| Message ID | 20221215115207.14784-1-wei.w.wang@intel.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [v1] KVM: x86: add KVM_CAP_DEVICE_CTRL | expand |
On Thu, Dec 15, 2022, Wei Wang wrote: > KVM_CAP_DEVICE_CTRL allows userspace to create emulated device in KVM. > For example, userspace VFIO implementation needs to create a kvm_device > (i.e. KVM_DEV_TYPE_VFIO) on x86. So add the cap to allow userspace for > such use cases. > > Signed-off-by: Wei Wang <wei.w.wang@intel.com> > --- > arch/x86/kvm/x86.c | 1 + > 1 file changed, 1 insertion(+) > > diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c > index 69227f77b201..1cdc4469652c 100644 > --- a/arch/x86/kvm/x86.c > +++ b/arch/x86/kvm/x86.c > @@ -4410,6 +4410,7 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext) > case KVM_CAP_VAPIC: > case KVM_CAP_ENABLE_CAP: > case KVM_CAP_VM_DISABLE_NX_HUGE_PAGES: > + case KVM_CAP_DEVICE_CTRL: Rather than hardcode this in x86, I think it would be better to add an #ifdef'd version in the generic check. E.g. if MIPS or RISC-V ever gains KVM_VFIO support then they'll need to enumerate KVM_CAP_DEVICE_CTRL too, and odds are we'll forget to to do. diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c index 13e88297f999..f70b9cea95d9 100644 --- a/virt/kvm/kvm_main.c +++ b/virt/kvm/kvm_main.c @@ -4525,6 +4525,10 @@ static long kvm_vm_ioctl_check_extension_generic(struct kvm *kvm, long arg) case KVM_CAP_BINARY_STATS_FD: case KVM_CAP_SYSTEM_EVENT_DATA: return 1; +#ifdef CONFIG_KVM_VFIO + case KVM_CAP_DEVICE_CTRL: + return 1; +#endif default: break; } The other potentially bad idea would be to detect the presence of a device_ops and delete all of the arch hooks, e.g. diff --git a/arch/arm64/kvm/arm.c b/arch/arm64/kvm/arm.c index 9c5573bc4614..190e9c3b10a7 100644 --- a/arch/arm64/kvm/arm.c +++ b/arch/arm64/kvm/arm.c @@ -212,7 +212,6 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext) r = vgic_present; break; case KVM_CAP_IOEVENTFD: - case KVM_CAP_DEVICE_CTRL: case KVM_CAP_USER_MEMORY: case KVM_CAP_SYNC_MMU: case KVM_CAP_DESTROY_MEMORY_REGION_WORKS: diff --git a/arch/powerpc/kvm/powerpc.c b/arch/powerpc/kvm/powerpc.c index 04494a4fb37a..21f9fbe96f6a 100644 --- a/arch/powerpc/kvm/powerpc.c +++ b/arch/powerpc/kvm/powerpc.c @@ -541,7 +541,6 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext) case KVM_CAP_ENABLE_CAP: case KVM_CAP_ONE_REG: case KVM_CAP_IOEVENTFD: - case KVM_CAP_DEVICE_CTRL: case KVM_CAP_IMMEDIATE_EXIT: case KVM_CAP_SET_GUEST_DEBUG: r = 1; diff --git a/arch/riscv/kvm/vm.c b/arch/riscv/kvm/vm.c index 65a964d7e70d..6efe93b282e1 100644 --- a/arch/riscv/kvm/vm.c +++ b/arch/riscv/kvm/vm.c @@ -57,7 +57,6 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext) switch (ext) { case KVM_CAP_IOEVENTFD: - case KVM_CAP_DEVICE_CTRL: case KVM_CAP_USER_MEMORY: case KVM_CAP_SYNC_MMU: case KVM_CAP_DESTROY_MEMORY_REGION_WORKS: diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c index e4890e04b210..191d220b6a30 100644 --- a/arch/s390/kvm/kvm-s390.c +++ b/arch/s390/kvm/kvm-s390.c @@ -567,7 +567,6 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext) case KVM_CAP_ENABLE_CAP: case KVM_CAP_S390_CSS_SUPPORT: case KVM_CAP_IOEVENTFD: - case KVM_CAP_DEVICE_CTRL: case KVM_CAP_S390_IRQCHIP: case KVM_CAP_VM_ATTRIBUTES: case KVM_CAP_MP_STATE: diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c index 13e88297f999..99e3da9ce42d 100644 --- a/virt/kvm/kvm_main.c +++ b/virt/kvm/kvm_main.c @@ -4525,6 +4525,15 @@ static long kvm_vm_ioctl_check_extension_generic(struct kvm *kvm, long arg) case KVM_CAP_BINARY_STATS_FD: case KVM_CAP_SYSTEM_EVENT_DATA: return 1; + case KVM_CAP_DEVICE_CTRL: { + int i; + + for (i = 0; i < ARRAY_SIZE(kvm_device_ops_table); ++) { + if (kvm_device_ops_table[i]) + return 1; + } + return 0; + } default: break; }
On Saturday, December 17, 2022 1:13 AM, Sean Christopherson wrote: > Rather than hardcode this in x86, I think it would be better to add an #ifdef'd > version in the generic check. E.g. if MIPS or RISC-V ever gains KVM_VFIO > support then they'll need to enumerate KVM_CAP_DEVICE_CTRL too, and odds > are we'll forget to to do. > > diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c index > 13e88297f999..f70b9cea95d9 100644 > --- a/virt/kvm/kvm_main.c > +++ b/virt/kvm/kvm_main.c > @@ -4525,6 +4525,10 @@ static long > kvm_vm_ioctl_check_extension_generic(struct kvm *kvm, long arg) > case KVM_CAP_BINARY_STATS_FD: > case KVM_CAP_SYSTEM_EVENT_DATA: > return 1; > +#ifdef CONFIG_KVM_VFIO > + case KVM_CAP_DEVICE_CTRL: > + return 1; > +#endif > default: > break; > } > > The other potentially bad idea would be to detect the presence of a > device_ops and delete all of the arch hooks, e.g. > > diff --git a/arch/arm64/kvm/arm.c b/arch/arm64/kvm/arm.c index > 9c5573bc4614..190e9c3b10a7 100644 > --- a/arch/arm64/kvm/arm.c > +++ b/arch/arm64/kvm/arm.c > @@ -212,7 +212,6 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, > long ext) > r = vgic_present; > break; > case KVM_CAP_IOEVENTFD: > - case KVM_CAP_DEVICE_CTRL: > case KVM_CAP_USER_MEMORY: > case KVM_CAP_SYNC_MMU: > case KVM_CAP_DESTROY_MEMORY_REGION_WORKS: > diff --git a/arch/powerpc/kvm/powerpc.c b/arch/powerpc/kvm/powerpc.c > index 04494a4fb37a..21f9fbe96f6a 100644 > --- a/arch/powerpc/kvm/powerpc.c > +++ b/arch/powerpc/kvm/powerpc.c > @@ -541,7 +541,6 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, > long ext) > case KVM_CAP_ENABLE_CAP: > case KVM_CAP_ONE_REG: > case KVM_CAP_IOEVENTFD: > - case KVM_CAP_DEVICE_CTRL: > case KVM_CAP_IMMEDIATE_EXIT: > case KVM_CAP_SET_GUEST_DEBUG: > r = 1; > diff --git a/arch/riscv/kvm/vm.c b/arch/riscv/kvm/vm.c index > 65a964d7e70d..6efe93b282e1 100644 > --- a/arch/riscv/kvm/vm.c > +++ b/arch/riscv/kvm/vm.c > @@ -57,7 +57,6 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, > long ext) > > switch (ext) { > case KVM_CAP_IOEVENTFD: > - case KVM_CAP_DEVICE_CTRL: > case KVM_CAP_USER_MEMORY: > case KVM_CAP_SYNC_MMU: > case KVM_CAP_DESTROY_MEMORY_REGION_WORKS: > diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c index > e4890e04b210..191d220b6a30 100644 > --- a/arch/s390/kvm/kvm-s390.c > +++ b/arch/s390/kvm/kvm-s390.c > @@ -567,7 +567,6 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, > long ext) > case KVM_CAP_ENABLE_CAP: > case KVM_CAP_S390_CSS_SUPPORT: > case KVM_CAP_IOEVENTFD: > - case KVM_CAP_DEVICE_CTRL: > case KVM_CAP_S390_IRQCHIP: > case KVM_CAP_VM_ATTRIBUTES: > case KVM_CAP_MP_STATE: > diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c index > 13e88297f999..99e3da9ce42d 100644 > --- a/virt/kvm/kvm_main.c > +++ b/virt/kvm/kvm_main.c > @@ -4525,6 +4525,15 @@ static long > kvm_vm_ioctl_check_extension_generic(struct kvm *kvm, long arg) > case KVM_CAP_BINARY_STATS_FD: > case KVM_CAP_SYSTEM_EVENT_DATA: > return 1; > + case KVM_CAP_DEVICE_CTRL: { > + int i; > + > + for (i = 0; i < ARRAY_SIZE(kvm_device_ops_table); ++) { > + if (kvm_device_ops_table[i]) > + return 1; > + } > + return 0; > + } > default: > break; > } Yes, it looks better to move it to the generic check, but I'm not sure if it would be necessary to do the per-device check here either via CONFIG_KVM_VFIO (for example, if more non-arch-specific usages are added, we would end up with lots of such #ifdef to be added, which doesn't seem nice) or kvm_device_ops_table. I think fundamentally KVM_CAP_DEVICE_CTRL is used to check if the generic kvm_device framework (e.g. KVM_CREATE_DEVICE) is supported by KVM (older KVM before 2013 doesn't have it). The per-device type (KVM_DEV_TYPE_VFIO, KVM_DEV_TYPE_ARM_PV_TIME etc.) support can be checked via KVM_CREATE_DEVICE, which reports -ENODEV if the device type doesn't have an entry in kvm_device_ops_table.
On Mon, Dec 19, 2022, Wang, Wei W wrote: > On Saturday, December 17, 2022 1:13 AM, Sean Christopherson wrote: > > Rather than hardcode this in x86, I think it would be better to add an #ifdef'd > > version in the generic check. E.g. if MIPS or RISC-V ever gains KVM_VFIO > > support then they'll need to enumerate KVM_CAP_DEVICE_CTRL too, and odds > > are we'll forget to to do. ... > > The other potentially bad idea would be to detect the presence of a > > device_ops and delete all of the arch hooks, e.g. > Yes, it looks better to move it to the generic check, but I'm not sure if it > would be necessary to do the per-device check here either via CONFIG_KVM_VFIO > (for example, if more non-arch-specific usages are added, we would end up > with lots of such #ifdef to be added, which doesn't seem nice) or > kvm_device_ops_table. > > I think fundamentally KVM_CAP_DEVICE_CTRL is used to check if the generic > kvm_device framework (e.g. KVM_CREATE_DEVICE) is supported by KVM (older KVM > before 2013 doesn't have it). The per-device type (KVM_DEV_TYPE_VFIO, > KVM_DEV_TYPE_ARM_PV_TIME etc.) support can be checked via KVM_CREATE_DEVICE, > which reports -ENODEV if the device type doesn't have an entry in > kvm_device_ops_table. If that's how we want to retroactively define things, then KVM should unconditionally return 1/true for KVM_CAP_DEVICE_CTRL since KVM_CREATE_DEVICE is provided by generic code.
On Tuesday, December 20, 2022 4:36 AM, Sean Christopherson wrote: > > Yes, it looks better to move it to the generic check, but I'm not sure > > if it would be necessary to do the per-device check here either via > > CONFIG_KVM_VFIO (for example, if more non-arch-specific usages are > > added, we would end up with lots of such #ifdef to be added, which > > doesn't seem nice) or kvm_device_ops_table. > > > > I think fundamentally KVM_CAP_DEVICE_CTRL is used to check if the > > generic kvm_device framework (e.g. KVM_CREATE_DEVICE) is supported by > > KVM (older KVM before 2013 doesn't have it). The per-device type > > (KVM_DEV_TYPE_VFIO, KVM_DEV_TYPE_ARM_PV_TIME etc.) support can be > > checked via KVM_CREATE_DEVICE, which reports -ENODEV if the device > > type doesn't have an entry in kvm_device_ops_table. > > If that's how we want to retroactively define things, then KVM should > unconditionally return 1/true for KVM_CAP_DEVICE_CTRL since > KVM_CREATE_DEVICE is provided by generic code. Yes. Also, since we have KVM_DEV_TYPE_VFIO the generic use case, it should be better to move the CAP check to the generic kvm_vm_ioctl_check_extension_generic.
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c index 69227f77b201..1cdc4469652c 100644 --- a/arch/x86/kvm/x86.c +++ b/arch/x86/kvm/x86.c @@ -4410,6 +4410,7 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext) case KVM_CAP_VAPIC: case KVM_CAP_ENABLE_CAP: case KVM_CAP_VM_DISABLE_NX_HUGE_PAGES: + case KVM_CAP_DEVICE_CTRL: r = 1; break; case KVM_CAP_EXIT_HYPERCALL:
KVM_CAP_DEVICE_CTRL allows userspace to create emulated device in KVM. For example, userspace VFIO implementation needs to create a kvm_device (i.e. KVM_DEV_TYPE_VFIO) on x86. So add the cap to allow userspace for such use cases. Signed-off-by: Wei Wang <wei.w.wang@intel.com> --- arch/x86/kvm/x86.c | 1 + 1 file changed, 1 insertion(+)