Message-ID: <aQUHXotIjne3vHm_@google.com>
Date: Fri, 31 Oct 2025 12:00:46 -0700
From: Namhyung Kim <namhyung@...nel.org>
To: Ian Rogers <irogers@...gle.com>
Cc: Shuai Xue <xueshuai@...ux.alibaba.com>,
	alexander.shishkin@...ux.intel.com, peterz@...radead.org,
	james.clark@....com, leo.yan@...aro.org, mingo@...hat.com,
	baolin.wang@...ux.alibaba.com, acme@...nel.org,
	mark.rutland@....com, jolsa@...nel.org, adrian.hunter@...el.com,
	linux-perf-users@...r.kernel.org, linux-kernel@...r.kernel.org,
	nathan@...nel.org, bpf@...r.kernel.org
Subject: Re: [PATCH] perf record: skip synthesize event when open evsel failed

Hello,

On Fri, Oct 31, 2025 at 09:04:38AM -0700, Ian Rogers wrote:
> On Thu, Oct 30, 2025 at 7:36 PM Shuai Xue <xueshuai@...ux.alibaba.com> wrote:
> >
> > 在 2025/10/31 01:32, Ian Rogers 写道:
> > > On Wed, Oct 29, 2025 at 5:55 AM Shuai Xue <xueshuai@...ux.alibaba.com> wrote:
> > >>
> > >>
> > >>
> > >> 在 2025/10/24 10:45, Shuai Xue 写道:
> > >>>
> > >>>
> > >>> 在 2025/10/24 00:08, Ian Rogers 写道:
> > >>>> On Wed, Oct 22, 2025 at 6:50 PM Shuai Xue <xueshuai@...ux.alibaba.com> wrote:
> > >>>>>
> > >>>>> When using perf record with the `--overwrite` option, a segmentation fault
> > >>>>> occurs if an event fails to open. For example:
> > >>>>>
> > >>>>>     perf record -e cycles-ct -F 1000 -a --overwrite
> > >>>>>     Error:
> > >>>>>     cycles-ct:H: PMU Hardware doesn't support sampling/overflow-interrupts. Try 'perf stat'
> > >>>>>     perf: Segmentation fault
> > >>>>>         #0 0x6466b6 in dump_stack debug.c:366
> > >>>>>         #1 0x646729 in sighandler_dump_stack debug.c:378
> > >>>>>         #2 0x453fd1 in sigsegv_handler builtin-record.c:722
> > >>>>>         #3 0x7f8454e65090 in __restore_rt libc-2.32.so[54090]
> > >>>>>         #4 0x6c5671 in __perf_event__synthesize_id_index synthetic-events.c:1862
> > >>>>>         #5 0x6c5ac0 in perf_event__synthesize_id_index synthetic-events.c:1943
> > >>>>>         #6 0x458090 in record__synthesize builtin-record.c:2075
> > >>>>>         #7 0x45a85a in __cmd_record builtin-record.c:2888
> > >>>>>         #8 0x45deb6 in cmd_record builtin-record.c:4374
> > >>>>>         #9 0x4e5e33 in run_builtin perf.c:349
> > >>>>>         #10 0x4e60bf in handle_internal_command perf.c:401
> > >>>>>         #11 0x4e6215 in run_argv perf.c:448
> > >>>>>         #12 0x4e653a in main perf.c:555
> > >>>>>         #13 0x7f8454e4fa72 in __libc_start_main libc-2.32.so[3ea72]
> > >>>>>         #14 0x43a3ee in _start ??:0
> > >>>>>
> > >>>>> The --overwrite option implies --tail-synthesize, which collects non-sample
> > >>>>> events reflecting the system status when recording finishes. However, when
> > >>>>> evsel opening fails (e.g., unsupported event 'cycles-ct'), session->evlist
> > >>>>> is not initialized and remains NULL. The code unconditionally calls
> > >>>>> record__synthesize() in the error path, which iterates through the NULL
> > >>>>> evlist pointer and causes a segfault.
> > >>>>>
> > >>>>> To fix it, move the record__synthesize() call inside the error check block, so
> > >>>>> it's only called when there was no error during recording, ensuring that evlist
> > >>>>> is properly initialized.
> > >>>>>
> > >>>>> Fixes: 4ea648aec019 ("perf record: Add --tail-synthesize option")
> > >>>>> Signed-off-by: Shuai Xue <xueshuai@...ux.alibaba.com>
> > >>>>
> > >>>> This looks great! I wonder if we can add a test, perhaps here:
> > >>>> https://web.git.kernel.org/pub/scm/linux/kernel/git/perf/perf-tools-next.git/tree/tools/perf/tests/shell/record.sh?h=perf-tools-next#n435
> > >>>> something like:
> > >>>> ```
> > >>>> $ perf record -e foobar -F 1000 -a --overwrite -o /dev/null -- sleep 0.1
> > >>>> ```
> > >>>> in a new test subsection for test_overwrite? foobar would be an event
> > >>>> that we could assume isn't present. Could you help with a test
> > >>>> covering the problems you've uncovered and perhaps related flags?
> > >>>>
> > >>>
> > >>> Hi, Ian,
> > >>>
> > >>> Good suggestion, I'd like to add a test. But foobar may not a good case.
> > >>>
> > >>> Regarding your example:
> > >>>
> > >>>     perf record -e foobar -a --overwrite -o /dev/null -- sleep 0.1
> > >>>     event syntax error: 'foobar'
> > >>>                          \___ Bad event name
> > >>>
> > >>>     Unable to find event on a PMU of 'foobar'
> > >>>     Run 'perf list' for a list of valid events
> > >>>
> > >>>      Usage: perf record [<options>] [<command>]
> > >>>         or: perf record [<options>] -- <command> [<options>]
> > >>>
> > >>>         -e, --event <event>   event selector. use 'perf list' to list available events
> > >>>
> > >>>
> > >>> The issue with using foobar is that it's an invalid event name, and the
> > >>> perf parser will reject it much earlier. This means the test would exit
> > >>> before reaching the part of the code path we want to verify (where
> > >>> record__synthesize() could be called).
> > >>>
> > >>> A potential alternative could be testing an error case such as EACCES:
> > >>>
> > >>>     perf record -e cycles -C 0 --overwrite -o /dev/null -- sleep 0.1
> > >>>
> > >>> This could reproduce the scenario of a failure when attempting to access
> > >>> a valid event, such as due to permission restrictions. However, the
> > >>> limitation here is that users may override
> > >>> /proc/sys/kernel/perf_event_paranoid, which affects whether or not this
> > >>> test would succeed in triggering an EACCES error.
> > >>>
> > >>>
> > >>> If you have any other suggestions or ideas for a better way to simulate
> > >>> this situation, I'd love to hear them.
> > >>>
> > >>> Thanks.
> > >>> Shuai
> > >>
> > >> Hi, Ian,
> > >>
> > >> Gentle ping.
> > >
> > > Sorry, for the delay. I was trying to think of a better way given the
> > > problems you mention and then got distracted. I wonder if a legacy
> > > event that core PMUs never implement would be a good candidate to
> > > test. For example, the event "node-prefetch-misses" is for "Local
> > > memory prefetch misses" but the memory controller tends to be a
> > > separate PMU and this event is never implemented to my knowledge.
> > > Running this locally I see:
> > >
> > > ```
> > > $ perf record -e node-prefetch-misses -a --overwrite -o /dev/null -- sleep 0.1
> > > Lowering default frequency rate from 4000 to 1750.
> > > Please consider tweaking /proc/sys/kernel/perf_event_max_sample_rate.
> > > Error:
> > > Failure to open event 'cpu_atom/node-prefetch-misses/' on PMU
> > > 'cpu_atom' which will be removed.
> > > No fallback found for 'cpu_atom/node-prefetch-misses/' for error 2
> > > Error:
> > > Failure to open event 'cpu_core/node-prefetch-misses/' on PMU
> > > 'cpu_core' which will be removed.
> > > No fallback found for 'cpu_core/node-prefetch-misses/' for error 2
> > > Error:
> > > Failure to open any events for recording.
> > > perf: Segmentation fault
> > >     #0 0x55a487ad8b87 in dump_stack debug.c:366
> > >     #1 0x55a487ad8bfd in sighandler_dump_stack debug.c:378
> > >     #2 0x55a4878c6f94 in sigsegv_handler builtin-record.c:722
> > >     #3 0x7f72aae49df0 in __restore_rt libc_sigaction.c:0
> > >     #4 0x55a487b57ef8 in __perf_event__synthesize_id_index
> > > synthetic-events.c:1862
> > >     #5 0x55a487b58346 in perf_event__synthesize_id_index synthetic-events.c:1943
> > >     #6 0x55a4878cb2a3 in record__synthesize builtin-record.c:2150
> > >     #7 0x55a4878cdada in __cmd_record builtin-record.c:2963
> > >     #8 0x55a4878d11ca in cmd_record builtin-record.c:4453
> > >     #9 0x55a48795b3cc in run_builtin perf.c:349
> > >     #10 0x55a48795b664 in handle_internal_command perf.c:401
> > >     #11 0x55a48795b7bd in run_argv perf.c:448
> > >     #12 0x55a48795bb06 in main perf.c:555
> > >     #13 0x7f72aae33ca8 in __libc_start_call_main libc_start_call_main.h:74
> > >     #14 0x7f72aae33d65 in __libc_start_main_alias_2 libc-start.c:128
> > >     #15 0x55a4878acf41 in _start perf[52f41]
> > > Segmentation fault
> > > ```
> >
> >
> > Hi, Ian，
> >
> > Is node-prefetch-misses a platform specific event? Running it on ARM Yitian 710
> > and Intel SPR platform, I see:
> >
> > $sudo perf record -e node-prefetch-misses
> > Error:
> > The node-prefetch-misses event is not supported.
> 
> Hi Shuai,
> 
> So node-prefetch-misses is a legacy event. Perf has a notion of events
> that are inbuilt to the kernel/PMU driver and get special fixed
> encodings. That said, the PMU driver in the kernel can just fail to
> support the events and I think that's uniformly the case for
> node-prefetch-misses. As shown by my reproduction of the crash, which
> I hope this suffices for a test - i.e. it is an event that parses but
> one that is never supported.

Maybe it's platform dependent.  I have no idea what's the best for this
test.  Any uncore event would work as well but it's not standardized.

I'll merge this fix first.

Thanks,
Namhyung

Powered by blists - more mailing lists