From mboxrd@z Thu Jan 1 00:00:00 1970 From: Keir Fraser Subject: Re: [PATCH, v2] reduce 'd' debug key's global impact Date: Thu, 6 May 2010 10:09:21 +0100 Message-ID: References: <4BE294CF0200007800001825@vpn.id2.novell.com> Mime-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <4BE294CF0200007800001825@vpn.id2.novell.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: Jan Beulich Cc: "xen-devel@lists.xensource.com" List-Id: xen-devel@lists.xenproject.org On 06/05/2010 09:07, "Jan Beulich" wrote: >>>> Keir Fraser 06.05.10 09:42 >>> >> Pick something rare like SPURIOUS_APIC_VECTOR. In that particular handler >> you can even put your check on the unlikely path that checks for and acks a >> real vectored interrupt. > > Hmm, yes, that's indeed better than the event check one. > > However, I noticed that even without that change on_selected_cpus() > is being used in interrupt context, hence I wonder whether indeed only > the new mechanism should be going the call-func-less route. If you add this new mechanism to call into __dump_execstate() then there is no reason not to use it for both the 'old' and 'alt' dumping methods. Indeed doing it for both will avoid needing your hack to the call_function handler to pass regs when the info parameter is NULL (which I could understand but was still icky, and I would probably have changed the interface to the on_selected/each_cpu functions instead). -- Keir