[PATCH v9 00/14] KVM: arm64: Add support for FEAT_TLBIRANGE

[Sean Christopherson <seanjc@google.com>]

On Mon, Aug 14, 2023, Raghavendra Rao Ananta wrote:
> On Mon, Aug 14, 2023 at 4:16?PM Sean Christopherson <seanjc@google.com> wrote:
> >
> > On Fri, Aug 11, 2023, Raghavendra Rao Ananta wrote:
> > > The series is based off of upstream v6.5-rc1.
> >
> > Lies!  :-)
> >
> > This is based off one of the kvmarm.git topic branches (I didn't bother to figure
> > out which one), not v6.5-rc1.
> >
> Sorry, what am I missing here? My git log is as follows:

Hmm, not sure what's going on.  Maybe I misinterpreted why `git am` was failing?
I assumed it was because there were objects in kvmarm that I didn't have locally,
and fetching kvmarm allowed am to complete, though with 3-way merges, which IIUC
shouldn't have been necessary if I was using the exact same base.  Or maybe I
messed up and didn't actually reset to 6.5-rc1.

> $ git log --oneline upstream_tlbi_range_v9
> 5025857507abe (upstream_tlbi_range_v9) KVM: arm64: Use TLBI
> range-based instructions for unmap
> 5c0291b99a8fc KVM: arm64: Invalidate the table entries upon a range
> 8c46b54d4aaec KVM: arm64: Flush only the memslot after write-protect
> 231abaeb7ffc2 KVM: arm64: Implement kvm_arch_flush_remote_tlbs_range()
> 5ec291b863309 KVM: arm64: Define kvm_tlb_flush_vmid_range()
> 5bcd7a085c34e KVM: arm64: Implement  __kvm_tlb_flush_vmid_range()
> ea08f9dff7e5b arm64: tlb: Implement __flush_s2_tlb_range_op()
> b3178687947c9 arm64: tlb: Refactor the core flush algorithm of __flush_tlb_range
> a4850fa988eef KVM: Move kvm_arch_flush_remote_tlbs_memslot() to common code
> 306dc4e6afd37 KVM: Allow range-based TLB invalidation from common code
> d02785a0a1e01 KVM: Remove CONFIG_HAVE_KVM_ARCH_TLB_FLUSH_ALL
> 136fa2d254537 KVM: arm64: Use kvm_arch_flush_remote_tlbs()
> e35c68a75170d KVM: Declare kvm_arch_flush_remote_tlbs() globally
> 5d592777b9bba KVM: Rename kvm_arch_flush_remote_tlb() to
> kvm_arch_flush_remote_tlbs()
> 06c2afb862f9d (tag: v6.5-rc1, tag: linux/v6.5-rc1) Linux 6.5-rc1
> c192ac7357683 MAINTAINERS 2: Electric Boogaloo
> f71f64210d698 Merge tag 'dma-mapping-6.5-2023-07-09' of
> git://git.infradead.org/users/hch/dma-mapping
> ...
> 
> Isn't the commit, 06c2afb862f9d (06c2afb862f9d (tag: v6.5-rc1, tag:
> linux/v6.5-rc1) Linux 6.5-rc1) the 'base' commit?

Ya, should be.

Either way, even if this is PEBKAC on my end, using --base would be nice, e.g.
then you can definitely say it's my fault ;-)

> > Please try to incorporate git format-patch's "--base" option into your workflow,
> > e.g. I do "git format-patch --base=HEAD~$nr" where $nr is the number of patches
> > I am posting.
> >
> > It's not foolproof, e.g. my approach doesn't help if I have a local patch that
> > I'm not posting, but 99% of the time it Just Works and eliminates any ambuitity.
> >
> > You can also do "--base=auto", but that only does the right thing if your series
> > has its upstream branch set to the base/tree that you want your patches applied
> > to (I use the upstream branch for a completely different purpose for my dev branches).