op_model_amd.c index 42fb46f83883..a011bcc0f943 100644 --- a/arch/x86/oprofile/op_model_amd.c +++ b/arch/x86/oprofile/op_model_amd.c @@ -48,17 +48,24 @@ static unsigned long reset_value[NUM_VIRT_COUNTERS]; static u32 ibs_caps; -struct op_ibs_config { +struct ibs_config { unsigned long op_enabled; unsigned long fetch_enabled; unsigned long max_cnt_fetch; unsigned long max_cnt_op; unsigned long rand_en; unsigned long dispatched_ops; + unsigned long branch_target; }; -static struct op_ibs_config ibs_config; -static u64 ibs_op_ctl; +struct ibs_state { + u64 ibs_op_ctl; + int branch_target; + unsigned long sample_size; +}; + +static struct ibs_config ibs_config; +static struct ibs_state ibs_state; /* * IBS cpuid feature detection @@ -71,8 +78,16 @@ static u64 ibs_op_ctl; * bit 0 is used to indicate the existence of IBS. */ #define IBS_CAPS_AVAIL (1U<<0) +#define IBS_CAPS_FETCHSAM (1U<<1) +#define IBS_CAPS_OPSAM (1U<<2) #define IBS_CAPS_RDWROPCNT (1U<<3) #define IBS_CAPS_OPCNT (1U<<4) +#define IBS_CAPS_BRNTRGT (1U<<5) +#define IBS_CAPS_OPCNTEXT (1U<<6) + +#define IBS_CAPS_DEFAULT (IBS_CAPS_AVAIL \ + | IBS_CAPS_FETCHSAM \ + | IBS_CAPS_OPSAM) /* * IBS APIC setup @@ -99,12 +114,12 @@ static u32 get_ibs_caps(void) /* check IBS cpuid feature flags */ max_level = cpuid_eax(0x80000000); if (max_level < IBS_CPUID_FEATURES) - return IBS_CAPS_AVAIL; + return IBS_CAPS_DEFAULT; ibs_caps = cpuid_eax(IBS_CPUID_FEATURES); if (!(ibs_caps & IBS_CAPS_AVAIL)) /* cpuid flags not valid */ - return IBS_CAPS_AVAIL; + return IBS_CAPS_DEFAULT; return ibs_caps; } @@ -197,8 +212,8 @@ op_amd_handle_ibs(struct pt_regs * const regs, rdmsrl(MSR_AMD64_IBSOPCTL, ctl); if (ctl & IBS_OP_VAL) { rdmsrl(MSR_AMD64_IBSOPRIP, val); - oprofile_write_reserve(&entry, regs, val, - IBS_OP_CODE, IBS_OP_SIZE); + oprofile_write_reserve(&entry, regs, val, IBS_OP_CODE, + ibs_state.sample_size); oprofile_add_data64(&entry, val); rdmsrl(MSR_AMD64_IBSOPDATA, val); oprofile_add_data64(&entry, val); @@ -210,10 +225,14 @@ op_amd_handle_ibs(struct pt_regs * const regs, oprofile_add_data64(&entry, val); rdmsrl(MSR_AMD64_IBSDCPHYSAD, val); oprofile_add_data64(&entry, val); + if (ibs_state.branch_target) { + rdmsrl(MSR_AMD64_IBSBRTARGET, val); + oprofile_add_data(&entry, (unsigned long)val); + } oprofile_write_commit(&entry); /* reenable the IRQ */ - ctl = op_amd_randomize_ibs_op(ibs_op_ctl); + ctl = op_amd_randomize_ibs_op(ibs_state.ibs_op_ctl); wrmsrl(MSR_AMD64_IBSOPCTL, ctl); } } @@ -226,21 +245,32 @@ static inline void op_amd_start_ibs(void) if (!ibs_caps) return; + memset(&ibs_state, 0, sizeof(ibs_state)); + + /* + * Note: Since the max count settings may out of range we + * write back the actual used values so that userland can read + * it. + */ + if (ibs_config.fetch_enabled) { - val = (ibs_config.max_cnt_fetch >> 4) & IBS_FETCH_MAX_CNT; + val = ibs_config.max_cnt_fetch >> 4; + val = min(val, IBS_FETCH_MAX_CNT); + ibs_config.max_cnt_fetch = val << 4; val |= ibs_config.rand_en ? IBS_FETCH_RAND_EN : 0; val |= IBS_FETCH_ENABLE; wrmsrl(MSR_AMD64_IBSFETCHCTL, val); } if (ibs_config.op_enabled) { - ibs_op_ctl = ibs_config.max_cnt_op >> 4; + val = ibs_config.max_cnt_op >> 4; if (!(ibs_caps & IBS_CAPS_RDWROPCNT)) { /* * IbsOpCurCnt not supported. See * op_amd_randomize_ibs_op() for details. */ - ibs_op_ctl = clamp(ibs_op_ctl, 0x0081ULL, 0xFF80ULL); + val = clamp(val, 0x0081ULL, 0xFF80ULL); + ibs_config.max_cnt_op = val << 4; } else { /* * The start value is randomized with a @@ -248,13 +278,24 @@ static inline void op_amd_start_ibs(void) * with the half of the randomized range. Also * avoid underflows. */ - ibs_op_ctl = min(ibs_op_ctl + IBS_RANDOM_MAXCNT_OFFSET, - IBS_OP_MAX_CNT); + val += IBS_RANDOM_MAXCNT_OFFSET; + if (ibs_caps & IBS_CAPS_OPCNTEXT) + val = min(val, IBS_OP_MAX_CNT_EXT); + else + val = min(val, IBS_OP_MAX_CNT); + ibs_config.max_cnt_op = + (val - IBS_RANDOM_MAXCNT_OFFSET) << 4; + } + val = ((val & ~IBS_OP_MAX_CNT) << 4) | (val & IBS_OP_MAX_CNT); + val |= ibs_config.dispatched_ops ? IBS_OP_CNT_CTL : 0; + val |= IBS_OP_ENABLE; + ibs_state.ibs_op_ctl = val; + ibs_state.sample_size = IBS_OP_SIZE; + if (ibs_config.branch_target) { + ibs_state.branch_target = 1; + ibs_state.sample_size++; } - if (ibs_caps & IBS_CAPS_OPCNT && ibs_config.dispatched_ops) - ibs_op_ctl |= IBS_OP_CNT_CTL; - ibs_op_ctl |= IBS_OP_ENABLE; - val = op_amd_randomize_ibs_op(ibs_op_ctl); + val = op_amd_randomize_ibs_op(ibs_state.ibs_op_ctl); wrmsrl(MSR_AMD64_IBSOPCTL, val); } } @@ -281,29 +322,25 @@ static inline int eilvt_is_available(int offset) static inline int ibs_eilvt_valid(void) { - u64 val; int offset; + u64 val; rdmsrl(MSR_AMD64_IBSCTL, val); + offset = val & IBSCTL_LVT_OFFSET_MASK; + if (!(val & IBSCTL_LVT_OFFSET_VALID)) { - pr_err(FW_BUG "cpu %d, invalid IBS " - "interrupt offset %d (MSR%08X=0x%016llx)", - smp_processor_id(), offset, - MSR_AMD64_IBSCTL, val); + pr_err(FW_BUG "cpu %d, invalid IBS interrupt offset %d (MSR%08X=0x%016llx)\n", + smp_processor_id(), offset, MSR_AMD64_IBSCTL, val); return 0; } - offset = val & IBSCTL_LVT_OFFSET_MASK; - - if (eilvt_is_available(offset)) - return !0; - - pr_err(FW_BUG "cpu %d, IBS interrupt offset %d " - "not available (MSR%08X=0x%016llx)", - smp_processor_id(), offset, - MSR_AMD64_IBSCTL, val); + if (!eilvt_is_available(offset)) { + pr_err(FW_BUG "cpu %d, IBS interrupt offset %d not available (MSR%08X=0x%016llx)\n", + smp_processor_id(), offset, MSR_AMD64_IBSCTL, val); + return 0; + } - return 0; + return 1; } static inline int get_ibs_offset(void) @@ -630,28 +667,33 @@ static int setup_ibs_files(struct super_block *sb, struct dentry *root) /* model specific files */ /* setup some reasonable defaults */ + memset(&ibs_config, 0, sizeof(ibs_config)); ibs_config.max_cnt_fetch = 250000; - ibs_config.fetch_enabled = 0; ibs_config.max_cnt_op = 250000; - ibs_config.op_enabled = 0; - ibs_config.dispatched_ops = 0; - - dir = oprofilefs_mkdir(sb, root, "ibs_fetch"); - oprofilefs_create_ulong(sb, dir, "enable", - &ibs_config.fetch_enabled); - oprofilefs_create_ulong(sb, dir, "max_count", - &ibs_config.max_cnt_fetch); - oprofilefs_create_ulong(sb, dir, "rand_enable", - &ibs_config.rand_en); - - dir = oprofilefs_mkdir(sb, root, "ibs_op"); - oprofilefs_create_ulong(sb, dir, "enable", - &ibs_config.op_enabled); - oprofilefs_create_ulong(sb, dir, "max_count", - &ibs_config.max_cnt_op); - if (ibs_caps & IBS_CAPS_OPCNT) - oprofilefs_create_ulong(sb, dir, "dispatched_ops", - &ibs_config.dispatched_ops); + + if (ibs_caps & IBS_CAPS_FETCHSAM) { + dir = oprofilefs_mkdir(sb, root, "ibs_fetch"); + oprofilefs_create_ulong(sb, dir, "enable", + &ibs_config.fetch_enabled); + oprofilefs_create_ulong(sb, dir, "max_count", + &ibs_config.max_cnt_fetch); + oprofilefs_create_ulong(sb, dir, "rand_enable", + &ibs_config.rand_en); + } + + if (ibs_caps & IBS_CAPS_OPSAM) { + dir = oprofilefs_mkdir(sb, root, "ibs_op"); + oprofilefs_create_ulong(sb, dir, "enable", + &ibs_config.op_enabled); + oprofilefs_create_ulong(sb, dir, "max_count", + &ibs_config.max_cnt_op); + if (ibs_caps & IBS_CAPS_OPCNT) + oprofilefs_create_ulong(sb, dir, "dispatched_ops", + &ibs_config.dispatched_ops); + if (ibs_caps & IBS_CAPS_BRNTRGT) + oprofilefs_create_ulong(sb, dir, "branch_target", + &ibs_config.branch_target); + } return 0; } |