KVM: arm/arm64: Skip updating PTE entry if no change
authorPunit Agrawal <punit.agrawal@arm.com>
Mon, 13 Aug 2018 10:43:51 +0000 (11:43 +0100)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 5 Sep 2018 07:26:36 +0000 (09:26 +0200)
commit 976d34e2dab10ece5ea8fe7090b7692913f89084 upstream.

When there is contention on faulting in a particular page table entry
at stage 2, the break-before-make requirement of the architecture can
lead to additional refaulting due to TLB invalidation.

Avoid this by skipping a page table update if the new value of the PTE
matches the previous value.

Cc: stable@vger.kernel.org
Fixes: d5d8184d35c9 ("KVM: ARM: Memory virtualization setup")
Reviewed-by: Suzuki Poulose <suzuki.poulose@arm.com>
Acked-by: Christoffer Dall <christoffer.dall@arm.com>
Signed-off-by: Punit Agrawal <punit.agrawal@arm.com>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
virt/kvm/arm/mmu.c

index d88f1e3fa52c17d355cf6003989ae8d25da8aec9..ec275b8472a9b7f26704960b87a5e62588f02a4f 100644 (file)
@@ -985,6 +985,10 @@ static int stage2_set_pte(struct kvm *kvm, struct kvm_mmu_memory_cache *cache,
        /* Create 2nd stage page table mapping - Level 3 */
        old_pte = *pte;
        if (pte_present(old_pte)) {
+               /* Skip page table update if there is no change */
+               if (pte_val(old_pte) == pte_val(*new_pte))
+                       return 0;
+
                kvm_set_pte(pte, __pte(0));
                kvm_tlb_flush_vmid_ipa(kvm, addr);
        } else {