From: Brijesh Singh <brijesh.singh@amd.com>
Date: Mon, 4 Dec 2017 10:57:40 -0600
Subject: KVM: X86: Restart the guest when insn_len is zero and SEV is enabled
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Git-commit: 00b10fe1046c4b2232097a7ffaa9238c7e479388
Patch-mainline: v4.16-rc1
References: bsc#1063068

On AMD platforms, under certain conditions insn_len may be zero on #NPF.
This can happen if a guest gets a page-fault on data access but the HW
table walker is not able to read the instruction page (e.g instruction
page is not present in memory).

Typically, when insn_len is zero, x86_emulate_instruction() walks the
guest page table and fetches the instruction bytes from guest memory.
When SEV is enabled, the guest memory is encrypted with guest-specific
key hence hypervisor will not able to fetch the instruction bytes.
In those cases we simply restart the guest.

I have encountered this issue when running kernbench inside the guest.

Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: "Radim Krčmář" <rkrcmar@redhat.com>
Cc: Joerg Roedel <joro@8bytes.org>
Cc: Borislav Petkov <bp@suse.de>
Cc: Tom Lendacky <thomas.lendacky@amd.com>
Cc: x86@kernel.org
Cc: kvm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Brijesh Singh <brijesh.singh@amd.com>
Acked-by: Borislav Petkov <bp@suse.de>
---
 arch/x86/kvm/mmu.c |   10 ++++++++++
 arch/x86/kvm/svm.c |    6 ++++--
 2 files changed, 14 insertions(+), 2 deletions(-)

--- a/arch/x86/kvm/mmu.c
+++ b/arch/x86/kvm/mmu.c
@@ -4951,6 +4951,16 @@ int kvm_mmu_page_fault(struct kvm_vcpu *
 	if (mmio_info_in_cache(vcpu, cr2, direct))
 		emulation_type = 0;
 emulate:
+	/*
+	 * On AMD platforms, under certain conditions insn_len may be zero on #NPF.
+	 * This can happen if a guest gets a page-fault on data access but the HW
+	 * table walker is not able to read the instruction page (e.g instruction
+	 * page is not present in memory). In those cases we simply restart the
+	 * guest.
+	 */
+	if (unlikely(insn && !insn_len))
+		return 1;
+
 	er = x86_emulate_instruction(vcpu, cr2, emulation_type, insn, insn_len);
 
 	switch (er) {
--- a/arch/x86/kvm/svm.c
+++ b/arch/x86/kvm/svm.c
@@ -2435,7 +2435,8 @@ static int pf_interception(struct vcpu_s
 	u64 error_code = svm->vmcb->control.exit_info_1;
 
 	return kvm_handle_page_fault(&svm->vcpu, error_code, fault_address,
-			svm->vmcb->control.insn_bytes,
+			static_cpu_has(X86_FEATURE_DECODEASSISTS) ?
+			svm->vmcb->control.insn_bytes : NULL,
 			svm->vmcb->control.insn_len);
 }
 
@@ -2446,7 +2447,8 @@ static int npf_interception(struct vcpu_
 
 	trace_kvm_page_fault(fault_address, error_code);
 	return kvm_mmu_page_fault(&svm->vcpu, fault_address, error_code,
-			svm->vmcb->control.insn_bytes,
+			static_cpu_has(X86_FEATURE_DECODEASSISTS) ?
+			svm->vmcb->control.insn_bytes : NULL,
 			svm->vmcb->control.insn_len);
 }