Re: [PATCH v1] perf lock contention: skip traceiter functions

[Anne Macedo <retpolanne@posteo.net>]

Namhyung Kim <namhyung@kernel.org> writes:

> Hello Anne,
>
> On Sat, Mar 9, 2024 at 5:27 AM Anne Macedo <retpolanne@posteo.net> wrote:
>>
>> The perf lock contention program currently shows the caller of the locks
>> as __traceiter_contention_begin+0x??. This caller can be ignored, as it is
>> from the traceiter itself. Instead, it should show the real callers for
>> the locks.
>>
>> When fiddling with the --stack-skip parameter, the actual callers for
>> the locks start to show up. However, just ignore the
>> __traceiter_contention_begin and the __traceiter_contention_end symbols
>> so the actual callers will show up.
>>
>> Before this patch is applied:
>>
>> sudo perf lock con -a -b -- sleep 3
>>  contended   total wait     max wait     avg wait         type   caller
>>
>>          8      2.33 s       2.28 s     291.18 ms     rwlock:W   __traceiter_contention_begin+0x44
>>          4      2.33 s       2.28 s     582.35 ms     rwlock:W   __traceiter_contention_begin+0x44
>>          7    140.30 ms     46.77 ms     20.04 ms     rwlock:W   __traceiter_contention_begin+0x44
>>          2     63.35 ms     33.76 ms     31.68 ms        mutex   trace_contention_begin+0x84
>>          2     46.74 ms     46.73 ms     23.37 ms     rwlock:W   __traceiter_contention_begin+0x44
>>          1     13.54 us     13.54 us     13.54 us        mutex   trace_contention_begin+0x84
>>          1      3.67 us      3.67 us      3.67 us      rwsem:R   __traceiter_contention_begin+0x44
>>
>> Before this patch is applied - using --stack-skip 5
>>
>> sudo perf lock con --stack-skip 5 -a -b -- sleep 3
>>  contended   total wait     max wait     avg wait         type   caller
>>
>>          2      2.24 s       2.24 s       1.12 s      rwlock:W   do_epoll_wait+0x5a0
>>          4      1.65 s     824.21 ms    412.08 ms     rwlock:W   do_exit+0x338
>>          2    824.35 ms    824.29 ms    412.17 ms     spinlock   get_signal+0x108
>>          2    824.14 ms    824.14 ms    412.07 ms     rwlock:W   release_task+0x68
>>          1     25.22 ms     25.22 ms     25.22 ms        mutex   cgroup_kn_lock_live+0x58
>>          1     24.71 us     24.71 us     24.71 us     spinlock   do_exit+0x44
>>          1     22.04 us     22.04 us     22.04 us      rwsem:R   lock_mm_and_find_vma+0xb0
>>
>> After this patch is applied:
>>
>> sudo ./perf lock con -a -b -- sleep 3
>>  contended   total wait     max wait     avg wait         type   caller
>>
>>          4      4.13 s       2.07 s       1.03 s      rwlock:W   release_task+0x68
>>          2      2.07 s       2.07 s       1.03 s      rwlock:R   mm_update_next_owner+0x50
>>          2      2.07 s       2.07 s       1.03 s      rwlock:W   do_exit+0x338
>>          1     41.56 ms     41.56 ms     41.56 ms        mutex   cgroup_kn_lock_live+0x58
>>          2     36.12 us     18.83 us     18.06 us     rwlock:W   do_exit+0x338
>>
>> changes since v0:
>>
>> - skip trace_contention functions
>> - use sym->end instead of __traceiter_contention_end for text_end
>>
>> Signed-off-by: Anne Macedo <retpolanne@posteo.net>
>> ---
>>  tools/perf/util/machine.c | 17 +++++++++++++++++
>>  tools/perf/util/machine.h |  2 +-
>>  2 files changed, 18 insertions(+), 1 deletion(-)
>>
>> diff --git a/tools/perf/util/machine.c b/tools/perf/util/machine.c
>> index 527517db3182..db443947ccd1 100644
>> --- a/tools/perf/util/machine.c
>> +++ b/tools/perf/util/machine.c
>> @@ -3266,6 +3266,14 @@ bool machine__is_lock_function(struct machine *machine, u64 addr)
>>
>>                 sym = machine__find_kernel_symbol_by_name(machine, "__lock_text_end", &kmap);
>>                 machine->lock.text_end = map__unmap_ip(kmap, sym->start);
>> +
>> +               sym = machine__find_kernel_symbol_by_name(machine, "__traceiter_contention_begin", &kmap);
>
> Unlike __sched_text_{start,end} and __lock_text_{start,end} I guess
> this traceiter thing is optional so it might not be there.  You need to
> handle if it's NULL.
>

Let me test with CONFIG_TRACEPOINTS=n then.

> I think it depends on the kernel version and configuration.  I remember
> I saw a different symbol on old kernels.  But it'd be hard to handle all
> the cases.  Let's have a single trace text section in the struct machine
> and use __traceiter_contention_begin only.  If it's not found you can
> fallback to trace_contention_begin.
>
ACK :)
>
> Thanks,
> Namhyung
>
>
>> +               machine->traceiter.text_start = map__unmap_ip(kmap, sym->start);
>> +               machine->traceiter.text_end = map__unmap_ip(kmap, sym->end);
>> +
>> +               sym = machine__find_kernel_symbol_by_name(machine, "trace_contention_begin", &kmap);
>> +               machine->trace.text_start = map__unmap_ip(kmap, sym->start);
>> +               machine->trace.text_end = map__unmap_ip(kmap, sym->end);
>>         }
>>
>>         /* failed to get kernel symbols */
>> @@ -3280,5 +3288,14 @@ bool machine__is_lock_function(struct machine *machine, u64 addr)
>>         if (machine->lock.text_start <= addr && addr < machine->lock.text_end)
>>                 return true;
>>
>> +       /* traceiter functions currently don't have their own section
>> +        * but we consider them lock functions
>> +        */
>> +       if (machine->traceiter.text_start <= addr && addr < machine->traceiter.text_end)
>> +               return true;
>> +
>> +       if (machine->trace.text_start <= addr && addr < machine->trace.text_end)
>> +               return true;
>> +
>>         return false;
>>  }
>> diff --git a/tools/perf/util/machine.h b/tools/perf/util/machine.h
>> index e28c787616fe..4312f6db6de0 100644
>> --- a/tools/perf/util/machine.h
>> +++ b/tools/perf/util/machine.h
>> @@ -49,7 +49,7 @@ struct machine {
>>         struct {
>>                 u64       text_start;
>>                 u64       text_end;
>> -       } sched, lock;
>> +       } sched, lock, traceiter, trace;
>>         pid_t             *current_tid;
>>         size_t            current_tid_sz;
>>         union { /* Tool specific area */
>> --
>> 2.39.2
>>