diff options
Diffstat (limited to 'arch/powerpc/perf')
-rw-r--r-- | arch/powerpc/perf/core-book3s.c | 40 | ||||
-rw-r--r-- | arch/powerpc/perf/imc-pmu.c | 108 | ||||
-rw-r--r-- | arch/powerpc/perf/isa207-common.c | 12 | ||||
-rw-r--r-- | arch/powerpc/perf/isa207-common.h | 5 | ||||
-rw-r--r-- | arch/powerpc/perf/power9-pmu.c | 54 | ||||
-rw-r--r-- | arch/powerpc/perf/req-gen/_begin.h | 2 | ||||
-rw-r--r-- | arch/powerpc/perf/req-gen/perf.h | 1 |
7 files changed, 63 insertions, 159 deletions
diff --git a/arch/powerpc/perf/core-book3s.c b/arch/powerpc/perf/core-book3s.c index 3f66fcf8ad99..81f8a0c838ae 100644 --- a/arch/powerpc/perf/core-book3s.c +++ b/arch/powerpc/perf/core-book3s.c @@ -128,10 +128,6 @@ static inline void power_pmu_bhrb_disable(struct perf_event *event) {} static void power_pmu_sched_task(struct perf_event_context *ctx, bool sched_in) {} static inline void power_pmu_bhrb_read(struct cpu_hw_events *cpuhw) {} static void pmao_restore_workaround(bool ebb) { } -static bool use_ic(u64 event) -{ - return false; -} #endif /* CONFIG_PPC32 */ static bool regs_use_siar(struct pt_regs *regs) @@ -714,14 +710,6 @@ static void pmao_restore_workaround(bool ebb) mtspr(SPRN_PMC6, pmcs[5]); } -static bool use_ic(u64 event) -{ - if (cpu_has_feature(CPU_FTR_POWER9_DD1) && - (event == 0x200f2 || event == 0x300f2)) - return true; - - return false; -} #endif /* CONFIG_PPC64 */ static void perf_event_interrupt(struct pt_regs *regs); @@ -1046,7 +1034,6 @@ static u64 check_and_compute_delta(u64 prev, u64 val) static void power_pmu_read(struct perf_event *event) { s64 val, delta, prev; - struct cpu_hw_events *cpuhw = this_cpu_ptr(&cpu_hw_events); if (event->hw.state & PERF_HES_STOPPED) return; @@ -1056,13 +1043,6 @@ static void power_pmu_read(struct perf_event *event) if (is_ebb_event(event)) { val = read_pmc(event->hw.idx); - if (use_ic(event->attr.config)) { - val = mfspr(SPRN_IC); - if (val > cpuhw->ic_init) - val = val - cpuhw->ic_init; - else - val = val + (0 - cpuhw->ic_init); - } local64_set(&event->hw.prev_count, val); return; } @@ -1076,13 +1056,6 @@ static void power_pmu_read(struct perf_event *event) prev = local64_read(&event->hw.prev_count); barrier(); val = read_pmc(event->hw.idx); - if (use_ic(event->attr.config)) { - val = mfspr(SPRN_IC); - if (val > cpuhw->ic_init) - val = val - cpuhw->ic_init; - else - val = val + (0 - cpuhw->ic_init); - } delta = check_and_compute_delta(prev, val); if (!delta) return; @@ -1469,7 +1442,7 @@ static int collect_events(struct perf_event *group, int max_count, } /* - * Add a event to the PMU. + * Add an event to the PMU. * If all events are not already frozen, then we disable and * re-enable the PMU in order to get hw_perf_enable to do the * actual work of reconfiguring the PMU. @@ -1535,20 +1508,13 @@ nocheck: event->attr.branch_sample_type); } - /* - * Workaround for POWER9 DD1 to use the Instruction Counter - * register value for instruction counting - */ - if (use_ic(event->attr.config)) - cpuhw->ic_init = mfspr(SPRN_IC); - perf_pmu_enable(event->pmu); local_irq_restore(flags); return ret; } /* - * Remove a event from the PMU. + * Remove an event from the PMU. */ static void power_pmu_del(struct perf_event *event, int ef_flags) { @@ -1742,7 +1708,7 @@ static int power_pmu_commit_txn(struct pmu *pmu) /* * Return 1 if we might be able to put event on a limited PMC, * or 0 if not. - * A event can only go on a limited PMC if it counts something + * An event can only go on a limited PMC if it counts something * that a limited PMC can count, doesn't require interrupts, and * doesn't exclude any processor mode. */ diff --git a/arch/powerpc/perf/imc-pmu.c b/arch/powerpc/perf/imc-pmu.c index d1977b61f827..1fafc32b12a0 100644 --- a/arch/powerpc/perf/imc-pmu.c +++ b/arch/powerpc/perf/imc-pmu.c @@ -867,59 +867,6 @@ static int thread_imc_cpu_init(void) ppc_thread_imc_cpu_offline); } -void thread_imc_pmu_sched_task(struct perf_event_context *ctx, - bool sched_in) -{ - int core_id; - struct imc_pmu_ref *ref; - - if (!is_core_imc_mem_inited(smp_processor_id())) - return; - - core_id = smp_processor_id() / threads_per_core; - /* - * imc pmus are enabled only when it is used. - * See if this is triggered for the first time. - * If yes, take the mutex lock and enable the counters. - * If not, just increment the count in ref count struct. - */ - ref = &core_imc_refc[core_id]; - if (!ref) - return; - - if (sched_in) { - mutex_lock(&ref->lock); - if (ref->refc == 0) { - if (opal_imc_counters_start(OPAL_IMC_COUNTERS_CORE, - get_hard_smp_processor_id(smp_processor_id()))) { - mutex_unlock(&ref->lock); - pr_err("thread-imc: Unable to start the counter\ - for core %d\n", core_id); - return; - } - } - ++ref->refc; - mutex_unlock(&ref->lock); - } else { - mutex_lock(&ref->lock); - ref->refc--; - if (ref->refc == 0) { - if (opal_imc_counters_stop(OPAL_IMC_COUNTERS_CORE, - get_hard_smp_processor_id(smp_processor_id()))) { - mutex_unlock(&ref->lock); - pr_err("thread-imc: Unable to stop the counters\ - for core %d\n", core_id); - return; - } - } else if (ref->refc < 0) { - ref->refc = 0; - } - mutex_unlock(&ref->lock); - } - - return; -} - static int thread_imc_event_init(struct perf_event *event) { u32 config = event->attr.config; @@ -1046,22 +993,70 @@ static int imc_event_add(struct perf_event *event, int flags) static int thread_imc_event_add(struct perf_event *event, int flags) { + int core_id; + struct imc_pmu_ref *ref; + if (flags & PERF_EF_START) imc_event_start(event, flags); - /* Enable the sched_task to start the engine */ - perf_sched_cb_inc(event->ctx->pmu); + if (!is_core_imc_mem_inited(smp_processor_id())) + return -EINVAL; + + core_id = smp_processor_id() / threads_per_core; + /* + * imc pmus are enabled only when it is used. + * See if this is triggered for the first time. + * If yes, take the mutex lock and enable the counters. + * If not, just increment the count in ref count struct. + */ + ref = &core_imc_refc[core_id]; + if (!ref) + return -EINVAL; + + mutex_lock(&ref->lock); + if (ref->refc == 0) { + if (opal_imc_counters_start(OPAL_IMC_COUNTERS_CORE, + get_hard_smp_processor_id(smp_processor_id()))) { + mutex_unlock(&ref->lock); + pr_err("thread-imc: Unable to start the counter\ + for core %d\n", core_id); + return -EINVAL; + } + } + ++ref->refc; + mutex_unlock(&ref->lock); return 0; } static void thread_imc_event_del(struct perf_event *event, int flags) { + + int core_id; + struct imc_pmu_ref *ref; + /* * Take a snapshot and calculate the delta and update * the event counter values. */ imc_event_update(event); - perf_sched_cb_dec(event->ctx->pmu); + + core_id = smp_processor_id() / threads_per_core; + ref = &core_imc_refc[core_id]; + + mutex_lock(&ref->lock); + ref->refc--; + if (ref->refc == 0) { + if (opal_imc_counters_stop(OPAL_IMC_COUNTERS_CORE, + get_hard_smp_processor_id(smp_processor_id()))) { + mutex_unlock(&ref->lock); + pr_err("thread-imc: Unable to stop the counters\ + for core %d\n", core_id); + return; + } + } else if (ref->refc < 0) { + ref->refc = 0; + } + mutex_unlock(&ref->lock); } /* update_pmu_ops : Populate the appropriate operations for "pmu" */ @@ -1087,7 +1082,6 @@ static int update_pmu_ops(struct imc_pmu *pmu) break; case IMC_DOMAIN_THREAD: pmu->pmu.event_init = thread_imc_event_init; - pmu->pmu.sched_task = thread_imc_pmu_sched_task; pmu->pmu.add = thread_imc_event_add; pmu->pmu.del = thread_imc_event_del; pmu->pmu.start_txn = thread_imc_pmu_start_txn; diff --git a/arch/powerpc/perf/isa207-common.c b/arch/powerpc/perf/isa207-common.c index 2efee3f196f5..177de814286f 100644 --- a/arch/powerpc/perf/isa207-common.c +++ b/arch/powerpc/perf/isa207-common.c @@ -59,7 +59,7 @@ static bool is_event_valid(u64 event) { u64 valid_mask = EVENT_VALID_MASK; - if (cpu_has_feature(CPU_FTR_ARCH_300) && !cpu_has_feature(CPU_FTR_POWER9_DD1)) + if (cpu_has_feature(CPU_FTR_ARCH_300)) valid_mask = p9_EVENT_VALID_MASK; return !(event & ~valid_mask); @@ -86,8 +86,6 @@ static void mmcra_sdar_mode(u64 event, unsigned long *mmcra) * Incase of Power9: * Marked event: MMCRA[SDAR_MODE] will be set to 0b00 ('No Updates'), * or if group already have any marked events. - * Non-Marked events (for DD1): - * MMCRA[SDAR_MODE] will be set to 0b01 * For rest * MMCRA[SDAR_MODE] will be set from event code. * If sdar_mode from event is zero, default to 0b01. Hardware @@ -96,7 +94,7 @@ static void mmcra_sdar_mode(u64 event, unsigned long *mmcra) if (cpu_has_feature(CPU_FTR_ARCH_300)) { if (is_event_marked(event) || (*mmcra & MMCRA_SAMPLE_ENABLE)) *mmcra &= MMCRA_SDAR_MODE_NO_UPDATES; - else if (!cpu_has_feature(CPU_FTR_POWER9_DD1) && p9_SDAR_MODE(event)) + else if (p9_SDAR_MODE(event)) *mmcra |= p9_SDAR_MODE(event) << MMCRA_SDAR_MODE_SHIFT; else *mmcra |= MMCRA_SDAR_MODE_DCACHE; @@ -106,7 +104,7 @@ static void mmcra_sdar_mode(u64 event, unsigned long *mmcra) static u64 thresh_cmp_val(u64 value) { - if (cpu_has_feature(CPU_FTR_ARCH_300) && !cpu_has_feature(CPU_FTR_POWER9_DD1)) + if (cpu_has_feature(CPU_FTR_ARCH_300)) return value << p9_MMCRA_THR_CMP_SHIFT; return value << MMCRA_THR_CMP_SHIFT; @@ -114,7 +112,7 @@ static u64 thresh_cmp_val(u64 value) static unsigned long combine_from_event(u64 event) { - if (cpu_has_feature(CPU_FTR_ARCH_300) && !cpu_has_feature(CPU_FTR_POWER9_DD1)) + if (cpu_has_feature(CPU_FTR_ARCH_300)) return p9_EVENT_COMBINE(event); return EVENT_COMBINE(event); @@ -122,7 +120,7 @@ static unsigned long combine_from_event(u64 event) static unsigned long combine_shift(unsigned long pmc) { - if (cpu_has_feature(CPU_FTR_ARCH_300) && !cpu_has_feature(CPU_FTR_POWER9_DD1)) + if (cpu_has_feature(CPU_FTR_ARCH_300)) return p9_MMCR1_COMBINE_SHIFT(pmc); return MMCR1_COMBINE_SHIFT(pmc); diff --git a/arch/powerpc/perf/isa207-common.h b/arch/powerpc/perf/isa207-common.h index 6a0b586c935a..0028f4b9490d 100644 --- a/arch/powerpc/perf/isa207-common.h +++ b/arch/powerpc/perf/isa207-common.h @@ -158,11 +158,6 @@ CNST_PMC_VAL(1) | CNST_PMC_VAL(2) | CNST_PMC_VAL(3) | \ CNST_PMC_VAL(4) | CNST_PMC_VAL(5) | CNST_PMC_VAL(6) | CNST_NC_VAL -/* - * Lets restrict use of PMC5 for instruction counting. - */ -#define P9_DD1_TEST_ADDER (ISA207_TEST_ADDER | CNST_PMC_VAL(5)) - /* Bits in MMCR1 for PowerISA v2.07 */ #define MMCR1_UNIT_SHIFT(pmc) (60 - (4 * ((pmc) - 1))) #define MMCR1_COMBINE_SHIFT(pmc) (35 - ((pmc) - 1)) diff --git a/arch/powerpc/perf/power9-pmu.c b/arch/powerpc/perf/power9-pmu.c index 2ca0b33b4efb..e012b1030a5b 100644 --- a/arch/powerpc/perf/power9-pmu.c +++ b/arch/powerpc/perf/power9-pmu.c @@ -219,12 +219,6 @@ static struct attribute_group power9_pmu_events_group = { .attrs = power9_events_attr, }; -static const struct attribute_group *power9_isa207_pmu_attr_groups[] = { - &isa207_pmu_format_group, - &power9_pmu_events_group, - NULL, -}; - PMU_FORMAT_ATTR(event, "config:0-51"); PMU_FORMAT_ATTR(pmcxsel, "config:0-7"); PMU_FORMAT_ATTR(mark, "config:8"); @@ -267,17 +261,6 @@ static const struct attribute_group *power9_pmu_attr_groups[] = { NULL, }; -static int power9_generic_events_dd1[] = { - [PERF_COUNT_HW_CPU_CYCLES] = PM_CYC, - [PERF_COUNT_HW_STALLED_CYCLES_FRONTEND] = PM_ICT_NOSLOT_CYC, - [PERF_COUNT_HW_STALLED_CYCLES_BACKEND] = PM_CMPLU_STALL, - [PERF_COUNT_HW_INSTRUCTIONS] = PM_INST_DISP, - [PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = PM_BR_CMPL_ALT, - [PERF_COUNT_HW_BRANCH_MISSES] = PM_BR_MPRED_CMPL, - [PERF_COUNT_HW_CACHE_REFERENCES] = PM_LD_REF_L1, - [PERF_COUNT_HW_CACHE_MISSES] = PM_LD_MISS_L1_FIN, -}; - static int power9_generic_events[] = { [PERF_COUNT_HW_CPU_CYCLES] = PM_CYC, [PERF_COUNT_HW_STALLED_CYCLES_FRONTEND] = PM_ICT_NOSLOT_CYC, @@ -439,25 +422,6 @@ static int power9_cache_events[C(MAX)][C(OP_MAX)][C(RESULT_MAX)] = { #undef C -static struct power_pmu power9_isa207_pmu = { - .name = "POWER9", - .n_counter = MAX_PMU_COUNTERS, - .add_fields = ISA207_ADD_FIELDS, - .test_adder = P9_DD1_TEST_ADDER, - .compute_mmcr = isa207_compute_mmcr, - .config_bhrb = power9_config_bhrb, - .bhrb_filter_map = power9_bhrb_filter_map, - .get_constraint = isa207_get_constraint, - .get_alternatives = power9_get_alternatives, - .disable_pmc = isa207_disable_pmc, - .flags = PPMU_NO_SIAR | PPMU_ARCH_207S, - .n_generic = ARRAY_SIZE(power9_generic_events_dd1), - .generic_events = power9_generic_events_dd1, - .cache_events = &power9_cache_events, - .attr_groups = power9_isa207_pmu_attr_groups, - .bhrb_nr = 32, -}; - static struct power_pmu power9_pmu = { .name = "POWER9", .n_counter = MAX_PMU_COUNTERS, @@ -500,23 +464,7 @@ static int __init init_power9_pmu(void) } } - if (cpu_has_feature(CPU_FTR_POWER9_DD1)) { - /* - * Since PM_INST_CMPL may not provide right counts in all - * sampling scenarios in power9 DD1, instead use PM_INST_DISP. - */ - EVENT_VAR(PM_INST_CMPL, _g).id = PM_INST_DISP; - /* - * Power9 DD1 should use PM_BR_CMPL_ALT event code for - * "branches" to provide correct counter value. - */ - EVENT_VAR(PM_BR_CMPL, _g).id = PM_BR_CMPL_ALT; - EVENT_VAR(PM_BR_CMPL, _c).id = PM_BR_CMPL_ALT; - rc = register_power_pmu(&power9_isa207_pmu); - } else { - rc = register_power_pmu(&power9_pmu); - } - + rc = register_power_pmu(&power9_pmu); if (rc) return rc; diff --git a/arch/powerpc/perf/req-gen/_begin.h b/arch/powerpc/perf/req-gen/_begin.h index 549f8782c52d..a200b86eba3b 100644 --- a/arch/powerpc/perf/req-gen/_begin.h +++ b/arch/powerpc/perf/req-gen/_begin.h @@ -3,6 +3,8 @@ #ifndef POWERPC_PERF_REQ_GEN_H_ #define POWERPC_PERF_REQ_GEN_H_ +#include <linux/stringify.h> + #define CAT2_STR_(t, s) __stringify(t/s) #define CAT2_STR(t, s) CAT2_STR_(t, s) #define I(...) __VA_ARGS__ diff --git a/arch/powerpc/perf/req-gen/perf.h b/arch/powerpc/perf/req-gen/perf.h index 871a9a1766c2..fa9bc804e67a 100644 --- a/arch/powerpc/perf/req-gen/perf.h +++ b/arch/powerpc/perf/req-gen/perf.h @@ -3,6 +3,7 @@ #define LINUX_POWERPC_PERF_REQ_GEN_PERF_H_ #include <linux/perf_event.h> +#include <linux/stringify.h> #ifndef REQUEST_FILE #error "REQUEST_FILE must be defined before including" |