| Message ID | 20230725122601.424738-2-ajones@ventanamicro.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [v2] kvm: Remove KVM_CREATE_IRQCHIP support assumption | expand |
On 25/7/23 14:26, Andrew Jones wrote: > Since Linux commit 00f918f61c56 ("RISC-V: KVM: Skeletal in-kernel AIA > irqchip support") checking KVM_CAP_IRQCHIP returns non-zero when the > RISC-V platform has AIA. The cap indicates KVM supports at least one > of the following ioctls: > > KVM_CREATE_IRQCHIP > KVM_IRQ_LINE > KVM_GET_IRQCHIP > KVM_SET_IRQCHIP > KVM_GET_LAPIC > KVM_SET_LAPIC > > but the cap doesn't imply that KVM must support any of those ioctls > in particular. However, QEMU was assuming the KVM_CREATE_IRQCHIP > ioctl was supported. Stop making that assumption by introducing a > KVM parameter that each architecture which supports KVM_CREATE_IRQCHIP > sets. Adding parameters isn't awesome, but given how the > KVM_CAP_IRQCHIP isn't very helpful on its own, we don't have a lot of > options. > > Signed-off-by: Andrew Jones <ajones@ventanamicro.com> > --- > > While this fixes booting guests on riscv KVM with AIA it's unlikely > to get merged before the QEMU support for KVM AIA[1] lands, which > would also fix the issue. I think this patch is still worth considering > though since QEMU's assumption is wrong. > > [1] https://lore.kernel.org/all/20230714084429.22349-1-yongxuan.wang@sifive.com/ > > v2: > - Move the s390x code to an s390x file. [Thomas] > - Drop the KVM_CAP_IRQCHIP check from the top of kvm_irqchip_create(), > as it's no longer necessary. > > accel/kvm/kvm-all.c | 16 ++++------------ > include/sysemu/kvm.h | 1 + > target/arm/kvm.c | 3 +++ > target/i386/kvm/kvm.c | 2 ++ > target/s390x/kvm/kvm.c | 11 +++++++++++ > 5 files changed, 21 insertions(+), 12 deletions(-) Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
On 25/07/2023 14.26, Andrew Jones wrote: > Since Linux commit 00f918f61c56 ("RISC-V: KVM: Skeletal in-kernel AIA > irqchip support") checking KVM_CAP_IRQCHIP returns non-zero when the > RISC-V platform has AIA. The cap indicates KVM supports at least one > of the following ioctls: > > KVM_CREATE_IRQCHIP > KVM_IRQ_LINE > KVM_GET_IRQCHIP > KVM_SET_IRQCHIP > KVM_GET_LAPIC > KVM_SET_LAPIC > > but the cap doesn't imply that KVM must support any of those ioctls > in particular. However, QEMU was assuming the KVM_CREATE_IRQCHIP > ioctl was supported. Stop making that assumption by introducing a > KVM parameter that each architecture which supports KVM_CREATE_IRQCHIP > sets. Adding parameters isn't awesome, but given how the > KVM_CAP_IRQCHIP isn't very helpful on its own, we don't have a lot of > options. > > Signed-off-by: Andrew Jones <ajones@ventanamicro.com> > --- > > While this fixes booting guests on riscv KVM with AIA it's unlikely > to get merged before the QEMU support for KVM AIA[1] lands, which > would also fix the issue. I think this patch is still worth considering > though since QEMU's assumption is wrong. > > [1] https://lore.kernel.org/all/20230714084429.22349-1-yongxuan.wang@sifive.com/ > > v2: > - Move the s390x code to an s390x file. [Thomas] > - Drop the KVM_CAP_IRQCHIP check from the top of kvm_irqchip_create(), > as it's no longer necessary. Looks good now! Reviewed-by: Thomas Huth <thuth@redhat.com>
On 7/25/23 09:26, Andrew Jones wrote: > Since Linux commit 00f918f61c56 ("RISC-V: KVM: Skeletal in-kernel AIA > irqchip support") checking KVM_CAP_IRQCHIP returns non-zero when the > RISC-V platform has AIA. The cap indicates KVM supports at least one > of the following ioctls: > > KVM_CREATE_IRQCHIP > KVM_IRQ_LINE > KVM_GET_IRQCHIP > KVM_SET_IRQCHIP > KVM_GET_LAPIC > KVM_SET_LAPIC > > but the cap doesn't imply that KVM must support any of those ioctls > in particular. However, QEMU was assuming the KVM_CREATE_IRQCHIP > ioctl was supported. Stop making that assumption by introducing a > KVM parameter that each architecture which supports KVM_CREATE_IRQCHIP > sets. Adding parameters isn't awesome, but given how the > KVM_CAP_IRQCHIP isn't very helpful on its own, we don't have a lot of > options. > > Signed-off-by: Andrew Jones <ajones@ventanamicro.com> > --- Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com> > > While this fixes booting guests on riscv KVM with AIA it's unlikely > to get merged before the QEMU support for KVM AIA[1] lands, which > would also fix the issue. I think this patch is still worth considering > though since QEMU's assumption is wrong. > > [1] https://lore.kernel.org/all/20230714084429.22349-1-yongxuan.wang@sifive.com/ > > v2: > - Move the s390x code to an s390x file. [Thomas] > - Drop the KVM_CAP_IRQCHIP check from the top of kvm_irqchip_create(), > as it's no longer necessary. > > accel/kvm/kvm-all.c | 16 ++++------------ > include/sysemu/kvm.h | 1 + > target/arm/kvm.c | 3 +++ > target/i386/kvm/kvm.c | 2 ++ > target/s390x/kvm/kvm.c | 11 +++++++++++ > 5 files changed, 21 insertions(+), 12 deletions(-) > > diff --git a/accel/kvm/kvm-all.c b/accel/kvm/kvm-all.c > index 373d876c0580..cddcb6eca641 100644 > --- a/accel/kvm/kvm-all.c > +++ b/accel/kvm/kvm-all.c > @@ -86,6 +86,7 @@ struct KVMParkedVcpu { > }; > > KVMState *kvm_state; > +bool kvm_has_create_irqchip; > bool kvm_kernel_irqchip; > bool kvm_split_irqchip; > bool kvm_async_interrupts_allowed; > @@ -2358,17 +2359,6 @@ static void kvm_irqchip_create(KVMState *s) > int ret; > > assert(s->kernel_irqchip_split != ON_OFF_AUTO_AUTO); > - if (kvm_check_extension(s, KVM_CAP_IRQCHIP)) { > - ; > - } else if (kvm_check_extension(s, KVM_CAP_S390_IRQCHIP)) { > - ret = kvm_vm_enable_cap(s, KVM_CAP_S390_IRQCHIP, 0); > - if (ret < 0) { > - fprintf(stderr, "Enable kernel irqchip failed: %s\n", strerror(-ret)); > - exit(1); > - } > - } else { > - return; > - } > > /* First probe and see if there's a arch-specific hook to create the > * in-kernel irqchip for us */ > @@ -2377,8 +2367,10 @@ static void kvm_irqchip_create(KVMState *s) > if (s->kernel_irqchip_split == ON_OFF_AUTO_ON) { > error_report("Split IRQ chip mode not supported."); > exit(1); > - } else { > + } else if (kvm_has_create_irqchip) { > ret = kvm_vm_ioctl(s, KVM_CREATE_IRQCHIP); > + } else { > + return; > } > } > if (ret < 0) { > diff --git a/include/sysemu/kvm.h b/include/sysemu/kvm.h > index 115f0cca79d1..84b1bb3dc91e 100644 > --- a/include/sysemu/kvm.h > +++ b/include/sysemu/kvm.h > @@ -32,6 +32,7 @@ > #ifdef CONFIG_KVM_IS_POSSIBLE > > extern bool kvm_allowed; > +extern bool kvm_has_create_irqchip; > extern bool kvm_kernel_irqchip; > extern bool kvm_split_irqchip; > extern bool kvm_async_interrupts_allowed; > diff --git a/target/arm/kvm.c b/target/arm/kvm.c > index b4c7654f4980..2fa87b495d68 100644 > --- a/target/arm/kvm.c > +++ b/target/arm/kvm.c > @@ -250,6 +250,9 @@ int kvm_arm_get_max_vm_ipa_size(MachineState *ms, bool *fixed_ipa) > int kvm_arch_init(MachineState *ms, KVMState *s) > { > int ret = 0; > + > + kvm_has_create_irqchip = kvm_check_extension(s, KVM_CAP_IRQCHIP); > + > /* For ARM interrupt delivery is always asynchronous, > * whether we are using an in-kernel VGIC or not. > */ > diff --git a/target/i386/kvm/kvm.c b/target/i386/kvm/kvm.c > index ebfaf3d24c79..6363e67f092d 100644 > --- a/target/i386/kvm/kvm.c > +++ b/target/i386/kvm/kvm.c > @@ -2771,6 +2771,8 @@ int kvm_arch_init(MachineState *ms, KVMState *s) > } > } > > + kvm_has_create_irqchip = kvm_check_extension(s, KVM_CAP_IRQCHIP); > + > return 0; > } > > diff --git a/target/s390x/kvm/kvm.c b/target/s390x/kvm/kvm.c > index a9e5880349d9..bcc735227f7d 100644 > --- a/target/s390x/kvm/kvm.c > +++ b/target/s390x/kvm/kvm.c > @@ -391,6 +391,17 @@ int kvm_arch_init(MachineState *ms, KVMState *s) > } > > kvm_set_max_memslot_size(KVM_SLOT_MAX_BYTES); > + > + kvm_has_create_irqchip = kvm_check_extension(s, KVM_CAP_S390_IRQCHIP); > + if (kvm_has_create_irqchip) { > + int ret = kvm_vm_enable_cap(s, KVM_CAP_S390_IRQCHIP, 0); > + > + if (ret < 0) { > + fprintf(stderr, "Enable kernel irqchip failed: %s\n", strerror(-ret)); > + exit(1); > + } > + } > + > return 0; > } >
Hi Paolo, Is this good for 8.1? Thanks, drew On Tue, Jul 25, 2023 at 02:26:02PM +0200, Andrew Jones wrote: > Since Linux commit 00f918f61c56 ("RISC-V: KVM: Skeletal in-kernel AIA > irqchip support") checking KVM_CAP_IRQCHIP returns non-zero when the > RISC-V platform has AIA. The cap indicates KVM supports at least one > of the following ioctls: > > KVM_CREATE_IRQCHIP > KVM_IRQ_LINE > KVM_GET_IRQCHIP > KVM_SET_IRQCHIP > KVM_GET_LAPIC > KVM_SET_LAPIC > > but the cap doesn't imply that KVM must support any of those ioctls > in particular. However, QEMU was assuming the KVM_CREATE_IRQCHIP > ioctl was supported. Stop making that assumption by introducing a > KVM parameter that each architecture which supports KVM_CREATE_IRQCHIP > sets. Adding parameters isn't awesome, but given how the > KVM_CAP_IRQCHIP isn't very helpful on its own, we don't have a lot of > options. > > Signed-off-by: Andrew Jones <ajones@ventanamicro.com> > --- > > While this fixes booting guests on riscv KVM with AIA it's unlikely > to get merged before the QEMU support for KVM AIA[1] lands, which > would also fix the issue. I think this patch is still worth considering > though since QEMU's assumption is wrong. > > [1] https://lore.kernel.org/all/20230714084429.22349-1-yongxuan.wang@sifive.com/ > > v2: > - Move the s390x code to an s390x file. [Thomas] > - Drop the KVM_CAP_IRQCHIP check from the top of kvm_irqchip_create(), > as it's no longer necessary. > > accel/kvm/kvm-all.c | 16 ++++------------ > include/sysemu/kvm.h | 1 + > target/arm/kvm.c | 3 +++ > target/i386/kvm/kvm.c | 2 ++ > target/s390x/kvm/kvm.c | 11 +++++++++++ > 5 files changed, 21 insertions(+), 12 deletions(-) > > diff --git a/accel/kvm/kvm-all.c b/accel/kvm/kvm-all.c > index 373d876c0580..cddcb6eca641 100644 > --- a/accel/kvm/kvm-all.c > +++ b/accel/kvm/kvm-all.c > @@ -86,6 +86,7 @@ struct KVMParkedVcpu { > }; > > KVMState *kvm_state; > +bool kvm_has_create_irqchip; > bool kvm_kernel_irqchip; > bool kvm_split_irqchip; > bool kvm_async_interrupts_allowed; > @@ -2358,17 +2359,6 @@ static void kvm_irqchip_create(KVMState *s) > int ret; > > assert(s->kernel_irqchip_split != ON_OFF_AUTO_AUTO); > - if (kvm_check_extension(s, KVM_CAP_IRQCHIP)) { > - ; > - } else if (kvm_check_extension(s, KVM_CAP_S390_IRQCHIP)) { > - ret = kvm_vm_enable_cap(s, KVM_CAP_S390_IRQCHIP, 0); > - if (ret < 0) { > - fprintf(stderr, "Enable kernel irqchip failed: %s\n", strerror(-ret)); > - exit(1); > - } > - } else { > - return; > - } > > /* First probe and see if there's a arch-specific hook to create the > * in-kernel irqchip for us */ > @@ -2377,8 +2367,10 @@ static void kvm_irqchip_create(KVMState *s) > if (s->kernel_irqchip_split == ON_OFF_AUTO_ON) { > error_report("Split IRQ chip mode not supported."); > exit(1); > - } else { > + } else if (kvm_has_create_irqchip) { > ret = kvm_vm_ioctl(s, KVM_CREATE_IRQCHIP); > + } else { > + return; > } > } > if (ret < 0) { > diff --git a/include/sysemu/kvm.h b/include/sysemu/kvm.h > index 115f0cca79d1..84b1bb3dc91e 100644 > --- a/include/sysemu/kvm.h > +++ b/include/sysemu/kvm.h > @@ -32,6 +32,7 @@ > #ifdef CONFIG_KVM_IS_POSSIBLE > > extern bool kvm_allowed; > +extern bool kvm_has_create_irqchip; > extern bool kvm_kernel_irqchip; > extern bool kvm_split_irqchip; > extern bool kvm_async_interrupts_allowed; > diff --git a/target/arm/kvm.c b/target/arm/kvm.c > index b4c7654f4980..2fa87b495d68 100644 > --- a/target/arm/kvm.c > +++ b/target/arm/kvm.c > @@ -250,6 +250,9 @@ int kvm_arm_get_max_vm_ipa_size(MachineState *ms, bool *fixed_ipa) > int kvm_arch_init(MachineState *ms, KVMState *s) > { > int ret = 0; > + > + kvm_has_create_irqchip = kvm_check_extension(s, KVM_CAP_IRQCHIP); > + > /* For ARM interrupt delivery is always asynchronous, > * whether we are using an in-kernel VGIC or not. > */ > diff --git a/target/i386/kvm/kvm.c b/target/i386/kvm/kvm.c > index ebfaf3d24c79..6363e67f092d 100644 > --- a/target/i386/kvm/kvm.c > +++ b/target/i386/kvm/kvm.c > @@ -2771,6 +2771,8 @@ int kvm_arch_init(MachineState *ms, KVMState *s) > } > } > > + kvm_has_create_irqchip = kvm_check_extension(s, KVM_CAP_IRQCHIP); > + > return 0; > } > > diff --git a/target/s390x/kvm/kvm.c b/target/s390x/kvm/kvm.c > index a9e5880349d9..bcc735227f7d 100644 > --- a/target/s390x/kvm/kvm.c > +++ b/target/s390x/kvm/kvm.c > @@ -391,6 +391,17 @@ int kvm_arch_init(MachineState *ms, KVMState *s) > } > > kvm_set_max_memslot_size(KVM_SLOT_MAX_BYTES); > + > + kvm_has_create_irqchip = kvm_check_extension(s, KVM_CAP_S390_IRQCHIP); > + if (kvm_has_create_irqchip) { > + int ret = kvm_vm_enable_cap(s, KVM_CAP_S390_IRQCHIP, 0); > + > + if (ret < 0) { > + fprintf(stderr, "Enable kernel irqchip failed: %s\n", strerror(-ret)); > + exit(1); > + } > + } > + > return 0; > } > > -- > 2.41.0 >
On Thu, 10 Aug 2023 at 12:29, Andrew Jones <ajones@ventanamicro.com> wrote: > > > Hi Paolo, > > Is this good for 8.1? Is it fixing a regression since 8.0 ? thanks -- PMM > > Thanks, > drew > > > On Tue, Jul 25, 2023 at 02:26:02PM +0200, Andrew Jones wrote: > > Since Linux commit 00f918f61c56 ("RISC-V: KVM: Skeletal in-kernel AIA > > irqchip support") checking KVM_CAP_IRQCHIP returns non-zero when the > > RISC-V platform has AIA. The cap indicates KVM supports at least one > > of the following ioctls: > > > > KVM_CREATE_IRQCHIP > > KVM_IRQ_LINE > > KVM_GET_IRQCHIP > > KVM_SET_IRQCHIP > > KVM_GET_LAPIC > > KVM_SET_LAPIC > > > > but the cap doesn't imply that KVM must support any of those ioctls > > in particular. However, QEMU was assuming the KVM_CREATE_IRQCHIP > > ioctl was supported. Stop making that assumption by introducing a > > KVM parameter that each architecture which supports KVM_CREATE_IRQCHIP > > sets. Adding parameters isn't awesome, but given how the > > KVM_CAP_IRQCHIP isn't very helpful on its own, we don't have a lot of > > options. > > > > Signed-off-by: Andrew Jones <ajones@ventanamicro.com> > > --- > > > > While this fixes booting guests on riscv KVM with AIA it's unlikely > > to get merged before the QEMU support for KVM AIA[1] lands, which > > would also fix the issue. I think this patch is still worth considering > > though since QEMU's assumption is wrong. > > > > [1] https://lore.kernel.org/all/20230714084429.22349-1-yongxuan.wang@sifive.com/ > > > > v2: > > - Move the s390x code to an s390x file. [Thomas] > > - Drop the KVM_CAP_IRQCHIP check from the top of kvm_irqchip_create(), > > as it's no longer necessary. > > > > accel/kvm/kvm-all.c | 16 ++++------------ > > include/sysemu/kvm.h | 1 + > > target/arm/kvm.c | 3 +++ > > target/i386/kvm/kvm.c | 2 ++ > > target/s390x/kvm/kvm.c | 11 +++++++++++ > > 5 files changed, 21 insertions(+), 12 deletions(-) > > > > diff --git a/accel/kvm/kvm-all.c b/accel/kvm/kvm-all.c > > index 373d876c0580..cddcb6eca641 100644 > > --- a/accel/kvm/kvm-all.c > > +++ b/accel/kvm/kvm-all.c > > @@ -86,6 +86,7 @@ struct KVMParkedVcpu { > > }; > > > > KVMState *kvm_state; > > +bool kvm_has_create_irqchip; > > bool kvm_kernel_irqchip; > > bool kvm_split_irqchip; > > bool kvm_async_interrupts_allowed; > > @@ -2358,17 +2359,6 @@ static void kvm_irqchip_create(KVMState *s) > > int ret; > > > > assert(s->kernel_irqchip_split != ON_OFF_AUTO_AUTO); > > - if (kvm_check_extension(s, KVM_CAP_IRQCHIP)) { > > - ; > > - } else if (kvm_check_extension(s, KVM_CAP_S390_IRQCHIP)) { > > - ret = kvm_vm_enable_cap(s, KVM_CAP_S390_IRQCHIP, 0); > > - if (ret < 0) { > > - fprintf(stderr, "Enable kernel irqchip failed: %s\n", strerror(-ret)); > > - exit(1); > > - } > > - } else { > > - return; > > - } > > > > /* First probe and see if there's a arch-specific hook to create the > > * in-kernel irqchip for us */ > > @@ -2377,8 +2367,10 @@ static void kvm_irqchip_create(KVMState *s) > > if (s->kernel_irqchip_split == ON_OFF_AUTO_ON) { > > error_report("Split IRQ chip mode not supported."); > > exit(1); > > - } else { > > + } else if (kvm_has_create_irqchip) { > > ret = kvm_vm_ioctl(s, KVM_CREATE_IRQCHIP); > > + } else { > > + return; > > } > > } > > if (ret < 0) { > > diff --git a/include/sysemu/kvm.h b/include/sysemu/kvm.h > > index 115f0cca79d1..84b1bb3dc91e 100644 > > --- a/include/sysemu/kvm.h > > +++ b/include/sysemu/kvm.h > > @@ -32,6 +32,7 @@ > > #ifdef CONFIG_KVM_IS_POSSIBLE > > > > extern bool kvm_allowed; > > +extern bool kvm_has_create_irqchip; > > extern bool kvm_kernel_irqchip; > > extern bool kvm_split_irqchip; > > extern bool kvm_async_interrupts_allowed; > > diff --git a/target/arm/kvm.c b/target/arm/kvm.c > > index b4c7654f4980..2fa87b495d68 100644 > > --- a/target/arm/kvm.c > > +++ b/target/arm/kvm.c > > @@ -250,6 +250,9 @@ int kvm_arm_get_max_vm_ipa_size(MachineState *ms, bool *fixed_ipa) > > int kvm_arch_init(MachineState *ms, KVMState *s) > > { > > int ret = 0; > > + > > + kvm_has_create_irqchip = kvm_check_extension(s, KVM_CAP_IRQCHIP); > > + > > /* For ARM interrupt delivery is always asynchronous, > > * whether we are using an in-kernel VGIC or not. > > */ > > diff --git a/target/i386/kvm/kvm.c b/target/i386/kvm/kvm.c > > index ebfaf3d24c79..6363e67f092d 100644 > > --- a/target/i386/kvm/kvm.c > > +++ b/target/i386/kvm/kvm.c > > @@ -2771,6 +2771,8 @@ int kvm_arch_init(MachineState *ms, KVMState *s) > > } > > } > > > > + kvm_has_create_irqchip = kvm_check_extension(s, KVM_CAP_IRQCHIP); > > + > > return 0; > > } > > > > diff --git a/target/s390x/kvm/kvm.c b/target/s390x/kvm/kvm.c > > index a9e5880349d9..bcc735227f7d 100644 > > --- a/target/s390x/kvm/kvm.c > > +++ b/target/s390x/kvm/kvm.c > > @@ -391,6 +391,17 @@ int kvm_arch_init(MachineState *ms, KVMState *s) > > } > > > > kvm_set_max_memslot_size(KVM_SLOT_MAX_BYTES); > > + > > + kvm_has_create_irqchip = kvm_check_extension(s, KVM_CAP_S390_IRQCHIP); > > + if (kvm_has_create_irqchip) { > > + int ret = kvm_vm_enable_cap(s, KVM_CAP_S390_IRQCHIP, 0); > > + > > + if (ret < 0) { > > + fprintf(stderr, "Enable kernel irqchip failed: %s\n", strerror(-ret)); > > + exit(1); > > + } > > + } > > + > > return 0; > > } > > > > -- > > 2.41.0 > >
On Thu, Aug 10, 2023 at 12:38:28PM +0100, Peter Maydell wrote: > On Thu, 10 Aug 2023 at 12:29, Andrew Jones <ajones@ventanamicro.com> wrote: > > > > > > Hi Paolo, > > > > Is this good for 8.1? > > Is it fixing a regression since 8.0 ? The fix is for running with KVM which includes 00f918f61c56, which is v6.5-rc1 and later. All QEMU versions will fail to start, so this fix isn't fixing a QEMU regression, but until this fix, or the KVM AIA support series, is merged, then we won't have any QEMU version we can use with later KVM. Thanks, drew > > thanks > -- PMM > > > > > Thanks, > > drew > > > > > > On Tue, Jul 25, 2023 at 02:26:02PM +0200, Andrew Jones wrote: > > > Since Linux commit 00f918f61c56 ("RISC-V: KVM: Skeletal in-kernel AIA > > > irqchip support") checking KVM_CAP_IRQCHIP returns non-zero when the > > > RISC-V platform has AIA. The cap indicates KVM supports at least one > > > of the following ioctls: > > > > > > KVM_CREATE_IRQCHIP > > > KVM_IRQ_LINE > > > KVM_GET_IRQCHIP > > > KVM_SET_IRQCHIP > > > KVM_GET_LAPIC > > > KVM_SET_LAPIC > > > > > > but the cap doesn't imply that KVM must support any of those ioctls > > > in particular. However, QEMU was assuming the KVM_CREATE_IRQCHIP > > > ioctl was supported. Stop making that assumption by introducing a > > > KVM parameter that each architecture which supports KVM_CREATE_IRQCHIP > > > sets. Adding parameters isn't awesome, but given how the > > > KVM_CAP_IRQCHIP isn't very helpful on its own, we don't have a lot of > > > options. > > > > > > Signed-off-by: Andrew Jones <ajones@ventanamicro.com> > > > --- > > > > > > While this fixes booting guests on riscv KVM with AIA it's unlikely > > > to get merged before the QEMU support for KVM AIA[1] lands, which > > > would also fix the issue. I think this patch is still worth considering > > > though since QEMU's assumption is wrong. > > > > > > [1] https://lore.kernel.org/all/20230714084429.22349-1-yongxuan.wang@sifive.com/ > > > > > > v2: > > > - Move the s390x code to an s390x file. [Thomas] > > > - Drop the KVM_CAP_IRQCHIP check from the top of kvm_irqchip_create(), > > > as it's no longer necessary. > > > > > > accel/kvm/kvm-all.c | 16 ++++------------ > > > include/sysemu/kvm.h | 1 + > > > target/arm/kvm.c | 3 +++ > > > target/i386/kvm/kvm.c | 2 ++ > > > target/s390x/kvm/kvm.c | 11 +++++++++++ > > > 5 files changed, 21 insertions(+), 12 deletions(-) > > > > > > diff --git a/accel/kvm/kvm-all.c b/accel/kvm/kvm-all.c > > > index 373d876c0580..cddcb6eca641 100644 > > > --- a/accel/kvm/kvm-all.c > > > +++ b/accel/kvm/kvm-all.c > > > @@ -86,6 +86,7 @@ struct KVMParkedVcpu { > > > }; > > > > > > KVMState *kvm_state; > > > +bool kvm_has_create_irqchip; > > > bool kvm_kernel_irqchip; > > > bool kvm_split_irqchip; > > > bool kvm_async_interrupts_allowed; > > > @@ -2358,17 +2359,6 @@ static void kvm_irqchip_create(KVMState *s) > > > int ret; > > > > > > assert(s->kernel_irqchip_split != ON_OFF_AUTO_AUTO); > > > - if (kvm_check_extension(s, KVM_CAP_IRQCHIP)) { > > > - ; > > > - } else if (kvm_check_extension(s, KVM_CAP_S390_IRQCHIP)) { > > > - ret = kvm_vm_enable_cap(s, KVM_CAP_S390_IRQCHIP, 0); > > > - if (ret < 0) { > > > - fprintf(stderr, "Enable kernel irqchip failed: %s\n", strerror(-ret)); > > > - exit(1); > > > - } > > > - } else { > > > - return; > > > - } > > > > > > /* First probe and see if there's a arch-specific hook to create the > > > * in-kernel irqchip for us */ > > > @@ -2377,8 +2367,10 @@ static void kvm_irqchip_create(KVMState *s) > > > if (s->kernel_irqchip_split == ON_OFF_AUTO_ON) { > > > error_report("Split IRQ chip mode not supported."); > > > exit(1); > > > - } else { > > > + } else if (kvm_has_create_irqchip) { > > > ret = kvm_vm_ioctl(s, KVM_CREATE_IRQCHIP); > > > + } else { > > > + return; > > > } > > > } > > > if (ret < 0) { > > > diff --git a/include/sysemu/kvm.h b/include/sysemu/kvm.h > > > index 115f0cca79d1..84b1bb3dc91e 100644 > > > --- a/include/sysemu/kvm.h > > > +++ b/include/sysemu/kvm.h > > > @@ -32,6 +32,7 @@ > > > #ifdef CONFIG_KVM_IS_POSSIBLE > > > > > > extern bool kvm_allowed; > > > +extern bool kvm_has_create_irqchip; > > > extern bool kvm_kernel_irqchip; > > > extern bool kvm_split_irqchip; > > > extern bool kvm_async_interrupts_allowed; > > > diff --git a/target/arm/kvm.c b/target/arm/kvm.c > > > index b4c7654f4980..2fa87b495d68 100644 > > > --- a/target/arm/kvm.c > > > +++ b/target/arm/kvm.c > > > @@ -250,6 +250,9 @@ int kvm_arm_get_max_vm_ipa_size(MachineState *ms, bool *fixed_ipa) > > > int kvm_arch_init(MachineState *ms, KVMState *s) > > > { > > > int ret = 0; > > > + > > > + kvm_has_create_irqchip = kvm_check_extension(s, KVM_CAP_IRQCHIP); > > > + > > > /* For ARM interrupt delivery is always asynchronous, > > > * whether we are using an in-kernel VGIC or not. > > > */ > > > diff --git a/target/i386/kvm/kvm.c b/target/i386/kvm/kvm.c > > > index ebfaf3d24c79..6363e67f092d 100644 > > > --- a/target/i386/kvm/kvm.c > > > +++ b/target/i386/kvm/kvm.c > > > @@ -2771,6 +2771,8 @@ int kvm_arch_init(MachineState *ms, KVMState *s) > > > } > > > } > > > > > > + kvm_has_create_irqchip = kvm_check_extension(s, KVM_CAP_IRQCHIP); > > > + > > > return 0; > > > } > > > > > > diff --git a/target/s390x/kvm/kvm.c b/target/s390x/kvm/kvm.c > > > index a9e5880349d9..bcc735227f7d 100644 > > > --- a/target/s390x/kvm/kvm.c > > > +++ b/target/s390x/kvm/kvm.c > > > @@ -391,6 +391,17 @@ int kvm_arch_init(MachineState *ms, KVMState *s) > > > } > > > > > > kvm_set_max_memslot_size(KVM_SLOT_MAX_BYTES); > > > + > > > + kvm_has_create_irqchip = kvm_check_extension(s, KVM_CAP_S390_IRQCHIP); > > > + if (kvm_has_create_irqchip) { > > > + int ret = kvm_vm_enable_cap(s, KVM_CAP_S390_IRQCHIP, 0); > > > + > > > + if (ret < 0) { > > > + fprintf(stderr, "Enable kernel irqchip failed: %s\n", strerror(-ret)); > > > + exit(1); > > > + } > > > + } > > > + > > > return 0; > > > } > > > > > > -- > > > 2.41.0 > > >
diff --git a/accel/kvm/kvm-all.c b/accel/kvm/kvm-all.c index 373d876c0580..cddcb6eca641 100644 --- a/accel/kvm/kvm-all.c +++ b/accel/kvm/kvm-all.c @@ -86,6 +86,7 @@ struct KVMParkedVcpu { }; KVMState *kvm_state; +bool kvm_has_create_irqchip; bool kvm_kernel_irqchip; bool kvm_split_irqchip; bool kvm_async_interrupts_allowed; @@ -2358,17 +2359,6 @@ static void kvm_irqchip_create(KVMState *s) int ret; assert(s->kernel_irqchip_split != ON_OFF_AUTO_AUTO); - if (kvm_check_extension(s, KVM_CAP_IRQCHIP)) { - ; - } else if (kvm_check_extension(s, KVM_CAP_S390_IRQCHIP)) { - ret = kvm_vm_enable_cap(s, KVM_CAP_S390_IRQCHIP, 0); - if (ret < 0) { - fprintf(stderr, "Enable kernel irqchip failed: %s\n", strerror(-ret)); - exit(1); - } - } else { - return; - } /* First probe and see if there's a arch-specific hook to create the * in-kernel irqchip for us */ @@ -2377,8 +2367,10 @@ static void kvm_irqchip_create(KVMState *s) if (s->kernel_irqchip_split == ON_OFF_AUTO_ON) { error_report("Split IRQ chip mode not supported."); exit(1); - } else { + } else if (kvm_has_create_irqchip) { ret = kvm_vm_ioctl(s, KVM_CREATE_IRQCHIP); + } else { + return; } } if (ret < 0) { diff --git a/include/sysemu/kvm.h b/include/sysemu/kvm.h index 115f0cca79d1..84b1bb3dc91e 100644 --- a/include/sysemu/kvm.h +++ b/include/sysemu/kvm.h @@ -32,6 +32,7 @@ #ifdef CONFIG_KVM_IS_POSSIBLE extern bool kvm_allowed; +extern bool kvm_has_create_irqchip; extern bool kvm_kernel_irqchip; extern bool kvm_split_irqchip; extern bool kvm_async_interrupts_allowed; diff --git a/target/arm/kvm.c b/target/arm/kvm.c index b4c7654f4980..2fa87b495d68 100644 --- a/target/arm/kvm.c +++ b/target/arm/kvm.c @@ -250,6 +250,9 @@ int kvm_arm_get_max_vm_ipa_size(MachineState *ms, bool *fixed_ipa) int kvm_arch_init(MachineState *ms, KVMState *s) { int ret = 0; + + kvm_has_create_irqchip = kvm_check_extension(s, KVM_CAP_IRQCHIP); + /* For ARM interrupt delivery is always asynchronous, * whether we are using an in-kernel VGIC or not. */ diff --git a/target/i386/kvm/kvm.c b/target/i386/kvm/kvm.c index ebfaf3d24c79..6363e67f092d 100644 --- a/target/i386/kvm/kvm.c +++ b/target/i386/kvm/kvm.c @@ -2771,6 +2771,8 @@ int kvm_arch_init(MachineState *ms, KVMState *s) } } + kvm_has_create_irqchip = kvm_check_extension(s, KVM_CAP_IRQCHIP); + return 0; } diff --git a/target/s390x/kvm/kvm.c b/target/s390x/kvm/kvm.c index a9e5880349d9..bcc735227f7d 100644 --- a/target/s390x/kvm/kvm.c +++ b/target/s390x/kvm/kvm.c @@ -391,6 +391,17 @@ int kvm_arch_init(MachineState *ms, KVMState *s) } kvm_set_max_memslot_size(KVM_SLOT_MAX_BYTES); + + kvm_has_create_irqchip = kvm_check_extension(s, KVM_CAP_S390_IRQCHIP); + if (kvm_has_create_irqchip) { + int ret = kvm_vm_enable_cap(s, KVM_CAP_S390_IRQCHIP, 0); + + if (ret < 0) { + fprintf(stderr, "Enable kernel irqchip failed: %s\n", strerror(-ret)); + exit(1); + } + } + return 0; }
Since Linux commit 00f918f61c56 ("RISC-V: KVM: Skeletal in-kernel AIA irqchip support") checking KVM_CAP_IRQCHIP returns non-zero when the RISC-V platform has AIA. The cap indicates KVM supports at least one of the following ioctls: KVM_CREATE_IRQCHIP KVM_IRQ_LINE KVM_GET_IRQCHIP KVM_SET_IRQCHIP KVM_GET_LAPIC KVM_SET_LAPIC but the cap doesn't imply that KVM must support any of those ioctls in particular. However, QEMU was assuming the KVM_CREATE_IRQCHIP ioctl was supported. Stop making that assumption by introducing a KVM parameter that each architecture which supports KVM_CREATE_IRQCHIP sets. Adding parameters isn't awesome, but given how the KVM_CAP_IRQCHIP isn't very helpful on its own, we don't have a lot of options. Signed-off-by: Andrew Jones <ajones@ventanamicro.com> --- While this fixes booting guests on riscv KVM with AIA it's unlikely to get merged before the QEMU support for KVM AIA[1] lands, which would also fix the issue. I think this patch is still worth considering though since QEMU's assumption is wrong. [1] https://lore.kernel.org/all/20230714084429.22349-1-yongxuan.wang@sifive.com/ v2: - Move the s390x code to an s390x file. [Thomas] - Drop the KVM_CAP_IRQCHIP check from the top of kvm_irqchip_create(), as it's no longer necessary. accel/kvm/kvm-all.c | 16 ++++------------ include/sysemu/kvm.h | 1 + target/arm/kvm.c | 3 +++ target/i386/kvm/kvm.c | 2 ++ target/s390x/kvm/kvm.c | 11 +++++++++++ 5 files changed, 21 insertions(+), 12 deletions(-)