This is the mail archive of the gcc-bugs@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[Bug tree-optimization/56294] BOOT_CFLAGS='-O2 -g -fno-ipa-sra' leads to bootstrap comparison failure


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56294

--- Comment #10 from Martin Jambor <jamborm at gcc dot gnu.org> 2013-02-23 11:50:46 UTC ---
Created attachment 29529
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=29529
Another unrelated issue

Irrespective of ho we decide to deal with the first issue, there is
another problem that hinders bootstrap with IPA-SRA disabled of
compilation of tree-ssa-loop-ivopts.c with -fno-ipa-sra and
-fcompare-debug.

I have reduced the issue with multidelta to the attached pre-processed
form that fails with O2 -fno-ipa-sra -fcompare-debug and started
failing with revision 195015:

    2013-01-08  Martin Jambor  <mjambor@suse.cz>

        PR debug/55579
        * tree-sra.c (analyze_access_subtree): Return true also after
        potentially creating a debug-only replacement.

As far I can see, dumps begin to diverge in the release_ssa early pass
where, in the -g case, a control_var SSA names gets a version number
28 instead of 27 - that is also the problematic difference in the
final dump.  So far I have not figured out what is the one extra live
SSA_NAME that lives in the debug case nor I have observed anything
apparently wrong in the behavior of early SRA.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]