From mboxrd@z Thu Jan 1 00:00:00 1970 From: Arnd Bergmann Subject: Re: uapi __NR_syscalls for microblaze Date: Thu, 3 Jan 2013 22:27:05 +0000 Message-ID: <201301032227.05553.arnd@arndb.de> References: <201301031601.00594.arnd@arndb.de> Mime-Version: 1.0 Content-Type: Text/Plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: Sender: linux-kernel-owner@vger.kernel.org To: Michal Simek Cc: David Howells , LKML , Benjamin Herrenschmidt , linux-arch@vger.kernel.org List-Id: linux-arch.vger.kernel.org On Thursday 03 January 2013, Michal Simek wrote: > > 2013/1/3 Arnd Bergmann : > > On Thursday 03 January 2013, Michal Simek wrote: > >> 2013/1/3 David Howells : > >> > Michal Simek wrote: > >> > > >> >> just want to check with you if __NR_syscalls is necessary for user space. I > >> >> see that powerpc and arm have this macro in asm not in uapi like Microblaze. > >> >> If is not needed by user space, I should move it to asm/unistd.h > >> > > >> > It isn't as far as I know... I recommend putting your question on the > >> > linux-arch mailing list. > >> > >> I have add linux-arch to CC. > > > > Right, I'm pretty sure it's not needed, but it has traditionally been > > exported on a lot of platforms, so you can keep it in UAPI to be > > on the safe side. > > Isn't it better to be consistent across all architectures? Yes, certainly. > If it is wrong to exported and there is no reason to use it in userspace > then all architecture should move it out of uapi. > If there is any reason then it is should be in uapi. I don't really care which way we do it, I think you can rightfully argue either way. AFAICT x86 doesn't even define __NR_syscalls at all, so it's clearly not part of the ABI. Arnd