From patchwork Fri Nov 9 14:20:17 2012 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Takashi Iwai X-Patchwork-Id: 1720851 Return-Path: X-Original-To: patchwork-kvm@patchwork.kernel.org Delivered-To: patchwork-process-083081@patchwork1.kernel.org Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by patchwork1.kernel.org (Postfix) with ESMTP id 22EDA3FCDF for ; Fri, 9 Nov 2012 14:23:22 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753624Ab2KIOUU (ORCPT ); Fri, 9 Nov 2012 09:20:20 -0500 Received: from cantor2.suse.de ([195.135.220.15]:46313 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753534Ab2KIOUT (ORCPT ); Fri, 9 Nov 2012 09:20:19 -0500 Received: from relay2.suse.de (unknown [195.135.220.254]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by mx2.suse.de (Postfix) with ESMTP id 97AA6912E4; Fri, 9 Nov 2012 15:20:17 +0100 (CET) Date: Fri, 09 Nov 2012 15:20:17 +0100 Message-ID: From: Takashi Iwai To: Avi Kivity Cc: Marcelo Tosatti , Junjie Mao , Fengguang Wu , kvm@vger.kernel.org, linux-kernel@vger.kernel.org Subject: [PATCH] KVM: x86: Fix invalid secondary exec controls in vmx_cpuid_update() User-Agent: Wanderlust/2.15.9 (Almost Unreal) SEMI/1.14.6 (Maruoka) FLIM/1.14.9 (=?UTF-8?B?R29qxY0=?=) APEL/10.8 Emacs/24.2 (x86_64-suse-linux-gnu) MULE/6.0 (HANACHIRUSATO) MIME-Version: 1.0 (generated by SEMI 1.14.6 - "Maruoka") Sender: kvm-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: kvm@vger.kernel.org The commit [ad756a16: KVM: VMX: Implement PCID/INVPCID for guests with EPT] introduced the unconditional access to SECONDARY_VM_EXEC_CONTROL, and this triggers kernel warnings like below on old CPUs: vmwrite error: reg 401e value a0568000 (err 12) Pid: 13649, comm: qemu-kvm Not tainted 3.7.0-rc4-test2+ #154 Call Trace: [] vmwrite_error+0x27/0x29 [kvm_intel] [] vmcs_writel+0x1b/0x20 [kvm_intel] [] vmx_cpuid_update+0x74/0x170 [kvm_intel] [] kvm_vcpu_ioctl_set_cpuid2+0x76/0x90 [kvm] [] kvm_arch_vcpu_ioctl+0xc37/0xed0 [kvm] [] ? __vunmap+0x9c/0x110 [] ? vmx_vcpu_load+0x39/0x1a0 [kvm_intel] [] ? kvm_arch_vcpu_load+0x52/0x1a0 [kvm] [] ? vcpu_load+0x74/0xd0 [kvm] [] kvm_vcpu_ioctl+0x110/0x5e0 [kvm] [] ? kvm_dev_ioctl+0x4d/0x4a0 [kvm] [] do_vfs_ioctl+0x8f/0x530 [] ? remove_vma+0x56/0x60 [] ? do_munmap+0x328/0x400 [] ? fget_light+0x4c/0x100 [] sys_ioctl+0x91/0xb0 [] system_call_fastpath+0x1a/0x1f This patch adds a check for the availability of secondary exec control to avoid these warnings. Cc: [v3.6+] Signed-off-by: Takashi Iwai --- arch/x86/kvm/vmx.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-) diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c index ad6b1dd..f858159 100644 --- a/arch/x86/kvm/vmx.c +++ b/arch/x86/kvm/vmx.c @@ -6549,19 +6549,22 @@ static void vmx_cpuid_update(struct kvm_vcpu *vcpu) } } - exec_control = vmcs_read32(SECONDARY_VM_EXEC_CONTROL); /* Exposing INVPCID only when PCID is exposed */ best = kvm_find_cpuid_entry(vcpu, 0x7, 0); if (vmx_invpcid_supported() && best && (best->ebx & bit(X86_FEATURE_INVPCID)) && guest_cpuid_has_pcid(vcpu)) { + exec_control = vmcs_read32(SECONDARY_VM_EXEC_CONTROL); exec_control |= SECONDARY_EXEC_ENABLE_INVPCID; vmcs_write32(SECONDARY_VM_EXEC_CONTROL, exec_control); } else { - exec_control &= ~SECONDARY_EXEC_ENABLE_INVPCID; - vmcs_write32(SECONDARY_VM_EXEC_CONTROL, - exec_control); + if (cpu_has_secondary_exec_ctrls()) { + exec_control = vmcs_read32(SECONDARY_VM_EXEC_CONTROL); + exec_control &= ~SECONDARY_EXEC_ENABLE_INVPCID; + vmcs_write32(SECONDARY_VM_EXEC_CONTROL, + exec_control); + } if (best) best->ebx &= ~bit(X86_FEATURE_INVPCID); }