| Message ID | 20200129234640.8147-5-sean.j.christopherson@intel.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | None | expand |
Sean Christopherson <sean.j.christopherson@intel.com> writes: > Add a hook, ->has_virtualized_msr(), to allow moving vendor specific > checks into SVM/VMX and ultimately facilitate the removal of the > piecemeal ->*_supported() hooks. > > No functional change intended. > > Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com> > --- > arch/x86/include/asm/kvm_host.h | 1 + > arch/x86/kvm/svm.c | 6 ++++++ > arch/x86/kvm/vmx/vmx.c | 6 ++++++ > arch/x86/kvm/x86.c | 2 ++ > 4 files changed, 15 insertions(+) > > diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h > index 5c2ad3fa0980..8fb32c27fa44 100644 > --- a/arch/x86/include/asm/kvm_host.h > +++ b/arch/x86/include/asm/kvm_host.h > @@ -1050,6 +1050,7 @@ struct kvm_x86_ops { > int (*hardware_setup)(void); /* __init */ > void (*hardware_unsetup)(void); /* __exit */ > bool (*cpu_has_accelerated_tpr)(void); > + bool (*has_virtualized_msr)(u32 index); > bool (*has_emulated_msr)(u32 index); > void (*cpuid_update)(struct kvm_vcpu *vcpu); > > diff --git a/arch/x86/kvm/svm.c b/arch/x86/kvm/svm.c > index a7b944a3a0e2..1f9323fbad81 100644 > --- a/arch/x86/kvm/svm.c > +++ b/arch/x86/kvm/svm.c > @@ -5985,6 +5985,11 @@ static bool svm_cpu_has_accelerated_tpr(void) > return false; > } > > +static bool svm_has_virtualized_msr(u32 index) > +{ > + return true; > +} > + > static bool svm_has_emulated_msr(u32 index) > { > switch (index) { > @@ -7379,6 +7384,7 @@ static struct kvm_x86_ops svm_x86_ops __ro_after_init = { > .hardware_enable = svm_hardware_enable, > .hardware_disable = svm_hardware_disable, > .cpu_has_accelerated_tpr = svm_cpu_has_accelerated_tpr, > + .has_virtualized_msr = svm_has_virtualized_msr, > .has_emulated_msr = svm_has_emulated_msr, > > .vcpu_create = svm_create_vcpu, > diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c > index f5bb1ad2e9fa..3f2c094434e8 100644 > --- a/arch/x86/kvm/vmx/vmx.c > +++ b/arch/x86/kvm/vmx/vmx.c > @@ -6274,6 +6274,11 @@ static void vmx_handle_exit_irqoff(struct kvm_vcpu *vcpu, > *exit_fastpath = handle_fastpath_set_msr_irqoff(vcpu); > } > > +static bool vmx_has_virtualized_msr(u32 index) > +{ > + return true; > +} > + > static bool vmx_has_emulated_msr(u32 index) > { > switch (index) { > @@ -7754,6 +7759,7 @@ static struct kvm_x86_ops vmx_x86_ops __ro_after_init = { > .hardware_enable = hardware_enable, > .hardware_disable = hardware_disable, > .cpu_has_accelerated_tpr = report_flexpriority, > + .has_virtualized_msr = vmx_has_virtualized_msr, > .has_emulated_msr = vmx_has_emulated_msr, > > .vm_init = vmx_vm_init, > diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c > index 3d4a5326d84e..94f90fe1c0de 100644 > --- a/arch/x86/kvm/x86.c > +++ b/arch/x86/kvm/x86.c > @@ -5279,6 +5279,8 @@ static void kvm_init_msr_list(void) > continue; > break; > default: > + if (!kvm_x86_ops->has_virtualized_msr(msr_index)) > + continue; > break; > } Shouldn't break anything by itself, so Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
On Wed, Feb 05, 2020 at 03:34:29PM +0100, Vitaly Kuznetsov wrote: > Sean Christopherson <sean.j.christopherson@intel.com> writes: > > Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com> Stooooooop! Everything from this point on is obsoleted by kvm_cpu_caps!
Sean Christopherson <sean.j.christopherson@intel.com> writes: > On Wed, Feb 05, 2020 at 03:34:29PM +0100, Vitaly Kuznetsov wrote: >> Sean Christopherson <sean.j.christopherson@intel.com> writes: >> >> Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com> > > Stooooooop! Everything from this point on is obsoleted by kvm_cpu_caps! > Oops, this was only a week old series! Patches are rottening fast nowadays!
On Wed, Feb 05, 2020 at 04:22:48PM +0100, Vitaly Kuznetsov wrote: > Sean Christopherson <sean.j.christopherson@intel.com> writes: > > > On Wed, Feb 05, 2020 at 03:34:29PM +0100, Vitaly Kuznetsov wrote: > >> Sean Christopherson <sean.j.christopherson@intel.com> writes: > >> > >> Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com> > > > > Stooooooop! Everything from this point on is obsoleted by kvm_cpu_caps! > > > > Oops, this was only a week old series! Patches are rottening fast > nowadays! Sorry :-( I dug deeper into the CPUID crud after posting this series because I really didn't like the end result for vendor-specific leafs, and ended up coming up with (IMO) a much more elegant solution. https://lkml.kernel.org/r/20200201185218.24473-1-sean.j.christopherson@intel.com/ or on patchwork https://patchwork.kernel.org/cover/11361361/
Sean Christopherson <sean.j.christopherson@intel.com> writes: > On Wed, Feb 05, 2020 at 04:22:48PM +0100, Vitaly Kuznetsov wrote: >> Sean Christopherson <sean.j.christopherson@intel.com> writes: >> >> > On Wed, Feb 05, 2020 at 03:34:29PM +0100, Vitaly Kuznetsov wrote: >> >> Sean Christopherson <sean.j.christopherson@intel.com> writes: >> >> >> >> Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com> >> > >> > Stooooooop! Everything from this point on is obsoleted by kvm_cpu_caps! >> > >> >> Oops, this was only a week old series! Patches are rottening fast >> nowadays! > > Sorry :-( > > I dug deeper into the CPUID crud after posting this series because I really > didn't like the end result for vendor-specific leafs, and ended up coming > up with (IMO) a much more elegant solution. > > https://lkml.kernel.org/r/20200201185218.24473-1-sean.j.christopherson@intel.com/ > > or on patchwork > > https://patchwork.kernel.org/cover/11361361/ > Thanks, I saw it. I tried applying it to kvm/next earlier today but failed. Do you by any chance have a git branch somewhere? I'll try to review it and test at least AMD stuff (if AMD people don't beat me to it of course).
On Wed, Feb 05, 2020 at 05:55:32PM +0100, Vitaly Kuznetsov wrote: > Sean Christopherson <sean.j.christopherson@intel.com> writes: > > I dug deeper into the CPUID crud after posting this series because I really > > didn't like the end result for vendor-specific leafs, and ended up coming > > up with (IMO) a much more elegant solution. > > > > https://lkml.kernel.org/r/20200201185218.24473-1-sean.j.christopherson@intel.com/ > > > > or on patchwork > > > > https://patchwork.kernel.org/cover/11361361/ > > > > Thanks, I saw it. I tried applying it to kvm/next earlier today but > failed. Do you by any chance have a git branch somewhere? I'll try to > review it and test at least AMD stuff (if AMD people don't beat me to it > of course). Have you tried kvm/queue? I'm pretty sure I based the code on kvm/queue. If that doesn't work, I'll push a tag to my github repo. This is exactly why I usually note the base for large series. *sigh*
Sean Christopherson <sean.j.christopherson@intel.com> writes: > On Wed, Feb 05, 2020 at 05:55:32PM +0100, Vitaly Kuznetsov wrote: >> Sean Christopherson <sean.j.christopherson@intel.com> writes: >> > I dug deeper into the CPUID crud after posting this series because I really >> > didn't like the end result for vendor-specific leafs, and ended up coming >> > up with (IMO) a much more elegant solution. >> > >> > https://lkml.kernel.org/r/20200201185218.24473-1-sean.j.christopherson@intel.com/ >> > >> > or on patchwork >> > >> > https://patchwork.kernel.org/cover/11361361/ >> > >> >> Thanks, I saw it. I tried applying it to kvm/next earlier today but >> failed. Do you by any chance have a git branch somewhere? I'll try to >> review it and test at least AMD stuff (if AMD people don't beat me to it >> of course). > > Have you tried kvm/queue? I'm pretty sure I based the code on kvm/queue. > If that doesn't work, I'll push a tag to my github repo. My bad, kvm/queue worked like a charm! > > This is exactly why I usually note the base for large series. *sigh* Pull requests, anyone? :-)
diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h index 5c2ad3fa0980..8fb32c27fa44 100644 --- a/arch/x86/include/asm/kvm_host.h +++ b/arch/x86/include/asm/kvm_host.h @@ -1050,6 +1050,7 @@ struct kvm_x86_ops { int (*hardware_setup)(void); /* __init */ void (*hardware_unsetup)(void); /* __exit */ bool (*cpu_has_accelerated_tpr)(void); + bool (*has_virtualized_msr)(u32 index); bool (*has_emulated_msr)(u32 index); void (*cpuid_update)(struct kvm_vcpu *vcpu); diff --git a/arch/x86/kvm/svm.c b/arch/x86/kvm/svm.c index a7b944a3a0e2..1f9323fbad81 100644 --- a/arch/x86/kvm/svm.c +++ b/arch/x86/kvm/svm.c @@ -5985,6 +5985,11 @@ static bool svm_cpu_has_accelerated_tpr(void) return false; } +static bool svm_has_virtualized_msr(u32 index) +{ + return true; +} + static bool svm_has_emulated_msr(u32 index) { switch (index) { @@ -7379,6 +7384,7 @@ static struct kvm_x86_ops svm_x86_ops __ro_after_init = { .hardware_enable = svm_hardware_enable, .hardware_disable = svm_hardware_disable, .cpu_has_accelerated_tpr = svm_cpu_has_accelerated_tpr, + .has_virtualized_msr = svm_has_virtualized_msr, .has_emulated_msr = svm_has_emulated_msr, .vcpu_create = svm_create_vcpu, diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c index f5bb1ad2e9fa..3f2c094434e8 100644 --- a/arch/x86/kvm/vmx/vmx.c +++ b/arch/x86/kvm/vmx/vmx.c @@ -6274,6 +6274,11 @@ static void vmx_handle_exit_irqoff(struct kvm_vcpu *vcpu, *exit_fastpath = handle_fastpath_set_msr_irqoff(vcpu); } +static bool vmx_has_virtualized_msr(u32 index) +{ + return true; +} + static bool vmx_has_emulated_msr(u32 index) { switch (index) { @@ -7754,6 +7759,7 @@ static struct kvm_x86_ops vmx_x86_ops __ro_after_init = { .hardware_enable = hardware_enable, .hardware_disable = hardware_disable, .cpu_has_accelerated_tpr = report_flexpriority, + .has_virtualized_msr = vmx_has_virtualized_msr, .has_emulated_msr = vmx_has_emulated_msr, .vm_init = vmx_vm_init, diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c index 3d4a5326d84e..94f90fe1c0de 100644 --- a/arch/x86/kvm/x86.c +++ b/arch/x86/kvm/x86.c @@ -5279,6 +5279,8 @@ static void kvm_init_msr_list(void) continue; break; default: + if (!kvm_x86_ops->has_virtualized_msr(msr_index)) + continue; break; }
Add a hook, ->has_virtualized_msr(), to allow moving vendor specific checks into SVM/VMX and ultimately facilitate the removal of the piecemeal ->*_supported() hooks. No functional change intended. Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com> --- arch/x86/include/asm/kvm_host.h | 1 + arch/x86/kvm/svm.c | 6 ++++++ arch/x86/kvm/vmx/vmx.c | 6 ++++++ arch/x86/kvm/x86.c | 2 ++ 4 files changed, 15 insertions(+)