summaryrefslogtreecommitdiffstats
path: root/tools/perf/util/intel-pt.c
diff options
context:
space:
mode:
Diffstat (limited to 'tools/perf/util/intel-pt.c')
-rw-r--r--tools/perf/util/intel-pt.c266
1 files changed, 237 insertions, 29 deletions
diff --git a/tools/perf/util/intel-pt.c b/tools/perf/util/intel-pt.c
index 535d86f8e4d1..97f963a3dcb9 100644
--- a/tools/perf/util/intel-pt.c
+++ b/tools/perf/util/intel-pt.c
@@ -22,6 +22,7 @@
#include "../perf.h"
#include "session.h"
#include "machine.h"
+#include "sort.h"
#include "tool.h"
#include "event.h"
#include "evlist.h"
@@ -63,6 +64,7 @@ struct intel_pt {
bool data_queued;
bool est_tsc;
bool sync_switch;
+ bool mispred_all;
int have_sched_switch;
u32 pmu_type;
u64 kernel_start;
@@ -115,6 +117,9 @@ struct intel_pt_queue {
void *decoder;
const struct intel_pt_state *state;
struct ip_callchain *chain;
+ struct branch_stack *last_branch;
+ struct branch_stack *last_branch_rb;
+ size_t last_branch_pos;
union perf_event *event_buf;
bool on_heap;
bool stop;
@@ -675,6 +680,19 @@ static struct intel_pt_queue *intel_pt_alloc_queue(struct intel_pt *pt,
goto out_free;
}
+ if (pt->synth_opts.last_branch) {
+ size_t sz = sizeof(struct branch_stack);
+
+ sz += pt->synth_opts.last_branch_sz *
+ sizeof(struct branch_entry);
+ ptq->last_branch = zalloc(sz);
+ if (!ptq->last_branch)
+ goto out_free;
+ ptq->last_branch_rb = zalloc(sz);
+ if (!ptq->last_branch_rb)
+ goto out_free;
+ }
+
ptq->event_buf = malloc(PERF_SAMPLE_MAX_SIZE);
if (!ptq->event_buf)
goto out_free;
@@ -720,7 +738,7 @@ static struct intel_pt_queue *intel_pt_alloc_queue(struct intel_pt *pt,
if (!params.period) {
params.period_type = INTEL_PT_PERIOD_INSTRUCTIONS;
- params.period = 1000;
+ params.period = 1;
}
}
@@ -732,6 +750,8 @@ static struct intel_pt_queue *intel_pt_alloc_queue(struct intel_pt *pt,
out_free:
zfree(&ptq->event_buf);
+ zfree(&ptq->last_branch);
+ zfree(&ptq->last_branch_rb);
zfree(&ptq->chain);
free(ptq);
return NULL;
@@ -746,6 +766,8 @@ static void intel_pt_free_queue(void *priv)
thread__zput(ptq->thread);
intel_pt_decoder_free(ptq->decoder);
zfree(&ptq->event_buf);
+ zfree(&ptq->last_branch);
+ zfree(&ptq->last_branch_rb);
zfree(&ptq->chain);
free(ptq);
}
@@ -876,6 +898,58 @@ static int intel_pt_setup_queues(struct intel_pt *pt)
return 0;
}
+static inline void intel_pt_copy_last_branch_rb(struct intel_pt_queue *ptq)
+{
+ struct branch_stack *bs_src = ptq->last_branch_rb;
+ struct branch_stack *bs_dst = ptq->last_branch;
+ size_t nr = 0;
+
+ bs_dst->nr = bs_src->nr;
+
+ if (!bs_src->nr)
+ return;
+
+ nr = ptq->pt->synth_opts.last_branch_sz - ptq->last_branch_pos;
+ memcpy(&bs_dst->entries[0],
+ &bs_src->entries[ptq->last_branch_pos],
+ sizeof(struct branch_entry) * nr);
+
+ if (bs_src->nr >= ptq->pt->synth_opts.last_branch_sz) {
+ memcpy(&bs_dst->entries[nr],
+ &bs_src->entries[0],
+ sizeof(struct branch_entry) * ptq->last_branch_pos);
+ }
+}
+
+static inline void intel_pt_reset_last_branch_rb(struct intel_pt_queue *ptq)
+{
+ ptq->last_branch_pos = 0;
+ ptq->last_branch_rb->nr = 0;
+}
+
+static void intel_pt_update_last_branch_rb(struct intel_pt_queue *ptq)
+{
+ const struct intel_pt_state *state = ptq->state;
+ struct branch_stack *bs = ptq->last_branch_rb;
+ struct branch_entry *be;
+
+ if (!ptq->last_branch_pos)
+ ptq->last_branch_pos = ptq->pt->synth_opts.last_branch_sz;
+
+ ptq->last_branch_pos -= 1;
+
+ be = &bs->entries[ptq->last_branch_pos];
+ be->from = state->from_ip;
+ be->to = state->to_ip;
+ be->flags.abort = !!(state->flags & INTEL_PT_ABORT_TX);
+ be->flags.in_tx = !!(state->flags & INTEL_PT_IN_TX);
+ /* No support for mispredict */
+ be->flags.mispred = ptq->pt->mispred_all;
+
+ if (bs->nr < ptq->pt->synth_opts.last_branch_sz)
+ bs->nr += 1;
+}
+
static int intel_pt_inject_event(union perf_event *event,
struct perf_sample *sample, u64 type,
bool swapped)
@@ -890,6 +964,13 @@ static int intel_pt_synth_branch_sample(struct intel_pt_queue *ptq)
struct intel_pt *pt = ptq->pt;
union perf_event *event = ptq->event_buf;
struct perf_sample sample = { .ip = 0, };
+ struct dummy_branch_stack {
+ u64 nr;
+ struct branch_entry entries;
+ } dummy_bs;
+
+ if (pt->branches_filter && !(pt->branches_filter & ptq->flags))
+ return 0;
event->sample.header.type = PERF_RECORD_SAMPLE;
event->sample.header.misc = PERF_RECORD_MISC_USER;
@@ -909,8 +990,20 @@ static int intel_pt_synth_branch_sample(struct intel_pt_queue *ptq)
sample.flags = ptq->flags;
sample.insn_len = ptq->insn_len;
- if (pt->branches_filter && !(pt->branches_filter & ptq->flags))
- return 0;
+ /*
+ * perf report cannot handle events without a branch stack when using
+ * SORT_MODE__BRANCH so make a dummy one.
+ */
+ if (pt->synth_opts.last_branch && sort__mode == SORT_MODE__BRANCH) {
+ dummy_bs = (struct dummy_branch_stack){
+ .nr = 1,
+ .entries = {
+ .from = sample.ip,
+ .to = sample.addr,
+ },
+ };
+ sample.branch_stack = (struct branch_stack *)&dummy_bs;
+ }
if (pt->synth_opts.inject) {
ret = intel_pt_inject_event(event, &sample,
@@ -961,6 +1054,11 @@ static int intel_pt_synth_instruction_sample(struct intel_pt_queue *ptq)
sample.callchain = ptq->chain;
}
+ if (pt->synth_opts.last_branch) {
+ intel_pt_copy_last_branch_rb(ptq);
+ sample.branch_stack = ptq->last_branch;
+ }
+
if (pt->synth_opts.inject) {
ret = intel_pt_inject_event(event, &sample,
pt->instructions_sample_type,
@@ -974,6 +1072,9 @@ static int intel_pt_synth_instruction_sample(struct intel_pt_queue *ptq)
pr_err("Intel Processor Trace: failed to deliver instruction event, error %d\n",
ret);
+ if (pt->synth_opts.last_branch)
+ intel_pt_reset_last_branch_rb(ptq);
+
return ret;
}
@@ -1008,6 +1109,11 @@ static int intel_pt_synth_transaction_sample(struct intel_pt_queue *ptq)
sample.callchain = ptq->chain;
}
+ if (pt->synth_opts.last_branch) {
+ intel_pt_copy_last_branch_rb(ptq);
+ sample.branch_stack = ptq->last_branch;
+ }
+
if (pt->synth_opts.inject) {
ret = intel_pt_inject_event(event, &sample,
pt->transactions_sample_type,
@@ -1021,6 +1127,9 @@ static int intel_pt_synth_transaction_sample(struct intel_pt_queue *ptq)
pr_err("Intel Processor Trace: failed to deliver transaction event, error %d\n",
ret);
+ if (pt->synth_opts.callchain)
+ intel_pt_reset_last_branch_rb(ptq);
+
return ret;
}
@@ -1116,6 +1225,9 @@ static int intel_pt_sample(struct intel_pt_queue *ptq)
return err;
}
+ if (pt->synth_opts.last_branch)
+ intel_pt_update_last_branch_rb(ptq);
+
if (!pt->sync_switch)
return 0;
@@ -1145,16 +1257,18 @@ static int intel_pt_sample(struct intel_pt_queue *ptq)
return 0;
}
-static u64 intel_pt_switch_ip(struct machine *machine, u64 *ptss_ip)
+static u64 intel_pt_switch_ip(struct intel_pt *pt, u64 *ptss_ip)
{
+ struct machine *machine = pt->machine;
struct map *map;
struct symbol *sym, *start;
u64 ip, switch_ip = 0;
+ const char *ptss;
if (ptss_ip)
*ptss_ip = 0;
- map = machine__kernel_map(machine, MAP__FUNCTION);
+ map = machine__kernel_map(machine);
if (!map)
return 0;
@@ -1177,8 +1291,13 @@ static u64 intel_pt_switch_ip(struct machine *machine, u64 *ptss_ip)
if (!switch_ip || !ptss_ip)
return 0;
+ if (pt->have_sched_switch == 1)
+ ptss = "perf_trace_sched_switch";
+ else
+ ptss = "__perf_event_task_sched_out";
+
for (sym = start; sym; sym = dso__next_symbol(sym)) {
- if (!strcmp(sym->name, "perf_trace_sched_switch")) {
+ if (!strcmp(sym->name, ptss)) {
ip = map->unmap_ip(map, sym->start);
if (ip >= map->start && ip < map->end) {
*ptss_ip = ip;
@@ -1198,11 +1317,11 @@ static int intel_pt_run_decoder(struct intel_pt_queue *ptq, u64 *timestamp)
if (!pt->kernel_start) {
pt->kernel_start = machine__kernel_start(pt->machine);
- if (pt->per_cpu_mmaps && pt->have_sched_switch &&
+ if (pt->per_cpu_mmaps &&
+ (pt->have_sched_switch == 1 || pt->have_sched_switch == 3) &&
!pt->timeless_decoding && intel_pt_tracing_kernel(pt) &&
!pt->sampling_mode) {
- pt->switch_ip = intel_pt_switch_ip(pt->machine,
- &pt->ptss_ip);
+ pt->switch_ip = intel_pt_switch_ip(pt, &pt->ptss_ip);
if (pt->switch_ip) {
intel_pt_log("switch_ip: %"PRIx64" ptss_ip: %"PRIx64"\n",
pt->switch_ip, pt->ptss_ip);
@@ -1387,31 +1506,18 @@ static struct intel_pt_queue *intel_pt_cpu_to_ptq(struct intel_pt *pt, int cpu)
return NULL;
}
-static int intel_pt_process_switch(struct intel_pt *pt,
- struct perf_sample *sample)
+static int intel_pt_sync_switch(struct intel_pt *pt, int cpu, pid_t tid,
+ u64 timestamp)
{
struct intel_pt_queue *ptq;
- struct perf_evsel *evsel;
- pid_t tid;
- int cpu, err;
-
- evsel = perf_evlist__id2evsel(pt->session->evlist, sample->id);
- if (evsel != pt->switch_evsel)
- return 0;
-
- tid = perf_evsel__intval(evsel, sample, "next_pid");
- cpu = sample->cpu;
-
- intel_pt_log("sched_switch: cpu %d tid %d time %"PRIu64" tsc %#"PRIx64"\n",
- cpu, tid, sample->time, perf_time_to_tsc(sample->time,
- &pt->tc));
+ int err;
if (!pt->sync_switch)
- goto out;
+ return 1;
ptq = intel_pt_cpu_to_ptq(pt, cpu);
if (!ptq)
- goto out;
+ return 1;
switch (ptq->switch_state) {
case INTEL_PT_SS_NOT_TRACING:
@@ -1424,7 +1530,7 @@ static int intel_pt_process_switch(struct intel_pt *pt,
return 0;
case INTEL_PT_SS_EXPECTING_SWITCH_EVENT:
if (!ptq->on_heap) {
- ptq->timestamp = perf_time_to_tsc(sample->time,
+ ptq->timestamp = perf_time_to_tsc(timestamp,
&pt->tc);
err = auxtrace_heap__add(&pt->heap, ptq->queue_nr,
ptq->timestamp);
@@ -1441,10 +1547,76 @@ static int intel_pt_process_switch(struct intel_pt *pt,
default:
break;
}
-out:
+
+ return 1;
+}
+
+static int intel_pt_process_switch(struct intel_pt *pt,
+ struct perf_sample *sample)
+{
+ struct perf_evsel *evsel;
+ pid_t tid;
+ int cpu, ret;
+
+ evsel = perf_evlist__id2evsel(pt->session->evlist, sample->id);
+ if (evsel != pt->switch_evsel)
+ return 0;
+
+ tid = perf_evsel__intval(evsel, sample, "next_pid");
+ cpu = sample->cpu;
+
+ intel_pt_log("sched_switch: cpu %d tid %d time %"PRIu64" tsc %#"PRIx64"\n",
+ cpu, tid, sample->time, perf_time_to_tsc(sample->time,
+ &pt->tc));
+
+ ret = intel_pt_sync_switch(pt, cpu, tid, sample->time);
+ if (ret <= 0)
+ return ret;
+
return machine__set_current_tid(pt->machine, cpu, -1, tid);
}
+static int intel_pt_context_switch(struct intel_pt *pt, union perf_event *event,
+ struct perf_sample *sample)
+{
+ bool out = event->header.misc & PERF_RECORD_MISC_SWITCH_OUT;
+ pid_t pid, tid;
+ int cpu, ret;
+
+ cpu = sample->cpu;
+
+ if (pt->have_sched_switch == 3) {
+ if (!out)
+ return 0;
+ if (event->header.type != PERF_RECORD_SWITCH_CPU_WIDE) {
+ pr_err("Expecting CPU-wide context switch event\n");
+ return -EINVAL;
+ }
+ pid = event->context_switch.next_prev_pid;
+ tid = event->context_switch.next_prev_tid;
+ } else {
+ if (out)
+ return 0;
+ pid = sample->pid;
+ tid = sample->tid;
+ }
+
+ if (tid == -1) {
+ pr_err("context_switch event has no tid\n");
+ return -EINVAL;
+ }
+
+ intel_pt_log("context_switch: cpu %d pid %d tid %d time %"PRIu64" tsc %#"PRIx64"\n",
+ cpu, pid, tid, sample->time, perf_time_to_tsc(sample->time,
+ &pt->tc));
+
+ ret = intel_pt_sync_switch(pt, cpu, tid, sample->time);
+ if (ret <= 0)
+ return ret;
+
+ return machine__set_current_tid(pt->machine, cpu, pid, tid);
+}
+
static int intel_pt_process_itrace_start(struct intel_pt *pt,
union perf_event *event,
struct perf_sample *sample)
@@ -1515,6 +1687,9 @@ static int intel_pt_process_event(struct perf_session *session,
err = intel_pt_process_switch(pt, sample);
else if (event->header.type == PERF_RECORD_ITRACE_START)
err = intel_pt_process_itrace_start(pt, event, sample);
+ else if (event->header.type == PERF_RECORD_SWITCH ||
+ event->header.type == PERF_RECORD_SWITCH_CPU_WIDE)
+ err = intel_pt_context_switch(pt, event, sample);
intel_pt_log("event %s (%u): cpu %d time %"PRIu64" tsc %#"PRIx64"\n",
perf_event__name(event->header.type), event->header.type,
@@ -1700,6 +1875,8 @@ static int intel_pt_synth_events(struct intel_pt *pt,
pt->instructions_sample_period = attr.sample_period;
if (pt->synth_opts.callchain)
attr.sample_type |= PERF_SAMPLE_CALLCHAIN;
+ if (pt->synth_opts.last_branch)
+ attr.sample_type |= PERF_SAMPLE_BRANCH_STACK;
pr_debug("Synthesizing 'instructions' event with id %" PRIu64 " sample type %#" PRIx64 "\n",
id, (u64)attr.sample_type);
err = intel_pt_synth_event(session, &attr, id);
@@ -1719,6 +1896,8 @@ static int intel_pt_synth_events(struct intel_pt *pt,
attr.sample_period = 1;
if (pt->synth_opts.callchain)
attr.sample_type |= PERF_SAMPLE_CALLCHAIN;
+ if (pt->synth_opts.last_branch)
+ attr.sample_type |= PERF_SAMPLE_BRANCH_STACK;
pr_debug("Synthesizing 'transactions' event with id %" PRIu64 " sample type %#" PRIx64 "\n",
id, (u64)attr.sample_type);
err = intel_pt_synth_event(session, &attr, id);
@@ -1745,6 +1924,7 @@ static int intel_pt_synth_events(struct intel_pt *pt,
attr.sample_period = 1;
attr.sample_type |= PERF_SAMPLE_ADDR;
attr.sample_type &= ~(u64)PERF_SAMPLE_CALLCHAIN;
+ attr.sample_type &= ~(u64)PERF_SAMPLE_BRANCH_STACK;
pr_debug("Synthesizing 'branches' event with id %" PRIu64 " sample type %#" PRIx64 "\n",
id, (u64)attr.sample_type);
err = intel_pt_synth_event(session, &attr, id);
@@ -1777,6 +1957,28 @@ static struct perf_evsel *intel_pt_find_sched_switch(struct perf_evlist *evlist)
return NULL;
}
+static bool intel_pt_find_switch(struct perf_evlist *evlist)
+{
+ struct perf_evsel *evsel;
+
+ evlist__for_each(evlist, evsel) {
+ if (evsel->attr.context_switch)
+ return true;
+ }
+
+ return false;
+}
+
+static int intel_pt_perf_config(const char *var, const char *value, void *data)
+{
+ struct intel_pt *pt = data;
+
+ if (!strcmp(var, "intel-pt.mispred-all"))
+ pt->mispred_all = perf_config_bool(var, value);
+
+ return 0;
+}
+
static const char * const intel_pt_info_fmts[] = {
[INTEL_PT_PMU_TYPE] = " PMU Type %"PRId64"\n",
[INTEL_PT_TIME_SHIFT] = " Time Shift %"PRIu64"\n",
@@ -1821,6 +2023,8 @@ int intel_pt_process_auxtrace_info(union perf_event *event,
if (!pt)
return -ENOMEM;
+ perf_config(intel_pt_perf_config, pt);
+
err = auxtrace_queues__init(&pt->queues);
if (err)
goto err_free;
@@ -1888,6 +2092,10 @@ int intel_pt_process_auxtrace_info(union perf_event *event,
pr_err("%s: missing sched_switch event\n", __func__);
goto err_delete_thread;
}
+ } else if (pt->have_sched_switch == 2 &&
+ !intel_pt_find_switch(session->evlist)) {
+ pr_err("%s: missing context_switch attribute flag\n", __func__);
+ goto err_delete_thread;
}
if (session->itrace_synth_opts && session->itrace_synth_opts->set) {
OpenPOWER on IntegriCloud