From: Jan Kiszka <jan.kiszka@web.de>
To: Claudio Fontana <claudio.fontana@huawei.com>,
Peter Maydell <peter.maydell@linaro.org>
Cc: Marc Zyngier <marc.zyngier@arm.com>,
kvm-devel <kvm@vger.kernel.org>,
"kvmarm@lists.cs.columbia.edu" <kvmarm@lists.cs.columbia.edu>,
arm-mail-list <linux-arm-kernel@lists.infradead.org>
Subject: Re: [RFC PATCH] KVM: arm/arm64: Don't let userspace update CNTVOFF once guest is running
Date: Fri, 26 Jun 2015 06:49:34 +0200 [thread overview]
Message-ID: <558CD9DE.3010609@web.de> (raw)
In-Reply-To: <558BC90B.4060806@huawei.com>
[-- Attachment #1.1: Type: text/plain, Size: 1897 bytes --]
On 2015-06-25 11:25, Claudio Fontana wrote:
> On 25.06.2015 11:10, Peter Maydell wrote:
>> On 25 June 2015 at 09:59, Claudio Fontana <claudio.fontana@huawei.com> wrote:
>>> Once the VM is created, I think QEMU should not request kvm to
>>> change the virtual offset of the VM anymore: maybe an unexpected
>>> consequence of QEMU's target-arm/kvm64.c::kvm_arch_put_registers ?
>>
>> Hmm. In general we assume that we can:
>> * stop the VM
>> * read all the guest system registers
>> * write those values back again
>> * restart the VM
>>
>> if we need to. Is that what's happening here, or are we doing
>> something odder?
>>
>> -- PMM
>>
>
> What I guess could be happening by looking at the code in linux
>
> virt/kvm/arm/arch_timer.c::kvm_arm_timer_set_reg
>
> is that QEMU tries to set the KVM_REG_ARM_TIMER_CNT register from exactly the previous value,
> but just because of the fact that the set function is called, cntvoff is updated,
> since the value provided by the user is apparently assumed to be _relative_ to the physical timer.
>
> This is apparent to me in the code in that function which says:
>
> case KVM_REG_ARM_TIMER_CNT: {
> /* ... */
> u64 cntvoff = kvm_phys_timer_read() - value;
> /* ... */
> }
>
> And this is matched by the corresponding get function kvm_arm_timer_get_reg where it says:
>
> case KVM_REG_ARM_TIMER_CNT:
> return kvm_phys_timer_read() - vcpu->kvm->arch.timer.cntvoff;
>
> The time difference between when the GET is issued by QEMU and when the PUT is issued then would account for the difference.
QEMU has the concept of write-back levels: KVM_PUT_RUNTIME_STATE,
KVM_PUT_RESET_STATE and KVM_PUT_FULL_STATE. I suspect this registers is
just sorted into the wrong category, thus written as part of the
RUNTIME_STATE. We had such bug patterns during the x86 maturing phase as
well.
Jan
[-- Attachment #1.2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]
[-- Attachment #2: Type: text/plain, Size: 151 bytes --]
_______________________________________________
kvmarm mailing list
kvmarm@lists.cs.columbia.edu
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
WARNING: multiple messages have this Message-ID (diff)
From: jan.kiszka@web.de (Jan Kiszka)
To: linux-arm-kernel@lists.infradead.org
Subject: [RFC PATCH] KVM: arm/arm64: Don't let userspace update CNTVOFF once guest is running
Date: Fri, 26 Jun 2015 06:49:34 +0200 [thread overview]
Message-ID: <558CD9DE.3010609@web.de> (raw)
In-Reply-To: <558BC90B.4060806@huawei.com>
On 2015-06-25 11:25, Claudio Fontana wrote:
> On 25.06.2015 11:10, Peter Maydell wrote:
>> On 25 June 2015 at 09:59, Claudio Fontana <claudio.fontana@huawei.com> wrote:
>>> Once the VM is created, I think QEMU should not request kvm to
>>> change the virtual offset of the VM anymore: maybe an unexpected
>>> consequence of QEMU's target-arm/kvm64.c::kvm_arch_put_registers ?
>>
>> Hmm. In general we assume that we can:
>> * stop the VM
>> * read all the guest system registers
>> * write those values back again
>> * restart the VM
>>
>> if we need to. Is that what's happening here, or are we doing
>> something odder?
>>
>> -- PMM
>>
>
> What I guess could be happening by looking at the code in linux
>
> virt/kvm/arm/arch_timer.c::kvm_arm_timer_set_reg
>
> is that QEMU tries to set the KVM_REG_ARM_TIMER_CNT register from exactly the previous value,
> but just because of the fact that the set function is called, cntvoff is updated,
> since the value provided by the user is apparently assumed to be _relative_ to the physical timer.
>
> This is apparent to me in the code in that function which says:
>
> case KVM_REG_ARM_TIMER_CNT: {
> /* ... */
> u64 cntvoff = kvm_phys_timer_read() - value;
> /* ... */
> }
>
> And this is matched by the corresponding get function kvm_arm_timer_get_reg where it says:
>
> case KVM_REG_ARM_TIMER_CNT:
> return kvm_phys_timer_read() - vcpu->kvm->arch.timer.cntvoff;
>
> The time difference between when the GET is issued by QEMU and when the PUT is issued then would account for the difference.
QEMU has the concept of write-back levels: KVM_PUT_RUNTIME_STATE,
KVM_PUT_RESET_STATE and KVM_PUT_FULL_STATE. I suspect this registers is
just sorted into the wrong category, thus written as part of the
RUNTIME_STATE. We had such bug patterns during the x86 maturing phase as
well.
Jan
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 181 bytes
Desc: OpenPGP digital signature
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20150626/50b303fc/attachment.sig>
next prev parent reply other threads:[~2015-06-26 4:38 UTC|newest]
Thread overview: 46+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-06-24 14:54 [RFC PATCH] KVM: arm/arm64: Don't let userspace update CNTVOFF once guest is running Marc Zyngier
2015-06-24 14:54 ` Marc Zyngier
2015-06-25 8:04 ` Christoffer Dall
2015-06-25 8:04 ` Christoffer Dall
2015-06-25 8:48 ` Marc Zyngier
2015-06-25 8:48 ` Marc Zyngier
2015-06-25 8:59 ` Claudio Fontana
2015-06-25 8:59 ` Claudio Fontana
2015-06-25 9:10 ` Peter Maydell
2015-06-25 9:10 ` Peter Maydell
2015-06-25 9:25 ` Claudio Fontana
2015-06-25 9:25 ` Claudio Fontana
2015-06-26 4:49 ` Jan Kiszka [this message]
2015-06-26 4:49 ` Jan Kiszka
2015-06-29 17:20 ` Claudio Fontana
2015-06-29 17:20 ` Claudio Fontana
2015-06-29 17:37 ` Peter Maydell
2015-06-29 17:37 ` Peter Maydell
2015-07-08 15:56 ` Marc Zyngier
2015-07-08 15:56 ` Marc Zyngier
2015-07-08 16:06 ` Peter Maydell
2015-07-08 16:06 ` Peter Maydell
2015-07-08 16:37 ` Marc Zyngier
2015-07-08 16:37 ` Marc Zyngier
2015-07-08 19:13 ` Peter Maydell
2015-07-08 19:13 ` Peter Maydell
2015-07-09 10:22 ` Christoffer Dall
2015-07-09 10:22 ` Christoffer Dall
2015-07-09 10:38 ` Peter Maydell
2015-07-09 10:38 ` Peter Maydell
2015-07-09 12:05 ` Christoffer Dall
2015-07-09 12:05 ` Christoffer Dall
2015-07-09 12:07 ` Peter Maydell
2015-07-09 12:07 ` Peter Maydell
2015-07-09 12:24 ` Christoffer Dall
2015-07-09 12:24 ` Christoffer Dall
2015-07-09 14:17 ` Christoffer Dall
2015-07-09 14:17 ` Christoffer Dall
2015-07-09 14:26 ` Peter Maydell
2015-07-09 14:26 ` Peter Maydell
2015-07-09 16:06 ` Christoffer Dall
2015-07-09 16:06 ` Christoffer Dall
2015-07-09 10:40 ` Jan Kiszka
2015-07-09 10:40 ` Jan Kiszka
2015-07-09 12:08 ` Christoffer Dall
2015-07-09 12:08 ` Christoffer Dall
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=558CD9DE.3010609@web.de \
--to=jan.kiszka@web.de \
--cc=claudio.fontana@huawei.com \
--cc=kvm@vger.kernel.org \
--cc=kvmarm@lists.cs.columbia.edu \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=marc.zyngier@arm.com \
--cc=peter.maydell@linaro.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.