kvm.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Marcelo Tosatti <mtosatti@redhat.com>
To: kvm@vger.kernel.org
Cc: Paolo Bonzini <pbonzini@redhat.com>,
	Alex Williamson <alex.williamson@redhat.com>,
	Sean Christopherson <seanjc@google.com>,
	Pei Zhang <pezhang@redhat.com>
Subject: Re: [patch 2/2] KVM: VMX: update vcpu posted-interrupt descriptor when assigning device
Date: Thu, 6 May 2021 16:11:08 -0300	[thread overview]
Message-ID: <20210506191108.GA349654@fuller.cnet> (raw)
In-Reply-To: <20210506190419.481236922@redhat.com>

On Thu, May 06, 2021 at 03:57:34PM -0300, Marcelo Tosatti wrote:
> For VMX, when a vcpu enters HLT emulation, pi_post_block will:
> 
> 1) Add vcpu to per-cpu list of blocked vcpus.
> 
> 2) Program the posted-interrupt descriptor "notification vector" 
> to POSTED_INTR_WAKEUP_VECTOR
> 
> With interrupt remapping, an interrupt will set the PIR bit for the 
> vector programmed for the device on the CPU, test-and-set the 
> ON bit on the posted interrupt descriptor, and if the ON bit is clear
> generate an interrupt for the notification vector.
> 
> This way, the target CPU wakes upon a device interrupt and wakes up
> the target vcpu.
> 
> Problem is that pi_post_block only programs the notification vector
> if kvm_arch_has_assigned_device() is true. Its possible for the
> following to happen:
> 
> 1) vcpu V HLTs on pcpu P, kvm_arch_has_assigned_device is false,
> notification vector is not programmed
> 2) device is assigned to VM
> 3) device interrupts vcpu V, sets ON bit (notification vector not programmed,
> so pcpu P remains in idle)
> 4) vcpu 0 IPIs vcpu V (in guest), but since pi descriptor ON bit is set,
> kvm_vcpu_kick is skipped
> 5) vcpu 0 busy spins on vcpu V's response for several seconds, until
> RCU watchdog NMIs all vCPUs.
> 
> To fix this, use the start_assignment kvm_x86_ops callback to program the
> notification vector when assigned device count changes from 0 to 1.
> 
> Reported-by: Pei Zhang <pezhang@redhat.com>
> Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

Argh, missing setting vmx_pi_start_assignment, will resend v2.


  reply	other threads:[~2021-05-06 19:11 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-05-06 18:57 [patch 0/2] VMX: configure posted interrupt " Marcelo Tosatti
2021-05-06 18:57 ` [patch 1/2] KVM: x86: add start_assignment hook to kvm_x86_ops Marcelo Tosatti
2021-05-06 19:54   ` Sean Christopherson
2021-05-06 18:57 ` [patch 2/2] KVM: VMX: update vcpu posted-interrupt descriptor when assigning device Marcelo Tosatti
2021-05-06 19:11   ` Marcelo Tosatti [this message]
2021-05-06 19:21   ` [patch 2/2 V2] " Marcelo Tosatti
2021-05-06 21:35     ` Sean Christopherson
2021-05-07 12:11       ` Marcelo Tosatti

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20210506191108.GA349654@fuller.cnet \
    --to=mtosatti@redhat.com \
    --cc=alex.williamson@redhat.com \
    --cc=kvm@vger.kernel.org \
    --cc=pbonzini@redhat.com \
    --cc=pezhang@redhat.com \
    --cc=seanjc@google.com \
    --subject='Re: [patch 2/2] KVM: VMX: update vcpu posted-interrupt descriptor when assigning device' \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).