This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug target/45670] [4.6 Regression] Less efficient x86 addressing mode selection on 4.6, causes -Os size regression from 4.5
- From: "ebotcazou at gcc dot gnu.org" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: Fri, 5 Nov 2010 11:17:27 +0000
- Subject: [Bug target/45670] [4.6 Regression] Less efficient x86 addressing mode selection on 4.6, causes -Os size regression from 4.5
- Auto-submitted: auto-generated
- References: <bug-45670-4@http.gcc.gnu.org/bugzilla/>
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45670
--- Comment #5 from Eric Botcazou <ebotcazou at gcc dot gnu.org> 2010-11-05 11:17:11 UTC ---
> Well, it is actually very much related to that reversion. While it didn't
> regress in between 161907 and 162617, it regressed from the MEM_REF merge till
> 161906 and from 162618 onwards.
Yes, that isn't different from what I said, it had been caused by something
else.
> The problem is that due to the MEM load combiner isn't able to fix this up, so
> if we don't get it right during expansion, we are out of luck. Looking at it
> now...
Thanks!