------- Comment #11 from anirkko at insel dot ch 2007-04-30 09:41 -------
(In reply to comment #10)
> > Therefore, I still think there is a bug in -O3. (alternatively, in the
> > combination of the two). Would you consider reopening this bug if it was
> > -O3?
> > (In this case, I might re-run the bootstrap with only '-O3', next weekend).
>
> Yes, I'm going to reopen it, but I'm not sure someone will tackle it.
>
Testing several combinations, it turns out the bug is indeed in '-O3' alone:
Machine: Ultrasparc
Base compiler: gcc-4.1.2 (bootstrapped with -O2)
Using this setup, bootstrapping
-works with '-O2'
-works with '-O2 -mcpu=supersparc -no-app-regs'
-FAILS with '-O3'
-FAILS with '-O3 -mcpu=supersparc -no-app-regs'
(exactly same ICE as reported in the first place, above)
Thus, it is reproduced several times, both with gcc-3.4.3 and gcc-4.1.2 itself
as base compiler.
Would it help to individually test the various sub-options that -O3
switches on? If you could give me these sub-options ranked according
to 'suspectedness' then I could run the script to test them one by one.
Reason: if the responsible sub-option is identified, it could be discarded from
the -O3 defaults for being tested and found unsafe, even if no one has time
to tackle/mend it.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31523
