kvmstealtime.patch

   1 Return-Path: kvm-owner@vger.kernel.org
   2 Received: from mailpro.odiso.net (LHLO mailpro.odiso.net) (10.1.31.112) by
   3  mailpro.odiso.net with LMTP; Thu, 15 Oct 2015 00:33:53 +0200 (CEST)
   4 Received: from localhost (localhost [127.0.0.1])
   5         by mailpro.odiso.net (Postfix) with ESMTP id 1F3454439326A
   6         for <aderumier@oxygem.tv>; Thu, 15 Oct 2015 00:33:53 +0200 (CEST)
   7 X-Virus-Scanned: amavisd-new at mailpro.odiso.com
   8 X-Spam-Flag: NO
   9 X-Spam-Score: 0.971
  10 X-Spam-Level:
  11 X-Spam-Status: No, score=0.971 tagged_above=-10 required=4
  12         tests=[HEADER_FROM_DIFFERENT_DOMAINS=0.001,
  13         KAM_LAZY_DOMAIN_SECURITY=1, RCVD_IN_MSPIKE_H3=-0.01,
  14         RCVD_IN_MSPIKE_WL=-0.01, T_RP_MATCHES_RCVD=-0.01]
  15         autolearn=no autolearn_force=no
  16 Received: from mailpro.odiso.net ([127.0.0.1])
  17         by localhost (mailpro.odiso.net [127.0.0.1]) (amavisd-new, port 10024)
  18         with ESMTP id hMmmKVYj2S2m for <aderumier@oxygem.tv>;
  19         Thu, 15 Oct 2015 00:33:47 +0200 (CEST)
  20 Received: from vger.kernel.org (vger.kernel.org [209.132.180.67])
  21         by mailpro.odiso.net (Postfix) with ESMTP id 09D3D4439324A
  22         for <aderumier@odiso.com>; Thu, 15 Oct 2015 00:33:46 +0200 (CEST)
  23 Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand
  24         id S1754065AbbJNWdh (ORCPT <rfc822;aderumier@odiso.com>);
  25         Wed, 14 Oct 2015 18:33:37 -0400
  26 Received: from mx1.redhat.com ([209.132.183.28]:48432 "EHLO mx1.redhat.com"
  27         rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP
  28         id S1753882AbbJNWdg (ORCPT <rfc822;kvm@vger.kernel.org>);
  29         Wed, 14 Oct 2015 18:33:36 -0400
  30 Received: from int-mx11.intmail.prod.int.phx2.redhat.com (int-mx11.intmail.prod.int.phx2.redhat.com [10.5.11.24])
  31         by mx1.redhat.com (Postfix) with ESMTPS id 35096C0BFD02
  32         for <kvm@vger.kernel.org>; Wed, 14 Oct 2015 22:33:36 +0000 (UTC)
  33 Received: from amt.cnet (vpn1-4-185.ams2.redhat.com [10.36.4.185])
  34         by int-mx11.intmail.prod.int.phx2.redhat.com (8.14.4/8.14.4) with ESMTP id t9EMXYFs002263;
  35         Wed, 14 Oct 2015 18:33:35 -0400
  36 Received: from amt.cnet (localhost [127.0.0.1])
  37         by amt.cnet (Postfix) with ESMTP id ADFE61009C7;
  38         Wed, 14 Oct 2015 19:33:15 -0300 (BRT)
  39 Received: (from marcelo@localhost)
  40         by amt.cnet (8.14.7/8.14.7/Submit) id t9EMXCRq014112;
  41         Wed, 14 Oct 2015 19:33:12 -0300
  42 Date:   Wed, 14 Oct 2015 19:33:09 -0300
  43 From:   Marcelo Tosatti <mtosatti@redhat.com>
  44 To:     kvm@vger.kernel.org
  45 Cc:     Rik van Riel <riel@redhat.com>,
  46         "Dr. David Alan Gilbert" <dgilbert@redhat.com>,
  47         Paolo Bonzini <pbonzini@redhat.com>,
  48         Amit Shah <amit.shah@redhat.com>
  49 Subject: [PATCH] KVM: x86: move steal time initialization to vcpu entry time
  50 Message-ID: <20151014223309.GA13900@amt.cnet>
  51 MIME-Version: 1.0
  52 Content-Type: text/plain; charset=us-ascii
  53 Content-Disposition: inline
  54 User-Agent: Mutt/1.5.21 (2010-09-15)
  55 X-Scanned-By: MIMEDefang 2.68 on 10.5.11.24
  56 Sender: kvm-owner@vger.kernel.org
  57 Precedence: bulk
  58 List-ID: <kvm.vger.kernel.org>
  59 X-Mailing-List: kvm@vger.kernel.org
  60
  61
  62 As reported at https://bugs.launchpad.net/qemu/+bug/1494350,
  63 it is possible to have vcpu->arch.st.last_steal initialized
  64 from a thread other than vcpu thread, say the iothread, via
  65 KVM_SET_MSRS.
  66
  67 Which can cause an overflow later (when subtracting from vcpu threads
  68 sched_info.run_delay).
  69
  70 To avoid that, move steal time accumulation to vcpu entry time,
  71 before copying steal time data to guest.
  72
  73 Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
  74
  75 diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
  76 index 8f0f6ec..0e0332e 100644
  77 --- a/arch/x86/kvm/x86.c
  78 +++ b/arch/x86/kvm/x86.c
  79 @@ -2030,6 +2030,8 @@ static void accumulate_steal_time(struct kvm_vcpu *vcpu)
  80
  81  static void record_steal_time(struct kvm_vcpu *vcpu)
  82  {
  83 +       accumulate_steal_time(vcpu);
  84 +
  85         if (!(vcpu->arch.st.msr_val & KVM_MSR_ENABLED))
  86                 return;
  87
  88 @@ -2182,12 +2184,6 @@ int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
  89                 if (!(data & KVM_MSR_ENABLED))
  90                         break;
  91
  92 -               vcpu->arch.st.last_steal = current->sched_info.run_delay;
  93 -
  94 -               preempt_disable();
  95 -               accumulate_steal_time(vcpu);
  96 -               preempt_enable();
  97 -
  98                 kvm_make_request(KVM_REQ_STEAL_UPDATE, vcpu);
  99
 100                 break;
 101 @@ -2830,7 +2826,6 @@ void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
 102                 vcpu->cpu = cpu;
 103         }
 104
 105 -       accumulate_steal_time(vcpu);
 106         kvm_make_request(KVM_REQ_STEAL_UPDATE, vcpu);
 107  }
 108
 109 --
 110 To unsubscribe from this list: send the line "unsubscribe kvm" in
 111 the body of a message to majordomo@vger.kernel.org
 112 More majordomo info at  http://vger.kernel.org/majordomo-info.html