[PATCH v1 6/6] KVM: arm64: Use kvm_s2_fault_vma_info in pkvm_mem_abort()

[Vincent Donnefort]

To paraphrase Marc in commit 08f97454b7fa ("KVM: arm64: Fix protected
mode handling of pages larger than 4kB"), pkvm_pgtable_stage2_map()
assumes the address passed as a parameter is aligned to the size of the
intended mapping, while HPFAR_EL2 gives the IPA minus the bottom 12
bits, regardless of the system page size configuration.

To fix this alignment, bring support for kvm_s2_fault_vma_info in
pkvm_mem_abort() and use its members where possible. They do contain the
page-alignment we need to fix the fault on system with pages larger than
4K.

Also, add a check at the start of pkvm_pgtable_stage2_map(), it does not
support !PAGE_ALIGNED arguments.

Fixes: ea03466e806f ("KVM: arm64: Handle aborts from protected VMs")
Cc: stable at vger.kernel.org
Signed-off-by: Vincent Donnefort <vdonnefort at google.com>
---
 arch/arm64/kvm/mmu.c  | 24 +++++++++++++++---------
 arch/arm64/kvm/pkvm.c |  3 +++
 2 files changed, 18 insertions(+), 9 deletions(-)

diff --git a/arch/arm64/kvm/mmu.c b/arch/arm64/kvm/mmu.c
index ee0fcca838c1..167b769107ee 100644
--- a/arch/arm64/kvm/mmu.c
+++ b/arch/arm64/kvm/mmu.c
@@ -1374,9 +1374,11 @@ static bool fault_supports_stage2_huge_mapping(const struct kvm_s2_fault_desc *s
 	if (map_size == PAGE_SIZE)
 		return true;
 
-	/* pKVM only supports PMD_SIZE huge-mappings */
-	if (is_protected_kvm_enabled() && map_size != PMD_SIZE)
-		return false;
+	/* pKVM only supports PMD_SIZE huge-mappings for non-protected VMs */
+	if (is_protected_kvm_enabled()) {
+		if (vcpu_is_protected(s2fd->vcpu) || map_size != PMD_SIZE)
+			return false;
+	}
 
 	size = memslot->npages * PAGE_SIZE;
 
@@ -2167,12 +2169,12 @@ int kvm_handle_guest_sea(struct kvm_vcpu *vcpu)
 static int pkvm_mem_abort(const struct kvm_s2_fault_desc *s2fd)
 {
 	unsigned int flags = FOLL_HWPOISON | FOLL_LONGTERM | FOLL_WRITE;
+	struct kvm_s2_fault_vma_info s2vi = {};
 	struct kvm_vcpu *vcpu = s2fd->vcpu;
 	struct kvm_pgtable *pgt = vcpu->arch.hw_mmu->pgt;
 	struct mm_struct *mm = current->mm;
 	struct kvm *kvm = vcpu->kvm;
 	void *hyp_memcache;
-	struct page *page;
 	int ret;
 
 	hyp_memcache = get_mmu_memcache(vcpu);
@@ -2180,12 +2182,16 @@ static int pkvm_mem_abort(const struct kvm_s2_fault_desc *s2fd)
 	if (ret)
 		return -ENOMEM;
 
+	ret = kvm_s2_fault_get_vma_info(s2fd, &s2vi);
+	if (ret)
+		return ret;
+
 	ret = account_locked_vm(mm, 1, true);
 	if (ret)
 		return ret;
 
 	mmap_read_lock(mm);
-	ret = pin_user_pages(s2fd->hva, 1, flags, &page);
+	ret = pin_user_pages(s2fd->hva, 1, flags, &s2vi.page);
 	mmap_read_unlock(mm);
 
 	if (ret == -EHWPOISON) {
@@ -2195,7 +2201,7 @@ static int pkvm_mem_abort(const struct kvm_s2_fault_desc *s2fd)
 	} else if (ret != 1) {
 		ret = -EFAULT;
 		goto dec_account;
-	} else if (!folio_test_swapbacked(page_folio(page))) {
+	} else if (!folio_test_swapbacked(page_folio(s2vi.page))) {
 		/*
 		 * We really can't deal with page-cache pages returned by GUP
 		 * because (a) we may trigger writeback of a page for which we
@@ -2215,8 +2221,8 @@ static int pkvm_mem_abort(const struct kvm_s2_fault_desc *s2fd)
 	}
 
 	write_lock(&kvm->mmu_lock);
-	ret = pkvm_pgtable_stage2_map(pgt, s2fd->fault_ipa, PAGE_SIZE,
-				      page_to_phys(page), KVM_PGTABLE_PROT_RWX,
+	ret = pkvm_pgtable_stage2_map(pgt, gfn_to_gpa(s2vi.gfn), PAGE_SIZE,
+				      page_to_phys(s2vi.page), KVM_PGTABLE_PROT_RWX,
 				      hyp_memcache, 0);
 	write_unlock(&kvm->mmu_lock);
 	if (ret) {
@@ -2227,7 +2233,7 @@ static int pkvm_mem_abort(const struct kvm_s2_fault_desc *s2fd)
 
 	return 0;
 unpin:
-	unpin_user_pages(&page, 1);
+	unpin_user_page(s2vi.page);
 dec_account:
 	account_locked_vm(mm, 1, false);
 	return ret;
diff --git a/arch/arm64/kvm/pkvm.c b/arch/arm64/kvm/pkvm.c
index 8e4c6e4bec12..c298a5b9b12a 100644
--- a/arch/arm64/kvm/pkvm.c
+++ b/arch/arm64/kvm/pkvm.c
@@ -414,6 +414,9 @@ int pkvm_pgtable_stage2_map(struct kvm_pgtable *pgt, u64 addr, u64 size,
 	u64 end = addr + size;
 	int ret;
 
+	if (WARN_ON_ONCE(!PAGE_ALIGNED(addr | size)))
+		return -EINVAL;
+
 	lockdep_assert_held_write(&kvm->mmu_lock);
 	mapping = pkvm_mapping_iter_first(&pgt->pkvm_mappings, addr, end - 1);
 
-- 
2.55.0.1082.g2b9226bbc0-goog