This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [patch] i386.c: Fix for PR 10795.
On Tue, 2003-07-15 at 06:46, Kazu Hirata wrote:
> Another reason is that one of the if conversions runs after the
> register allocation, and I wasn't sure if force_reg() can be used
> after register allocation.
I am not sure if this is possible. It looks like the ix86_expand_*
functions are only called from define_expand patterns, which are only
used in initial RTL generation, and hence things like force_reg are
safe. Looking at other related functions, I see quite a few that call
force_reg and/or force_operand.
If this does happen, there are ways to work around it. For instance,
checking reload_completed and doing something different if true.
I see only one counterexample, which is that ix86_expand_fp_movcc is
called from a post-reload splitter, but that function has a check for
that case.
--
Jim Wilson, GNU Tools Support, http://www.specifix.com