[PATCH][PR67666] Handle single restrict pointer in struct in create_variable_info_for_1

[PATCH][PR67666] Handle single restrict pointer in struct in create_variable_info_for_1

[Tom de Vries]

[-- Attachment #1: Type: text/plain, Size: 467 bytes --]

Hi,

Consider this test-case:

struct ps
{
   int *__restrict__ p;
};

void
f (struct ps &__restrict__ ps1)
{
   *(ps1.p) = 1;
}


Atm, the restrict on p has no effect. Now, say we add a field to the struct:

struct ps
{
   int *__restrict__ p;
   int a;
};


Then the restrict on p does have the desired effect.


This patch fixes the handling of structs with a single field in alias 
analysis.

Bootstrapped and reg-tested on x86_64.

OK for trunk?

Thanks,
- Tom


[-- Attachment #2: 0001-Handle-single-restrict-pointer-in-struct-in-create_v.patch --]
[-- Type: text/x-patch, Size: 2584 bytes --]

Handle single restrict pointer in struct in create_variable_info_for_1

2015-09-22  Tom de Vries  <tom@codesourcery.com>

	PR tree-optimization/67666
	* tree-ssa-structalias.c (create_variable_info_for_1): Handle struct
	with single field non-conservative.

	* g++.dg/pr67666.C: New test.
---
 gcc/testsuite/g++.dg/pr67666.C | 17 +++++++++++++++++
 gcc/tree-ssa-structalias.c     | 25 ++++++++++++++++---------
 2 files changed, 33 insertions(+), 9 deletions(-)
 create mode 100644 gcc/testsuite/g++.dg/pr67666.C

diff --git a/gcc/testsuite/g++.dg/pr67666.C b/gcc/testsuite/g++.dg/pr67666.C
new file mode 100644
index 0000000..ad162f4
--- /dev/null
+++ b/gcc/testsuite/g++.dg/pr67666.C
@@ -0,0 +1,17 @@
+// { dg-do compile }
+// { dg-options "-O2 -fdump-tree-ealias-all" }
+
+struct ps
+{
+  int *__restrict__ p;
+};
+
+void
+f (struct ps &__restrict__ ps1)
+{
+  *(ps1.p) = 1;
+}
+
+// { dg-final { scan-tree-dump-times "clique 1 base 1" 1 "ealias" } }
+// { dg-final { scan-tree-dump-times "clique 1 base 2" 1 "ealias" } }
+// { dg-final { scan-tree-dump-times "(?n)clique .* base .*" 2 "ealias" } }
diff --git a/gcc/tree-ssa-structalias.c b/gcc/tree-ssa-structalias.c
index b5b9d0a..94016b9 100644
--- a/gcc/tree-ssa-structalias.c
+++ b/gcc/tree-ssa-structalias.c
@@ -5675,7 +5675,7 @@ create_variable_info_for_1 (tree decl, const char *name)
 
   /* If we didn't end up collecting sub-variables create a full
      variable for the decl.  */
-  if (fieldstack.length () <= 1
+  if (fieldstack.length () == 0
       || fieldstack.length () > MAX_FIELDS_FOR_FIELD_SENSITIVE)
     {
       vi = new_var_info (decl, name);
@@ -5694,19 +5694,26 @@ create_variable_info_for_1 (tree decl, const char *name)
        fieldstack.iterate (i, &fo);
        ++i, newvi = vi_next (newvi))
     {
-      const char *newname = "NULL";
+      const char *newname = NULL;
       char *tempname;
 
       if (dump_file)
 	{
-	  tempname
-	    = xasprintf ("%s." HOST_WIDE_INT_PRINT_DEC
-			 "+" HOST_WIDE_INT_PRINT_DEC, name,
-			 fo->offset, fo->size);
-	  newname = ggc_strdup (tempname);
-	  free (tempname);
+	  if (fieldstack.length () != 1)
+	    {
+	      tempname
+		= xasprintf ("%s." HOST_WIDE_INT_PRINT_DEC
+			     "+" HOST_WIDE_INT_PRINT_DEC, name,
+			     fo->offset, fo->size);
+	      newname = ggc_strdup (tempname);
+	      free (tempname);
+	    }
 	}
-      newvi->name = newname;
+      else
+	newname = "NULL";
+
+      if (newname)
+	  newvi->name = newname;
       newvi->offset = fo->offset;
       newvi->size = fo->size;
       newvi->fullsize = vi->fullsize;
-- 
1.9.1

Re: [PATCH][PR67666] Handle single restrict pointer in struct in create_variable_info_for_1

[Richard Biener]

On Tue, 22 Sep 2015, Tom de Vries wrote:

> Hi,
> 
> Consider this test-case:
> 
> struct ps
> {
>   int *__restrict__ p;
> };
> 
> void
> f (struct ps &__restrict__ ps1)
> {
>   *(ps1.p) = 1;
> }
> 
> 
> Atm, the restrict on p has no effect. Now, say we add a field to the struct:
> 
> struct ps
> {
>   int *__restrict__ p;
>   int a;
> };
> 
> 
> Then the restrict on p does have the desired effect.
> 
> 
> This patch fixes the handling of structs with a single field in alias
> analysis.
> 
> Bootstrapped and reg-tested on x86_64.
> 
> OK for trunk?

Ok.

Thanks,
Richard.

Re: [PATCH][PR67666] Handle single restrict pointer in struct in create_variable_info_for_1

[Tom de Vries]

On 22/09/15 09:49, Richard Biener wrote:
> On Tue, 22 Sep 2015, Tom de Vries wrote:
>
>> Hi,
>>
>> Consider this test-case:
>>
>> struct ps
>> {
>>    int *__restrict__ p;
>> };
>>
>> void
>> f (struct ps &__restrict__ ps1)
>> {
>>    *(ps1.p) = 1;
>> }
>>
>>
>> Atm, the restrict on p has no effect. Now, say we add a field to the struct:
>>
>> struct ps
>> {
>>    int *__restrict__ p;
>>    int a;
>> };
>>
>>
>> Then the restrict on p does have the desired effect.
>>
>>
>> This patch fixes the handling of structs with a single field in alias
>> analysis.
>>
>> Bootstrapped and reg-tested on x86_64.
>>
>> OK for trunk?
>
> Ok.
>

Hi,

I wonder if this follow-up patch is necessary.

Now that we handle structs with one field in the final loop of 
create_variable_info_for_1, should we set the is_full_var field as well? 
It used to be set for such structs before I committed the "Handle single 
restrict pointer in struct in create_variable_info_for_1" patch.

Thanks,
- Tom

diff --git a/gcc/tree-ssa-structalias.c b/gcc/tree-ssa-structalias.c
index 8d86dcb..26d97a3 100644
--- a/gcc/tree-ssa-structalias.c
+++ b/gcc/tree-ssa-structalias.c
@@ -5720,6 +5720,8 @@ create_variable_info_for_1 (tree decl, const char 
*name)
        newvi->offset = fo->offset;
        newvi->size = fo->size;
        newvi->fullsize = vi->fullsize;
+      if (fieldstack.length () == 1)
+       newvi->is_full_var = true;
        newvi->may_have_pointers = fo->may_have_pointers;
        newvi->only_restrict_pointers = fo->only_restrict_pointers;
        if (i + 1 < fieldstack.length ())

Re: [PATCH][PR67666] Handle single restrict pointer in struct in create_variable_info_for_1

[Richard Biener]

On Tue, 29 Sep 2015, Tom de Vries wrote:

> On 22/09/15 09:49, Richard Biener wrote:
> > On Tue, 22 Sep 2015, Tom de Vries wrote:
> > 
> > > Hi,
> > > 
> > > Consider this test-case:
> > > 
> > > struct ps
> > > {
> > >    int *__restrict__ p;
> > > };
> > > 
> > > void
> > > f (struct ps &__restrict__ ps1)
> > > {
> > >    *(ps1.p) = 1;
> > > }
> > > 
> > > 
> > > Atm, the restrict on p has no effect. Now, say we add a field to the
> > > struct:
> > > 
> > > struct ps
> > > {
> > >    int *__restrict__ p;
> > >    int a;
> > > };
> > > 
> > > 
> > > Then the restrict on p does have the desired effect.
> > > 
> > > 
> > > This patch fixes the handling of structs with a single field in alias
> > > analysis.
> > > 
> > > Bootstrapped and reg-tested on x86_64.
> > > 
> > > OK for trunk?
> > 
> > Ok.
> > 
> 
> Hi,
> 
> I wonder if this follow-up patch is necessary.
> 
> Now that we handle structs with one field in the final loop of
> create_variable_info_for_1, should we set the is_full_var field as well? It
> used to be set for such structs before I committed the "Handle single restrict
> pointer in struct in create_variable_info_for_1" patch.

Yeah, I suppose so.  But I'd set vi->is_full_var to true when
allocating 'vi':

  vi = new_var_info (decl, name);
  vi->fullsize = tree_to_uhwi (declsize);
 +      if (fieldstack.length () == 1) 
 +       vi->is_full_var = true;


Ok with that change.

Thanks,
Richard.

> Thanks,
> - Tom
> 
> diff --git a/gcc/tree-ssa-structalias.c b/gcc/tree-ssa-structalias.c
> index 8d86dcb..26d97a3 100644
> --- a/gcc/tree-ssa-structalias.c
> +++ b/gcc/tree-ssa-structalias.c
> @@ -5720,6 +5720,8 @@ create_variable_info_for_1 (tree decl, const char *name)
>        newvi->offset = fo->offset;
>        newvi->size = fo->size;
>        newvi->fullsize = vi->fullsize;
> +      if (fieldstack.length () == 1)
> +       newvi->is_full_var = true;
>        newvi->may_have_pointers = fo->may_have_pointers;
>        newvi->only_restrict_pointers = fo->only_restrict_pointers;
>        if (i + 1 < fieldstack.length ())
> 
> 

-- 
Richard Biener <rguenther@suse.de>
SUSE LINUX GmbH, GF: Felix Imendoerffer, Jane Smithard, Graham Norton, HRB 21284 (AG Nuernberg)

Re: [PATCH][PR67666] Handle single restrict pointer in struct in create_variable_info_for_1

[Tom de Vries]

[-- Attachment #1: Type: text/plain, Size: 1480 bytes --]

On 29/09/15 10:29, Richard Biener wrote:
> On Tue, 29 Sep 2015, Tom de Vries wrote:
>
>> On 22/09/15 09:49, Richard Biener wrote:
>>> On Tue, 22 Sep 2015, Tom de Vries wrote:
>>>
>>>> Hi,
>>>>
>>>> Consider this test-case:
>>>>
>>>> struct ps
>>>> {
>>>>     int *__restrict__ p;
>>>> };
>>>>
>>>> void
>>>> f (struct ps &__restrict__ ps1)
>>>> {
>>>>     *(ps1.p) = 1;
>>>> }
>>>>
>>>>
>>>> Atm, the restrict on p has no effect. Now, say we add a field to the
>>>> struct:
>>>>
>>>> struct ps
>>>> {
>>>>     int *__restrict__ p;
>>>>     int a;
>>>> };
>>>>
>>>>
>>>> Then the restrict on p does have the desired effect.
>>>>
>>>>
>>>> This patch fixes the handling of structs with a single field in alias
>>>> analysis.
>>>>
>>>> Bootstrapped and reg-tested on x86_64.
>>>>
>>>> OK for trunk?
>>>
>>> Ok.
>>>
>>
>> Hi,
>>
>> I wonder if this follow-up patch is necessary.
>>
>> Now that we handle structs with one field in the final loop of
>> create_variable_info_for_1, should we set the is_full_var field as well? It
>> used to be set for such structs before I committed the "Handle single restrict
>> pointer in struct in create_variable_info_for_1" patch.
>
> Yeah, I suppose so.  But I'd set vi->is_full_var to true when
> allocating 'vi':
>
>    vi = new_var_info (decl, name);
>    vi->fullsize = tree_to_uhwi (declsize);
>   +      if (fieldstack.length () == 1)
>   +       vi->is_full_var = true;
>
>
> Ok with that change.
>

Committed as attached.

Thanks,
- Tom


[-- Attachment #2: 0002-Add-missing-is_full_var-setting-in-create_variable_i.patch --]
[-- Type: text/x-patch, Size: 805 bytes --]

Add missing is_full_var setting in create_variable_info_for_1

2015-10-23  Tom de Vries  <tom@codesourcery.com>

	* tree-ssa-structalias.c (create_variable_info_for_1): Add missing
	setting of is_full_var in case of a single field.
---
 gcc/tree-ssa-structalias.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/gcc/tree-ssa-structalias.c b/gcc/tree-ssa-structalias.c
index 8d86dcb..db0ab1e 100644
--- a/gcc/tree-ssa-structalias.c
+++ b/gcc/tree-ssa-structalias.c
@@ -5693,6 +5693,8 @@ create_variable_info_for_1 (tree decl, const char *name)
 
   vi = new_var_info (decl, name);
   vi->fullsize = tree_to_uhwi (declsize);
+  if (fieldstack.length () == 1)
+    vi->is_full_var = true;
   for (i = 0, newvi = vi;
        fieldstack.iterate (i, &fo);
        ++i, newvi = vi_next (newvi))
-- 
1.9.1