Re: new syscall stub support for ia64 libc

[David Mosberger <davidm at napali dot hpl dot hp dot com>]

>>>>> On Mon, 17 Nov 2003 16:53:10 -0800, Ulrich Drepper <drepper@redhat.com> said:

  Uli> David Mosberger wrote:

  >> Or is NPTL simply not making any correctness-guarantees when
  >> PTHREAD_CANCEL_ASYNCHRONOUS is in effect?

  Uli> It is not allowed to call any library function except for
  Uli> pthread_cancel(), pthread_setcancelstate(), and
  Uli> pthread_setcanceltype() (XSH 2.9.5.4).

The HTML version doesn't have section numbers, but I think you're
referring to this section:

  Async-Cancel Safety

  The pthread_cancel(), pthread_setcancelstate(), and
  pthread_setcanceltype() functions are defined to be async-cancel
  safe.

  No other functions in this volume of IEEE Std 1003.1-2001 are
  required to be async-cancel-safe.

I don't see anything that would prevent NPTL from providing better
async-cancel-safety, but juding by your response, it doesn't.

  Uli> If you want to discuss correctness issues I request you first
  Uli> read the POSIX standard.

Hey, I'm trying...

I still don't understand the interaction between signals and
thread-cancellation and I couldn't find where this is being discussed
in the standard.  Any pointers?

BTW: All NPTL-tests now succeed with libunwind, except that
tst-cancel10 occasionaly segfaults due to some sort of memory
corruption/race.  I'll see now if I can fix GCC's unwind-ia64.c to
work also.

Thanks,

	--david