From mboxrd@z Thu Jan 1 00:00:00 1970 From: George Dunlap Subject: Re: [PATCH 00/03 V2] arm: host SMP support Date: Wed, 17 Apr 2013 14:45:46 +0100 Message-ID: <516EA78A.9090800@eu.citrix.com> References: <1366203132.25579.18.camel@zakaz.uk.xensource.com> <516EA196.3090501@eu.citrix.com> <1366205327.25579.41.camel@zakaz.uk.xensource.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; Format="flowed" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <1366205327.25579.41.camel@zakaz.uk.xensource.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xen.org Errors-To: xen-devel-bounces@lists.xen.org To: Ian Campbell Cc: Julien Grall , "Tim (Xen.org)" , xen-devel , "Keir (Xen.org)" , Stefano Stabellini List-Id: xen-devel@lists.xenproject.org On 17/04/13 14:28, Ian Campbell wrote: > On Wed, 2013-04-17 at 14:20 +0100, George Dunlap wrote: >> On 17/04/13 13:52, Ian Campbell wrote: >>> Most of v1 one went in already. This rebases Rebases the remainder of v1 >>> and addressed the comments by dropping the use of the GROUP1 bit. >>> >>> I think SMP is a substantial feature of Xen on ARM for 4.3 and it is >>> therefore worthy of a freeze exception. It touches only ARM specific >>> code. >> Yeah, I think ARM without SMP is almost a bug. :-) So for this series: >> >> Acked-by: George Dunlap >> >>> I would like to apply Julien's "implement smp_call_function" series[0] >>> on top of it which fixes reboot/shutdown but which makes the x86 >>> smp_call_function generic and so necessarily touches x86 and generic >>> code. If that is considered too risky (I personally think it should be >>> OK, it's a pretty obvious move) then we could duplicate that code for >>> ARM. >>> >>> [0] v2 at <1366119508-9227-3-git-send-email-julien.grall@citrix.com> >> Let me take a look. The most important question is, if there's a bug, >> will we find it before the release? Since smp_call_function() is called >> by basically everyone all the time, I think we can be pretty confident >> that we'll find any bugs in x86 code. Would you agree? > It's not called much on ARM but I believe it is on x86 so Yes. > > It is also a pretty simple move with the only change being to refactor > the one arch specific line in to a per arch function. If Keir or Jan think that if there is a regression we are highly likely to catch it before the release, I'm OK with Julien's series. -George