diff options
Diffstat (limited to 'tools/perf/arch/x86/util')
-rw-r--r-- | tools/perf/arch/x86/util/event.c | 46 | ||||
-rw-r--r-- | tools/perf/arch/x86/util/evlist.c | 24 | ||||
-rw-r--r-- | tools/perf/arch/x86/util/evsel.c | 46 | ||||
-rw-r--r-- | tools/perf/arch/x86/util/topdown.c | 59 | ||||
-rw-r--r-- | tools/perf/arch/x86/util/topdown.h | 6 |
5 files changed, 80 insertions, 101 deletions
diff --git a/tools/perf/arch/x86/util/event.c b/tools/perf/arch/x86/util/event.c index a0400707180c..3cd384317739 100644 --- a/tools/perf/arch/x86/util/event.c +++ b/tools/perf/arch/x86/util/event.c @@ -91,49 +91,3 @@ int perf_event__synthesize_extra_kmaps(const struct perf_tool *tool, } #endif - -void arch_perf_parse_sample_weight(struct perf_sample *data, - const __u64 *array, u64 type) -{ - union perf_sample_weight weight; - - weight.full = *array; - if (type & PERF_SAMPLE_WEIGHT) - data->weight = weight.full; - else { - data->weight = weight.var1_dw; - data->ins_lat = weight.var2_w; - data->retire_lat = weight.var3_w; - } -} - -void arch_perf_synthesize_sample_weight(const struct perf_sample *data, - __u64 *array, u64 type) -{ - *array = data->weight; - - if (type & PERF_SAMPLE_WEIGHT_STRUCT) { - *array &= 0xffffffff; - *array |= ((u64)data->ins_lat << 32); - *array |= ((u64)data->retire_lat << 48); - } -} - -const char *arch_perf_header_entry(const char *se_header) -{ - if (!strcmp(se_header, "Local Pipeline Stage Cycle")) - return "Local Retire Latency"; - else if (!strcmp(se_header, "Pipeline Stage Cycle")) - return "Retire Latency"; - - return se_header; -} - -int arch_support_sort_key(const char *sort_key) -{ - if (!strcmp(sort_key, "p_stage_cyc")) - return 1; - if (!strcmp(sort_key, "local_p_stage_cyc")) - return 1; - return 0; -} diff --git a/tools/perf/arch/x86/util/evlist.c b/tools/perf/arch/x86/util/evlist.c index 1969758cc8c1..75e9d00a1494 100644 --- a/tools/perf/arch/x86/util/evlist.c +++ b/tools/perf/arch/x86/util/evlist.c @@ -81,3 +81,27 @@ int arch_evlist__cmp(const struct evsel *lhs, const struct evsel *rhs) /* Default ordering by insertion index. */ return lhs->core.idx - rhs->core.idx; } + +int arch_evlist__add_required_events(struct list_head *list) +{ + struct evsel *pos, *metric_event = NULL; + int idx = 0; + + if (!topdown_sys_has_perf_metrics()) + return 0; + + list_for_each_entry(pos, list, core.node) { + if (arch_is_topdown_slots(pos)) { + /* Slots event already present, nothing to do. */ + return 0; + } + if (metric_event == NULL && arch_is_topdown_metrics(pos)) + metric_event = pos; + idx++; + } + if (metric_event == NULL) { + /* No topdown metric events, nothing to do. */ + return 0; + } + return topdown_insert_slots_event(list, idx + 1, metric_event); +} diff --git a/tools/perf/arch/x86/util/evsel.c b/tools/perf/arch/x86/util/evsel.c index 3dd29ba2c23b..9bc80fff3aa0 100644 --- a/tools/perf/arch/x86/util/evsel.c +++ b/tools/perf/arch/x86/util/evsel.c @@ -23,47 +23,25 @@ void arch_evsel__set_sample_weight(struct evsel *evsel) bool evsel__sys_has_perf_metrics(const struct evsel *evsel) { struct perf_pmu *pmu; - u32 type = evsel->core.attr.type; - /* - * The PERF_TYPE_RAW type is the core PMU type, e.g., "cpu" PMU - * on a non-hybrid machine, "cpu_core" PMU on a hybrid machine. - * The slots event is only available for the core PMU, which - * supports the perf metrics feature. - * Checking both the PERF_TYPE_RAW type and the slots event - * should be good enough to detect the perf metrics feature. - */ -again: - switch (type) { - case PERF_TYPE_HARDWARE: - case PERF_TYPE_HW_CACHE: - type = evsel->core.attr.config >> PERF_PMU_TYPE_SHIFT; - if (type) - goto again; - break; - case PERF_TYPE_RAW: - break; - default: + if (!topdown_sys_has_perf_metrics()) return false; - } - - pmu = evsel->pmu; - if (pmu && perf_pmu__is_fake(pmu)) - pmu = NULL; - if (!pmu) { - while ((pmu = perf_pmus__scan_core(pmu)) != NULL) { - if (pmu->type == PERF_TYPE_RAW) - break; - } - } - return pmu && perf_pmu__have_event(pmu, "slots"); + /* + * The PERF_TYPE_RAW type is the core PMU type, e.g., "cpu" PMU on a + * non-hybrid machine, "cpu_core" PMU on a hybrid machine. The + * topdown_sys_has_perf_metrics checks the slots event is only available + * for the core PMU, which supports the perf metrics feature. Checking + * both the PERF_TYPE_RAW type and the slots event should be good enough + * to detect the perf metrics feature. + */ + pmu = evsel__find_pmu(evsel); + return pmu && pmu->type == PERF_TYPE_RAW; } bool arch_evsel__must_be_in_group(const struct evsel *evsel) { - if (!evsel__sys_has_perf_metrics(evsel) || !evsel->name || - strcasestr(evsel->name, "uops_retired.slots")) + if (!evsel__sys_has_perf_metrics(evsel)) return false; return arch_is_topdown_metrics(evsel) || arch_is_topdown_slots(evsel); diff --git a/tools/perf/arch/x86/util/topdown.c b/tools/perf/arch/x86/util/topdown.c index d1c654839049..0d01b662627a 100644 --- a/tools/perf/arch/x86/util/topdown.c +++ b/tools/perf/arch/x86/util/topdown.c @@ -1,6 +1,4 @@ // SPDX-License-Identifier: GPL-2.0 -#include "api/fs/fs.h" -#include "util/evsel.h" #include "util/evlist.h" #include "util/pmu.h" #include "util/pmus.h" @@ -8,6 +6,9 @@ #include "topdown.h" #include "evsel.h" +// cmask=0, inv=0, pc=0, edge=0, umask=4, event=0 +#define TOPDOWN_SLOTS 0x0400 + /* Check whether there is a PMU which supports the perf metrics. */ bool topdown_sys_has_perf_metrics(void) { @@ -32,31 +33,19 @@ bool topdown_sys_has_perf_metrics(void) return has_perf_metrics; } -#define TOPDOWN_SLOTS 0x0400 bool arch_is_topdown_slots(const struct evsel *evsel) { - if (evsel->core.attr.config == TOPDOWN_SLOTS) - return true; - - return false; + return evsel->core.attr.type == PERF_TYPE_RAW && + evsel->core.attr.config == TOPDOWN_SLOTS && + evsel->core.attr.config1 == 0; } bool arch_is_topdown_metrics(const struct evsel *evsel) { - int config = evsel->core.attr.config; - const char *name_from_config; - struct perf_pmu *pmu; - - /* All topdown events have an event code of 0. */ - if ((config & 0xFF) != 0) - return false; - - pmu = evsel__find_pmu(evsel); - if (!pmu || !pmu->is_core) - return false; - - name_from_config = perf_pmu__name_from_config(pmu, config); - return name_from_config && strcasestr(name_from_config, "topdown"); + // cmask=0, inv=0, pc=0, edge=0, umask=0x80-0x87, event=0 + return evsel->core.attr.type == PERF_TYPE_RAW && + (evsel->core.attr.config & 0xFFFFF8FF) == 0x8000 && + evsel->core.attr.config1 == 0; } /* @@ -88,3 +77,31 @@ bool arch_topdown_sample_read(struct evsel *leader) return false; } + +/* + * Make a copy of the topdown metric event metric_event with the given index but + * change its configuration to be a topdown slots event. Copying from + * metric_event ensures modifiers are the same. + */ +int topdown_insert_slots_event(struct list_head *list, int idx, struct evsel *metric_event) +{ + struct evsel *evsel = evsel__new_idx(&metric_event->core.attr, idx); + + if (!evsel) + return -ENOMEM; + + evsel->core.attr.config = TOPDOWN_SLOTS; + evsel->core.cpus = perf_cpu_map__get(metric_event->core.cpus); + evsel->core.pmu_cpus = perf_cpu_map__get(metric_event->core.pmu_cpus); + evsel->core.is_pmu_core = true; + evsel->pmu = metric_event->pmu; + evsel->name = strdup("slots"); + evsel->precise_max = metric_event->precise_max; + evsel->sample_read = metric_event->sample_read; + evsel->weak_group = metric_event->weak_group; + evsel->bpf_counter = metric_event->bpf_counter; + evsel->retire_lat = metric_event->retire_lat; + evsel__set_leader(evsel, evsel__leader(metric_event)); + list_add_tail(&evsel->core.node, list); + return 0; +} diff --git a/tools/perf/arch/x86/util/topdown.h b/tools/perf/arch/x86/util/topdown.h index 1bae9b1822d7..69035565e649 100644 --- a/tools/perf/arch/x86/util/topdown.h +++ b/tools/perf/arch/x86/util/topdown.h @@ -2,8 +2,14 @@ #ifndef _TOPDOWN_H #define _TOPDOWN_H 1 +#include <stdbool.h> + +struct evsel; +struct list_head; + bool topdown_sys_has_perf_metrics(void); bool arch_is_topdown_slots(const struct evsel *evsel); bool arch_is_topdown_metrics(const struct evsel *evsel); +int topdown_insert_slots_event(struct list_head *list, int idx, struct evsel *metric_event); #endif |