From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jan Kiszka Subject: [PATCH uq/master] kvm: x86: Separately write feature control MSR on reset Date: Tue, 17 Dec 2013 20:05:13 +0100 Message-ID: <52B0A069.8060807@siemens.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Cc: kvm , qemu-devel To: Paolo Bonzini , Gleb Natapov Return-path: Received: from thoth.sbs.de ([192.35.17.2]:34844 "EHLO thoth.sbs.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752012Ab3LQTF0 (ORCPT ); Tue, 17 Dec 2013 14:05:26 -0500 Sender: kvm-owner@vger.kernel.org List-ID: If the guest is running in nested mode on system reset, clearing the feature MSR signals the kernel to leave this mode. Recent kernels processes this properly, but leave the VCPU state undefined behind. It is the job of userspace to bring it to a proper shape. Therefore, write this specific MSR first so that no state transfer gets lost. This allows to cleanly reset a guest with VMX in use. Signed-off-by: Jan Kiszka --- target-i386/kvm.c | 32 ++++++++++++++++++++++++++++---- 1 file changed, 28 insertions(+), 4 deletions(-) diff --git a/target-i386/kvm.c b/target-i386/kvm.c index 1188482..ec51447 100644 --- a/target-i386/kvm.c +++ b/target-i386/kvm.c @@ -1104,6 +1104,25 @@ static int kvm_put_tscdeadline_msr(X86CPU *cpu) return kvm_vcpu_ioctl(CPU(cpu), KVM_SET_MSRS, &msr_data); } +/* + * Provide a separate write service for the feature control MSR in order to + * kick the VCPU out of VMXON or even guest mode on reset. This has to be done + * before writing any other state because forcibly leaving nested mode + * invalidates the VCPU state. + */ +static int kvm_put_msr_feature_control(X86CPU *cpu) +{ + struct { + struct kvm_msrs info; + struct kvm_msr_entry entry; + } msr_data; + + kvm_msr_entry_set(&msr_data.entry, MSR_IA32_FEATURE_CONTROL, + cpu->env.msr_ia32_feature_control); + msr_data.info.nmsrs = 1; + return kvm_vcpu_ioctl(CPU(cpu), KVM_SET_MSRS, &msr_data); +} + static int kvm_put_msrs(X86CPU *cpu, int level) { CPUX86State *env = &cpu->env; @@ -1204,10 +1223,8 @@ static int kvm_put_msrs(X86CPU *cpu, int level) if (cpu->hyperv_vapic) { kvm_msr_entry_set(&msrs[n++], HV_X64_MSR_APIC_ASSIST_PAGE, 0); } - if (has_msr_feature_control) { - kvm_msr_entry_set(&msrs[n++], MSR_IA32_FEATURE_CONTROL, - env->msr_ia32_feature_control); - } + /* Note: MSR_IA32_FEATURE_CONTROL is written separately, see + * kvm_put_msr_feature_control. */ } if (env->mcg_cap) { int i; @@ -1801,6 +1818,13 @@ int kvm_arch_put_registers(CPUState *cpu, int level) assert(cpu_is_stopped(cpu) || qemu_cpu_is_self(cpu)); + if (level >= KVM_PUT_RESET_STATE && has_msr_feature_control) { + ret = kvm_put_msr_feature_control(x86_cpu); + if (ret < 0) { + return ret; + } + } + ret = kvm_getput_regs(x86_cpu, 1); if (ret < 0) { return ret; -- 1.8.1.1.298.ge7eed54 From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:37701) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1VszxP-00011t-Pn for qemu-devel@nongnu.org; Tue, 17 Dec 2013 14:05:33 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1VszxJ-0005Iq-WA for qemu-devel@nongnu.org; Tue, 17 Dec 2013 14:05:27 -0500 Received: from thoth.sbs.de ([192.35.17.2]:33311) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1VszxJ-0005Ih-Lw for qemu-devel@nongnu.org; Tue, 17 Dec 2013 14:05:21 -0500 Message-ID: <52B0A069.8060807@siemens.com> Date: Tue, 17 Dec 2013 20:05:13 +0100 From: Jan Kiszka MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Subject: [Qemu-devel] [PATCH uq/master] kvm: x86: Separately write feature control MSR on reset List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Paolo Bonzini , Gleb Natapov Cc: qemu-devel , kvm If the guest is running in nested mode on system reset, clearing the feature MSR signals the kernel to leave this mode. Recent kernels processes this properly, but leave the VCPU state undefined behind. It is the job of userspace to bring it to a proper shape. Therefore, write this specific MSR first so that no state transfer gets lost. This allows to cleanly reset a guest with VMX in use. Signed-off-by: Jan Kiszka --- target-i386/kvm.c | 32 ++++++++++++++++++++++++++++---- 1 file changed, 28 insertions(+), 4 deletions(-) diff --git a/target-i386/kvm.c b/target-i386/kvm.c index 1188482..ec51447 100644 --- a/target-i386/kvm.c +++ b/target-i386/kvm.c @@ -1104,6 +1104,25 @@ static int kvm_put_tscdeadline_msr(X86CPU *cpu) return kvm_vcpu_ioctl(CPU(cpu), KVM_SET_MSRS, &msr_data); } +/* + * Provide a separate write service for the feature control MSR in order to + * kick the VCPU out of VMXON or even guest mode on reset. This has to be done + * before writing any other state because forcibly leaving nested mode + * invalidates the VCPU state. + */ +static int kvm_put_msr_feature_control(X86CPU *cpu) +{ + struct { + struct kvm_msrs info; + struct kvm_msr_entry entry; + } msr_data; + + kvm_msr_entry_set(&msr_data.entry, MSR_IA32_FEATURE_CONTROL, + cpu->env.msr_ia32_feature_control); + msr_data.info.nmsrs = 1; + return kvm_vcpu_ioctl(CPU(cpu), KVM_SET_MSRS, &msr_data); +} + static int kvm_put_msrs(X86CPU *cpu, int level) { CPUX86State *env = &cpu->env; @@ -1204,10 +1223,8 @@ static int kvm_put_msrs(X86CPU *cpu, int level) if (cpu->hyperv_vapic) { kvm_msr_entry_set(&msrs[n++], HV_X64_MSR_APIC_ASSIST_PAGE, 0); } - if (has_msr_feature_control) { - kvm_msr_entry_set(&msrs[n++], MSR_IA32_FEATURE_CONTROL, - env->msr_ia32_feature_control); - } + /* Note: MSR_IA32_FEATURE_CONTROL is written separately, see + * kvm_put_msr_feature_control. */ } if (env->mcg_cap) { int i; @@ -1801,6 +1818,13 @@ int kvm_arch_put_registers(CPUState *cpu, int level) assert(cpu_is_stopped(cpu) || qemu_cpu_is_self(cpu)); + if (level >= KVM_PUT_RESET_STATE && has_msr_feature_control) { + ret = kvm_put_msr_feature_control(x86_cpu); + if (ret < 0) { + return ret; + } + } + ret = kvm_getput_regs(x86_cpu, 1); if (ret < 0) { return ret; -- 1.8.1.1.298.ge7eed54