https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100263
Jakub Jelinek <jakub at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Last reconfirmed| |2021-04-26
Ever confirmed|0 |1
Status|UNCONFIRMED |NEW
Summary|Wrong removal of statement |[11/12 Regression] Wrong
|in copyprop3 |removal of statement in
| |copyprop3
Target Milestone|--- |11.2
CC| |jakub at gcc dot gnu.org,
| |rguenth at gcc dot gnu.org
Priority|P3 |P2
--- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
In can confirm it is r11-39-gf9e1ea10e657af9fb02fafecf1a600740fd34409
Before lim2 the dump is identical, lim2 dump difference between r11-38 and
r11-39
is:
--- pr100263.c.131t.lim2.r11-38 2021-04-26 15:54:38.623719858 +0200
+++ pr100263.c.131t.lim2.r11-39 2021-04-26 15:54:53.982545411 +0200
@@ -165,26 +165,23 @@ main ()
j_lsm_flag.23_69 = 0;
k_lsm.24_70 = k;
k_lsm_flag.25_71 = 0;
- l_lsm.26_72 = l;
- l_lsm_flag.27_73 = 0;
- c__lsm.20_74 = MEM[(char *)&c + 107B];
- c__lsm_flag.21_75 = 0;
+ l_lsm_flag.27_72 = 0;
+ c__lsm_flag.21_73 = 0;
h.4_5 = h;
- b_lsm.28_76 = b;
- b_lsm_flag.29_77 = 0;
+ b_lsm_flag.29_74 = 0;
<bb 3> [local count: 10321168952]:
# b.12_44 = PHI <b.0_1(18), _32(23)>
- # c__lsm.20_50 = PHI <c__lsm.20_74(18), c__lsm.20_51(23)>
- # c__lsm_flag.21_52 = PHI <c__lsm_flag.21_75(18), c__lsm_flag.21_53(23)>
+ # c__lsm.20_50 = PHI <c__lsm.20_75(D)(18), c__lsm.20_51(23)>
+ # c__lsm_flag.21_52 = PHI <c__lsm_flag.21_73(18), c__lsm_flag.21_53(23)>
# j_lsm.22_54 = PHI <j_lsm.22_68(18), j_lsm.22_55(23)>
# j_lsm_flag.23_56 = PHI <j_lsm_flag.23_69(18), j_lsm_flag.23_57(23)>
# k_lsm.24_58 = PHI <k_lsm.24_70(18), k_lsm.24_59(23)>
# k_lsm_flag.25_60 = PHI <k_lsm_flag.25_71(18), k_lsm_flag.25_61(23)>
- # l_lsm.26_62 = PHI <l_lsm.26_72(18), l_lsm.26_63(23)>
- # l_lsm_flag.27_64 = PHI <l_lsm_flag.27_73(18), l_lsm_flag.27_65(23)>
- # b_lsm.28_66 = PHI <b_lsm.28_76(18), b_lsm.28_101(23)>
- # b_lsm_flag.29_67 = PHI <b_lsm_flag.29_77(18), b_lsm_flag.29_102(23)>
+ # l_lsm.26_62 = PHI <l_lsm.26_76(D)(18), l_lsm.26_63(23)>
+ # l_lsm_flag.27_64 = PHI <l_lsm_flag.27_72(18), l_lsm_flag.27_65(23)>
+ # b_lsm.28_66 = PHI <b_lsm.28_77(D)(18), b_lsm.28_101(23)>
+ # b_lsm_flag.29_67 = PHI <b_lsm_flag.29_74(18), b_lsm_flag.29_102(23)>
if (a.1_7 != 0)
goto <bb 4>; [50.00%]
else
I can reproduce the same problem even when replacing
char *n = &c[3][8][2];
*n = 1;
with
c[3][8][2] = 1;
