------- Comment #9 from bonzini at gnu dot org 2008-07-14 16:29 -------
In fact, using modified_between_p would not fix the bug. The problem is that
use_killed_between gives the wrong answer *before* it reaches the code that
"redoes" modified_between_p using df data structures.
I'll regtest this patch tomorrow because the machine is right now too busy.
But if anyone wants to beat me, feel free.
Index: ../../peak-gcc-src/gcc/fwprop.c
===================================================================
--- ../../peak-gcc-src/gcc/fwprop.c (revision 136756)
+++ ../../peak-gcc-src/gcc/fwprop.c (working copy)
@@ -480,10 +480,15 @@ use_killed_between (struct df_ref *use,
return true;
/* Check if the reg in USE has only one definition. We already
- know that this definition reaches use, or we wouldn't be here. */
+ know that this definition reaches use, or we wouldn't be here.
+ However, this is not true for hard registers, because if they are
+ live at the beginning of the function it does not mean that we
+ have an uninitialized access. */
regno = DF_REF_REGNO (use);
def = DF_REG_DEF_CHAIN (regno);
- if (def && (def->next_reg == NULL))
+ if (def
+ && (def->next_reg == NULL)
+ && regno >= FIRST_PSEUDO_REGISTER)
return false;
/* Check locally if we are in the same basic block. */
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36753
