Re: [PATCH] perf tools: Fix ordering with unstable tsc

[Stephane Eranian <eranian@google.com>]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset=UTF-8, Size: 15998 bytes --]

On Sat, Feb 18, 2012 at 5:50 PM, Frederic Weisbecker <fweisbec@gmail.com> wrote:
> On a system with a TSC considered as unstable, one can encounter this
> kind of warning:
>
>     $ perf sched rec pong 2
>     $ perf sched lat
>     Warning: Timestamp below last timeslice flush
>
> This happens when trace events trigger with a potentially high period,
> such as sched_stat_sleep, sched_stat_runtime, sched_stat_wait, etc...
> The perf event core then implement that weight by sending as many events
> as the given period. For example as many as the time the task has been
> sleeping in sched_stat_sleep event.
>
> If this happens while irqs are disabled with an unstable tsc and this takes
> more time than a jiffy, then the timestamps of the events get stuck to
> the value of that next jiffy because sched_clock_local() bounds the timestamp
> to that maximum. The local timer tick is supposed to update that boundary but
> it can't given that irqs are disabled.
>
> We can then meet this kind of scenario in perf record:
>
> ===== CPU 0 =====      ==== CPU 1 ====
>
>              PASS n
>     ...                    ...
>      1                      1
>      1                      2
>      1                      3 <-- max recorded
>
>           finished round event
>            PASS n + 1
>
>      1                      4
>      1                      5
>      1                      6
>
>           finished round event
>            PASS n + 2
>
>      1                      7
>     ...                    ...
>
> CPU 0 is stuck sending events with irqs disabled and with the stale
> timestamp. When we do the events reordering for perf script for example,
> we flush all the events before timestamp 3 when we reach PASS n + 2,
> considering we can't anymore have timestamps below 3 now.
> But we still do have timestamps below 3 on PASS n + 2.
>
> To solve that issue, instead of considering that timestamps are globally
> monotonic, we assume they are locally monotonic. Instead of recording
> the max timestamp on each pass, we check the max one per CPU on each
> pass and keep the smallest over these as the new barrier up to which
> we flush the events on the PASS n + 2. This still relies on a bit of
> global monotonicity because if some CPU doesn't have events in PASS n,
> we expect it not to have event in PASS n + 2 past the barrier recorded
> in PASS n. So this is still not a totally robust ordering but it's still
> better than what we had before.
>
> The only way to have a deterministic and solid ordering will be to use
> per cpu perf.data files.
>
> Reported-by: Stephane Eranian <eranian@google.com>
> Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
> Cc: David Ahern <dsahern@gmail.com>
> Cc: Peter Zijlstra <peterz@infradead.org>
> Cc: Stephane Eranian <eranian@google.com>
> Cc: Ingo Molnar <mingo@elte.hu>
> Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
> ---
>  tools/perf/util/evsel.c   |    5 +-
>  tools/perf/util/session.c |  146 +++++++++++++++++++++++++++++++++-----------
>  tools/perf/util/session.h |    3 +-
>  3 files changed, 115 insertions(+), 39 deletions(-)
>
> diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c
> index 302d49a..1c8eb4b 100644
> --- a/tools/perf/util/evsel.c
> +++ b/tools/perf/util/evsel.c
> @@ -119,9 +119,12 @@ void perf_evsel__config(struct perf_evsel *evsel, struct perf_record_opts *opts)
>        if (opts->raw_samples) {
>                attr->sample_type       |= PERF_SAMPLE_TIME;
>                attr->sample_type       |= PERF_SAMPLE_RAW;
> -               attr->sample_type       |= PERF_SAMPLE_CPU;
>        }
>
I don't get this bit here. You may want CPU information when capturing
in raw + per-thread mode.


> +       /* Need to know the CPU for tools that need to order events */
> +       if (attr->sample_type & PERF_SAMPLE_TIME)
> +               attr->sample_type       |= PERF_SAMPLE_CPU;
> +
>        if (opts->no_delay) {
>                attr->watermark = 0;
>                attr->wakeup_events = 1;
> diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c
> index 9f833cf..f297342 100644
> --- a/tools/perf/util/session.c
> +++ b/tools/perf/util/session.c
> @@ -494,6 +494,8 @@ static void perf_session_free_sample_buffers(struct perf_session *session)
>                list_del(&sq->list);
>                free(sq);
>        }
> +
> +       free(os->last_cpu_timestamp);
>  }
>
>  static int perf_session_deliver_event(struct perf_session *session,
> @@ -549,56 +551,89 @@ static void flush_sample_queue(struct perf_session *s,
>  }
>
>  /*
> - * When perf record finishes a pass on every buffers, it records this pseudo
> - * event.
> - * We record the max timestamp t found in the pass n.
> - * Assuming these timestamps are monotonic across cpus, we know that if
> - * a buffer still has events with timestamps below t, they will be all
> - * available and then read in the pass n + 1.
> - * Hence when we start to read the pass n + 2, we can safely flush every
> - * events with timestamps below t.
> + * We make the assumption that timestamps are not globally monotonic but locally
> + * non-strictly monotonic. In practice, this is because if we are dealing with a
> + * machine with unstable TSC, the kernel bounds the result of the tsc between
> + * last_tick_time < tsc < next_tick_time. Thus, if a CPU disables interrupts for more
> + * than one jiffy, all of its timestamps will be equal to next_tick_time after we
> + * cross that jiffy, without any further progress whereas the other CPU continue
> + * with normal timestamps. This can happen if a CPU sends crazillions of events
> + * while interrupts are disabled. But there are potentially other random scenarios
> + * with unstable TSC that drives us to assume the monotonicity of time only per CPU
> + * and not globally.
> + *
> + * To solve this, when perf record finishes a round of write on every buffers, it
> + * records a pseudo event named "finished round". The frame of events that happen
> + * between two finished rounds is called a "pass".
> + * We record the max timestamp T[cpu] per CPU found over the events in the pass n.
> + * Then when we finish a round, we iterate over these T[cpu]and keep the smallest
> + * one: min(T).
> + *
> + * Assuming these timestamps are locally monotonic (non strictly), we can flush all
> + * queued events having a timestamp below min(T) when we start to process PASS n + 1.
> + * But we actually wait until we start PASS n + 2 in case a CPU did not have any
> + * event in PASS n but came in PASS n + 1 with events below min(T). We truly
> + * hope no CPU will come with events below min(T) after pass n + 1. This
> + * heuristicly rely on some minimal global consistancy. This should work in most
> + * real world case, the only way to ensure a truly safe ordering with regular
> + * flush will be to switch to per CPU record files.
> + *
>  *
> - *    ============ PASS n =================
> - *       CPU 0         |   CPU 1
> - *                     |
> - *    cnt1 timestamps  |   cnt2 timestamps
> - *          1          |         2
> - *          2          |         3
> - *          -          |         4  <--- max recorded
> + *    ========================== PASS n ============================
> + *       CPU 0                   |   CPU 1
> + *                               |
> + *    cnt1 timestamps            |   cnt2 timestamps
> + *          1                    |         2
> + *          2 <--- max recorded  |         3
> + *          -                    |         4 <--- max recorded
> + *                          min(T) = 2
>  *
> - *    ============ PASS n + 1 ==============
> - *       CPU 0         |   CPU 1
> - *                     |
> - *    cnt1 timestamps  |   cnt2 timestamps
> - *          3          |         5
> - *          4          |         6
> - *          5          |         7 <---- max recorded
> + *    ========================== PASS n + 1 ========================
> + *       CPU 0                   |   CPU 1
> + *                               |
> + *    cnt1 timestamps            |   cnt2 timestamps
> + *          3                    |         5
> + *          4                    |         6
> + *          5 <--- max record    |         7 <---- max recorded
> + *                          min(T) = 5
>  *
> - *      Flush every events below timestamp 4
> + *                Flush every events below timestamp 2
>  *
> - *    ============ PASS n + 2 ==============
> - *       CPU 0         |   CPU 1
> - *                     |
> - *    cnt1 timestamps  |   cnt2 timestamps
> - *          6          |         8
> - *          7          |         9
> - *          -          |         10
> + *    ========================== PASS n + 2 ========================
> + *       CPU 0                   |   CPU 1
> + *                               |
> + *    cnt1 timestamps            |   cnt2 timestamps
> + *          6                    |         8
> + *          7                    |         9
> + *          -                    |         10
>  *
> - *      Flush every events below timestamp 7
> - *      etc...
> + *                Flush every events below timestamp 5, etc...
>  */
>  static int process_finished_round(struct perf_tool *tool,
>                                  union perf_event *event __used,
>                                  struct perf_session *session)
>  {
> +       unsigned int i;
> +       u64 min = ULLONG_MAX;
> +       struct ordered_samples *os = &session->ordered_samples;
> +
>        flush_sample_queue(session, tool);
> -       session->ordered_samples.next_flush = session->ordered_samples.max_timestamp;
> +
> +       for (i = 0; i < session->nr_cpus; i++) {
> +               if (os->last_cpu_timestamp[i] < min)
> +                       min = os->last_cpu_timestamp[i];
> +
> +               os->last_cpu_timestamp[i] = ULLONG_MAX;
> +       }
> +
> +       if (min != ULLONG_MAX)
> +               os->next_flush = min;
>
>        return 0;
>  }
>
>  /* The queue is ordered by time */
> -static void __queue_event(struct sample_queue *new, struct perf_session *s)
> +static void __queue_event(struct sample_queue *new, struct perf_session *s, int cpu)
>  {
>        struct ordered_samples *os = &s->ordered_samples;
>        struct sample_queue *sample = os->last_sample;
> @@ -607,10 +642,10 @@ static void __queue_event(struct sample_queue *new, struct perf_session *s)
>
>        ++os->nr_samples;
>        os->last_sample = new;
> +       os->last_cpu_timestamp[cpu] = timestamp;
>
>        if (!sample) {
>                list_add(&new->list, &os->samples);
> -               os->max_timestamp = timestamp;
>                return;
>        }
>
> @@ -624,7 +659,6 @@ static void __queue_event(struct sample_queue *new, struct perf_session *s)
>                        p = sample->list.next;
>                        if (p == &os->samples) {
>                                list_add_tail(&new->list, &os->samples);
> -                               os->max_timestamp = timestamp;
>                                return;
>                        }
>                        sample = list_entry(p, struct sample_queue, list);
> @@ -643,6 +677,34 @@ static void __queue_event(struct sample_queue *new, struct perf_session *s)
>        }
>  }
>
> +static int alloc_cpus_timestamp_array(struct perf_session *s,
> +                                     struct perf_sample *sample,
> +                                     struct ordered_samples *os)
> +{
> +       int i;
> +       int nr_cpus;
> +
> +       if (sample->cpu < s->nr_cpus)
> +               return 0;
> +
> +       nr_cpus = sample->cpu + 1;
> +
> +       if (!os->last_cpu_timestamp)
> +               os->last_cpu_timestamp = malloc(sizeof(u64) * nr_cpus);
> +       else
> +               os->last_cpu_timestamp = realloc(os->last_cpu_timestamp,
> +                                                sizeof(u64) * nr_cpus);
> +       if (!os->last_cpu_timestamp)
> +               return -ENOMEM;
> +
> +       for (i = s->nr_cpus; i < nr_cpus; i++)
> +               os->last_cpu_timestamp[i] = ULLONG_MAX;
> +
> +       s->nr_cpus = nr_cpus;
> +
> +       return 0;
> +}
> +
>  #define MAX_SAMPLE_BUFFER      (64 * 1024 / sizeof(struct sample_queue))
>
>  static int perf_session_queue_event(struct perf_session *s, union perf_event *event,
> @@ -652,6 +714,12 @@ static int perf_session_queue_event(struct perf_session *s, union perf_event *ev
>        struct list_head *sc = &os->sample_cache;
>        u64 timestamp = sample->time;
>        struct sample_queue *new;
> +       int err;
> +
> +       if (!(s->sample_type & PERF_SAMPLE_CPU)) {
> +               pr_err("Warning: Need to record CPU on samples for ordering\n");
> +               return -EINVAL;
> +       }
>
>        if (!timestamp || timestamp == ~0ULL)
>                return -ETIME;
> @@ -661,6 +729,10 @@ static int perf_session_queue_event(struct perf_session *s, union perf_event *ev
>                return -EINVAL;
>        }
>
> +       err = alloc_cpus_timestamp_array(s, sample, os);
> +       if (err)
> +               return err;
> +
>        if (!list_empty(sc)) {
>                new = list_entry(sc->next, struct sample_queue, list);
>                list_del(&new->list);
> @@ -681,7 +753,7 @@ static int perf_session_queue_event(struct perf_session *s, union perf_event *ev
>        new->file_offset = file_offset;
>        new->event = event;
>
> -       __queue_event(new, s);
> +       __queue_event(new, s, sample->cpu);
>
>        return 0;
>  }
> diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h
> index c8d9017..642591b 100644
> --- a/tools/perf/util/session.h
> +++ b/tools/perf/util/session.h
> @@ -16,7 +16,7 @@ struct thread;
>  struct ordered_samples {
>        u64                     last_flush;
>        u64                     next_flush;
> -       u64                     max_timestamp;
> +       u64                     *last_cpu_timestamp;
>        struct list_head        samples;
>        struct list_head        sample_cache;
>        struct list_head        to_free;
> @@ -50,6 +50,7 @@ struct perf_session {
>        int                     cwdlen;
>        char                    *cwd;
>        struct ordered_samples  ordered_samples;
> +       unsigned int            nr_cpus;
>        char                    filename[1];
>  };
>
> --
> 1.7.5.4
>
ÿôèº{.nÇ+‰·Ÿ®‰­†+%ŠËÿ±éݶ\x17¥Šwÿº{.nÇ+‰·¥Š{±þG«éÿŠ{ayº\x1dʇڙë,j\a­¢f£¢·hšïêÿ‘êçz_è®\x03(­éšŽŠÝ¢j"ú\x1a¶^[m§ÿÿ¾\a«þG«éÿ¢¸?™¨è­Ú&£ø§~á¶iO•æ¬z·švØ^\x14\x04\x1a¶^[m§ÿÿÃ\fÿ¶ìÿ¢¸?–I¥