KVM: arm64: Fix handling of merging tables into a block entry
authorYanan Wang <wangyanan55@huawei.com>
Tue, 1 Dec 2020 20:10:33 +0000 (04:10 +0800)
committerMarc Zyngier <maz@kernel.org>
Wed, 2 Dec 2020 09:42:36 +0000 (09:42 +0000)
commit3a0b870e3448302ca2ba703bea1b79b61c3f33c6
tree3820f50b0ad6b3d780ac5ea45d6ae09b614f32ef
parent5c646b7e1d8bcb12317426287c516dfa4c5171c2
KVM: arm64: Fix handling of merging tables into a block entry

When dirty logging is enabled, we collapse block entries into tables
as necessary. If dirty logging gets canceled, we can end-up merging
tables back into block entries.

When this happens, we must not only free the non-huge page-table
pages but also invalidate all the TLB entries that can potentially
cover the block. Otherwise, we end-up with multiple possible translations
for the same physical page, which can legitimately result in a TLB
conflict.

To address this, replease the bogus invalidation by IPA with a full
VM invalidation. Although this is pretty heavy handed, it happens
very infrequently and saves a bunch of invalidations by IPA.

Signed-off-by: Yanan Wang <wangyanan55@huawei.com>
[maz: fixup commit message]
Signed-off-by: Marc Zyngier <maz@kernel.org>
Acked-by: Will Deacon <will@kernel.org>
Link: https://lore.kernel.org/r/20201201201034.116760-3-wangyanan55@huawei.com
arch/arm64/kvm/hyp/pgtable.c