Re: rthreads patch #7: redirect kill(getpid(), sig) to the current thread

On Tue, May 6, 2008 at 6:46 AM, Ted Unangst <ted.unangst@gmail.com> wrote:
> On 5/6/08, Philip Guenther <guenther@gmail.com> wrote:
> > That's true of pthreads, but not of rthreads. Currently with
> > rthreads, kill(getpid(),sig) will send the signal to the main thread,
> > period. If the main thread has the signal blocked then it'll be left
> > pending there, even if the current thread doesn't have it blocked.
>
> oops. I forget now if that was intentional or just implementation
> fallout, but Mark's right, we should fix this in the kernel. The
> library is the wrong place to workaround this.


Yep. To quote myself:
> There has to be some way to differentiate between
> kill(getpid(), sig);
> and
> raise(sig); [in the original thread]
>
> in the kernel. There are two ways to do that:
> 1) give the main thread an id distinct from that of the process, or
> 2) use distinct system calls for pthread_kill() and kill().
>
> The former has the advantage of being generic for any other cases
> where there may be a pid vs thread-id distinction. The latter would
> convey the intent of the call more clearly, perhaps simplifying the
> checks in each case.

What's your preference?

Philip Guenther