public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
From: Kai Tietz <ktietz70@googlemail.com>
To: Gabriel Dos Reis <gdr@integrable-solutions.net>
Cc: Dave Korn <dave.korn.cygwin@gmail.com>,
	GCC Patches <gcc-patches@gcc.gnu.org>,
		"Joseph S. Myers" <joseph@codesourcery.com>,
	Jason Merrill <jason@redhat.com>,
		Richard Henderson <rth@redhat.com>
Subject: Re: [patch c,c++,i386]:PR/15774 - Conflicting function decls not diagnosed
Date: Wed, 22 Dec 2010 16:38:00 -0000	[thread overview]
Message-ID: <AANLkTimX2iCERap2asvia4Ft-k0-1qqEt3hVhx0CBTwp@mail.gmail.com> (raw)
In-Reply-To: <AANLkTimVn_=uqH7x=0YL3Wqs9CHmrXHY0F9qz-DE4HQW@mail.gmail.com>

[-- Attachment #1: Type: text/plain, Size: 1352 bytes --]

2010/12/22 Gabriel Dos Reis <gdr@integrable-solutions.net>:
> On Wed, Dec 22, 2010 at 9:34 AM, Kai Tietz <ktietz70@googlemail.com> wrote:
>> 2010/12/22 Gabriel Dos Reis <gdr@integrable-solutions.net>:
>>> Any reason why is_first is of type `int' as opposed to `bool'?
>>> rename pp_c_attributes_calling_convention to
>>> pp_c_calling_convention_attributes.
>>> You should probably modify cp/cxx-pretty-print.c too
>>
>> Well, I was first think here to implement logic different, but you are
>> right, bool is better here. I adjusted patch. Also renamed function.
>>
>>> I really prefer we don't add pretty printing code to cp/error.c.
>>> Rather, we should be deferring to the pretty printers in
>>> cp/cxx-pretty-print.c.  But, I suspect that is a battle for
>>> another day.
>>
>> Well, I tried to do thing here first in cxx-pretty-print.c, but it
>> doesn't had any effect for function pointers, so I did it just in
>> error.c file.  This seems to be a different battle for a different
>> day.
>
> The problem is we are going to have the same regression when
> the cp/error.c to what it should be (as originally plan).
> So, my suggestion to is to add similar code to cp/cxx-pretty-print.c.
> Patch approved with that modification.
>


Good, I added code to cp/cxx-pretty-print.c (pp_cxx_ptr_operator).

Regards,
Kai

[-- Attachment #2: dispcallcvt.txt --]
[-- Type: text/plain, Size: 7582 bytes --]

Index: gcc/gcc/c-family/c-pretty-print.c
===================================================================
--- gcc.orig/gcc/c-family/c-pretty-print.c	2010-12-22 16:29:55.231892000 +0100
+++ gcc/gcc/c-family/c-pretty-print.c	2010-12-22 16:39:57.923249700 +0100
@@ -29,6 +29,7 @@ along with GCC; see the file COPYING3.  
 #include "tree-pretty-print.h"
 #include "tree-iterator.h"
 #include "diagnostic.h"
+#include "target.h"
 
 /* Translate if being used for diagnostics, but not for dump files or
    __PRETTY_FUNCTION.  */
@@ -460,6 +461,7 @@ pp_c_specifier_qualifier_list (c_pretty_
 	  {
 	    pp_c_whitespace (pp);
 	    pp_c_left_paren (pp);
+	    pp_c_calling_convention_attributes (pp, TYPE_ATTRIBUTES (pointee));
 	  }
 	else if (!c_dialect_cxx ())
 	  pp_c_whitespace (pp);
@@ -790,6 +792,45 @@ pp_c_attributes (c_pretty_printer *pp, t
   pp_c_right_paren (pp);
 }
 
+/* Pretty-print ATTRIBUTES using GNU C extension syntax for calling
+   convention affecting attributes.  */
+
+void
+pp_c_calling_convention_attributes (c_pretty_printer *pp, tree a)
+{
+  bool is_first = true;
+
+  if (a == NULL_TREE)
+    return;
+
+  for (; a != NULL_TREE; a = TREE_CHAIN (a))
+    {
+      if (!targetm.attribute_affects_calling_convention (TREE_PURPOSE (a)))
+        continue;
+      if (is_first)
+	{
+	  pp_c_ws_string (pp, "__attribute__");
+	  pp_c_left_paren (pp);
+	  pp_c_left_paren (pp);
+	  is_first = false;
+	}
+      else
+	{
+	  pp_separate_with (pp, ',');
+	}
+      pp_tree_identifier (pp, TREE_PURPOSE (a));
+      if (TREE_VALUE (a))
+	pp_c_call_argument_list (pp, TREE_VALUE (a));
+    }
+
+  if (!is_first)
+    {
+      pp_c_right_paren (pp);
+      pp_c_right_paren (pp);
+      pp_c_whitespace (pp);
+    }
+}
+
 /* function-definition:
       declaration-specifiers declarator compound-statement  */
 
Index: gcc/gcc/cp/error.c
===================================================================
--- gcc.orig/gcc/cp/error.c	2010-12-22 16:29:55.237892000 +0100
+++ gcc/gcc/cp/error.c	2010-12-22 16:39:58.001375200 +0100
@@ -661,6 +661,8 @@ dump_type_prefix (tree t, int flags)
 	  {
 	    pp_cxx_whitespace (cxx_pp);
 	    pp_cxx_left_paren (cxx_pp);
+	    pp_c_calling_convention_attributes (pp_c_base (cxx_pp),
+						TYPE_ATTRIBUTES (sub));
 	  }
 	if (TREE_CODE (t) == POINTER_TYPE)
 	  pp_character(cxx_pp, '*');
Index: gcc/gcc/c-family/c-pretty-print.h
===================================================================
--- gcc.orig/gcc/c-family/c-pretty-print.h	2010-12-22 16:29:55.232892000 +0100
+++ gcc/gcc/c-family/c-pretty-print.h	2010-12-22 16:39:58.032625400 +0100
@@ -176,6 +176,7 @@ void pp_c_space_for_pointer_operator (c_
 void pp_c_tree_decl_identifier (c_pretty_printer *, tree);
 void pp_c_function_definition (c_pretty_printer *, tree);
 void pp_c_attributes (c_pretty_printer *, tree);
+void pp_c_calling_convention_attributes (c_pretty_printer *, tree);
 void pp_c_cv_qualifiers (c_pretty_printer *pp, int qualifiers, bool func_type);
 void pp_c_type_qualifier_list (c_pretty_printer *, tree);
 void pp_c_parameter_type_list (c_pretty_printer *, tree);
Index: gcc/gcc/config/i386/i386.c
===================================================================
--- gcc.orig/gcc/config/i386/i386.c	2010-12-22 16:29:55.235892000 +0100
+++ gcc/gcc/config/i386/i386.c	2010-12-22 16:39:58.048250500 +0100
@@ -5116,6 +5116,46 @@ ix86_function_ok_for_sibcall (tree decl,
   return true;
 }
 
+static bool
+ix86_attribute_affects_calling_convention (tree name)
+{
+  int ident_len;
+  const char *p;
+
+  gcc_assert (TREE_CODE (name) == IDENTIFIER_NODE);
+  p = IDENTIFIER_POINTER (name);
+  ident_len = IDENTIFIER_LENGTH (name);
+
+  if (ident_len > 4 && p[0] == '_' && p[1] == '_' && p[ident_len - 1] == '_'
+      && p[ident_len - 2] == '_')
+    {
+      ident_len -= 4;
+      p += 2;
+    }
+  switch (ident_len)
+    {
+    case 5:
+      if (!strncmp (p, "cdecl", 5))
+        return true;
+      break;
+    case 6:
+      if (!strncmp (p, "ms_abi", 6))
+        return true;
+      break;
+    case 7:
+      if (!strncmp (p, "regparm", 7) || !strncmp (p, "stdcall", 7))
+        return true;
+      break;
+    case 8:
+      if (!strncmp (p, "thiscall", 8) || !strncmp (p, "fastcall", 8)
+          || !strncmp (p, "sysv_abi", 8))
+        return true;
+      break;
+    }
+
+  return false;
+}
+
 /* Handle "cdecl", "stdcall", "fastcall", "regparm", "thiscall",
    and "sseregparm" calling convention attributes;
    arguments as in struct attribute_spec.handler.  */
@@ -34813,6 +34853,10 @@ ix86_autovectorize_vector_sizes (void)
 #undef TARGET_CONDITIONAL_REGISTER_USAGE
 #define TARGET_CONDITIONAL_REGISTER_USAGE ix86_conditional_register_usage
 
+#undef TARGET_ATTRIBUTE_AFFECTS_CALLING_CONVENTION
+#define TARGET_ATTRIBUTE_AFFECTS_CALLING_CONVENTION \
+  ix86_attribute_affects_calling_convention
+
 struct gcc_target targetm = TARGET_INITIALIZER;
 \f
 #include "gt-i386.h"
Index: gcc/gcc/doc/tm.texi
===================================================================
--- gcc.orig/gcc/doc/tm.texi	2010-12-22 16:29:55.239892000 +0100
+++ gcc/gcc/doc/tm.texi	2010-12-22 16:39:58.157626200 +0100
@@ -9765,6 +9765,10 @@ to perform initial processing of the @sa
 @file{i386/i386.c}, for example.
 @end deftypefn
 
+@deftypefn {Target Hook} bool TARGET_ATTRIBUTE_AFFECTS_CALLING_CONVENTION (const_tree @var{name})
+Returns @code{true} if given attribute by its @var{name} affects calling convention, otherwise @code{false}.
+@end deftypefn
+
 @deftypefn {Target Hook} bool TARGET_VALID_DLLIMPORT_ATTRIBUTE_P (const_tree @var{decl})
 @var{decl} is a variable or function with @code{__attribute__((dllimport))} specified.  Use this hook if the target needs to add extra validation checks to @code{handle_dll_attribute}.
 @end deftypefn
Index: gcc/gcc/doc/tm.texi.in
===================================================================
--- gcc.orig/gcc/doc/tm.texi.in	2010-12-22 16:29:55.241892000 +0100
+++ gcc/gcc/doc/tm.texi.in	2010-12-22 16:39:58.282627000 +0100
@@ -9729,6 +9729,8 @@ to perform initial processing of the @sa
 @file{i386/i386.c}, for example.
 @end deftypefn
 
+@hook TARGET_ATTRIBUTE_AFFECTS_CALLING_CONVENTION
+
 @hook TARGET_VALID_DLLIMPORT_ATTRIBUTE_P
 
 @defmac TARGET_DECLSPEC
Index: gcc/gcc/target.def
===================================================================
--- gcc.orig/gcc/target.def	2010-12-22 16:29:55.243892000 +0100
+++ gcc/gcc/target.def	2010-12-22 16:39:58.282627000 +0100
@@ -1113,6 +1113,14 @@ DEFHOOK
  tree, (tree olddecl, tree newdecl),
  merge_decl_attributes)
 
+/* Return true iff attribute NAME affects calling convention. */
+DEFHOOK
+(attribute_affects_calling_convention,
+ "Returns @code{true} if given attribute by its @var{name} affects calling\
+ convention, otherwise @code{false}.",
+ bool, (const_tree name),
+ hook_bool_const_tree_false)
+
 /* Given two types, merge their attributes and return the result.  */
 DEFHOOK
 (merge_type_attributes,
Index: gcc/gcc/cp/cxx-pretty-print.c
===================================================================
--- gcc.orig/gcc/cp/cxx-pretty-print.c	2010-12-22 12:10:20.000000000 +0100
+++ gcc/gcc/cp/cxx-pretty-print.c	2010-12-22 16:48:14.020174700 +0100
@@ -1323,6 +1323,9 @@ pp_cxx_ptr_operator (cxx_pretty_printer 
       if (TREE_CODE (TREE_TYPE (t)) == POINTER_TYPE
 	  || TYPE_PTR_TO_MEMBER_P (TREE_TYPE (t)))
 	pp_cxx_ptr_operator (pp, TREE_TYPE (t));
+      pp_c_calling_convention_attributes (pp_c_base (pp),
+					  TYPE_ATTRIBUTES (TREE_TYPE (t)));
+
       if (TREE_CODE (t) == POINTER_TYPE)
 	{
 	  pp_star (pp);

  reply	other threads:[~2010-12-22 15:54 UTC|newest]

Thread overview: 19+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-12-22 14:53 Kai Tietz
2010-12-22 14:58 ` Dave Korn
2010-12-22 15:11   ` Kai Tietz
2010-12-22 15:54     ` Gabriel Dos Reis
2010-12-22 16:21       ` Kai Tietz
2010-12-22 16:31         ` Gabriel Dos Reis
2010-12-22 16:38           ` Kai Tietz [this message]
2010-12-22 17:21 ` Joseph S. Myers
2010-12-22 17:38   ` Gabriel Dos Reis
2010-12-22 17:55   ` Kai Tietz
2010-12-22 17:57     ` Joseph S. Myers
2010-12-22 18:03       ` Kai Tietz
2010-12-22 18:10         ` Kai Tietz
2010-12-22 19:13         ` Joseph S. Myers
2010-12-22 18:24   ` Jason Merrill
2010-12-22 19:23     ` Kai Tietz
2010-12-22 19:25       ` Kai Tietz
2010-12-22 20:37         ` Kai Tietz
2010-12-23  8:19       ` Gabriel Dos Reis

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=AANLkTimX2iCERap2asvia4Ft-k0-1qqEt3hVhx0CBTwp@mail.gmail.com \
    --to=ktietz70@googlemail.com \
    --cc=dave.korn.cygwin@gmail.com \
    --cc=gcc-patches@gcc.gnu.org \
    --cc=gdr@integrable-solutions.net \
    --cc=jason@redhat.com \
    --cc=joseph@codesourcery.com \
    --cc=rth@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
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).