[PATCH 2/2] KVM: arm64: Destroy the stage-2 page-table periodically

[Oliver Upton]

On Thu, Aug 07, 2025 at 11:58:01AM -0700, Raghavendra Rao Ananta wrote:
> Hi Oliver,
> 
> >
> > Protected mode is affected by the same problem, potentially even worse
> > due to the overheads of calling into EL2. Both protected and
> > non-protected flows should use stage2_destroy_range().
> >
> I experimented with this (see diff below), and it looks like it takes
> significantly longer to finish the destruction even for a very small
> VM. For instance, it takes ~140 seconds on an Ampere Altra machine.
> This is probably because we run cond_resched() for every breakup in
> the entire sweep of the possible address range, 0 to  ~(0ULL), even
> though there are no actual mappings there, and we context switch out
> more often.

This seems more like an issue with the upper bound on a pKVM walk rather
than a problem with the suggestion. The information in pgt->ia_bits is
actually derived from the VTCR value of the owning MMU.

Even though we never use the VTCR value in hardware, pKVM MMUs have a
valid VTCR value that encodes the size of the IPA space and we use that
in the common stage-2 abort path.

I'm attaching some fixups that I have on top of your series that'd allow
the resched logic to remain common, like it is in other MMU flows.

>From 421468dcaa4692208c3f708682b058cfc072a984 Mon Sep 17 00:00:00 2001
From: Oliver Upton <oliver.upton at linux.dev>
Date: Fri, 8 Aug 2025 11:43:12 -0700
Subject: [PATCH 4/4] fixup! KVM: arm64: Destroy the stage-2 page-table
 periodically

---
 arch/arm64/kvm/mmu.c | 60 ++++++++++++++++++--------------------------
 1 file changed, 25 insertions(+), 35 deletions(-)

diff --git a/arch/arm64/kvm/mmu.c b/arch/arm64/kvm/mmu.c
index b82412323054..fc93cc256bd8 100644
--- a/arch/arm64/kvm/mmu.c
+++ b/arch/arm64/kvm/mmu.c
@@ -383,40 +383,6 @@ static void stage2_flush_vm(struct kvm *kvm)
 	srcu_read_unlock(&kvm->srcu, idx);
 }
 
-/*
- * Assume that @pgt is valid and unlinked from the KVM MMU to free the
- * page-table without taking the kvm_mmu_lock and without performing any
- * TLB invalidations.
- *
- * Also, the range of addresses can be large enough to cause need_resched
- * warnings, for instance on CONFIG_PREEMPT_NONE kernels. Hence, invoke
- * cond_resched() periodically to prevent hogging the CPU for a long time
- * and schedule something else, if required.
- */
-static void stage2_destroy_range(struct kvm_pgtable *pgt, phys_addr_t addr,
-			      phys_addr_t end)
-{
-	u64 next;
-
-	do {
-		next = stage2_range_addr_end(addr, end);
-		kvm_pgtable_stage2_destroy_range(pgt, addr, next - addr);
-
-		if (next != end)
-			cond_resched();
-	} while (addr = next, addr != end);
-}
-
-static void kvm_destroy_stage2_pgt(struct kvm_pgtable *pgt)
-{
-	if (!is_protected_kvm_enabled()) {
-		stage2_destroy_range(pgt, 0, BIT(pgt->ia_bits));
-		kvm_pgtable_stage2_destroy_pgd(pgt);
-	} else {
-		pkvm_pgtable_stage2_destroy(pgt);
-	}
-}
-
 /**
  * free_hyp_pgds - free Hyp-mode page tables
  */
@@ -938,11 +904,35 @@ static int kvm_init_ipa_range(struct kvm_s2_mmu *mmu, unsigned long type)
 	return 0;
 }
 
+/*
+ * Assume that @pgt is valid and unlinked from the KVM MMU to free the
+ * page-table without taking the kvm_mmu_lock and without performing any
+ * TLB invalidations.
+ *
+ * Also, the range of addresses can be large enough to cause need_resched
+ * warnings, for instance on CONFIG_PREEMPT_NONE kernels. Hence, invoke
+ * cond_resched() periodically to prevent hogging the CPU for a long time
+ * and schedule something else, if required.
+ */
+static void stage2_destroy_range(struct kvm_pgtable *pgt, phys_addr_t addr,
+				 phys_addr_t end)
+{
+	u64 next;
+
+	do {
+		next = stage2_range_addr_end(addr, end);
+		KVM_PGT_FN(kvm_pgtable_stage2_destroy_range)(pgt, addr, next - addr);
+
+		if (next != end)
+			cond_resched();
+	} while (addr = next, addr != end);
+}
+
 static void kvm_stage2_destroy(struct kvm_pgtable *pgt)
 {
 	unsigned int ia_bits = VTCR_EL2_IPA(pgt->mmu->vtcr);
 
-	KVM_PGT_FN(kvm_pgtable_stage2_destroy_range)(pgt, 0, BIT(ia_bits));
+	stage2_destroy_range(pgt, 0, BIT(ia_bits));
 	KVM_PGT_FN(kvm_pgtable_stage2_destroy_pgd)(pgt);
 }
 
-- 
2.39.5