linux-kernel - Re: [PATCH] perf/core: Fix WARN in perf

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <1412f058-17e2-4345-a8ad-e8c968d57184@I-love.SAKURA.ne.jp>
Date: Wed, 9 Jul 2025 19:47:29 +0900
From: Tetsuo Handa <penguin-kernel@...ove.SAKURA.ne.jp>
To: Peter Zijlstra <peterz@...radead.org>
Cc: syzbot <syzbot+2fe61cb2a86066be6985@...kaller.appspotmail.com>,
        linux-perf-users@...r.kernel.org, LKML <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] perf/core: Fix WARN in perf_sigtrap()

On 2025/07/09 18:44, Peter Zijlstra wrote:
> On Sat, Jul 05, 2025 at 11:43:37PM +0900, Tetsuo Handa wrote:
>> Since commit 4f6fc7821283 ("perf: Fix sample vs do_exit()") has moved
>> perf_event_exit_task() call from after exit_task_work() to before
>> exit_task_work(), 
> 
>> task_work_add() from perf_event_exit_task() now returns
> 
> There is no task_work_add() in perf_event_exit_task().

Since the function which triggers BUG_ON() is perf_sigtrap(), I guessed
that the location that queued the task work is task_work_add() in
__perf_event_overflow(). But since testing again with

--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -10324,6 +10324,7 @@ static int __perf_event_overflow(struct perf_event *event,
 
                if (!event->pending_work &&
                    !task_work_add(current, &event->pending_task, notify_mode)) {
+                       BUG_ON(event->ctx && event->ctx->task == TASK_TOMBSTONE);
                        event->pending_work = pending_id;
                        local_inc(&event->ctx->nr_no_switch_fast);
                        WARN_ON_ONCE(!atomic_long_inc_not_zero(&event->refcount));

did not trigger, it seems that __perf_event_overflow() is not called from
perf_event_exit_task() path.



> Did you mean to say something like:
> 
> Because exit_task_work() now runs after perf_event_exit_task(), it is
> possible for an already queued perf_pending_task()->perf_sigtrap() to
> observe a dead task context.

Yes.



> Also, isn't it better to simply swap the early exit tests in that
> function like so:

Yes, that will be OK if you prefer it
( https://lkml.kernel.org/r/ed888189-dad4-47e1-bfc8-4f2213eda32d@I-love.SAKURA.ne.jp ).