[Bug tree-optimization/52091] [4.7 Regression] ICE: verify_gimple failed
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52091 Ira Rosen changed: What|Removed |Added Status|NEW |RESOLVED Resolution||FIXED --- Comment #9 from Ira Rosen 2012-02-06 08:42:00 UTC --- Fixed.
[Bug tree-optimization/52091] [4.7 Regression] ICE: verify_gimple failed
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52091 --- Comment #8 from irar at gcc dot gnu.org 2012-02-05 05:58:22 UTC --- Author: irar Date: Sun Feb 5 05:58:18 2012 New Revision: 183902 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=183902 Log: PR tree-optimization/52091 * tree-vectorizer.h (vect_is_simple_use): Add an argument. (vect_is_simple_use_1): Likewise. * tree-vect-loop.c (vectorizable_reduction): Update calls to vect_is_simple_use_1 and vect_is_simple_use. (vectorizable_live_operation): Likewise. * tree-vect-patterns.c (widened_name_p, vect_recog_vector_vector_shift_pattern, check_bool_pattern): Likewise. * tree-vect-stmts.c (process_use, vect_get_vec_def_for_operand, vectorizable_call, vectorizable_conversion, vectorizable_assignment, vectorizable_shift, vectorizable_operation, vectorizable_store, vectorizable_load): Likewise. (vect_is_simple_cond): Add an argument, pass it to vect_is_simple_use_1. (vectorizable_condition): Update calls to vect_is_simple_cond, vect_is_simple_use. (vect_is_simple_use): Add an argument, the statement in which OPERAND is used. Check that if OPERAND's def stmt is a double reduction phi node, the use is a phi node too. (vect_is_simple_use_1): Add an argument, pass it to vect_is_simple_use. * tree-vect-slp.c (vect_get_and_check_slp_defs): Update a call to vect_is_simple_use. Added: trunk/gcc/testsuite/gcc.dg/vect/pr52091.c Modified: trunk/gcc/ChangeLog trunk/gcc/testsuite/ChangeLog trunk/gcc/tree-vect-loop.c trunk/gcc/tree-vect-patterns.c trunk/gcc/tree-vect-slp.c trunk/gcc/tree-vect-stmts.c trunk/gcc/tree-vectorizer.h
[Bug tree-optimization/52091] [4.7 Regression] ICE: verify_gimple failed
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52091
--- Comment #7 from Ira Rosen 2012-02-02 15:16:18 UTC
---
(In reply to comment #6)
> (In reply to comment #5)
> > I think we should fail to vectorize D.2030_19 = prephitmp.17_21 & 65535, or
> > any
> > other non-phi/not vect_double_reduction_def stmt with a double reduction
> > phi as
> > a def_stmt.
> >
> > We can either check this in every vectorizable_* for every operand, like
> > this:
> > Index: tree-vect-stmts.c
> > ===
> > --- tree-vect-stmts.c (revision 183125)
> > +++ tree-vect-stmts.c (working copy)
> > @@ -3326,7 +3326,8 @@ vectorizable_operation (gimple stmt, gimple_stmt_i
> >
> >op0 = gimple_assign_rhs1 (stmt);
> >if (!vect_is_simple_use_1 (op0, loop_vinfo, bb_vinfo,
> > -&def_stmt, &def, &dt[0], &vectype))
> > +&def_stmt, &def, &dt[0], &vectype)
> > + || dt[0] == vect_double_reduction_def)
> > {
> >if (vect_print_dump_info (REPORT_DETAILS))
> > fprintf (vect_dump, "use not simple.");
> >
> >
> > or pass stmt or stmt_info to vect_is_simple_use and check it there.
>
> Are you going to write a patch for this? Not sure how exactly would you like
> it to look up.
>
We need to check that if def_stmt is vect_double_reduction_def, the stmt itself
needs to be vect_double_reduction_def. We know def_type (dt in the above patch)
of def_stmt from vect_is_simple_use. We call it from all the vectorizable_*
functions. We don't pass the stmt itself to vect_is_simple_use, therefore, we
should either do that and perform the check in vect_is_simple_use, or check
this in all the vectorizable_* (except perhaps vectorizable_reduction). I
prefer to pass stmt or stmt_info to vect_is_simple_use.
I can do this on Sunday. You are welcome to do this yourself. Whatever you
prefer.
[Bug tree-optimization/52091] [4.7 Regression] ICE: verify_gimple failed
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52091
--- Comment #6 from Jakub Jelinek 2012-02-02
13:41:01 UTC ---
(In reply to comment #5)
> I think we should fail to vectorize D.2030_19 = prephitmp.17_21 & 65535, or
> any
> other non-phi/not vect_double_reduction_def stmt with a double reduction phi
> as
> a def_stmt.
>
> We can either check this in every vectorizable_* for every operand, like this:
> Index: tree-vect-stmts.c
> ===
> --- tree-vect-stmts.c (revision 183125)
> +++ tree-vect-stmts.c (working copy)
> @@ -3326,7 +3326,8 @@ vectorizable_operation (gimple stmt, gimple_stmt_i
>
>op0 = gimple_assign_rhs1 (stmt);
>if (!vect_is_simple_use_1 (op0, loop_vinfo, bb_vinfo,
> -&def_stmt, &def, &dt[0], &vectype))
> +&def_stmt, &def, &dt[0], &vectype)
> + || dt[0] == vect_double_reduction_def)
> {
>if (vect_print_dump_info (REPORT_DETAILS))
> fprintf (vect_dump, "use not simple.");
>
>
> or pass stmt or stmt_info to vect_is_simple_use and check it there.
Are you going to write a patch for this? Not sure how exactly would you like
it to look up.
> > OT, it is strange that we are creating a reduction for a loop which loops
> > exactly as many times as there are units in the vector, that doesn't seem
> > to be
> > profitable.
> >
>
> Right, but doesn't cost model catch this?
For simple testcases it does apparently.
[Bug tree-optimization/52091] [4.7 Regression] ICE: verify_gimple failed
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52091
Ira Rosen changed:
What|Removed |Added
CC||irar at il dot ibm.com
--- Comment #5 from Ira Rosen 2012-02-02 13:22:53 UTC
---
(In reply to comment #4)
I think that the problem here is that
D.2030_19 = prephitmp.17_21 & 65535
is falsely initialized as reduction, while it isn't a reduction. The mistake
occurs because its def_stmt,
prephitmp.17_21 = PHI ,
is a phi node marked as double reduction:
:
# f.6_36 = PHI
# prephitmp.17_21 = PHI - double reduction phi
...
D.2030_19 = prephitmp.17_21 & 65535; - not a reduction stmt
...
:
# b.4_41 = PHI
...
d.3_9 = d_lsm.23_31 & 1;
...
:
# d.3_44 = PHI- double reduction stmt
I think we should fail to vectorize D.2030_19 = prephitmp.17_21 & 65535, or any
other non-phi/not vect_double_reduction_def stmt with a double reduction phi as
a def_stmt.
We can either check this in every vectorizable_* for every operand, like this:
Index: tree-vect-stmts.c
===
--- tree-vect-stmts.c (revision 183125)
+++ tree-vect-stmts.c (working copy)
@@ -3326,7 +3326,8 @@ vectorizable_operation (gimple stmt, gimple_stmt_i
op0 = gimple_assign_rhs1 (stmt);
if (!vect_is_simple_use_1 (op0, loop_vinfo, bb_vinfo,
-&def_stmt, &def, &dt[0], &vectype))
+&def_stmt, &def, &dt[0], &vectype)
+ || dt[0] == vect_double_reduction_def)
{
if (vect_print_dump_info (REPORT_DETAILS))
fprintf (vect_dump, "use not simple.");
or pass stmt or stmt_info to vect_is_simple_use and check it there.
> OT, it is strange that we are creating a reduction for a loop which loops
> exactly as many times as there are units in the vector, that doesn't seem to
> be
> profitable.
>
Right, but doesn't cost model catch this?
[Bug tree-optimization/52091] [4.7 Regression] ICE: verify_gimple failed
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52091
Jakub Jelinek changed:
What|Removed |Added
Status|ASSIGNED|NEW
CC||irar at gcc dot gnu.org
AssignedTo|jakub at gcc dot gnu.org|unassigned at gcc dot
||gnu.org
--- Comment #4 from Jakub Jelinek 2012-02-02
12:06:27 UTC ---
Executable testcase:
/* PR tree-optimization/52091 */
int b, c, d, f;
unsigned h;
extern void abort (void);
int
main ()
{
d = -1;
h = 65;
asm volatile ("" : : : "memory");
for (f = 0; f < 4; f++)
{
h &= (unsigned short) d;
for (b = 0; b <= 1; b++)
{
c = 0;
d &= 1;
}
}
asm volatile ("" : : : "memory");
if (b != 2 || c != 0 || d != 1 || f != 4 || h != 1)
abort ();
return 0;
}
Following patch fixes the ICE:
--- gcc/tree-vect-stmts.c.jj2012-02-01 10:33:58.0 +0100
+++ gcc/tree-vect-stmts.c2012-02-02 11:50:42.623948066 +0100
@@ -1294,7 +1294,11 @@ vect_get_vec_def_for_operand (tree op, g
/* Get the def before the loop */
op = PHI_ARG_DEF_FROM_EDGE (def_stmt, loop_preheader_edge (loop));
-return get_initial_def_for_reduction (stmt, op, scalar_def);
+vec_oprnd = get_initial_def_for_reduction (stmt, op, scalar_def);
+if (!is_gimple_val (vec_oprnd))
+ vec_oprnd = vect_init_vector (stmt, vec_oprnd,
+TREE_TYPE (vec_oprnd), NULL);
+return vec_oprnd;
}
/* Case 5: operand is defined by loop-header phi - induction. */
but just turns ice-on-valid-code into wrong-code, h is 65 when vectorizing.
Ira, could you please look at this? Thanks.
OT, it is strange that we are creating a reduction for a loop which loops
exactly as many times as there are units in the vector, that doesn't seem to be
profitable.
My other attempt to fix the ICE was:
--- gcc/tree-vect-loop.c.jj2011-12-16 08:37:45.0 +0100
+++ gcc/tree-vect-loop.c2012-02-02 11:57:25.354607843 +0100
@@ -3370,7 +3370,10 @@ get_initial_def_for_reduction (gimple st
if (TREE_CONSTANT (init_val))
init_def = build_vector (vectype, t);
else
- init_def = build_constructor_from_list (vectype, t);
+ {
+init_def = build_constructor_from_list (vectype, t);
+init_def = vect_init_vector (stmt, init_def, vectype, NULL);
+ }
break;
but that ICEs on the testcase, because the other caller of
get_initial_def_for_reduction is calling vect_init_vector immediately after it,
with a different stmt than was passed to get_initial_def_for_reduction, and
this resulted in definition not dominating the use. Wonder if that isn't
another bug.
[Bug tree-optimization/52091] [4.7 Regression] ICE: verify_gimple failed
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52091 --- Comment #3 from Jakub Jelinek 2012-02-02 09:00:36 UTC --- Started with http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=178728
[Bug tree-optimization/52091] [4.7 Regression] ICE: verify_gimple failed
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52091 Jakub Jelinek changed: What|Removed |Added Status|NEW |ASSIGNED CC||jakub at gcc dot gnu.org AssignedTo|unassigned at gcc dot |jakub at gcc dot gnu.org |gnu.org | --- Comment #2 from Jakub Jelinek 2012-02-02 08:29:41 UTC --- Looking at it.
[Bug tree-optimization/52091] [4.7 Regression] ICE: verify_gimple failed
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52091
Andrew Pinski changed:
What|Removed |Added
Status|UNCONFIRMED |NEW
Last reconfirmed||2012-02-02
Host|x86_64-unknown-linux-gnu|
Ever Confirmed|0 |1
Build|x86_64-unknown-linux-gnu|
--- Comment #1 from Andrew Pinski 2012-02-02
04:53:25 UTC ---
Confirmed reduced testcase:
int b,c,d,f;
unsigned h;
void fn52 (void)
{
for (f = 4; f; f--)
{
h &= (unsigned short)d;
for (b = 0; b <= 1; b++)
{
c = 0;
d &= 1;
}
}
}
CUT
Note also:
int b,c,d,f;
unsigned h;
void fn52 (void)
{
for (f = 4; f; f--)
{
h &= d;
for (b = 0; b <= 1; b++)
{
c = 0;
d &= 1;
}
}
}
CUT
Gives a slightly different ICE:
t.c: In function ‘fn52’:
t.c:4:6: internal compiler error: in get_initial_def_for_reduction, at
tree-vect-loop.c:3391
Please submit a full bug report,
with preprocessed source if appropriate.
See for instructions.
[Bug tree-optimization/52091] [4.7 Regression] ICE: verify_gimple failed
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52091 Andrew Pinski changed: What|Removed |Added Target Milestone|--- |4.7.0 Summary|ICE: verify_gimple failed |[4.7 Regression] ICE: ||verify_gimple failed
