Shouldn't <sys/uio.h> be available everywhere?

Shouldn't <sys/uio.h> be available everywhere?

[Dionna Amalie Glaze via newlib]

I'm looking at the POSIX standard for readv and writev, and it looks
like there should be definitions for each of these given
__POSIX_VISIBLE >= 200112, not just being on a PowerPC platform with
an SPU. There should additionally be a definition of IOV_MAX in
<limits.h>.
Shall I go ahead and create a patch, or am I in error?

-- 
-Dionna Glaze, PhD (she/her)

Re: Shouldn't <sys/uio.h> be available everywhere?

[Sebastian Huber]

On 12/07/17 21:42, Dionna Amalie Glaze via newlib wrote:

> I'm looking at the POSIX standard for readv and writev, and it looks
> like there should be definitions for each of these given
> __POSIX_VISIBLE >= 200112, not just being on a PowerPC platform with
> an SPU. There should additionally be a definition of IOV_MAX in
> <limits.h>.
> Shall I go ahead and create a patch, or am I in error?
>

There are some uio.h in Newlib/Cygwin:

./winsup/cygwin/include/sys/uio.h
./newlib/libc/machine/spu/sys/uio.h
./newlib/libc/sys/rtems/include/sys/uio.h
./newlib/libc/sys/phoenix/sys/uio.h

It would be nice to unify this and move it to

./newlib/libc/include/sys/uio.h

For the #ifdef _KERNEL part (and similar) we can use something like this:

#ifdef _KERNEL
/* Header file provided outside of Newlib */
#include <machine/_kernel_uio.h>
#endif

-- 
Sebastian Huber, embedded brains GmbH

Address : Dornierstr. 4, D-82178 Puchheim, Germany
Phone   : +49 89 189 47 41-16
Fax     : +49 89 189 47 41-09
E-Mail  : sebastian.huber@embedded-brains.de
PGP     : Public key available on request.

Diese Nachricht ist keine geschäftliche Mitteilung im Sinne des EHUG.