Re: [PATCH 4/5] perf tools: Allow to ignore missing pid

[Arnaldo Carvalho de Melo <acme@kernel.org>]

Em Mon, Dec 12, 2016 at 11:35:42AM +0100, Jiri Olsa escreveu:
> Adding perf_evsel::ignore_missing_cpu_thread bool.
> 
> When set true, it allows perf to ignore error of missing
> pid of perf event syscall.
> 
> We remove missing thread id from the thread_map, so the
> rest of the processing like ioctl and mmap won't get
> disturbed with -1 fd

What was the test performed? I tried it here the following while doing
a 'make -j4 allmodconfig' in the background, i.e. tons of threads coming
and going, and I sometimes get:

[root@jouet ~]# perf record -u acme 
Error:
The sys_perf_event_open() syscall returned with 3 (No such process) for
event (cycles:ppp).
/bin/dmesg may provide additional information.
No CONFIG_PERF_EVENTS=y kernel support configured?
[root@jouet ~]#
 
> The reason for supporting this is to ease up monitoring
> group of pids, that 'disappear' before perf opens their
> event. This currently leads perf to report error and exit
> and makes perf record's -u option unusable under certain
> setup.
> 
> With this change we will allow this race and ignore such
> failure with following warning:
> 
>   WARNING: Ignored open failure for pid 8605
> 
> Signed-off-by: Jiri Olsa <jolsa@kernel.org>
> Cc: David Ahern <dsahern@gmail.com>
> Cc: Namhyung Kim <namhyung@kernel.org>
> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
> Link: http://lkml.kernel.org/n/tip-atmzsjyhrt1j6xjwwmb84upg@git.kernel.org
> ---
>  tools/perf/perf.h       |  1 +
>  tools/perf/util/evsel.c | 44 ++++++++++++++++++++++++++++++++++++++++++++
>  tools/perf/util/evsel.h |  1 +
>  3 files changed, 46 insertions(+)
> 
> diff --git a/tools/perf/perf.h b/tools/perf/perf.h
> index 9a0236a4cf95..1c27d947c2fe 100644
> --- a/tools/perf/perf.h
> +++ b/tools/perf/perf.h
> @@ -55,6 +55,7 @@ struct record_opts {
>  	bool	     all_user;
>  	bool	     tail_synthesize;
>  	bool	     overwrite;
> +	bool	     ignore_missing_thread;
>  	unsigned int freq;
>  	unsigned int mmap_pages;
>  	unsigned int auxtrace_mmap_pages;
> diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c
> index fd61ebd77c26..ac87f1637ae9 100644
> --- a/tools/perf/util/evsel.c
> +++ b/tools/perf/util/evsel.c
> @@ -990,6 +990,8 @@ void perf_evsel__config(struct perf_evsel *evsel, struct record_opts *opts,
>  	 * it overloads any global configuration.
>  	 */
>  	apply_config_terms(evsel, opts);
> +
> +	evsel->ignore_missing_thread = opts->ignore_missing_thread;
>  }
>  
>  static int perf_evsel__alloc_fd(struct perf_evsel *evsel, int ncpus, int nthreads)
> @@ -1419,6 +1421,33 @@ static int __open_attr__fprintf(FILE *fp, const char *name, const char *val,
>  	return fprintf(fp, "  %-32s %s\n", name, val);
>  }
>  
> +static bool ignore_missing_thread(struct perf_evsel *evsel,
> +				  struct thread_map *threads,
> +				  int thread, int err)
> +{
> +	if (!evsel->ignore_missing_thread)
> +		return false;
> +
> +	/* The system wide setup does not work with threads. */
> +	if (!evsel->system_wide)
> +		return false;
> +
> +	/* The -ESRCH is perf event syscall errno for pid's not found. */
> +	if (err != -ESRCH)
> +		return false;
> +
> +	/* If there's only one thread, let it fail. */
> +	if (threads->nr == 1)
> +		return false;
> +
> +	if (thread_map__remove(threads, thread))
> +		return false;
> +
> +	pr_warning("WARNING: Ignored open failure for pid %d\n",
> +		   thread_map__pid(threads, thread));
> +	return true;
> +}
> +
>  static int __perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus,
>  			      struct thread_map *threads)
>  {
> @@ -1491,6 +1520,21 @@ static int __perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus,
>  
>  			if (fd < 0) {
>  				err = -errno;
> +
> +				if (ignore_missing_thread(evsel, threads, thread, err)) {
> +					/*
> +					 * We just removed 1 thread, so take a step
> +					 * back on thread index and lower the upper
> +					 * nthreads limit.
> +					 */
> +					nthreads--;
> +					thread--;
> +
> +					/* ... and pretend like nothing have happened. */
> +					err = 0;
> +					continue;
> +				}
> +
>  				pr_debug2("\nsys_perf_event_open failed, error %d\n",
>  					  err);
>  				goto try_fallback;
> diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h
> index 6abb89cd27f9..06ef6f29efa1 100644
> --- a/tools/perf/util/evsel.h
> +++ b/tools/perf/util/evsel.h
> @@ -120,6 +120,7 @@ struct perf_evsel {
>  	bool			tracking;
>  	bool			per_pkg;
>  	bool			precise_max;
> +	bool			ignore_missing_thread;
>  	/* parse modifier helper */
>  	int			exclude_GH;
>  	int			nr_members;
> -- 
> 2.7.4