Skip to contentSkip to conversation

KVM: arm64: vgic: Missing checks for REDIST/CPU and ITS regions above the VM IPA size

PATCH Superseded Ricardo Koller 3 participants 38 messages Sep 28, 2021 – Oct 5, 2021

Subsystems (4)

v3 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

v3 cover KVM: arm64: vgic: Missing checks for REDIST/CPU and ITS regions above the VM IPA size

KVM doesn't check for redist, CPU interface, and ITS regions that extend partially above the guest addressable IPA range (phys_size). This can happen when using the V[2|3]_ADDR_TYPE_CPU, ADDR_TYPE_REDIST[_REGION], or ITS_ADDR_TYPE attributes to set a new region that extends partially above phys_size (with the base below phys_size). The issue is that vcpus can potentially run into a situation where some redistributors are addressable and others are not, or just the first half of the ITS is addressable.

Patches 1-4 fixes the issue for GICv2 and GICv3 (and the ITS). Patches 5-10 add some selftests for all these fixes. While adding these tests, these add support for some extra GICv2 and ITS device tests.

Changes:
v3: add missing checks for GICv2 and the ITS, plus tests for the fixes.
v2: adding a test for KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, and returning E2BIG
instead of EINVAL (thanks Alexandru and Eric).

Ricardo Koller (10):
kvm: arm64: vgic: Introduce vgic_check_iorange
KVM: arm64: vgic-v3: Check redist region is not above the VM IPA size
KVM: arm64: vgic-v2: Check cpu interface region is not above the VM
IPA size
KVM: arm64: vgic-v3: Check ITS region is not above the VM IPA size
KVM: arm64: selftests: Make vgic_init gic version agnostic
KVM: arm64: selftests: Make vgic_init/vm_gic_create version agnostic
KVM: arm64: selftests: Add some tests for GICv2 in vgic_init
KVM: arm64: selftests: Add tests for GIC redist/cpuif partially above
IPA range
KVM: arm64: selftests: Add test for legacy GICv3 REDIST base partially
above IPA range
KVM: arm64: selftests: Add basic ITS device tests

 arch/arm64/kvm/vgic/vgic-its.c                |   4 +-
 arch/arm64/kvm/vgic/vgic-kvm-device.c         |  29 +-
 arch/arm64/kvm/vgic/vgic-mmio-v3.c            |   6 +-
 arch/arm64/kvm/vgic/vgic-v3.c                 |   4 +
 arch/arm64/kvm/vgic/vgic.h                    |   4 +
 .../testing/selftests/kvm/aarch64/vgic_init.c | 372 +++++++++++++-----
 6 files changed, 317 insertions(+), 102 deletions(-)
Signature · 1 line
2.33.0.685.g46640cef36-goog
#1 lore raw

PATCH v3 1/10 kvm: arm64: vgic: Introduce vgic_check_iorange

Add the new vgic_check_iorange helper that checks that an iorange is
sane: the start address and size have valid alignments, the range is within the addressable PA range, start+size doesn't overflow, and the start wasn't already defined.

No functional change.

---
 arch/arm64/kvm/vgic/vgic-kvm-device.c | 22 ++++++++++++++++++++++
 arch/arm64/kvm/vgic/vgic.h            |  4 ++++
 2 files changed, 26 insertions(+)
arch/arm64/kvm/vgic/vgic-kvm-device.c +22 -0
index 7740995de982..f714aded67b2 100644
@@ -29,6 +29,28 @@ int vgic_check_ioaddr(struct kvm *kvm, phys_addr_t *ioaddr,
 	return 0;
 }
 
+int vgic_check_iorange(struct kvm *kvm, phys_addr_t *ioaddr,
+		       phys_addr_t addr, phys_addr_t alignment,
+		       phys_addr_t size)
+{
+	int ret;
+
+	ret = vgic_check_ioaddr(kvm, ioaddr, addr, alignment);
+	if (ret)
+		return ret;
+
+	if (!IS_ALIGNED(size, alignment))
+		return -EINVAL;
+
+	if (addr + size < addr)
+		return -EINVAL;
+
+	if (addr + size > kvm_phys_size(kvm))
+		return -E2BIG;
+
+	return 0;
+}
+
 static int vgic_check_type(struct kvm *kvm, int type_needed)
 {
 	if (kvm->arch.vgic.vgic_model != type_needed)
arch/arm64/kvm/vgic/vgic.h +4 -0
index 14a9218641f5..c4df4dcef31f 100644
@@ -175,6 +175,10 @@ void vgic_irq_handle_resampling(struct vgic_irq *irq,
 int vgic_check_ioaddr(struct kvm *kvm, phys_addr_t *ioaddr,
 		      phys_addr_t addr, phys_addr_t alignment);
 
+int vgic_check_iorange(struct kvm *kvm, phys_addr_t *ioaddr,
+		       phys_addr_t addr, phys_addr_t alignment,
+		       phys_addr_t size);
+
 void vgic_v2_fold_lr_state(struct kvm_vcpu *vcpu);
 void vgic_v2_populate_lr(struct kvm_vcpu *vcpu, struct vgic_irq *irq, int lr);
 void vgic_v2_clear_lr(struct kvm_vcpu *vcpu, int lr);
Signature · 1 line
2.33.0.685.g46640cef36-goog
Reviewed-by on v3 1/10
#2 lore raw

Re: [PATCH v3 01/10] kvm: arm64: vgic: Introduce vgic_check_iorange

Hi Ricardo,

Quoted 18 lines · Ricardo Koller

Add the new vgic_check_iorange helper that checks that an iorange is
sane: the start address and size have valid alignments, the range is within the addressable PA range, start+size doesn't overflow, and the start wasn't already defined.

No functional change.

---
 arch/arm64/kvm/vgic/vgic-kvm-device.c | 22 ++++++++++++++++++++++
 arch/arm64/kvm/vgic/vgic.h            |  4 ++++
 2 files changed, 26 insertions(+)
arch/arm64/kvm/vgic/vgic-kvm-device.c +0 -0
index 7740995de982..f714aded67b2 100644

arch/arm64/kvm/vgic/vgic-kvm-device.c int vgic_check_ioaddr(struct kvm *kvm, phys_addr_t *ioaddr, · line 38

Quoted 10 lines
arch/arm64/kvm/vgic/vgic-kvm-device.c +7 -0
@@ -29,6 +29,28 @@ int vgic_check_ioaddr(struct kvm *kvm, phys_addr_t *ioaddr,
 	return 0;
 }
 
+int vgic_check_iorange(struct kvm *kvm, phys_addr_t *ioaddr,
+		       phys_addr_t addr, phys_addr_t alignment,
+		       phys_addr_t size)
+{
+	int ret;
+
+	ret = vgic_check_ioaddr(kvm, ioaddr, addr, alignment);

nit: not related to this patch but I am just wondering why we are passing phys_addr_t *ioaddr downto vgic_check_ioaddr and thus to

vgic_check_iorange()? This must be a leftover of some old code?

Quoted 23 lines
arch/arm64/kvm/vgic/vgic-kvm-device.c +15 -0
+	if (ret)
+		return ret;
+
+	if (!IS_ALIGNED(size, alignment))
+		return -EINVAL;
+
+	if (addr + size < addr)
+		return -EINVAL;
+
+	if (addr + size > kvm_phys_size(kvm))
+		return -E2BIG;
+
+	return 0;
+}
+
 static int vgic_check_type(struct kvm *kvm, int type_needed)
 {
 	if (kvm->arch.vgic.vgic_model != type_needed)
arch/arm64/kvm/vgic/vgic.h +0 -0
index 14a9218641f5..c4df4dcef31f 100644

arch/arm64/kvm/vgic/vgic.h void vgic_irq_handle_resampling(struct vgic_irq *irq, · line 184

Quoted 10 lines
arch/arm64/kvm/vgic/vgic.h +4 -0
@@ -175,6 +175,10 @@ void vgic_irq_handle_resampling(struct vgic_irq *irq,
 int vgic_check_ioaddr(struct kvm *kvm, phys_addr_t *ioaddr,
 		      phys_addr_t addr, phys_addr_t alignment);
 
+int vgic_check_iorange(struct kvm *kvm, phys_addr_t *ioaddr,
+		       phys_addr_t addr, phys_addr_t alignment,
+		       phys_addr_t size);
+
 void vgic_v2_fold_lr_state(struct kvm_vcpu *vcpu);
 void vgic_v2_populate_lr(struct kvm_vcpu *vcpu, struct vgic_irq *irq, int lr);
 void vgic_v2_clear_lr(struct kvm_vcpu *vcpu, int lr);

Besides

Eric

#3 lore raw

Re: [PATCH v3 01/10] kvm: arm64: vgic: Introduce vgic_check_iorange

Quoted 34 lines · Eric Auger

Hi Ricardo,

Quoted 18 lines · Ricardo Koller

Add the new vgic_check_iorange helper that checks that an iorange is
sane: the start address and size have valid alignments, the range is within the addressable PA range, start+size doesn't overflow, and the start wasn't already defined.

No functional change.

---
 arch/arm64/kvm/vgic/vgic-kvm-device.c | 22 ++++++++++++++++++++++
 arch/arm64/kvm/vgic/vgic.h            |  4 ++++
 2 files changed, 26 insertions(+)
arch/arm64/kvm/vgic/vgic-kvm-device.c +0 -0
index 7740995de982..f714aded67b2 100644

arch/arm64/kvm/vgic/vgic-kvm-device.c int vgic_check_ioaddr(struct kvm *kvm, phys_addr_t *ioaddr, · line 38

Quoted 10 lines
arch/arm64/kvm/vgic/vgic-kvm-device.c +7 -0
@@ -29,6 +29,28 @@ int vgic_check_ioaddr(struct kvm *kvm, phys_addr_t *ioaddr,
 	return 0;
 }
 
+int vgic_check_iorange(struct kvm *kvm, phys_addr_t *ioaddr,
+		       phys_addr_t addr, phys_addr_t alignment,
+		       phys_addr_t size)
+{
+	int ret;
+
+	ret = vgic_check_ioaddr(kvm, ioaddr, addr, alignment);

nit: not related to this patch but I am just wondering why we are passing phys_addr_t *ioaddr downto vgic_check_ioaddr and thus to

Quoted 2 lines

vgic_check_iorange()? This must be a leftover of some old code?

It's used to check that the base of a region is not already set. kvm_vgic_addr() uses it to make that check; vgic_v3_alloc_redist_region() does not:

rdreg->base = VGIC_ADDR_UNDEF; // so the "not already defined" check passes ret = vgic_check_ioaddr(kvm, &rdreg->base, base, SZ_64K);

Thanks,
Ricardo

Quoted 37 lines
Quoted 23 lines
arch/arm64/kvm/vgic/vgic-kvm-device.c +15 -0
+	if (ret)
+		return ret;
+
+	if (!IS_ALIGNED(size, alignment))
+		return -EINVAL;
+
+	if (addr + size < addr)
+		return -EINVAL;
+
+	if (addr + size > kvm_phys_size(kvm))
+		return -E2BIG;
+
+	return 0;
+}
+
 static int vgic_check_type(struct kvm *kvm, int type_needed)
 {
 	if (kvm->arch.vgic.vgic_model != type_needed)
arch/arm64/kvm/vgic/vgic.h +0 -0
index 14a9218641f5..c4df4dcef31f 100644

arch/arm64/kvm/vgic/vgic.h void vgic_irq_handle_resampling(struct vgic_irq *irq, · line 184

Quoted 10 lines
arch/arm64/kvm/vgic/vgic.h +4 -0
@@ -175,6 +175,10 @@ void vgic_irq_handle_resampling(struct vgic_irq *irq,
 int vgic_check_ioaddr(struct kvm *kvm, phys_addr_t *ioaddr,
 		      phys_addr_t addr, phys_addr_t alignment);
 
+int vgic_check_iorange(struct kvm *kvm, phys_addr_t *ioaddr,
+		       phys_addr_t addr, phys_addr_t alignment,
+		       phys_addr_t size);
+
 void vgic_v2_fold_lr_state(struct kvm_vcpu *vcpu);
 void vgic_v2_populate_lr(struct kvm_vcpu *vcpu, struct vgic_irq *irq, int lr);
 void vgic_v2_clear_lr(struct kvm_vcpu *vcpu, int lr);

Besides

Eric

#4 lore raw

Re: [PATCH v3 01/10] kvm: arm64: vgic: Introduce vgic_check_iorange

Hi,

Quoted 35 lines · Ricardo Koller
Quoted 34 lines · Eric Auger

Hi Ricardo,

Quoted 18 lines · Ricardo Koller

Add the new vgic_check_iorange helper that checks that an iorange is
sane: the start address and size have valid alignments, the range is within the addressable PA range, start+size doesn't overflow, and the start wasn't already defined.

No functional change.

---
 arch/arm64/kvm/vgic/vgic-kvm-device.c | 22 ++++++++++++++++++++++
 arch/arm64/kvm/vgic/vgic.h            |  4 ++++
 2 files changed, 26 insertions(+)
arch/arm64/kvm/vgic/vgic-kvm-device.c +0 -0
index 7740995de982..f714aded67b2 100644

arch/arm64/kvm/vgic/vgic-kvm-device.c int vgic_check_ioaddr(struct kvm *kvm, phys_addr_t *ioaddr, · line 38

Quoted 10 lines
arch/arm64/kvm/vgic/vgic-kvm-device.c +7 -0
@@ -29,6 +29,28 @@ int vgic_check_ioaddr(struct kvm *kvm, phys_addr_t *ioaddr,
 	return 0;
 }
 
+int vgic_check_iorange(struct kvm *kvm, phys_addr_t *ioaddr,
+		       phys_addr_t addr, phys_addr_t alignment,
+		       phys_addr_t size)
+{
+	int ret;
+
+	ret = vgic_check_ioaddr(kvm, ioaddr, addr, alignment);

nit: not related to this patch but I am just wondering why we are passing phys_addr_t *ioaddr downto vgic_check_ioaddr and thus to

Quoted 8 lines
Quoted 2 lines

vgic_check_iorange()? This must be a leftover of some old code?

It's used to check that the base of a region is not already set. kvm_vgic_addr() uses it to make that check; vgic_v3_alloc_redist_region() does not:

rdreg->base = VGIC_ADDR_UNDEF; // so the "not already defined" check passes ret = vgic_check_ioaddr(kvm, &rdreg->base, base, SZ_64K);

Yes but I meant why a pointer?

Eric

Quoted 41 lines

Thanks,
Ricardo

Quoted 37 lines
Quoted 23 lines
arch/arm64/kvm/vgic/vgic-kvm-device.c +15 -0
+	if (ret)
+		return ret;
+
+	if (!IS_ALIGNED(size, alignment))
+		return -EINVAL;
+
+	if (addr + size < addr)
+		return -EINVAL;
+
+	if (addr + size > kvm_phys_size(kvm))
+		return -E2BIG;
+
+	return 0;
+}
+
 static int vgic_check_type(struct kvm *kvm, int type_needed)
 {
 	if (kvm->arch.vgic.vgic_model != type_needed)
arch/arm64/kvm/vgic/vgic.h +0 -0
index 14a9218641f5..c4df4dcef31f 100644

arch/arm64/kvm/vgic/vgic.h void vgic_irq_handle_resampling(struct vgic_irq *irq, · line 184

Quoted 10 lines
arch/arm64/kvm/vgic/vgic.h +4 -0
@@ -175,6 +175,10 @@ void vgic_irq_handle_resampling(struct vgic_irq *irq,
 int vgic_check_ioaddr(struct kvm *kvm, phys_addr_t *ioaddr,
 		      phys_addr_t addr, phys_addr_t alignment);
 
+int vgic_check_iorange(struct kvm *kvm, phys_addr_t *ioaddr,
+		       phys_addr_t addr, phys_addr_t alignment,
+		       phys_addr_t size);
+
 void vgic_v2_fold_lr_state(struct kvm_vcpu *vcpu);
 void vgic_v2_populate_lr(struct kvm_vcpu *vcpu, struct vgic_irq *irq, int lr);
 void vgic_v2_clear_lr(struct kvm_vcpu *vcpu, int lr);

Besides

Eric

#5 lore raw

Re: [PATCH v3 01/10] kvm: arm64: vgic: Introduce vgic_check_iorange

Quoted 40 lines · Eric Auger

Hi,

Quoted 37 lines · Ricardo Koller
Quoted 36 lines · Eric Auger

Hi Ricardo,

Quoted 18 lines · Ricardo Koller

Add the new vgic_check_iorange helper that checks that an iorange is
sane: the start address and size have valid alignments, the range is within the addressable PA range, start+size doesn't overflow, and the start wasn't already defined.

No functional change.

---
 arch/arm64/kvm/vgic/vgic-kvm-device.c | 22 ++++++++++++++++++++++
 arch/arm64/kvm/vgic/vgic.h            |  4 ++++
 2 files changed, 26 insertions(+)
arch/arm64/kvm/vgic/vgic-kvm-device.c +0 -0
index 7740995de982..f714aded67b2 100644

arch/arm64/kvm/vgic/vgic-kvm-device.c int vgic_check_ioaddr(struct kvm *kvm, phys_addr_t *ioaddr, · line 38

Quoted 10 lines
arch/arm64/kvm/vgic/vgic-kvm-device.c +7 -0
@@ -29,6 +29,28 @@ int vgic_check_ioaddr(struct kvm *kvm, phys_addr_t *ioaddr,
 	return 0;
 }
 
+int vgic_check_iorange(struct kvm *kvm, phys_addr_t *ioaddr,
+		       phys_addr_t addr, phys_addr_t alignment,
+		       phys_addr_t size)
+{
+	int ret;
+
+	ret = vgic_check_ioaddr(kvm, ioaddr, addr, alignment);

nit: not related to this patch but I am just wondering why we are passing phys_addr_t *ioaddr downto vgic_check_ioaddr and thus to

vgic_check_iorange()? This must be a leftover of some old code?

Quoted 7 lines
Quoted 6 lines

It's used to check that the base of a region is not already set. kvm_vgic_addr() uses it to make that check; vgic_v3_alloc_redist_region() does not:

rdreg->base = VGIC_ADDR_UNDEF; // so the "not already defined" check passes ret = vgic_check_ioaddr(kvm, &rdreg->base, base, SZ_64K);

Yes but I meant why a pointer?

I can't think of any good reason. It must be some leftover as you said.

Quoted 44 lines

Eric

Quoted 41 lines

Thanks,
Ricardo

Quoted 37 lines
Quoted 23 lines
arch/arm64/kvm/vgic/vgic-kvm-device.c +15 -0
+	if (ret)
+		return ret;
+
+	if (!IS_ALIGNED(size, alignment))
+		return -EINVAL;
+
+	if (addr + size < addr)
+		return -EINVAL;
+
+	if (addr + size > kvm_phys_size(kvm))
+		return -E2BIG;
+
+	return 0;
+}
+
 static int vgic_check_type(struct kvm *kvm, int type_needed)
 {
 	if (kvm->arch.vgic.vgic_model != type_needed)
arch/arm64/kvm/vgic/vgic.h +0 -0
index 14a9218641f5..c4df4dcef31f 100644

arch/arm64/kvm/vgic/vgic.h void vgic_irq_handle_resampling(struct vgic_irq *irq, · line 184

Quoted 10 lines
arch/arm64/kvm/vgic/vgic.h +4 -0
@@ -175,6 +175,10 @@ void vgic_irq_handle_resampling(struct vgic_irq *irq,
 int vgic_check_ioaddr(struct kvm *kvm, phys_addr_t *ioaddr,
 		      phys_addr_t addr, phys_addr_t alignment);
 
+int vgic_check_iorange(struct kvm *kvm, phys_addr_t *ioaddr,
+		       phys_addr_t addr, phys_addr_t alignment,
+		       phys_addr_t size);
+
 void vgic_v2_fold_lr_state(struct kvm_vcpu *vcpu);
 void vgic_v2_populate_lr(struct kvm_vcpu *vcpu, struct vgic_irq *irq, int lr);
 void vgic_v2_clear_lr(struct kvm_vcpu *vcpu, int lr);

Besides

Eric

maintainer
#6 lore raw

Re: [PATCH v3 01/10] kvm: arm64: vgic: Introduce vgic_check_iorange

Quoted 46 lines · Ricardo Koller
Quoted 44 lines · Eric Auger

Hi,

Quoted 41 lines · Ricardo Koller
Quoted 34 lines · Eric Auger

Hi Ricardo,

Quoted 18 lines · Ricardo Koller

Add the new vgic_check_iorange helper that checks that an iorange is
sane: the start address and size have valid alignments, the range is within the addressable PA range, start+size doesn't overflow, and the start wasn't already defined.

No functional change.

---
 arch/arm64/kvm/vgic/vgic-kvm-device.c | 22 ++++++++++++++++++++++
 arch/arm64/kvm/vgic/vgic.h            |  4 ++++
 2 files changed, 26 insertions(+)
arch/arm64/kvm/vgic/vgic-kvm-device.c +0 -0
index 7740995de982..f714aded67b2 100644

arch/arm64/kvm/vgic/vgic-kvm-device.c int vgic_check_ioaddr(struct kvm *kvm, phys_addr_t *ioaddr, · line 38

Quoted 10 lines
arch/arm64/kvm/vgic/vgic-kvm-device.c +7 -0
@@ -29,6 +29,28 @@ int vgic_check_ioaddr(struct kvm *kvm, phys_addr_t *ioaddr,
 	return 0;
 }
 
+int vgic_check_iorange(struct kvm *kvm, phys_addr_t *ioaddr,
+		       phys_addr_t addr, phys_addr_t alignment,
+		       phys_addr_t size)
+{
+	int ret;
+
+	ret = vgic_check_ioaddr(kvm, ioaddr, addr, alignment);

nit: not related to this patch but I am just wondering why we are passing phys_addr_t *ioaddr downto vgic_check_ioaddr and thus to

Quoted 2 lines

vgic_check_iorange()? This must be a leftover of some old code?

It's used to check that the base of a region is not already set. kvm_vgic_addr() uses it to make that check; vgic_v3_alloc_redist_region() does not:

Quoted 5 lines
Quoted 3 lines
Quoted 2 lines

rdreg->base = VGIC_ADDR_UNDEF; // so the "not already defined" check passes ret = vgic_check_ioaddr(kvm, &rdreg->base, base, SZ_64K);

Yes but I meant why a pointer?

I can't think of any good reason. It must be some leftover as you said.

It definitely is. Please have a patch to fix that. Also, it doesn't look like vgic_check_ioaddr() has any other user at the end of the series. Worth getting rid of.

	M.
Signature · 1 line
Without deviation from the norm, progress is not possible.
#7 lore raw

Re: [PATCH v3 01/10] kvm: arm64: vgic: Introduce vgic_check_iorange

Quoted 50 lines · Marc Zyngier
Quoted 48 lines · Ricardo Koller
Quoted 46 lines · Eric Auger

Hi,

Quoted 43 lines · Ricardo Koller
Quoted 34 lines · Eric Auger

Hi Ricardo,

Quoted 18 lines · Ricardo Koller

Add the new vgic_check_iorange helper that checks that an iorange is
sane: the start address and size have valid alignments, the range is within the addressable PA range, start+size doesn't overflow, and the start wasn't already defined.

No functional change.

---
 arch/arm64/kvm/vgic/vgic-kvm-device.c | 22 ++++++++++++++++++++++
 arch/arm64/kvm/vgic/vgic.h            |  4 ++++
 2 files changed, 26 insertions(+)
arch/arm64/kvm/vgic/vgic-kvm-device.c +0 -0
index 7740995de982..f714aded67b2 100644

arch/arm64/kvm/vgic/vgic-kvm-device.c int vgic_check_ioaddr(struct kvm *kvm, phys_addr_t *ioaddr, · line 38

Quoted 10 lines
arch/arm64/kvm/vgic/vgic-kvm-device.c +7 -0
@@ -29,6 +29,28 @@ int vgic_check_ioaddr(struct kvm *kvm, phys_addr_t *ioaddr,
 	return 0;
 }
 
+int vgic_check_iorange(struct kvm *kvm, phys_addr_t *ioaddr,
+		       phys_addr_t addr, phys_addr_t alignment,
+		       phys_addr_t size)
+{
+	int ret;
+
+	ret = vgic_check_ioaddr(kvm, ioaddr, addr, alignment);

nit: not related to this patch but I am just wondering why we are passing phys_addr_t *ioaddr downto vgic_check_ioaddr and thus to

Quoted 2 lines

vgic_check_iorange()? This must be a leftover of some old code?

It's used to check that the base of a region is not already set. kvm_vgic_addr() uses it to make that check; vgic_v3_alloc_redist_region() does not:

rdreg->base = VGIC_ADDR_UNDEF; // so the "not already defined" check passes ret = vgic_check_ioaddr(kvm, &rdreg->base, base, SZ_64K);

Quoted 7 lines
Quoted 3 lines
Quoted 1 line

Yes but I meant why a pointer?

I can't think of any good reason. It must be some leftover as you said.

It definitely is. Please have a patch to fix that. Also, it doesn't look like vgic_check_ioaddr() has any other user at the end of the series. Worth getting rid of.

ACK fixing that and getting rid of vgic_check_ioaddr().

Thanks,
Ricardo

Quoted 5 lines
	M.
Signature · 1 line
Without deviation from the norm, progress is not possible.
#8 lore raw

PATCH v3 2/10 KVM: arm64: vgic-v3: Check redist region is not above the VM IPA size

In reply to Ricardo Koller

Verify that the redistributor regions do not extend beyond the VM-specified IPA range (phys_size). This can happen when using KVM_VGIC_V3_ADDR_TYPE_REDIST or KVM_VGIC_V3_ADDR_TYPE_REDIST_REGIONS with:

base + size > phys_size AND base < phys_size

Add the missing check into vgic_v3_alloc_redist_region() which is called when setting the regions, and into vgic_v3_check_base() which is called when attempting the first vcpu-run. The vcpu-run check does not apply to KVM_VGIC_V3_ADDR_TYPE_REDIST_REGIONS because the regions size is known before the first vcpu-run. Note that using the REDIST_REGIONS API results in a different check, which already exists, at first vcpu run:
that the number of redist regions is enough for all vcpus.

Finally, this patch also enables some extra tests in vgic_v3_alloc_redist_region() by calculating "size" early for the legacy redist api: like checking that the REDIST region can fit all the already created vcpus.

---
 arch/arm64/kvm/vgic/vgic-mmio-v3.c | 6 ++++--
 arch/arm64/kvm/vgic/vgic-v3.c      | 4 ++++
 2 files changed, 8 insertions(+), 2 deletions(-)
arch/arm64/kvm/vgic/vgic-mmio-v3.c +4 -2
index a09cdc0b953c..9be02bf7865e 100644
@@ -796,7 +796,9 @@ static int vgic_v3_alloc_redist_region(struct kvm *kvm, uint32_t index,
 	struct vgic_dist *d = &kvm->arch.vgic;
 	struct vgic_redist_region *rdreg;
 	struct list_head *rd_regions = &d->rd_regions;
-	size_t size = count * KVM_VGIC_V3_REDIST_SIZE;
+	int nr_vcpus = atomic_read(&kvm->online_vcpus);
+	size_t size = count ? count * KVM_VGIC_V3_REDIST_SIZE
+			    : nr_vcpus * KVM_VGIC_V3_REDIST_SIZE;
 	int ret;
 
 	/* cross the end of memory ? */
@@ -840,7 +842,7 @@ static int vgic_v3_alloc_redist_region(struct kvm *kvm, uint32_t index,
 
 	rdreg->base = VGIC_ADDR_UNDEF;
 
-	ret = vgic_check_ioaddr(kvm, &rdreg->base, base, SZ_64K);
+	ret = vgic_check_iorange(kvm, &rdreg->base, base, SZ_64K, size);
 	if (ret)
 		goto free;
 
arch/arm64/kvm/vgic/vgic-v3.c +4 -0
index 21a6207fb2ee..27ee674631b3 100644
@@ -486,6 +486,10 @@ bool vgic_v3_check_base(struct kvm *kvm)
 		if (rdreg->base + vgic_v3_rd_region_size(kvm, rdreg) <
 			rdreg->base)
 			return false;
+
+		if (rdreg->base + vgic_v3_rd_region_size(kvm, rdreg) >
+			kvm_phys_size(kvm))
+			return false;
 	}
 
 	if (IS_VGIC_ADDR_UNDEF(d->vgic_dist_base))
Signature · 1 line
2.33.0.685.g46640cef36-goog
#9 lore raw

Re: [PATCH v3 02/10] KVM: arm64: vgic-v3: Check redist region is not above the VM IPA size

Hi Ricardo,

Quoted 31 lines · Ricardo Koller

Verify that the redistributor regions do not extend beyond the VM-specified IPA range (phys_size). This can happen when using KVM_VGIC_V3_ADDR_TYPE_REDIST or KVM_VGIC_V3_ADDR_TYPE_REDIST_REGIONS with:

base + size > phys_size AND base < phys_size

Add the missing check into vgic_v3_alloc_redist_region() which is called when setting the regions, and into vgic_v3_check_base() which is called when attempting the first vcpu-run. The vcpu-run check does not apply to KVM_VGIC_V3_ADDR_TYPE_REDIST_REGIONS because the regions size is known before the first vcpu-run. Note that using the REDIST_REGIONS API results in a different check, which already exists, at first vcpu run:
that the number of redist regions is enough for all vcpus.

Finally, this patch also enables some extra tests in vgic_v3_alloc_redist_region() by calculating "size" early for the legacy redist api: like checking that the REDIST region can fit all the already created vcpus.

---
 arch/arm64/kvm/vgic/vgic-mmio-v3.c | 6 ++++--
 arch/arm64/kvm/vgic/vgic-v3.c      | 4 ++++
 2 files changed, 8 insertions(+), 2 deletions(-)
arch/arm64/kvm/vgic/vgic-mmio-v3.c +0 -0
index a09cdc0b953c..9be02bf7865e 100644

arch/arm64/kvm/vgic/vgic-mmio-v3.c static int vgic_v3_alloc_redist_region(struct kvm *kvm, uint32_t index, · line 801

Quoted 7 lines
arch/arm64/kvm/vgic/vgic-mmio-v3.c +3 -1
@@ -796,7 +796,9 @@ static int vgic_v3_alloc_redist_region(struct kvm *kvm, uint32_t index,
 	struct vgic_dist *d = &kvm->arch.vgic;
 	struct vgic_redist_region *rdreg;
 	struct list_head *rd_regions = &d->rd_regions;
-	size_t size = count * KVM_VGIC_V3_REDIST_SIZE;
+	int nr_vcpus = atomic_read(&kvm->online_vcpus);
+	size_t size = count ? count * KVM_VGIC_V3_REDIST_SIZE
+			    : nr_vcpus * KVM_VGIC_V3_REDIST_SIZE;

This actually fixes the  vgic_dist_overlap(kvm, base, size=0)

in case the number of online-vcpus at that time is less than the final one (1st run), if count=0 (legacy API) do we ever check that the RDIST (with accumulated vcpu rdists) does not overlap with dist. in other words shouldn't we call vgic_dist_overlap(kvm, base, size) again in vgic_v3_check_base().

Thanks

Eric

Quoted 27 lines
arch/arm64/kvm/vgic/vgic-mmio-v3.c +1 -1
 	int ret;
 
 	/* cross the end of memory ? */
@@ -840,7 +842,7 @@ static int vgic_v3_alloc_redist_region(struct kvm *kvm, uint32_t index,
 
 	rdreg->base = VGIC_ADDR_UNDEF;
 
-	ret = vgic_check_ioaddr(kvm, &rdreg->base, base, SZ_64K);
+	ret = vgic_check_iorange(kvm, &rdreg->base, base, SZ_64K, size);
 	if (ret)
 		goto free;
 
arch/arm64/kvm/vgic/vgic-v3.c +4 -0
index 21a6207fb2ee..27ee674631b3 100644
@@ -486,6 +486,10 @@ bool vgic_v3_check_base(struct kvm *kvm)
 		if (rdreg->base + vgic_v3_rd_region_size(kvm, rdreg) <
 			rdreg->base)
 			return false;
+
+		if (rdreg->base + vgic_v3_rd_region_size(kvm, rdreg) >
+			kvm_phys_size(kvm))
+			return false;
 	}
 
 	if (IS_VGIC_ADDR_UNDEF(d->vgic_dist_base))
#10 lore raw

Re: [PATCH v3 02/10] KVM: arm64: vgic-v3: Check redist region is not above the VM IPA size

Hi Eric,

Quoted 44 lines · Eric Auger

Hi Ricardo,

Quoted 31 lines · Ricardo Koller

Verify that the redistributor regions do not extend beyond the VM-specified IPA range (phys_size). This can happen when using KVM_VGIC_V3_ADDR_TYPE_REDIST or KVM_VGIC_V3_ADDR_TYPE_REDIST_REGIONS with:

base + size > phys_size AND base < phys_size

Add the missing check into vgic_v3_alloc_redist_region() which is called when setting the regions, and into vgic_v3_check_base() which is called when attempting the first vcpu-run. The vcpu-run check does not apply to KVM_VGIC_V3_ADDR_TYPE_REDIST_REGIONS because the regions size is known before the first vcpu-run. Note that using the REDIST_REGIONS API results in a different check, which already exists, at first vcpu run:
that the number of redist regions is enough for all vcpus.

Finally, this patch also enables some extra tests in vgic_v3_alloc_redist_region() by calculating "size" early for the legacy redist api: like checking that the REDIST region can fit all the already created vcpus.

---
 arch/arm64/kvm/vgic/vgic-mmio-v3.c | 6 ++++--
 arch/arm64/kvm/vgic/vgic-v3.c      | 4 ++++
 2 files changed, 8 insertions(+), 2 deletions(-)
arch/arm64/kvm/vgic/vgic-mmio-v3.c +0 -0
index a09cdc0b953c..9be02bf7865e 100644

arch/arm64/kvm/vgic/vgic-mmio-v3.c static int vgic_v3_alloc_redist_region(struct kvm *kvm, uint32_t index, · line 801

Quoted 7 lines
arch/arm64/kvm/vgic/vgic-mmio-v3.c +3 -1
@@ -796,7 +796,9 @@ static int vgic_v3_alloc_redist_region(struct kvm *kvm, uint32_t index,
 	struct vgic_dist *d = &kvm->arch.vgic;
 	struct vgic_redist_region *rdreg;
 	struct list_head *rd_regions = &d->rd_regions;
-	size_t size = count * KVM_VGIC_V3_REDIST_SIZE;
+	int nr_vcpus = atomic_read(&kvm->online_vcpus);
+	size_t size = count ? count * KVM_VGIC_V3_REDIST_SIZE
+			    : nr_vcpus * KVM_VGIC_V3_REDIST_SIZE;

This actually fixes the  vgic_dist_overlap(kvm, base, size=0)

Quoted 6 lines

in case the number of online-vcpus at that time is less than the final one (1st run), if count=0 (legacy API) do we ever check that the RDIST (with accumulated vcpu rdists) does not overlap with dist. in other words shouldn't we call vgic_dist_overlap(kvm, base, size) again in vgic_v3_check_base().

I think we're good; that's checked by vgic_v3_rdist_overlap(dist_base) in vgic_v3_check_base(). This function uses the only region (legacy case) using a size based on the online_vcpus (in vgic_v3_rd_region_size()). This exact situation is tested by test_vgic_then_vcpus() in the vgic_init selftest.

Thanks,
Ricardo

Quoted 32 lines

Thanks

Eric

Quoted 27 lines
arch/arm64/kvm/vgic/vgic-mmio-v3.c +1 -1
 	int ret;
 
 	/* cross the end of memory ? */
@@ -840,7 +842,7 @@ static int vgic_v3_alloc_redist_region(struct kvm *kvm, uint32_t index,
 
 	rdreg->base = VGIC_ADDR_UNDEF;
 
-	ret = vgic_check_ioaddr(kvm, &rdreg->base, base, SZ_64K);
+	ret = vgic_check_iorange(kvm, &rdreg->base, base, SZ_64K, size);
 	if (ret)
 		goto free;
 
arch/arm64/kvm/vgic/vgic-v3.c +4 -0
index 21a6207fb2ee..27ee674631b3 100644
@@ -486,6 +486,10 @@ bool vgic_v3_check_base(struct kvm *kvm)
 		if (rdreg->base + vgic_v3_rd_region_size(kvm, rdreg) <
 			rdreg->base)
 			return false;
+
+		if (rdreg->base + vgic_v3_rd_region_size(kvm, rdreg) >
+			kvm_phys_size(kvm))
+			return false;
 	}
 
 	if (IS_VGIC_ADDR_UNDEF(d->vgic_dist_base))
Reviewed-by
#11 lore raw

Re: [PATCH v3 02/10] KVM: arm64: vgic-v3: Check redist region is not above the VM IPA size

Hi,

Quoted 52 lines · Ricardo Koller

Hi Eric,

Quoted 43 lines · Eric Auger

Hi Ricardo,

Quoted 31 lines · Ricardo Koller

Verify that the redistributor regions do not extend beyond the VM-specified IPA range (phys_size). This can happen when using KVM_VGIC_V3_ADDR_TYPE_REDIST or KVM_VGIC_V3_ADDR_TYPE_REDIST_REGIONS with:

base + size > phys_size AND base < phys_size

Add the missing check into vgic_v3_alloc_redist_region() which is called when setting the regions, and into vgic_v3_check_base() which is called when attempting the first vcpu-run. The vcpu-run check does not apply to KVM_VGIC_V3_ADDR_TYPE_REDIST_REGIONS because the regions size is known before the first vcpu-run. Note that using the REDIST_REGIONS API results in a different check, which already exists, at first vcpu run:
that the number of redist regions is enough for all vcpus.

Finally, this patch also enables some extra tests in vgic_v3_alloc_redist_region() by calculating "size" early for the legacy redist api: like checking that the REDIST region can fit all the already created vcpus.

---
 arch/arm64/kvm/vgic/vgic-mmio-v3.c | 6 ++++--
 arch/arm64/kvm/vgic/vgic-v3.c      | 4 ++++
 2 files changed, 8 insertions(+), 2 deletions(-)
arch/arm64/kvm/vgic/vgic-mmio-v3.c +0 -0
index a09cdc0b953c..9be02bf7865e 100644

arch/arm64/kvm/vgic/vgic-mmio-v3.c static int vgic_v3_alloc_redist_region(struct kvm *kvm, uint32_t index, · line 801

Quoted 7 lines
arch/arm64/kvm/vgic/vgic-mmio-v3.c +3 -1
@@ -796,7 +796,9 @@ static int vgic_v3_alloc_redist_region(struct kvm *kvm, uint32_t index,
 	struct vgic_dist *d = &kvm->arch.vgic;
 	struct vgic_redist_region *rdreg;
 	struct list_head *rd_regions = &d->rd_regions;
-	size_t size = count * KVM_VGIC_V3_REDIST_SIZE;
+	int nr_vcpus = atomic_read(&kvm->online_vcpus);
+	size_t size = count ? count * KVM_VGIC_V3_REDIST_SIZE
+			    : nr_vcpus * KVM_VGIC_V3_REDIST_SIZE;

This actually fixes the  vgic_dist_overlap(kvm, base, size=0)

Quoted 6 lines

in case the number of online-vcpus at that time is less than the final one (1st run), if count=0 (legacy API) do we ever check that the RDIST (with accumulated vcpu rdists) does not overlap with dist. in other words shouldn't we call vgic_dist_overlap(kvm, base, size) again in vgic_v3_check_base().

Quoted 5 lines

I think we're good; that's checked by vgic_v3_rdist_overlap(dist_base) in vgic_v3_check_base(). This function uses the only region (legacy case) using a size based on the online_vcpus (in vgic_v3_rd_region_size()). This exact situation is tested by test_vgic_then_vcpus() in the vgic_init selftest.

Ah OK so that's fine then.

So looks good to me

Eric

Quoted 35 lines

Thanks,
Ricardo

Quoted 31 lines

Thanks

Eric

Quoted 27 lines
arch/arm64/kvm/vgic/vgic-mmio-v3.c +1 -1
 	int ret;
 
 	/* cross the end of memory ? */
@@ -840,7 +842,7 @@ static int vgic_v3_alloc_redist_region(struct kvm *kvm, uint32_t index,
 
 	rdreg->base = VGIC_ADDR_UNDEF;
 
-	ret = vgic_check_ioaddr(kvm, &rdreg->base, base, SZ_64K);
+	ret = vgic_check_iorange(kvm, &rdreg->base, base, SZ_64K, size);
 	if (ret)
 		goto free;
 
arch/arm64/kvm/vgic/vgic-v3.c +4 -0
index 21a6207fb2ee..27ee674631b3 100644
@@ -486,6 +486,10 @@ bool vgic_v3_check_base(struct kvm *kvm)
 		if (rdreg->base + vgic_v3_rd_region_size(kvm, rdreg) <
 			rdreg->base)
 			return false;
+
+		if (rdreg->base + vgic_v3_rd_region_size(kvm, rdreg) >
+			kvm_phys_size(kvm))
+			return false;
 	}
 
 	if (IS_VGIC_ADDR_UNDEF(d->vgic_dist_base))
maintainer
#12 lore raw

Re: [PATCH v3 02/10] KVM: arm64: vgic-v3: Check redist region is not above the VM IPA size

In reply to Ricardo Koller
Quoted 56 lines · Ricardo Koller

Verify that the redistributor regions do not extend beyond the VM-specified IPA range (phys_size). This can happen when using KVM_VGIC_V3_ADDR_TYPE_REDIST or KVM_VGIC_V3_ADDR_TYPE_REDIST_REGIONS with:

base + size > phys_size AND base < phys_size

Add the missing check into vgic_v3_alloc_redist_region() which is called when setting the regions, and into vgic_v3_check_base() which is called when attempting the first vcpu-run. The vcpu-run check does not apply to KVM_VGIC_V3_ADDR_TYPE_REDIST_REGIONS because the regions size is known before the first vcpu-run. Note that using the REDIST_REGIONS API results in a different check, which already exists, at first vcpu run:
that the number of redist regions is enough for all vcpus.

Finally, this patch also enables some extra tests in vgic_v3_alloc_redist_region() by calculating "size" early for the legacy redist api: like checking that the REDIST region can fit all the already created vcpus.

---
 arch/arm64/kvm/vgic/vgic-mmio-v3.c | 6 ++++--
 arch/arm64/kvm/vgic/vgic-v3.c      | 4 ++++
 2 files changed, 8 insertions(+), 2 deletions(-)
arch/arm64/kvm/vgic/vgic-mmio-v3.c +4 -2
index a09cdc0b953c..9be02bf7865e 100644
@@ -796,7 +796,9 @@ static int vgic_v3_alloc_redist_region(struct kvm *kvm, uint32_t index,
 	struct vgic_dist *d = &kvm->arch.vgic;
 	struct vgic_redist_region *rdreg;
 	struct list_head *rd_regions = &d->rd_regions;
-	size_t size = count * KVM_VGIC_V3_REDIST_SIZE;
+	int nr_vcpus = atomic_read(&kvm->online_vcpus);
+	size_t size = count ? count * KVM_VGIC_V3_REDIST_SIZE
+			    : nr_vcpus * KVM_VGIC_V3_REDIST_SIZE;
 	int ret;
 
 	/* cross the end of memory ? */
@@ -840,7 +842,7 @@ static int vgic_v3_alloc_redist_region(struct kvm *kvm, uint32_t index,
 
 	rdreg->base = VGIC_ADDR_UNDEF;
 
-	ret = vgic_check_ioaddr(kvm, &rdreg->base, base, SZ_64K);
+	ret = vgic_check_iorange(kvm, &rdreg->base, base, SZ_64K, size);
 	if (ret)
 		goto free;
 
arch/arm64/kvm/vgic/vgic-v3.c +0 -0
index 21a6207fb2ee..27ee674631b3 100644

arch/arm64/kvm/vgic/vgic-v3.c bool vgic_v3_check_base(struct kvm *kvm) · line 492

Quoted 7 lines
arch/arm64/kvm/vgic/vgic-v3.c +4 -0
@@ -486,6 +486,10 @@ bool vgic_v3_check_base(struct kvm *kvm)
 		if (rdreg->base + vgic_v3_rd_region_size(kvm, rdreg) <
 			rdreg->base)
 			return false;
+
+		if (rdreg->base + vgic_v3_rd_region_size(kvm, rdreg) >
+			kvm_phys_size(kvm))
+			return false;

Why can't we replace these two checks with a single call to your new fancy helper?

Thanks,

	M.
Signature · 1 line
Without deviation from the norm, progress is not possible.
#13 lore raw

Re: [PATCH v3 02/10] KVM: arm64: vgic-v3: Check redist region is not above the VM IPA size

Quoted 53 lines · Marc Zyngier
Quoted 51 lines · Ricardo Koller

Verify that the redistributor regions do not extend beyond the VM-specified IPA range (phys_size). This can happen when using KVM_VGIC_V3_ADDR_TYPE_REDIST or KVM_VGIC_V3_ADDR_TYPE_REDIST_REGIONS with:

base + size > phys_size AND base < phys_size

Add the missing check into vgic_v3_alloc_redist_region() which is called when setting the regions, and into vgic_v3_check_base() which is called when attempting the first vcpu-run. The vcpu-run check does not apply to KVM_VGIC_V3_ADDR_TYPE_REDIST_REGIONS because the regions size is known before the first vcpu-run. Note that using the REDIST_REGIONS API results in a different check, which already exists, at first vcpu run:
that the number of redist regions is enough for all vcpus.

Finally, this patch also enables some extra tests in vgic_v3_alloc_redist_region() by calculating "size" early for the legacy redist api: like checking that the REDIST region can fit all the already created vcpus.

---
 arch/arm64/kvm/vgic/vgic-mmio-v3.c | 6 ++++--
 arch/arm64/kvm/vgic/vgic-v3.c      | 4 ++++
 2 files changed, 8 insertions(+), 2 deletions(-)
arch/arm64/kvm/vgic/vgic-mmio-v3.c +4 -2
index a09cdc0b953c..9be02bf7865e 100644
@@ -796,7 +796,9 @@ static int vgic_v3_alloc_redist_region(struct kvm *kvm, uint32_t index,
 	struct vgic_dist *d = &kvm->arch.vgic;
 	struct vgic_redist_region *rdreg;
 	struct list_head *rd_regions = &d->rd_regions;
-	size_t size = count * KVM_VGIC_V3_REDIST_SIZE;
+	int nr_vcpus = atomic_read(&kvm->online_vcpus);
+	size_t size = count ? count * KVM_VGIC_V3_REDIST_SIZE
+			    : nr_vcpus * KVM_VGIC_V3_REDIST_SIZE;
 	int ret;
 
 	/* cross the end of memory ? */
@@ -840,7 +842,7 @@ static int vgic_v3_alloc_redist_region(struct kvm *kvm, uint32_t index,
 
 	rdreg->base = VGIC_ADDR_UNDEF;
 
-	ret = vgic_check_ioaddr(kvm, &rdreg->base, base, SZ_64K);
+	ret = vgic_check_iorange(kvm, &rdreg->base, base, SZ_64K, size);
 	if (ret)
 		goto free;
 
Quoted 15 lines

arch/arm64/kvm/vgic/vgic-v3.c bool vgic_v3_check_base(struct kvm *kvm) · line 492

Quoted 12 lines
arch/arm64/kvm/vgic/vgic-v3.c +4 -0
index 21a6207fb2ee..27ee674631b3 100644
@@ -486,6 +486,10 @@ bool vgic_v3_check_base(struct kvm *kvm)
 		if (rdreg->base + vgic_v3_rd_region_size(kvm, rdreg) <
 			rdreg->base)
 			return false;
+
+		if (rdreg->base + vgic_v3_rd_region_size(kvm, rdreg) >
+			kvm_phys_size(kvm))
+			return false;

Why can't we replace these two checks with a single call to your new fancy helper?

ACK using the new helper (on rdreg base and size).

Thanks,
Ricardo

Quoted 7 lines

Thanks,

	M.
Signature · 1 line
Without deviation from the norm, progress is not possible.
#14 lore raw

PATCH v3 3/10 KVM: arm64: vgic-v2: Check cpu interface region is not above the VM IPA size

In reply to Ricardo Koller

Verify that the GICv2 CPU interface does not extend beyond the VM-specified IPA range (phys_size).

base + size > phys_size AND base < phys_size

Add the missing check into kvm_vgic_addr() which is called when setting the region. This patch also enables some superfluous checks for the distributor (vgic_check_ioaddr was enough as alignment == size for the distributors).

---
 arch/arm64/kvm/vgic/vgic-kvm-device.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
arch/arm64/kvm/vgic/vgic-kvm-device.c +5 -2
index f714aded67b2..b379eb81fddb 100644
@@ -79,7 +79,7 @@ int kvm_vgic_addr(struct kvm *kvm, unsigned long type, u64 *addr, bool write)
 {
 	int r = 0;
 	struct vgic_dist *vgic = &kvm->arch.vgic;
-	phys_addr_t *addr_ptr, alignment;
+	phys_addr_t *addr_ptr, alignment, size;
 	u64 undef_value = VGIC_ADDR_UNDEF;
 
 	mutex_lock(&kvm->lock);
@@ -88,16 +88,19 @@ int kvm_vgic_addr(struct kvm *kvm, unsigned long type, u64 *addr, bool write)
 		r = vgic_check_type(kvm, KVM_DEV_TYPE_ARM_VGIC_V2);
 		addr_ptr = &vgic->vgic_dist_base;
 		alignment = SZ_4K;
+		size = KVM_VGIC_V2_DIST_SIZE;
 		break;
 	case KVM_VGIC_V2_ADDR_TYPE_CPU:
 		r = vgic_check_type(kvm, KVM_DEV_TYPE_ARM_VGIC_V2);
 		addr_ptr = &vgic->vgic_cpu_base;
 		alignment = SZ_4K;
+		size = KVM_VGIC_V2_CPU_SIZE;
 		break;
 	case KVM_VGIC_V3_ADDR_TYPE_DIST:
 		r = vgic_check_type(kvm, KVM_DEV_TYPE_ARM_VGIC_V3);
 		addr_ptr = &vgic->vgic_dist_base;
 		alignment = SZ_64K;
+		size = KVM_VGIC_V3_DIST_SIZE;
 		break;
 	case KVM_VGIC_V3_ADDR_TYPE_REDIST: {
 		struct vgic_redist_region *rdreg;
@@ -162,7 +165,7 @@ int kvm_vgic_addr(struct kvm *kvm, unsigned long type, u64 *addr, bool write)
 		goto out;
 
 	if (write) {
-		r = vgic_check_ioaddr(kvm, addr_ptr, *addr, alignment);
+		r = vgic_check_iorange(kvm, addr_ptr, *addr, alignment, size);
 		if (!r)
 			*addr_ptr = *addr;
 	} else {
Signature · 1 line
2.33.0.685.g46640cef36-goog
Reviewed-by on v3 3/10
#15 lore raw

Re: [PATCH v3 03/10] KVM: arm64: vgic-v2: Check cpu interface region is not above the VM IPA size

Hi Ricardo,

Quoted 49 lines · Ricardo Koller

Verify that the GICv2 CPU interface does not extend beyond the VM-specified IPA range (phys_size).

base + size > phys_size AND base < phys_size

Add the missing check into kvm_vgic_addr() which is called when setting the region. This patch also enables some superfluous checks for the distributor (vgic_check_ioaddr was enough as alignment == size for the distributors).

---
 arch/arm64/kvm/vgic/vgic-kvm-device.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
arch/arm64/kvm/vgic/vgic-kvm-device.c +4 -1
index f714aded67b2..b379eb81fddb 100644
@@ -79,7 +79,7 @@ int kvm_vgic_addr(struct kvm *kvm, unsigned long type, u64 *addr, bool write)
 {
 	int r = 0;
 	struct vgic_dist *vgic = &kvm->arch.vgic;
-	phys_addr_t *addr_ptr, alignment;
+	phys_addr_t *addr_ptr, alignment, size;
 	u64 undef_value = VGIC_ADDR_UNDEF;
 
 	mutex_lock(&kvm->lock);
@@ -88,16 +88,19 @@ int kvm_vgic_addr(struct kvm *kvm, unsigned long type, u64 *addr, bool write)
 		r = vgic_check_type(kvm, KVM_DEV_TYPE_ARM_VGIC_V2);
 		addr_ptr = &vgic->vgic_dist_base;
 		alignment = SZ_4K;
+		size = KVM_VGIC_V2_DIST_SIZE;
 		break;
 	case KVM_VGIC_V2_ADDR_TYPE_CPU:
 		r = vgic_check_type(kvm, KVM_DEV_TYPE_ARM_VGIC_V2);
 		addr_ptr = &vgic->vgic_cpu_base;
 		alignment = SZ_4K;
+		size = KVM_VGIC_V2_CPU_SIZE;
 		break;
 	case KVM_VGIC_V3_ADDR_TYPE_DIST:
 		r = vgic_check_type(kvm, KVM_DEV_TYPE_ARM_VGIC_V3);
 		addr_ptr = &vgic->vgic_dist_base;
 		alignment = SZ_64K;
+		size = KVM_VGIC_V3_DIST_SIZE;
 		break;
 	case KVM_VGIC_V3_ADDR_TYPE_REDIST: {
 		struct vgic_redist_region *rdreg;

arch/arm64/kvm/vgic/vgic-kvm-device.c int kvm_vgic_addr(struct kvm *kvm, unsigned long type, u64 *addr, bool write) · line 171

Quoted 8 lines
arch/arm64/kvm/vgic/vgic-kvm-device.c +1 -1
@@ -162,7 +165,7 @@ int kvm_vgic_addr(struct kvm *kvm, unsigned long type, u64 *addr, bool write)
 		goto out;
 
 	if (write) {
-		r = vgic_check_ioaddr(kvm, addr_ptr, *addr, alignment);
+		r = vgic_check_iorange(kvm, addr_ptr, *addr, alignment, size);
 		if (!r)
 			*addr_ptr = *addr;
 	} else {

Looks god to me

Eric

#16 lore raw

PATCH v3 4/10 KVM: arm64: vgic-v3: Check ITS region is not above the VM IPA size

In reply to Ricardo Koller

Verify that the ITS region does not extend beyond the VM-specified IPA range (phys_size).

base + size > phys_size AND base < phys_size

Add the missing check into vgic_its_set_attr() which is called when setting the region.

---
 arch/arm64/kvm/vgic/vgic-its.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
arch/arm64/kvm/vgic/vgic-its.c +2 -2
index 61728c543eb9..321743b87002 100644
@@ -2710,8 +2710,8 @@ static int vgic_its_set_attr(struct kvm_device *dev,
 		if (copy_from_user(&addr, uaddr, sizeof(addr)))
 			return -EFAULT;
 
-		ret = vgic_check_ioaddr(dev->kvm, &its->vgic_its_base,
-					addr, SZ_64K);
+		ret = vgic_check_iorange(dev->kvm, &its->vgic_its_base,
+					 addr, SZ_64K, KVM_VGIC_V3_ITS_SIZE);
 		if (ret)
 			return ret;
 
Signature · 1 line
2.33.0.685.g46640cef36-goog
Reviewed-by on v3 4/10
#17 lore raw

Re: [PATCH v3 04/10] KVM: arm64: vgic-v3: Check ITS region is not above the VM IPA size

Quoted 18 lines · Ricardo Koller

Verify that the ITS region does not extend beyond the VM-specified IPA range (phys_size).

base + size > phys_size AND base < phys_size

Add the missing check into vgic_its_set_attr() which is called when setting the region.

---
 arch/arm64/kvm/vgic/vgic-its.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
arch/arm64/kvm/vgic/vgic-its.c +0 -0
index 61728c543eb9..321743b87002 100644

arch/arm64/kvm/vgic/vgic-its.c static int vgic_its_set_attr(struct kvm_device *dev, · line 2717

Quoted 10 lines
arch/arm64/kvm/vgic/vgic-its.c +2 -2
@@ -2710,8 +2710,8 @@ static int vgic_its_set_attr(struct kvm_device *dev,
 		if (copy_from_user(&addr, uaddr, sizeof(addr)))
 			return -EFAULT;
 
-		ret = vgic_check_ioaddr(dev->kvm, &its->vgic_its_base,
-					addr, SZ_64K);
+		ret = vgic_check_iorange(dev->kvm, &its->vgic_its_base,
+					 addr, SZ_64K, KVM_VGIC_V3_ITS_SIZE);
 		if (ret)
 			return ret;
 

Eric

#18 lore raw

PATCH v3 5/10 KVM: arm64: selftests: Make vgic_init gic version agnostic

In reply to Ricardo Koller

As a preparation for the next commits which will add some tests for GICv2, make aarch64/vgic_init GIC version agnostic. Add a new generic run_tests function(gic_dev_type) that starts all applicable tests using GICv3 or GICv2. GICv2 tests are attempted if GICv3 is not available in the system. There are currently no GICv2 tests, but the test passes now in GICv2 systems.

---
 .../testing/selftests/kvm/aarch64/vgic_init.c | 156 +++++++++++-------
 1 file changed, 95 insertions(+), 61 deletions(-)
tools/testing/selftests/kvm/aarch64/vgic_init.c +95 -61
index 623f31a14326..896a29f2503d 100644
@@ -22,6 +22,9 @@
 
 #define GICR_TYPER 0x8
 
+#define VGIC_DEV_IS_V2(_d) ((_d) == KVM_DEV_TYPE_ARM_VGIC_V2)
+#define VGIC_DEV_IS_V3(_d) ((_d) == KVM_DEV_TYPE_ARM_VGIC_V3)
+
 struct vm_gic {
 	struct kvm_vm *vm;
 	int gic_fd;
@@ -30,8 +33,8 @@ struct vm_gic {
 static int max_ipa_bits;
 
 /* helper to access a redistributor register */
-static int access_redist_reg(int gicv3_fd, int vcpu, int offset,
-			     uint32_t *val, bool write)
+static int access_v3_redist_reg(int gicv3_fd, int vcpu, int offset,
+				uint32_t *val, bool write)
 {
 	uint64_t attr = REG_OFFSET(vcpu, offset);
 
@@ -58,7 +61,7 @@ static int run_vcpu(struct kvm_vm *vm, uint32_t vcpuid)
 	return 0;
 }
 
-static struct vm_gic vm_gic_create(void)
+static struct vm_gic vm_gic_v3_create(void)
 {
 	struct vm_gic v;
 
@@ -80,7 +83,7 @@ static void vm_gic_destroy(struct vm_gic *v)
  * device gets created, a legacy RDIST region is set at @0x0
  * and a DIST region is set @0x60000
  */
-static void subtest_dist_rdist(struct vm_gic *v)
+static void subtest_v3_dist_rdist(struct vm_gic *v)
 {
 	int ret;
 	uint64_t addr;
@@ -145,7 +148,7 @@ static void subtest_dist_rdist(struct vm_gic *v)
 }
 
 /* Test the new REDIST region API */
-static void subtest_redist_regions(struct vm_gic *v)
+static void subtest_v3_redist_regions(struct vm_gic *v)
 {
 	uint64_t addr, expected_addr;
 	int ret;
@@ -249,7 +252,7 @@ static void subtest_redist_regions(struct vm_gic *v)
  * VGIC KVM device is created and initialized before the secondary CPUs
  * get created
  */
-static void test_vgic_then_vcpus(void)
+static void test_v3_vgic_then_vcpus(uint32_t gic_dev_type)
 {
 	struct vm_gic v;
 	int ret, i;
@@ -257,7 +260,7 @@ static void test_vgic_then_vcpus(void)
 	v.vm = vm_create_default(0, 0, guest_code);
 	v.gic_fd = kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_V3, false);
 
-	subtest_dist_rdist(&v);
+	subtest_v3_dist_rdist(&v);
 
 	/* Add the rest of the VCPUs */
 	for (i = 1; i < NR_VCPUS; ++i)
@@ -270,14 +273,14 @@ static void test_vgic_then_vcpus(void)
 }
 
 /* All the VCPUs are created before the VGIC KVM device gets initialized */
-static void test_vcpus_then_vgic(void)
+static void test_v3_vcpus_then_vgic(uint32_t gic_dev_type)
 {
 	struct vm_gic v;
 	int ret;
 
-	v = vm_gic_create();
+	v = vm_gic_v3_create();
 
-	subtest_dist_rdist(&v);
+	subtest_v3_dist_rdist(&v);
 
 	ret = run_vcpu(v.vm, 3);
 	TEST_ASSERT(ret == -EINVAL, "dist/rdist overlap detected on 1st vcpu run");
@@ -285,15 +288,15 @@ static void test_vcpus_then_vgic(void)
 	vm_gic_destroy(&v);
 }
 
-static void test_new_redist_regions(void)
+static void test_v3_new_redist_regions(void)
 {
 	void *dummy = NULL;
 	struct vm_gic v;
 	uint64_t addr;
 	int ret;
 
-	v = vm_gic_create();
-	subtest_redist_regions(&v);
+	v = vm_gic_v3_create();
+	subtest_v3_redist_regions(&v);
 	kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_CTRL,
 			  KVM_DEV_ARM_VGIC_CTRL_INIT, NULL, true);
 
@@ -303,8 +306,8 @@ static void test_new_redist_regions(void)
 
 	/* step2 */
 
-	v = vm_gic_create();
-	subtest_redist_regions(&v);
+	v = vm_gic_v3_create();
+	subtest_v3_redist_regions(&v);
 
 	addr = REDIST_REGION_ATTR_ADDR(1, 0x280000, 0, 2);
 	kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
@@ -317,8 +320,8 @@ static void test_new_redist_regions(void)
 
 	/* step 3 */
 
-	v = vm_gic_create();
-	subtest_redist_regions(&v);
+	v = vm_gic_v3_create();
+	subtest_v3_redist_regions(&v);
 
 	_kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
 			  KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, dummy, true);
@@ -338,7 +341,7 @@ static void test_new_redist_regions(void)
 	vm_gic_destroy(&v);
 }
 
-static void test_typer_accesses(void)
+static void test_v3_typer_accesses(void)
 {
 	struct vm_gic v;
 	uint64_t addr;
@@ -351,12 +354,12 @@ static void test_typer_accesses(void)
 
 	vm_vcpu_add_default(v.vm, 3, guest_code);
 
-	ret = access_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
 	TEST_ASSERT(ret && errno == EINVAL, "attempting to read GICR_TYPER of non created vcpu");
 
 	vm_vcpu_add_default(v.vm, 1, guest_code);
 
-	ret = access_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
 	TEST_ASSERT(ret && errno == EBUSY, "read GICR_TYPER before GIC initialized");
 
 	vm_vcpu_add_default(v.vm, 2, guest_code);
@@ -365,7 +368,7 @@ static void test_typer_accesses(void)
 			  KVM_DEV_ARM_VGIC_CTRL_INIT, NULL, true);
 
 	for (i = 0; i < NR_VCPUS ; i++) {
-		ret = access_redist_reg(v.gic_fd, 0, GICR_TYPER, &val, false);
+		ret = access_v3_redist_reg(v.gic_fd, 0, GICR_TYPER, &val, false);
 		TEST_ASSERT(!ret && !val, "read GICR_TYPER before rdist region setting");
 	}
 
@@ -374,10 +377,10 @@ static void test_typer_accesses(void)
 			  KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);
 
 	/* The 2 first rdists should be put there (vcpu 0 and 3) */
-	ret = access_redist_reg(v.gic_fd, 0, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 0, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && !val, "read typer of rdist #0");
 
-	ret = access_redist_reg(v.gic_fd, 3, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 3, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x310, "read typer of rdist #1");
 
 	addr = REDIST_REGION_ATTR_ADDR(10, 0x100000, 0, 1);
@@ -385,11 +388,11 @@ static void test_typer_accesses(void)
 				 KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);
 	TEST_ASSERT(ret && errno == EINVAL, "collision with previous rdist region");
 
-	ret = access_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x100,
 		    "no redist region attached to vcpu #1 yet, last cannot be returned");
 
-	ret = access_redist_reg(v.gic_fd, 2, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 2, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x200,
 		    "no redist region attached to vcpu #2, last cannot be returned");
 
@@ -397,10 +400,10 @@ static void test_typer_accesses(void)
 	kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
 			  KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);
 
-	ret = access_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x100, "read typer of rdist #1");
 
-	ret = access_redist_reg(v.gic_fd, 2, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 2, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x210,
 		    "read typer of rdist #1, last properly returned");
 
@@ -417,7 +420,7 @@ static void test_typer_accesses(void)
  * rdist region #2 @0x200000 2 rdist capacity
  *     rdists: 1, 2
  */
-static void test_last_bit_redist_regions(void)
+static void test_v3_last_bit_redist_regions(void)
 {
 	uint32_t vcpuids[] = { 0, 3, 5, 4, 1, 2 };
 	struct vm_gic v;
@@ -444,29 +447,29 @@ static void test_last_bit_redist_regions(void)
 	kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
 			  KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);
 
-	ret = access_redist_reg(v.gic_fd, 0, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 0, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x000, "read typer of rdist #0");
 
-	ret = access_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x100, "read typer of rdist #1");
 
-	ret = access_redist_reg(v.gic_fd, 2, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 2, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x200, "read typer of rdist #2");
 
-	ret = access_redist_reg(v.gic_fd, 3, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 3, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x310, "read typer of rdist #3");
 
-	ret = access_redist_reg(v.gic_fd, 5, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 5, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x500, "read typer of rdist #5");
 
-	ret = access_redist_reg(v.gic_fd, 4, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 4, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x410, "read typer of rdist #4");
 
 	vm_gic_destroy(&v);
 }
 
 /* Test last bit with legacy region */
-static void test_last_bit_single_rdist(void)
+static void test_v3_last_bit_single_rdist(void)
 {
 	uint32_t vcpuids[] = { 0, 3, 5, 4, 1, 2 };
 	struct vm_gic v;
@@ -485,28 +488,32 @@ static void test_last_bit_single_rdist(void)
 	kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
 			  KVM_VGIC_V3_ADDR_TYPE_REDIST, &addr, true);
 
-	ret = access_redist_reg(v.gic_fd, 0, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 0, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x000, "read typer of rdist #0");
 
-	ret = access_redist_reg(v.gic_fd, 3, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 3, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x300, "read typer of rdist #1");
 
-	ret = access_redist_reg(v.gic_fd, 5, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 5, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x500, "read typer of rdist #2");
 
-	ret = access_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x100, "read typer of rdist #3");
 
-	ret = access_redist_reg(v.gic_fd, 2, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 2, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x210, "read typer of rdist #3");
 
 	vm_gic_destroy(&v);
 }
 
-void test_kvm_device(void)
+/*
+ * Returns 0 if it's possible to create GIC device of a given type (V2 or V3).
+ */
+int test_kvm_device(uint32_t gic_dev_type)
 {
 	struct vm_gic v;
 	int ret, fd;
+	uint32_t other;
 
 	v.vm = vm_create_default_with_vcpus(NR_VCPUS, 0, 0, guest_code, NULL);
 
@@ -514,38 +521,65 @@ void test_kvm_device(void)
 	ret = _kvm_create_device(v.vm, 0, true, &fd);
 	TEST_ASSERT(ret && errno == ENODEV, "unsupported device");
 
-	/* trial mode with VGIC_V3 device */
-	ret = _kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_V3, true, &fd);
-	if (ret) {
-		print_skip("GICv3 not supported");
-		exit(KSFT_SKIP);
-	}
-	v.gic_fd = kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_V3, false);
+	/* trial mode */
+	ret = _kvm_create_device(v.vm, gic_dev_type, true, &fd);
+	if (ret)
+		return ret;
+	v.gic_fd = kvm_create_device(v.vm, gic_dev_type, false);
+
+	ret = _kvm_create_device(v.vm, gic_dev_type, false, &fd);
+	TEST_ASSERT(ret && errno == EEXIST, "create GIC device twice");
 
-	ret = _kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_V3, false, &fd);
-	TEST_ASSERT(ret && errno == EEXIST, "create GICv3 device twice");
+	kvm_create_device(v.vm, gic_dev_type, true);
 
-	kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_V3, true);
+	/* try to create the other gic_dev_type */
+	other = VGIC_DEV_IS_V2(gic_dev_type) ? KVM_DEV_TYPE_ARM_VGIC_V3
+					     : KVM_DEV_TYPE_ARM_VGIC_V2;
 
-	if (!_kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_V2, true, &fd)) {
-		ret = _kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_V2, false, &fd);
-		TEST_ASSERT(ret && errno == EINVAL, "create GICv2 while v3 exists");
+	if (!_kvm_create_device(v.vm, other, true, &fd)) {
+		ret = _kvm_create_device(v.vm, other, false, &fd);
+		TEST_ASSERT(ret && errno == EINVAL,
+				"create GIC device while other version exists");
 	}
 
 	vm_gic_destroy(&v);
+
+	return 0;
+}
+
+void run_tests(uint32_t gic_dev_type)
+{
+	if (VGIC_DEV_IS_V3(gic_dev_type)) {
+		test_v3_vcpus_then_vgic(gic_dev_type);
+		test_v3_vgic_then_vcpus(gic_dev_type);
+		test_v3_new_redist_regions();
+		test_v3_typer_accesses();
+		test_v3_last_bit_redist_regions();
+		test_v3_last_bit_single_rdist();
+	}
 }
 
 int main(int ac, char **av)
 {
+	int ret;
+
 	max_ipa_bits = kvm_check_cap(KVM_CAP_ARM_VM_IPA_SIZE);
 
-	test_kvm_device();
-	test_vcpus_then_vgic();
-	test_vgic_then_vcpus();
-	test_new_redist_regions();
-	test_typer_accesses();
-	test_last_bit_redist_regions();
-	test_last_bit_single_rdist();
+	ret = test_kvm_device(KVM_DEV_TYPE_ARM_VGIC_V3);
+	if (!ret) {
+		pr_info("Running GIC_v3 tests.\n");
+		run_tests(KVM_DEV_TYPE_ARM_VGIC_V3);
+		return 0;
+	}
+
+	ret = test_kvm_device(KVM_DEV_TYPE_ARM_VGIC_V2);
+	if (!ret) {
+		pr_info("Running GIC_v2 tests.\n");
+		run_tests(KVM_DEV_TYPE_ARM_VGIC_V2);
+		return 0;
+	}
 
+	print_skip("No GICv2 nor GICv3 support");
+	exit(KSFT_SKIP);
 	return 0;
 }
Signature · 1 line
2.33.0.685.g46640cef36-goog
Reviewed-by on v3 5/10
#19 lore raw

Re: [PATCH v3 05/10] KVM: arm64: selftests: Make vgic_init gic version agnostic

Quoted 362 lines · Ricardo Koller

As a preparation for the next commits which will add some tests for GICv2, make aarch64/vgic_init GIC version agnostic. Add a new generic run_tests function(gic_dev_type) that starts all applicable tests using GICv3 or GICv2. GICv2 tests are attempted if GICv3 is not available in the system. There are currently no GICv2 tests, but the test passes now in GICv2 systems.

---
 .../testing/selftests/kvm/aarch64/vgic_init.c | 156 +++++++++++-------
 1 file changed, 95 insertions(+), 61 deletions(-)
tools/testing/selftests/kvm/aarch64/vgic_init.c +80 -56
index 623f31a14326..896a29f2503d 100644
@@ -22,6 +22,9 @@
 
 #define GICR_TYPER 0x8
 
+#define VGIC_DEV_IS_V2(_d) ((_d) == KVM_DEV_TYPE_ARM_VGIC_V2)
+#define VGIC_DEV_IS_V3(_d) ((_d) == KVM_DEV_TYPE_ARM_VGIC_V3)
+
 struct vm_gic {
 	struct kvm_vm *vm;
 	int gic_fd;
@@ -30,8 +33,8 @@ struct vm_gic {
 static int max_ipa_bits;
 
 /* helper to access a redistributor register */
-static int access_redist_reg(int gicv3_fd, int vcpu, int offset,
-			     uint32_t *val, bool write)
+static int access_v3_redist_reg(int gicv3_fd, int vcpu, int offset,
+				uint32_t *val, bool write)
 {
 	uint64_t attr = REG_OFFSET(vcpu, offset);
 
@@ -58,7 +61,7 @@ static int run_vcpu(struct kvm_vm *vm, uint32_t vcpuid)
 	return 0;
 }
 
-static struct vm_gic vm_gic_create(void)
+static struct vm_gic vm_gic_v3_create(void)
 {
 	struct vm_gic v;
 
@@ -80,7 +83,7 @@ static void vm_gic_destroy(struct vm_gic *v)
  * device gets created, a legacy RDIST region is set at @0x0
  * and a DIST region is set @0x60000
  */
-static void subtest_dist_rdist(struct vm_gic *v)
+static void subtest_v3_dist_rdist(struct vm_gic *v)
 {
 	int ret;
 	uint64_t addr;
@@ -145,7 +148,7 @@ static void subtest_dist_rdist(struct vm_gic *v)
 }
 
 /* Test the new REDIST region API */
-static void subtest_redist_regions(struct vm_gic *v)
+static void subtest_v3_redist_regions(struct vm_gic *v)
 {
 	uint64_t addr, expected_addr;
 	int ret;
@@ -249,7 +252,7 @@ static void subtest_redist_regions(struct vm_gic *v)
  * VGIC KVM device is created and initialized before the secondary CPUs
  * get created
  */
-static void test_vgic_then_vcpus(void)
+static void test_v3_vgic_then_vcpus(uint32_t gic_dev_type)
 {
 	struct vm_gic v;
 	int ret, i;
@@ -257,7 +260,7 @@ static void test_vgic_then_vcpus(void)
 	v.vm = vm_create_default(0, 0, guest_code);
 	v.gic_fd = kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_V3, false);
 
-	subtest_dist_rdist(&v);
+	subtest_v3_dist_rdist(&v);
 
 	/* Add the rest of the VCPUs */
 	for (i = 1; i < NR_VCPUS; ++i)
@@ -270,14 +273,14 @@ static void test_vgic_then_vcpus(void)
 }
 
 /* All the VCPUs are created before the VGIC KVM device gets initialized */
-static void test_vcpus_then_vgic(void)
+static void test_v3_vcpus_then_vgic(uint32_t gic_dev_type)
 {
 	struct vm_gic v;
 	int ret;
 
-	v = vm_gic_create();
+	v = vm_gic_v3_create();
 
-	subtest_dist_rdist(&v);
+	subtest_v3_dist_rdist(&v);
 
 	ret = run_vcpu(v.vm, 3);
 	TEST_ASSERT(ret == -EINVAL, "dist/rdist overlap detected on 1st vcpu run");
@@ -285,15 +288,15 @@ static void test_vcpus_then_vgic(void)
 	vm_gic_destroy(&v);
 }
 
-static void test_new_redist_regions(void)
+static void test_v3_new_redist_regions(void)
 {
 	void *dummy = NULL;
 	struct vm_gic v;
 	uint64_t addr;
 	int ret;
 
-	v = vm_gic_create();
-	subtest_redist_regions(&v);
+	v = vm_gic_v3_create();
+	subtest_v3_redist_regions(&v);
 	kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_CTRL,
 			  KVM_DEV_ARM_VGIC_CTRL_INIT, NULL, true);
 
@@ -303,8 +306,8 @@ static void test_new_redist_regions(void)
 
 	/* step2 */
 
-	v = vm_gic_create();
-	subtest_redist_regions(&v);
+	v = vm_gic_v3_create();
+	subtest_v3_redist_regions(&v);
 
 	addr = REDIST_REGION_ATTR_ADDR(1, 0x280000, 0, 2);
 	kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
@@ -317,8 +320,8 @@ static void test_new_redist_regions(void)
 
 	/* step 3 */
 
-	v = vm_gic_create();
-	subtest_redist_regions(&v);
+	v = vm_gic_v3_create();
+	subtest_v3_redist_regions(&v);
 
 	_kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
 			  KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, dummy, true);
@@ -338,7 +341,7 @@ static void test_new_redist_regions(void)
 	vm_gic_destroy(&v);
 }
 
-static void test_typer_accesses(void)
+static void test_v3_typer_accesses(void)
 {
 	struct vm_gic v;
 	uint64_t addr;
@@ -351,12 +354,12 @@ static void test_typer_accesses(void)
 
 	vm_vcpu_add_default(v.vm, 3, guest_code);
 
-	ret = access_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
 	TEST_ASSERT(ret && errno == EINVAL, "attempting to read GICR_TYPER of non created vcpu");
 
 	vm_vcpu_add_default(v.vm, 1, guest_code);
 
-	ret = access_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
 	TEST_ASSERT(ret && errno == EBUSY, "read GICR_TYPER before GIC initialized");
 
 	vm_vcpu_add_default(v.vm, 2, guest_code);
@@ -365,7 +368,7 @@ static void test_typer_accesses(void)
 			  KVM_DEV_ARM_VGIC_CTRL_INIT, NULL, true);
 
 	for (i = 0; i < NR_VCPUS ; i++) {
-		ret = access_redist_reg(v.gic_fd, 0, GICR_TYPER, &val, false);
+		ret = access_v3_redist_reg(v.gic_fd, 0, GICR_TYPER, &val, false);
 		TEST_ASSERT(!ret && !val, "read GICR_TYPER before rdist region setting");
 	}
 
@@ -374,10 +377,10 @@ static void test_typer_accesses(void)
 			  KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);
 
 	/* The 2 first rdists should be put there (vcpu 0 and 3) */
-	ret = access_redist_reg(v.gic_fd, 0, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 0, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && !val, "read typer of rdist #0");
 
-	ret = access_redist_reg(v.gic_fd, 3, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 3, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x310, "read typer of rdist #1");
 
 	addr = REDIST_REGION_ATTR_ADDR(10, 0x100000, 0, 1);
@@ -385,11 +388,11 @@ static void test_typer_accesses(void)
 				 KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);
 	TEST_ASSERT(ret && errno == EINVAL, "collision with previous rdist region");
 
-	ret = access_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x100,
 		    "no redist region attached to vcpu #1 yet, last cannot be returned");
 
-	ret = access_redist_reg(v.gic_fd, 2, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 2, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x200,
 		    "no redist region attached to vcpu #2, last cannot be returned");
 
@@ -397,10 +400,10 @@ static void test_typer_accesses(void)
 	kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
 			  KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);
 
-	ret = access_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x100, "read typer of rdist #1");
 
-	ret = access_redist_reg(v.gic_fd, 2, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 2, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x210,
 		    "read typer of rdist #1, last properly returned");
 
@@ -417,7 +420,7 @@ static void test_typer_accesses(void)
  * rdist region #2 @0x200000 2 rdist capacity
  *     rdists: 1, 2
  */
-static void test_last_bit_redist_regions(void)
+static void test_v3_last_bit_redist_regions(void)
 {
 	uint32_t vcpuids[] = { 0, 3, 5, 4, 1, 2 };
 	struct vm_gic v;
@@ -444,29 +447,29 @@ static void test_last_bit_redist_regions(void)
 	kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
 			  KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);
 
-	ret = access_redist_reg(v.gic_fd, 0, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 0, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x000, "read typer of rdist #0");
 
-	ret = access_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x100, "read typer of rdist #1");
 
-	ret = access_redist_reg(v.gic_fd, 2, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 2, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x200, "read typer of rdist #2");
 
-	ret = access_redist_reg(v.gic_fd, 3, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 3, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x310, "read typer of rdist #3");
 
-	ret = access_redist_reg(v.gic_fd, 5, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 5, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x500, "read typer of rdist #5");
 
-	ret = access_redist_reg(v.gic_fd, 4, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 4, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x410, "read typer of rdist #4");
 
 	vm_gic_destroy(&v);
 }
 
 /* Test last bit with legacy region */
-static void test_last_bit_single_rdist(void)
+static void test_v3_last_bit_single_rdist(void)
 {
 	uint32_t vcpuids[] = { 0, 3, 5, 4, 1, 2 };
 	struct vm_gic v;
@@ -485,28 +488,32 @@ static void test_last_bit_single_rdist(void)
 	kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
 			  KVM_VGIC_V3_ADDR_TYPE_REDIST, &addr, true);
 
-	ret = access_redist_reg(v.gic_fd, 0, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 0, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x000, "read typer of rdist #0");
 
-	ret = access_redist_reg(v.gic_fd, 3, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 3, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x300, "read typer of rdist #1");
 
-	ret = access_redist_reg(v.gic_fd, 5, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 5, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x500, "read typer of rdist #2");
 
-	ret = access_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x100, "read typer of rdist #3");
 
-	ret = access_redist_reg(v.gic_fd, 2, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 2, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x210, "read typer of rdist #3");
 
 	vm_gic_destroy(&v);
 }
 
-void test_kvm_device(void)
+/*
+ * Returns 0 if it's possible to create GIC device of a given type (V2 or V3).
+ */
+int test_kvm_device(uint32_t gic_dev_type)
 {
 	struct vm_gic v;
 	int ret, fd;
+	uint32_t other;
 
 	v.vm = vm_create_default_with_vcpus(NR_VCPUS, 0, 0, guest_code, NULL);
 
@@ -514,38 +521,65 @@ void test_kvm_device(void)
 	ret = _kvm_create_device(v.vm, 0, true, &fd);
 	TEST_ASSERT(ret && errno == ENODEV, "unsupported device");
 
-	/* trial mode with VGIC_V3 device */
-	ret = _kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_V3, true, &fd);
-	if (ret) {
-		print_skip("GICv3 not supported");
-		exit(KSFT_SKIP);
-	}
-	v.gic_fd = kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_V3, false);
+	/* trial mode */
+	ret = _kvm_create_device(v.vm, gic_dev_type, true, &fd);
+	if (ret)
+		return ret;
+	v.gic_fd = kvm_create_device(v.vm, gic_dev_type, false);
+
+	ret = _kvm_create_device(v.vm, gic_dev_type, false, &fd);
+	TEST_ASSERT(ret && errno == EEXIST, "create GIC device twice");
 
-	ret = _kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_V3, false, &fd);
-	TEST_ASSERT(ret && errno == EEXIST, "create GICv3 device twice");
+	kvm_create_device(v.vm, gic_dev_type, true);
 
-	kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_V3, true);
+	/* try to create the other gic_dev_type */
+	other = VGIC_DEV_IS_V2(gic_dev_type) ? KVM_DEV_TYPE_ARM_VGIC_V3
+					     : KVM_DEV_TYPE_ARM_VGIC_V2;
 
-	if (!_kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_V2, true, &fd)) {
-		ret = _kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_V2, false, &fd);
-		TEST_ASSERT(ret && errno == EINVAL, "create GICv2 while v3 exists");
+	if (!_kvm_create_device(v.vm, other, true, &fd)) {
+		ret = _kvm_create_device(v.vm, other, false, &fd);
+		TEST_ASSERT(ret && errno == EINVAL,
+				"create GIC device while other version exists");
 	}
 
 	vm_gic_destroy(&v);
+
+	return 0;
+}
+
+void run_tests(uint32_t gic_dev_type)
+{
+	if (VGIC_DEV_IS_V3(gic_dev_type)) {
+		test_v3_vcpus_then_vgic(gic_dev_type);
+		test_v3_vgic_then_vcpus(gic_dev_type);
+		test_v3_new_redist_regions();
+		test_v3_typer_accesses();
+		test_v3_last_bit_redist_regions();
+		test_v3_last_bit_single_rdist();
+	}
 }
 
 int main(int ac, char **av)
 {
+	int ret;
+
 	max_ipa_bits = kvm_check_cap(KVM_CAP_ARM_VM_IPA_SIZE);
 
-	test_kvm_device();
-	test_vcpus_then_vgic();

tools/testing/selftests/kvm/aarch64/vgic_init.c void test_kvm_device(void) · line 572

Quoted 10 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +5 -5
-	test_vgic_then_vcpus();
-	test_new_redist_regions();
-	test_typer_accesses();
-	test_last_bit_redist_regions();
-	test_last_bit_single_rdist();
+	ret = test_kvm_device(KVM_DEV_TYPE_ARM_VGIC_V3);
+	if (!ret) {
+		pr_info("Running GIC_v3 tests.\n");
+		run_tests(KVM_DEV_TYPE_ARM_VGIC_V3);
+		return 0;

If the GICv3 supports compat with GICv2, I think you could be able to run both tests consecutively. So maybe don't return?

Besides looks good

Eric

Quoted 14 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +10 -0
+	}
+
+	ret = test_kvm_device(KVM_DEV_TYPE_ARM_VGIC_V2);
+	if (!ret) {
+		pr_info("Running GIC_v2 tests.\n");
+		run_tests(KVM_DEV_TYPE_ARM_VGIC_V2);
+		return 0;
+	}
 
 
+	print_skip("No GICv2 nor GICv3 support");
+	exit(KSFT_SKIP);
 	return 0;

}

#20 lore raw

Re: [PATCH v3 05/10] KVM: arm64: selftests: Make vgic_init gic version agnostic

Quoted 363 lines · Eric Auger
Quoted 360 lines · Ricardo Koller

As a preparation for the next commits which will add some tests for GICv2, make aarch64/vgic_init GIC version agnostic. Add a new generic run_tests function(gic_dev_type) that starts all applicable tests using GICv3 or GICv2. GICv2 tests are attempted if GICv3 is not available in the system. There are currently no GICv2 tests, but the test passes now in GICv2 systems.

---
 .../testing/selftests/kvm/aarch64/vgic_init.c | 156 +++++++++++-------
 1 file changed, 95 insertions(+), 61 deletions(-)
tools/testing/selftests/kvm/aarch64/vgic_init.c +80 -54
index 623f31a14326..896a29f2503d 100644
@@ -22,6 +22,9 @@
 
 #define GICR_TYPER 0x8
 
+#define VGIC_DEV_IS_V2(_d) ((_d) == KVM_DEV_TYPE_ARM_VGIC_V2)
+#define VGIC_DEV_IS_V3(_d) ((_d) == KVM_DEV_TYPE_ARM_VGIC_V3)
+
 struct vm_gic {
 	struct kvm_vm *vm;
 	int gic_fd;
@@ -30,8 +33,8 @@ struct vm_gic {
 static int max_ipa_bits;
 
 /* helper to access a redistributor register */
-static int access_redist_reg(int gicv3_fd, int vcpu, int offset,
-			     uint32_t *val, bool write)
+static int access_v3_redist_reg(int gicv3_fd, int vcpu, int offset,
+				uint32_t *val, bool write)
 {
 	uint64_t attr = REG_OFFSET(vcpu, offset);
 
@@ -58,7 +61,7 @@ static int run_vcpu(struct kvm_vm *vm, uint32_t vcpuid)
 	return 0;
 }
 
-static struct vm_gic vm_gic_create(void)
+static struct vm_gic vm_gic_v3_create(void)
 {
 	struct vm_gic v;
 
@@ -80,7 +83,7 @@ static void vm_gic_destroy(struct vm_gic *v)
  * device gets created, a legacy RDIST region is set at @0x0
  * and a DIST region is set @0x60000
  */
-static void subtest_dist_rdist(struct vm_gic *v)
+static void subtest_v3_dist_rdist(struct vm_gic *v)
 {
 	int ret;
 	uint64_t addr;
@@ -145,7 +148,7 @@ static void subtest_dist_rdist(struct vm_gic *v)
 }
 
 /* Test the new REDIST region API */
-static void subtest_redist_regions(struct vm_gic *v)
+static void subtest_v3_redist_regions(struct vm_gic *v)
 {
 	uint64_t addr, expected_addr;
 	int ret;
@@ -249,7 +252,7 @@ static void subtest_redist_regions(struct vm_gic *v)
  * VGIC KVM device is created and initialized before the secondary CPUs
  * get created
  */
-static void test_vgic_then_vcpus(void)
+static void test_v3_vgic_then_vcpus(uint32_t gic_dev_type)
 {
 	struct vm_gic v;
 	int ret, i;
@@ -257,7 +260,7 @@ static void test_vgic_then_vcpus(void)
 	v.vm = vm_create_default(0, 0, guest_code);
 	v.gic_fd = kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_V3, false);
 
-	subtest_dist_rdist(&v);
+	subtest_v3_dist_rdist(&v);
 
 	/* Add the rest of the VCPUs */
 	for (i = 1; i < NR_VCPUS; ++i)
@@ -270,14 +273,14 @@ static void test_vgic_then_vcpus(void)
 }
 
 /* All the VCPUs are created before the VGIC KVM device gets initialized */
-static void test_vcpus_then_vgic(void)
+static void test_v3_vcpus_then_vgic(uint32_t gic_dev_type)
 {
 	struct vm_gic v;
 	int ret;
 
-	v = vm_gic_create();
+	v = vm_gic_v3_create();
 
-	subtest_dist_rdist(&v);
+	subtest_v3_dist_rdist(&v);
 
 	ret = run_vcpu(v.vm, 3);
 	TEST_ASSERT(ret == -EINVAL, "dist/rdist overlap detected on 1st vcpu run");
@@ -285,15 +288,15 @@ static void test_vcpus_then_vgic(void)
 	vm_gic_destroy(&v);
 }
 
-static void test_new_redist_regions(void)
+static void test_v3_new_redist_regions(void)
 {
 	void *dummy = NULL;
 	struct vm_gic v;
 	uint64_t addr;
 	int ret;
 
-	v = vm_gic_create();
-	subtest_redist_regions(&v);
+	v = vm_gic_v3_create();
+	subtest_v3_redist_regions(&v);
 	kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_CTRL,
 			  KVM_DEV_ARM_VGIC_CTRL_INIT, NULL, true);
 
@@ -303,8 +306,8 @@ static void test_new_redist_regions(void)
 
 	/* step2 */
 
-	v = vm_gic_create();
-	subtest_redist_regions(&v);
+	v = vm_gic_v3_create();
+	subtest_v3_redist_regions(&v);
 
 	addr = REDIST_REGION_ATTR_ADDR(1, 0x280000, 0, 2);
 	kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
@@ -317,8 +320,8 @@ static void test_new_redist_regions(void)
 
 	/* step 3 */
 
-	v = vm_gic_create();
-	subtest_redist_regions(&v);
+	v = vm_gic_v3_create();
+	subtest_v3_redist_regions(&v);
 
 	_kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
 			  KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, dummy, true);
@@ -338,7 +341,7 @@ static void test_new_redist_regions(void)
 	vm_gic_destroy(&v);
 }
 
-static void test_typer_accesses(void)
+static void test_v3_typer_accesses(void)
 {
 	struct vm_gic v;
 	uint64_t addr;
@@ -351,12 +354,12 @@ static void test_typer_accesses(void)
 
 	vm_vcpu_add_default(v.vm, 3, guest_code);
 
-	ret = access_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
 	TEST_ASSERT(ret && errno == EINVAL, "attempting to read GICR_TYPER of non created vcpu");
 
 	vm_vcpu_add_default(v.vm, 1, guest_code);
 
-	ret = access_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
 	TEST_ASSERT(ret && errno == EBUSY, "read GICR_TYPER before GIC initialized");
 
 	vm_vcpu_add_default(v.vm, 2, guest_code);
@@ -365,7 +368,7 @@ static void test_typer_accesses(void)
 			  KVM_DEV_ARM_VGIC_CTRL_INIT, NULL, true);
 
 	for (i = 0; i < NR_VCPUS ; i++) {
-		ret = access_redist_reg(v.gic_fd, 0, GICR_TYPER, &val, false);
+		ret = access_v3_redist_reg(v.gic_fd, 0, GICR_TYPER, &val, false);
 		TEST_ASSERT(!ret && !val, "read GICR_TYPER before rdist region setting");
 	}
 
@@ -374,10 +377,10 @@ static void test_typer_accesses(void)
 			  KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);
 
 	/* The 2 first rdists should be put there (vcpu 0 and 3) */
-	ret = access_redist_reg(v.gic_fd, 0, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 0, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && !val, "read typer of rdist #0");
 
-	ret = access_redist_reg(v.gic_fd, 3, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 3, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x310, "read typer of rdist #1");
 
 	addr = REDIST_REGION_ATTR_ADDR(10, 0x100000, 0, 1);
@@ -385,11 +388,11 @@ static void test_typer_accesses(void)
 				 KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);
 	TEST_ASSERT(ret && errno == EINVAL, "collision with previous rdist region");
 
-	ret = access_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x100,
 		    "no redist region attached to vcpu #1 yet, last cannot be returned");
 
-	ret = access_redist_reg(v.gic_fd, 2, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 2, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x200,
 		    "no redist region attached to vcpu #2, last cannot be returned");
 
@@ -397,10 +400,10 @@ static void test_typer_accesses(void)
 	kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
 			  KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);
 
-	ret = access_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x100, "read typer of rdist #1");
 
-	ret = access_redist_reg(v.gic_fd, 2, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 2, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x210,
 		    "read typer of rdist #1, last properly returned");
 
@@ -417,7 +420,7 @@ static void test_typer_accesses(void)
  * rdist region #2 @0x200000 2 rdist capacity
  *     rdists: 1, 2
  */
-static void test_last_bit_redist_regions(void)
+static void test_v3_last_bit_redist_regions(void)
 {
 	uint32_t vcpuids[] = { 0, 3, 5, 4, 1, 2 };
 	struct vm_gic v;
@@ -444,29 +447,29 @@ static void test_last_bit_redist_regions(void)
 	kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
 			  KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);
 
-	ret = access_redist_reg(v.gic_fd, 0, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 0, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x000, "read typer of rdist #0");
 
-	ret = access_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x100, "read typer of rdist #1");
 
-	ret = access_redist_reg(v.gic_fd, 2, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 2, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x200, "read typer of rdist #2");
 
-	ret = access_redist_reg(v.gic_fd, 3, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 3, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x310, "read typer of rdist #3");
 
-	ret = access_redist_reg(v.gic_fd, 5, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 5, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x500, "read typer of rdist #5");
 
-	ret = access_redist_reg(v.gic_fd, 4, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 4, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x410, "read typer of rdist #4");
 
 	vm_gic_destroy(&v);
 }
 
 /* Test last bit with legacy region */
-static void test_last_bit_single_rdist(void)
+static void test_v3_last_bit_single_rdist(void)
 {
 	uint32_t vcpuids[] = { 0, 3, 5, 4, 1, 2 };
 	struct vm_gic v;
@@ -485,28 +488,32 @@ static void test_last_bit_single_rdist(void)
 	kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
 			  KVM_VGIC_V3_ADDR_TYPE_REDIST, &addr, true);
 
-	ret = access_redist_reg(v.gic_fd, 0, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 0, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x000, "read typer of rdist #0");
 
-	ret = access_redist_reg(v.gic_fd, 3, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 3, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x300, "read typer of rdist #1");
 
-	ret = access_redist_reg(v.gic_fd, 5, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 5, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x500, "read typer of rdist #2");
 
-	ret = access_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 1, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x100, "read typer of rdist #3");
 
-	ret = access_redist_reg(v.gic_fd, 2, GICR_TYPER, &val, false);
+	ret = access_v3_redist_reg(v.gic_fd, 2, GICR_TYPER, &val, false);
 	TEST_ASSERT(!ret && val == 0x210, "read typer of rdist #3");
 
 	vm_gic_destroy(&v);
 }
 
-void test_kvm_device(void)
+/*
+ * Returns 0 if it's possible to create GIC device of a given type (V2 or V3).
+ */
+int test_kvm_device(uint32_t gic_dev_type)
 {
 	struct vm_gic v;
 	int ret, fd;
+	uint32_t other;
 
 	v.vm = vm_create_default_with_vcpus(NR_VCPUS, 0, 0, guest_code, NULL);
 
@@ -514,38 +521,65 @@ void test_kvm_device(void)
 	ret = _kvm_create_device(v.vm, 0, true, &fd);
 	TEST_ASSERT(ret && errno == ENODEV, "unsupported device");
 
-	/* trial mode with VGIC_V3 device */
-	ret = _kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_V3, true, &fd);
-	if (ret) {
-		print_skip("GICv3 not supported");
-		exit(KSFT_SKIP);
-	}
-	v.gic_fd = kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_V3, false);
+	/* trial mode */
+	ret = _kvm_create_device(v.vm, gic_dev_type, true, &fd);
+	if (ret)
+		return ret;
+	v.gic_fd = kvm_create_device(v.vm, gic_dev_type, false);
+
+	ret = _kvm_create_device(v.vm, gic_dev_type, false, &fd);
+	TEST_ASSERT(ret && errno == EEXIST, "create GIC device twice");
 
-	ret = _kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_V3, false, &fd);
-	TEST_ASSERT(ret && errno == EEXIST, "create GICv3 device twice");
+	kvm_create_device(v.vm, gic_dev_type, true);
 
-	kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_V3, true);
+	/* try to create the other gic_dev_type */
+	other = VGIC_DEV_IS_V2(gic_dev_type) ? KVM_DEV_TYPE_ARM_VGIC_V3
+					     : KVM_DEV_TYPE_ARM_VGIC_V2;
 
-	if (!_kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_V2, true, &fd)) {
-		ret = _kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_V2, false, &fd);
-		TEST_ASSERT(ret && errno == EINVAL, "create GICv2 while v3 exists");
+	if (!_kvm_create_device(v.vm, other, true, &fd)) {
+		ret = _kvm_create_device(v.vm, other, false, &fd);
+		TEST_ASSERT(ret && errno == EINVAL,
+				"create GIC device while other version exists");
 	}
 
 	vm_gic_destroy(&v);
+
+	return 0;
+}
+
+void run_tests(uint32_t gic_dev_type)
+{
+	if (VGIC_DEV_IS_V3(gic_dev_type)) {
+		test_v3_vcpus_then_vgic(gic_dev_type);
+		test_v3_vgic_then_vcpus(gic_dev_type);
+		test_v3_new_redist_regions();
+		test_v3_typer_accesses();
+		test_v3_last_bit_redist_regions();
+		test_v3_last_bit_single_rdist();
+	}
 }
 
 int main(int ac, char **av)
 {
+	int ret;
+
 	max_ipa_bits = kvm_check_cap(KVM_CAP_ARM_VM_IPA_SIZE);
 
Quoted 15 lines

tools/testing/selftests/kvm/aarch64/vgic_init.c void test_kvm_device(void) · line 572

Quoted 12 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +5 -7
-	test_kvm_device();
-	test_vcpus_then_vgic();
-	test_vgic_then_vcpus();
-	test_new_redist_regions();
-	test_typer_accesses();
-	test_last_bit_redist_regions();
-	test_last_bit_single_rdist();
+	ret = test_kvm_device(KVM_DEV_TYPE_ARM_VGIC_V3);
+	if (!ret) {
+		pr_info("Running GIC_v3 tests.\n");
+		run_tests(KVM_DEV_TYPE_ARM_VGIC_V3);
+		return 0;

If the GICv3 supports compat with GICv2, I think you could be able to run both tests consecutively. So maybe don't return?

Good point, will try that (not returning).

Thanks,
Ricardo

Quoted 15 lines

Besides looks good

Eric

Quoted 10 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +8 -0
+	}
+
+	ret = test_kvm_device(KVM_DEV_TYPE_ARM_VGIC_V2);
+	if (!ret) {
+		pr_info("Running GIC_v2 tests.\n");
+		run_tests(KVM_DEV_TYPE_ARM_VGIC_V2);
+		return 0;
+	}
 
 
Quoted 5 lines
Quoted 4 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +2 -0
+	print_skip("No GICv2 nor GICv3 support");
+	exit(KSFT_SKIP);
 	return 0;

}

_______________________________________________
kvmarm mailing list
[email protected]
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm

maintainer
#21 lore raw

Re: [PATCH v3 05/10] KVM: arm64: selftests: Make vgic_init gic version agnostic

In reply to Eric Auger
Quoted 3 lines · Eric Auger

If the GICv3 supports compat with GICv2, I think you could be able to run both tests consecutively. So maybe don't return?

You'll need to recreate a full VM though. Even if the HW supports the compat mode, our GICv3 emulation doesn't.

Thanks,

	M.
Signature · 5 lines
Without deviation from the norm, progress is not possible.
_______________________________________________
kvmarm mailing list
[email protected]
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
#22 lore raw

Re: [PATCH v3 05/10] KVM: arm64: selftests: Make vgic_init gic version agnostic

Quoted 8 lines · Marc Zyngier
Quoted 3 lines · Eric Auger

If the GICv3 supports compat with GICv2, I think you could be able to run both tests consecutively. So maybe don't return?

You'll need to recreate a full VM though. Even if the HW supports the compat mode, our GICv3 emulation doesn't.

I'm not sure how much work would that entail, but sounds like it might be too much for a selftest.

Thanks,
Ricardo

Quoted 7 lines

Thanks,

	M.
Signature · 1 line
Without deviation from the norm, progress is not possible.
#23 lore raw

PATCH v3 6/10 KVM: arm64: selftests: Make vgic_init/vm_gic_create version agnostic

In reply to Ricardo Koller

Make vm_gic_create GIC version agnostic in the vgic_init test. Also add a nr_vcpus arg into it instead of defaulting to NR_VCPUS.

No functional change.

---
 .../testing/selftests/kvm/aarch64/vgic_init.c | 21 +++++++++++--------
 1 file changed, 12 insertions(+), 9 deletions(-)
tools/testing/selftests/kvm/aarch64/vgic_init.c +12 -9
index 896a29f2503d..b24067dbdac0 100644
@@ -28,6 +28,7 @@
 struct vm_gic {
 	struct kvm_vm *vm;
 	int gic_fd;
+	uint32_t gic_dev_type;
 };
 
 static int max_ipa_bits;
@@ -61,12 +62,13 @@ static int run_vcpu(struct kvm_vm *vm, uint32_t vcpuid)
 	return 0;
 }
 
-static struct vm_gic vm_gic_v3_create(void)
+static struct vm_gic vm_gic_create_with_vcpus(uint32_t gic_dev_type, uint32_t nr_vcpus)
 {
 	struct vm_gic v;
 
-	v.vm = vm_create_default_with_vcpus(NR_VCPUS, 0, 0, guest_code, NULL);
-	v.gic_fd = kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_V3, false);
+	v.gic_dev_type = gic_dev_type;
+	v.vm = vm_create_default_with_vcpus(nr_vcpus, 0, 0, guest_code, NULL);
+	v.gic_fd = kvm_create_device(v.vm, gic_dev_type, false);
 
 	return v;
 }
@@ -153,6 +155,8 @@ static void subtest_v3_redist_regions(struct vm_gic *v)
 	uint64_t addr, expected_addr;
 	int ret;
 
+	TEST_ASSERT(VGIC_DEV_IS_V3(v->gic_dev_type), "Only applies to v3");
+
 	ret = kvm_device_check_attr(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
 				     KVM_VGIC_V3_ADDR_TYPE_REDIST);
 	TEST_ASSERT(!ret, "Multiple redist regions advertised");
@@ -257,8 +261,7 @@ static void test_v3_vgic_then_vcpus(uint32_t gic_dev_type)
 	struct vm_gic v;
 	int ret, i;
 
-	v.vm = vm_create_default(0, 0, guest_code);
-	v.gic_fd = kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_V3, false);
+	v = vm_gic_create_with_vcpus(gic_dev_type, 1);
 
 	subtest_v3_dist_rdist(&v);
 
@@ -278,7 +281,7 @@ static void test_v3_vcpus_then_vgic(uint32_t gic_dev_type)
 	struct vm_gic v;
 	int ret;
 
-	v = vm_gic_v3_create();
+	v = vm_gic_create_with_vcpus(gic_dev_type, NR_VCPUS);
 
 	subtest_v3_dist_rdist(&v);
 
@@ -295,7 +298,7 @@ static void test_v3_new_redist_regions(void)
 	uint64_t addr;
 	int ret;
 
-	v = vm_gic_v3_create();
+	v = vm_gic_create_with_vcpus(KVM_DEV_TYPE_ARM_VGIC_V3, NR_VCPUS);
 	subtest_v3_redist_regions(&v);
 	kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_CTRL,
 			  KVM_DEV_ARM_VGIC_CTRL_INIT, NULL, true);
@@ -306,7 +309,7 @@ static void test_v3_new_redist_regions(void)
 
 	/* step2 */
 
-	v = vm_gic_v3_create();
+	v = vm_gic_create_with_vcpus(KVM_DEV_TYPE_ARM_VGIC_V3, NR_VCPUS);
 	subtest_v3_redist_regions(&v);
 
 	addr = REDIST_REGION_ATTR_ADDR(1, 0x280000, 0, 2);
@@ -320,7 +323,7 @@ static void test_v3_new_redist_regions(void)
 
 	/* step 3 */
 
-	v = vm_gic_v3_create();
+	v = vm_gic_create_with_vcpus(KVM_DEV_TYPE_ARM_VGIC_V3, NR_VCPUS);
 	subtest_v3_redist_regions(&v);
 
 	_kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
Signature · 1 line
2.33.0.685.g46640cef36-goog
Reviewed-by on v3 6/10
#24 lore raw

Re: [PATCH v3 06/10] KVM: arm64: selftests: Make vgic_init/vm_gic_create version agnostic

Quoted 40 lines · Ricardo Koller

Make vm_gic_create GIC version agnostic in the vgic_init test. Also add a nr_vcpus arg into it instead of defaulting to NR_VCPUS.

No functional change.

---
 .../testing/selftests/kvm/aarch64/vgic_init.c | 21 +++++++++++--------
 1 file changed, 12 insertions(+), 9 deletions(-)
tools/testing/selftests/kvm/aarch64/vgic_init.c +5 -3
index 896a29f2503d..b24067dbdac0 100644
@@ -28,6 +28,7 @@
 struct vm_gic {
 	struct kvm_vm *vm;
 	int gic_fd;
+	uint32_t gic_dev_type;
 };
 
 static int max_ipa_bits;
@@ -61,12 +62,13 @@ static int run_vcpu(struct kvm_vm *vm, uint32_t vcpuid)
 	return 0;
 }
 
-static struct vm_gic vm_gic_v3_create(void)
+static struct vm_gic vm_gic_create_with_vcpus(uint32_t gic_dev_type, uint32_t nr_vcpus)
 {
 	struct vm_gic v;
 
-	v.vm = vm_create_default_with_vcpus(NR_VCPUS, 0, 0, guest_code, NULL);
-	v.gic_fd = kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_V3, false);
+	v.gic_dev_type = gic_dev_type;
+	v.vm = vm_create_default_with_vcpus(nr_vcpus, 0, 0, guest_code, NULL);
+	v.gic_fd = kvm_create_device(v.vm, gic_dev_type, false);
 
 	return v;
 }

tools/testing/selftests/kvm/aarch64/vgic_init.c static void subtest_v3_redist_regions(struct vm_gic *v) · line 158

Quoted 4 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +1 -0
@@ -153,6 +155,8 @@ static void subtest_v3_redist_regions(struct vm_gic *v)
 	uint64_t addr, expected_addr;
 	int ret;
 
+	TEST_ASSERT(VGIC_DEV_IS_V3(v->gic_dev_type), "Only applies to v3");

Is that really needed? why here and not in other locations. I would remove. Besides

Eric

Quoted 50 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +6 -6
+
 	ret = kvm_device_check_attr(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
 				     KVM_VGIC_V3_ADDR_TYPE_REDIST);
 	TEST_ASSERT(!ret, "Multiple redist regions advertised");
@@ -257,8 +261,7 @@ static void test_v3_vgic_then_vcpus(uint32_t gic_dev_type)
 	struct vm_gic v;
 	int ret, i;
 
-	v.vm = vm_create_default(0, 0, guest_code);
-	v.gic_fd = kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_V3, false);
+	v = vm_gic_create_with_vcpus(gic_dev_type, 1);
 
 	subtest_v3_dist_rdist(&v);
 
@@ -278,7 +281,7 @@ static void test_v3_vcpus_then_vgic(uint32_t gic_dev_type)
 	struct vm_gic v;
 	int ret;
 
-	v = vm_gic_v3_create();
+	v = vm_gic_create_with_vcpus(gic_dev_type, NR_VCPUS);
 
 	subtest_v3_dist_rdist(&v);
 
@@ -295,7 +298,7 @@ static void test_v3_new_redist_regions(void)
 	uint64_t addr;
 	int ret;
 
-	v = vm_gic_v3_create();
+	v = vm_gic_create_with_vcpus(KVM_DEV_TYPE_ARM_VGIC_V3, NR_VCPUS);
 	subtest_v3_redist_regions(&v);
 	kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_CTRL,
 			  KVM_DEV_ARM_VGIC_CTRL_INIT, NULL, true);
@@ -306,7 +309,7 @@ static void test_v3_new_redist_regions(void)
 
 	/* step2 */
 
-	v = vm_gic_v3_create();
+	v = vm_gic_create_with_vcpus(KVM_DEV_TYPE_ARM_VGIC_V3, NR_VCPUS);
 	subtest_v3_redist_regions(&v);
 
 	addr = REDIST_REGION_ATTR_ADDR(1, 0x280000, 0, 2);
@@ -320,7 +323,7 @@ static void test_v3_new_redist_regions(void)
 
 	/* step 3 */
 
-	v = vm_gic_v3_create();
+	v = vm_gic_create_with_vcpus(KVM_DEV_TYPE_ARM_VGIC_V3, NR_VCPUS);
 	subtest_v3_redist_regions(&v);
 
 	_kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
#25 lore raw

PATCH v3 7/10 KVM: arm64: selftests: Add some tests for GICv2 in vgic_init

In reply to Ricardo Koller

Add some GICv2 tests: general KVM device tests and DIST/REDIST overlap tests. Do this by making test_vcpus_then_vgic and test_vgic_then_vcpus in vgic_init GIC version agnostic.

---
 .../testing/selftests/kvm/aarch64/vgic_init.c | 107 ++++++++++++------
 1 file changed, 75 insertions(+), 32 deletions(-)
tools/testing/selftests/kvm/aarch64/vgic_init.c +75 -32
index b24067dbdac0..92f5c6ca6b8b 100644
@@ -79,74 +79,116 @@ static void vm_gic_destroy(struct vm_gic *v)
 	kvm_vm_free(v->vm);
 }
 
+struct vgic_region_attr {
+	uint64_t attr;
+	uint64_t size;
+	uint64_t alignment;
+};
+
+struct vgic_region_attr gic_v3_dist_region = {
+	.attr = KVM_VGIC_V3_ADDR_TYPE_DIST,
+	.size = 0x10000,
+	.alignment = 0x10000,
+};
+
+struct vgic_region_attr gic_v3_redist_region = {
+	.attr = KVM_VGIC_V3_ADDR_TYPE_REDIST,
+	.size = NR_VCPUS * 0x20000,
+	.alignment = 0x10000,
+};
+
+struct vgic_region_attr gic_v2_dist_region = {
+	.attr = KVM_VGIC_V2_ADDR_TYPE_DIST,
+	.size = 0x1000,
+	.alignment = 0x1000,
+};
+
+struct vgic_region_attr gic_v2_cpu_region = {
+	.attr = KVM_VGIC_V2_ADDR_TYPE_CPU,
+	.size = 0x2000,
+	.alignment = 0x1000,
+};
+
 /**
- * Helper routine that performs KVM device tests in general and
- * especially ARM_VGIC_V3 ones. Eventually the ARM_VGIC_V3
- * device gets created, a legacy RDIST region is set at @0x0
- * and a DIST region is set @0x60000
+ * Helper routine that performs KVM device tests in general. Eventually the
+ * ARM_VGIC (GICv2 or GICv3) device gets created with an overlapping
+ * DIST/REDIST. A RDIST region (legacy in the case of GICv3) is set at @0x0 and
+ * a DIST region is set @0x70000 for GICv3 and @0x1000 for GICv2.
  */
-static void subtest_v3_dist_rdist(struct vm_gic *v)
+static void subtest_dist_rdist(struct vm_gic *v)
 {
 	int ret;
 	uint64_t addr;
+	struct vgic_region_attr rdist; /* CPU interface in GICv2*/
+	struct vgic_region_attr dist;
+
+	rdist = VGIC_DEV_IS_V3(v->gic_dev_type) ? gic_v3_redist_region
+						: gic_v2_cpu_region;
+	dist = VGIC_DEV_IS_V3(v->gic_dev_type) ? gic_v3_dist_region
+						: gic_v2_dist_region;
 
 	/* Check existing group/attributes */
 	kvm_device_check_attr(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-			      KVM_VGIC_V3_ADDR_TYPE_DIST);
+			      dist.attr);
 
 	kvm_device_check_attr(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-			      KVM_VGIC_V3_ADDR_TYPE_REDIST);
+			      rdist.attr);
 
 	/* check non existing attribute */
-	ret = _kvm_device_check_attr(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR, 0);
+	ret = _kvm_device_check_attr(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR, -1);
 	TEST_ASSERT(ret && errno == ENXIO, "attribute not supported");
 
 	/* misaligned DIST and REDIST address settings */
-	addr = 0x1000;
+	addr = dist.alignment / 0x10;
 	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-				 KVM_VGIC_V3_ADDR_TYPE_DIST, &addr, true);
-	TEST_ASSERT(ret && errno == EINVAL, "GICv3 dist base not 64kB aligned");
+				 dist.attr, &addr, true);
+	TEST_ASSERT(ret && errno == EINVAL, "GIC dist base not aligned");
 
+	addr = rdist.alignment / 0x10;
 	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-				 KVM_VGIC_V3_ADDR_TYPE_REDIST, &addr, true);
-	TEST_ASSERT(ret && errno == EINVAL, "GICv3 redist base not 64kB aligned");
+				 rdist.attr, &addr, true);
+	TEST_ASSERT(ret && errno == EINVAL, "GIC redist/cpu base not aligned");
 
 	/* out of range address */
 	if (max_ipa_bits) {
 		addr = 1ULL << max_ipa_bits;
 		ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-					 KVM_VGIC_V3_ADDR_TYPE_DIST, &addr, true);
+					 dist.attr, &addr, true);
 		TEST_ASSERT(ret && errno == E2BIG, "dist address beyond IPA limit");
 
 		ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-					 KVM_VGIC_V3_ADDR_TYPE_REDIST, &addr, true);
+					 rdist.attr, &addr, true);
 		TEST_ASSERT(ret && errno == E2BIG, "redist address beyond IPA limit");
 	}
 
 	/* set REDIST base address @0x0*/
 	addr = 0x00000;
 	kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-			  KVM_VGIC_V3_ADDR_TYPE_REDIST, &addr, true);
+			  rdist.attr, &addr, true);
 
 	/* Attempt to create a second legacy redistributor region */
 	addr = 0xE0000;
 	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-				 KVM_VGIC_V3_ADDR_TYPE_REDIST, &addr, true);
-	TEST_ASSERT(ret && errno == EEXIST, "GICv3 redist base set again");
+				 rdist.attr, &addr, true);
+	TEST_ASSERT(ret && errno == EEXIST, "GIC redist base set again");
 
-	/* Attempt to mix legacy and new redistributor regions */
-	addr = REDIST_REGION_ATTR_ADDR(NR_VCPUS, 0x100000, 0, 0);
-	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-				 KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);
-	TEST_ASSERT(ret && errno == EINVAL, "attempt to mix GICv3 REDIST and REDIST_REGION");
+	if (VGIC_DEV_IS_V3(v->gic_dev_type)) {
+		/* Attempt to mix legacy and new redistributor regions */
+		addr = REDIST_REGION_ATTR_ADDR(NR_VCPUS, 0x100000, 0, 0);
+		ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+					 KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION,
+					 &addr, true);
+		TEST_ASSERT(ret && errno == EINVAL,
+			    "attempt to mix GICv3 REDIST and REDIST_REGION");
+	}
 
 	/*
 	 * Set overlapping DIST / REDIST, cannot be detected here. Will be detected
 	 * on first vcpu run instead.
 	 */
-	addr = 3 * 2 * 0x10000;
-	kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR, KVM_VGIC_V3_ADDR_TYPE_DIST,
-			  &addr, true);
+	addr = rdist.size - rdist.alignment;
+	kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  dist.attr, &addr, true);
 }
 
 /* Test the new REDIST region API */
@@ -256,14 +298,14 @@ static void subtest_v3_redist_regions(struct vm_gic *v)
  * VGIC KVM device is created and initialized before the secondary CPUs
  * get created
  */
-static void test_v3_vgic_then_vcpus(uint32_t gic_dev_type)
+static void test_vgic_then_vcpus(uint32_t gic_dev_type)
 {
 	struct vm_gic v;
 	int ret, i;
 
 	v = vm_gic_create_with_vcpus(gic_dev_type, 1);
 
-	subtest_v3_dist_rdist(&v);
+	subtest_dist_rdist(&v);
 
 	/* Add the rest of the VCPUs */
 	for (i = 1; i < NR_VCPUS; ++i)
@@ -276,14 +318,14 @@ static void test_v3_vgic_then_vcpus(uint32_t gic_dev_type)
 }
 
 /* All the VCPUs are created before the VGIC KVM device gets initialized */
-static void test_v3_vcpus_then_vgic(uint32_t gic_dev_type)
+static void test_vcpus_then_vgic(uint32_t gic_dev_type)
 {
 	struct vm_gic v;
 	int ret;
 
 	v = vm_gic_create_with_vcpus(gic_dev_type, NR_VCPUS);
 
-	subtest_v3_dist_rdist(&v);
+	subtest_dist_rdist(&v);
 
 	ret = run_vcpu(v.vm, 3);
 	TEST_ASSERT(ret == -EINVAL, "dist/rdist overlap detected on 1st vcpu run");
@@ -552,9 +594,10 @@ int test_kvm_device(uint32_t gic_dev_type)
 
 void run_tests(uint32_t gic_dev_type)
 {
+	test_vcpus_then_vgic(gic_dev_type);
+	test_vgic_then_vcpus(gic_dev_type);
+
 	if (VGIC_DEV_IS_V3(gic_dev_type)) {
-		test_v3_vcpus_then_vgic(gic_dev_type);
-		test_v3_vgic_then_vcpus(gic_dev_type);
 		test_v3_new_redist_regions();
 		test_v3_typer_accesses();
 		test_v3_last_bit_redist_regions();
Signature · 6 lines
2.33.0.685.g46640cef36-goog

_______________________________________________
kvmarm mailing list
[email protected]
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
#26 lore raw

Re: [PATCH v3 07/10] KVM: arm64: selftests: Add some tests for GICv2 in vgic_init

Hi Ricardo,

Quoted 46 lines · Ricardo Koller

Add some GICv2 tests: general KVM device tests and DIST/REDIST overlap tests. Do this by making test_vcpus_then_vgic and test_vgic_then_vcpus in vgic_init GIC version agnostic.

---
 .../testing/selftests/kvm/aarch64/vgic_init.c | 107 ++++++++++++------
 1 file changed, 75 insertions(+), 32 deletions(-)
tools/testing/selftests/kvm/aarch64/vgic_init.c +29 -0
index b24067dbdac0..92f5c6ca6b8b 100644
@@ -79,74 +79,116 @@ static void vm_gic_destroy(struct vm_gic *v)
 	kvm_vm_free(v->vm);
 }
 
+struct vgic_region_attr {
+	uint64_t attr;
+	uint64_t size;
+	uint64_t alignment;
+};
+
+struct vgic_region_attr gic_v3_dist_region = {
+	.attr = KVM_VGIC_V3_ADDR_TYPE_DIST,
+	.size = 0x10000,
+	.alignment = 0x10000,
+};
+
+struct vgic_region_attr gic_v3_redist_region = {
+	.attr = KVM_VGIC_V3_ADDR_TYPE_REDIST,
+	.size = NR_VCPUS * 0x20000,
+	.alignment = 0x10000,
+};
+
+struct vgic_region_attr gic_v2_dist_region = {
+	.attr = KVM_VGIC_V2_ADDR_TYPE_DIST,
+	.size = 0x1000,
+	.alignment = 0x1000,
+};
+
+struct vgic_region_attr gic_v2_cpu_region = {
+	.attr = KVM_VGIC_V2_ADDR_TYPE_CPU,
+	.size = 0x2000,
+	.alignment = 0x1000,
+};

tools/testing/selftests/kvm/aarch64/vgic_init.c static void vm_gic_destroy(struct vm_gic *v) · line 116

Quoted 10 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +5 -4
+
 /**
- * Helper routine that performs KVM device tests in general and
- * especially ARM_VGIC_V3 ones. Eventually the ARM_VGIC_V3
- * device gets created, a legacy RDIST region is set at @0x0
- * and a DIST region is set @0x60000
+ * Helper routine that performs KVM device tests in general. Eventually the
+ * ARM_VGIC (GICv2 or GICv3) device gets created with an overlapping
+ * DIST/REDIST. A RDIST region (legacy in the case of GICv3) is set at @0x0 and
+ * a DIST region is set @0x70000 for GICv3 and @0x1000 for GICv2.

I would add "Assumption is 4 vcpus are going to be used hence the overlap". Also the RDIST is GICv3 only. In the above comment also mention the CPU I/F.

Quoted 16 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +8 -1
  */
-static void subtest_v3_dist_rdist(struct vm_gic *v)
+static void subtest_dist_rdist(struct vm_gic *v)
 {
 	int ret;
 	uint64_t addr;
+	struct vgic_region_attr rdist; /* CPU interface in GICv2*/
+	struct vgic_region_attr dist;
+
+	rdist = VGIC_DEV_IS_V3(v->gic_dev_type) ? gic_v3_redist_region
+						: gic_v2_cpu_region;
+	dist = VGIC_DEV_IS_V3(v->gic_dev_type) ? gic_v3_dist_region
+						: gic_v2_dist_region;
 
 	/* Check existing group/attributes */
 	kvm_device_check_attr(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,

tools/testing/selftests/kvm/aarch64/vgic_init.c static void vm_gic_destroy(struct vm_gic *v) · line 138

Quoted 10 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +3 -3
-			      KVM_VGIC_V3_ADDR_TYPE_DIST);
+			      dist.attr);
 
 	kvm_device_check_attr(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-			      KVM_VGIC_V3_ADDR_TYPE_REDIST);
+			      rdist.attr);
 
 	/* check non existing attribute */
-	ret = _kvm_device_check_attr(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR, 0);
+	ret = _kvm_device_check_attr(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR, -1);

was that necessary?

Quoted 42 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +9 -9
 	TEST_ASSERT(ret && errno == ENXIO, "attribute not supported");
 
 	/* misaligned DIST and REDIST address settings */
-	addr = 0x1000;
+	addr = dist.alignment / 0x10;
 	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-				 KVM_VGIC_V3_ADDR_TYPE_DIST, &addr, true);
-	TEST_ASSERT(ret && errno == EINVAL, "GICv3 dist base not 64kB aligned");
+				 dist.attr, &addr, true);
+	TEST_ASSERT(ret && errno == EINVAL, "GIC dist base not aligned");
 
+	addr = rdist.alignment / 0x10;
 	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-				 KVM_VGIC_V3_ADDR_TYPE_REDIST, &addr, true);
-	TEST_ASSERT(ret && errno == EINVAL, "GICv3 redist base not 64kB aligned");
+				 rdist.attr, &addr, true);
+	TEST_ASSERT(ret && errno == EINVAL, "GIC redist/cpu base not aligned");
 
 	/* out of range address */
 	if (max_ipa_bits) {
 		addr = 1ULL << max_ipa_bits;
 		ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-					 KVM_VGIC_V3_ADDR_TYPE_DIST, &addr, true);
+					 dist.attr, &addr, true);
 		TEST_ASSERT(ret && errno == E2BIG, "dist address beyond IPA limit");
 
 		ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-					 KVM_VGIC_V3_ADDR_TYPE_REDIST, &addr, true);
+					 rdist.attr, &addr, true);
 		TEST_ASSERT(ret && errno == E2BIG, "redist address beyond IPA limit");
 	}
 
 	/* set REDIST base address @0x0*/
 	addr = 0x00000;
 	kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-			  KVM_VGIC_V3_ADDR_TYPE_REDIST, &addr, true);
+			  rdist.attr, &addr, true);
 
 	/* Attempt to create a second legacy redistributor region */
 	addr = 0xE0000;
 	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-				 KVM_VGIC_V3_ADDR_TYPE_REDIST, &addr, true);

tools/testing/selftests/kvm/aarch64/vgic_init.c static void vm_gic_destroy(struct vm_gic *v) · line 175

Quoted 10 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +3 -6
-	TEST_ASSERT(ret && errno == EEXIST, "GICv3 redist base set again");
+				 rdist.attr, &addr, true);
+	TEST_ASSERT(ret && errno == EEXIST, "GIC redist base set again");
 
-	/* Attempt to mix legacy and new redistributor regions */
-	addr = REDIST_REGION_ATTR_ADDR(NR_VCPUS, 0x100000, 0, 0);
-	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-				 KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);
-	TEST_ASSERT(ret && errno == EINVAL, "attempt to mix GICv3 REDIST and REDIST_REGION");
+	if (VGIC_DEV_IS_V3(v->gic_dev_type)) {

Instead you could check
    ret = kvm_device_check_attr(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR, KVM_VGIC_V3_ADDR_TYPE_REDIST);

Quoted 59 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +15 -7
+		/* Attempt to mix legacy and new redistributor regions */
+		addr = REDIST_REGION_ATTR_ADDR(NR_VCPUS, 0x100000, 0, 0);
+		ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+					 KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION,
+					 &addr, true);
+		TEST_ASSERT(ret && errno == EINVAL,
+			    "attempt to mix GICv3 REDIST and REDIST_REGION");
+	}
 
 	/*
 	 * Set overlapping DIST / REDIST, cannot be detected here. Will be detected
 	 * on first vcpu run instead.
 	 */
-	addr = 3 * 2 * 0x10000;
-	kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR, KVM_VGIC_V3_ADDR_TYPE_DIST,
-			  &addr, true);
+	addr = rdist.size - rdist.alignment;
+	kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  dist.attr, &addr, true);
 }
 
 /* Test the new REDIST region API */
@@ -256,14 +298,14 @@ static void subtest_v3_redist_regions(struct vm_gic *v)
  * VGIC KVM device is created and initialized before the secondary CPUs
  * get created
  */
-static void test_v3_vgic_then_vcpus(uint32_t gic_dev_type)
+static void test_vgic_then_vcpus(uint32_t gic_dev_type)
 {
 	struct vm_gic v;
 	int ret, i;
 
 	v = vm_gic_create_with_vcpus(gic_dev_type, 1);
 
-	subtest_v3_dist_rdist(&v);
+	subtest_dist_rdist(&v);
 
 	/* Add the rest of the VCPUs */
 	for (i = 1; i < NR_VCPUS; ++i)
@@ -276,14 +318,14 @@ static void test_v3_vgic_then_vcpus(uint32_t gic_dev_type)
 }
 
 /* All the VCPUs are created before the VGIC KVM device gets initialized */
-static void test_v3_vcpus_then_vgic(uint32_t gic_dev_type)
+static void test_vcpus_then_vgic(uint32_t gic_dev_type)
 {
 	struct vm_gic v;
 	int ret;
 
 	v = vm_gic_create_with_vcpus(gic_dev_type, NR_VCPUS);
 
-	subtest_v3_dist_rdist(&v);
+	subtest_dist_rdist(&v);
 
 	ret = run_vcpu(v.vm, 3);
 	TEST_ASSERT(ret == -EINVAL, "dist/rdist overlap detected on 1st vcpu run");
@@ -552,9 +594,10 @@ int test_kvm_device(uint32_t gic_dev_type)
 
 void run_tests(uint32_t gic_dev_type)

tools/testing/selftests/kvm/aarch64/vgic_init.c int test_kvm_device(uint32_t gic_dev_type) · line 603

Quoted 10 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +3 -2
 {
+	test_vcpus_then_vgic(gic_dev_type);
+	test_vgic_then_vcpus(gic_dev_type);
+
 	if (VGIC_DEV_IS_V3(gic_dev_type)) {
-		test_v3_vcpus_then_vgic(gic_dev_type);
-		test_v3_vgic_then_vcpus(gic_dev_type);
 		test_v3_new_redist_regions();
 		test_v3_typer_accesses();
 		test_v3_last_bit_redist_regions();

Thanks

Eric

#27 lore raw

Re: [PATCH v3 07/10] KVM: arm64: selftests: Add some tests for GICv2 in vgic_init

Quoted 52 lines · Eric Auger

Hi Ricardo,

Quoted 49 lines · Ricardo Koller

Add some GICv2 tests: general KVM device tests and DIST/REDIST overlap tests. Do this by making test_vcpus_then_vgic and test_vgic_then_vcpus in vgic_init GIC version agnostic.

---
 .../testing/selftests/kvm/aarch64/vgic_init.c | 107 ++++++++++++------
 1 file changed, 75 insertions(+), 32 deletions(-)
tools/testing/selftests/kvm/aarch64/vgic_init.c +30 -1
index b24067dbdac0..92f5c6ca6b8b 100644
@@ -79,74 +79,116 @@ static void vm_gic_destroy(struct vm_gic *v)
 	kvm_vm_free(v->vm);
 }
 
+struct vgic_region_attr {
+	uint64_t attr;
+	uint64_t size;
+	uint64_t alignment;
+};
+
+struct vgic_region_attr gic_v3_dist_region = {
+	.attr = KVM_VGIC_V3_ADDR_TYPE_DIST,
+	.size = 0x10000,
+	.alignment = 0x10000,
+};
+
+struct vgic_region_attr gic_v3_redist_region = {
+	.attr = KVM_VGIC_V3_ADDR_TYPE_REDIST,
+	.size = NR_VCPUS * 0x20000,
+	.alignment = 0x10000,
+};
+
+struct vgic_region_attr gic_v2_dist_region = {
+	.attr = KVM_VGIC_V2_ADDR_TYPE_DIST,
+	.size = 0x1000,
+	.alignment = 0x1000,
+};
+
+struct vgic_region_attr gic_v2_cpu_region = {
+	.attr = KVM_VGIC_V2_ADDR_TYPE_CPU,
+	.size = 0x2000,
+	.alignment = 0x1000,
+};
+
 /**
- * Helper routine that performs KVM device tests in general and
Quoted 8 lines

tools/testing/selftests/kvm/aarch64/vgic_init.c static void vm_gic_destroy(struct vm_gic *v) · line 116

Quoted 7 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +4 -3
- * especially ARM_VGIC_V3 ones. Eventually the ARM_VGIC_V3
- * device gets created, a legacy RDIST region is set at @0x0
- * and a DIST region is set @0x60000
+ * Helper routine that performs KVM device tests in general. Eventually the
+ * ARM_VGIC (GICv2 or GICv3) device gets created with an overlapping
+ * DIST/REDIST. A RDIST region (legacy in the case of GICv3) is set at @0x0 and
+ * a DIST region is set @0x70000 for GICv3 and @0x1000 for GICv2.

I would add "Assumption is 4 vcpus are going to be used hence the overlap".

ACK

Quoted 1 line

Also the RDIST is GICv3 only. In the above comment also mention the CPU I/F.

Will do. I wasn't sure if it was better to refer to both of them as CPU interfaces, or redistributors (half correct in both cases).

Quoted 19 lines
Quoted 19 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +9 -2
  */
-static void subtest_v3_dist_rdist(struct vm_gic *v)
+static void subtest_dist_rdist(struct vm_gic *v)
 {
 	int ret;
 	uint64_t addr;
+	struct vgic_region_attr rdist; /* CPU interface in GICv2*/
+	struct vgic_region_attr dist;
+
+	rdist = VGIC_DEV_IS_V3(v->gic_dev_type) ? gic_v3_redist_region
+						: gic_v2_cpu_region;
+	dist = VGIC_DEV_IS_V3(v->gic_dev_type) ? gic_v3_dist_region
+						: gic_v2_dist_region;
 
 	/* Check existing group/attributes */
 	kvm_device_check_attr(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-			      KVM_VGIC_V3_ADDR_TYPE_DIST);
+			      dist.attr);
 
Quoted 8 lines

tools/testing/selftests/kvm/aarch64/vgic_init.c static void vm_gic_destroy(struct vm_gic *v) · line 138

Quoted 7 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +2 -2
 	kvm_device_check_attr(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-			      KVM_VGIC_V3_ADDR_TYPE_REDIST);
+			      rdist.attr);
 
 	/* check non existing attribute */
-	ret = _kvm_device_check_attr(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR, 0);
+	ret = _kvm_device_check_attr(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR, -1);

was that necessary?

Yeah, turns out that 0 is used in v2:

#define KVM_VGIC_V2_ADDR_TYPE_DIST 0

Quoted 46 lines
Quoted 46 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +11 -10
 	TEST_ASSERT(ret && errno == ENXIO, "attribute not supported");
 
 	/* misaligned DIST and REDIST address settings */
-	addr = 0x1000;
+	addr = dist.alignment / 0x10;
 	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-				 KVM_VGIC_V3_ADDR_TYPE_DIST, &addr, true);
-	TEST_ASSERT(ret && errno == EINVAL, "GICv3 dist base not 64kB aligned");
+				 dist.attr, &addr, true);
+	TEST_ASSERT(ret && errno == EINVAL, "GIC dist base not aligned");
 
+	addr = rdist.alignment / 0x10;
 	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-				 KVM_VGIC_V3_ADDR_TYPE_REDIST, &addr, true);
-	TEST_ASSERT(ret && errno == EINVAL, "GICv3 redist base not 64kB aligned");
+				 rdist.attr, &addr, true);
+	TEST_ASSERT(ret && errno == EINVAL, "GIC redist/cpu base not aligned");
 
 	/* out of range address */
 	if (max_ipa_bits) {
 		addr = 1ULL << max_ipa_bits;
 		ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-					 KVM_VGIC_V3_ADDR_TYPE_DIST, &addr, true);
+					 dist.attr, &addr, true);
 		TEST_ASSERT(ret && errno == E2BIG, "dist address beyond IPA limit");
 
 		ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-					 KVM_VGIC_V3_ADDR_TYPE_REDIST, &addr, true);
+					 rdist.attr, &addr, true);
 		TEST_ASSERT(ret && errno == E2BIG, "redist address beyond IPA limit");
 	}
 
 	/* set REDIST base address @0x0*/
 	addr = 0x00000;
 	kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-			  KVM_VGIC_V3_ADDR_TYPE_REDIST, &addr, true);
+			  rdist.attr, &addr, true);
 
 	/* Attempt to create a second legacy redistributor region */
 	addr = 0xE0000;
 	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-				 KVM_VGIC_V3_ADDR_TYPE_REDIST, &addr, true);
-	TEST_ASSERT(ret && errno == EEXIST, "GICv3 redist base set again");
+				 rdist.attr, &addr, true);
+	TEST_ASSERT(ret && errno == EEXIST, "GIC redist base set again");
 
Quoted 9 lines

tools/testing/selftests/kvm/aarch64/vgic_init.c static void vm_gic_destroy(struct vm_gic *v) · line 175

Quoted 6 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +1 -5
-	/* Attempt to mix legacy and new redistributor regions */
-	addr = REDIST_REGION_ATTR_ADDR(NR_VCPUS, 0x100000, 0, 0);
-	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-				 KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);
-	TEST_ASSERT(ret && errno == EINVAL, "attempt to mix GICv3 REDIST and REDIST_REGION");
+	if (VGIC_DEV_IS_V3(v->gic_dev_type)) {

Instead you could check
    ret = kvm_device_check_attr(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR, KVM_VGIC_V3_ADDR_TYPE_REDIST);

ACK, will do that instead.

Quoted 71 lines
Quoted 59 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +15 -7
+		/* Attempt to mix legacy and new redistributor regions */
+		addr = REDIST_REGION_ATTR_ADDR(NR_VCPUS, 0x100000, 0, 0);
+		ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+					 KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION,
+					 &addr, true);
+		TEST_ASSERT(ret && errno == EINVAL,
+			    "attempt to mix GICv3 REDIST and REDIST_REGION");
+	}
 
 	/*
 	 * Set overlapping DIST / REDIST, cannot be detected here. Will be detected
 	 * on first vcpu run instead.
 	 */
-	addr = 3 * 2 * 0x10000;
-	kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR, KVM_VGIC_V3_ADDR_TYPE_DIST,
-			  &addr, true);
+	addr = rdist.size - rdist.alignment;
+	kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  dist.attr, &addr, true);
 }
 
 /* Test the new REDIST region API */
@@ -256,14 +298,14 @@ static void subtest_v3_redist_regions(struct vm_gic *v)
  * VGIC KVM device is created and initialized before the secondary CPUs
  * get created
  */
-static void test_v3_vgic_then_vcpus(uint32_t gic_dev_type)
+static void test_vgic_then_vcpus(uint32_t gic_dev_type)
 {
 	struct vm_gic v;
 	int ret, i;
 
 	v = vm_gic_create_with_vcpus(gic_dev_type, 1);
 
-	subtest_v3_dist_rdist(&v);
+	subtest_dist_rdist(&v);
 
 	/* Add the rest of the VCPUs */
 	for (i = 1; i < NR_VCPUS; ++i)
@@ -276,14 +318,14 @@ static void test_v3_vgic_then_vcpus(uint32_t gic_dev_type)
 }
 
 /* All the VCPUs are created before the VGIC KVM device gets initialized */
-static void test_v3_vcpus_then_vgic(uint32_t gic_dev_type)
+static void test_vcpus_then_vgic(uint32_t gic_dev_type)
 {
 	struct vm_gic v;
 	int ret;
 
 	v = vm_gic_create_with_vcpus(gic_dev_type, NR_VCPUS);
 
-	subtest_v3_dist_rdist(&v);
+	subtest_dist_rdist(&v);
 
 	ret = run_vcpu(v.vm, 3);
 	TEST_ASSERT(ret == -EINVAL, "dist/rdist overlap detected on 1st vcpu run");
@@ -552,9 +594,10 @@ int test_kvm_device(uint32_t gic_dev_type)
 
 void run_tests(uint32_t gic_dev_type)

tools/testing/selftests/kvm/aarch64/vgic_init.c int test_kvm_device(uint32_t gic_dev_type) · line 603

Quoted 10 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +3 -2
 {
+	test_vcpus_then_vgic(gic_dev_type);
+	test_vgic_then_vcpus(gic_dev_type);
+
 	if (VGIC_DEV_IS_V3(gic_dev_type)) {
-		test_v3_vcpus_then_vgic(gic_dev_type);
-		test_v3_vgic_then_vcpus(gic_dev_type);
 		test_v3_new_redist_regions();
 		test_v3_typer_accesses();
 		test_v3_last_bit_redist_regions();

Thanks

Quoted 2 lines

Eric

Thanks,
Ricardo

#28 lore raw

PATCH v3 8/10 KVM: arm64: selftests: Add tests for GIC redist/cpuif partially above IPA range

In reply to Ricardo Koller

Add tests for checking that KVM returns the right error when trying to set GICv2 CPU interfaces or GICv3 Redistributors partially above the addressable IPA range. Also tighten the IPA range by replacing KVM_CAP_ARM_VM_IPA_SIZE with the IPA range currently configured for the guest (i.e., the default).

The check for the GICv3 redistributor created using the REDIST legacy API is not sufficient as this new test only checks the check done using vcpus already created when setting the base. The next commit will add the missing test which verifies that the KVM check is done at first vcpu run.

---
 .../testing/selftests/kvm/aarch64/vgic_init.c | 46 ++++++++++++++-----
 1 file changed, 35 insertions(+), 11 deletions(-)
tools/testing/selftests/kvm/aarch64/vgic_init.c +35 -11
index 92f5c6ca6b8b..77a1941e61fa 100644
@@ -31,7 +31,7 @@ struct vm_gic {
 	uint32_t gic_dev_type;
 };
 
-static int max_ipa_bits;
+static uint64_t max_phys_size;
 
 /* helper to access a redistributor register */
 static int access_v3_redist_reg(int gicv3_fd, int vcpu, int offset,
@@ -150,16 +150,21 @@ static void subtest_dist_rdist(struct vm_gic *v)
 	TEST_ASSERT(ret && errno == EINVAL, "GIC redist/cpu base not aligned");
 
 	/* out of range address */
-	if (max_ipa_bits) {
-		addr = 1ULL << max_ipa_bits;
-		ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-					 dist.attr, &addr, true);
-		TEST_ASSERT(ret && errno == E2BIG, "dist address beyond IPA limit");
+	addr = max_phys_size;
+	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+				 dist.attr, &addr, true);
+	TEST_ASSERT(ret && errno == E2BIG, "dist address beyond IPA limit");
 
-		ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-					 rdist.attr, &addr, true);
-		TEST_ASSERT(ret && errno == E2BIG, "redist address beyond IPA limit");
-	}
+	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+				 rdist.attr, &addr, true);
+	TEST_ASSERT(ret && errno == E2BIG, "redist address beyond IPA limit");
+
+	/* Space for half a rdist (a rdist is: 2 * rdist.alignment). */
+	addr = max_phys_size - dist.alignment;
+	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+				 rdist.attr, &addr, true);
+	TEST_ASSERT(ret && errno == E2BIG,
+			"half of the redist is beyond IPA limit");
 
 	/* set REDIST base address @0x0*/
 	addr = 0x00000;
@@ -248,7 +253,21 @@ static void subtest_v3_redist_regions(struct vm_gic *v)
 	kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
 			  KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);
 
-	addr = REDIST_REGION_ATTR_ADDR(1, 1ULL << max_ipa_bits, 0, 2);
+	addr = REDIST_REGION_ATTR_ADDR(1, max_phys_size, 0, 2);
+	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+				 KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);
+	TEST_ASSERT(ret && errno == E2BIG,
+		    "register redist region with base address beyond IPA range");
+
+	/* The last redist is above the pa range. */
+	addr = REDIST_REGION_ATTR_ADDR(1, max_phys_size - 0x10000, 0, 2);
+	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+				 KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);
+	TEST_ASSERT(ret && errno == E2BIG,
+		    "register redist region with base address beyond IPA range");
+
+	/* The last redist is above the pa range. */
+	addr = REDIST_REGION_ATTR_ADDR(2, max_phys_size - 0x30000, 0, 2);
 	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
 				 KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);
 	TEST_ASSERT(ret && errno == E2BIG,
@@ -608,8 +627,13 @@ void run_tests(uint32_t gic_dev_type)
 int main(int ac, char **av)
 {
 	int ret;
+	int max_ipa_bits, pa_bits;
 
 	max_ipa_bits = kvm_check_cap(KVM_CAP_ARM_VM_IPA_SIZE);
+	pa_bits = vm_guest_mode_params[VM_MODE_DEFAULT].pa_bits;
+	TEST_ASSERT(max_ipa_bits && pa_bits <= max_ipa_bits,
+		"The default PA range should not be larger than the max.");
+	max_phys_size = 1ULL << pa_bits;
 
 	ret = test_kvm_device(KVM_DEV_TYPE_ARM_VGIC_V3);
 	if (!ret) {
Signature · 1 line
2.33.0.685.g46640cef36-goog
#29 lore raw

Re: [PATCH v3 08/10] KVM: arm64: selftests: Add tests for GIC redist/cpuif partially above IPA range

Hi Ricardo,

Quoted 68 lines · Ricardo Koller

Add tests for checking that KVM returns the right error when trying to set GICv2 CPU interfaces or GICv3 Redistributors partially above the addressable IPA range. Also tighten the IPA range by replacing KVM_CAP_ARM_VM_IPA_SIZE with the IPA range currently configured for the guest (i.e., the default).

The check for the GICv3 redistributor created using the REDIST legacy API is not sufficient as this new test only checks the check done using vcpus already created when setting the base. The next commit will add the missing test which verifies that the KVM check is done at first vcpu run.

---
 .../testing/selftests/kvm/aarch64/vgic_init.c | 46 ++++++++++++++-----
 1 file changed, 35 insertions(+), 11 deletions(-)
tools/testing/selftests/kvm/aarch64/vgic_init.c +17 -11
index 92f5c6ca6b8b..77a1941e61fa 100644
@@ -31,7 +31,7 @@ struct vm_gic {
 	uint32_t gic_dev_type;
 };
 
-static int max_ipa_bits;
+static uint64_t max_phys_size;
 
 /* helper to access a redistributor register */
 static int access_v3_redist_reg(int gicv3_fd, int vcpu, int offset,
@@ -150,16 +150,21 @@ static void subtest_dist_rdist(struct vm_gic *v)
 	TEST_ASSERT(ret && errno == EINVAL, "GIC redist/cpu base not aligned");
 
 	/* out of range address */
-	if (max_ipa_bits) {
-		addr = 1ULL << max_ipa_bits;
-		ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-					 dist.attr, &addr, true);
-		TEST_ASSERT(ret && errno == E2BIG, "dist address beyond IPA limit");
+	addr = max_phys_size;
+	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+				 dist.attr, &addr, true);
+	TEST_ASSERT(ret && errno == E2BIG, "dist address beyond IPA limit");
 
-		ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-					 rdist.attr, &addr, true);
-		TEST_ASSERT(ret && errno == E2BIG, "redist address beyond IPA limit");
-	}
+	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+				 rdist.attr, &addr, true);
+	TEST_ASSERT(ret && errno == E2BIG, "redist address beyond IPA limit");
+
+	/* Space for half a rdist (a rdist is: 2 * rdist.alignment). */
+	addr = max_phys_size - dist.alignment;
+	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+				 rdist.attr, &addr, true);
+	TEST_ASSERT(ret && errno == E2BIG,
+			"half of the redist is beyond IPA limit");
 
 	/* set REDIST base address @0x0*/
 	addr = 0x00000;
@@ -248,7 +253,21 @@ static void subtest_v3_redist_regions(struct vm_gic *v)
 	kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
 			  KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);
 
-	addr = REDIST_REGION_ATTR_ADDR(1, 1ULL << max_ipa_bits, 0, 2);
+	addr = REDIST_REGION_ATTR_ADDR(1, max_phys_size, 0, 2);
+	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,

tools/testing/selftests/kvm/aarch64/vgic_init.c static void subtest_v3_redist_regions(struct vm_gic *v) · line 267

Quoted 10 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +10 -0
+				 KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);
+	TEST_ASSERT(ret && errno == E2BIG,
+		    "register redist region with base address beyond IPA range");
+
+	/* The last redist is above the pa range. */
+	addr = REDIST_REGION_ATTR_ADDR(1, max_phys_size - 0x10000, 0, 2);
+	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+				 KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);
+	TEST_ASSERT(ret && errno == E2BIG,
+		    "register redist region with base address beyond IPA range");

s/base address/top address

tools/testing/selftests/kvm/aarch64/vgic_init.c static void subtest_v3_redist_regions(struct vm_gic *v) · line 272

Quoted 5 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +3 -0
+
+	/* The last redist is above the pa range. */
+	addr = REDIST_REGION_ATTR_ADDR(2, max_phys_size - 0x30000, 0, 2);
 	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
 				 KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);

Why this second check?

tools/testing/selftests/kvm/aarch64/vgic_init.c void run_tests(uint32_t gic_dev_type) · line 635

Quoted 11 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +4 -0
 	TEST_ASSERT(ret && errno == E2BIG,
@@ -608,8 +627,13 @@ void run_tests(uint32_t gic_dev_type)
 int main(int ac, char **av)
 {
 	int ret;
+	int max_ipa_bits, pa_bits;
 
 	max_ipa_bits = kvm_check_cap(KVM_CAP_ARM_VM_IPA_SIZE);
+	pa_bits = vm_guest_mode_params[VM_MODE_DEFAULT].pa_bits;
+	TEST_ASSERT(max_ipa_bits && pa_bits <= max_ipa_bits,
+		"The default PA range should not be larger than the max.");

Isn't it already enforced in the test infra instead? I see in lib/kvm_util.c

#ifdef __aarch64__
        if (vm->pa_bits != 40)
                vm->type = KVM_VM_TYPE_ARM_IPA_SIZE(vm->pa_bits); #endif

vm_open()

tools/testing/selftests/kvm/aarch64/vgic_init.c void run_tests(uint32_t gic_dev_type) · line 639

Quoted 4 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +1 -0
+	max_phys_size = 1ULL << pa_bits;
 
 	ret = test_kvm_device(KVM_DEV_TYPE_ARM_VGIC_V3);
 	if (!ret) {

Eric

#30 lore raw

Re: [PATCH v3 08/10] KVM: arm64: selftests: Add tests for GIC redist/cpuif partially above IPA range

Quoted 68 lines · Eric Auger

Hi Ricardo,

Quoted 65 lines · Ricardo Koller

Add tests for checking that KVM returns the right error when trying to set GICv2 CPU interfaces or GICv3 Redistributors partially above the addressable IPA range. Also tighten the IPA range by replacing KVM_CAP_ARM_VM_IPA_SIZE with the IPA range currently configured for the guest (i.e., the default).

The check for the GICv3 redistributor created using the REDIST legacy API is not sufficient as this new test only checks the check done using vcpus already created when setting the base. The next commit will add the missing test which verifies that the KVM check is done at first vcpu run.

---
 .../testing/selftests/kvm/aarch64/vgic_init.c | 46 ++++++++++++++-----
 1 file changed, 35 insertions(+), 11 deletions(-)
tools/testing/selftests/kvm/aarch64/vgic_init.c +15 -10
index 92f5c6ca6b8b..77a1941e61fa 100644
@@ -31,7 +31,7 @@ struct vm_gic {
 	uint32_t gic_dev_type;
 };
 
-static int max_ipa_bits;
+static uint64_t max_phys_size;
 
 /* helper to access a redistributor register */
 static int access_v3_redist_reg(int gicv3_fd, int vcpu, int offset,
@@ -150,16 +150,21 @@ static void subtest_dist_rdist(struct vm_gic *v)
 	TEST_ASSERT(ret && errno == EINVAL, "GIC redist/cpu base not aligned");
 
 	/* out of range address */
-	if (max_ipa_bits) {
-		addr = 1ULL << max_ipa_bits;
-		ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-					 dist.attr, &addr, true);
-		TEST_ASSERT(ret && errno == E2BIG, "dist address beyond IPA limit");
+	addr = max_phys_size;
+	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+				 dist.attr, &addr, true);
+	TEST_ASSERT(ret && errno == E2BIG, "dist address beyond IPA limit");
 
-		ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
-					 rdist.attr, &addr, true);
-		TEST_ASSERT(ret && errno == E2BIG, "redist address beyond IPA limit");
-	}
+	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+				 rdist.attr, &addr, true);
+	TEST_ASSERT(ret && errno == E2BIG, "redist address beyond IPA limit");
+
+	/* Space for half a rdist (a rdist is: 2 * rdist.alignment). */
+	addr = max_phys_size - dist.alignment;
+	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+				 rdist.attr, &addr, true);
+	TEST_ASSERT(ret && errno == E2BIG,
+			"half of the redist is beyond IPA limit");
 
 	/* set REDIST base address @0x0*/
 	addr = 0x00000;
@@ -248,7 +253,21 @@ static void subtest_v3_redist_regions(struct vm_gic *v)
 	kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
 			  KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);
 
Quoted 14 lines

tools/testing/selftests/kvm/aarch64/vgic_init.c static void subtest_v3_redist_regions(struct vm_gic *v) · line 267

Quoted 13 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +12 -1
-	addr = REDIST_REGION_ATTR_ADDR(1, 1ULL << max_ipa_bits, 0, 2);
+	addr = REDIST_REGION_ATTR_ADDR(1, max_phys_size, 0, 2);
+	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+				 KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);
+	TEST_ASSERT(ret && errno == E2BIG,
+		    "register redist region with base address beyond IPA range");
+
+	/* The last redist is above the pa range. */
+	addr = REDIST_REGION_ATTR_ADDR(1, max_phys_size - 0x10000, 0, 2);
+	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+				 KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);
+	TEST_ASSERT(ret && errno == E2BIG,
+		    "register redist region with base address beyond IPA range");

s/base address/top address

ACK

Quoted 6 lines

tools/testing/selftests/kvm/aarch64/vgic_init.c static void subtest_v3_redist_regions(struct vm_gic *v) · line 272

Quoted 5 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +3 -0
+
+	/* The last redist is above the pa range. */
+	addr = REDIST_REGION_ATTR_ADDR(2, max_phys_size - 0x30000, 0, 2);
 	ret = _kvm_device_access(v->gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
 				 KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION, &addr, true);

Why this second check?

Wanted to test having two redists (just a bonus test). This one is sligthtly more interesting than the previous one, so will just remove the previous one.

Quoted 14 lines

tools/testing/selftests/kvm/aarch64/vgic_init.c void run_tests(uint32_t gic_dev_type) · line 635

Quoted 11 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +4 -0
 	TEST_ASSERT(ret && errno == E2BIG,
@@ -608,8 +627,13 @@ void run_tests(uint32_t gic_dev_type)
 int main(int ac, char **av)
 {
 	int ret;
+	int max_ipa_bits, pa_bits;
 
 	max_ipa_bits = kvm_check_cap(KVM_CAP_ARM_VM_IPA_SIZE);
+	pa_bits = vm_guest_mode_params[VM_MODE_DEFAULT].pa_bits;
+	TEST_ASSERT(max_ipa_bits && pa_bits <= max_ipa_bits,
+		"The default PA range should not be larger than the max.");

Isn't it already enforced in the test infra instead? I see in lib/kvm_util.c

Quoted 4 lines

#ifdef __aarch64__
        if (vm->pa_bits != 40)
                vm->type = KVM_VM_TYPE_ARM_IPA_SIZE(vm->pa_bits); #endif

You are right, and this is a weird place to make that check.

Quoted 8 lines

vm_open()

tools/testing/selftests/kvm/aarch64/vgic_init.c void run_tests(uint32_t gic_dev_type) · line 639

Quoted 4 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +1 -0
+	max_phys_size = 1ULL << pa_bits;
 
 	ret = test_kvm_device(KVM_DEV_TYPE_ARM_VGIC_V3);
 	if (!ret) {

Eric

Thanks,
Ricardo

#31 lore raw

PATCH v3 9/10 KVM: arm64: selftests: Add test for legacy GICv3 REDIST base partially above IPA range

In reply to Ricardo Koller

Add a new test into vgic_init which checks that the first vcpu fails to run if there is not sufficient REDIST space below the addressable IPA range. This only applies to the KVM_VGIC_V3_ADDR_TYPE_REDIST legacy API as the required REDIST space is not know when setting the DIST region.

Note that using the REDIST_REGION API results in a different check at first vcpu run: that the number of redist regions is enough for all vcpus. And there is already a test for that case in, the first step of test_v3_new_redist_regions.

---
 .../testing/selftests/kvm/aarch64/vgic_init.c | 34 +++++++++++++++++++
 1 file changed, 34 insertions(+)
tools/testing/selftests/kvm/aarch64/vgic_init.c +34 -0
index 77a1941e61fa..417a9a515cad 100644
@@ -570,6 +570,39 @@ static void test_v3_last_bit_single_rdist(void)
 	vm_gic_destroy(&v);
 }
 
+/* Uses the legacy REDIST region API. */
+static void test_v3_redist_ipa_range_check_at_vcpu_run(void)
+{
+	struct vm_gic v;
+	int ret, i;
+	uint64_t addr;
+
+	v = vm_gic_create_with_vcpus(KVM_DEV_TYPE_ARM_VGIC_V3, 1);
+
+	/* Set space for 3 redists, we have 1 vcpu, so this succeeds. */
+	addr = max_phys_size - (3 * 2 * 0x10000);
+	kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+				 KVM_VGIC_V3_ADDR_TYPE_REDIST, &addr, true);
+
+	addr = 0x00000;
+	kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			KVM_VGIC_V3_ADDR_TYPE_DIST, &addr, true);
+
+	/* Add the rest of the VCPUs */
+	for (i = 1; i < NR_VCPUS; ++i)
+		vm_vcpu_add_default(v.vm, i, guest_code);
+
+	kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_CTRL,
+			  KVM_DEV_ARM_VGIC_CTRL_INIT, NULL, true);
+
+	/* Attempt to run a vcpu without enough redist space. */
+	ret = run_vcpu(v.vm, 2);
+	TEST_ASSERT(ret && errno == EINVAL,
+		"redist base+size above PA range detected on 1st vcpu run");
+
+	vm_gic_destroy(&v);
+}
+
 /*
  * Returns 0 if it's possible to create GIC device of a given type (V2 or V3).
  */
@@ -621,6 +654,7 @@ void run_tests(uint32_t gic_dev_type)
 		test_v3_typer_accesses();
 		test_v3_last_bit_redist_regions();
 		test_v3_last_bit_single_rdist();
+		test_v3_redist_ipa_range_check_at_vcpu_run();
 	}
 }
 
Signature · 1 line
2.33.0.685.g46640cef36-goog
Reviewed-by on v3 9/10
#32 lore raw

Re: [PATCH v3 09/10] KVM: arm64: selftests: Add test for legacy GICv3 REDIST base partially above IPA range

Quoted 60 lines · Ricardo Koller

Add a new test into vgic_init which checks that the first vcpu fails to run if there is not sufficient REDIST space below the addressable IPA range. This only applies to the KVM_VGIC_V3_ADDR_TYPE_REDIST legacy API as the required REDIST space is not know when setting the DIST region.

Note that using the REDIST_REGION API results in a different check at first vcpu run: that the number of redist regions is enough for all vcpus. And there is already a test for that case in, the first step of test_v3_new_redist_regions.

---
 .../testing/selftests/kvm/aarch64/vgic_init.c | 34 +++++++++++++++++++
 1 file changed, 34 insertions(+)
tools/testing/selftests/kvm/aarch64/vgic_init.c +33 -0
index 77a1941e61fa..417a9a515cad 100644
@@ -570,6 +570,39 @@ static void test_v3_last_bit_single_rdist(void)
 	vm_gic_destroy(&v);
 }
 
+/* Uses the legacy REDIST region API. */
+static void test_v3_redist_ipa_range_check_at_vcpu_run(void)
+{
+	struct vm_gic v;
+	int ret, i;
+	uint64_t addr;
+
+	v = vm_gic_create_with_vcpus(KVM_DEV_TYPE_ARM_VGIC_V3, 1);
+
+	/* Set space for 3 redists, we have 1 vcpu, so this succeeds. */
+	addr = max_phys_size - (3 * 2 * 0x10000);
+	kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+				 KVM_VGIC_V3_ADDR_TYPE_REDIST, &addr, true);
+
+	addr = 0x00000;
+	kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			KVM_VGIC_V3_ADDR_TYPE_DIST, &addr, true);
+
+	/* Add the rest of the VCPUs */
+	for (i = 1; i < NR_VCPUS; ++i)
+		vm_vcpu_add_default(v.vm, i, guest_code);
+
+	kvm_device_access(v.gic_fd, KVM_DEV_ARM_VGIC_GRP_CTRL,
+			  KVM_DEV_ARM_VGIC_CTRL_INIT, NULL, true);
+
+	/* Attempt to run a vcpu without enough redist space. */
+	ret = run_vcpu(v.vm, 2);
+	TEST_ASSERT(ret && errno == EINVAL,
+		"redist base+size above PA range detected on 1st vcpu run");
+
+	vm_gic_destroy(&v);
+}
+
 /*
  * Returns 0 if it's possible to create GIC device of a given type (V2 or V3).
  */

tools/testing/selftests/kvm/aarch64/vgic_init.c void run_tests(uint32_t gic_dev_type) · line 660

Quoted 7 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +1 -0
@@ -621,6 +654,7 @@ void run_tests(uint32_t gic_dev_type)
 		test_v3_typer_accesses();
 		test_v3_last_bit_redist_regions();
 		test_v3_last_bit_single_rdist();
+		test_v3_redist_ipa_range_check_at_vcpu_run();
 	}
 }
 

Eric

_______________________________________________
kvmarm mailing list
[email protected]
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm

#33 lore raw

PATCH v3 10/10 KVM: arm64: selftests: Add basic ITS device tests

In reply to Ricardo Koller

Add some ITS device tests: general KVM device tests (address not defined already, address aligned) and tests for the ITS region being within the addressable IPA range.

---
 .../testing/selftests/kvm/aarch64/vgic_init.c | 42 +++++++++++++++++++
 1 file changed, 42 insertions(+)
tools/testing/selftests/kvm/aarch64/vgic_init.c +42 -0
index 417a9a515cad..180221ec325d 100644
@@ -603,6 +603,47 @@ static void test_v3_redist_ipa_range_check_at_vcpu_run(void)
 	vm_gic_destroy(&v);
 }
 
+static void test_v3_its_region(void)
+{
+	struct vm_gic v;
+	uint64_t addr;
+	int its_fd, ret;
+
+	v = vm_gic_create_with_vcpus(KVM_DEV_TYPE_ARM_VGIC_V3, NR_VCPUS);
+	its_fd = kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_ITS, false);
+
+	addr = 0x401000;
+	ret = _kvm_device_access(its_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  KVM_VGIC_ITS_ADDR_TYPE, &addr, true);
+	TEST_ASSERT(ret && errno == EINVAL,
+		"ITS region with misaligned address");
+
+	addr = max_phys_size;
+	ret = _kvm_device_access(its_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  KVM_VGIC_ITS_ADDR_TYPE, &addr, true);
+	TEST_ASSERT(ret && errno == E2BIG,
+		"register ITS region with base address beyond IPA range");
+
+	addr = max_phys_size - 0x10000;
+	ret = _kvm_device_access(its_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  KVM_VGIC_ITS_ADDR_TYPE, &addr, true);
+	TEST_ASSERT(ret && errno == E2BIG,
+		"Half of ITS region is beyond IPA range");
+
+	/* This one succeeds setting the ITS base */
+	addr = 0x400000;
+	kvm_device_access(its_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  KVM_VGIC_ITS_ADDR_TYPE, &addr, true);
+
+	addr = 0x300000;
+	ret = _kvm_device_access(its_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  KVM_VGIC_ITS_ADDR_TYPE, &addr, true);
+	TEST_ASSERT(ret && errno == EEXIST, "ITS base set again");
+
+	close(its_fd);
+	vm_gic_destroy(&v);
+}
+
 /*
  * Returns 0 if it's possible to create GIC device of a given type (V2 or V3).
  */
@@ -655,6 +696,7 @@ void run_tests(uint32_t gic_dev_type)
 		test_v3_last_bit_redist_regions();
 		test_v3_last_bit_single_rdist();
 		test_v3_redist_ipa_range_check_at_vcpu_run();
+		test_v3_its_region();
 	}
 }
 
Signature · 1 line
2.33.0.685.g46640cef36-goog
#34 lore raw

Re: [PATCH v3 10/10] KVM: arm64: selftests: Add basic ITS device tests

Hi Ricardo,

Quoted 15 lines · Ricardo Koller

Add some ITS device tests: general KVM device tests (address not defined already, address aligned) and tests for the ITS region being within the addressable IPA range.

---
 .../testing/selftests/kvm/aarch64/vgic_init.c | 42 +++++++++++++++++++
 1 file changed, 42 insertions(+)
tools/testing/selftests/kvm/aarch64/vgic_init.c +0 -0
index 417a9a515cad..180221ec325d 100644
@@ -603,6 +603,47 @@ static void test_v3_redist_ipa_range_check_at_vcpu_run(void)
 	vm_gic_destroy(&v);

tools/testing/selftests/kvm/aarch64/vgic_init.c static void test_v3_redist_ipa_range_check_at_vcpu_run(void) · line 613

Quoted 10 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +8 -0
 }
 
+static void test_v3_its_region(void)
+{
+	struct vm_gic v;
+	uint64_t addr;
+	int its_fd, ret;
+
+	v = vm_gic_create_with_vcpus(KVM_DEV_TYPE_ARM_VGIC_V3, NR_VCPUS);
+	its_fd = kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_ITS, false);

this may fail if the ITS device has not been registered by KVM (host GICv2)

Maybe refine the patch title mentionning this is an ITS device "init" test. as per Documentation/virt/kvm/devices/arm-vgic-its.rst we could also try instantiating the ITS before the GIC and try instantiating several ITSs with overlapping addresses.
But I would totally understand if you consider this out of the scope of your  fixes + tests.

Thanks!

Eric

Quoted 44 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +34 -0
+
+	addr = 0x401000;
+	ret = _kvm_device_access(its_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  KVM_VGIC_ITS_ADDR_TYPE, &addr, true);
+	TEST_ASSERT(ret && errno == EINVAL,
+		"ITS region with misaligned address");
+
+	addr = max_phys_size;
+	ret = _kvm_device_access(its_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  KVM_VGIC_ITS_ADDR_TYPE, &addr, true);
+	TEST_ASSERT(ret && errno == E2BIG,
+		"register ITS region with base address beyond IPA range");
+
+	addr = max_phys_size - 0x10000;
+	ret = _kvm_device_access(its_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  KVM_VGIC_ITS_ADDR_TYPE, &addr, true);
+	TEST_ASSERT(ret && errno == E2BIG,
+		"Half of ITS region is beyond IPA range");
+
+	/* This one succeeds setting the ITS base */
+	addr = 0x400000;
+	kvm_device_access(its_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  KVM_VGIC_ITS_ADDR_TYPE, &addr, true);
+
+	addr = 0x300000;
+	ret = _kvm_device_access(its_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  KVM_VGIC_ITS_ADDR_TYPE, &addr, true);
+	TEST_ASSERT(ret && errno == EEXIST, "ITS base set again");
+
+	close(its_fd);
+	vm_gic_destroy(&v);
+}
+
 /*
  * Returns 0 if it's possible to create GIC device of a given type (V2 or V3).
  */
@@ -655,6 +696,7 @@ void run_tests(uint32_t gic_dev_type)
 		test_v3_last_bit_redist_regions();
 		test_v3_last_bit_single_rdist();
 		test_v3_redist_ipa_range_check_at_vcpu_run();
+		test_v3_its_region();
 	}
 }
 
#35 lore raw

Re: [PATCH v3 10/10] KVM: arm64: selftests: Add basic ITS device tests

Hi Eric,

Quoted 20 lines · Eric Auger

Hi Ricardo,

Quoted 17 lines · Ricardo Koller

Add some ITS device tests: general KVM device tests (address not defined already, address aligned) and tests for the ITS region being within the addressable IPA range.

---
 .../testing/selftests/kvm/aarch64/vgic_init.c | 42 +++++++++++++++++++
 1 file changed, 42 insertions(+)
tools/testing/selftests/kvm/aarch64/vgic_init.c +0 -0
index 417a9a515cad..180221ec325d 100644
@@ -603,6 +603,47 @@ static void test_v3_redist_ipa_range_check_at_vcpu_run(void)
 	vm_gic_destroy(&v);
 }
 
Quoted 9 lines

tools/testing/selftests/kvm/aarch64/vgic_init.c static void test_v3_redist_ipa_range_check_at_vcpu_run(void) · line 613

Quoted 8 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +8 -0
+static void test_v3_its_region(void)
+{
+	struct vm_gic v;
+	uint64_t addr;
+	int its_fd, ret;
+
+	v = vm_gic_create_with_vcpus(KVM_DEV_TYPE_ARM_VGIC_V3, NR_VCPUS);
+	its_fd = kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_ITS, false);

this may fail if the ITS device has not been registered by KVM (host GICv2)

At the moment it's just called in the GICv3 case. It seems that registering a GICv3 device results in having an ITS registered as well (from kvm_register_vgic_device()). I'm assuming this won't change; we might as well check that assumption. What do you think?

Thanks,
Ricardo

Quoted 56 lines

Maybe refine the patch title mentionning this is an ITS device "init" test. as per Documentation/virt/kvm/devices/arm-vgic-its.rst we could also try instantiating the ITS before the GIC and try instantiating several ITSs with overlapping addresses.
But I would totally understand if you consider this out of the scope of your  fixes + tests.

Thanks!

Eric

Quoted 44 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +34 -0
+
+	addr = 0x401000;
+	ret = _kvm_device_access(its_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  KVM_VGIC_ITS_ADDR_TYPE, &addr, true);
+	TEST_ASSERT(ret && errno == EINVAL,
+		"ITS region with misaligned address");
+
+	addr = max_phys_size;
+	ret = _kvm_device_access(its_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  KVM_VGIC_ITS_ADDR_TYPE, &addr, true);
+	TEST_ASSERT(ret && errno == E2BIG,
+		"register ITS region with base address beyond IPA range");
+
+	addr = max_phys_size - 0x10000;
+	ret = _kvm_device_access(its_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  KVM_VGIC_ITS_ADDR_TYPE, &addr, true);
+	TEST_ASSERT(ret && errno == E2BIG,
+		"Half of ITS region is beyond IPA range");
+
+	/* This one succeeds setting the ITS base */
+	addr = 0x400000;
+	kvm_device_access(its_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  KVM_VGIC_ITS_ADDR_TYPE, &addr, true);
+
+	addr = 0x300000;
+	ret = _kvm_device_access(its_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  KVM_VGIC_ITS_ADDR_TYPE, &addr, true);
+	TEST_ASSERT(ret && errno == EEXIST, "ITS base set again");
+
+	close(its_fd);
+	vm_gic_destroy(&v);
+}
+
 /*
  * Returns 0 if it's possible to create GIC device of a given type (V2 or V3).
  */
@@ -655,6 +696,7 @@ void run_tests(uint32_t gic_dev_type)
 		test_v3_last_bit_redist_regions();
 		test_v3_last_bit_single_rdist();
 		test_v3_redist_ipa_range_check_at_vcpu_run();
+		test_v3_its_region();
 	}
 }
 
#36 lore raw

Re: [PATCH v3 10/10] KVM: arm64: selftests: Add basic ITS device tests

Hi Ricardo,

Quoted 32 lines · Ricardo Koller

Hi Eric,

Quoted 20 lines · Eric Auger

Hi Ricardo,

Quoted 17 lines · Ricardo Koller

Add some ITS device tests: general KVM device tests (address not defined already, address aligned) and tests for the ITS region being within the addressable IPA range.

---
 .../testing/selftests/kvm/aarch64/vgic_init.c | 42 +++++++++++++++++++
 1 file changed, 42 insertions(+)
tools/testing/selftests/kvm/aarch64/vgic_init.c +0 -0
index 417a9a515cad..180221ec325d 100644
@@ -603,6 +603,47 @@ static void test_v3_redist_ipa_range_check_at_vcpu_run(void)
 	vm_gic_destroy(&v);
 }
 
Quoted 9 lines

tools/testing/selftests/kvm/aarch64/vgic_init.c static void test_v3_redist_ipa_range_check_at_vcpu_run(void) · line 613

Quoted 8 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +8 -0
+static void test_v3_its_region(void)
+{
+	struct vm_gic v;
+	uint64_t addr;
+	int its_fd, ret;
+
+	v = vm_gic_create_with_vcpus(KVM_DEV_TYPE_ARM_VGIC_V3, NR_VCPUS);
+	its_fd = kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_ITS, false);

this may fail if the ITS device has not been registered by KVM (host GICv2)

Quoted 1 line

At the moment it's just called in the GICv3 case. It seems that

OK I missed that. in that case that's fine.

Thanks

Eric

Quoted 61 lines

registering a GICv3 device results in having an ITS registered as well (from kvm_register_vgic_device()). I'm assuming this won't change; we might as well check that assumption. What do you think?

Thanks,
Ricardo

Quoted 54 lines

Maybe refine the patch title mentionning this is an ITS device "init" test. as per Documentation/virt/kvm/devices/arm-vgic-its.rst we could also try instantiating the ITS before the GIC and try instantiating several ITSs with overlapping addresses.
But I would totally understand if you consider this out of the scope of your  fixes + tests.

Thanks!

Eric

Quoted 44 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +34 -0
+
+	addr = 0x401000;
+	ret = _kvm_device_access(its_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  KVM_VGIC_ITS_ADDR_TYPE, &addr, true);
+	TEST_ASSERT(ret && errno == EINVAL,
+		"ITS region with misaligned address");
+
+	addr = max_phys_size;
+	ret = _kvm_device_access(its_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  KVM_VGIC_ITS_ADDR_TYPE, &addr, true);
+	TEST_ASSERT(ret && errno == E2BIG,
+		"register ITS region with base address beyond IPA range");
+
+	addr = max_phys_size - 0x10000;
+	ret = _kvm_device_access(its_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  KVM_VGIC_ITS_ADDR_TYPE, &addr, true);
+	TEST_ASSERT(ret && errno == E2BIG,
+		"Half of ITS region is beyond IPA range");
+
+	/* This one succeeds setting the ITS base */
+	addr = 0x400000;
+	kvm_device_access(its_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  KVM_VGIC_ITS_ADDR_TYPE, &addr, true);
+
+	addr = 0x300000;
+	ret = _kvm_device_access(its_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  KVM_VGIC_ITS_ADDR_TYPE, &addr, true);
+	TEST_ASSERT(ret && errno == EEXIST, "ITS base set again");
+
+	close(its_fd);
+	vm_gic_destroy(&v);
+}
+
 /*
  * Returns 0 if it's possible to create GIC device of a given type (V2 or V3).
  */
@@ -655,6 +696,7 @@ void run_tests(uint32_t gic_dev_type)
 		test_v3_last_bit_redist_regions();
 		test_v3_last_bit_single_rdist();
 		test_v3_redist_ipa_range_check_at_vcpu_run();
+		test_v3_its_region();
 	}
 }
 
#37 lore raw

Re: [PATCH v3 10/10] KVM: arm64: selftests: Add basic ITS device tests

In reply to Eric Auger
Quoted 30 lines · Eric Auger

Hi Ricardo,

Quoted 15 lines · Ricardo Koller

Add some ITS device tests: general KVM device tests (address not defined already, address aligned) and tests for the ITS region being within the addressable IPA range.

---
 .../testing/selftests/kvm/aarch64/vgic_init.c | 42 +++++++++++++++++++
 1 file changed, 42 insertions(+)
tools/testing/selftests/kvm/aarch64/vgic_init.c +0 -0
index 417a9a515cad..180221ec325d 100644
@@ -603,6 +603,47 @@ static void test_v3_redist_ipa_range_check_at_vcpu_run(void)
 	vm_gic_destroy(&v);

tools/testing/selftests/kvm/aarch64/vgic_init.c static void test_v3_redist_ipa_range_check_at_vcpu_run(void) · line 613

Quoted 10 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +8 -0
 }
 
+static void test_v3_its_region(void)
+{
+	struct vm_gic v;
+	uint64_t addr;
+	int its_fd, ret;
+
+	v = vm_gic_create_with_vcpus(KVM_DEV_TYPE_ARM_VGIC_V3, NR_VCPUS);
+	its_fd = kvm_create_device(v.vm, KVM_DEV_TYPE_ARM_VGIC_ITS, false);

this may fail if the ITS device has not been registered by KVM (host GICv2)

Quoted 6 lines

Maybe refine the patch title mentionning this is an ITS device "init" test. as per Documentation/virt/kvm/devices/arm-vgic-its.rst we could also try instantiating the ITS before the GIC and try instantiating several ITSs with overlapping addresses.
But I would totally understand if you consider this out of the scope of your  fixes + tests.

Will just send a v4 with init tests for now. ACK on changing the patch title.

Thanks,
Ricardo

Quoted 49 lines

Thanks!

Eric

Quoted 44 lines
tools/testing/selftests/kvm/aarch64/vgic_init.c +34 -0
+
+	addr = 0x401000;
+	ret = _kvm_device_access(its_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  KVM_VGIC_ITS_ADDR_TYPE, &addr, true);
+	TEST_ASSERT(ret && errno == EINVAL,
+		"ITS region with misaligned address");
+
+	addr = max_phys_size;
+	ret = _kvm_device_access(its_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  KVM_VGIC_ITS_ADDR_TYPE, &addr, true);
+	TEST_ASSERT(ret && errno == E2BIG,
+		"register ITS region with base address beyond IPA range");
+
+	addr = max_phys_size - 0x10000;
+	ret = _kvm_device_access(its_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  KVM_VGIC_ITS_ADDR_TYPE, &addr, true);
+	TEST_ASSERT(ret && errno == E2BIG,
+		"Half of ITS region is beyond IPA range");
+
+	/* This one succeeds setting the ITS base */
+	addr = 0x400000;
+	kvm_device_access(its_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  KVM_VGIC_ITS_ADDR_TYPE, &addr, true);
+
+	addr = 0x300000;
+	ret = _kvm_device_access(its_fd, KVM_DEV_ARM_VGIC_GRP_ADDR,
+			  KVM_VGIC_ITS_ADDR_TYPE, &addr, true);
+	TEST_ASSERT(ret && errno == EEXIST, "ITS base set again");
+
+	close(its_fd);
+	vm_gic_destroy(&v);
+}
+
 /*
  * Returns 0 if it's possible to create GIC device of a given type (V2 or V3).
  */
@@ -655,6 +696,7 @@ void run_tests(uint32_t gic_dev_type)
 		test_v3_last_bit_redist_regions();
 		test_v3_last_bit_single_rdist();
 		test_v3_redist_ipa_range_check_at_vcpu_run();
+		test_v3_its_region();
 	}
 }