Command line: $ gcc -O2 -fgraphite-identity -fprofile-generate testcase.c or $ gcc -O -funswitch-loops -fgraphite-identity -fprofile-generate testcase.c Compiler output: $ gcc -O -funswitch-loops -fgraphite-identity -fprofile-generate testcase.c testcase.c: In function 'extend_options': testcase.c:2:1: internal compiler error: in loop_entry_phi_arg, at graphite-sese-to-poly.c:89 Please submit a full bug report, with preprocessed source if appropriate. See <http://gcc.gnu.org/bugs.html> for instructions. Tested revisions: r161383 - crash r158095 - crash r153685 - OK 4.5 r158978 - crash 4.4 r160770 - OK (with checking)
Created attachment 21011 [details] reduced testcase (from gcc.dg/pr28935.c) Command line: $ gcc -O -funswitch-loops -fgraphite-identity -fprofile-generate pr44676.c
Mine. On trunk I am able to reproduce the ICE. This is fixed in the Graphite branch, probably by the cleanup of the IV canonicalization. I will investigate a little bit more what happens in trunk, and if this is really fixed by my changes in the Graphite branch I will close this PR when these changes will be committed to trunk. Sebastian
GCC 4.5.1 is being released, adjusting target milestone.
This seems to have disappeared between r161659 and r163636. I can't reproduce it anymore with r166868/x86_64-linux.
Git bisect points out that on trunk this was fixed by http://gcc.gnu.org/viewcvs?view=revision&revision=163105
Author: spop Date: Tue Dec 7 15:18:39 2010 New Revision: 167542 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=167542 Log: Fix PR44676: fix invariant phi node removal. Backport from trunk http://gcc.gnu.org/viewcvs?view=revision&revision=163105 2010-05-07 Sebastian Pop <sebastian.pop@amd.com> Backport from mainline: 2010-05-07 Sebastian Pop <sebastian.pop@amd.com> PR tree-optimization/44676 * graphite-sese-to-poly.c (loop_entry_phi_arg): Renamed phi_arg_in_outermost_loop. (remove_simple_copy_phi): Call phi_arg_in_outermost_loop. (remove_invariant_phi): Same. Modified: branches/gcc-4_5-branch/gcc/ChangeLog branches/gcc-4_5-branch/gcc/graphite-sese-to-poly.c
Fixed.
Missing testcases for both trunk and 4.5 branch.
Author: spop Date: Tue Dec 7 16:27:28 2010 New Revision: 167554 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=167554 Log: Add missing testcase for PR44676. 2010-12-07 Sebastian Pop <sebastian.pop@amd.com> PR tree-optimization/44676 * gcc.dg/graphite/id-pr44676.c: New. Added: trunk/gcc/testsuite/gcc.dg/graphite/id-pr44676.c Modified: trunk/gcc/testsuite/ChangeLog
Author: spop Date: Tue Dec 7 16:32:59 2010 New Revision: 167555 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=167555 Log: Add missing testcase for PR44676. 2010-12-07 Sebastian Pop <sebastian.pop@amd.com> PR tree-optimization/44676 * gcc.dg/graphite/id-pr44676.c: New. Added: branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/graphite/id-pr44676.c Modified: branches/gcc-4_5-branch/gcc/testsuite/ChangeLog