public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
From: Jason Merrill <jason@redhat.com>
To: Marek Polacek <polacek@redhat.com>
Cc: Jakub Jelinek <jakub@redhat.com>,
	GCC Patches <gcc-patches@gcc.gnu.org>,
	Joseph Myers <joseph@codesourcery.com>
Subject: Re: [PATCH v3] attribs: Fix wrong error with -Wno-attribute=A::b [PR103649]
Date: Fri, 17 Dec 2021 14:33:01 -0500	[thread overview]
Message-ID: <6f8ef9bc-47c2-ccb5-6238-a039114d7047@redhat.com> (raw)
In-Reply-To: <YbzTm2OK51VQvqUt@redhat.com>

On 12/17/21 13:14, Marek Polacek wrote:
> On Thu, Dec 16, 2021 at 08:06:43PM -0500, Jason Merrill wrote:
>> On 12/16/21 19:52, Marek Polacek wrote:
>>> --- /dev/null
>>> +++ b/gcc/testsuite/c-c++-common/Wno-attributes-6.c
>>> @@ -0,0 +1,14 @@
>>> +/* PR c/103649 */
>>> +/* { dg-do compile { target { c || c++11 } } } */
>>> +/* { dg-additional-options "-Wno-attributes=foo::bar" } */
>>> +/* { dg-additional-options "-Wno-attributes=baz::" } */
>>> +/* { dg-additional-options "-Wno-attributes=womp::womp" } */
>>> +/* { dg-additional-options "-Wno-attributes=qux::foo" } */
>>> +
>>> +[[foo::bar(1, 2)]]; /* { dg-warning "attribute ignored" } */
>>> +[[baz::bar(1, 2)]]; /* { dg-warning "attribute ignored" } */
>>> +[[foo::bar(1, 2)]] void f1();
>>> +[[baz::bar(1, 2)]] void f2();
>>> +[[qux::foo({t})]] void f3();
>>> +[[womp::womp (another::directive (threadprivate (t)))]] void f4();
>>> +[[womp::womp (another::directive (threadprivate (t)))]]; /* { dg-warning "attribute ignored" } */
>>
>> If we're ignoring these attributes, we should ignore them everywhere;
>> perhaps using them for an attribute-declaration is meaningful in another
>> compiler.
> 
> Sounds good.  Reminds me of P177, Portable assumptions...
> 
> Bootstrapped/regtested on x86_64-pc-linux-gnu, ok for trunk?
> 
> -- >8 --
> My patch to implement -Wno-attribute=A::b caused a bogus error when
> parsing
> 
>    [[foo::bar(1, 2)]];
> 
> when -Wno-attributes=foo::bar was specified on the command line, because
> when we create a fake foo::bar attribute and insert it into our attribute
> table, it is created with max_length == 0 which doesn't allow any args.
> That is wrong -- we know nothing about the attribute, so we shouldn't
> require any specific number of arguments.  And since unknown attributes
> can be rather complex (see for example omp::{directive,sequence}), we
> must skip parsing their arguments.  To that end, I'm using max_length
> with value -2.
> 
> Also let's not warn about things like
> 
>    [[vendor::assume(true)]];
> 
> because they may have some meaning (this is reminiscent of C++ Portable
> Assumptions).
> 
> 	PR c/103649
> 
> gcc/ChangeLog:
> 
> 	* attribs.c (handle_ignored_attributes_option): Create the fake
> 	attribute with max_length == -2.
> 	(attribute_ignored_p): New function.
> 	* attribs.h (attribute_ignored_p): Declare.
> 	* tree-core.h (struct attribute_spec): Document that max_length
> 	can be -2.
> 
> gcc/c/ChangeLog:
> 
> 	* c-decl.c (c_warn_unused_attributes): Don't warn for
> 	attribute_ignored_p.
> 	* c-parser.c (c_parser_std_attribute): Skip parsing of the attribute
> 	arguments when max_length == -2.
> 
> gcc/cp/ChangeLog:
> 
> 	* parser.c (cp_parser_declaration): Don't warn for attribute_ignored_p.
> 	(cp_parser_std_attribute): Skip parsing of the attribute
> 	arguments when max_length == -2.
> 
> gcc/testsuite/ChangeLog:
> 
> 	* c-c++-common/Wno-attributes-6.c: New test.
> ---
>   gcc/attribs.c                                 | 20 ++++++++++++++++++-
>   gcc/attribs.h                                 |  1 +
>   gcc/c/c-decl.c                                |  2 +-
>   gcc/c/c-parser.c                              |  4 +++-
>   gcc/cp/parser.c                               |  6 ++++--
>   gcc/testsuite/c-c++-common/Wno-attributes-6.c | 16 +++++++++++++++
>   gcc/tree-core.h                               |  4 +++-
>   7 files changed, 47 insertions(+), 6 deletions(-)
>   create mode 100644 gcc/testsuite/c-c++-common/Wno-attributes-6.c
> 
> diff --git a/gcc/attribs.c b/gcc/attribs.c
> index 29703e75fba..82527869093 100644
> --- a/gcc/attribs.c
> +++ b/gcc/attribs.c
> @@ -304,7 +304,7 @@ handle_ignored_attributes_option (vec<char *> *v)
>   	 We can't free it here, so squirrel away the pointers.  */
>         attribute_spec *table = new attribute_spec[2];
>         ignored_attributes_table.safe_push (table);
> -      table[0] = { attr, 0, 0, false, false, false, false, nullptr, nullptr };
> +      table[0] = { attr, 0, -2, false, false, false, false, nullptr, nullptr };
>         table[1] = { nullptr, 0, 0, false, false, false, false, nullptr,
>   		   nullptr };
>         register_scoped_attributes (table, IDENTIFIER_POINTER (vendor_id), !attr);
> @@ -569,6 +569,24 @@ attr_namespace_ignored_p (tree ns)
>     return r && r->ignored_p;
>   }
>   
> +/* Return true if the attribute ATTR should not be warned about.  */
> +
> +bool
> +attribute_ignored_p (tree attr)
> +{
> +  if (!cxx11_attribute_p (attr))
> +    return false;
> +  if (tree ns = get_attribute_namespace (attr))
> +    {
> +      if (attr_namespace_ignored_p (ns))
> +	return true;
> +      const attribute_spec *as = lookup_attribute_spec (TREE_PURPOSE (attr));
> +      if (as && as->max_length == -2)
> +	return true;
> +    }
> +  return false;
> +}
> +
>   /* Process the attributes listed in ATTRIBUTES and install them in *NODE,
>      which is either a DECL (including a TYPE_DECL) or a TYPE.  If a DECL,
>      it should be modified in place; if a TYPE, a copy should be created
> diff --git a/gcc/attribs.h b/gcc/attribs.h
> index f5899d83c0b..d3449d5bd1d 100644
> --- a/gcc/attribs.h
> +++ b/gcc/attribs.h
> @@ -39,6 +39,7 @@ extern tree get_attribute_name (const_tree);
>   extern tree get_attribute_namespace (const_tree);
>   extern void apply_tm_attr (tree, tree);
>   extern tree make_attribute (const char *, const char *, tree);
> +extern bool attribute_ignored_p (tree);
>   
>   extern struct scoped_attributes* register_scoped_attributes (const struct attribute_spec *,
>   							     const char *,
> diff --git a/gcc/c/c-decl.c b/gcc/c/c-decl.c
> index 4b5481ce8f4..516e3c26e21 100644
> --- a/gcc/c/c-decl.c
> +++ b/gcc/c/c-decl.c
> @@ -4643,7 +4643,7 @@ c_warn_unused_attributes (tree attrs)
>   	 constraint violation.  */
>         pedwarn (input_location, OPT_Wattributes, "%qE attribute ignored",
>   	       get_attribute_name (t));
> -    else
> +    else if (!attribute_ignored_p (t))
>         warning (OPT_Wattributes, "%qE attribute ignored",
>   	       get_attribute_name (t));
>   }
> diff --git a/gcc/c/c-parser.c b/gcc/c/c-parser.c
> index d7e5f051ac0..c9068bdbb8a 100644
> --- a/gcc/c/c-parser.c
> +++ b/gcc/c/c-parser.c
> @@ -4943,7 +4943,9 @@ c_parser_std_attribute (c_parser *parser, bool for_tm)
>   	parens.skip_until_found_close (parser);
>   	return error_mark_node;
>         }
> -    if (as)
> +    /* When MAX_LENGTH is -2, this is a fake attribute created to
> +       handle -Wno-attributes, and we must skip parsing the arguments.  */
> +    if (as && as->max_length != -2)
>         {
>   	bool takes_identifier
>   	  = (ns != NULL_TREE
> diff --git a/gcc/cp/parser.c b/gcc/cp/parser.c
> index 44eed7ea638..a843c953da8 100644
> --- a/gcc/cp/parser.c
> +++ b/gcc/cp/parser.c
> @@ -14776,7 +14776,7 @@ cp_parser_declaration (cp_parser* parser, tree prefix_attrs)
>   	    }
>   	}
>   
> -      if (std_attrs != NULL_TREE)
> +      if (std_attrs != NULL_TREE && !attribute_ignored_p (std_attrs))
>   	warning_at (make_location (attrs_loc, attrs_loc, parser->lexer),
>   		    OPT_Wattributes, "attribute ignored");
>         if (cp_lexer_next_token_is (parser->lexer, CPP_SEMICOLON))
> @@ -28979,7 +28979,9 @@ cp_parser_std_attribute (cp_parser *parser, tree attr_ns)
>         /* A GNU attribute that takes an identifier in parameter.  */
>         attr_flag = id_attr;
>   
> -    if (as == NULL)
> +    /* When MAX_LENGTH is -2, this is a fake attribute created to
> +       handle -Wno-attributes, and we must skip parsing the arguments.  */
> +    if (as == NULL || as->max_length == -2)

Rather than require multiple places to know about the magic -2, could we 
have an overload of attribute_ignored_p that takes an attribute_spec*? 
OK with that change.

>         {
>   	if ((flag_openmp || flag_openmp_simd) && attr_ns == omp_identifier)
>   	  {
> diff --git a/gcc/testsuite/c-c++-common/Wno-attributes-6.c b/gcc/testsuite/c-c++-common/Wno-attributes-6.c
> new file mode 100644
> index 00000000000..ca3c7bebb99
> --- /dev/null
> +++ b/gcc/testsuite/c-c++-common/Wno-attributes-6.c
> @@ -0,0 +1,16 @@
> +/* PR c/103649 */
> +/* { dg-do compile { target { c || c++11 } } } */
> +/* { dg-additional-options "-Wno-attributes=foo::bar" } */
> +/* { dg-additional-options "-Wno-attributes=baz::" } */
> +/* { dg-additional-options "-Wno-attributes=womp::womp" } */
> +/* { dg-additional-options "-Wno-attributes=qux::foo" } */
> +/* { dg-additional-options "-Wno-attributes=vendor::assume" } */
> +
> +[[vendor::assume(1 + 1 == 2)]];
> +[[foo::bar(1, 2)]];
> +[[baz::bar(1, 2)]];
> +[[foo::bar(1, 2)]] void f1();
> +[[baz::bar(1, 2)]] void f2();
> +[[qux::foo({t})]] void f3();
> +[[womp::womp (another::directive (threadprivate (t)))]] void f4();
> +[[womp::womp (another::directive (threadprivate (t)))]];
> diff --git a/gcc/tree-core.h b/gcc/tree-core.h
> index 91ae5237d7e..9b37a065d18 100644
> --- a/gcc/tree-core.h
> +++ b/gcc/tree-core.h
> @@ -2077,7 +2077,9 @@ struct attribute_spec {
>     /* The minimum length of the list of arguments of the attribute.  */
>     int min_length;
>     /* The maximum length of the list of arguments of the attribute
> -     (-1 for no maximum).  */
> +     (-1 for no maximum).  It can also be -2 for fake attributes
> +     created for the sake of -Wno-attributes; in that case, we
> +     should skip the balanced token sequence when parsing the attribute.  */
>     int max_length;
>     /* Whether this attribute requires a DECL.  If it does, it will be passed
>        from types of DECLs, function return types and array element types to
> 
> base-commit: d7ca2a79b82c6500ead6ab983d14c609e2124eee


  reply	other threads:[~2021-12-17 19:33 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-12-16 22:35 [PATCH] " Marek Polacek
2021-12-16 22:53 ` Jakub Jelinek
2021-12-17  0:52   ` [PATCH v2] " Marek Polacek
2021-12-17  1:06     ` Jason Merrill
2021-12-17 18:14       ` [PATCH v3] " Marek Polacek
2021-12-17 19:33         ` Jason Merrill [this message]
2021-12-17 22:56           ` [PATCH v4] " Marek Polacek

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=6f8ef9bc-47c2-ccb5-6238-a039114d7047@redhat.com \
    --to=jason@redhat.com \
    --cc=gcc-patches@gcc.gnu.org \
    --cc=jakub@redhat.com \
    --cc=joseph@codesourcery.com \
    --cc=polacek@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).