svn commit: trunk/uClibc/include/sys

Erik Andersen andersen at codepoet.org
Fri Dec 8 20:40:56 UTC 2006


On Fri Dec 08, 2006 at 02:57:08PM -0500, Mike Frysinger wrote:
> On Friday 08 December 2006 14:28, andersen at uclibc.org wrote:
> > kernel provided syscalls
> > are NOT suitable for user-space.  The libc ones are the ones that
> > application code must use.
>
> the push i think is to stop using _syscall#() completely and to have user
> applications call syscall() instead

Fully agreed -- nobody but libc should use the _syscall#() macros
-- those are purely for uClibc internal use only.  And uClibc
should of course only use our own macros, not the kernel ones.

The problem that led me to make this change was that when trying
to build LTP the syscall01.c test failed on syscall(SYS_*) as the
traditional 'SYS_<name>' syscall names were missing.  The cause
may have been some other entirely different transient thing.

I suppose I could remove bits/sysnum.h entirely from
libc/sysdeps/linux/*/bits/syscalls.h and then change
sys/syscall.h to something like:

	#include <features.h>
	#include <bits/sysnum.h>
	#if defined _LIBC && (defined IS_IN_libc || defined NOT_IN_libc)
	# include <bits/syscalls.h>
	#endif

Agreeable?

 -Erik

--
Erik B. Andersen             http://codepoet-consulting.com/
--This message was written using 73% post-consumer electrons--
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://lists.busybox.net/pipermail/uclibc/attachments/20061208/795809e2/attachment-0002.pgp 


More information about the uClibc mailing list