Re: Fw: 2.5.61 oops running SDET

From: Linus Torvalds (torvalds@transmeta.com)
Date: Sat Feb 15 2003 - 20:35:21 EST


On Sat, 15 Feb 2003, Andrew Morton wrote:
>
> The recent change to fs/proc/array.c:task_sig()?
>
> buffer += sprintf(buffer, "ShdPnd:\t");
> buffer = render_sigset_t(&p->signal->shared_pending.signal, buffer);

Yeah, but I think the bug has existed for much longer.

It looks like "proc_pid_status()" doesn't actually lock the task at all,
nor even bother to test whether the task has signal state. Which has
_always_ been a bug. I don't know why it would start happening now, but it
might just be unlucky timing.

I think the proper fix is to put a

        task_lock()
        ..
        task_unlock()

around the whole proc_pid_status() function, _and_ then verify that
"tsk->sighand" is non-NULL.

(Oh, careful, that's already what "get_task_mm()" does internally, so look
out for deadlocks - you'd need to open-code the get_task_mm() in there
too, so the end result is something like

        task_lock(task)
        if (task->mm) {
                .. mm state
        }
        if (task->sighand) {
                .. signal state
        }
        ..
        task_unlock(task);

instead).

                Linus

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



This archive was generated by hypermail 2b29 : Sat Feb 15 2003 - 22:01:05 EST