[PATCH v5 00/21] KVM: arm64: Rewrite page-table code and fault handling

[Will Deacon]

Hi Alex,

On Thu, Oct 01, 2020 at 11:21:43AM +0100, Alexandru Elisei wrote:
> On 9/11/20 2:25 PM, Will Deacon wrote:
> > It's me again, with version five of the KVM page-table rework previously
> > seen at:
> >
> >   v1: https://lore.kernel.org/r/20200730153406.25136-1-will@kernel.org
> >   v2: https://lore.kernel.org/r/20200818132818.16065-1-will@kernel.org
> >   v3: https://lore.kernel.org/r/20200825093953.26493-1-will@kernel.org
> >   v4: https://lore.kernel.org/r/20200907152344.12978-1-will@kernel.org
> >
> > Changes since v4 include:
> >
> >   * Add comments to the kerneldoc describing alignment behaviour for
> >     addresses and size parameters
> >   * Fix formatting of IPA size messages
> >   * Fix handling of unaligned addresses in kvm_phys_addr_ioremap()
> >   * Add DSB after zeroing stage-2 PGD pages
> >   * Add reviewer tags
> >
> > Once again, thanks to Alex, Gavin and Andrew for their comments.
> 
> I didn't realize that there won't be another iteration of the series. I reviewed
> all the patches with the exception of patch #5 ("KVM: arm64: Use generic allocator
> for hyp stage-1 page-tables") because I don't know much about what needs to mapped
> at EL2 stage 1, everything else looks alright to me. If the commit message can be
> modified, you can add my Reviewed-by tag to the series (sans #5); it's fine if
> that's not possible.

The patches ended up in kvmarm/next as the remaining comments were pretty
minor, so I've just been sending patches on top to clear up the leftovers
(e.g. passing the TLBI level when handling a stage-2 permission fault).

Cheers for your review, the series definitely ended up in a much better
shape for it. The hyp code is a walk in the park compared with the stage-2
parts, so don't worry about skipping that.

Will