------- Additional Comments From amylaar at gcc dot gnu dot org 2005-06-20 13:05 ------- (In reply to comment #4) > 2005-05-31 Pat Haugen <[EMAIL PROTECTED]> > > * loop.c (loop_invariant_p, valid_initial_value_p): Revert last > change. > > > The last change was: > 2005-05-30 Pat Haugen <[EMAIL PROTECTED]> > > * loop.c (loop_invariant_p, valid_initial_value_p): Use > regs_invalidated_by_call instead of call_used_regs. >
The failure that I analyzed involved killing stores to memory, not fixed registers. Therefore, the underlying problem has not been fixed, it has only become latent (again?). I don't think we should daclare a latent wrong-code bug as fixed. We might change the severity amd/or priority, though. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21848
