From mboxrd@z Thu Jan 1 00:00:00 1970 From: Arnd Bergmann Subject: Re: [PATCH v2 1/7] ia64: add __NR_old_getpagesize macro Date: Mon, 1 Oct 2018 16:44:16 +0200 Message-ID: References: <1538057720-3392-1-git-send-email-firoz.khan@linaro.org> <1538057720-3392-2-git-send-email-firoz.khan@linaro.org> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Return-path: In-Reply-To: Sender: linux-kernel-owner@vger.kernel.org To: Firoz Khan Cc: linux-ia64@vger.kernel.org, Tony Luck , Fenghua Yu , Thomas Gleixner , gregkh , Philippe Ombredanne , Kate Stewart , y2038 Mailman List , Linux Kernel Mailing List , linux-arch , Deepa Dinamani , Marcin Juszkiewicz List-Id: linux-arch.vger.kernel.org On Mon, Oct 1, 2018 at 4:18 PM Firoz Khan wrote: > > Hi Arnd, > > On Fri, 28 Sep 2018 at 17:41, Arnd Bergmann wrote: > > > > On Thu, Sep 27, 2018 at 4:16 PM Firoz Khan wrote: > > > > > > Add __NR_old_getpagesize in order to not break old user space > > > as it is reserved for backwards compatibility with old __NR_ > > > getpagesize. > > > > That description doesn't really make sense here. > > Sure I'll some more details. But rest of the patch looks good? > This is some I was bit confused! In header there is no entry, but > in system call table has the entry for the system call. I think adding some macro for it makes sense, and the __NR_old_getpagesize name you picked is fine for that. I think generally speaking we want the unistd.h macros to match up with the system call entry points, and there are currently multiple ways they diverge: - System calls have been removed from the kernels, and the macros in unistd.h are still there on some architectures but not others. This is something we should clean up /after/ you are done. - System calls have a number but are not implemented on all architectures because they don't make sense (e.g. pk pkey_add). These should probably have an entry in kernel/sys_ni.c if they don't already have one. - System calls have an entry but no macro, like the ia64 old_getpagesize. I think if you encounter one of these, we need to fix them up before your series so we can complete the conversion to generated tables. Arnd From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-qt1-f196.google.com ([209.85.160.196]:42395 "EHLO mail-qt1-f196.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1729457AbeJAVWo (ORCPT ); Mon, 1 Oct 2018 17:22:44 -0400 MIME-Version: 1.0 References: <1538057720-3392-1-git-send-email-firoz.khan@linaro.org> <1538057720-3392-2-git-send-email-firoz.khan@linaro.org> In-Reply-To: From: Arnd Bergmann Date: Mon, 1 Oct 2018 16:44:16 +0200 Message-ID: Subject: Re: [PATCH v2 1/7] ia64: add __NR_old_getpagesize macro Content-Type: text/plain; charset="UTF-8" Sender: linux-arch-owner@vger.kernel.org List-ID: To: Firoz Khan Cc: linux-ia64@vger.kernel.org, Tony Luck , Fenghua Yu , Thomas Gleixner , gregkh , Philippe Ombredanne , Kate Stewart , y2038 Mailman List , Linux Kernel Mailing List , linux-arch , Deepa Dinamani , Marcin Juszkiewicz Message-ID: <20181001144416.yuCDQO2153pH1KBW9x5WAIZyICrzWl2eLc80F2loPjk@z> On Mon, Oct 1, 2018 at 4:18 PM Firoz Khan wrote: > > Hi Arnd, > > On Fri, 28 Sep 2018 at 17:41, Arnd Bergmann wrote: > > > > On Thu, Sep 27, 2018 at 4:16 PM Firoz Khan wrote: > > > > > > Add __NR_old_getpagesize in order to not break old user space > > > as it is reserved for backwards compatibility with old __NR_ > > > getpagesize. > > > > That description doesn't really make sense here. > > Sure I'll some more details. But rest of the patch looks good? > This is some I was bit confused! In header there is no entry, but > in system call table has the entry for the system call. I think adding some macro for it makes sense, and the __NR_old_getpagesize name you picked is fine for that. I think generally speaking we want the unistd.h macros to match up with the system call entry points, and there are currently multiple ways they diverge: - System calls have been removed from the kernels, and the macros in unistd.h are still there on some architectures but not others. This is something we should clean up /after/ you are done. - System calls have a number but are not implemented on all architectures because they don't make sense (e.g. pk pkey_add). These should probably have an entry in kernel/sys_ni.c if they don't already have one. - System calls have an entry but no macro, like the ia64 old_getpagesize. I think if you encounter one of these, we need to fix them up before your series so we can complete the conversion to generated tables. Arnd