Re: [PATCH V5 2/2] Enabling IA32_TSC_ADJUST for KVM guest VM support
On Thu, Nov 29, 2012 at 07:21:28PM +, Auld, Will wrote: Marcelo, The behavior on reset is to return the TSC_AJUST msr value to 0x0. I am currently initializing this emulated msr in kvm_arch_vcpu_init(). Will, Reset is handled by QEMU. kvm_arch_vcpu_init is only called during vcpu allocation (ie vm creation). See x86_cpu_reset in QEMU's target-i386/cpu.c file. - Behaviour on reset: what is the behaviour on RESET? I am testing the rebase now. I would like to get any needed changes for this initialization into my next patch set version (v6) if possible. For the kernel patch, everything is fine. Apparently for the QEMU patch it is necessary to set value to zero on reset. -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
RE: [PATCH V5 2/2] Enabling IA32_TSC_ADJUST for KVM guest VM support
Marcelo, My latest QEMU patch ([PATCH V5] target-i386: Enabling IA32_TSC_ADJUST for QEMU KVM guest VMs) seems to be OK with regards to reset. The tsc_adjust variable is being zeroed by the memset() in x86_cpu_reset(). Later code seems to write these values through kvm_put_msr() by way of cpu_synchronize_all_post_reset(). Andreas said: I'm expecting this to go through Marcello's queue unless I'm told otherwise. Does this meet your expectations as well? Thanks, Will -Original Message- From: Marcelo Tosatti [mailto:mtosa...@redhat.com] Sent: Thursday, November 29, 2012 12:44 PM To: Auld, Will Cc: kvm@vger.kernel.org; Dugger, Donald D; Liu, Jinsong; Zhang, Xiantao; a...@redhat.com; qemu-devel; Gleb Subject: Re: [PATCH V5 2/2] Enabling IA32_TSC_ADJUST for KVM guest VM support On Thu, Nov 29, 2012 at 07:21:28PM +, Auld, Will wrote: Marcelo, The behavior on reset is to return the TSC_AJUST msr value to 0x0. I am currently initializing this emulated msr in kvm_arch_vcpu_init(). Will, Reset is handled by QEMU. kvm_arch_vcpu_init is only called during vcpu allocation (ie vm creation). See x86_cpu_reset in QEMU's target- i386/cpu.c file. - Behaviour on reset: what is the behaviour on RESET? I am testing the rebase now. I would like to get any needed changes for this initialization into my next patch set version (v6) if possible. For the kernel patch, everything is fine. Apparently for the QEMU patch it is necessary to set value to zero on reset. -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
[PATCH V5 2/2] Enabling IA32_TSC_ADJUST for KVM guest VM support
CPUID.7.0.EBX[1]=1 indicates IA32_TSC_ADJUST MSR 0x3b is supported Basic design is to emulate the MSR by allowing reads and writes to a guest vcpu specific location to store the value of the emulated MSR while adding the value to the vmcs tsc_offset. In this way the IA32_TSC_ADJUST value will be included in all reads to the TSC MSR whether through rdmsr or rdtsc. This is of course as long as the use TSC counter offsetting VM-execution control is enabled as well as the IA32_TSC_ADJUST control. However, because hardware will only return the TSC + IA32_TSC_ADJUST + vmsc tsc_offset for a guest process when it does and rdtsc (with the correct settings) the value of our virtualized IA32_TSC_ADJUST must be stored in one of these three locations. The argument against storing it in the actual MSR is performance. This is likely to be seldom used while the save/restore is required on every transition. IA32_TSC_ADJUST was created as a way to solve some issues with writing TSC itself so that is not an option either. The remaining option, defined above as our solution has the problem of returning incorrect vmcs tsc_offset values (unless we intercept and fix, not done here) as mentioned above. However, more problematic is that storing the data in vmcs tsc_offset will have a different semantic effect on the system than does using the actual MSR. This is illustrated in the following example: The hypervisor set the IA32_TSC_ADJUST, then the guest sets it and a guest process performs a rdtsc. In this case the guest process will get TSC + IA32_TSC_ADJUST_hyperviser + vmsc tsc_offset including IA32_TSC_ADJUST_guest. While the total system semantics changed the semantics as seen by the guest do not and hence this will not cause a problem. Signed-off-by: Will Auld will.a...@intel.com --- arch/x86/include/asm/cpufeature.h | 1 + arch/x86/include/asm/kvm_host.h | 3 +++ arch/x86/include/asm/msr-index.h | 1 + arch/x86/kvm/cpuid.c | 2 ++ arch/x86/kvm/cpuid.h | 8 arch/x86/kvm/svm.c| 7 +++ arch/x86/kvm/vmx.c| 9 + arch/x86/kvm/x86.c| 22 ++ 8 files changed, 53 insertions(+) diff --git a/arch/x86/include/asm/cpufeature.h b/arch/x86/include/asm/cpufeature.h index 6b7ee5f..e574d81 100644 --- a/arch/x86/include/asm/cpufeature.h +++ b/arch/x86/include/asm/cpufeature.h @@ -199,6 +199,7 @@ /* Intel-defined CPU features, CPUID level 0x0007:0 (ebx), word 9 */ #define X86_FEATURE_FSGSBASE (9*32+ 0) /* {RD/WR}{FS/GS}BASE instructions*/ +#define X86_FEATURE_TSC_ADJUST (9*32+ 1) /* TSC adjustment MSR 0x3b */ #define X86_FEATURE_BMI1 (9*32+ 3) /* 1st group bit manipulation extensions */ #define X86_FEATURE_HLE(9*32+ 4) /* Hardware Lock Elision */ #define X86_FEATURE_AVX2 (9*32+ 5) /* AVX2 instructions */ diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h index da34027..cf8c7e0 100644 --- a/arch/x86/include/asm/kvm_host.h +++ b/arch/x86/include/asm/kvm_host.h @@ -442,6 +442,8 @@ struct kvm_vcpu_arch { u32 virtual_tsc_mult; u32 virtual_tsc_khz; + s64 ia32_tsc_adjust_msr; + atomic_t nmi_queued; /* unprocessed asynchronous NMIs */ unsigned nmi_pending; /* NMI queued after currently running handler */ bool nmi_injected;/* Trying to inject an NMI this entry */ @@ -690,6 +692,7 @@ struct kvm_x86_ops { bool (*has_wbinvd_exit)(void); void (*set_tsc_khz)(struct kvm_vcpu *vcpu, u32 user_tsc_khz, bool scale); + u64 (*read_tsc_offset)(struct kvm_vcpu *vcpu); void (*write_tsc_offset)(struct kvm_vcpu *vcpu, u64 offset); u64 (*compute_tsc_offset)(struct kvm_vcpu *vcpu, u64 target_tsc); diff --git a/arch/x86/include/asm/msr-index.h b/arch/x86/include/asm/msr-index.h index 957ec87..6486569 100644 --- a/arch/x86/include/asm/msr-index.h +++ b/arch/x86/include/asm/msr-index.h @@ -231,6 +231,7 @@ #define MSR_IA32_EBL_CR_POWERON0x002a #define MSR_EBC_FREQUENCY_ID 0x002c #define MSR_IA32_FEATURE_CONTROL0x003a +#define MSR_IA32_TSC_ADJUST 0x003b #define FEATURE_CONTROL_LOCKED (10) #define FEATURE_CONTROL_VMXON_ENABLED_INSIDE_SMX (11) diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c index 0595f13..e817bac 100644 --- a/arch/x86/kvm/cpuid.c +++ b/arch/x86/kvm/cpuid.c @@ -320,6 +320,8 @@ static int do_cpuid_ent(struct kvm_cpuid_entry2 *entry, u32 function, if (index == 0) { entry-ebx = kvm_supported_word9_x86_features; cpuid_mask(entry-ebx, 9); + // TSC_ADJUST is emulated + entry-ebx |= F(TSC_ADJUST); } else entry-ebx = 0; entry-eax = 0; diff --git a/arch/x86/kvm/cpuid.h b/arch/x86/kvm/cpuid.h index
Re: [PATCH V5 2/2] Enabling IA32_TSC_ADJUST for KVM guest VM support
Hi Will, On Tue, Nov 27, 2012 at 11:09:00AM -0800, Will Auld wrote: CPUID.7.0.EBX[1]=1 indicates IA32_TSC_ADJUST MSR 0x3b is supported Basic design is to emulate the MSR by allowing reads and writes to a guest vcpu specific location to store the value of the emulated MSR while adding the value to the vmcs tsc_offset. In this way the IA32_TSC_ADJUST value will be included in all reads to the TSC MSR whether through rdmsr or rdtsc. This is of course as long as the use TSC counter offsetting VM-execution control is enabled as well as the IA32_TSC_ADJUST control. However, because hardware will only return the TSC + IA32_TSC_ADJUST + vmsc tsc_offset for a guest process when it does and rdtsc (with the correct settings) the value of our virtualized IA32_TSC_ADJUST must be stored in one of these three locations. The argument against storing it in the actual MSR is performance. This is likely to be seldom used while the save/restore is required on every transition. IA32_TSC_ADJUST was created as a way to solve some issues with writing TSC itself so that is not an option either. The remaining option, defined above as our solution has the problem of returning incorrect vmcs tsc_offset values (unless we intercept and fix, not done here) as mentioned above. However, more problematic is that storing the data in vmcs tsc_offset will have a different semantic effect on the system than does using the actual MSR. This is illustrated in the following example: The hypervisor set the IA32_TSC_ADJUST, then the guest sets it and a guest process performs a rdtsc. In this case the guest process will get TSC + IA32_TSC_ADJUST_hyperviser + vmsc tsc_offset including IA32_TSC_ADJUST_guest. While the total system semantics changed the semantics as seen by the guest do not and hence this will not cause a problem. Signed-off-by: Will Auld will.a...@intel.com --- arch/x86/include/asm/cpufeature.h | 1 + arch/x86/include/asm/kvm_host.h | 3 +++ arch/x86/include/asm/msr-index.h | 1 + arch/x86/kvm/cpuid.c | 2 ++ arch/x86/kvm/cpuid.h | 8 arch/x86/kvm/svm.c| 7 +++ arch/x86/kvm/vmx.c| 9 + arch/x86/kvm/x86.c| 22 ++ 8 files changed, 53 insertions(+) diff --git a/arch/x86/include/asm/cpufeature.h b/arch/x86/include/asm/cpufeature.h index 6b7ee5f..e574d81 100644 --- a/arch/x86/include/asm/cpufeature.h +++ b/arch/x86/include/asm/cpufeature.h @@ -199,6 +199,7 @@ /* Intel-defined CPU features, CPUID level 0x0007:0 (ebx), word 9 */ #define X86_FEATURE_FSGSBASE (9*32+ 0) /* {RD/WR}{FS/GS}BASE instructions*/ +#define X86_FEATURE_TSC_ADJUST (9*32+ 1) /* TSC adjustment MSR 0x3b */ #define X86_FEATURE_BMI1 (9*32+ 3) /* 1st group bit manipulation extensions */ #define X86_FEATURE_HLE (9*32+ 4) /* Hardware Lock Elision */ #define X86_FEATURE_AVX2 (9*32+ 5) /* AVX2 instructions */ diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h index da34027..cf8c7e0 100644 --- a/arch/x86/include/asm/kvm_host.h +++ b/arch/x86/include/asm/kvm_host.h @@ -442,6 +442,8 @@ struct kvm_vcpu_arch { u32 virtual_tsc_mult; u32 virtual_tsc_khz; + s64 ia32_tsc_adjust_msr; + atomic_t nmi_queued; /* unprocessed asynchronous NMIs */ unsigned nmi_pending; /* NMI queued after currently running handler */ bool nmi_injected;/* Trying to inject an NMI this entry */ @@ -690,6 +692,7 @@ struct kvm_x86_ops { bool (*has_wbinvd_exit)(void); void (*set_tsc_khz)(struct kvm_vcpu *vcpu, u32 user_tsc_khz, bool scale); + u64 (*read_tsc_offset)(struct kvm_vcpu *vcpu); void (*write_tsc_offset)(struct kvm_vcpu *vcpu, u64 offset); u64 (*compute_tsc_offset)(struct kvm_vcpu *vcpu, u64 target_tsc); diff --git a/arch/x86/include/asm/msr-index.h b/arch/x86/include/asm/msr-index.h index 957ec87..6486569 100644 --- a/arch/x86/include/asm/msr-index.h +++ b/arch/x86/include/asm/msr-index.h @@ -231,6 +231,7 @@ #define MSR_IA32_EBL_CR_POWERON 0x002a #define MSR_EBC_FREQUENCY_ID 0x002c #define MSR_IA32_FEATURE_CONTROL0x003a +#define MSR_IA32_TSC_ADJUST 0x003b #define FEATURE_CONTROL_LOCKED (10) #define FEATURE_CONTROL_VMXON_ENABLED_INSIDE_SMX (11) diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c index 0595f13..e817bac 100644 --- a/arch/x86/kvm/cpuid.c +++ b/arch/x86/kvm/cpuid.c @@ -320,6 +320,8 @@ static int do_cpuid_ent(struct kvm_cpuid_entry2 *entry, u32 function, if (index == 0) { entry-ebx = kvm_supported_word9_x86_features; cpuid_mask(entry-ebx, 9); + // TSC_ADJUST is emulated + entry-ebx |= F(TSC_ADJUST); } else
RE: [PATCH V5 2/2] Enabling IA32_TSC_ADJUST for KVM guest VM support
Thanks Marcelo, I'll address these items. - Please rebase against queue branch on kvm.git. I am not sure how to do this. The repository I have been working against is: git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git I assume that I need to change this in some way but I am not sure what I need to do. Will you explain or give me the needed path? Thanks, Will -Original Message- From: Marcelo Tosatti [mailto:mtosa...@redhat.com] Sent: Tuesday, November 27, 2012 5:00 PM To: Auld, Will Cc: kvm@vger.kernel.org; Dugger, Donald D; Liu, Jinsong; Zhang, Xiantao; a...@redhat.com; qemu-devel; Gleb Subject: Re: [PATCH V5 2/2] Enabling IA32_TSC_ADJUST for KVM guest VM support Hi Will, On Tue, Nov 27, 2012 at 11:09:00AM -0800, Will Auld wrote: CPUID.7.0.EBX[1]=1 indicates IA32_TSC_ADJUST MSR 0x3b is supported Basic design is to emulate the MSR by allowing reads and writes to a guest vcpu specific location to store the value of the emulated MSR while adding the value to the vmcs tsc_offset. In this way the IA32_TSC_ADJUST value will be included in all reads to the TSC MSR whether through rdmsr or rdtsc. This is of course as long as the use TSC counter offsetting VM-execution control is enabled as well as the IA32_TSC_ADJUST control. However, because hardware will only return the TSC + IA32_TSC_ADJUST + vmsc tsc_offset for a guest process when it does and rdtsc (with the correct settings) the value of our virtualized IA32_TSC_ADJUST must be stored in one of these three locations. The argument against storing it in the actual MSR is performance. This is likely to be seldom used while the save/restore is required on every transition. IA32_TSC_ADJUST was created as a way to solve some issues with writing TSC itself so that is not an option either. The remaining option, defined above as our solution has the problem of returning incorrect vmcs tsc_offset values (unless we intercept and fix, not done here) as mentioned above. However, more problematic is that storing the data in vmcs tsc_offset will have a different semantic effect on the system than does using the actual MSR. This is illustrated in the following example: The hypervisor set the IA32_TSC_ADJUST, then the guest sets it and a guest process performs a rdtsc. In this case the guest process will get TSC + IA32_TSC_ADJUST_hyperviser + vmsc tsc_offset including IA32_TSC_ADJUST_guest. While the total system semantics changed the semantics as seen by the guest do not and hence this will not cause a problem. Signed-off-by: Will Auld will.a...@intel.com --- arch/x86/include/asm/cpufeature.h | 1 + arch/x86/include/asm/kvm_host.h | 3 +++ arch/x86/include/asm/msr-index.h | 1 + arch/x86/kvm/cpuid.c | 2 ++ arch/x86/kvm/cpuid.h | 8 arch/x86/kvm/svm.c| 7 +++ arch/x86/kvm/vmx.c| 9 + arch/x86/kvm/x86.c| 22 ++ 8 files changed, 53 insertions(+) diff --git a/arch/x86/include/asm/cpufeature.h b/arch/x86/include/asm/cpufeature.h index 6b7ee5f..e574d81 100644 --- a/arch/x86/include/asm/cpufeature.h +++ b/arch/x86/include/asm/cpufeature.h @@ -199,6 +199,7 @@ /* Intel-defined CPU features, CPUID level 0x0007:0 (ebx), word 9 */ #define X86_FEATURE_FSGSBASE (9*32+ 0) /* {RD/WR}{FS/GS}BASE instructions*/ +#define X86_FEATURE_TSC_ADJUST (9*32+ 1) /* TSC adjustment MSR 0x3b +*/ #define X86_FEATURE_BMI1 (9*32+ 3) /* 1st group bit manipulation extensions */ #define X86_FEATURE_HLE(9*32+ 4) /* Hardware Lock Elision */ #define X86_FEATURE_AVX2 (9*32+ 5) /* AVX2 instructions */ diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h index da34027..cf8c7e0 100644 --- a/arch/x86/include/asm/kvm_host.h +++ b/arch/x86/include/asm/kvm_host.h @@ -442,6 +442,8 @@ struct kvm_vcpu_arch { u32 virtual_tsc_mult; u32 virtual_tsc_khz; + s64 ia32_tsc_adjust_msr; + atomic_t nmi_queued; /* unprocessed asynchronous NMIs */ unsigned nmi_pending; /* NMI queued after currently running handler */ bool nmi_injected;/* Trying to inject an NMI this entry */ @@ -690,6 +692,7 @@ struct kvm_x86_ops { bool (*has_wbinvd_exit)(void); void (*set_tsc_khz)(struct kvm_vcpu *vcpu, u32 user_tsc_khz, bool scale); + u64 (*read_tsc_offset)(struct kvm_vcpu *vcpu); void (*write_tsc_offset)(struct kvm_vcpu *vcpu, u64 offset); u64 (*compute_tsc_offset)(struct kvm_vcpu *vcpu, u64 target_tsc); diff --git a/arch/x86/include/asm/msr-index.h b/arch/x86/include/asm/msr-index.h index 957ec87..6486569 100644 --- a/arch/x86/include/asm/msr-index.h +++ b/arch/x86/include/asm/msr-index.h @@ -231,6 +231,7 @@ #define MSR_IA32_EBL_CR_POWERON0x002a #define MSR_EBC_FREQUENCY_ID