* Fix crash in vect_is_slp_reduction
@ 2011-05-20 14:54 Ryan Mansfield
2011-05-20 15:11 ` Richard Guenther
0 siblings, 1 reply; 5+ messages in thread
From: Ryan Mansfield @ 2011-05-20 14:54 UTC (permalink / raw)
To: gcc-patches
There is a crash in vect_is_slp_reduction where use_stmt doesn't get
initialized in the FOR_EACH_IMM_USE_FAST loop.
1718 FOR_EACH_IMM_USE_FAST (use_p, imm_iter, lhs)
(gdb) call debug_tree(lhs)
<ssa_name 0x7ffff6e3a5a0
type <integer_type 0x7ffff7f8c7e0 int32_t sizes-gimplified public SI
size <integer_cst 0x7ffff7ec47e0 constant 32>
unit size <integer_cst 0x7ffff7ec4580 constant 4>
align 32 symtab 0 alias set -1 canonical type 0x7ffff7ecb498
precision 32 min <integer_cst 0x7ffff7ec4780 -2147483648> max
<integer_cst 0x7ffff7ec47a0 2147483647> context <translation_unit_decl
0x7ffff7f98508 D.1314>
pointer_to_this <pointer_type 0x7ffff7f9f930>>
var <var_decl 0x7ffff6e1ce60 g_29_lsm.18>def_stmt g_29_lsm.18_137 =
PHI <g_29_lsm.18_87(15), D.2113_96(18)>
version 137>
(gdb) n
1742 if (found)
(gdb)
1746 if (gimple_code (use_stmt) == GIMPLE_PHI)
(gdb)
Program received signal SIGSEGV, Segmentation fault.
0x0000000000980f22 in gimple_code (g=0x0) at ../../gcc/gimple.h:1098
1098 return g->gsbase.code;
Is the following sufficient?
Index: tree-vect-loop.c
===================================================================
--- tree-vect-loop.c (revision 173945)
+++ tree-vect-loop.c (working copy)
@@ -1743,7 +1743,7 @@
break;
/* This is a loop exit phi, and we haven't reached the reduction
phi. */
- if (gimple_code (use_stmt) == GIMPLE_PHI)
+ if (!use_stmt || gimple_code (use_stmt) == GIMPLE_PHI)
return false;
if (!is_gimple_assign (use_stmt)
Regards,
Ryan Mansfield
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: Fix crash in vect_is_slp_reduction
2011-05-20 14:54 Fix crash in vect_is_slp_reduction Ryan Mansfield
@ 2011-05-20 15:11 ` Richard Guenther
2011-05-20 16:51 ` Ira Rosen
2011-05-20 19:22 ` Fix crash in vect_is_slp_reduction Ryan Mansfield
0 siblings, 2 replies; 5+ messages in thread
From: Richard Guenther @ 2011-05-20 15:11 UTC (permalink / raw)
To: Ryan Mansfield; +Cc: gcc-patches
On Fri, May 20, 2011 at 4:06 PM, Ryan Mansfield <rmansfield@qnx.com> wrote:
> There is a crash in vect_is_slp_reduction where use_stmt doesn't get
> initialized in the FOR_EACH_IMM_USE_FAST loop.
>
> 1718 FOR_EACH_IMM_USE_FAST (use_p, imm_iter, lhs)
> (gdb) call debug_tree(lhs)
> <ssa_name 0x7ffff6e3a5a0
> type <integer_type 0x7ffff7f8c7e0 int32_t sizes-gimplified public SI
> size <integer_cst 0x7ffff7ec47e0 constant 32>
> unit size <integer_cst 0x7ffff7ec4580 constant 4>
> align 32 symtab 0 alias set -1 canonical type 0x7ffff7ecb498
> precision 32 min <integer_cst 0x7ffff7ec4780 -2147483648> max <integer_cst
> 0x7ffff7ec47a0 2147483647> context <translation_unit_decl 0x7ffff7f98508
> D.1314>
> pointer_to_this <pointer_type 0x7ffff7f9f930>>
> var <var_decl 0x7ffff6e1ce60 g_29_lsm.18>def_stmt g_29_lsm.18_137 = PHI
> <g_29_lsm.18_87(15), D.2113_96(18)>
>
> version 137>
> (gdb) n
> 1742 if (found)
> (gdb)
> 1746 if (gimple_code (use_stmt) == GIMPLE_PHI)
> (gdb)
>
> Program received signal SIGSEGV, Segmentation fault.
> 0x0000000000980f22 in gimple_code (g=0x0) at ../../gcc/gimple.h:1098
> 1098 return g->gsbase.code;
>
> Is the following sufficient?
No, we shouldn't arrive with a NULL use_stmt here.
Richard.
> Index: tree-vect-loop.c
> ===================================================================
> --- tree-vect-loop.c (revision 173945)
> +++ tree-vect-loop.c (working copy)
> @@ -1743,7 +1743,7 @@
> break;
>
> /* This is a loop exit phi, and we haven't reached the reduction phi.
> */
> - if (gimple_code (use_stmt) == GIMPLE_PHI)
> + if (!use_stmt || gimple_code (use_stmt) == GIMPLE_PHI)
> return false;
>
> if (!is_gimple_assign (use_stmt)
>
> Regards,
>
> Ryan Mansfield
>
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: Fix crash in vect_is_slp_reduction
2011-05-20 15:11 ` Richard Guenther
@ 2011-05-20 16:51 ` Ira Rosen
2011-05-22 15:39 ` [patch] Fix PR 49087 (was Re: Fix crash in vect_is_slp_reduction) Ira Rosen
2011-05-20 19:22 ` Fix crash in vect_is_slp_reduction Ryan Mansfield
1 sibling, 1 reply; 5+ messages in thread
From: Ira Rosen @ 2011-05-20 16:51 UTC (permalink / raw)
To: Richard Guenther; +Cc: gcc-patches, Ryan Mansfield
gcc-patches-owner@gcc.gnu.org wrote on 20/05/2011 05:17:47 PM:
>
> On Fri, May 20, 2011 at 4:06 PM, Ryan Mansfield <rmansfield@qnx.com>
wrote:
> > There is a crash in vect_is_slp_reduction where use_stmt doesn't get
> > initialized in the FOR_EACH_IMM_USE_FAST loop.
> >
> > 1718 FOR_EACH_IMM_USE_FAST (use_p, imm_iter, lhs)
> > (gdb) call debug_tree(lhs)
> > <ssa_name 0x7ffff6e3a5a0
> > type <integer_type 0x7ffff7f8c7e0 int32_t sizes-gimplified public SI
> > size <integer_cst 0x7ffff7ec47e0 constant 32>
> > unit size <integer_cst 0x7ffff7ec4580 constant 4>
> > align 32 symtab 0 alias set -1 canonical type 0x7ffff7ecb498
> > precision 32 min <integer_cst 0x7ffff7ec4780 -2147483648> max
<integer_cst
> > 0x7ffff7ec47a0 2147483647> context <translation_unit_decl
0x7ffff7f98508
> > D.1314>
> > pointer_to_this <pointer_type 0x7ffff7f9f930>>
> > var <var_decl 0x7ffff6e1ce60 g_29_lsm.18>def_stmt g_29_lsm.18_137 =
PHI
> > <g_29_lsm.18_87(15), D.2113_96(18)>
> >
> > version 137>
> > (gdb) n
> > 1742 if (found)
> > (gdb)
> > 1746 if (gimple_code (use_stmt) == GIMPLE_PHI)
> > (gdb)
> >
> > Program received signal SIGSEGV, Segmentation fault.
> > 0x0000000000980f22 in gimple_code (g=0x0) at ../../gcc/gimple.h:1098
> > 1098 return g->gsbase.code;
> >
> > Is the following sufficient?
>
> No, we shouldn't arrive with a NULL use_stmt here.
I think a proper fix will be to fail if there are no uses.
I'll prepare a patch on Sunday.
Ira
Index: tree-vect-loop.c
===================================================================
--- tree-vect-loop.c (revision 173856)
+++ tree-vect-loop.c (working copy)
@@ -1704,7 +1704,7 @@ vect_is_slp_reduction (loop_vec_info loop_info, gi
tree lhs;
imm_use_iterator imm_iter;
use_operand_p use_p;
- int nloop_uses, size = 0;
+ int nloop_uses, size = 0, nuses;
bool found = false;
if (loop != vect_loop)
@@ -1715,9 +1715,11 @@ vect_is_slp_reduction (loop_vec_info loop_info, gi
while (1)
{
nloop_uses = 0;
+ nuses = 0;
FOR_EACH_IMM_USE_FAST (use_p, imm_iter, lhs)
{
use_stmt = USE_STMT (use_p);
+ nuses++;
if (is_gimple_debug (use_stmt))
continue;
@@ -1739,6 +1741,10 @@ vect_is_slp_reduction (loop_vec_info loop_info, gi
return false;
}
+ /* The statement has no uses. */
+ if (nuses == 0)
+ return false;
+
if (found)
break;
>
> Richard.
>
> > Index: tree-vect-loop.c
> > ===================================================================
> > --- tree-vect-loop.c (revision 173945)
> > +++ tree-vect-loop.c (working copy)
> > @@ -1743,7 +1743,7 @@
> > break;
> >
> > /* This is a loop exit phi, and we haven't reached the reduction
phi.
> > */
> > - if (gimple_code (use_stmt) == GIMPLE_PHI)
> > + if (!use_stmt || gimple_code (use_stmt) == GIMPLE_PHI)
> > return false;
> >
> > if (!is_gimple_assign (use_stmt)
> >
> > Regards,
> >
> > Ryan Mansfield
> >
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: Fix crash in vect_is_slp_reduction
2011-05-20 15:11 ` Richard Guenther
2011-05-20 16:51 ` Ira Rosen
@ 2011-05-20 19:22 ` Ryan Mansfield
1 sibling, 0 replies; 5+ messages in thread
From: Ryan Mansfield @ 2011-05-20 19:22 UTC (permalink / raw)
To: Richard Guenther; +Cc: gcc-patches
On 11-05-20 10:17 AM, Richard Guenther wrote:
> On Fri, May 20, 2011 at 4:06 PM, Ryan Mansfield<rmansfield@qnx.com> wrote:
>> There is a crash in vect_is_slp_reduction where use_stmt doesn't get
>> initialized in the FOR_EACH_IMM_USE_FAST loop.
>>
>> 1718 FOR_EACH_IMM_USE_FAST (use_p, imm_iter, lhs)
>> (gdb) call debug_tree(lhs)
>> <ssa_name 0x7ffff6e3a5a0
>> type<integer_type 0x7ffff7f8c7e0 int32_t sizes-gimplified public SI
>> size<integer_cst 0x7ffff7ec47e0 constant 32>
>> unit size<integer_cst 0x7ffff7ec4580 constant 4>
>> align 32 symtab 0 alias set -1 canonical type 0x7ffff7ecb498
>> precision 32 min<integer_cst 0x7ffff7ec4780 -2147483648> max<integer_cst
>> 0x7ffff7ec47a0 2147483647> context<translation_unit_decl 0x7ffff7f98508
>> D.1314>
>> pointer_to_this<pointer_type 0x7ffff7f9f930>>
>> var<var_decl 0x7ffff6e1ce60 g_29_lsm.18>def_stmt g_29_lsm.18_137 = PHI
>> <g_29_lsm.18_87(15), D.2113_96(18)>
>>
>> version 137>
>> (gdb) n
>> 1742 if (found)
>> (gdb)
>> 1746 if (gimple_code (use_stmt) == GIMPLE_PHI)
>> (gdb)
>>
>> Program received signal SIGSEGV, Segmentation fault.
>> 0x0000000000980f22 in gimple_code (g=0x0) at ../../gcc/gimple.h:1098
>> 1098 return g->gsbase.code;
>>
>> Is the following sufficient?
>
> No, we shouldn't arrive with a NULL use_stmt here.
OK, thanks. I've filed PR49087 with a testcase.
Regards,
Ryan Mansfield
^ permalink raw reply [flat|nested] 5+ messages in thread
* [patch] Fix PR 49087 (was Re: Fix crash in vect_is_slp_reduction)
2011-05-20 16:51 ` Ira Rosen
@ 2011-05-22 15:39 ` Ira Rosen
0 siblings, 0 replies; 5+ messages in thread
From: Ira Rosen @ 2011-05-22 15:39 UTC (permalink / raw)
To: Ira Rosen; +Cc: gcc-patches, Richard Guenther, Ryan Mansfield
> >
> > No, we shouldn't arrive with a NULL use_stmt here.
>
> I think a proper fix will be to fail if there are no uses.
> I'll prepare a patch on Sunday.
>
Here is the patch. It bails out if LHS has no uses.
Bootstrapped and tested on powerpc64-suse-linux.
Committed.
Ira
ChangeLog:
PR tree-optimization/49087
* tree-vect-loop.c (vect_is_slp_reduction): Fail if
LHS has no uses.
testsuite/ChangeLog:
PR tree-optimization/49087
* gcc.dg/vect/O3-pr49087.c: New test.
Index: tree-vect-loop.c
===================================================================
--- tree-vect-loop.c (revision 174025)
+++ tree-vect-loop.c (working copy)
@@ -1704,7 +1704,7 @@ vect_is_slp_reduction (loop_vec_info loop_info, gi
tree lhs;
imm_use_iterator imm_iter;
use_operand_p use_p;
- int nloop_uses, size = 0;
+ int nloop_uses, size = 0, nuses;
bool found = false;
if (loop != vect_loop)
@@ -1715,9 +1715,11 @@ vect_is_slp_reduction (loop_vec_info loop_info, gi
while (1)
{
nloop_uses = 0;
+ nuses = 0;
FOR_EACH_IMM_USE_FAST (use_p, imm_iter, lhs)
{
use_stmt = USE_STMT (use_p);
+ nuses++;
if (is_gimple_debug (use_stmt))
continue;
@@ -1739,6 +1741,10 @@ vect_is_slp_reduction (loop_vec_info loop_info, gi
return false;
}
+ /* We reached a statement with no uses. */
+ if (nuses == 0)
+ return false;
+
if (found)
break;
Index: testsuite/gcc.dg/vect/O3-pr49087.c
===================================================================
--- testsuite/gcc.dg/vect/O3-pr49087.c (revision 0)
+++ testsuite/gcc.dg/vect/O3-pr49087.c (revision 0)
@@ -0,0 +1,37 @@
+/* { dg-do compile } */
+
+static char func2() { }
+
+struct S0
+{
+ int t;
+};
+
+int g;
+
+struct S0 s0;
+
+int
+foo (int arg)
+{
+ int *ptr = &g;
+ int i, j;
+ for (i = 0; i < 10; i += 1)
+ {
+ for (j = 0; j < 1; j += 1)
+ {
+ int k;
+ if (arg)
+ {
+ int l;
+ for (k = 1; arg < 10; arg = func2 ())
+ {
+ return l;
+ }
+ }
+ *ptr = func2 () ^ s0.t;
+ }
+ }
+}
+
+/* { dg-final { cleanup-tree-dump "vect" } } */
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2011-05-22 8:11 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2011-05-20 14:54 Fix crash in vect_is_slp_reduction Ryan Mansfield
2011-05-20 15:11 ` Richard Guenther
2011-05-20 16:51 ` Ira Rosen
2011-05-22 15:39 ` [patch] Fix PR 49087 (was Re: Fix crash in vect_is_slp_reduction) Ira Rosen
2011-05-20 19:22 ` Fix crash in vect_is_slp_reduction Ryan Mansfield
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).