This is the mail archive of the
mailing list for the GCC project.
Re: fix alpha bootstrap
- From: Jan Hubicka <hubicka at ucw dot cz>
- To: Richard Henderson <rth at twiddle dot net>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Fri, 20 May 2005 09:57:02 +0200
- Subject: Re: fix alpha bootstrap
- References: <20050520024216.GA15794@twiddle.net>
> This patch,
> 2005-05-19 Jan Hubicka <email@example.com>
> * basic-block.h (REG_BR_PROB_BASE): Define.
> * rtl.h (REG_BR_PROB_BASE): Kill.
> broke alpha bootstrap, in that REG_BR_PROB_BASE was no longer available
> in the generated insn-emit.c. Rather than revert this one part, or
> include basic-block.h into insn-emit.c, I took the opportunity to do
> some code sharing and simplification.
Oops, thanks for fixing this!
REG_BR_PROB_BASE is now definitly more related to profile than it is to
RTL, so it seemed to make sense to move the definition around (it might
make sense to rename it too, i guess I will do that in stage 3 along
with the other renaming of cfun accestors), but I forgot that there
might be .md code dealing with profile.
For a while I would like to make the profile information available for
insn splitters/output templates so code size/performance choice can be
done based on hotness of the basic block insn is in, so we might end up
including basic-block.h there anyway. (I am not sure about the
interface here, perhaps simply adding bb argument to splitter/output
tempaltes might work, but some mds still destroy CFG before final pass)