From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-qk1-f193.google.com (mail-qk1-f193.google.com [209.85.222.193]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by lists.ozlabs.org (Postfix) with ESMTPS id 42K9RM2Rt1zF31R for ; Tue, 25 Sep 2018 16:00:46 +1000 (AEST) Received: by mail-qk1-f193.google.com with SMTP id b19-v6so12813619qkc.6 for ; Mon, 24 Sep 2018 23:00:46 -0700 (PDT) MIME-Version: 1.0 References: <1536913980-4811-1-git-send-email-firoz.khan@linaro.org> <1536913980-4811-3-git-send-email-firoz.khan@linaro.org> <87efdi5rm6.fsf@concordia.ellerman.id.au> In-Reply-To: <87efdi5rm6.fsf@concordia.ellerman.id.au> From: Arnd Bergmann Date: Tue, 25 Sep 2018 08:00:28 +0200 Message-ID: Subject: Re: [PATCH 2/3] powerpc: Add system call table generation support To: Michael Ellerman Cc: Firoz Khan , linuxppc-dev , Benjamin Herrenschmidt , Paul Mackerras , linuxram@us.ibm.com, leitao@debian.org, Boqun Feng , gregkh , Philippe Ombredanne , Thomas Gleixner , Kate Stewart , y2038 Mailman List , Linux Kernel Mailing List , linux-arch , Deepa Dinamani , Marcin Juszkiewicz Content-Type: text/plain; charset="UTF-8" List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Tue, Sep 25, 2018 at 2:48 AM Michael Ellerman wrote: > Arnd Bergmann writes: > > On Tue, Sep 18, 2018 at 2:15 PM Firoz Khan wrote: > >> On 14 September 2018 at 15:31, Arnd Bergmann wrote: > >> > On Fri, Sep 14, 2018 at 10:33 AM Firoz Khan wrote: > > > > But all three existing architectures (x86, s390 and arm) already > > have the capability to parse the table and generate different output > > from that. > > Yeah, we want that on powerpc too. > > If the script needs to be more complex that's fine, if it can't be > shared across arches that's fine, the main thing for me is that wiring > up a syscall can be done by adding a single line in a single file. Yes, that's definitely the idea, we want to make it easier for everyone. We need at least a special case for mips, which needs three separate input files (the tables are completely different) to generate four versions of the output, plus a future extension to use the generic table for their new one. For powerpc, I'm hoping that both the table format and script can be completely generic and not need a special case that is different from the others, but if we need some extra magic to handle the SPU syscalls, we can still do that with a private script. Arnd