Re: [PATCH] posix_cpu_timers: fix timer never expires when executesclock_nanosleep

From: Stanislaw Gruszka
Date: Thu Aug 01 2013 - 07:28:04 EST


Hi Dong Zhu

On Thu, Aug 01, 2013 at 06:10:19PM +0800, Dong Zhu wrote:
> diff --git a/kernel/posix-cpu-timers.c b/kernel/posix-cpu-timers.c
> index c7f31aa..cc03290 100644
> --- a/kernel/posix-cpu-timers.c
> +++ b/kernel/posix-cpu-timers.c
> @@ -1413,9 +1413,9 @@ static int posix_cpu_nsleep(const clockid_t which_clock, int flags,
> /*
> * Diagnose required errors first.
> */
> - if (CPUCLOCK_PERTHREAD(which_clock) &&
> - (CPUCLOCK_PID(which_clock) == 0 ||
> - CPUCLOCK_PID(which_clock) == current->pid))
> + if (CPUCLOCK_PID(which_clock) == current->pid ||
> + (CPUCLOCK_PERTHREAD(which_clock) &&
> + CPUCLOCK_PID(which_clock) == 0))
> return -EINVAL;

Nope, this is wrong. We have to allow own pid process clock, because it
can be used correctly on multi-threaded processes. Own tid thread clock
has no sense and we correctly return -EINVAL in such case.

We could possibly add check for own pid together with check if process
consist of one thread, but that is too complicated IMHO especially
taking into account that threads on the process can be destroyed and
created dynamically.

Stanislaw

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/