From: Andrew Jones <drjones@redhat.com>
To: Marc Zyngier <maz@kernel.org>
Cc: kvmarm@lists.cs.columbia.edu, kvm@vger.kernel.org, mark.rutland@arm.com
Subject: Re: [PATCH] arm64: KVM: Only skip MMIO insn once
Date: Thu, 22 Aug 2019 11:25:14 +0200 [thread overview]
Message-ID: <20190822092514.5opwahkjjpqbbayd@kamzik.brq.redhat.com> (raw)
In-Reply-To: <177091d5-2d2c-6a75-472c-92702ee98e86@kernel.org>
On Thu, Aug 22, 2019 at 09:30:44AM +0100, Marc Zyngier wrote:
> Hi Drew,
>
> On 21/08/2019 20:50, Andrew Jones wrote:
> > If after an MMIO exit to userspace a VCPU is immediately run with an
> > immediate_exit request, such as when a signal is delivered or an MMIO
> > emulation completion is needed, then the VCPU completes the MMIO
> > emulation and immediately returns to userspace. As the exit_reason
> > does not get changed from KVM_EXIT_MMIO in these cases we have to
> > be careful not to complete the MMIO emulation again, when the VCPU is
> > eventually run again, because the emulation does an instruction skip
> > (and doing too many skips would be a waste of guest code :-) We need
> > to use additional VCPU state to track if the emulation is complete.
> > As luck would have it, we already have 'mmio_needed', which even
> > appears to be used in this way by other architectures already.
> >
> > Fixes: 0d640732dbeb ("arm64: KVM: Skip MMIO insn after emulation")
> > Signed-off-by: Andrew Jones <drjones@redhat.com>
> > ---
> > virt/kvm/arm/arm.c | 3 ++-
> > virt/kvm/arm/mmio.c | 1 +
> > 2 files changed, 3 insertions(+), 1 deletion(-)
> >
> > diff --git a/virt/kvm/arm/arm.c b/virt/kvm/arm/arm.c
> > index 35a069815baf..322cf9030bbe 100644
> > --- a/virt/kvm/arm/arm.c
> > +++ b/virt/kvm/arm/arm.c
> > @@ -669,7 +669,8 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu, struct kvm_run *run)
> > if (ret)
> > return ret;
> >
> > - if (run->exit_reason == KVM_EXIT_MMIO) {
> > + if (vcpu->mmio_needed) {
> > + vcpu->mmio_needed = 0;
> > ret = kvm_handle_mmio_return(vcpu, vcpu->run);
> > if (ret)
> > return ret;
> > diff --git a/virt/kvm/arm/mmio.c b/virt/kvm/arm/mmio.c
> > index a8a6a0c883f1..2d9b5e064ae0 100644
> > --- a/virt/kvm/arm/mmio.c
> > +++ b/virt/kvm/arm/mmio.c
> > @@ -201,6 +201,7 @@ int io_mem_abort(struct kvm_vcpu *vcpu, struct kvm_run *run,
> > if (is_write)
> > memcpy(run->mmio.data, data_buf, len);
> > vcpu->stat.mmio_exit_user++;
> > + vcpu->mmio_needed = 1;
> > run->exit_reason = KVM_EXIT_MMIO;
> > return 0;
> > }
> >
>
> Thanks for this. That's quite embarrassing. Out of curiosity,
> how was this spotted?
avocado has a guest execution state snapshotting feature. The feature
simply periodically uses QEMU's 'info registers' monitor command while
a guest is running. The monitor command kicks the vcpu to userspace with
a signal, and since avocado's snapshot rate was set relatively high that
increased the probability of causing a noticeable (weird things / guest
crashes) event during guest boot (when MMIO activity is also high). The
signals correlated with guest crashes lead me to this code.
>
> Patch wise, I'd have a small preference for the following (untested)
> patch, as it keeps the mmio_needed accesses close together, making
> it easier to read (at least for me). What do you think?
>
> M.
>
> diff --git a/virt/kvm/arm/mmio.c b/virt/kvm/arm/mmio.c
> index a8a6a0c883f1..6af5c91337f2 100644
> --- a/virt/kvm/arm/mmio.c
> +++ b/virt/kvm/arm/mmio.c
> @@ -86,6 +86,12 @@ int kvm_handle_mmio_return(struct kvm_vcpu *vcpu, struct kvm_run *run)
> unsigned int len;
> int mask;
>
> + /* Detect an already handled MMIO return */
> + if (unlikely(!vcpu->mmio_needed))
> + return 0;
> +
> + vcpu->mmio_needed = 0;
> +
> if (!run->mmio.is_write) {
> len = run->mmio.len;
> if (len > sizeof(unsigned long))
> @@ -188,6 +194,7 @@ int io_mem_abort(struct kvm_vcpu *vcpu, struct kvm_run *run,
> run->mmio.is_write = is_write;
> run->mmio.phys_addr = fault_ipa;
> run->mmio.len = len;
> + vcpu->mmio_needed = 1;
>
> if (!ret) {
> /* We handled the access successfully in the kernel. */
That looks good to me. Should I repost?
Thanks,
drew
next prev parent reply other threads:[~2019-08-22 9:25 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-08-21 19:50 [PATCH] arm64: KVM: Only skip MMIO insn once Andrew Jones
2019-08-22 8:30 ` Marc Zyngier
2019-08-22 9:25 ` Andrew Jones [this message]
2019-08-22 9:38 ` Marc Zyngier
2019-08-22 10:42 ` Andrew Jones
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=20190822092514.5opwahkjjpqbbayd@kamzik.brq.redhat.com \
--to=drjones@redhat.com \
--cc=kvm@vger.kernel.org \
--cc=kvmarm@lists.cs.columbia.edu \
--cc=mark.rutland@arm.com \
--cc=maz@kernel.org \
/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
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
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).