This is the mail archive of the
mailing list for the GCC project.
Re: powerpc-unknown-linux-gnu bootstrap fix
- From: Geoff Keating <geoffk at geoffk dot org>
- To: David Edelsohn <dje at watson dot ibm dot com>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: 15 May 2003 15:46:58 -0700
- Subject: Re: powerpc-unknown-linux-gnu bootstrap fix
- References: <firstname.lastname@example.org> <200305141720.NAA29554@makai.watson.ibm.com>
David Edelsohn <email@example.com> writes:
> >>>>> Richard Henderson writes:
> Richard> On Wed, May 14, 2003 at 01:03:12PM -0400, David Edelsohn wrote:
> >> Richard, would it be better to change g_switch_value to unsigned
> >> int or unsigned HOST_WIDE_INT?
> Richard> Yes.
> I just noticed that this is fairly ugly. Half the time
> g_switch_value is tested against SIZE (unsigned HOST_WIDE_INT) and the
> other half again int_size_in_bytes (signed HOST_WIDE_INT). It's not going
> to be easy to reliably rationalize this either way. Maybe we should just
> punt with a cast in sysv4.h -- alpha.c does that already. Comments?
I think g_switch_value should be unsigned, because the reason
int_size_in_bytes is signed is that it can return -1 meaning "I
dunno", and that's a special case that the port maintainer should take
care to handle specially.
- Geoffrey Keating <firstname.lastname@example.org>