Re: [PATCH 5.2 000/413] 5.2.3-stable review

[Paolo Bonzini]

On 25/07/19 13:34, Greg Kroah-Hartman wrote:
> Any chance you can run 'git bisect' to find the offending patch?  Or
> just try reverting a few, you can ignore the ppc ones, so that only
> leaves you 7 different commits.
> 
> Does this same test pass in 5.3-rc1?

Anders, are you running the same kvm-unit-tests commit that passed for 
5.2.2?  My suspicion is that your previous test didn't have this commit

    commit 95d6d2c3228891537ee8e35d2e2984964ee0cf6b
    Author:     Krish Sadhukhan <krish.sadhukhan@xxxxxxxxxx>
    AuthorDate: Fri Jun 28 18:14:47 2019 -0400
    Commit:     Paolo Bonzini <pbonzini@xxxxxxxxxx>
    CommitDate: Thu Jul 11 14:26:53 2019 +0200

    nVMX: Test Host Segment Registers and Descriptor Tables on vmentry of nested guests
    

since the symptoms match and the corresponding fix was made in 5.3.

I think Linaro's tests would be helped by making kvm-unit-tests.git a submodule of
Linux, but I'm a bit wary since it would be the first submodule and I wouldn't know
where to put it...

Paolo