[Bug target/27770] New: wrong code in spec tests for -ftree-vectorize -maltivec

["janis at gcc dot gnu dot org" <gcc-bugzilla at gcc dot gnu dot org>]

Two tests from SPEC CPU2000, parser and gzip, fail at runtime when built on
powerpc64-linux with "-O2 -ftree-vectorize -maltivec -mabi=altivec".  parser
fails with both -m32 and -m64, gzip with -m64.  A regression hunt identified
this patch as the start of the failures:

    http://gcc.gnu.org/viewcvs?view=rev&rev=113395

    r113395 | dje | 2006-04-30 19:23:13 +0000 (Sun, 30 Apr 2006)

I've been able to duplicate the failure with 32-bit parser by compiling only
post-process.c with "-m32 -O2 -ftree-vectorize -maltivec -mabi=altivec" and
everything else with "-m32 -O2".  In addition, if all static variables in
post-process.c are made global, the test passes, but with only word_links and
visited left static it fails. I can continue to work on coming up with a
self-contained test case, but perhaps this is enough information; let me know.


-- 
           Summary: wrong code in spec tests for -ftree-vectorize -maltivec
           Product: gcc
           Version: 4.2.0
            Status: UNCONFIRMED
          Keywords: wrong-code
          Severity: normal
          Priority: P3
         Component: target
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: janis at gcc dot gnu dot org
GCC target triplet: powerpc-linux


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27770