------- Comment #3 from vmakarov at redhat dot com 2008-05-06 19:11 ------- Thanks for decreasing the test case. The problem was in wrong mode of save/restore stack slot when it was reused for another hard register. It resulted in generation of wrong save/restore insn and after that in a wrong secondary reload insn needed for this insn and the last insn caused a crash in the postreload pass.
The patch solving the problem will follow soon. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36028