[Bug middle-end/38271] [4.4 Regression] Spurious / missing ... used uninitialized in this function warning
--- Comment #9 from aoliva at gcc dot gnu dot org 2008-12-10 17:22 --- Subject: Bug 38271 Author: aoliva Date: Wed Dec 10 17:20:50 2008 New Revision: 142651 URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=142651 Log: gcc/ChangeLog: PR middle-end/38271 * tree-sra.c (sra_build_bf_assignment): Avoid warnings for variables initialized from SRAed bit fields. gcc/testsuite/ChangeLog: PR middle-end/38271 * gcc.dg/torture/pr38271.c: New. Added: trunk/gcc/testsuite/gcc.dg/torture/pr38271.c Modified: trunk/gcc/ChangeLog trunk/gcc/testsuite/ChangeLog trunk/gcc/tree-sra.c -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38271
[Bug middle-end/38271] [4.4 Regression] Spurious / missing ... used uninitialized in this function warning
--- Comment #10 from aoliva at gcc dot gnu dot org 2008-12-10 17:30 --- Fixed. -- aoliva at gcc dot gnu dot org changed: What|Removed |Added Status|NEW |RESOLVED Resolution||FIXED http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38271
[Bug middle-end/38271] [4.4 Regression] Spurious / missing ... used uninitialized in this function warning
-- mmitchel at gcc dot gnu dot org changed: What|Removed |Added Priority|P3 |P2 http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38271
[Bug middle-end/38271] [4.4 Regression] Spurious / missing ... used uninitialized in this function warning
--- Comment #8 from d dot g dot gorbachev at gmail dot com 2008-12-08
20:34 ---
(In reply to comment #7)
The patch causes segfault. This is how it happens:
tree-sra.c:1612
for (f = TYPE_FIELDS (elt-type);
f; f = TREE_CHAIN (f))
{
tree-sra.c:1700
if (size != TREE_INT_CST_LOW (TYPE_SIZE (TREE_TYPE (elt-element
return f; // NULL
tree-sra.c:1764
for (f = TYPE_FIELDS (type); f ; f = TREE_CHAIN (f))
if (TREE_CODE (f) == FIELD_DECL)
{
tree last = try_instantiate_multiple_fields (elt, f);
if (last != f)
{
f = last;
continue;
}
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38271
[Bug middle-end/38271] [4.4 Regression] Spurious / missing ... used uninitialized in this function warning
--- Comment #6 from rguenth at gcc dot gnu dot org 2008-12-05 12:57 ---
This is sra_build_assignment/sra_build_bf_assignments way of lowering a
BIT_FIELD_REF to integer arithmetic. We hit the !INTEGRAL_TYPE_P (TREE_TYPE
(var)) paths and for
BIT_FIELD_REF *p_1(D), 32, 0
we create VIEW_CONVERT_EXPR long long (*p) and extract the lower/upper 32
bits by shifting and masking. This obviously uses possibly uninitialized
parts of *p or s0. The problem here is that we interestingly split the
initialization of s0 to the upper part
s0.c ={v} s0$c_7;
and the lower part
SR.3_8 = VIEW_CONVERT_EXPRlong long unsigned int(s0);
SR.4_9 = SR.3_8 0x;
SR.5_10 = (long long unsigned int) s0$B0F32_5;
SR.6_11 = SR.4_9 | SR.5_10;
s0 ={v} VIEW_CONVERT_EXPRstruct xxx(SR.6_11);
this is likely because SRA really wants to have two 32bit fields but it
cannot deal with the larger struct with a VIEW_CONVERT_EXPR.
With just disabling all these code-paths we get
bb 2:
s0$B0F32_4 = 0;
s0$c_5 = p_1(D)-c;
s0$B0F32_6 = BIT_FIELD_REF *p_1(D), 32, 0;
s0.c ={v} s0$c_5;
BIT_FIELD_REF s0, 32, 0 ={v} s0$B0F32_6;
s ={v} s0;
s$a_7 = s.a;
D.1242_2 = s$a_7;
and no warning. But of course we now have bitfield refs that are not optimized
(not scalarizing in this case would be better). Of course even with the
current scalarization we finally arrive with
bb 2:
# VUSE SMT.16_13(D)
s0$c_7 = p_1(D)-c;
# VUSE SMT.16_13(D)
D.1249_4 = VIEW_CONVERT_EXPRlong long unsigned int(*p_1(D));
s0$B0F32_5 = (unsigned int) D.1249_4;
# VUSE s0_14(D)
SR.21_24 = VIEW_CONVERT_EXPRlong long unsigned int(s0);
SR.22_25 = SR.21_24 0x;
# s0_28 = VDEF s0_14(D)
s0 = VIEW_CONVERT_EXPRstruct xxx(SR.22_25);
# s0_29 = VDEF s0_28
s0.c = s0$c_7;
# VUSE s0_29
SR.3_8 = VIEW_CONVERT_EXPRlong long unsigned int(s0);
SR.4_9 = SR.3_8 0x;
SR.5_10 = (long long unsigned int) s0$B0F32_5;
SR.6_11 = SR.5_10 | SR.4_9;
s0$B0F32_21 = (unsigned int) SR.6_11;
s0$c_30 = VIEW_CONVERT_EXPRstruct xxx(SR.6_11).c;
# VUSE s0_29
SR.25_31 = VIEW_CONVERT_EXPRlong long unsigned int(s0);
SR.26_32 = SR.25_31 0x;
SR.27_33 = (long long unsigned int) s0$B0F32_21;
SR.28_34 = SR.27_33 | SR.26_32;
# s0_35 = VDEF s0_29
s0 = VIEW_CONVERT_EXPRstruct xxx(SR.28_34);
# s0_36 = VDEF s0_35
s0.c = s0$c_30;
# VUSE s0_36
# s_18 = VDEF s_17(D)
s = s0;
# VUSE s_18
s$a_37 = s.a;
# s_39 = VDEF s_18
s.a = s$a_37;
# VUSE s_39
# SMT.16_20 = VDEF SMT.16_13(D)
bar (s);
return;
because we seem to scalarize again. Ugh.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38271
[Bug middle-end/38271] [4.4 Regression] Spurious / missing ... used uninitialized in this function warning
--- Comment #7 from rguenth at gcc dot gnu dot org 2008-12-05 13:05 --- The following would make SRA only combine all of the struct or nothing. That avoids these messy situations: Index: tree-sra.c === --- tree-sra.c (revision 142469) +++ tree-sra.c (working copy) @@ -1695,7 +1695,10 @@ try_instantiate_multiple_fields (struct gcc_assert (~alchk align); - /* Create the field group as a single variable. */ + /* Create the field group as a single variable if the group covers + the whole element. */ + if (size != TREE_INT_CST_LOW (TYPE_SIZE (TREE_TYPE (elt-element +return f; /* We used to create a type for the mode above, but size turns to be out not of mode-size. As we need a matching type -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38271
[Bug middle-end/38271] [4.4 Regression] Spurious / missing ... used uninitialized in this function warning
--- Comment #4 from jakub at gcc dot gnu dot org 2008-12-03 16:28 --- I can't reproduce this any longer since http://gcc.gnu.org/viewcvs?view=revrevision=142396 -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38271
[Bug middle-end/38271] [4.4 Regression] Spurious / missing ... used uninitialized in this function warning
--- Comment #5 from rguenth at gcc dot gnu dot org 2008-12-03 16:38 ---
Still warns for
truct xxx {
short a;
short b;
void *c;
};
void bar(struct xxx);
void foo(struct xxx *p, int i)
{
struct xxx s0 = *p;
struct xxx s = s0;
if (s.a) i++;
bar(s);
}
at -O -m32 -Wuninitialized.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38271
[Bug middle-end/38271] [4.4 Regression] Spurious / missing ... used uninitialized in this function warning
--- Comment #3 from rguenth at gcc dot gnu dot org 2008-11-26 10:00 ---
Doesn't happen on 64bit x86_64. Testcase which doesn't warn with 4.3:
struct xxx {
short a;
short b;
void *c;
};
void bar(struct xxx);
void foo(struct xxx *p, int i)
{
struct xxx s = *p;
if (s.a) i++;
bar(s);
}
but indeed, SRA does bullshit here, cost-wise and correctness-wise.
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
GCC target triplet||i686-*-*
Known to work||4.3.2
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38271
[Bug middle-end/38271] [4.4 Regression] Spurious / missing ... used uninitialized in this function warning
--- Comment #2 from pinskia at gcc dot gnu dot org 2008-11-26 02:54 ---
What the *#$^*(#^$*(:
SR.3_5 = p_1(D)-c;
SR.4_6 = VIEW_CONVERT_EXPRlong long unsigned int(*p_1(D));
SR.4_7 = SR.4_6 4294967295;
SR.5_8 = (unsigned int) SR.4_7;
s.0.c ={v} SR.3_5;
SR.6_10 = VIEW_CONVERT_EXPRlong long unsigned int(s.0);
SR.7_11 = SR.6_10 0x0;
SR.8_12 = (long long unsigned int) SR.5_8;
SR.9_13 = SR.7_11 | SR.8_12;
s.0 ={v} VIEW_CONVERT_EXPRstruct xxx(SR.9_13);
This is so wrong.
--
pinskia at gcc dot gnu dot org changed:
What|Removed |Added
CC||aoliva at gcc dot gnu dot
||org
Status|UNCONFIRMED |NEW
Ever Confirmed|0 |1
Keywords||diagnostic
Last reconfirmed|-00-00 00:00:00 |2008-11-26 02:54:24
date||
Summary|Spurious / missing ... used|[4.4 Regression] Spurious /
|uninitialized in this |missing ... used
|function warning |uninitialized in this
||function warning
Target Milestone|--- |4.4.0
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38271
