Re: new syscall stub support for ia64 libc

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

>>>>> On Mon, 17 Nov 2003 18:08:03 -0800, Ulrich Drepper <drepper@redhat.com> said:

  Uli> David Mosberger wrote:

  >> The one I mentioned: signal handler gets called in this code
  >> right before the _IO_flockfile():

  >> _IO_FILE *_IO_acquire_lock_file \ __attribute__((cleanup
  >> (_IO_acquire_lock_fct))) \ = (_fp); \ _IO_flockfile
  >> (_IO_acquire_lock_file);

  >> and then the signal handler calls write(), which ends up getting
  >> cancelled.  What prevents this from happening?

  Uli> Why should it be prevented?  If you call write in a signal
  Uli> handler you either disable cancellation of live with it.

Ah, I see.  I guess that's OK as long as programmer's using
pthread_cancel() are aware of this.  Reading the POSIX description, it
would not have occurred to me that calling a cancellable function from
a (asynchronous) signal handler is effectively equivalent to enabling
asynchronous cancellation.

	--david