From: "Dr. David Alan Gilbert" <dgilbert@redhat.com>
To: Vitaly Kuznetsov <vkuznets@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>,
Marcelo Tosatti <mtosatti@redhat.com>,
Richard Henderson <richard.henderson@linaro.org>,
qemu-devel@nongnu.org, Eduardo Habkost <ehabkost@redhat.com>
Subject: Re: [PATCH v2] i386: Make migration fail when Hyper-V reenlightenment was enabled but 'user_tsc_khz' is unset
Date: Mon, 29 Mar 2021 15:49:13 +0100 [thread overview]
Message-ID: <YGHo6WiRVcmVrnSZ@work-vm> (raw)
In-Reply-To: <87y2e69lg8.fsf@vitty.brq.redhat.com>
* Vitaly Kuznetsov (vkuznets@redhat.com) wrote:
> "Dr. David Alan Gilbert" <dgilbert@redhat.com> writes:
>
> > * Vitaly Kuznetsov (vkuznets@redhat.com) wrote:
> >> KVM doesn't fully support Hyper-V reenlightenment notifications on
> >> migration. In particular, it doesn't support emulating TSC frequency
> >> of the source host by trapping all TSC accesses so unless TSC scaling
> >> is supported on the destination host and KVM_SET_TSC_KHZ succeeds, it
> >> is unsafe to proceed with migration.
> >>
> >> KVM_SET_TSC_KHZ is called from two sites: kvm_arch_init_vcpu() and
> >> kvm_arch_put_registers(). The later (intentionally) doesn't propagate
> >> errors allowing migrations to succeed even when TSC scaling is not
> >> supported on the destination. This doesn't suit 're-enlightenment'
> >> use-case as we have to guarantee that TSC frequency stays constant.
> >>
> >> Require 'tsc-frequency=' command line option to be specified for successful
> >> migration when re-enlightenment was enabled by the guest.
> >>
> >> Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
> >> ---
> >> This patch is a successor of "[PATCH 3/3] i386: Make sure
> >> kvm_arch_set_tsc_khz() succeeds on migration when 'hv-reenlightenment'
> >> was exposed" taking a different approach suggested by Paolo.
> >> ---
> >> docs/hyperv.txt | 5 +++++
> >> target/i386/kvm/hyperv-proto.h | 1 +
> >> target/i386/machine.c | 20 ++++++++++++++++++++
> >> 3 files changed, 26 insertions(+)
> >>
> >> diff --git a/docs/hyperv.txt b/docs/hyperv.txt
> >> index 5df00da54fc4..e53c581f4586 100644
> >> --- a/docs/hyperv.txt
> >> +++ b/docs/hyperv.txt
> >> @@ -160,6 +160,11 @@ the hypervisor) until it is ready to switch to the new one. This, in conjunction
> >> with hv-frequencies, allows Hyper-V on KVM to pass stable clocksource (Reference
> >> TSC page) to its own guests.
> >>
> >> +Note, KVM doesn't fully support re-enlightenment notifications and doesn't
> >> +emulate TSC accesses after migration so 'tsc-frequency=' CPU option also has to
> >> +be specified to make migration succeed. The destination host has to either have
> >> +the same TSC frequency or support TSC scaling CPU feature.
> >> +
> >> Recommended: hv-frequencies
> >>
> >> 3.16. hv-evmcs
> >> diff --git a/target/i386/kvm/hyperv-proto.h b/target/i386/kvm/hyperv-proto.h
> >> index 056a305be38c..e30d64b4ade4 100644
> >> --- a/target/i386/kvm/hyperv-proto.h
> >> +++ b/target/i386/kvm/hyperv-proto.h
> >> @@ -139,6 +139,7 @@
> >> * Reenlightenment notification MSRs
> >> */
> >> #define HV_X64_MSR_REENLIGHTENMENT_CONTROL 0x40000106
> >> +#define HV_REENLIGHTENMENT_ENABLE_BIT (1u << 16)
> >> #define HV_X64_MSR_TSC_EMULATION_CONTROL 0x40000107
> >> #define HV_X64_MSR_TSC_EMULATION_STATUS 0x40000108
> >>
> >> diff --git a/target/i386/machine.c b/target/i386/machine.c
> >> index 7259fe6868c6..137604ddb898 100644
> >> --- a/target/i386/machine.c
> >> +++ b/target/i386/machine.c
> >> @@ -883,11 +883,31 @@ static bool hyperv_reenlightenment_enable_needed(void *opaque)
> >> env->msr_hv_tsc_emulation_status != 0;
> >> }
> >>
> >> +static int hyperv_reenlightenment_post_load(void *opaque, int version_id)
> >> +{
> >> + X86CPU *cpu = opaque;
> >> + CPUX86State *env = &cpu->env;
> >> +
> >> + /*
> >> + * KVM doesn't fully support re-enlightenment notifications so we need to
> >> + * make sure TSC frequency doesn't change upon migration.
> >> + */
> >> + if ((env->msr_hv_reenlightenment_control & HV_REENLIGHTENMENT_ENABLE_BIT) &&
> >> + !env->user_tsc_khz) {
> >> + error_report("Guest enabled re-enlightenment notifications, "
> >> + "'tsc-frequency=' has to be specified");
> >
> > It's unusual to fail on the destination for a valid configuration but
> > guest state; wouldn't it be better to always insist on tsc-frequency if
> > that hv feature is exposed; failing early before reeiving the state?
> >
>
> Doing so would make a number of currently existing configurations
> invalid, even when re-enlightenment is not to be used by the
> guest. AFAIR Windows without Hyper-V doesn't enable it. Generally, we
> just advise people to 'enable all currently supported hyper-v
> enlightenments' to make things easier so reenlightenment may end up
> being added for no particular reason.
Ouch, that's difficult - the problem with testing this late is that the
migration fails right at the end so it's an unpleasent surprise.
Could you disallow re-enlightenment without tsc-frequency on new machine
types?
Dave
> FWIW, v1 of this patch used a slightly different approach: I was
> requiring kvm_arch_set_tsc_khz() to succeed. This is not exactly the
> same as e.g. even when the destination host doesn't support TSC scaling,
> it may happen to have the same TSC frequency.
>
> --
> Vitaly
>
--
Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK
next prev parent reply other threads:[~2021-03-29 14:51 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-03-19 12:38 [PATCH v2] i386: Make migration fail when Hyper-V reenlightenment was enabled but 'user_tsc_khz' is unset Vitaly Kuznetsov
2021-03-29 12:21 ` Dr. David Alan Gilbert
2021-03-29 12:46 ` Vitaly Kuznetsov
2021-03-29 14:49 ` Dr. David Alan Gilbert [this message]
2021-03-30 11:51 ` Vitaly Kuznetsov
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=YGHo6WiRVcmVrnSZ@work-vm \
--to=dgilbert@redhat.com \
--cc=ehabkost@redhat.com \
--cc=mtosatti@redhat.com \
--cc=pbonzini@redhat.com \
--cc=qemu-devel@nongnu.org \
--cc=richard.henderson@linaro.org \
--cc=vkuznets@redhat.com \
/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).