From mboxrd@z Thu Jan 1 00:00:00 1970 From: Paul Mackerras Subject: Re: [PATCH 3/11] powerpc: convert to generic helpers for IPI function calls Date: Tue, 22 Apr 2008 22:03:24 +1000 Message-ID: <18445.54284.194023.553595@cargo.ozlabs.ibm.com> References: <1208851058-8500-1-git-send-email-jens.axboe@oracle.com> <1208851058-8500-4-git-send-email-jens.axboe@oracle.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <1208851058-8500-4-git-send-email-jens.axboe-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org> Sender: linux-arch-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org List-ID: To: Jens Axboe Cc: linux-arch-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, npiggin-l3A5Bk7waGM@public.gmane.org, torvalds-de/tnXTf+JLsfHDXvbKv3WD2FQJk+8+b@public.gmane.org, anton-eUNUBHrolfbYtjvyW6yDsg@public.gmane.org Jens Axboe writes: > This converts ppc to use the new helpers for smp_call_function() and > friends, and adds support for smp_call_function_single(). Looks OK, and runs on a dual G5 powermac here. I'll try on some bigger IBM boxes tomorrow, or maybe Anton can do that. I notice that you have changed the semantics slightly in that previously, the powerpc implementation would wait for the other cpus to have taken the interrupt, and print a message if they didn't all take the interrupt (this was different from the function of the `wait' parameter, which said whether to wait for all cpus to finish executing the specified function). Now the generic smp_call_function_mask doesn't wait for the other cpus to take the interrupt. I don't think that's likely to be a problem, but I do think that's worth mentioning in the patch description. It's probably also worth mentioning that smp_send_stop no longer has the behaviour of doing its stuff even if someone is holding the lock, but that is OK (presumably) because now the lock (call_function_lock in the new code) is held for much shorter periods and there doesn't appear to be much danger of other cpus getting stuck or panicking while holding that lock. You can put Acked-by: Paul Mackerras on the ppc patch, but please extend the description a bit as outlined above. Thanks for doing this, BTW. Regards, Paul. From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from ozlabs.org ([203.10.76.45]:50333 "EHLO ozlabs.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752285AbYDVMFT (ORCPT ); Tue, 22 Apr 2008 08:05:19 -0400 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Message-ID: <18445.54284.194023.553595@cargo.ozlabs.ibm.com> Date: Tue, 22 Apr 2008 22:03:24 +1000 From: Paul Mackerras Subject: Re: [PATCH 3/11] powerpc: convert to generic helpers for IPI function calls In-Reply-To: <1208851058-8500-4-git-send-email-jens.axboe@oracle.com> References: <1208851058-8500-1-git-send-email-jens.axboe@oracle.com> <1208851058-8500-4-git-send-email-jens.axboe@oracle.com> Sender: linux-arch-owner@vger.kernel.org List-ID: To: Jens Axboe Cc: linux-arch@vger.kernel.org, linux-kernel@vger.kernel.org, npiggin@suse.de, torvalds@linux-foundation.org, anton@samba.org Message-ID: <20080422120324.R1eU8D8_PfO_XNaSMaJGo6W_HaZEL46JSBI9SAytdxU@z> Jens Axboe writes: > This converts ppc to use the new helpers for smp_call_function() and > friends, and adds support for smp_call_function_single(). Looks OK, and runs on a dual G5 powermac here. I'll try on some bigger IBM boxes tomorrow, or maybe Anton can do that. I notice that you have changed the semantics slightly in that previously, the powerpc implementation would wait for the other cpus to have taken the interrupt, and print a message if they didn't all take the interrupt (this was different from the function of the `wait' parameter, which said whether to wait for all cpus to finish executing the specified function). Now the generic smp_call_function_mask doesn't wait for the other cpus to take the interrupt. I don't think that's likely to be a problem, but I do think that's worth mentioning in the patch description. It's probably also worth mentioning that smp_send_stop no longer has the behaviour of doing its stuff even if someone is holding the lock, but that is OK (presumably) because now the lock (call_function_lock in the new code) is held for much shorter periods and there doesn't appear to be much danger of other cpus getting stuck or panicking while holding that lock. You can put Acked-by: Paul Mackerras on the ppc patch, but please extend the description a bit as outlined above. Thanks for doing this, BTW. Regards, Paul.