[patch] Fix PR tree-optimization/45902

[patch] Fix PR tree-optimization/45902

[Ira Rosen]

Hi,

This patch fixes a bug in creation of a vector of constants in SLP. The
problem is in the type of created vector. It should be set to the vector
type of the statement unless it's a pointer.

Bootstrapped and tested on x86_64-suse-linux and checked that the failure
is fixed on powerpc64-suse-linux.

Committed to mainline, ok for 4.5?

Thanks,
Ira

ChangeLog:

	PR tree-optimization/45902
	* tree-vect-slp.c (vect_get_constant_vectors): Use statement's
	vector type for constants, unless it's a pointer.

testsuite/ChangeLog:

	PR tree-optimization/45902
	* gcc.dg/vect/45902.c: New test.


4.6 patch

Index: tree-vect-slp.c
===================================================================
--- tree-vect-slp.c     (revision 164987)
+++ tree-vect-slp.c     (working copy)
@@ -1894,13 +1894,20 @@ vect_get_constant_vectors (slp_tree slp_
     }

   if (CONSTANT_CLASS_P (op))
-    constant_p = true;
+    {
+      constant_p = true;
+      if (POINTER_TYPE_P (TREE_TYPE (gimple_assign_lhs (stmt))))
+        vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
+      else
+        vector_type = STMT_VINFO_VECTYPE (stmt_vinfo);
+    }
   else
-    constant_p = false;
+    {
+      constant_p = false;
+      vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
+    }

-  vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
   gcc_assert (vector_type);
-
   nunits = TYPE_VECTOR_SUBPARTS (vector_type);

   /* NUMBER_OF_COPIES is the number of times we need to use the same
values in
Index: testsuite/gcc.dg/vect/pr45902.c
===================================================================
--- testsuite/gcc.dg/vect/pr45902.c     (revision 0)
+++ testsuite/gcc.dg/vect/pr45902.c     (revision 0)
@@ -0,0 +1,44 @@
+/* { dg-require-effective-target vect_int } */
+
+#include <stdarg.h>
+#include <stdlib.h>
+#include "tree-vect.h"
+
+#define N 128
+
+short res[N];
+short a[N];
+
+int
+main1 ()
+{
+  int i;
+
+  for (i = 0; i < N/4; i+=4)
+    {
+      res[i] = a[i] >> 8;
+      res[i+1] = a[i+1] >> 8;
+      res[i+2] = a[i+2] >> 8;
+      res[i+3] = a[i+3] >> 8;
+    }
+}
+
+int
+main ()
+{
+  int i;
+
+  for (i = 0; i < N; i++)
+    a[i] = i;
+
+  main1 ();
+
+  for (i = 0; i < N; i++)
+    if (res[i] != a[i] >> 8)
+      abort ();
+
+  return 0;
+}
+
+/* { dg-final { cleanup-tree-dump "vect" } } */

4.5 patch

Index: tree-vect-slp.c
===================================================================
--- tree-vect-slp.c     (revision 165269)
+++ tree-vect-slp.c     (working copy)
@@ -1459,13 +1459,20 @@ vect_get_constant_vectors (slp_tree slp_
     }

   if (CONSTANT_CLASS_P (op))
-    constant_p = true;
+    {
+      constant_p = true;
+      if (POINTER_TYPE_P (TREE_TYPE (gimple_assign_lhs (stmt))))
+        vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
+      else
+        vector_type = STMT_VINFO_VECTYPE (stmt_vinfo);
+    }
   else
-    constant_p = false;
+    {
+      constant_p = false;
+      vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
+    }

-  vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
   gcc_assert (vector_type);
-
   nunits = TYPE_VECTOR_SUBPARTS (vector_type);

   /* NUMBER_OF_COPIES is the number of times we need to use the same
values in
Index: testsuite/gcc.dg/vect/pr45902.c
===================================================================
--- testsuite/gcc.dg/vect/pr45902.c     (revision 0)
+++ testsuite/gcc.dg/vect/pr45902.c     (revision 0)
@@ -0,0 +1,44 @@
+/* { dg-require-effective-target vect_int } */
+
+#include <stdarg.h>
+#include <stdlib.h>
+#include "tree-vect.h"
+
+#define N 128
+
+short res[N];
+short a[N];
+
+int
+main1 ()
+{
+  int i;
+
+  for (i = 0; i < N/4; i+=4)
+    {
+      res[i] = a[i] >> 8;
+      res[i+1] = a[i+1] >> 8;
+      res[i+2] = a[i+2] >> 8;
+      res[i+3] = a[i+3] >> 8;
+    }
+}
+
+int
+main ()
+{
+  int i;
+
+  for (i = 0; i < N; i++)
+    a[i] = i;
+
+  main1 ();
+
+  for (i = 0; i < N; i++)
+    if (res[i] != a[i] >> 8)
+      abort ();
+
+  return 0;
+}
+
+/* { dg-final { cleanup-tree-dump "vect" } } */
+

Re: [patch] Fix PR tree-optimization/45902

[Richard Guenther]

On Mon, Oct 11, 2010 at 2:19 PM, Ira Rosen <IRAR@il.ibm.com> wrote:
>
> Hi,
>
> This patch fixes a bug in creation of a vector of constants in SLP. The
> problem is in the type of created vector. It should be set to the vector
> type of the statement unless it's a pointer.
>
> Bootstrapped and tested on x86_64-suse-linux and checked that the failure
> is fixed on powerpc64-suse-linux.
>
> Committed to mainline, ok for 4.5?

I don't think this makes sense.  If at all the selection of which type
to use should be based on the operation code and the operand
position, but not on CONSTANT_CLASS_P or pointer-type-ness.

So - what operation code / operand position is currently mishandled?

Thanks,
Richard.

> Thanks,
> Ira
>
> ChangeLog:
>
>        PR tree-optimization/45902
>        * tree-vect-slp.c (vect_get_constant_vectors): Use statement's
>        vector type for constants, unless it's a pointer.
>
> testsuite/ChangeLog:
>
>        PR tree-optimization/45902
>        * gcc.dg/vect/45902.c: New test.
>
>
> 4.6 patch
>
> Index: tree-vect-slp.c
> ===================================================================
> --- tree-vect-slp.c     (revision 164987)
> +++ tree-vect-slp.c     (working copy)
> @@ -1894,13 +1894,20 @@ vect_get_constant_vectors (slp_tree slp_
>     }
>
>   if (CONSTANT_CLASS_P (op))
> -    constant_p = true;
> +    {
> +      constant_p = true;
> +      if (POINTER_TYPE_P (TREE_TYPE (gimple_assign_lhs (stmt))))
> +        vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
> +      else
> +        vector_type = STMT_VINFO_VECTYPE (stmt_vinfo);
> +    }
>   else
> -    constant_p = false;
> +    {
> +      constant_p = false;
> +      vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
> +    }
>
> -  vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
>   gcc_assert (vector_type);
> -
>   nunits = TYPE_VECTOR_SUBPARTS (vector_type);
>
>   /* NUMBER_OF_COPIES is the number of times we need to use the same
> values in
> Index: testsuite/gcc.dg/vect/pr45902.c
> ===================================================================
> --- testsuite/gcc.dg/vect/pr45902.c     (revision 0)
> +++ testsuite/gcc.dg/vect/pr45902.c     (revision 0)
> @@ -0,0 +1,44 @@
> +/* { dg-require-effective-target vect_int } */
> +
> +#include <stdarg.h>
> +#include <stdlib.h>
> +#include "tree-vect.h"
> +
> +#define N 128
> +
> +short res[N];
> +short a[N];
> +
> +int
> +main1 ()
> +{
> +  int i;
> +
> +  for (i = 0; i < N/4; i+=4)
> +    {
> +      res[i] = a[i] >> 8;
> +      res[i+1] = a[i+1] >> 8;
> +      res[i+2] = a[i+2] >> 8;
> +      res[i+3] = a[i+3] >> 8;
> +    }
> +}
> +
> +int
> +main ()
> +{
> +  int i;
> +
> +  for (i = 0; i < N; i++)
> +    a[i] = i;
> +
> +  main1 ();
> +
> +  for (i = 0; i < N; i++)
> +    if (res[i] != a[i] >> 8)
> +      abort ();
> +
> +  return 0;
> +}
> +
> +/* { dg-final { cleanup-tree-dump "vect" } } */
>
> 4.5 patch
>
> Index: tree-vect-slp.c
> ===================================================================
> --- tree-vect-slp.c     (revision 165269)
> +++ tree-vect-slp.c     (working copy)
> @@ -1459,13 +1459,20 @@ vect_get_constant_vectors (slp_tree slp_
>     }
>
>   if (CONSTANT_CLASS_P (op))
> -    constant_p = true;
> +    {
> +      constant_p = true;
> +      if (POINTER_TYPE_P (TREE_TYPE (gimple_assign_lhs (stmt))))
> +        vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
> +      else
> +        vector_type = STMT_VINFO_VECTYPE (stmt_vinfo);
> +    }
>   else
> -    constant_p = false;
> +    {
> +      constant_p = false;
> +      vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
> +    }
>
> -  vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
>   gcc_assert (vector_type);
> -
>   nunits = TYPE_VECTOR_SUBPARTS (vector_type);
>
>   /* NUMBER_OF_COPIES is the number of times we need to use the same
> values in
> Index: testsuite/gcc.dg/vect/pr45902.c
> ===================================================================
> --- testsuite/gcc.dg/vect/pr45902.c     (revision 0)
> +++ testsuite/gcc.dg/vect/pr45902.c     (revision 0)
> @@ -0,0 +1,44 @@
> +/* { dg-require-effective-target vect_int } */
> +
> +#include <stdarg.h>
> +#include <stdlib.h>
> +#include "tree-vect.h"
> +
> +#define N 128
> +
> +short res[N];
> +short a[N];
> +
> +int
> +main1 ()
> +{
> +  int i;
> +
> +  for (i = 0; i < N/4; i+=4)
> +    {
> +      res[i] = a[i] >> 8;
> +      res[i+1] = a[i+1] >> 8;
> +      res[i+2] = a[i+2] >> 8;
> +      res[i+3] = a[i+3] >> 8;
> +    }
> +}
> +
> +int
> +main ()
> +{
> +  int i;
> +
> +  for (i = 0; i < N; i++)
> +    a[i] = i;
> +
> +  main1 ();
> +
> +  for (i = 0; i < N; i++)
> +    if (res[i] != a[i] >> 8)
> +      abort ();
> +
> +  return 0;
> +}
> +
> +/* { dg-final { cleanup-tree-dump "vect" } } */
> +
>
>

Re: [patch] Fix PR tree-optimization/45902

[Ira Rosen]

Richard Guenther <richard.guenther@gmail.com> wrote on 11/10/2010 02:32:08
PM:

>
> On Mon, Oct 11, 2010 at 2:19 PM, Ira Rosen <IRAR@il.ibm.com> wrote:
> >
> > Hi,
> >
> > This patch fixes a bug in creation of a vector of constants in SLP. The
> > problem is in the type of created vector. It should be set to the
vector
> > type of the statement unless it's a pointer.
> >
> > Bootstrapped and tested on x86_64-suse-linux and checked that the
failure
> > is fixed on powerpc64-suse-linux.
> >
> > Committed to mainline, ok for 4.5?
>
> I don't think this makes sense.  If at all the selection of which type
> to use should be based on the operation code and the operand
> position, but not on CONSTANT_CLASS_P or pointer-type-ness.
>
> So - what operation code / operand position is currently mishandled?

This function creates vectors of constants or loop invariants. If it's a
variable (loop invariant), we know exactly the type, hence the check for
CONSTANT_CLASS_P.

I guess instead of checking the type we can check if it's POINTER_PLUS_EXPR
to catch the cases where we need the operand's type instead of the stmt's
vectype.

Does this make sense?

Index: tree-vect-slp.c
===================================================================
--- tree-vect-slp.c     (revision 165302)
+++ tree-vect-slp.c     (working copy)
@@ -1836,10 +1836,10 @@ vect_get_constant_vectors (slp_tree slp_
   VEC (tree, heap) *voprnds = VEC_alloc (tree, heap, number_of_vectors);
   bool constant_p, is_store;
   tree neutral_op = NULL;
+  enum tree_code code = gimple_assign_rhs_code (stmt);

   if (STMT_VINFO_DEF_TYPE (stmt_vinfo) == vect_reduction_def)
     {
-      enum tree_code code = gimple_assign_rhs_code (stmt);
       if (reduc_index == -1)
         {
           VEC_free (tree, heap, *vec_oprnds);
@@ -1895,18 +1895,14 @@ vect_get_constant_vectors (slp_tree slp_
     }

   if (CONSTANT_CLASS_P (op))
-    {
-      constant_p = true;
-      if (POINTER_TYPE_P (TREE_TYPE (gimple_assign_lhs (stmt))))
-        vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
-      else
-        vector_type = STMT_VINFO_VECTYPE (stmt_vinfo);
-    }
+    constant_p = true;
   else
-    {
-      constant_p = false;
-      vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
-    }
+    constant_p = false;
+
+  if (code == POINTER_PLUS_EXPR)
+    vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
+  else
+    vector_type = STMT_VINFO_VECTYPE (stmt_vinfo);

   gcc_assert (vector_type);
   nunits = TYPE_VECTOR_SUBPARTS (vector_type);

Thanks,
Ira

Re: [patch] Fix PR tree-optimization/45902

[Richard Guenther]

On Mon, Oct 11, 2010 at 3:03 PM, Ira Rosen <IRAR@il.ibm.com> wrote:
> Richard Guenther <richard.guenther@gmail.com> wrote on 11/10/2010 02:32:08
> PM:
>
>>
>> On Mon, Oct 11, 2010 at 2:19 PM, Ira Rosen <IRAR@il.ibm.com> wrote:
>> >
>> > Hi,
>> >
>> > This patch fixes a bug in creation of a vector of constants in SLP. The
>> > problem is in the type of created vector. It should be set to the
> vector
>> > type of the statement unless it's a pointer.
>> >
>> > Bootstrapped and tested on x86_64-suse-linux and checked that the
> failure
>> > is fixed on powerpc64-suse-linux.
>> >
>> > Committed to mainline, ok for 4.5?
>>
>> I don't think this makes sense.  If at all the selection of which type
>> to use should be based on the operation code and the operand
>> position, but not on CONSTANT_CLASS_P or pointer-type-ness.
>>
>> So - what operation code / operand position is currently mishandled?
>
> This function creates vectors of constants or loop invariants. If it's a
> variable (loop invariant), we know exactly the type, hence the check for
> CONSTANT_CLASS_P.
>
> I guess instead of checking the type we can check if it's POINTER_PLUS_EXPR
> to catch the cases where we need the operand's type instead of the stmt's
> vectype.
>
> Does this make sense?

That makes more sense, with ...

> Index: tree-vect-slp.c
> ===================================================================
> --- tree-vect-slp.c     (revision 165302)
> +++ tree-vect-slp.c     (working copy)
> @@ -1836,10 +1836,10 @@ vect_get_constant_vectors (slp_tree slp_
>   VEC (tree, heap) *voprnds = VEC_alloc (tree, heap, number_of_vectors);
>   bool constant_p, is_store;
>   tree neutral_op = NULL;
> +  enum tree_code code = gimple_assign_rhs_code (stmt);
>
>   if (STMT_VINFO_DEF_TYPE (stmt_vinfo) == vect_reduction_def)
>     {
> -      enum tree_code code = gimple_assign_rhs_code (stmt);
>       if (reduc_index == -1)
>         {
>           VEC_free (tree, heap, *vec_oprnds);
> @@ -1895,18 +1895,14 @@ vect_get_constant_vectors (slp_tree slp_
>     }
>
>   if (CONSTANT_CLASS_P (op))
> -    {
> -      constant_p = true;
> -      if (POINTER_TYPE_P (TREE_TYPE (gimple_assign_lhs (stmt))))
> -        vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
> -      else
> -        vector_type = STMT_VINFO_VECTYPE (stmt_vinfo);
> -    }
> +    constant_p = true;
>   else
> -    {
> -      constant_p = false;
> -      vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
> -    }
> +    constant_p = false;
> +
> +  if (code == POINTER_PLUS_EXPR)

checking

          && op_num == 2

(with a possible fixup for the reduction case where that doesn't seem to
be prevailing).

Thanks,
Richard.

> +    vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
> +  else
> +    vector_type = STMT_VINFO_VECTYPE (stmt_vinfo);
>
>   gcc_assert (vector_type);
>   nunits = TYPE_VECTOR_SUBPARTS (vector_type);
>
> Thanks,
> Ira
>
>
>

Re: [patch] Fix PR tree-optimization/45902

[Ira Rosen]

Richard Guenther <richard.guenther@gmail.com> wrote on 11/10/2010 04:03:07
PM:

>
> On Mon, Oct 11, 2010 at 3:03 PM, Ira Rosen <IRAR@il.ibm.com> wrote:
> > Richard Guenther <richard.guenther@gmail.com> wrote on 11/10/2010
02:32:08
> > PM:
> >
> >>
> >> On Mon, Oct 11, 2010 at 2:19 PM, Ira Rosen <IRAR@il.ibm.com> wrote:
> >> >
> >> > Hi,
> >> >
> >> > This patch fixes a bug in creation of a vector of constants in SLP.
The
> >> > problem is in the type of created vector. It should be set to the
> > vector
> >> > type of the statement unless it's a pointer.
> >> >
> >> > Bootstrapped and tested on x86_64-suse-linux and checked that the
> > failure
> >> > is fixed on powerpc64-suse-linux.
> >> >
> >> > Committed to mainline, ok for 4.5?
> >>
> >> I don't think this makes sense.  If at all the selection of which type
> >> to use should be based on the operation code and the operand
> >> position, but not on CONSTANT_CLASS_P or pointer-type-ness.
> >>
> >> So - what operation code / operand position is currently mishandled?
> >
> > This function creates vectors of constants or loop invariants. If it's
a
> > variable (loop invariant), we know exactly the type, hence the check
for
> > CONSTANT_CLASS_P.
> >
> > I guess instead of checking the type we can check if it's
POINTER_PLUS_EXPR
> > to catch the cases where we need the operand's type instead of the
stmt's
> > vectype.
> >
> > Does this make sense?
>
> That makes more sense, with ...
>
> > Index: tree-vect-slp.c
> > ===================================================================
> > --- tree-vect-slp.c     (revision 165302)
> > +++ tree-vect-slp.c     (working copy)
> > @@ -1836,10 +1836,10 @@ vect_get_constant_vectors (slp_tree slp_
> >   VEC (tree, heap) *voprnds = VEC_alloc (tree, heap,
number_of_vectors);
> >   bool constant_p, is_store;
> >   tree neutral_op = NULL;
> > +  enum tree_code code = gimple_assign_rhs_code (stmt);
> >
> >   if (STMT_VINFO_DEF_TYPE (stmt_vinfo) == vect_reduction_def)
> >     {
> > -      enum tree_code code = gimple_assign_rhs_code (stmt);
> >       if (reduc_index == -1)
> >         {
> >           VEC_free (tree, heap, *vec_oprnds);
> > @@ -1895,18 +1895,14 @@ vect_get_constant_vectors (slp_tree slp_
> >     }
> >
> >   if (CONSTANT_CLASS_P (op))
> > -    {
> > -      constant_p = true;
> > -      if (POINTER_TYPE_P (TREE_TYPE (gimple_assign_lhs (stmt))))
> > -        vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
> > -      else
> > -        vector_type = STMT_VINFO_VECTYPE (stmt_vinfo);
> > -    }
> > +    constant_p = true;
> >   else
> > -    {
> > -      constant_p = false;
> > -      vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
> > -    }
> > +    constant_p = false;
> > +
> > +  if (code == POINTER_PLUS_EXPR)
>
> checking
>
>           && op_num == 2

But if it's the first operand then its type is the type of the stmt, i.e.,
we get the same type anyway, right?

Thanks,
Ira

>
> (with a possible fixup for the reduction case where that doesn't seem to
> be prevailing).
>
> Thanks,
> Richard.
>
> > +    vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
> > +  else
> > +    vector_type = STMT_VINFO_VECTYPE (stmt_vinfo);
> >
> >   gcc_assert (vector_type);
> >   nunits = TYPE_VECTOR_SUBPARTS (vector_type);
> >
> > Thanks,
> > Ira
> >
> >
> >

Re: [patch] Fix PR tree-optimization/45902

[Richard Guenther]

On Mon, Oct 11, 2010 at 10:48 PM, Ira Rosen <IRAR@il.ibm.com> wrote:
>
>
> Richard Guenther <richard.guenther@gmail.com> wrote on 11/10/2010 04:03:07
> PM:
>
>>
>> On Mon, Oct 11, 2010 at 3:03 PM, Ira Rosen <IRAR@il.ibm.com> wrote:
>> > Richard Guenther <richard.guenther@gmail.com> wrote on 11/10/2010
> 02:32:08
>> > PM:
>> >
>> >>
>> >> On Mon, Oct 11, 2010 at 2:19 PM, Ira Rosen <IRAR@il.ibm.com> wrote:
>> >> >
>> >> > Hi,
>> >> >
>> >> > This patch fixes a bug in creation of a vector of constants in SLP.
> The
>> >> > problem is in the type of created vector. It should be set to the
>> > vector
>> >> > type of the statement unless it's a pointer.
>> >> >
>> >> > Bootstrapped and tested on x86_64-suse-linux and checked that the
>> > failure
>> >> > is fixed on powerpc64-suse-linux.
>> >> >
>> >> > Committed to mainline, ok for 4.5?
>> >>
>> >> I don't think this makes sense.  If at all the selection of which type
>> >> to use should be based on the operation code and the operand
>> >> position, but not on CONSTANT_CLASS_P or pointer-type-ness.
>> >>
>> >> So - what operation code / operand position is currently mishandled?
>> >
>> > This function creates vectors of constants or loop invariants. If it's
> a
>> > variable (loop invariant), we know exactly the type, hence the check
> for
>> > CONSTANT_CLASS_P.
>> >
>> > I guess instead of checking the type we can check if it's
> POINTER_PLUS_EXPR
>> > to catch the cases where we need the operand's type instead of the
> stmt's
>> > vectype.
>> >
>> > Does this make sense?
>>
>> That makes more sense, with ...
>>
>> > Index: tree-vect-slp.c
>> > ===================================================================
>> > --- tree-vect-slp.c     (revision 165302)
>> > +++ tree-vect-slp.c     (working copy)
>> > @@ -1836,10 +1836,10 @@ vect_get_constant_vectors (slp_tree slp_
>> >   VEC (tree, heap) *voprnds = VEC_alloc (tree, heap,
> number_of_vectors);
>> >   bool constant_p, is_store;
>> >   tree neutral_op = NULL;
>> > +  enum tree_code code = gimple_assign_rhs_code (stmt);
>> >
>> >   if (STMT_VINFO_DEF_TYPE (stmt_vinfo) == vect_reduction_def)
>> >     {
>> > -      enum tree_code code = gimple_assign_rhs_code (stmt);
>> >       if (reduc_index == -1)
>> >         {
>> >           VEC_free (tree, heap, *vec_oprnds);
>> > @@ -1895,18 +1895,14 @@ vect_get_constant_vectors (slp_tree slp_
>> >     }
>> >
>> >   if (CONSTANT_CLASS_P (op))
>> > -    {
>> > -      constant_p = true;
>> > -      if (POINTER_TYPE_P (TREE_TYPE (gimple_assign_lhs (stmt))))
>> > -        vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
>> > -      else
>> > -        vector_type = STMT_VINFO_VECTYPE (stmt_vinfo);
>> > -    }
>> > +    constant_p = true;
>> >   else
>> > -    {
>> > -      constant_p = false;
>> > -      vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
>> > -    }
>> > +    constant_p = false;
>> > +
>> > +  if (code == POINTER_PLUS_EXPR)
>>
>> checking
>>
>>           && op_num == 2
>
> But if it's the first operand then its type is the type of the stmt, i.e.,
> we get the same type anyway, right?

Yes.  So your patch looks ok, possibly with a comment that looking
at either operand is ok.

Thanks,
Richard.

> Thanks,
> Ira
>
>>
>> (with a possible fixup for the reduction case where that doesn't seem to
>> be prevailing).
>>
>> Thanks,
>> Richard.
>>
>> > +    vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
>> > +  else
>> > +    vector_type = STMT_VINFO_VECTYPE (stmt_vinfo);
>> >
>> >   gcc_assert (vector_type);
>> >   nunits = TYPE_VECTOR_SUBPARTS (vector_type);
>> >
>> > Thanks,
>> > Ira
>> >
>> >
>> >
>
>

Re: [patch] Fix PR tree-optimization/45902

[Ira Rosen]

Richard Guenther <richard.guenther@gmail.com> wrote on 12/10/2010 12:08:31
PM:

> >
> > But if it's the first operand then its type is the type of the stmt,
i.e.,
> > we get the same type anyway, right?
>
> Yes.  So your patch looks ok, possibly with a comment that looking
> at either operand is ok.
>

I committed the following:

	* tree-vect-slp.c (vect_get_constant_vectors): Fix comment.
	Use operand's type for POINTER_PLUS_EXPR.


Index: tree-vect-slp.c
===================================================================
--- tree-vect-slp.c     (revision 165411)
+++ tree-vect-slp.c     (working copy)
@@ -1811,8 +1811,8 @@ vect_update_slp_costs_according_to_vf (l

 /* For constant and loop invariant defs of SLP_NODE this function returns
    (vector) defs (VEC_OPRNDS) that will be used in the vectorized stmts.
-   OP_NUM determines if we gather defs for operand 0 or operand 1 of the
scalar
-   stmts. NUMBER_OF_VECTORS is the number of vector defs to create.
+   OP_NUM determines if we gather defs for operand 0 or operand 1 of the
RHS of
+   scalar stmts.  NUMBER_OF_VECTORS is the number of vector defs to
create.
    REDUC_INDEX is the index of the reduction operand in the statements,
unless
    it is -1.  */

@@ -1836,10 +1836,10 @@ vect_get_constant_vectors (slp_tree slp_
   VEC (tree, heap) *voprnds = VEC_alloc (tree, heap, number_of_vectors);
   bool constant_p, is_store;
   tree neutral_op = NULL;
+  enum tree_code code = gimple_assign_rhs_code (stmt);

   if (STMT_VINFO_DEF_TYPE (stmt_vinfo) == vect_reduction_def)
     {
-      enum tree_code code = gimple_assign_rhs_code (stmt);
       if (reduc_index == -1)
         {
           VEC_free (tree, heap, *vec_oprnds);
@@ -1895,18 +1895,18 @@ vect_get_constant_vectors (slp_tree slp_
     }

   if (CONSTANT_CLASS_P (op))
-    {
-      constant_p = true;
-      if (POINTER_TYPE_P (TREE_TYPE (gimple_assign_lhs (stmt))))
-        vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
-      else
-        vector_type = STMT_VINFO_VECTYPE (stmt_vinfo);
-    }
+    constant_p = true;
   else
-    {
-      constant_p = false;
-      vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
-    }
+    constant_p = false;
+
+  /* For POINTER_PLUS_EXPR we use the type of the constant/invariant
itself.
+     If OP is the first operand of POINTER_PLUS_EXPR, its type is the type
of
+     the statement, so it's OK to use OP's type for both first and second
+     operands.  */
+  if (code == POINTER_PLUS_EXPR)
+    vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
+  else
+    vector_type = STMT_VINFO_VECTYPE (stmt_vinfo);

   gcc_assert (vector_type);
   nunits = TYPE_VECTOR_SUBPARTS (vector_type);


And here is an updated patch for 4.5.
Bootstrapped and tested on x86_64-suse-linux.
OK to commit?

Thanks,
Ira

ChangeLog:

	PR tree-optimization/45902
	* tree-vect-slp.c (vect_get_constant_vectors): Use operand's type
	for POINTER_PLUS_EXPR.

testsuite/ChangeLog:

	PR tree-optimization/45902
	* gcc.dg/vect/45902.c: New test.


Index: tree-vect-slp.c
===================================================================
--- tree-vect-slp.c     (revision 165366)
+++ tree-vect-slp.c     (working copy)
@@ -1446,6 +1446,7 @@ vect_get_constant_vectors (slp_tree slp_
   int number_of_copies = 1;
   VEC (tree, heap) *voprnds = VEC_alloc (tree, heap, number_of_vectors);
   bool constant_p, is_store;
+  enum tree_code code = gimple_assign_rhs_code (stmt);

   if (STMT_VINFO_DATA_REF (stmt_vinfo))
     {
@@ -1463,9 +1464,16 @@ vect_get_constant_vectors (slp_tree slp_
   else
     constant_p = false;

-  vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
-  gcc_assert (vector_type);
+  /* For POINTER_PLUS_EXPR we use the type of the constant/invariant
itself.
+     If OP is the first operand of POINTER_PLUS_EXPR, its type is the type
of
+     the statement, so it's OK to use OP's type for both first and second
+     operands.  */
+  if (code == POINTER_PLUS_EXPR)
+    vector_type = get_vectype_for_scalar_type (TREE_TYPE (op));
+  else
+    vector_type = STMT_VINFO_VECTYPE (stmt_vinfo);

+  gcc_assert (vector_type);
   nunits = TYPE_VECTOR_SUBPARTS (vector_type);

   /* NUMBER_OF_COPIES is the number of times we need to use the same
values in
Index: testsuite/gcc.dg/vect/pr45902.c
===================================================================
--- testsuite/gcc.dg/vect/pr45902.c     (revision 0)
+++ testsuite/gcc.dg/vect/pr45902.c     (revision 0)
@@ -0,0 +1,43 @@
+/* { dg-require-effective-target vect_int } */
+
+#include <stdarg.h>
+#include <stdlib.h>
+#include "tree-vect.h"
+
+#define N 128
+
+short res[N];
+short a[N];
+
+int
+main1 ()
+{
+  int i;
+
+  for (i = 0; i < N/4; i+=4)
+    {
+      res[i] = a[i] >> 8;
+      res[i+1] = a[i+1] >> 8;
+      res[i+2] = a[i+2] >> 8;
+      res[i+3] = a[i+3] >> 8;
+    }
+}
+
+int
+main ()
+{
+  int i;
+
+  for (i = 0; i < N; i++)
+    a[i] = i;
+
+  main1 ();
+
+  for (i = 0; i < N; i++)
+    if (res[i] != a[i] >> 8)
+      abort ();
+
+  return 0;
+}
+
+/* { dg-final { cleanup-tree-dump "vect" } } */

Re: [patch] Fix PR tree-optimization/45902

[H.J. Lu]

On Wed, Oct 13, 2010 at 12:43 AM, Ira Rosen <IRAR@il.ibm.com> wrote:
>
>
> Richard Guenther <richard.guenther@gmail.com> wrote on 12/10/2010 12:08:31
> PM:
>
>> >
>> > But if it's the first operand then its type is the type of the stmt,
> i.e.,
>> > we get the same type anyway, right?
>>
>> Yes.  So your patch looks ok, possibly with a comment that looking
>> at either operand is ok.
>>
>
> I committed the following:
>
>        * tree-vect-slp.c (vect_get_constant_vectors): Fix comment.
>        Use operand's type for POINTER_PLUS_EXPR.
>
>

This caused:

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46049


-- 
H.J.

Re: [patch] Fix PR tree-optimization/45902

[H.J. Lu]

On Sat, Oct 16, 2010 at 10:45 AM, H.J. Lu <hjl.tools@gmail.com> wrote:
> On Wed, Oct 13, 2010 at 12:43 AM, Ira Rosen <IRAR@il.ibm.com> wrote:
>>
>>
>> Richard Guenther <richard.guenther@gmail.com> wrote on 12/10/2010 12:08:31
>> PM:
>>
>>> >
>>> > But if it's the first operand then its type is the type of the stmt,
>> i.e.,
>>> > we get the same type anyway, right?
>>>
>>> Yes.  So your patch looks ok, possibly with a comment that looking
>>> at either operand is ok.
>>>
>>
>> I committed the following:
>>
>>        * tree-vect-slp.c (vect_get_constant_vectors): Fix comment.
>>        Use operand's type for POINTER_PLUS_EXPR.
>>
>>
>
> This caused:
>
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46049
>
>

It also caused:

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46052


-- 
H.J.