Re: new syscall stub support for ia64 libc

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

>>>>> On Mon, 17 Nov 2003 10:12:22 -0800, Ulrich Drepper <drepper@redhat.com> said:

  Uli> But we already did unwind through signal handlers.  It must be
  Uli> something new your patch adds.  So likely the code in
  Uli> _dl_sysinfo_break, the .altrp etc.  If a signal is received the
  Uli> thread is usually in a syscall so we have to unwind through the
  Uli> signal handler frame, the sigreturn stuff, and then the frame
  Uli> around the break instruction.  It must be the step from
  Uli> sigreturn to _dl_sysinfo_break, everything else is the same.

It's likely altrp but the ia64 unwinder in libgcc is rather broken so
it could be a more complicated interaction (e.g., the unwinder won't
work completely correctly with label_state/copy_state which is
something the signal trampoline uses).  In any case, my first priority
is to get it to work with libunwind, since it's much easier to debug
the code with it.  Once that's working, I'll try to figure out what's
holding up GCC's unwinder.  If it's reasonably easy to fix, I'll make
a patch.  Otherwise, we may just want to give up on the builtin
unwinder (it has already has at least one or two serious bugs, which
would be hard to fix).

	--david