This is the mail archive of the
archer@sourceware.org
mailing list for the Archer project.
Re: ptrace improvement: PTRACE_O_INHERIT
On 02/22, Roland McGrath wrote:
>
> > Unless I missed something again, other threads are visible to wait().
>
> That's what it looks like in the code to me as well. But my recollection
> is otherwise. And IMHO it should be otherwise, that is that MT exec cannot
> be arbitrarily delayed by any userland act.
perhaps, I am not sure right now... but
> As long as the de_thread delay
> is an uninterruptible wait not even affected by SIGKILL, it looks like a
> DoS sort of vector
Aaahhh. good point, and I never thought about this... this is like vfork().
At first glance it is simple to turn this wait into TASK_KILLABLE, if we
stay with the current behaviour.
Oleg.