public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
* [PATCH RFA (cgraph)] c++: pragma target and static init [PR109753]
@ 2024-05-02 13:54 Jason Merrill
  2024-05-14 21:21 ` PING " Jason Merrill
  0 siblings, 1 reply; 4+ messages in thread
From: Jason Merrill @ 2024-05-02 13:54 UTC (permalink / raw)
  To: gcc-patches, Jan Hubicka

Tested x86_64-pc-linux-gnu, OK for trunk?  14.2?

This two-year-old thread seems relevant:
https://gcc.gnu.org/pipermail/gcc-patches/2022-April/593410.html

-- 8< --

 #pragma target and optimize should also apply to implicitly-generated
 functions like static initialization functions and defaulted special member
 functions.

At least one of the create_same_body_alias/handle_optimize_attribute changes
is necessary to avoid regressing g++.dg/opt/pr105306.C; maybe_clone_body
creates a cgraph_node for the ~B alias before handle_optimize_attribute, and
the alias never goes through finalize_function, so we need to adjust
semantic_interposition somewhere else.

	PR c++/109753

gcc/ChangeLog:

	* cgraph.cc (cgraph_node::create_same_body_alias): Set
	semantic_interposition.

gcc/c-family/ChangeLog:

	* c-attribs.cc (handle_optimize_attribute): Set
	cgraph_node::semantic_interposition.

gcc/cp/ChangeLog:

	* decl.cc (start_preparsed_function): Call decl_attributes.

gcc/testsuite/ChangeLog:

	* g++.dg/opt/always_inline1.C: New test.
---
 gcc/c-family/c-attribs.cc                 | 4 ++++
 gcc/cgraph.cc                             | 2 ++
 gcc/cp/decl.cc                            | 3 +++
 gcc/testsuite/g++.dg/opt/always_inline1.C | 8 ++++++++
 4 files changed, 17 insertions(+)
 create mode 100644 gcc/testsuite/g++.dg/opt/always_inline1.C

diff --git a/gcc/c-family/c-attribs.cc b/gcc/c-family/c-attribs.cc
index 04e39b41bdf..605469dd7dd 100644
--- a/gcc/c-family/c-attribs.cc
+++ b/gcc/c-family/c-attribs.cc
@@ -5971,6 +5971,10 @@ handle_optimize_attribute (tree *node, tree name, tree args,
       if (prev_target_node != target_node)
 	DECL_FUNCTION_SPECIFIC_TARGET (*node) = target_node;
 
+      /* Also update the cgraph_node, if it's already built.  */
+      if (cgraph_node *cn = cgraph_node::get (*node))
+	cn->semantic_interposition = flag_semantic_interposition;
+
       /* Restore current options.  */
       cl_optimization_restore (&global_options, &global_options_set,
 			       &cur_opts);
diff --git a/gcc/cgraph.cc b/gcc/cgraph.cc
index 473d8410bc9..f3bd2fa8ece 100644
--- a/gcc/cgraph.cc
+++ b/gcc/cgraph.cc
@@ -604,6 +604,8 @@ cgraph_node::create_same_body_alias (tree alias, tree decl)
 
   n = cgraph_node::create_alias (alias, decl);
   n->cpp_implicit_alias = true;
+  /* Aliases don't go through finalize_function.  */
+  n->semantic_interposition = opt_for_fn (decl, flag_semantic_interposition);
   if (symtab->cpp_implicit_aliases_done)
     n->resolve_alias (cgraph_node::get (decl));
   return n;
diff --git a/gcc/cp/decl.cc b/gcc/cp/decl.cc
index 378311c0f04..4531d830462 100644
--- a/gcc/cp/decl.cc
+++ b/gcc/cp/decl.cc
@@ -17796,6 +17796,9 @@ start_preparsed_function (tree decl1, tree attrs, int flags)
 	doing_friend = true;
     }
 
+  /* Adjust for #pragma target/optimize.  */
+  decl_attributes (&decl1, NULL_TREE, 0);
+
   if (DECL_DECLARED_INLINE_P (decl1)
       && lookup_attribute ("noinline", attrs))
     warning_at (DECL_SOURCE_LOCATION (decl1), 0,
diff --git a/gcc/testsuite/g++.dg/opt/always_inline1.C b/gcc/testsuite/g++.dg/opt/always_inline1.C
new file mode 100644
index 00000000000..a042a1cf0c6
--- /dev/null
+++ b/gcc/testsuite/g++.dg/opt/always_inline1.C
@@ -0,0 +1,8 @@
+// PR c++/109753
+// { dg-do compile { target x86_64-*-* } }
+
+#pragma GCC target("avx2")
+struct aa {
+    __attribute__((__always_inline__)) aa() {}
+};
+aa _M_impl;

base-commit: 2f15787f2e1a3afe2c2ad93d4eb0d3c1f73c8fbd
-- 
2.44.0


^ permalink raw reply	[flat|nested] 4+ messages in thread

* PING Re: [PATCH RFA (cgraph)] c++: pragma target and static init [PR109753]
  2024-05-02 13:54 [PATCH RFA (cgraph)] c++: pragma target and static init [PR109753] Jason Merrill
@ 2024-05-14 21:21 ` Jason Merrill
  2024-05-23 22:43   ` PING " Jason Merrill
  0 siblings, 1 reply; 4+ messages in thread
From: Jason Merrill @ 2024-05-14 21:21 UTC (permalink / raw)
  To: gcc-patches, Jan Hubicka

Ping

On 5/2/24 09:54, Jason Merrill wrote:
> Tested x86_64-pc-linux-gnu, OK for trunk?  14.2?
> 
> This two-year-old thread seems relevant:
> https://gcc.gnu.org/pipermail/gcc-patches/2022-April/593410.html
> 
> -- 8< --
> 
>   #pragma target and optimize should also apply to implicitly-generated
>   functions like static initialization functions and defaulted special member
>   functions.
> 
> At least one of the create_same_body_alias/handle_optimize_attribute changes
> is necessary to avoid regressing g++.dg/opt/pr105306.C; maybe_clone_body
> creates a cgraph_node for the ~B alias before handle_optimize_attribute, and
> the alias never goes through finalize_function, so we need to adjust
> semantic_interposition somewhere else.
> 
> 	PR c++/109753
> 
> gcc/ChangeLog:
> 
> 	* cgraph.cc (cgraph_node::create_same_body_alias): Set
> 	semantic_interposition.
> 
> gcc/c-family/ChangeLog:
> 
> 	* c-attribs.cc (handle_optimize_attribute): Set
> 	cgraph_node::semantic_interposition.
> 
> gcc/cp/ChangeLog:
> 
> 	* decl.cc (start_preparsed_function): Call decl_attributes.
> 
> gcc/testsuite/ChangeLog:
> 
> 	* g++.dg/opt/always_inline1.C: New test.
> ---
>   gcc/c-family/c-attribs.cc                 | 4 ++++
>   gcc/cgraph.cc                             | 2 ++
>   gcc/cp/decl.cc                            | 3 +++
>   gcc/testsuite/g++.dg/opt/always_inline1.C | 8 ++++++++
>   4 files changed, 17 insertions(+)
>   create mode 100644 gcc/testsuite/g++.dg/opt/always_inline1.C
> 
> diff --git a/gcc/c-family/c-attribs.cc b/gcc/c-family/c-attribs.cc
> index 04e39b41bdf..605469dd7dd 100644
> --- a/gcc/c-family/c-attribs.cc
> +++ b/gcc/c-family/c-attribs.cc
> @@ -5971,6 +5971,10 @@ handle_optimize_attribute (tree *node, tree name, tree args,
>         if (prev_target_node != target_node)
>   	DECL_FUNCTION_SPECIFIC_TARGET (*node) = target_node;
>   
> +      /* Also update the cgraph_node, if it's already built.  */
> +      if (cgraph_node *cn = cgraph_node::get (*node))
> +	cn->semantic_interposition = flag_semantic_interposition;
> +
>         /* Restore current options.  */
>         cl_optimization_restore (&global_options, &global_options_set,
>   			       &cur_opts);
> diff --git a/gcc/cgraph.cc b/gcc/cgraph.cc
> index 473d8410bc9..f3bd2fa8ece 100644
> --- a/gcc/cgraph.cc
> +++ b/gcc/cgraph.cc
> @@ -604,6 +604,8 @@ cgraph_node::create_same_body_alias (tree alias, tree decl)
>   
>     n = cgraph_node::create_alias (alias, decl);
>     n->cpp_implicit_alias = true;
> +  /* Aliases don't go through finalize_function.  */
> +  n->semantic_interposition = opt_for_fn (decl, flag_semantic_interposition);
>     if (symtab->cpp_implicit_aliases_done)
>       n->resolve_alias (cgraph_node::get (decl));
>     return n;
> diff --git a/gcc/cp/decl.cc b/gcc/cp/decl.cc
> index 378311c0f04..4531d830462 100644
> --- a/gcc/cp/decl.cc
> +++ b/gcc/cp/decl.cc
> @@ -17796,6 +17796,9 @@ start_preparsed_function (tree decl1, tree attrs, int flags)
>   	doing_friend = true;
>       }
>   
> +  /* Adjust for #pragma target/optimize.  */
> +  decl_attributes (&decl1, NULL_TREE, 0);
> +
>     if (DECL_DECLARED_INLINE_P (decl1)
>         && lookup_attribute ("noinline", attrs))
>       warning_at (DECL_SOURCE_LOCATION (decl1), 0,
> diff --git a/gcc/testsuite/g++.dg/opt/always_inline1.C b/gcc/testsuite/g++.dg/opt/always_inline1.C
> new file mode 100644
> index 00000000000..a042a1cf0c6
> --- /dev/null
> +++ b/gcc/testsuite/g++.dg/opt/always_inline1.C
> @@ -0,0 +1,8 @@
> +// PR c++/109753
> +// { dg-do compile { target x86_64-*-* } }
> +
> +#pragma GCC target("avx2")
> +struct aa {
> +    __attribute__((__always_inline__)) aa() {}
> +};
> +aa _M_impl;
> 
> base-commit: 2f15787f2e1a3afe2c2ad93d4eb0d3c1f73c8fbd


^ permalink raw reply	[flat|nested] 4+ messages in thread

* PING Re: PING Re: [PATCH RFA (cgraph)] c++: pragma target and static init [PR109753]
  2024-05-14 21:21 ` PING " Jason Merrill
@ 2024-05-23 22:43   ` Jason Merrill
  2024-05-29 13:49     ` [pushed] " Jason Merrill
  0 siblings, 1 reply; 4+ messages in thread
From: Jason Merrill @ 2024-05-23 22:43 UTC (permalink / raw)
  To: gcc-patches, Jan Hubicka

Ping

On 5/14/24 17:21, Jason Merrill wrote:
> Ping
> 
> On 5/2/24 09:54, Jason Merrill wrote:
>> Tested x86_64-pc-linux-gnu, OK for trunk?  14.2?
>>
>> This two-year-old thread seems relevant:
>> https://gcc.gnu.org/pipermail/gcc-patches/2022-April/593410.html
>>
>> -- 8< --
>>
>>   #pragma target and optimize should also apply to implicitly-generated
>>   functions like static initialization functions and defaulted special 
>> member
>>   functions.
>>
>> At least one of the create_same_body_alias/handle_optimize_attribute 
>> changes
>> is necessary to avoid regressing g++.dg/opt/pr105306.C; maybe_clone_body
>> creates a cgraph_node for the ~B alias before 
>> handle_optimize_attribute, and
>> the alias never goes through finalize_function, so we need to adjust
>> semantic_interposition somewhere else.
>>
>>     PR c++/109753
>>
>> gcc/ChangeLog:
>>
>>     * cgraph.cc (cgraph_node::create_same_body_alias): Set
>>     semantic_interposition.
>>
>> gcc/c-family/ChangeLog:
>>
>>     * c-attribs.cc (handle_optimize_attribute): Set
>>     cgraph_node::semantic_interposition.
>>
>> gcc/cp/ChangeLog:
>>
>>     * decl.cc (start_preparsed_function): Call decl_attributes.
>>
>> gcc/testsuite/ChangeLog:
>>
>>     * g++.dg/opt/always_inline1.C: New test.
>> ---
>>   gcc/c-family/c-attribs.cc                 | 4 ++++
>>   gcc/cgraph.cc                             | 2 ++
>>   gcc/cp/decl.cc                            | 3 +++
>>   gcc/testsuite/g++.dg/opt/always_inline1.C | 8 ++++++++
>>   4 files changed, 17 insertions(+)
>>   create mode 100644 gcc/testsuite/g++.dg/opt/always_inline1.C
>>
>> diff --git a/gcc/c-family/c-attribs.cc b/gcc/c-family/c-attribs.cc
>> index 04e39b41bdf..605469dd7dd 100644
>> --- a/gcc/c-family/c-attribs.cc
>> +++ b/gcc/c-family/c-attribs.cc
>> @@ -5971,6 +5971,10 @@ handle_optimize_attribute (tree *node, tree 
>> name, tree args,
>>         if (prev_target_node != target_node)
>>       DECL_FUNCTION_SPECIFIC_TARGET (*node) = target_node;
>> +      /* Also update the cgraph_node, if it's already built.  */
>> +      if (cgraph_node *cn = cgraph_node::get (*node))
>> +    cn->semantic_interposition = flag_semantic_interposition;
>> +
>>         /* Restore current options.  */
>>         cl_optimization_restore (&global_options, &global_options_set,
>>                      &cur_opts);
>> diff --git a/gcc/cgraph.cc b/gcc/cgraph.cc
>> index 473d8410bc9..f3bd2fa8ece 100644
>> --- a/gcc/cgraph.cc
>> +++ b/gcc/cgraph.cc
>> @@ -604,6 +604,8 @@ cgraph_node::create_same_body_alias (tree alias, 
>> tree decl)
>>     n = cgraph_node::create_alias (alias, decl);
>>     n->cpp_implicit_alias = true;
>> +  /* Aliases don't go through finalize_function.  */
>> +  n->semantic_interposition = opt_for_fn (decl, 
>> flag_semantic_interposition);
>>     if (symtab->cpp_implicit_aliases_done)
>>       n->resolve_alias (cgraph_node::get (decl));
>>     return n;
>> diff --git a/gcc/cp/decl.cc b/gcc/cp/decl.cc
>> index 378311c0f04..4531d830462 100644
>> --- a/gcc/cp/decl.cc
>> +++ b/gcc/cp/decl.cc
>> @@ -17796,6 +17796,9 @@ start_preparsed_function (tree decl1, tree 
>> attrs, int flags)
>>       doing_friend = true;
>>       }
>> +  /* Adjust for #pragma target/optimize.  */
>> +  decl_attributes (&decl1, NULL_TREE, 0);
>> +
>>     if (DECL_DECLARED_INLINE_P (decl1)
>>         && lookup_attribute ("noinline", attrs))
>>       warning_at (DECL_SOURCE_LOCATION (decl1), 0,
>> diff --git a/gcc/testsuite/g++.dg/opt/always_inline1.C 
>> b/gcc/testsuite/g++.dg/opt/always_inline1.C
>> new file mode 100644
>> index 00000000000..a042a1cf0c6
>> --- /dev/null
>> +++ b/gcc/testsuite/g++.dg/opt/always_inline1.C
>> @@ -0,0 +1,8 @@
>> +// PR c++/109753
>> +// { dg-do compile { target x86_64-*-* } }
>> +
>> +#pragma GCC target("avx2")
>> +struct aa {
>> +    __attribute__((__always_inline__)) aa() {}
>> +};
>> +aa _M_impl;
>>
>> base-commit: 2f15787f2e1a3afe2c2ad93d4eb0d3c1f73c8fbd
> 


^ permalink raw reply	[flat|nested] 4+ messages in thread

* [pushed] c++: pragma target and static init [PR109753]
  2024-05-23 22:43   ` PING " Jason Merrill
@ 2024-05-29 13:49     ` Jason Merrill
  0 siblings, 0 replies; 4+ messages in thread
From: Jason Merrill @ 2024-05-29 13:49 UTC (permalink / raw)
  To: gcc-patches

Revised to drop the cgraph change so I can self-approve the remaining patch.

Tested x86_64-pc-linux-gnu, applying to trunk.

-- 8< --

 #pragma target and optimize should also apply to implicitly-generated
 functions like static initialization functions and defaulted special member
 functions.

The handle_optimize_attribute change is necessary to avoid regressing
g++.dg/opt/pr105306.C; maybe_clone_body creates a cgraph_node for the ~B
alias before handle_optimize_attribute, and the alias never goes through
finalize_function, so we need to adjust semantic_interposition somewhere
else.

	PR c++/109753

gcc/c-family/ChangeLog:

	* c-attribs.cc (handle_optimize_attribute): Set
	cgraph_node::semantic_interposition.

gcc/cp/ChangeLog:

	* decl.cc (start_preparsed_function): Call decl_attributes.

gcc/testsuite/ChangeLog:

	* g++.dg/opt/always_inline1.C: New test.
---
 gcc/c-family/c-attribs.cc                 | 4 ++++
 gcc/cp/decl.cc                            | 3 +++
 gcc/testsuite/g++.dg/opt/always_inline1.C | 8 ++++++++
 3 files changed, 15 insertions(+)
 create mode 100644 gcc/testsuite/g++.dg/opt/always_inline1.C

diff --git a/gcc/c-family/c-attribs.cc b/gcc/c-family/c-attribs.cc
index 04e39b41bdf..605469dd7dd 100644
--- a/gcc/c-family/c-attribs.cc
+++ b/gcc/c-family/c-attribs.cc
@@ -5971,6 +5971,10 @@ handle_optimize_attribute (tree *node, tree name, tree args,
       if (prev_target_node != target_node)
 	DECL_FUNCTION_SPECIFIC_TARGET (*node) = target_node;
 
+      /* Also update the cgraph_node, if it's already built.  */
+      if (cgraph_node *cn = cgraph_node::get (*node))
+	cn->semantic_interposition = flag_semantic_interposition;
+
       /* Restore current options.  */
       cl_optimization_restore (&global_options, &global_options_set,
 			       &cur_opts);
diff --git a/gcc/cp/decl.cc b/gcc/cp/decl.cc
index a992d54dc8f..d481e1ec074 100644
--- a/gcc/cp/decl.cc
+++ b/gcc/cp/decl.cc
@@ -17832,6 +17832,9 @@ start_preparsed_function (tree decl1, tree attrs, int flags)
 	doing_friend = true;
     }
 
+  /* Adjust for #pragma target/optimize.  */
+  decl_attributes (&decl1, NULL_TREE, 0);
+
   if (DECL_DECLARED_INLINE_P (decl1)
       && lookup_attribute ("noinline", attrs))
     warning_at (DECL_SOURCE_LOCATION (decl1), 0,
diff --git a/gcc/testsuite/g++.dg/opt/always_inline1.C b/gcc/testsuite/g++.dg/opt/always_inline1.C
new file mode 100644
index 00000000000..a042a1cf0c6
--- /dev/null
+++ b/gcc/testsuite/g++.dg/opt/always_inline1.C
@@ -0,0 +1,8 @@
+// PR c++/109753
+// { dg-do compile { target x86_64-*-* } }
+
+#pragma GCC target("avx2")
+struct aa {
+    __attribute__((__always_inline__)) aa() {}
+};
+aa _M_impl;

base-commit: ff41abdca0ab9993b6170b9b1f46b3a40921f1b0
-- 
2.44.0


^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2024-05-29 13:51 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2024-05-02 13:54 [PATCH RFA (cgraph)] c++: pragma target and static init [PR109753] Jason Merrill
2024-05-14 21:21 ` PING " Jason Merrill
2024-05-23 22:43   ` PING " Jason Merrill
2024-05-29 13:49     ` [pushed] " Jason Merrill

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).