From 4fef41bfb1d8d2ada4a18eb3ab80c2682bcbae12 Mon Sep 17 00:00:00 2001 From: Adrian Hunter Date: Wed, 1 Apr 2020 13:16:06 +0300 Subject: perf thread-stack: Add thread_stack__sample_late() Add a thread stack function to create a call chain for hardware events where the sample records get created some time after the event occurred. Signed-off-by: Adrian Hunter Cc: Andi Kleen Cc: Jiri Olsa Link: http://lore.kernel.org/lkml/20200401101613.6201-10-adrian.hunter@intel.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/thread-stack.c | 57 ++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 57 insertions(+) (limited to 'tools/perf/util/thread-stack.c') diff --git a/tools/perf/util/thread-stack.c b/tools/perf/util/thread-stack.c index 0885967d5bc3..83f6c83f5617 100644 --- a/tools/perf/util/thread-stack.c +++ b/tools/perf/util/thread-stack.c @@ -497,6 +497,63 @@ void thread_stack__sample(struct thread *thread, int cpu, chain->nr = i; } +/* + * Hardware sample records, created some time after the event occurred, need to + * have subsequent addresses removed from the call chain. + */ +void thread_stack__sample_late(struct thread *thread, int cpu, + struct ip_callchain *chain, size_t sz, + u64 sample_ip, u64 kernel_start) +{ + struct thread_stack *ts = thread__stack(thread, cpu); + u64 sample_context = callchain_context(sample_ip, kernel_start); + u64 last_context, context, ip; + size_t nr = 0, j; + + if (sz < 2) { + chain->nr = 0; + return; + } + + if (!ts) + goto out; + + /* + * When tracing kernel space, kernel addresses occur at the top of the + * call chain after the event occurred but before tracing stopped. + * Skip them. + */ + for (j = 1; j <= ts->cnt; j++) { + ip = ts->stack[ts->cnt - j].ret_addr; + context = callchain_context(ip, kernel_start); + if (context == PERF_CONTEXT_USER || + (context == sample_context && ip == sample_ip)) + break; + } + + last_context = sample_ip; /* Use sample_ip as an invalid context */ + + for (; nr < sz && j <= ts->cnt; nr++, j++) { + ip = ts->stack[ts->cnt - j].ret_addr; + context = callchain_context(ip, kernel_start); + if (context != last_context) { + if (nr >= sz - 1) + break; + chain->ips[nr++] = context; + last_context = context; + } + chain->ips[nr] = ip; + } +out: + if (nr) { + chain->nr = nr; + } else { + chain->ips[0] = sample_context; + chain->ips[1] = sample_ip; + chain->nr = 2; + } +} + struct call_return_processor * call_return_processor__new(int (*process)(struct call_return *cr, u64 *parent_db_id, void *data), void *data) -- cgit v1.2.3 From 86d67180b920d178ae1c2923f50a0759d6ce1a10 Mon Sep 17 00:00:00 2001 From: Adrian Hunter Date: Wed, 29 Apr 2020 18:07:43 +0300 Subject: perf thread-stack: Add branch stack support Intel PT already has support for creating branch stacks for each context (per-cpu or per-thread). In the more common per-cpu case, the branch stack is not separated for different threads, instead being cleared in between each sample. That approach will not work very well for adding branch stacks to regular events. The branch stacks really need to be accumulated separately for each thread. As a start to accomplishing that, this patch adds support for putting branch stack support into the thread-stack. The advantages are: 1. the branches are accumulated separately for each thread 2. the branch stack is cleared only in between continuous traces This helps pave the way for adding branch stacks to regular events, not just synthesized events as at present. Signed-off-by: Adrian Hunter Cc: Andi Kleen Cc: Jiri Olsa Link: http://lore.kernel.org/lkml/20200429150751.12570-2-adrian.hunter@intel.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/intel-bts.c | 2 +- tools/perf/util/intel-pt.c | 2 +- tools/perf/util/thread-stack.c | 113 +++++++++++++++++++++++++++++++++++++---- tools/perf/util/thread-stack.h | 5 +- 4 files changed, 108 insertions(+), 14 deletions(-) (limited to 'tools/perf/util/thread-stack.c') diff --git a/tools/perf/util/intel-bts.c b/tools/perf/util/intel-bts.c index 059e1c805ed0..506112f52619 100644 --- a/tools/perf/util/intel-bts.c +++ b/tools/perf/util/intel-bts.c @@ -432,7 +432,7 @@ static int intel_bts_process_buffer(struct intel_bts_queue *btsq, le64_to_cpu(branch->from), le64_to_cpu(branch->to), btsq->intel_pt_insn.length, - buffer->buffer_nr + 1); + buffer->buffer_nr + 1, true, 0, 0); if (filter && !(filter & btsq->sample_flags)) continue; err = intel_bts_synth_branch_sample(btsq, branch); diff --git a/tools/perf/util/intel-pt.c b/tools/perf/util/intel-pt.c index 30e1ee6d3e40..bdb84470f7d0 100644 --- a/tools/perf/util/intel-pt.c +++ b/tools/perf/util/intel-pt.c @@ -2033,7 +2033,7 @@ static int intel_pt_sample(struct intel_pt_queue *ptq) pt->synth_opts.thread_stack) thread_stack__event(ptq->thread, ptq->cpu, ptq->flags, state->from_ip, state->to_ip, ptq->insn_len, - state->trace_nr); + state->trace_nr, true, 0, 0); else thread_stack__set_trace_nr(ptq->thread, ptq->cpu, state->trace_nr); diff --git a/tools/perf/util/thread-stack.c b/tools/perf/util/thread-stack.c index 83f6c83f5617..79698831b4b9 100644 --- a/tools/perf/util/thread-stack.c +++ b/tools/perf/util/thread-stack.c @@ -80,6 +80,10 @@ struct thread_stack_entry { * @comm: current comm * @arr_sz: size of array if this is the first element of an array * @rstate: used to detect retpolines + * @br_stack_rb: branch stack (ring buffer) + * @br_stack_sz: maximum branch stack size + * @br_stack_pos: current position in @br_stack_rb + * @mispred_all: mark all branches as mispredicted */ struct thread_stack { struct thread_stack_entry *stack; @@ -95,6 +99,10 @@ struct thread_stack { struct comm *comm; unsigned int arr_sz; enum retpoline_state_t rstate; + struct branch_stack *br_stack_rb; + unsigned int br_stack_sz; + unsigned int br_stack_pos; + bool mispred_all; }; /* @@ -126,13 +134,26 @@ static int thread_stack__grow(struct thread_stack *ts) } static int thread_stack__init(struct thread_stack *ts, struct thread *thread, - struct call_return_processor *crp) + struct call_return_processor *crp, + bool callstack, unsigned int br_stack_sz) { int err; - err = thread_stack__grow(ts); - if (err) - return err; + if (callstack) { + err = thread_stack__grow(ts); + if (err) + return err; + } + + if (br_stack_sz) { + size_t sz = sizeof(struct branch_stack); + + sz += br_stack_sz * sizeof(struct branch_entry); + ts->br_stack_rb = zalloc(sz); + if (!ts->br_stack_rb) + return -ENOMEM; + ts->br_stack_sz = br_stack_sz; + } if (thread->maps && thread->maps->machine) { struct machine *machine = thread->maps->machine; @@ -150,7 +171,9 @@ static int thread_stack__init(struct thread_stack *ts, struct thread *thread, } static struct thread_stack *thread_stack__new(struct thread *thread, int cpu, - struct call_return_processor *crp) + struct call_return_processor *crp, + bool callstack, + unsigned int br_stack_sz) { struct thread_stack *ts = thread->ts, *new_ts; unsigned int old_sz = ts ? ts->arr_sz : 0; @@ -176,7 +199,7 @@ static struct thread_stack *thread_stack__new(struct thread *thread, int cpu, ts += cpu; if (!ts->stack && - thread_stack__init(ts, thread, crp)) + thread_stack__init(ts, thread, crp, callstack, br_stack_sz)) return NULL; return ts; @@ -319,6 +342,9 @@ static int __thread_stack__flush(struct thread *thread, struct thread_stack *ts) if (!crp) { ts->cnt = 0; + ts->br_stack_pos = 0; + if (ts->br_stack_rb) + ts->br_stack_rb->nr = 0; return 0; } @@ -353,8 +379,33 @@ int thread_stack__flush(struct thread *thread) return err; } +static void thread_stack__update_br_stack(struct thread_stack *ts, u32 flags, + u64 from_ip, u64 to_ip) +{ + struct branch_stack *bs = ts->br_stack_rb; + struct branch_entry *be; + + if (!ts->br_stack_pos) + ts->br_stack_pos = ts->br_stack_sz; + + ts->br_stack_pos -= 1; + + be = &bs->entries[ts->br_stack_pos]; + be->from = from_ip; + be->to = to_ip; + be->flags.value = 0; + be->flags.abort = !!(flags & PERF_IP_FLAG_TX_ABORT); + be->flags.in_tx = !!(flags & PERF_IP_FLAG_IN_TX); + /* No support for mispredict */ + be->flags.mispred = ts->mispred_all; + + if (bs->nr < ts->br_stack_sz) + bs->nr += 1; +} + int thread_stack__event(struct thread *thread, int cpu, u32 flags, u64 from_ip, - u64 to_ip, u16 insn_len, u64 trace_nr) + u64 to_ip, u16 insn_len, u64 trace_nr, bool callstack, + unsigned int br_stack_sz, bool mispred_all) { struct thread_stack *ts = thread__stack(thread, cpu); @@ -362,12 +413,13 @@ int thread_stack__event(struct thread *thread, int cpu, u32 flags, u64 from_ip, return -EINVAL; if (!ts) { - ts = thread_stack__new(thread, cpu, NULL); + ts = thread_stack__new(thread, cpu, NULL, callstack, br_stack_sz); if (!ts) { pr_warning("Out of memory: no thread stack\n"); return -ENOMEM; } ts->trace_nr = trace_nr; + ts->mispred_all = mispred_all; } /* @@ -381,8 +433,14 @@ int thread_stack__event(struct thread *thread, int cpu, u32 flags, u64 from_ip, ts->trace_nr = trace_nr; } - /* Stop here if thread_stack__process() is in use */ - if (ts->crp) + if (br_stack_sz) + thread_stack__update_br_stack(ts, flags, from_ip, to_ip); + + /* + * Stop here if thread_stack__process() is in use, or not recording call + * stack. + */ + if (ts->crp || !callstack) return 0; if (flags & PERF_IP_FLAG_CALL) { @@ -430,6 +488,7 @@ static void __thread_stack__free(struct thread *thread, struct thread_stack *ts) { __thread_stack__flush(thread, ts); zfree(&ts->stack); + zfree(&ts->br_stack_rb); } static void thread_stack__reset(struct thread *thread, struct thread_stack *ts) @@ -554,6 +613,38 @@ out: } } +void thread_stack__br_sample(struct thread *thread, int cpu, + struct branch_stack *dst, unsigned int sz) +{ + struct thread_stack *ts = thread__stack(thread, cpu); + const size_t bsz = sizeof(struct branch_entry); + struct branch_stack *src; + struct branch_entry *be; + unsigned int nr; + + dst->nr = 0; + + if (!ts) + return; + + src = ts->br_stack_rb; + if (!src->nr) + return; + + dst->nr = min((unsigned int)src->nr, sz); + + be = &dst->entries[0]; + nr = min(ts->br_stack_sz - ts->br_stack_pos, (unsigned int)dst->nr); + memcpy(be, &src->entries[ts->br_stack_pos], bsz * nr); + + if (src->nr >= ts->br_stack_sz) { + sz -= nr; + be = &dst->entries[nr]; + nr = min(ts->br_stack_pos, sz); + memcpy(be, &src->entries[0], bsz * ts->br_stack_pos); + } +} + struct call_return_processor * call_return_processor__new(int (*process)(struct call_return *cr, u64 *parent_db_id, void *data), void *data) @@ -921,7 +1012,7 @@ int thread_stack__process(struct thread *thread, struct comm *comm, } if (!ts) { - ts = thread_stack__new(thread, sample->cpu, crp); + ts = thread_stack__new(thread, sample->cpu, crp, true, 0); if (!ts) return -ENOMEM; ts->comm = comm; diff --git a/tools/perf/util/thread-stack.h b/tools/perf/util/thread-stack.h index 8962ddc4e1ab..c279a0cbbdd0 100644 --- a/tools/perf/util/thread-stack.h +++ b/tools/perf/util/thread-stack.h @@ -81,13 +81,16 @@ struct call_return_processor { }; int thread_stack__event(struct thread *thread, int cpu, u32 flags, u64 from_ip, - u64 to_ip, u16 insn_len, u64 trace_nr); + u64 to_ip, u16 insn_len, u64 trace_nr, bool callstack, + unsigned int br_stack_sz, bool mispred_all); void thread_stack__set_trace_nr(struct thread *thread, int cpu, u64 trace_nr); void thread_stack__sample(struct thread *thread, int cpu, struct ip_callchain *chain, size_t sz, u64 ip, u64 kernel_start); void thread_stack__sample_late(struct thread *thread, int cpu, struct ip_callchain *chain, size_t sz, u64 ip, u64 kernel_start); +void thread_stack__br_sample(struct thread *thread, int cpu, + struct branch_stack *dst, unsigned int sz); int thread_stack__flush(struct thread *thread); void thread_stack__free(struct thread *thread); size_t thread_stack__depth(struct thread *thread, int cpu); -- cgit v1.2.3 From 3749e0bbdef24efbf1698bf0dbd9575fddb9ed22 Mon Sep 17 00:00:00 2001 From: Adrian Hunter Date: Wed, 29 Apr 2020 18:07:48 +0300 Subject: perf thread-stack: Add thread_stack__br_sample_late() Add a thread stack function to create a branch stack for hardware events where the sample records get created some time after the event occurred. Signed-off-by: Adrian Hunter Cc: Andi Kleen Cc: Jiri Olsa Link: http://lore.kernel.org/lkml/20200429150751.12570-7-adrian.hunter@intel.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/thread-stack.c | 104 +++++++++++++++++++++++++++++++++++++++++ tools/perf/util/thread-stack.h | 3 ++ 2 files changed, 107 insertions(+) (limited to 'tools/perf/util/thread-stack.c') diff --git a/tools/perf/util/thread-stack.c b/tools/perf/util/thread-stack.c index 79698831b4b9..1b992bbba4e8 100644 --- a/tools/perf/util/thread-stack.c +++ b/tools/perf/util/thread-stack.c @@ -645,6 +645,110 @@ void thread_stack__br_sample(struct thread *thread, int cpu, } } +/* Start of user space branch entries */ +static bool us_start(struct branch_entry *be, u64 kernel_start, bool *start) +{ + if (!*start) + *start = be->to && be->to < kernel_start; + + return *start; +} + +/* + * Start of branch entries after the ip fell in between 2 branches, or user + * space branch entries. + */ +static bool ks_start(struct branch_entry *be, u64 sample_ip, u64 kernel_start, + bool *start, struct branch_entry *nb) +{ + if (!*start) { + *start = (nb && sample_ip >= be->to && sample_ip <= nb->from) || + be->from < kernel_start || + (be->to && be->to < kernel_start); + } + + return *start; +} + +/* + * Hardware sample records, created some time after the event occurred, need to + * have subsequent addresses removed from the branch stack. + */ +void thread_stack__br_sample_late(struct thread *thread, int cpu, + struct branch_stack *dst, unsigned int sz, + u64 ip, u64 kernel_start) +{ + struct thread_stack *ts = thread__stack(thread, cpu); + struct branch_entry *d, *s, *spos, *ssz; + struct branch_stack *src; + unsigned int nr = 0; + bool start = false; + + dst->nr = 0; + + if (!ts) + return; + + src = ts->br_stack_rb; + if (!src->nr) + return; + + spos = &src->entries[ts->br_stack_pos]; + ssz = &src->entries[ts->br_stack_sz]; + + d = &dst->entries[0]; + s = spos; + + if (ip < kernel_start) { + /* + * User space sample: start copying branch entries when the + * branch is in user space. + */ + for (s = spos; s < ssz && nr < sz; s++) { + if (us_start(s, kernel_start, &start)) { + *d++ = *s; + nr += 1; + } + } + + if (src->nr >= ts->br_stack_sz) { + for (s = &src->entries[0]; s < spos && nr < sz; s++) { + if (us_start(s, kernel_start, &start)) { + *d++ = *s; + nr += 1; + } + } + } + } else { + struct branch_entry *nb = NULL; + + /* + * Kernel space sample: start copying branch entries when the ip + * falls in between 2 branches (or the branch is in user space + * because then the start must have been missed). + */ + for (s = spos; s < ssz && nr < sz; s++) { + if (ks_start(s, ip, kernel_start, &start, nb)) { + *d++ = *s; + nr += 1; + } + nb = s; + } + + if (src->nr >= ts->br_stack_sz) { + for (s = &src->entries[0]; s < spos && nr < sz; s++) { + if (ks_start(s, ip, kernel_start, &start, nb)) { + *d++ = *s; + nr += 1; + } + nb = s; + } + } + } + + dst->nr = nr; +} + struct call_return_processor * call_return_processor__new(int (*process)(struct call_return *cr, u64 *parent_db_id, void *data), void *data) diff --git a/tools/perf/util/thread-stack.h b/tools/perf/util/thread-stack.h index c279a0cbbdd0..3bc47a42af8e 100644 --- a/tools/perf/util/thread-stack.h +++ b/tools/perf/util/thread-stack.h @@ -91,6 +91,9 @@ void thread_stack__sample_late(struct thread *thread, int cpu, u64 kernel_start); void thread_stack__br_sample(struct thread *thread, int cpu, struct branch_stack *dst, unsigned int sz); +void thread_stack__br_sample_late(struct thread *thread, int cpu, + struct branch_stack *dst, unsigned int sz, + u64 sample_ip, u64 kernel_start); int thread_stack__flush(struct thread *thread); void thread_stack__free(struct thread *thread); size_t thread_stack__depth(struct thread *thread, int cpu); -- cgit v1.2.3