[PATCH v2 11/19] nptl: Use tidlock when accessing TID on pthread_setaffinity

[Florian Weimer]

* Adhemerval Zanella:

> Checked on x86_64-linux-gnu.
> ---
>  nptl/pthread_setaffinity.c | 22 +++++++++++++++-------
>  1 file changed, 15 insertions(+), 7 deletions(-)
>
> diff --git a/nptl/pthread_setaffinity.c b/nptl/pthread_setaffinity.c
> index 3bfdc63e19..beb61836a6 100644
> --- a/nptl/pthread_setaffinity.c
> +++ b/nptl/pthread_setaffinity.c
> @@ -27,15 +27,23 @@ int
>  __pthread_setaffinity_new (pthread_t th, size_t cpusetsize,
>  			   const cpu_set_t *cpuset)
>  {
> -  const struct pthread *pd = (const struct pthread *) th;
> -  int res;
> +  struct pthread *pd = (struct pthread *) th;
>  
> -  res = INTERNAL_SYSCALL_CALL (sched_setaffinity, pd->tid, cpusetsize,
> -			       cpuset);
> +  /* Block all signal, since the lock is recursive and used on pthread_cancel
> +     (which should be async-signal-safe).  */
> +  sigset_t oldmask;
> +  __libc_signal_block_all (&oldmask);
> +  lll_lock (pd->tidlock, LLL_PRIVATE);
>  
> -  return (INTERNAL_SYSCALL_ERROR_P (res)
> -	  ? INTERNAL_SYSCALL_ERRNO (res)
> -	  : 0);
> +  int res = pd->tid == 0
> +	    ? ESRCH
> +	    : INTERNAL_SYSCALL_CALL (sched_setaffinity, pd->tid, cpusetsize,
> +				     cpuset);
> +
> +  lll_unlock (pd->tidlock, LLL_PRIVATE);
> +  __libc_signal_restore_set (&oldmask);
> +
> +  return res;

Same issue regarding ESRCH and pthread_cancel.  Here we can just return
0 in case the thread is gone, I think.

Thanks,
Florian