On 2020-04-18 03:49, Zengtao (B) wrote:
-----Original Message-----
From: Marc Zyngier [mailto:maz@xxxxxxxxxx]
Sent: Friday, April 17, 2020 11:06 PM
To: Zengtao (B)
Cc: George Cherian; Dave.Martin@xxxxxxx; alexandru.elisei@xxxxxxx;
andre.przywara@xxxxxxx; christoffer.dall@xxxxxxx;
james.morse@xxxxxxx; jintack@xxxxxxxxxxxxxxx;
julien.thierry.kdev@xxxxxxxxx; kvm@xxxxxxxxxxxxxxx;
kvmarm@xxxxxxxxxxxxxxxxxxxxx; linux-arm-kernel@xxxxxxxxxxxxxxxxxxx;
suzuki.poulose@xxxxxxx; Anil Kumar Reddy H; Ganapatrao Kulkarni
Subject: Re: [PATCH v2 00/94] KVM: arm64: ARMv8.3/8.4 Nested
Virtualization support
On Thu, 16 Apr 2020 19:22:21 +0100
Marc Zyngier <maz@xxxxxxxxxx> wrote:
> Hi Zengtao,
>
> On 2020-04-16 02:38, Zengtao (B) wrote:
> > Hi Marc:
> >
> > Got it.
> > Really a bit patch set :)
>
> Well, yeah... ;-)
>
> >
> > BTW, I have done a basic kvm unit test
> > git://git.kernel.org/pub/scm/virt/kvm/kvm-unit-tests.git
> > And I find that after apply the patch KVM: arm64: VNCR-ize ELR_EL1,
> > The psci test failed for some reason, I can't understand why, this
> > is only the test result.(find the patch by git bisect + kvm test)
>
> That it is that mechanical, we should be able to quickly nail that one.
>
> > My platform: Hisilicon D06 board.
> > Linux kernel: Linux 5.6-rc6 + nv patches(some rebases)
> > Could you help to take a look?
>
> I'll have a look tomorrow. I'm in the middle of refactoring the series
> for 5.7, and things have changed quite a bit. Hopefully this isn't a VHE
> vs non-VHE issue.
So I've repeatedly tried with the current state of the NV patches[1],
on both an ARMv8.0 system (Seattle) and an ARMv8.2 pile of putrid junk
(vim3l). PSCI is pretty happy, although I can only test with at most 8
vcpus (GICv2 gets in the way).
Can you please:
- post the detailed error by running the PSCI unit test on its own
I tried to trace the error, and I found in kernel function
kvm_mpidr_to_vcpu,
casually, mpidr returns zero and we can't get the expected vcpu, and
psci
test failed due to this.
Can you post the exact error message from the unit test?
And as I mentioned in my last before, the psci error is introduced by
the
patch KVM: arm64: VNCR-ize ELR_EL1.(Only test result)
Maybe you have to try tens of times to reproduce. :)
Deep into the patch itself, I don't find any connection between the
patch
and the issue.
Me neither, and I haven't managed to reproduce your issue.
- test with the current state of the patches
I test with your nv-5.7-rc1-WIP branch and latest kvm_unit_test, the
error still exist.
How many vcpus do you create with this PSCI test?
Thanks,
M.
--
Jazz is not dead. It just smells funny...
