Re: [PATCH 5.10 096/126] KVM: x86/mmu: Use atomic ops to set SPTEs in TDP MMU map

[Sasha Levin]

On Tue, Apr 06, 2021 at 05:48:50PM +0200, Paolo Bonzini wrote:
> On 06/04/21 15:49, Sasha Levin wrote:
> > Yup. Is there anything wrong with those patches?
>
> The big issue, and the one that you ignoredz every time we discuss 
> this topic, is that this particular subset of 17 has AFAIK never been 
> tested by anyone.

Few of the CI systems that run on stable(-rc) releases run
kvm-unit-tests, which passed. So yes, this was tested.

> There's plenty of locking changes in here, one patch that you didn't 
> backport has this in its commit message:
>
>   This isn't technically a bug fix in the current code [...] but that
>   is all very, very subtle, and will break at the slightest sneeze,
>
> meaning that the locking in 5.10 and 5.11 was also less robust to 
> changes elsewhere in the code.
>
> Let's also talk about the process and the timing.  I got the "failed 
> to apply" automated message last Friday and I was going to work on the 
> backport today since yesterday was a holiday here.  I was *never* CCed 

There are a few more "FAILED:" mails that need attention that are older
than this one, I hope they're also in the queue.

> on a post of this backport for maintainers to review; you guys 

You're looking at it, this is the -rc cycle for stable kernels.

> *literally* took random subsets of patches from a feature that is new 
> and in active development, and hoped that they worked on a past 
> release.

Right, I looked at what needed to be backported, took it back to 5.4,
and ran kvm-unit-tests on it.

What other hoops should we jump through so we won't need to "hope"
anymore?

> I could be happy because you just provided me with a perfect example 
> of why to use my employer's franken-kernel instead of upstream stable 
> kernels... ;) but this is not how a world-class operating system is 
> developed.  Who cares if a VM breaks or even if my laptop panics; but 
> I'd seriously fear for my data if you applied the same attitude to XFS 
> or ext4fs.
>
> For now, please drop all 17 patches from 5.10 and 5.11.  I'll send a 
> tested backport as soon as possible.

Sure, I'll drop them. Please let us know when a backport is available.

--
Thanks,
Sasha