posix-cpu-timers: fix race between handle_posix_cpu_timers() and posix_cpu_timer_del()
A message that is not in the archive (sent off-list or not yet received): CAGCho0V0x_Y2+vg5G8-r45Xc6uftLbZK5K1=vpavd_4783fogQ@mail.gmail.com
PATCH 1/1 posix-cpu-timers: fix race between handle_posix_cpu_timers() and posix_cpu_timer_del()
If an exiting non-autoreaping task has already passed exit_notify() and calls handle_posix_cpu_timers() from IRQ, it can be reaped by its parent or debugger right after unlock_task_sighand().
If a concurrent posix_cpu_timer_del() runs at that moment, it won't be able to detect timer->it.cpu.firing != 0: cpu_timer_task_rcu() and/or lock_task_sighand() will fail.
Add the tsk->exit_state check into run_posix_cpu_timers() to fix this.
This fix is not needed if CONFIG_POSIX_CPU_TIMERS_TASK_WORK=y, because exit_task_work() is called before exit_notify(). But the check still makes sense, task_work_add(&tsk->posix_cputimers_work.work) will fail anyway in this case.
--- kernel/time/posix-cpu-timers.c | 9 +++++++++ 1 file changed, 9 insertions(+)
kernel/time/posix-cpu-timers.c +9 -0
@@ -1405,6 +1405,15 @@ void run_posix_cpu_timers(void)lockdep_assert_irqs_disabled (); +/* +* Ensure that release_task(tsk) can't happen while +* handle_posix_cpu_timers() is running. Otherwise, a concurrent +* posix_cpu_timer_del() may fail to lock_task_sighand(tsk) and +* miss timer->it.cpu.firing != 0. +*/ +if (tsk->exit_state) +return ; +/* * If the actual expiry is deferred to task work context and the * work is already scheduled there is no point to do anything here.
Signature · 1 line
2.25.1.362.g51ebf55
Quoted 2 lines · Oleg Nesterov
Add the
tsk->exit_statecheck intorun_posix_cpu_timers()to fix this.
Thanks, I'll just apply this directly since I've been cc'd on the whole discussion anyway.
Linus
Quoted 15 lines · Oleg Nesterov
If an exiting non-autoreaping task has already passed
exit_notify()and callshandle_posix_cpu_timers()from IRQ, it can be reaped by its parent or debugger right afterunlock_task_sighand().If a concurrent
posix_cpu_timer_del()runs at that moment, it won't be able to detecttimer->it.cpu.firing!= 0:cpu_timer_task_rcu()and/orlock_task_sighand()will fail.Add the
tsk->exit_statecheck intorun_posix_cpu_timers()to fix this.This fix is not needed if
CONFIG_POSIX_CPU_TIMERS_TASK_WORK=y, becauseexit_task_work()is called beforeexit_notify(). But the check still makes sense,task_work_add(&tsk->posix_cputimers_work.work)will fail anyway in this case.
Quoted 4 lines
Reported-by Benoît Sevens <[email protected]>Fixes 0bdd2ed4138e ("sched: run_posix_cpu_timers: Don't check ->exit_state, use lock_task_sighand()")Signed-off-by Oleg Nesterov <[email protected]>
Linus, I assume you take it directly or do you want me to play the intermediary?
Thanks,
tglx
Quoted 3 lines · Thomas Gleixner
Linus, I assume you take it directly or do you want me to play the intermediary?
I already took it just because it was easy and I was following this discussion anyway.
Linus