https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62025
--- Comment #15 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
So, what about this completely untested fix? Unfortunately the scheduler
change has been committed with no testcases. I guess I'll do a
bootstrap/regtest with some printout where this patch changes things and where
we consider it at all.
--- gcc/sched-deps.c.jj 2014-08-06 10:34:13.000000000 +0200
+++ gcc/sched-deps.c 2014-08-12 14:12:06.625193731 +0200
@@ -4751,6 +4751,24 @@ find_inc (struct mem_inc_info *mii, bool
"inc conflicts with store failure.\n");
goto next;
}
+
+ /* The inc instruction could have clobbers, make sure those
+ registers are not used in mem insn. */
+ FOR_EACH_INSN_DEF (def, mii->inc_insn)
+ if (!reg_overlap_mentioned_p (DF_REF_REG (def), mii->mem_reg0))
+ {
+ df_ref use;
+ FOR_EACH_INSN_USE (use, mii->mem_insn)
+ if (reg_overlap_mentioned_p (DF_REF_REG (def),
+ DF_REF_REG (use)))
+ {
+ if (sched_verbose >= 5)
+ fprintf (sched_dump,
+ "inc clobber used in store failure.\n");
+ goto next;
+ }
+ }
+
newaddr = mii->inc_input;
if (mii->mem_index != NULL_RTX)
newaddr = gen_rtx_PLUS (GET_MODE (newaddr), newaddr,