Since the process wide cputime counter is started locklessly from
posix_cpu_timer_rearm(), it can be concurrently stopped by operations
on other timers from the same thread group, such as in the following
unlucky scenario:
         CPU 0                                CPU 1
         -----                                -----
                                           timer_settime(TIMER B)
   posix_cpu_timer_rearm(TIMER A)
       cpu_clock_sample_group()
           (pct->timers_active already true)
                                           handle_posix_cpu_timers()
                                               check_process_timers()
                                                   stop_process_timers()
                                                       pct->timers_active = false
       arm_timer(TIMER A)
   tick -> run_posix_cpu_timers()
       // sees !pct->timers_active, ignore
       // our TIMER A
Fix this with simply locking process wide cputime counting start and
timer arm in the same block.
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Fixes: 60f2ceaa8111 ("posix-cpu-timers: Remove unnecessary locking around cpu_clock_sample_group")
Cc: stable@vger.kernel.org
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Eric W. Biederman <ebiederm@xmission.com>
        if (!p)
                goto out;
 
+       /* Protect timer list r/w in arm_timer() */
+       sighand = lock_task_sighand(p, &flags);
+       if (unlikely(sighand == NULL))
+               goto out;
+
        /*
         * Fetch the current sample and update the timer's expiry time.
         */
 
        bump_cpu_timer(timer, now);
 
-       /* Protect timer list r/w in arm_timer() */
-       sighand = lock_task_sighand(p, &flags);
-       if (unlikely(sighand == NULL))
-               goto out;
-
        /*
         * Now re-arm for the new expiry time.
         */