Defining _GNU_SOURCE hides the declaration of aligned_alloc

Defining _GNU_SOURCE hides the declaration of aligned_alloc

[Ken Brown]

The issue in the Subject line came up in connection with an emacs bug 
report.

Here's a test case:

$ cat test.c
#define _GNU_SOURCE
#include <stdlib.h>

int
main ()
{
   aligned_alloc (1, 1);
}

$ gcc test.c -Wimplicit-function-declaration
test.c: In function ‘main’:
test.c:7:3: warning: implicit declaration of function ‘aligned_alloc’

The cause is that the declaration of aligned_alloc in stdlib.h is 
guarded by #if __ISO_C_VISIBLE >= 2011 || __cplusplus >= 201103L; but 
defining _GNU_SOURCE causes __ISO_C_VISIBLE to be defined as 1999. 
Here's an excerpt from /usr/include/sys/cdefs.h showing how this happens:

/* Deal with _GNU_SOURCE, which implies everything and the kitchen sink */
#ifdef _GNU_SOURCE
[...]
#define    _XOPEN_SOURCE        700
[...]
#endif
[...]
#if _XOPEN_SOURCE - 0 >= 700
[...]
#define    _POSIX_C_SOURCE        200809
[...]
#endif
[...]
#if _POSIX_C_SOURCE >= 200809
[...]
#define    __ISO_C_VISIBLE        1999
[...]
#endif /* _POSIX_C_SOURCE */

According to the discussion of the emacs bug I mentioned, Linux and 
FreeBSD don't have this issue.  Should Cygwin's headers be changed to 
conform to those other platforms?

Ken

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Defining _GNU_SOURCE hides the declaration of aligned_alloc

[Ken Brown]

On 2/2/2016 5:27 PM, Ken Brown wrote:
> The issue in the Subject line came up in connection with an emacs bug
> report.
>
> Here's a test case:
>
> $ cat test.c
> #define _GNU_SOURCE
> #include <stdlib.h>
>
> int
> main ()
> {
>    aligned_alloc (1, 1);
> }
>
> $ gcc test.c -Wimplicit-function-declaration
> test.c: In function ‘main’:
> test.c:7:3: warning: implicit declaration of function ‘aligned_alloc’
>
> The cause is that the declaration of aligned_alloc in stdlib.h is
> guarded by #if __ISO_C_VISIBLE >= 2011 || __cplusplus >= 201103L; but
> defining _GNU_SOURCE causes __ISO_C_VISIBLE to be defined as 1999.
> Here's an excerpt from /usr/include/sys/cdefs.h showing how this happens:
>
> /* Deal with _GNU_SOURCE, which implies everything and the kitchen sink */
> #ifdef _GNU_SOURCE
> [...]
> #define    _XOPEN_SOURCE        700
> [...]
> #endif
> [...]
> #if _XOPEN_SOURCE - 0 >= 700
> [...]
> #define    _POSIX_C_SOURCE        200809
> [...]
> #endif
> [...]
> #if _POSIX_C_SOURCE >= 200809
> [...]
> #define    __ISO_C_VISIBLE        1999
> [...]
> #endif /* _POSIX_C_SOURCE */
>
> According to the discussion of the emacs bug I mentioned, Linux and
> FreeBSD don't have this issue.  Should Cygwin's headers be changed to
> conform to those other platforms?

Paul Eggert says they should:

> Defining _GNU_SOURCE should make aligned_alloc visible regardless of whether -std=c99 is specified. This is because defining _GNU_SOURCE means, "Make GNU symbols visible even when compiling pedantically." This is OK, since the C standard says the behavior is undefined whenever the user defines a reserved symbol like _GNU_SOURCE.

Ken

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: Defining _GNU_SOURCE hides the declaration of aligned_alloc

[Yaakov Selkowitz]

On 2016-02-02 16:27, Ken Brown wrote:
> The issue in the Subject line came up in connection with an emacs bug
> report.
>
> Here's a test case:
[snip]
> $ gcc test.c -Wimplicit-function-declaration
> test.c: In function ‘main’:
> test.c:7:3: warning: implicit declaration of function ‘aligned_alloc’
>
> The cause is that the declaration of aligned_alloc in stdlib.h is
> guarded by #if __ISO_C_VISIBLE >= 2011 || __cplusplus >= 201103L; but
> defining _GNU_SOURCE causes __ISO_C_VISIBLE to be defined as 1999.
> Here's an excerpt from /usr/include/sys/cdefs.h showing how this happens:
[snip]
> According to the discussion of the emacs bug I mentioned, Linux and
> FreeBSD don't have this issue.  Should Cygwin's headers be changed to
> conform to those other platforms?

Yes, it should.  I have added this to my queue of such fixes, which I 
hope to post soon.

-- 
Yaakov

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple