Skip to content

Was: Q: sys_perf_event_open() && PF_EXITING

Oleg Nesterov lkml 1 version In review

Get series with b4
b4 am [email protected]

Versions newest first

  1. v1 Latest2 patches 41 repliesIn review

Patches in v1 No applies recorded yet

#Subject and filesSizeRepliesApplied
1/2perf: fix find_get_context() vs perf_event_exit_task() racekernel/perf_event.c+20 -149 replies
2/2perf: fix perf_event_init_task()/perf_event_free_task() interactionkernel/perf_event.c+4 -52 replies

Reviews trailers per patch

All 2 patches have a Reviewed-by or Acked-by.

PatchPeter Zijlstra
1/2A-b A-b
2/2A-b A-b

Trailers given in replies are filled; those carried in the patches from earlier versions are outlined with dashes.

Read the v1 thread

Cover letter [PATCH 0/2] Was: Q: sys_perf_event_open() && PF_EXITING
Quoted 8 lines · Oleg Nesterov

I am puzzled by PF_EXITING check in find_lively_task_by_vpid().

How can it help? The task can call do_exit() right after the check.

And why do we need it? The comment only says "Can't attach events to a dying task". Maybe it tries protect sys_perf_event_open() against perf_event_exit_task_context(), but it can't.

Yes.

Please see 1/2. Well, I can't say I really like the idea to reuse task->perf_event_mutex, but I do not see a better fix.

Also, I have no idea how can I actually test the changes in the code I can hardly understand, please review.

Also. I believe there are more problems in perf_install_event(), but I need to recheck.

Quoted 4 lines

Hmm. jump_label_inc/dec looks obviously racy too. Say, free_event() races with perf_event_alloc(). There is a window between atomic_xxx() and jump_label_update(), afaics it is possible to call jump_label_disable() when perf_task_events/perf_swevent_enabled != 0.

Another issue...

Oleg.