KVM ARM Archive on lore.kernel.org
 help / color / Atom feed
From: Marc Zyngier <maz@kernel.org>
To: Alexander Graf <graf@amazon.com>
Cc: kvm@vger.kernel.org, linux-doc@vger.kernel.org,
	Catalin Marinas <catalin.marinas@arm.com>,
	linux-kernel@vger.kernel.org,
	Russell King <linux@armlinux.org.uk>,
	Paolo Bonzini <pbonzini@redhat.com>,
	Steven Price <steven.price@arm.com>,
	Will Deacon <will@kernel.org>,
	kvmarm@lists.cs.columbia.edu,
	linux-arm-kernel@lists.infradead.org
Subject: Re: [PATCH 0/9] arm64: Stolen time support
Date: Wed, 14 Aug 2019 15:19:02 +0100
Message-ID: <8636i3omnd.wl-maz@kernel.org> (raw)
In-Reply-To: <8ca5c106-7c12-4c6e-6d81-a90f281a9894@amazon.com>

On Wed, 14 Aug 2019 14:02:25 +0100,
Alexander Graf <graf@amazon.com> wrote:
> 
> 
> 
> On 05.08.19 15:06, Steven Price wrote:
> > On 03/08/2019 19:05, Marc Zyngier wrote:
> >> On Fri,  2 Aug 2019 15:50:08 +0100
> >> Steven Price <steven.price@arm.com> wrote:
> >> 
> >> Hi Steven,
> >> 
> >>> This series add support for paravirtualized time for arm64 guests and
> >>> KVM hosts following the specification in Arm's document DEN 0057A:
> >>> 
> >>> https://developer.arm.com/docs/den0057/a
> >>> 
> >>> It implements support for stolen time, allowing the guest to
> >>> identify time when it is forcibly not executing.
> >>> 
> >>> It doesn't implement support for Live Physical Time (LPT) as there are
> >>> some concerns about the overheads and approach in the above
> >>> specification, and I expect an updated version of the specification to
> >>> be released soon with just the stolen time parts.
> >> 
> >> Thanks for posting this.
> >> 
> >> My current concern with this series is around the fact that we allocate
> >> memory from the kernel on behalf of the guest. It is the first example
> >> of such thing in the ARM port, and I can't really say I'm fond of it.
> >> 
> >> x86 seems to get away with it by having the memory allocated from
> >> userspace, why I tend to like more. Yes, put_user is more
> >> expensive than a straight store, but this isn't done too often either.
> >> 
> >> What is the rational for your current approach?
> > 
> > As I see it there are 3 approaches that can be taken here:
> > 
> > 1. Hypervisor allocates memory and adds it to the virtual machine. This
> > means that everything to do with the 'device' is encapsulated behind the
> > KVM_CREATE_DEVICE / KVM_[GS]ET_DEVICE_ATTR ioctls. But since we want the
> > stolen time structure to be fast it cannot be a trapping region and has
> > to be backed by real memory - in this case allocated by the host kernel.
> > 
> > 2. Host user space allocates memory. Similar to above, but this time
> > user space needs to manage the memory region as well as the usual
> > KVM_CREATE_DEVICE dance. I've no objection to this, but it means
> > kvmtool/QEMU needs to be much more aware of what is going on (e.g. how
> > to size the memory region).
> 
> You ideally want to get the host overhead for a VM to as little as you
> can. I'm not terribly fond of the idea of reserving a full page just
> because we're too afraid of having the guest donate memory.

Well, reduce the amount of memory you give to the guest by one page,
and allocate that page to the stolen time device. Problem solved!

Seriously, if you're worried about the allocation of a single page,
you should first look at how many holes we have in the vcpu structure,
for example (even better, with the 8.4 NV patches applied). Just
fixing that would give you that page back *per vcpu*.

> > 3. Guest kernel "donates" the memory to the hypervisor for the
> > structure. As far as I'm aware this is what x86 does. The problems I see
> > this approach are:
> > 
> >   a) kexec becomes much more tricky - there needs to be a disabling
> > mechanism for the guest to stop the hypervisor scribbling on memory
> > before starting the new kernel.
> 
> I wouldn't call "quiesce a device" much more tricky. We have to do
> that for other devices as well today.

And since there is no standard way of doing it, we keep inventing
weird and wonderful ways of doing so -- cue the terrible GICv3 LPI
situation, and all the various hacks to keep existing IOMMU mappings
around across firmware/kernel handovers as well as kexec.

> 
> >   b) If there is more than one entity that is interested in the
> > information (e.g. firmware and kernel) then this requires some form of
> > arbitration in the guest because the hypervisor doesn't want to have to
> > track an arbitrary number of regions to update.
> 
> Why would FW care?

Exactly. It doesn't care. Not caring means it doesn't know about the
page the guest has allocated for stolen time, and starts using it for
its own purposes. Hello, memory corruption. Same thing goes if you
reboot into a non stolen time aware kernel.

> 
> >   c) Performance can suffer if the host kernel doesn't have a suitably
> > aligned/sized area to use. As you say - put_user() is more expensive.
> 
> Just define the interface to always require natural alignment when
> donating a memory location?
> 
> > The structure is updated on every return to the VM.
> 
> If you really do suffer from put_user(), there are alternatives. You
> could just map the page on the registration hcall and then leave it
> pinned until the vcpu gets destroyed again.

put_user() should be cheap enough. It is one of the things we tend to
optimise anyway. And yes, worse case, we pin the page.

> 
> > Of course x86 does prove the third approach can work, but I'm not sure
> > which is actually better. Avoid the kexec cancellation requirements was
> > the main driver of the current approach. Although many of the
> 
> I really don't understand the problem with kexec cancellation. Worst
> case, let guest FW set it up for you and propagate only the address
> down via ACPI/DT. That way you can mark the respective memory as
> reserved too.

We already went down that road with the LPI hack. I'm not going there
again if we can avoid it. And it turn out that we can. Just allocate
the stolen time page as a separate memblock, give it to KVM for that
purpose.

Your suggestion of letting the guest firmware set something up only
works if whatever you're booting after that understands it. If it
doesn't, you're screwed.

> But even with a Linux only mechanism, just take a look at
> arch/x86/kernel/kvmclock.c. All they do to remove the map is to hook
> into machine_crash_shutdown() and machine_shutdown().

I'm not going to take something that is Linux specific. It has to work
for all guests, at all times, whether they know about the hypervisor
service or not.

	M.

> 
> 
> Alex
> 
> > conversations about this were also tied up with Live Physical Time which
> > adds its own complications.
> > 
> > Steve
> > _______________________________________________
> > kvmarm mailing list
> > kvmarm@lists.cs.columbia.edu
> > https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
> > 

-- 
Jazz is not dead, it just smells funny.
_______________________________________________
kvmarm mailing list
kvmarm@lists.cs.columbia.edu
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm

  reply index

Thread overview: 45+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-08-02 14:50 Steven Price
2019-08-02 14:50 ` [PATCH 1/9] KVM: arm64: Document PV-time interface Steven Price
2019-08-03 11:13   ` Marc Zyngier
2019-08-05 13:06     ` Steven Price
2019-08-05  3:23   ` Zenghui Yu
2019-08-05 13:06     ` Steven Price
2019-08-05 16:40   ` Christophe de Dinechin
2019-08-07 13:21     ` Steven Price
2019-08-07 14:28       ` Christophe de Dinechin
2019-08-07 15:26         ` Steven Price
2019-08-02 14:50 ` [PATCH 2/9] KVM: arm/arm64: Factor out hypercall handling from PSCI code Steven Price
2019-08-02 14:50 ` [PATCH 3/9] KVM: arm64: Implement PV_FEATURES call Steven Price
2019-08-03 11:21   ` Marc Zyngier
2019-08-05 13:14     ` Steven Price
2019-08-02 14:50 ` [PATCH 4/9] KVM: arm64: Support stolen time reporting via shared structure Steven Price
2019-08-03 11:55   ` Marc Zyngier
2019-08-05 14:09     ` Steven Price
2019-08-03 17:58   ` Marc Zyngier
2019-08-03 18:13     ` Marc Zyngier
2019-08-05 14:18       ` Steven Price
2019-08-02 14:50 ` [PATCH 5/9] KVM: Allow kvm_device_ops to be const Steven Price
2019-08-02 14:50 ` [PATCH 6/9] KVM: arm64: Provide a PV_TIME device to user space Steven Price
2019-08-03 12:51   ` Marc Zyngier
2019-08-03 17:34     ` Marc Zyngier
2019-08-07 13:39       ` Steven Price
2019-08-07 13:51         ` Marc Zyngier
2019-08-05 16:10     ` Steven Price
2019-08-05 16:28       ` Marc Zyngier
2019-08-02 14:50 ` [PATCH 7/9] arm/arm64: Provide a wrapper for SMCCC 1.1 calls Steven Price
2019-08-05 10:03   ` Will Deacon
2019-08-02 14:50 ` [PATCH 8/9] arm/arm64: Make use of the SMCCC 1.1 wrapper Steven Price
2019-08-02 14:50 ` [PATCH 9/9] arm64: Retrieve stolen time as paravirtualized guest Steven Price
2019-08-04  9:53   ` Marc Zyngier
2019-08-08 15:29     ` Steven Price
2019-08-08 15:49       ` Marc Zyngier
2019-08-09 13:51   ` Zenghui Yu
2019-08-12 10:39     ` Steven Price
2019-08-13  6:06       ` Zenghui Yu
2019-08-03 18:05 ` [PATCH 0/9] arm64: Stolen time support Marc Zyngier
2019-08-05 13:06   ` Steven Price
2019-08-05 13:26     ` Marc Zyngier
2019-08-14 13:02     ` Alexander Graf
2019-08-14 14:19       ` Marc Zyngier [this message]
2019-08-14 14:52         ` [UNVERIFIED SENDER] " Alexander Graf
2019-08-16 10:23           ` Steven Price

Reply instructions:

You may reply publically 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=8636i3omnd.wl-maz@kernel.org \
    --to=maz@kernel.org \
    --cc=catalin.marinas@arm.com \
    --cc=graf@amazon.com \
    --cc=kvm@vger.kernel.org \
    --cc=kvmarm@lists.cs.columbia.edu \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-doc@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux@armlinux.org.uk \
    --cc=pbonzini@redhat.com \
    --cc=steven.price@arm.com \
    --cc=will@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

KVM ARM Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/kvmarm/0 kvmarm/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 kvmarm kvmarm/ https://lore.kernel.org/kvmarm \
		kvmarm@lists.cs.columbia.edu kvmarm@archiver.kernel.org
	public-inbox-index kvmarm


Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/edu.columbia.cs.lists.kvmarm


AGPL code for this site: git clone https://public-inbox.org/ public-inbox