Skip to contentSkip to conversation

ARM Nested Virt Support

RFC Superseded Eric Auger 4 participants 17 messages Feb 9, 2024 – Mar 26, 2024

v2 of 9 Superseded Latest is v9 →

lore
Share & tools
lore

Eric Auger

author
#0 lore raw

v2 cover ARM Nested Virt Support

This series adds ARM Nested Virtualization support in KVM mode. This is a respin of previous contributions from Miguel [1] and Haibo [2].

This was tested with Marc's v11 [3] on Ampere HW with fedora L1 guest and L2 guests booted without EDK2. However it does not work yet with EDK2 but it looks unrelated to this qemu integration (host hard lockups).

The host needs to be booted with "kvm-arm.mode=nested" option and qemu needs to be invoked with :

-machine virt,virtualization=on

There is a known issue with hosts supporting SVE. Kernel does not support both SVE and NV2 and the current qemu integration has an issue with the scratch_host_vcpu startup because both are enabled if exposed by the kernel. This is independent on whether sve is disabled on the command line. Unfortunately I lost access to the HW that expose that issue so I couldn't fix it in this version.

This series can be found at:
https://github.com/eauger/qemu/tree/v8.2-nv-rfcv2

Previous version from Miguel:
[1] https://lore.kernel.org/all/[email protected]/
Previous version from Haibo:
[2] https://lore.kernel.org/qemu-devel/[email protected]/
[3] Marc's kernel v11 series:

    [PATCH v11 00/43] KVM: arm64: Nested Virtualization support (FEAT_NV2 only)
    https://lore.kernel.org/linux-arm-kernel/[email protected]/T/
    available at: https://github.com/eauger/linux/tree/nv-6.8-nv2-v11

Haibo Xu (5):
[Placeholder] headers: Partial headers update for NV2 enablement hw/arm: Allow setting KVM vGIC maintenance IRQ target/arm/kvm: Add helper to detect EL2 when using KVM target/arm: Enable feature ARM_FEATURE_EL2 if EL2 is supported hw/arm/virt: Allow virt extensions with KVM

 hw/arm/virt.c                      |  6 +++++-
 hw/intc/arm_gicv3_common.c         |  1 +
 hw/intc/arm_gicv3_kvm.c            | 21 +++++++++++++++++++++
 include/hw/intc/arm_gicv3_common.h |  1 +
 linux-headers/asm-arm64/kvm.h      |  1 +
 linux-headers/linux/kvm.h          |  1 +
 target/arm/kvm.c                   | 21 +++++++++++++++++++++
 target/arm/kvm_arm.h               | 12 ++++++++++++
 8 files changed, 63 insertions(+), 1 deletion(-)
Signature · 1 line
2.41.0
author
#1 lore raw

RFC v2 1/5 headers: Partial headers update for NV2 enablement

From: Haibo Xu <[email protected]>

For now let's only import the pieces needed to run NV on KVM. Later on this will be replaced by a standard and comprehensive linux header update using scripts/update-linux-headers.sh.

For now the changes are taken from
https://github.com/eauger/linux/tree/nv-6.8-nv2-v11

---
 linux-headers/asm-arm64/kvm.h | 1 +
 linux-headers/linux/kvm.h     | 1 +
 2 files changed, 2 insertions(+)
linux-headers/asm-arm64/kvm.h +1 -0
index c59ea55cd8..d46839f1d9 100644
@@ -394,6 +394,7 @@ enum {
 #define KVM_DEV_ARM_VGIC_GRP_CPU_SYSREGS 6
 #define KVM_DEV_ARM_VGIC_GRP_LEVEL_INFO  7
 #define KVM_DEV_ARM_VGIC_GRP_ITS_REGS 8
+#define KVM_DEV_ARM_VGIC_GRP_MAINT_IRQ  9
 #define KVM_DEV_ARM_VGIC_LINE_LEVEL_INFO_SHIFT	10
 #define KVM_DEV_ARM_VGIC_LINE_LEVEL_INFO_MASK \
 			(0x3fffffULL << KVM_DEV_ARM_VGIC_LINE_LEVEL_INFO_SHIFT)
linux-headers/linux/kvm.h +1 -0
index 549fea3a97..5af601d7b8 100644
@@ -1197,6 +1197,7 @@ struct kvm_ppc_resize_hpt {
 #define KVM_CAP_ARM_EAGER_SPLIT_CHUNK_SIZE 228
 #define KVM_CAP_ARM_SUPPORTED_BLOCK_SIZES 229
 #define KVM_CAP_ARM_SUPPORTED_REG_MASK_RANGES 230
+#define KVM_CAP_ARM_EL2 231
 
 #ifdef KVM_CAP_IRQ_ROUTING
 
Signature · 1 line
2.41.0
author
#2 lore raw

RFC v2 2/5 hw/arm: Allow setting KVM vGIC maintenance IRQ

In reply to Eric Auger

From: Haibo Xu <[email protected]>

Allow virt arm machine to set the intid for the KVM GIC maintenance interrupt.

---
v1 -> v2:
- [Miguel] replaced the has_virt_extensions by the maintenance irq intid property. [Eric] restored kvm_device_check_attr and kvm_device_access standard usage and conditionally call those if the prop is set.

---
 hw/arm/virt.c                      |  3 +++
 hw/intc/arm_gicv3_common.c         |  1 +
 hw/intc/arm_gicv3_kvm.c            | 21 +++++++++++++++++++++
 include/hw/intc/arm_gicv3_common.h |  1 +
 4 files changed, 26 insertions(+)
hw/arm/virt.c +3 -0
index 368c2a415a..5214aca898 100644
@@ -750,6 +750,9 @@ static void create_gic(VirtMachineState *vms, MemoryRegion *mem)
                                          OBJECT(mem), &error_fatal);
                 qdev_prop_set_bit(vms->gic, "has-lpi", true);
             }
+        } else {
+       qdev_prop_set_uint32(vms->gic, "maintenance-interrupt-id",
+                            ARCH_GIC_MAINT_IRQ);
         }
     } else {
         if (!kvm_irqchip_in_kernel()) {
hw/intc/arm_gicv3_common.c +1 -0
index cb55c72681..df056dc35c 100644
@@ -564,6 +564,7 @@ static Property arm_gicv3_common_properties[] = {
     DEFINE_PROP_UINT32("revision", GICv3State, revision, 3),
     DEFINE_PROP_BOOL("has-lpi", GICv3State, lpi_enable, 0),
     DEFINE_PROP_BOOL("has-security-extensions", GICv3State, security_extn, 0),
+    DEFINE_PROP_UINT32("maintenance-interrupt-id", GICv3State, maint_irq, 0),
     /*
      * Compatibility property: force 8 bits of physical priority, even
      * if the CPU being emulated should have fewer.
hw/intc/arm_gicv3_kvm.c +21 -0
index 77eb37e131..23fad60515 100644
@@ -22,6 +22,7 @@
 #include "qemu/osdep.h"
 #include "qapi/error.h"
 #include "hw/intc/arm_gicv3_common.h"
+#include "hw/arm/virt.h"
 #include "qemu/error-report.h"
 #include "qemu/module.h"
 #include "sysemu/kvm.h"
@@ -820,6 +821,26 @@ static void kvm_arm_gicv3_realize(DeviceState *dev, Error **errp)
         return;
     }
 
+    if (s->maint_irq) {
+        int ret;
+
+        ret = kvm_device_check_attr(s->dev_fd,
+                                    KVM_DEV_ARM_VGIC_GRP_MAINT_IRQ, 0);
+        if (!ret) {
+            error_setg_errno(errp, errno,
+                             "VGICv3 setting maintenance IRQ is not "
+                             "supported by this host kernel");
+            return;
+        }
+
+        ret = kvm_device_access(s->dev_fd, KVM_DEV_ARM_VGIC_GRP_MAINT_IRQ, 0,
+                                &s->maint_irq, true, errp);
+        if (ret) {
+            error_setg_errno(errp, errno, "Failed to set VGIC maintenance IRQ");
+            return;
+       }
+    }
+
     multiple_redist_region_allowed =
         kvm_device_check_attr(s->dev_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
                               KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION);
include/hw/intc/arm_gicv3_common.h +1 -0
index 4e2fb518e7..4ff421a165 100644
@@ -246,6 +246,7 @@ struct GICv3State {
     uint32_t num_cpu;
     uint32_t num_irq;
     uint32_t revision;
+    uint32_t maint_irq;
     bool lpi_enable;
     bool security_extn;
     bool force_8bit_prio;
Signature · 1 line
2.41.0
#3 lore raw

Re: [RFC v2 2/5] hw/arm: Allow setting KVM vGIC maintenance IRQ

Quoted 10 lines · Eric Auger

From: Haibo Xu <[email protected]>

Allow virt arm machine to set the intid for the KVM GIC maintenance interrupt.

Quoted 6 lines

---
v1 -> v2:
- [Miguel] replaced the has_virt_extensions by the maintenance irq intid property. [Eric] restored kvm_device_check_attr and kvm_device_access standard usage and conditionally call those if the prop is set

This seems reasonable, but it's not the same way we opted to handle telling the kernel the IRQ number for the PMU interrupt (where we use kvm_arm_pmu_set_irq()). I guess we have to do it this way because it's a device attr so we need to set it in gic realize, though?

By the way, does the kernel automatically complain and fail if we try to enable nested-virt with a GICv2 or with a userspace GIC, or do we need to catch and produce error messages for those (invalid) combinations ourselves?

thanks
-- PMM

author
#4 lore raw

Re: [RFC v2 2/5] hw/arm: Allow setting KVM vGIC maintenance IRQ

Hi Peter,

Quoted 10 lines · Peter Maydell
Quoted 9 lines · Eric Auger

From: Haibo Xu <[email protected]>

Allow virt arm machine to set the intid for the KVM GIC maintenance interrupt.

Quoted 6 lines
Quoted 6 lines

---
v1 -> v2:
- [Miguel] replaced the has_virt_extensions by the maintenance irq intid property. [Eric] restored kvm_device_check_attr and kvm_device_access standard usage and conditionally call those if the prop is set

Please forgive me for the delay

Quoted 5 lines

This seems reasonable, but it's not the same way we opted to handle telling the kernel the IRQ number for the PMU interrupt (where we use kvm_arm_pmu_set_irq()). I guess we have to do it this way because it's a device attr so we need to set it in gic realize, though?

This cannot follow the same pattern as the

kvm_arm_pmu_set_irq() because the maintenance irq must be set between before the GICv3 KVM device creation and the KVM_DEV_ARM_VGIC_CTRL_INIT. The GICv3 realize function calls both so I cannot set the maintenance after the realize. It would fail with -EBUSY. Hope this helps.

Thanks

Eric

Quoted 9 lines

By the way, does the kernel automatically complain and fail if we try to enable nested-virt with a GICv2 or with a userspace GIC, or do we need to catch and produce error messages for those (invalid) combinations ourselves?

thanks
-- PMM

author
#5 lore raw

Re: [RFC v2 2/5] hw/arm: Allow setting KVM vGIC maintenance IRQ

In reply to Peter Maydell

Hi Peter,

Quoted 22 lines · Peter Maydell
Quoted 9 lines · Eric Auger

From: Haibo Xu <[email protected]>

Allow virt arm machine to set the intid for the KVM GIC maintenance interrupt.

Quoted 6 lines

---
v1 -> v2:
- [Miguel] replaced the has_virt_extensions by the maintenance irq intid property. [Eric] restored kvm_device_check_attr and kvm_device_access standard usage and conditionally call those if the prop is set

This seems reasonable, but it's not the same way we opted to handle telling the kernel the IRQ number for the PMU interrupt (where we use kvm_arm_pmu_set_irq()). I guess we have to do it this way because it's a device attr so we need to set it in gic realize, though?

Quoted 4 lines

By the way, does the kernel automatically complain and fail if we try to enable nested-virt with a GICv2 or with a userspace GIC, or do we need to catch and produce error messages for those (invalid) combinations ourselves?

I don't think there is any check of that kind in Marc's series yet. This may be added if GICv2 KVM device is created while kvm_mode is set to KVM_MODE_NV.

Wrt userspace irqchip compat, KVM_CAP_ARM_USER_IRQ extension may not be exposed in case of nested virt.

Marc, is that something you would like to integrate into the kernel series?

Thanks

Eric

Quoted 4 lines

thanks
-- PMM

author
#6 lore raw

RFC v2 3/5 target/arm/kvm: Add helper to detect EL2 when using KVM

In reply to Eric Auger

From: Haibo Xu <[email protected]>

Introduce query support for KVM_CAP_ARM_EL2.

---
 target/arm/kvm.c     |  5 +++++
 target/arm/kvm_arm.h | 12 ++++++++++++
 2 files changed, 17 insertions(+)
target/arm/kvm.c +5 -0
index 81813030a5..0996866afe 100644
@@ -1791,6 +1791,11 @@ bool kvm_arm_aarch32_supported(void)
     return kvm_check_extension(kvm_state, KVM_CAP_ARM_EL1_32BIT);
 }
 
+bool kvm_arm_el2_supported(void)
+{
+    return kvm_check_extension(kvm_state, KVM_CAP_ARM_EL2);
+}
+
 bool kvm_arm_sve_supported(void)
 {
     return kvm_check_extension(kvm_state, KVM_CAP_ARM_SVE);
target/arm/kvm_arm.h +12 -0
index cfaa0d9bc7..36e4b37ec0 100644
@@ -188,6 +188,13 @@ bool kvm_arm_pmu_supported(void);
  */
 bool kvm_arm_sve_supported(void);
 
+/**
+ * kvm_arm_el2_supported:
+ *
+ * Returns true if KVM can enable EL2 and false otherwise.
+ */
+bool kvm_arm_el2_supported(void);
+
 /**
  * kvm_arm_get_max_vm_ipa_size:
  * @ms: Machine state handle
@@ -235,6 +242,11 @@ static inline bool kvm_arm_sve_supported(void)
     return false;
 }
 
+static inline bool kvm_arm_el2_supported(void)
+{
+    return false;
+}
+
 /*
  * These functions should never actually be called without KVM support.
  */
Signature · 1 line
2.41.0
Reviewed-by on v2 3/5
#7 lore raw

Re: [RFC v2 3/5] target/arm/kvm: Add helper to detect EL2 when using KVM

Quoted 4 lines · Eric Auger

From: Haibo Xu <[email protected]>

Introduce query support for KVM_CAP_ARM_EL2.

Quoted 8 lines
---
  target/arm/kvm.c     |  5 +++++
  target/arm/kvm_arm.h | 12 ++++++++++++
  2 files changed, 17 insertions(+)
author
#8 lore raw

RFC v2 4/5 target/arm: Enable feature ARM_FEATURE_EL2 if EL2 is supported

In reply to Eric Auger

From: Haibo Xu <[email protected]>

KVM_CAP_ARM_EL2 must be supported by the cpu to enable ARM_FEATURE_EL2. In case the host does support NV, expose the feature.

---

v1 -> v2:
- remove isar_feature_aa64_aa32_el2 modif in target/arm/cpu.h [Richard] and use el2_supported in kvm_arch_init_vcpu

---
 target/arm/kvm.c | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)
target/arm/kvm.c +16 -0
index 0996866afe..a08bc68a3f 100644
@@ -238,6 +238,7 @@ static bool kvm_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf)
      */
     int fdarray[3];
     bool sve_supported;
+    bool el2_supported;
     bool pmu_supported = false;
     uint64_t features = 0;
     int err;
@@ -268,6 +269,14 @@ static bool kvm_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf)
         init.features[0] |= 1 << KVM_ARM_VCPU_SVE;
     }
 
+    /*
+     * Ask for EL2 if supported.
+     */
+    el2_supported = kvm_arm_el2_supported();
+    if (el2_supported) {
+        init.features[0] |= 1 << KVM_ARM_VCPU_HAS_EL2;
+    }
+
     /*
      * Ask for Pointer Authentication if supported, so that we get
      * the unsanitized field values for AA64ISAR1_EL1.
@@ -449,6 +458,10 @@ static bool kvm_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf)
     features |= 1ULL << ARM_FEATURE_PMU;
     features |= 1ULL << ARM_FEATURE_GENERIC_TIMER;
 
+    if (el2_supported) {
+        features |= 1ULL << ARM_FEATURE_EL2;
+    }
+
     ahcf->features = features;
 
     return true;
@@ -1912,6 +1925,9 @@ int kvm_arch_init_vcpu(CPUState *cs)
         cpu->kvm_init_features[0] |= (1 << KVM_ARM_VCPU_PTRAUTH_ADDRESS |
                                       1 << KVM_ARM_VCPU_PTRAUTH_GENERIC);
     }
+    if (kvm_arm_el2_supported()) {
+        cpu->kvm_init_features[0] |= 1 << KVM_ARM_VCPU_HAS_EL2;
+    }
 
     /* Do KVM_ARM_VCPU_INIT ioctl */
     ret = kvm_arm_vcpu_init(cpu);
Signature · 1 line
2.41.0
#9 lore raw

Re: [RFC v2 4/5] target/arm: Enable feature ARM_FEATURE_EL2 if EL2 is supported

Quoted 58 lines · Eric Auger

From: Haibo Xu <[email protected]>

KVM_CAP_ARM_EL2 must be supported by the cpu to enable ARM_FEATURE_EL2. In case the host does support NV, expose the feature.

---

v1 -> v2:
- remove isar_feature_aa64_aa32_el2 modif in target/arm/cpu.h [Richard] and use el2_supported in kvm_arch_init_vcpu

---
 target/arm/kvm.c | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)
target/arm/kvm.c +13 -0
index 0996866afe..a08bc68a3f 100644
@@ -238,6 +238,7 @@ static bool kvm_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf)
      */
     int fdarray[3];
     bool sve_supported;
+    bool el2_supported;
     bool pmu_supported = false;
     uint64_t features = 0;
     int err;
@@ -268,6 +269,14 @@ static bool kvm_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf)
         init.features[0] |= 1 << KVM_ARM_VCPU_SVE;
     }
 
+    /*
+     * Ask for EL2 if supported.
+     */
+    el2_supported = kvm_arm_el2_supported();
+    if (el2_supported) {
+        init.features[0] |= 1 << KVM_ARM_VCPU_HAS_EL2;
+    }
+
     /*
      * Ask for Pointer Authentication if supported, so that we get
      * the unsanitized field values for AA64ISAR1_EL1.
@@ -449,6 +458,10 @@ static bool kvm_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf)
     features |= 1ULL << ARM_FEATURE_PMU;
     features |= 1ULL << ARM_FEATURE_GENERIC_TIMER;
 
+    if (el2_supported) {
+        features |= 1ULL << ARM_FEATURE_EL2;
+    }
+
     ahcf->features = features;
 
     return true;

target/arm/kvm.c int kvm_arch_init_vcpu(CPUState *cs) · line 1933

Quoted 9 lines
target/arm/kvm.c +3 -0
@@ -1912,6 +1925,9 @@ int kvm_arch_init_vcpu(CPUState *cs)
         cpu->kvm_init_features[0] |= (1 << KVM_ARM_VCPU_PTRAUTH_ADDRESS |
                                       1 << KVM_ARM_VCPU_PTRAUTH_GENERIC);
     }
+    if (kvm_arm_el2_supported()) {
+        cpu->kvm_init_features[0] |= 1 << KVM_ARM_VCPU_HAS_EL2;
+    }
 
     /* Do KVM_ARM_VCPU_INIT ioctl */
     ret = kvm_arm_vcpu_init(cpu);

Am I reading this right that if the kernel supports FEAT_NV then we will always ask for a vCPU with that feature? Is that a good idea, or should we arrange to only do it if the user uses the 'virtualization=on' option to -M virt ? (Or does that happen already in some way I'm not seeing?)

thanks
-- PMM

author
#10 lore raw

Re: [RFC v2 4/5] target/arm: Enable feature ARM_FEATURE_EL2 if EL2 is supported

Hi Peter,

Quoted 65 lines · Peter Maydell
Quoted 64 lines · Eric Auger

From: Haibo Xu <[email protected]>

KVM_CAP_ARM_EL2 must be supported by the cpu to enable ARM_FEATURE_EL2. In case the host does support NV, expose the feature.

---

v1 -> v2:
- remove isar_feature_aa64_aa32_el2 modif in target/arm/cpu.h [Richard] and use el2_supported in kvm_arch_init_vcpu

---
 target/arm/kvm.c | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)
target/arm/kvm.c +16 -0
index 0996866afe..a08bc68a3f 100644
@@ -238,6 +238,7 @@ static bool kvm_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf)
      */
     int fdarray[3];
     bool sve_supported;
+    bool el2_supported;
     bool pmu_supported = false;
     uint64_t features = 0;
     int err;
@@ -268,6 +269,14 @@ static bool kvm_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf)
         init.features[0] |= 1 << KVM_ARM_VCPU_SVE;
     }
 
+    /*
+     * Ask for EL2 if supported.
+     */
+    el2_supported = kvm_arm_el2_supported();
+    if (el2_supported) {
+        init.features[0] |= 1 << KVM_ARM_VCPU_HAS_EL2;
+    }
+
     /*
      * Ask for Pointer Authentication if supported, so that we get
      * the unsanitized field values for AA64ISAR1_EL1.
@@ -449,6 +458,10 @@ static bool kvm_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf)
     features |= 1ULL << ARM_FEATURE_PMU;
     features |= 1ULL << ARM_FEATURE_GENERIC_TIMER;
 
+    if (el2_supported) {
+        features |= 1ULL << ARM_FEATURE_EL2;
+    }
+
     ahcf->features = features;
 
     return true;
@@ -1912,6 +1925,9 @@ int kvm_arch_init_vcpu(CPUState *cs)
         cpu->kvm_init_features[0] |= (1 << KVM_ARM_VCPU_PTRAUTH_ADDRESS |
                                       1 << KVM_ARM_VCPU_PTRAUTH_GENERIC);
     }
+    if (kvm_arm_el2_supported()) {
+        cpu->kvm_init_features[0] |= 1 << KVM_ARM_VCPU_HAS_EL2;
+    }
 
Quoted 7 lines

target/arm/kvm.c int kvm_arch_init_vcpu(CPUState *cs) · line 1933

Quoted 2 lines
target/arm/kvm.c +0 -0
     /* Do KVM_ARM_VCPU_INIT ioctl */
     ret = kvm_arm_vcpu_init(cpu);

Am I reading this right that if the kernel supports FEAT_NV then we will always ask for a vCPU with that feature? Is that a good idea, or should we arrange to only do it if the user uses the 'virtualization=on' option to -M virt ? (Or does that happen already in some way I'm not seeing?)

yes you're right, if the host supports it, the feature is currently set on the vcpu. I am not totaly clear under which conditions the features shall be instantiated in the scratch VM and when the host passthrough model shall be altered by machine option.

Thanks

Eric

Quoted 4 lines

thanks
-- PMM

author
#11 lore raw

RFC v2 5/5 hw/arm/virt: Allow virt extensions with KVM

In reply to Eric Auger

From: Haibo Xu <[email protected]>

Up to now virt support on guest has been only supported with TCG. Now it becomes feasible to use it with KVM acceleration.

---

v1 -> v2:
- fixed test ordering: virt && ((kvm && !kvm_el2) || hvf) [Richard]
- tweeked the commit title & message

---
 hw/arm/virt.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
hw/arm/virt.c +2 -1
index 5214aca898..ae7ac07301 100644
@@ -2109,7 +2109,8 @@ static void machvirt_init(MachineState *machine)
         exit(1);
     }
 
-    if (vms->virt && (kvm_enabled() || hvf_enabled())) {
+    if (vms->virt &&
+        ((kvm_enabled() && !kvm_arm_el2_supported()) || hvf_enabled())) {
         error_report("mach-virt: %s does not support providing "
                      "Virtualization extensions to the guest CPU",
                      current_accel_name());
Signature · 1 line
2.41.0
#12 lore raw
In reply to Eric Auger
Quoted 11 lines · Eric Auger

This series adds ARM Nested Virtualization support in KVM mode. This is a respin of previous contributions from Miguel [1] and Haibo [2].

This was tested with Marc's v11 [3] on Ampere HW with fedora L1 guest and L2 guests booted without EDK2. However it does not work yet with EDK2 but it looks unrelated to this qemu integration (host hard lockups).

The host needs to be booted with "kvm-arm.mode=nested" option and qemu needs to be invoked with :

Quoted 8 lines

-machine virt,virtualization=on

There is a known issue with hosts supporting SVE. Kernel does not support both SVE and NV2 and the current qemu integration has an issue with the scratch_host_vcpu startup because both are enabled if exposed by the kernel. This is independent on whether sve is disabled on the command line. Unfortunately I lost access to the HW that expose that issue so I couldn't fix it in this version.

You can probably repro that by running the whole setup under QEMU's FEAT_NV emulation, which will be able to give you a CPU with both FEAT_NV and SVE.

Personally I think that this is a kernel missing-feature that should really be fixed as part of getting the kernel patches upstreamed. There's no cause to force every userspace VMM to develop extra complications for this.

thanks
-- PMM

author
#13 lore raw

Hi Peter,

Quoted 19 lines · Peter Maydell
Quoted 10 lines · Eric Auger

This series adds ARM Nested Virtualization support in KVM mode. This is a respin of previous contributions from Miguel [1] and Haibo [2].

This was tested with Marc's v11 [3] on Ampere HW with fedora L1 guest and L2 guests booted without EDK2. However it does not work yet with EDK2 but it looks unrelated to this qemu integration (host hard lockups).

The host needs to be booted with "kvm-arm.mode=nested" option and qemu needs to be invoked with :

Quoted 8 lines

-machine virt,virtualization=on

There is a known issue with hosts supporting SVE. Kernel does not support both SVE and NV2 and the current qemu integration has an issue with the scratch_host_vcpu startup because both are enabled if exposed by the kernel. This is independent on whether sve is disabled on the command line. Unfortunately I lost access to the HW that expose that issue so I couldn't fix it in this version.

Quoted 3 lines

You can probably repro that by running the whole setup under QEMU's FEAT_NV emulation, which will be able to give you a CPU with both FEAT_NV and SVE.

indeed, this should work now we have FEAT_NV.

Quoted 5 lines

Personally I think that this is a kernel missing-feature that should really be fixed as part of getting the kernel patches upstreamed. There's no cause to force every userspace VMM to develop extra complications for this.

yes maybe this will be fixed later on.

Thanks!

Eric

Quoted 4 lines

thanks
-- PMM

#14 lore raw
In reply to Peter Maydell
Quoted 21 lines · Peter Maydell
Quoted 11 lines · Eric Auger

This series adds ARM Nested Virtualization support in KVM mode. This is a respin of previous contributions from Miguel [1] and Haibo [2].

This was tested with Marc's v11 [3] on Ampere HW with fedora L1 guest and L2 guests booted without EDK2. However it does not work yet with EDK2 but it looks unrelated to this qemu integration (host hard lockups).

The host needs to be booted with "kvm-arm.mode=nested" option and qemu needs to be invoked with :

Quoted 8 lines

-machine virt,virtualization=on

There is a known issue with hosts supporting SVE. Kernel does not support both SVE and NV2 and the current qemu integration has an issue with the scratch_host_vcpu startup because both are enabled if exposed by the kernel. This is independent on whether sve is disabled on the command line. Unfortunately I lost access to the HW that expose that issue so I couldn't fix it in this version.

Quoted 8 lines

You can probably repro that by running the whole setup under QEMU's FEAT_NV emulation, which will be able to give you a CPU with both FEAT_NV and SVE.

Personally I think that this is a kernel missing-feature that should really be fixed as part of getting the kernel patches upstreamed. There's no cause to force every userspace VMM to develop extra complications for this.

I don't plan to make NV visible to userspace before this is fixed. Which may delay KVM NV by another year or five, but I don't think anyone is really waiting for it anyway.

        M.
Signature · 1 line
Jazz is not dead. It just smells funny...
#15 lore raw
In reply to Eric Auger
Quoted 39 lines · Eric Auger

This series adds ARM Nested Virtualization support in KVM mode. This is a respin of previous contributions from Miguel [1] and Haibo [2].

This was tested with Marc's v11 [3] on Ampere HW with fedora L1 guest and L2 guests booted without EDK2. However it does not work yet with EDK2 but it looks unrelated to this qemu integration (host hard lockups).

The host needs to be booted with "kvm-arm.mode=nested" option and qemu needs to be invoked with :

-machine virt,virtualization=on

There is a known issue with hosts supporting SVE. Kernel does not support both SVE and NV2 and the current qemu integration has an issue with the scratch_host_vcpu startup because both are enabled if exposed by the kernel. This is independent on whether sve is disabled on the command line. Unfortunately I lost access to the HW that expose that issue so I couldn't fix it in this version.

This series can be found at:
https://github.com/eauger/qemu/tree/v8.2-nv-rfcv2

Previous version from Miguel:
[1] https://lore.kernel.org/all/[email protected]/
Previous version from Haibo:
[2] https://lore.kernel.org/qemu-devel/[email protected]/
[3] Marc's kernel v11 series:

    [PATCH v11 00/43] KVM: arm64: Nested Virtualization support (FEAT_NV2 only)
    https://lore.kernel.org/linux-arm-kernel/[email protected]/T/
    available at: https://github.com/eauger/linux/tree/nv-6.8-nv2-v11

Haibo Xu (5):
[Placeholder] headers: Partial headers update for NV2 enablement hw/arm: Allow setting KVM vGIC maintenance IRQ target/arm/kvm: Add helper to detect EL2 when using KVM target/arm: Enable feature ARM_FEATURE_EL2 if EL2 is supported hw/arm/virt: Allow virt extensions with KVM

Quoted 9 lines
 hw/arm/virt.c                      |  6 +++++-
 hw/intc/arm_gicv3_common.c         |  1 +
 hw/intc/arm_gicv3_kvm.c            | 21 +++++++++++++++++++++
 include/hw/intc/arm_gicv3_common.h |  1 +
 linux-headers/asm-arm64/kvm.h      |  1 +
 linux-headers/linux/kvm.h          |  1 +
 target/arm/kvm.c                   | 21 +++++++++++++++++++++
 target/arm/kvm_arm.h               | 12 ++++++++++++
 8 files changed, 63 insertions(+), 1 deletion(-)

All the patches in this series seem reasonable, but the series as a whole is so short I wonder if we're missing something :-) Does migration Just Work? (I guess as long as the kernel exposes all the EL2 sysregs via the ONE_REG ioctl interface it ought to...)

Anyway, I don't think there's anything that stood out as needing major changes, so for now I guess we just wait for whenever the KVM side patches eventually land.

thanks
-- PMM

author
#16 lore raw

Hi Peter,

Quoted 48 lines · Peter Maydell
Quoted 38 lines · Eric Auger

This series adds ARM Nested Virtualization support in KVM mode. This is a respin of previous contributions from Miguel [1] and Haibo [2].

This was tested with Marc's v11 [3] on Ampere HW with fedora L1 guest and L2 guests booted without EDK2. However it does not work yet with EDK2 but it looks unrelated to this qemu integration (host hard lockups).

The host needs to be booted with "kvm-arm.mode=nested" option and qemu needs to be invoked with :

-machine virt,virtualization=on

There is a known issue with hosts supporting SVE. Kernel does not support both SVE and NV2 and the current qemu integration has an issue with the scratch_host_vcpu startup because both are enabled if exposed by the kernel. This is independent on whether sve is disabled on the command line. Unfortunately I lost access to the HW that expose that issue so I couldn't fix it in this version.

This series can be found at:
https://github.com/eauger/qemu/tree/v8.2-nv-rfcv2

Previous version from Miguel:
[1] https://lore.kernel.org/all/[email protected]/
Previous version from Haibo:
[2] https://lore.kernel.org/qemu-devel/[email protected]/
[3] Marc's kernel v11 series:

    [PATCH v11 00/43] KVM: arm64: Nested Virtualization support (FEAT_NV2 only)
    https://lore.kernel.org/linux-arm-kernel/[email protected]/T/
    available at: https://github.com/eauger/linux/tree/nv-6.8-nv2-v11

Haibo Xu (5):
[Placeholder] headers: Partial headers update for NV2 enablement hw/arm: Allow setting KVM vGIC maintenance IRQ target/arm/kvm: Add helper to detect EL2 when using KVM target/arm: Enable feature ARM_FEATURE_EL2 if EL2 is supported hw/arm/virt: Allow virt extensions with KVM

Quoted 9 lines
 hw/arm/virt.c                      |  6 +++++-
 hw/intc/arm_gicv3_common.c         |  1 +
 hw/intc/arm_gicv3_kvm.c            | 21 +++++++++++++++++++++
 include/hw/intc/arm_gicv3_common.h |  1 +
 linux-headers/asm-arm64/kvm.h      |  1 +
 linux-headers/linux/kvm.h          |  1 +
 target/arm/kvm.c                   | 21 +++++++++++++++++++++
 target/arm/kvm_arm.h               | 12 ++++++++++++
 8 files changed, 63 insertions(+), 1 deletion(-)
Quoted 4 lines

All the patches in this series seem reasonable, but the series as a whole is so short I wonder if we're missing something :-) Does migration Just Work? (I guess as long as the kernel exposes all the EL2 sysregs via the ONE_REG ioctl interface it ought to...)

To be honest I have not tested yet. I would also guess there is no mig blocker but that definitively deserves some testing. I will check before sending the next version.

Quoted 4 lines

Anyway, I don't think there's anything that stood out as needing major changes, so for now I guess we just wait for whenever the KVM side patches eventually land.

yup

Thanks

Eric

Quoted 4 lines

thanks
-- PMM