KVM: PPC: Book3S PR: Better handling of host-side read-only pages

Currently we request write access to all pages that get mapped into the
guest, even if the guest is only loading from the page.  This reduces
the effectiveness of KSM because it means that we unshare every page we
access.  Also, we always set the changed (C) bit in the guest HPTE if
it allows writing, even for a guest load.

This fixes both these problems.  We pass an 'iswrite' flag to the
mmu.xlate() functions and to kvmppc_mmu_map_page() to indicate whether
the access is a load or a store.  The mmu.xlate() functions now only
set C for stores.  kvmppc_gfn_to_pfn() now calls gfn_to_pfn_prot()
instead of gfn_to_pfn() so that it can indicate whether we need write
access to the page, and get back a 'writable' flag to indicate whether
the page is writable or not.  If that 'writable' flag is clear, we then
make the host HPTE read-only even if the guest HPTE allowed writing.

This means that we can get a protection fault when the guest writes to a
page that it has mapped read-write but which is read-only on the host
side (perhaps due to KSM having merged the page).  Thus we now call
kvmppc_handle_pagefault() for protection faults as well as HPTE not found
faults.  In kvmppc_handle_pagefault(), if the access was allowed by the
guest HPTE and we thus need to install a new host HPTE, we then need to
remove the old host HPTE if there is one.  This is done with a new
function, kvmppc_mmu_unmap_page(), which uses kvmppc_mmu_pte_vflush() to
find and remove the old host HPTE.

Since the memslot-related functions require the KVM SRCU read lock to
be held, this adds srcu_read_lock/unlock pairs around the calls to
kvmppc_handle_pagefault().

Finally, this changes kvmppc_mmu_book3s_32_xlate_pte() to not ignore
guest HPTEs that don't permit access, and to return -EPERM for accesses
that are not permitted by the page protections.

Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>

1 files changed, 24 insertions, 5 deletions

diff --git a/arch/powerpc/kvm/book3s_pr.c b/arch/powerpc/kvm/book3s_pr.c
index 677d7e33b1ff..2f84ed807184 100644
--- a/arch/powerpc/kvm/book3s_pr.c
+++ b/arch/powerpc/kvm/book3s_pr.c
@@ -401,6 +401,7 @@ int kvmppc_handle_pagefault(struct kvm_run *run, struct kvm_vcpu *vcpu,
 			    ulong eaddr, int vec)
 {
 	bool data = (vec == BOOK3S_INTERRUPT_DATA_STORAGE);
+	bool iswrite = false;
 	int r = RESUME_GUEST;
 	int relocated;
 	int page_found = 0;
@@ -411,10 +412,12 @@ int kvmppc_handle_pagefault(struct kvm_run *run, struct kvm_vcpu *vcpu,
 	u64 vsid;
 
 	relocated = data ? dr : ir;
+	if (data && (vcpu->arch.fault_dsisr & DSISR_ISSTORE))
+		iswrite = true;
 
 	/* Resolve real address if translation turned on */
 	if (relocated) {
-		page_found = vcpu->arch.mmu.xlate(vcpu, eaddr, &pte, data);
+		page_found = vcpu->arch.mmu.xlate(vcpu, eaddr, &pte, data, iswrite);
 	} else {
 		pte.may_execute = true;
 		pte.may_read = true;
@@ -475,12 +478,20 @@ int kvmppc_handle_pagefault(struct kvm_run *run, struct kvm_vcpu *vcpu,
 		kvmppc_book3s_queue_irqprio(vcpu, vec + 0x80);
 	} else if (!is_mmio &&
 		   kvmppc_visible_gfn(vcpu, pte.raddr >> PAGE_SHIFT)) {
+		if (data && !(vcpu->arch.fault_dsisr & DSISR_NOHPTE)) {
+			/*
+			 * There is already a host HPTE there, presumably
+			 * a read-only one for a page the guest thinks
+			 * is writable, so get rid of it first.
+			 */
+			kvmppc_mmu_unmap_page(vcpu, &pte);
+		}
 		/* The guest's PTE is not mapped yet. Map on the host */
-		kvmppc_mmu_map_page(vcpu, &pte);
+		kvmppc_mmu_map_page(vcpu, &pte, iswrite);
 		if (data)
 			vcpu->stat.sp_storage++;
 		else if (vcpu->arch.mmu.is_dcbz32(vcpu) &&
-			(!(vcpu->arch.hflags & BOOK3S_HFLAG_DCBZ32)))
+			 (!(vcpu->arch.hflags & BOOK3S_HFLAG_DCBZ32)))
 			kvmppc_patch_dcbz(vcpu, &pte);
 	} else {
 		/* MMIO */
@@ -732,7 +743,9 @@ int kvmppc_handle_exit(struct kvm_run *run, struct kvm_vcpu *vcpu,
 
 		/* only care about PTEG not found errors, but leave NX alone */
 		if (shadow_srr1 & 0x40000000) {
+			int idx = srcu_read_lock(&vcpu->kvm->srcu);
 			r = kvmppc_handle_pagefault(run, vcpu, kvmppc_get_pc(vcpu), exit_nr);
+			srcu_read_unlock(&vcpu->kvm->srcu, idx);
 			vcpu->stat.sp_instruc++;
 		} else if (vcpu->arch.mmu.is_dcbz32(vcpu) &&
 			  (!(vcpu->arch.hflags & BOOK3S_HFLAG_DCBZ32))) {
@@ -774,9 +787,15 @@ int kvmppc_handle_exit(struct kvm_run *run, struct kvm_vcpu *vcpu,
 		}
 #endif
 
-		/* The only case we need to handle is missing shadow PTEs */
-		if (fault_dsisr & DSISR_NOHPTE) {
+		/*
+		 * We need to handle missing shadow PTEs, and
+		 * protection faults due to us mapping a page read-only
+		 * when the guest thinks it is writable.
+		 */
+		if (fault_dsisr & (DSISR_NOHPTE | DSISR_PROTFAULT)) {
+			int idx = srcu_read_lock(&vcpu->kvm->srcu);
 			r = kvmppc_handle_pagefault(run, vcpu, dar, exit_nr);
+			srcu_read_unlock(&vcpu->kvm->srcu, idx);
 		} else {
 			vcpu->arch.shared->dar = dar;
 			vcpu->arch.shared->dsisr = fault_dsisr;