Hi Bernhard, On Thu, Nov 07, 2013 at 12:32:49PM +0100, Bernhard Reutner-Fischer wrote: > On 7 November 2013 10:52, Baruch Siach <bar...@tkos.co.il> wrote: > > Not all architectures use 'ret' as function return instruction. For example, > > xtensa usually uses 'retw'. Use the ret_ERRVAL arch dependant macro instead. > > > > Signed-off-by: Baruch Siach <bar...@tkos.co.il> > > --- > > v3: Add a common ret_ERRVAL, and redefine for special archs (Bernhard > > Reutner-Fischer) > > > > v2: Add ret_ERRVAL to architectures missing it as noted by Bernhard > > Reutner-Fischer. > > --- > > libc/sysdeps/linux/arm/sysdep.h | 1 + > > libc/sysdeps/linux/common/sysdep.h | 4 ++++ > > libc/sysdeps/linux/i386/sysdep.h | 2 -- > > libc/sysdeps/linux/ia64/sysdep.h | 1 - > > libc/sysdeps/linux/mips/sysdep.h | 2 -- > > libc/sysdeps/linux/powerpc/powerpc32/sysdep.h | 1 + > > libc/sysdeps/linux/powerpc/powerpc64/sysdep.h | 1 + > > libc/sysdeps/linux/sh/sysdep.h | 2 -- > > libc/sysdeps/linux/x86_64/sysdep.h | 2 -- > > librt/mq_timedreceive.S | 2 +- > > librt/mq_timedsend.S | 2 +- > > 11 files changed, 9 insertions(+), 11 deletions(-) > > And where does e.g. xtensa pick that up?
xtensa defines ret_ERRVAL as 'retw' already, but the 'undef' is missing. I'll add it in my next version of this patch. > Can you commonize these, i.e. make sure that common/sysdep.h is > included first, #defining unconditional the defaults, arches > redefining stuff where they diverge. > Then, as a second, follow-up fix librt to make use of the new ret_ERRVAL. > Bonus points if you don't break any arch :) OK. I can't promise anything about not breaking other archs, though, but I'll try. baruch -- http://baruch.siach.name/blog/ ~. .~ Tk Open Systems =}------------------------------------------------ooO--U--Ooo------------{= - bar...@tkos.co.il - tel: +972.2.679.5364, http://www.tkos.co.il - _______________________________________________ uClibc mailing list uClibc@uclibc.org http://lists.busybox.net/mailman/listinfo/uclibc