[PATCH v1 06/13] KVM: arm64: Add support for KVM_MEM_USERFAULT

[James Houghton]

Adhering to the requirements of KVM Userfault:

1. When it is toggled (either on or off), zap the second stage with
   kvm_arch_flush_shadow_memslot(). This is to (1) respect
   userfault-ness and (2) to reconstruct block mappings.
2. While KVM_MEM_USERFAULT is enabled, restrict new second-stage mappings
   to be PAGE_SIZE, just like when dirty logging is enabled.

Signed-off-by: James Houghton <jthoughton@xxxxxxxxxx>
---
  I'm not 100% sure if kvm_arch_flush_shadow_memslot() is correct in
  this case (like if the host does not have S2FWB).
---
 arch/arm64/kvm/Kconfig |  1 +
 arch/arm64/kvm/mmu.c   | 23 ++++++++++++++++++++++-
 2 files changed, 23 insertions(+), 1 deletion(-)

diff --git a/arch/arm64/kvm/Kconfig b/arch/arm64/kvm/Kconfig
index ead632ad01b4..d89b4088b580 100644
--- a/arch/arm64/kvm/Kconfig
+++ b/arch/arm64/kvm/Kconfig
@@ -38,6 +38,7 @@ menuconfig KVM
 	select HAVE_KVM_VCPU_RUN_PID_CHANGE
 	select SCHED_INFO
 	select GUEST_PERF_EVENTS if PERF_EVENTS
+	select HAVE_KVM_USERFAULT
 	help
 	  Support hosting virtualized guest machines.
 
diff --git a/arch/arm64/kvm/mmu.c b/arch/arm64/kvm/mmu.c
index a71fe6f6bd90..53cee0bacb75 100644
--- a/arch/arm64/kvm/mmu.c
+++ b/arch/arm64/kvm/mmu.c
@@ -1482,7 +1482,7 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
 	 * logging_active is guaranteed to never be true for VM_PFNMAP
 	 * memslots.
 	 */
-	if (logging_active) {
+	if (logging_active || kvm_memslot_userfault(memslot)) {
 		force_pte = true;
 		vma_shift = PAGE_SHIFT;
 	} else {
@@ -1571,6 +1571,13 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
 	mmu_seq = vcpu->kvm->mmu_invalidate_seq;
 	mmap_read_unlock(current->mm);
 
+	if (kvm_gfn_userfault(kvm, memslot, gfn)) {
+		kvm_prepare_memory_fault_exit(vcpu, gfn << PAGE_SHIFT,
+					      PAGE_SIZE, write_fault,
+					      exec_fault, false, true);
+		return -EFAULT;
+	}
+
 	pfn = __kvm_faultin_pfn(memslot, gfn, write_fault ? FOLL_WRITE : 0,
 				&writable, &page);
 	if (pfn == KVM_PFN_ERR_HWPOISON) {
@@ -2062,6 +2069,20 @@ void kvm_arch_commit_memory_region(struct kvm *kvm,
 				   enum kvm_mr_change change)
 {
 	bool log_dirty_pages = new && new->flags & KVM_MEM_LOG_DIRTY_PAGES;
+	u32 changed_flags = (new ? new->flags : 0) ^ (old ? old->flags : 0);
+
+	/*
+	 * If KVM_MEM_USERFAULT changed, drop all the stage-2 mappings so that
+	 * we can (1) respect userfault-ness or (2) create block mappings.
+	 */
+	if ((changed_flags & KVM_MEM_USERFAULT) && change == KVM_MR_FLAGS_ONLY)
+		kvm_arch_flush_shadow_memslot(kvm, old);
+
+	/*
+	 * Nothing left to do if not toggling dirty logging.
+	 */
+	if (!(changed_flags & KVM_MEM_LOG_DIRTY_PAGES))
+		return;
 
 	/*
 	 * At this point memslot has been committed and there is an
-- 
2.47.0.338.g60cca15819-goog