From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754566Ab0HXJFa (ORCPT ); Tue, 24 Aug 2010 05:05:30 -0400 Received: from mx1.redhat.com ([209.132.183.28]:39107 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754433Ab0HXJF2 (ORCPT ); Tue, 24 Aug 2010 05:05:28 -0400 Message-ID: <4C738B23.6040205@redhat.com> Date: Tue, 24 Aug 2010 12:04:35 +0300 From: Avi Kivity User-Agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.2.7) Gecko/20100720 Fedora/3.1.1-1.fc13 Lightning/1.0b2pre Thunderbird/3.1.1 MIME-Version: 1.0 To: Gleb Natapov CC: kvm@vger.kernel.org, linux-mm@kvack.org, linux-kernel@vger.kernel.org, mingo@elte.hu, a.p.zijlstra@chello.nl, tglx@linutronix.de, hpa@zytor.com, riel@redhat.com, cl@linux-foundation.org, mtosatti@redhat.com Subject: Re: [PATCH v5 08/12] Inject asynchronous page fault into a guest if page is swapped out. References: <1279553462-7036-1-git-send-email-gleb@redhat.com> <1279553462-7036-9-git-send-email-gleb@redhat.com> <4C729F10.40005@redhat.com> <20100824075258.GX10499@redhat.com> In-Reply-To: <20100824075258.GX10499@redhat.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 08/24/2010 10:52 AM, Gleb Natapov wrote: >> This nice cache needs to be outside apf to reduce complexity for >> reviewers and since it is useful for others. >> >> Would be good to have memslot-cached kvm_put_guest() and kvm_get_guest(). > Will look into it. In the meantime, you can just drop the caching. >>> + struct kvm_arch_async_pf *arch) >>> +{ >>> + struct kvm_async_pf *work; >>> + >>> + if (vcpu->async_pf_queued>= ASYNC_PF_PER_VCPU) >>> + return 0; >> 100 == too high. At 16 vcpus, this allows 1600 kernel threads to >> wait for I/O. > Number of kernel threads are limited by other means. Slow work subsystem > has its own knobs to tune that. Here we limit how much slow work items > can be queued per vcpu. OK. >> Would have been best if we could ask for a page to be paged in >> asynchronously. >> > You mean to have core kernel facility for that? I agree it would be > nice, but much harder. Yes, that's what I meant. -- error compiling committee.c: too many arguments to function