Messages in this thread | | | From | Namhyung Kim <> | Date | Thu, 18 Aug 2022 09:41:44 -0700 | Subject | Re: [PATCH v1 5/6] perf mutex: Fix thread safety analysis |
| |
On Tue, Aug 16, 2022 at 10:39 PM Ian Rogers <irogers@google.com> wrote: > > Add annotations to describe lock behavior. Add missing unlocks to > perf_sched__replay. Alter hist_iter__top_callback as the thread-safety > analysis cannot follow pointers through local variables. > > Signed-off-by: Ian Rogers <irogers@google.com> > --- > tools/perf/builtin-sched.c | 8 ++++++++ > tools/perf/builtin-top.c | 5 +++-- > 2 files changed, 11 insertions(+), 2 deletions(-) > > diff --git a/tools/perf/builtin-sched.c b/tools/perf/builtin-sched.c > index 0f52f73be896..a8a765ed28ce 100644 > --- a/tools/perf/builtin-sched.c > +++ b/tools/perf/builtin-sched.c > @@ -658,6 +658,8 @@ static void *thread_func(void *ctx) > } > > static void create_tasks(struct perf_sched *sched) > + EXCLUSIVE_LOCK_FUNCTION(sched->start_work_mutex) > + EXCLUSIVE_LOCK_FUNCTION(sched->work_done_wait_mutex) > { > struct task_desc *task; > pthread_attr_t attr; > @@ -687,6 +689,8 @@ static void create_tasks(struct perf_sched *sched) > } > > static void wait_for_tasks(struct perf_sched *sched) > + EXCLUSIVE_LOCKS_REQUIRED(sched->work_done_wait_mutex) > + EXCLUSIVE_LOCKS_REQUIRED(sched->start_work_mutex) > { > u64 cpu_usage_0, cpu_usage_1; > struct task_desc *task; > @@ -738,6 +742,8 @@ static void wait_for_tasks(struct perf_sched *sched) > } > > static void run_one_test(struct perf_sched *sched) > + EXCLUSIVE_LOCKS_REQUIRED(sched->work_done_wait_mutex) > + EXCLUSIVE_LOCKS_REQUIRED(sched->start_work_mutex) > { > u64 T0, T1, delta, avg_delta, fluct; > > @@ -3314,6 +3320,8 @@ static int perf_sched__replay(struct perf_sched *sched) > for (i = 0; i < sched->replay_repeat; i++) > run_one_test(sched); > > + mutex_unlock(&sched->start_work_mutex); > + mutex_unlock(&sched->work_done_wait_mutex);
But this would wake up the replay tasks and let them burn cpus unnecessarily. Maybe we can make them exit at the moment.
> return 0; > } > > diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c > index 3757292bfe86..e832f04e3076 100644 > --- a/tools/perf/builtin-top.c > +++ b/tools/perf/builtin-top.c > @@ -196,6 +196,7 @@ static void perf_top__record_precise_ip(struct perf_top *top, > struct hist_entry *he, > struct perf_sample *sample, > struct evsel *evsel, u64 ip) > + EXCLUSIVE_LOCKS_REQUIRED(he->hists->lock) > { > struct annotation *notes; > struct symbol *sym = he->ms.sym; > @@ -724,13 +725,13 @@ static void *display_thread(void *arg) > static int hist_iter__top_callback(struct hist_entry_iter *iter, > struct addr_location *al, bool single, > void *arg) > + EXCLUSIVE_LOCKS_REQUIRED(iter->he->hists->lock) > { > struct perf_top *top = arg; > - struct hist_entry *he = iter->he; > struct evsel *evsel = iter->evsel; > > if (perf_hpp_list.sym && single) > - perf_top__record_precise_ip(top, he, iter->sample, evsel, al->addr); > + perf_top__record_precise_ip(top, iter->he, iter->sample, evsel, al->addr); > > hist__account_cycles(iter->sample->branch_stack, al, iter->sample, > !(top->record_opts.branch_stack & PERF_SAMPLE_BRANCH_ANY),
Looks like a separate change.
Thanks, Namhyung
> -- > 2.37.1.595.g718a3a8f04-goog >
| |