From mboxrd@z Thu Jan 1 00:00:00 1970 From: Brijesh Singh Subject: [Part2 PATCH v4 28/29] KVM: X86: Restart the guest when insn_len is zero and SEV is enabled Date: Tue, 19 Sep 2017 15:46:26 -0500 Message-ID: <20170919204627.3875-29-brijesh.singh@amd.com> References: <20170919204627.3875-1-brijesh.singh@amd.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Cc: Brijesh Singh , Thomas Gleixner , Ingo Molnar , "H. Peter Anvin" , Paolo Bonzini , =?UTF-8?q?Radim=20Kr=C4=8Dm=C3=A1=C5=99?= , Joerg Roedel , Borislav Petkov , Tom Lendacky , x86@kernel.org To: linux-kernel@vger.kernel.org, kvm@vger.kernel.org Return-path: In-Reply-To: <20170919204627.3875-1-brijesh.singh@amd.com> Sender: linux-kernel-owner@vger.kernel.org List-Id: kvm.vger.kernel.org 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 Cc: Ingo Molnar Cc: "H. Peter Anvin" Cc: Paolo Bonzini Cc: "Radim Krčmář" Cc: Joerg Roedel Cc: Borislav Petkov Cc: Tom Lendacky Cc: x86@kernel.org Cc: kvm@vger.kernel.org Cc: linux-kernel@vger.kernel.org Signed-off-by: Brijesh Singh --- arch/x86/kvm/mmu.c | 17 +++++++++++++++++ 1 file changed, 17 insertions(+) diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c index eca30c1eb1d9..2bc0fe84aca2 100644 --- a/arch/x86/kvm/mmu.c +++ b/arch/x86/kvm/mmu.c @@ -4953,6 +4953,23 @@ int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t cr2, u64 error_code, 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). + * + * 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. + */ + if (unlikely(!insn_len) && + kvm_x86_ops->mem_enc_enabled && + kvm_x86_ops->mem_enc_enabled(vcpu)) + return 1; + er = x86_emulate_instruction(vcpu, cr2, emulation_type, insn, insn_len); switch (er) { -- 2.9.5