Re: [PATCH v2] x86/kvm: Disable KVM_ASYNC_PF_SEND_ALWAYS

From: Paolo Bonzini <pbonzini@redhat.com>
To: Thomas Gleixner <tglx@linutronix.de>,
	Andy Lutomirski <luto@amacapital.net>,
	Vivek Goyal <vgoyal@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>,
	Andy Lutomirski <luto@kernel.org>,
	LKML <linux-kernel@vger.kernel.org>, X86 ML <x86@kernel.org>,
	kvm list <kvm@vger.kernel.org>, stable <stable@vger.kernel.org>
Subject: Re: [PATCH v2] x86/kvm: Disable KVM_ASYNC_PF_SEND_ALWAYS
Date: Wed, 8 Apr 2020 10:23:58 +0200	[thread overview]
Message-ID: <274f3d14-08ac-e5cc-0b23-e6e0274796c8@redhat.com> (raw)
In-Reply-To: <874ktukhku.fsf@nanos.tec.linutronix.de>

On 08/04/20 01:21, Thomas Gleixner wrote:
> Paolo Bonzini <pbonzini@redhat.com> writes:
> 
>> On 07/04/20 22:20, Thomas Gleixner wrote:
>>>>> Havind said that, I thought disabling interrupts does not mask exceptions.
>>>>> So page fault exception should have been delivered even with interrupts
>>>>> disabled. Is that correct? May be there was no vm exit/entry during
>>>>> those 10 seconds and that's why.
>>> No. Async PF is not a real exception. It has interrupt semantics and it
>>> can only be injected when the guest has interrupts enabled. It's bad
>>> design.
>>
>> Page-ready async PF has interrupt semantics.
>>
>> Page-not-present async PF however does not have interrupt semantics, it
>> has to be injected immediately or not at all (falling back to host page
>> fault in the latter case).
> 
> If interrupts are disabled in the guest then it is NOT injected and the
> guest is suspended. So it HAS interrupt semantics. Conditional ones,
> i.e. if interrupts are disabled, bail, if not then inject it.

Interrupts can be delayed by TPR or STI/MOV SS interrupt window, async
page faults cannot (again, not the page-ready kind).  Page-not-present
async page faults are almost a perfect match for the hardware use of #VE
(and it might even be possible to let the processor deliver the
exceptions).  There are other advantages:

- the only real problem with using #PF (with or without
KVM_ASYNC_PF_SEND_ALWAYS) seems to be the NMI reentrancy issue, which
would not be there for #VE.

- #VE are combined the right way with other exceptions (the
benign/contributory/pagefault stuff)

- adjusting KVM and Linux to use #VE instead of #PF would be less than
100 lines of code.

Paolo

> But that does not make it an exception by any means.
> 
> It never should have been hooked to #PF in the first place and it never
> should have been named that way. The functionality is to opportunisticly
> tell the guest to do some other stuff.
> 
> So the proper name for this seperate interrupt vector would be:
> 
>    VECTOR_OMG_DOS - Opportunisticly Make Guest Do Other Stuff
> 
> and the counter part
> 
>    VECTOR_STOP_DOS - Stop Doing Other Stuff 
> 
>> So page-not-present async PF definitely needs to be an exception, this
>> is independent of whether it can be injected when IF=0.
> 
> That wants to be a straight #PF. See my reply to Andy.
> 
>> Hypervisors do not have any reserved exception vector, and must use
>> vectors up to 31, which is why I believe #PF was used in the first place
>> (though that predates my involvement in KVM by a few years).
> 
> No. That was just bad taste or something worse. It has nothing to do
> with exceptions, see above. Stop proliferating the confusion.
> 
>> These days, #VE would be a much better exception to use instead (and
>> it also has a defined mechanism to avoid reentrancy).
> 
> #VE is not going to solve anything.
> 
> The idea of OMG_DOS is to (opportunisticly) avoid that the guest (and
> perhaps host) sit idle waiting for I/O until the fault has been
> resolved. That makes sense as there might be enough other stuff to do
> which does not depend on that particular page. If not then fine, the
> guest will go idle.
> 
> Thanks,
> 
>         tglx
> 