http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55358
Jakub Jelinek <jakub at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |NEW
Last reconfirmed| |2012-11-28
Ever Confirmed|0 |1
--- Comment #8 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-11-28
14:47:33 UTC ---
(In reply to comment #7)
> diff --git a/gcc/dse.c b/gcc/dse.c
> index f879adb..8d12e8d 100644
> --- a/gcc/dse.c
> +++ b/gcc/dse.c
> @@ -2869,7 +2869,7 @@ dse_step1 (void)
> INSN_UID (s_info->redundant_reason->insn));
> delete_dead_store_insn (ptr);
> }
> - if (s_info)
> + if (ptr->store_rec && s_info)
> s_info->redundant_reason = NULL;
> free_store_info (ptr);
> }
Actually, I fail to see what is that if (s_info) s_info->redundant_reason =
NULL;
good for, the following free_store_info (ptr) should pool_free s_info if
delete_dead_store_insn hasn't already called free_store_info. And it doesn't
seem that pool_alloced store info recs would rely on the field being cleared,
it seems we initialize it always:
store_info->redundant_reason = redundant_reason;
Thus, my preference would be just drop the two lines instead.
