This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug bootstrap/52674] [4.7 regression] segfault during profiled LTO bootstrap
- 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, 23 Mar 2012 21:30:26 +0000
- Subject: [Bug bootstrap/52674] [4.7 regression] segfault during profiled LTO bootstrap
- Auto-submitted: auto-generated
- References: <bug-52674-4@http.gcc.gnu.org/bugzilla/>
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52674
--- Comment #3 from Eric Botcazou <ebotcazou at gcc dot gnu.org> 2012-03-23 21:30:26 UTC ---
> I was doing profiledbootstrap with the bootstrap-lto config on x64 just fine
> until about 6 weeks ago. The speedup in the resulting binary was notable, and
> was very useful to have for my shorter gcc build iterations where I use
> --disable-bootstrap. If it's not supposed to work, then configure should not
> allow it IMO.
The speedup comes the profiled bootstrap, LTO doesn't add much. The average
user isn't supposed to use --with-build-config, it's mainly for GCC developers.
> If you can recommend an intermediate make target I can use to make the
> iteration time shorter, I can try and debug this issue further on my own and
> report back. I can't figure out how to capture the resulting binary that
> crashes, since the source isn't listed in the config.log when it fails, so any
> help on that front would be appreciated.
Right, debugging LTO is a pain. Do not waste your time on this I'd say, just
do a simple profiled bootstrap.