Re: KVM NV + SVE host OS warning

[Oliver Upton <oliver.upton@linux.dev>]

On Thu, Sep 25, 2025 at 04:35:13PM +0100, Marc Zyngier wrote:
> On Thu, 25 Sep 2025 16:10:22 +0100,
> Jan Kotas <jank@cadence.com> wrote:
> > 
> > Hi Marc,
> > 
> > Thank you for a quick response.
> > 
> > > On 25 Sep 2025, at 16:38, Marc Zyngier <maz@kernel.org> wrote:
> > > 
> > > EXTERNAL MAIL
> > > 
> > > 
> > > [+Oliver for the SVE stuff]
> > > 
> > > Hi Jan,
> 
> [...]
> 
> > > 
> > > I would also recommend to update to 6.17 -- 6.16 was the first release
> > > with NV, and while it may work, it will also have a lot of ugly bugs.
> > 
> > I will try, but as I don’t own this machine, changing kernels might
> > be… challenging.
> 
> Well, you're already passing something on the command-line that is
> documented as "experimental and should be used with extreme caution".
> Changing the kernel is a walk in the park compared to that ;-).

+1, also it'd be good to know what you're doing in the L1 and the L2
that leads to this bug.

I just tried the original test setup I used when doing NV SVE and things
appear to work (kvmarm/next + initramfs at L1 + L2, tested with
sve-stress selftest).

> > If there’s anything else I can provide to help track it down,
> > please let me know.
> 
> Testing the provided patch would be good -- I would if I could.

Although I didn't repro the bug, the patch itself LGTM.

Thanks,
Oliver