Skip to contentSkip to conversation

KVM: x86: fix for missing initialization of return status variable

PATCH Superseded Ameer Hamza 3 participants 11 messages Dec 5, 2021 – Dec 6, 2021

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 KVM: x86: fix for missing initialization of return status variable

If undefined ioctl number is passed to the kvm_vcpu_ioctl_device_attr function, it should return with error status.

Addresses-Coverity: 1494124 ("Uninitialized scalar variable")

---
 arch/x86/kvm/x86.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/kvm/x86.c +1 -1
index e0aa4dd53c7f..55b90c185717 100644
@@ -5001,7 +5001,7 @@ static int kvm_vcpu_ioctl_device_attr(struct kvm_vcpu *vcpu,
 				      void __user *argp)
 {
 	struct kvm_device_attr attr;
-	int r;
+	int r = -EINVAL;
 
 	if (copy_from_user(&attr, argp, sizeof(attr)))
 		return -EFAULT;
Signature · 1 line
2.25.1
#1 lore raw
Quoted 15 lines · Ameer Hamza

If undefined ioctl number is passed to the kvm_vcpu_ioctl_device_attr function, it should return with error status.

Addresses-Coverity: 1494124 ("Uninitialized scalar variable")

---
 arch/x86/kvm/x86.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/kvm/x86.c +0 -0
index e0aa4dd53c7f..55b90c185717 100644

arch/x86/kvm/x86.c static int kvm_vcpu_ioctl_device_attr(struct kvm_vcpu *vcpu, · line 5007

Quoted 8 lines
arch/x86/kvm/x86.c +1 -1
@@ -5001,7 +5001,7 @@ static int kvm_vcpu_ioctl_device_attr(struct kvm_vcpu *vcpu,
 				      void __user *argp)
 {
 	struct kvm_device_attr attr;
-	int r;
+	int r = -EINVAL;
 
 	if (copy_from_user(&attr, argp, sizeof(attr)))
 		return -EFAULT;

The reported issue is not real, kvm_vcpu_ioctl_device_attr() is never called with anything but [KVM_HAS_DEVICE_ATTR, KVM_GET_DEVICE_ATTR, KVM_SET_DEVICE_ATTR] as 'ioctl' and the switch below covers all three. Instead of initializing 'r' we could've added a 'default' case to the switch, either returning something like EINVAL or just BUG(). Hope it'll silence coverity.

Signature · 1 line
Vitaly
#2 lore raw
Quoted 23 lines · Vitaly Kuznetsov
Quoted 21 lines · Ameer Hamza

If undefined ioctl number is passed to the kvm_vcpu_ioctl_device_attr function, it should return with error status.

Addresses-Coverity: 1494124 ("Uninitialized scalar variable")

---
 arch/x86/kvm/x86.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/kvm/x86.c +1 -1
index e0aa4dd53c7f..55b90c185717 100644
@@ -5001,7 +5001,7 @@ static int kvm_vcpu_ioctl_device_attr(struct kvm_vcpu *vcpu,
 				      void __user *argp)
 {
 	struct kvm_device_attr attr;
-	int r;
+	int r = -EINVAL;
 
Quoted 9 lines

arch/x86/kvm/x86.c static int kvm_vcpu_ioctl_device_attr(struct kvm_vcpu *vcpu, · line 5007

Quoted 2 lines
arch/x86/kvm/x86.c +0 -0
 	if (copy_from_user(&attr, argp, sizeof(attr)))
 		return -EFAULT;

The reported issue is not real, kvm_vcpu_ioctl_device_attr() is never called with anything but [KVM_HAS_DEVICE_ATTR, KVM_GET_DEVICE_ATTR, KVM_SET_DEVICE_ATTR] as 'ioctl' and the switch below covers all three. Instead of initializing 'r' we could've added a 'default' case to the switch, either returning something like EINVAL or just BUG(). Hope it'll silence coverity.

Thank you for your kind response. I agree with you and I think its logical to add default case here. Let me update the patch.

Best Regards,
Hamza

#3 lore raw

PATCH v2 1/1 KVM: x86: fix for missing initialization of return status variable

In reply to Vitaly Kuznetsov

If undefined ioctl number is passed to the kvm_vcpu_ioctl_device_attr function, it should return with error status.

Addresses-Coverity: 1494124 ("Uninitialized scalar variable")

---
Added default case to return EINV for undefined ioctl number

---
 arch/x86/kvm/x86.c | 2 ++
 1 file changed, 2 insertions(+)
arch/x86/kvm/x86.c +2 -0
index e0aa4dd53c7f..e6e00f997b1f 100644
@@ -5019,6 +5019,8 @@ static int kvm_vcpu_ioctl_device_attr(struct kvm_vcpu *vcpu,
 	case KVM_SET_DEVICE_ATTR:
 		r = kvm_arch_tsc_set_attr(vcpu, &attr);
 		break;
+	default:
+		r = -EINVAL;
 	}
 
 	return r;
Signature · 1 line
2.25.1
#4 lore raw
Quoted 2 lines · Ameer Hamza

If undefined ioctl number is passed to the kvm_vcpu_ioctl_device_attr function, it should return with error status.

No, if anything KVM should do KVM_BUG_ON() and return -EIO, because @ioctl is completely KVM controlled. But I'd personally prefer we leave it as is, there's one call site that very clearly invokes the helper with only the three ioctls. It's not a strong preference though.

#5 lore raw
Quoted 8 lines · Sean Christopherson
Quoted 2 lines · Ameer Hamza

If undefined ioctl number is passed to the kvm_vcpu_ioctl_device_attr function, it should return with error status.

No, if anything KVM should do KVM_BUG_ON() and return -EIO, because @ioctl is completely KVM controlled. But I'd personally prefer we leave it as is, there's one call site that very clearly invokes the helper with only the three ioctls. It's not a strong preference though.

Thank you for your response. I agree with you, but I think in my opinion, it would be nice to resolve coverity warning. Let me update the patch according to your suggestions anyway.

Thanks,
Hamza.

#6 lore raw

PATCH v3 1/1 KVM: x86: fix for missing initialization of return status variable

If undefined ioctl number is passed to the kvm_vcpu_ioctl_device_attr ioctl, we should trigger KVM_BUG_ON() and return with EIO to silent coverity warning.

Addresses-Coverity: 1494124 ("Uninitialized scalar variable")

---
Changes in v3:
Added KVM_BUG_ON() as default case and returned -EIO

---
 arch/x86/kvm/x86.c | 3 +++
 1 file changed, 3 insertions(+)
arch/x86/kvm/x86.c +3 -0
index e0aa4dd53c7f..b37068f847ff 100644
@@ -5019,6 +5019,9 @@ static int kvm_vcpu_ioctl_device_attr(struct kvm_vcpu *vcpu,
 	case KVM_SET_DEVICE_ATTR:
 		r = kvm_arch_tsc_set_attr(vcpu, &attr);
 		break;
+	default:
+		KVM_BUG_ON(1, vcpu->kvm);
+		r = -EIO;
 	}
 
 	return r;
Signature · 1 line
2.25.1
#7 lore raw
Quoted 18 lines · Ameer Hamza

If undefined ioctl number is passed to the kvm_vcpu_ioctl_device_attr ioctl, we should trigger KVM_BUG_ON() and return with EIO to silent coverity warning.

Addresses-Coverity: 1494124 ("Uninitialized scalar variable")

---
Changes in v3:
Added KVM_BUG_ON() as default case and returned -EIO

---
 arch/x86/kvm/x86.c | 3 +++
 1 file changed, 3 insertions(+)
arch/x86/kvm/x86.c +0 -0
index e0aa4dd53c7f..b37068f847ff 100644

arch/x86/kvm/x86.c static int kvm_vcpu_ioctl_device_attr(struct kvm_vcpu *vcpu, · line 5024

Quoted 6 lines
arch/x86/kvm/x86.c +3 -0
@@ -5019,6 +5019,9 @@ static int kvm_vcpu_ioctl_device_attr(struct kvm_vcpu *vcpu,
 	case KVM_SET_DEVICE_ATTR:
 		r = kvm_arch_tsc_set_attr(vcpu, &attr);
 		break;
+	default:
+		KVM_BUG_ON(1, vcpu->kvm);
+		r = -EIO;

At least have a

		break;

if we're going to be pedantic about things.

Quoted 6 lines
arch/x86/kvm/x86.c +0 -0
 	}
 
 	return r;
Signature · 1 line
2.25.1
#8 lore raw
Quoted 28 lines · Sean Christopherson
Quoted 18 lines · Ameer Hamza

If undefined ioctl number is passed to the kvm_vcpu_ioctl_device_attr ioctl, we should trigger KVM_BUG_ON() and return with EIO to silent coverity warning.

Addresses-Coverity: 1494124 ("Uninitialized scalar variable")

---
Changes in v3:
Added KVM_BUG_ON() as default case and returned -EIO

---
 arch/x86/kvm/x86.c | 3 +++
 1 file changed, 3 insertions(+)
arch/x86/kvm/x86.c +0 -0
index e0aa4dd53c7f..b37068f847ff 100644

arch/x86/kvm/x86.c static int kvm_vcpu_ioctl_device_attr(struct kvm_vcpu *vcpu, · line 5024

Quoted 6 lines
arch/x86/kvm/x86.c +3 -0
@@ -5019,6 +5019,9 @@ static int kvm_vcpu_ioctl_device_attr(struct kvm_vcpu *vcpu,
 	case KVM_SET_DEVICE_ATTR:
 		r = kvm_arch_tsc_set_attr(vcpu, &attr);
 		break;
+	default:
+		KVM_BUG_ON(1, vcpu->kvm);
+		r = -EIO;

At least have a

Quoted 3 lines
		break;

if we're going to be pedantic about things.

I just started as a contributer in this community and trying to fix issues found by static analyzer tools. If you think that's not necessary, its totally fine :)

Quoted 6 lines
Quoted 6 lines
arch/x86/kvm/x86.c +0 -0
 	}
 
 	return r;
Signature · 1 line
2.25.1
#9 lore raw
Quoted 29 lines · Ameer Hamza
Quoted 28 lines · Sean Christopherson
Quoted 18 lines · Ameer Hamza

If undefined ioctl number is passed to the kvm_vcpu_ioctl_device_attr ioctl, we should trigger KVM_BUG_ON() and return with EIO to silent coverity warning.

Addresses-Coverity: 1494124 ("Uninitialized scalar variable")

---
Changes in v3:
Added KVM_BUG_ON() as default case and returned -EIO

---
 arch/x86/kvm/x86.c | 3 +++
 1 file changed, 3 insertions(+)
arch/x86/kvm/x86.c +0 -0
index e0aa4dd53c7f..b37068f847ff 100644

arch/x86/kvm/x86.c static int kvm_vcpu_ioctl_device_attr(struct kvm_vcpu *vcpu, · line 5024

Quoted 6 lines
arch/x86/kvm/x86.c +3 -0
@@ -5019,6 +5019,9 @@ static int kvm_vcpu_ioctl_device_attr(struct kvm_vcpu *vcpu,
 	case KVM_SET_DEVICE_ATTR:
 		r = kvm_arch_tsc_set_attr(vcpu, &attr);
 		break;
+	default:
+		KVM_BUG_ON(1, vcpu->kvm);
+		r = -EIO;

At least have a

Quoted 6 lines
Quoted 3 lines
		break;

if we're going to be pedantic about things.

I just started as a contributer in this community and trying to fix issues found by static analyzer tools. If you think that's not necessary, its totally fine :)

(Most) Static analyzers are great, they definitely find real bugs. But they also have a fair number of false positives, e.g. this is a firmly a false positive, so the results of any static analyzer needs to thought about critically, not blindly followed. It's completely understandable that Coverity got tripped up in this case, but that's exactly why having a human vet the bug report is necessary.

There is arguably value in having a default statement to ensure future KVM code doesn't end up adding a bad call, which is why I'm not completely opposed to the above addition.

Where folks, myself included, get a bit grumpy is when patches are sent to "fix" bug reports from static analyzers without evidence that the submitter has done their due dilegence to understand the code they are changing, e.g. even without any understanding of KVM, a search of kvm_vcpu_ioctl_device_attr() in the code base and reading of the function would have shown that the report was a false positive, albeit a somewhat odd one, and that returning -EINVAL was likely the wrong thing to do. If you're unsure if something is a real bug, please ask a question.

Rapid firing patches at the list also makes reviewers grumpy as it again suggests a lack of due dilegence, especially when the patches have typos ("EINV" in v2) and/or have obvious shortcomings (missing "break" in v3).

TL;DR: I have no objection whatsover to fixing (potential) bugs found by static analyzers, but please slow down and (a) make sure that it's actually a bug, (b) ask if you're unsure, and (c) do your best to ensure that what you're sending is an overall improvement.

#10 lore raw
Quoted 56 lines · Sean Christopherson
Quoted 29 lines · Ameer Hamza
Quoted 28 lines · Sean Christopherson
Quoted 18 lines · Ameer Hamza

If undefined ioctl number is passed to the kvm_vcpu_ioctl_device_attr ioctl, we should trigger KVM_BUG_ON() and return with EIO to silent coverity warning.

Addresses-Coverity: 1494124 ("Uninitialized scalar variable")

---
Changes in v3:
Added KVM_BUG_ON() as default case and returned -EIO

---
 arch/x86/kvm/x86.c | 3 +++
 1 file changed, 3 insertions(+)
arch/x86/kvm/x86.c +0 -0
index e0aa4dd53c7f..b37068f847ff 100644

arch/x86/kvm/x86.c static int kvm_vcpu_ioctl_device_attr(struct kvm_vcpu *vcpu, · line 5024

Quoted 6 lines
arch/x86/kvm/x86.c +3 -0
@@ -5019,6 +5019,9 @@ static int kvm_vcpu_ioctl_device_attr(struct kvm_vcpu *vcpu,
 	case KVM_SET_DEVICE_ATTR:
 		r = kvm_arch_tsc_set_attr(vcpu, &attr);
 		break;
+	default:
+		KVM_BUG_ON(1, vcpu->kvm);
+		r = -EIO;

At least have a

Quoted 6 lines
Quoted 3 lines
		break;

if we're going to be pedantic about things.

I just started as a contributer in this community and trying to fix issues found by static analyzer tools. If you think that's not necessary, its totally fine :)

(Most) Static analyzers are great, they definitely find real bugs. But they also have a fair number of false positives, e.g. this is a firmly a false positive, so the results of any static analyzer needs to thought about critically, not blindly followed. It's completely understandable that Coverity got tripped up in this case, but that's exactly why having a human vet the bug report is necessary.

There is arguably value in having a default statement to ensure future KVM code doesn't end up adding a bad call, which is why I'm not completely opposed to the above addition.

Where folks, myself included, get a bit grumpy is when patches are sent to "fix" bug reports from static analyzers without evidence that the submitter has done their due dilegence to understand the code they are changing, e.g. even without any understanding of KVM, a search of kvm_vcpu_ioctl_device_attr() in the code base and reading of the function would have shown that the report was a false positive, albeit a somewhat odd one, and that returning -EINVAL was likely the wrong thing to do. If you're unsure if something is a real bug, please ask a question.

Quoted 8 lines

Rapid firing patches at the list also makes reviewers grumpy as it again suggests a lack of due dilegence, especially when the patches have typos ("EINV" in v2) and/or have obvious shortcomings (missing "break" in v3).

TL;DR: I have no objection whatsover to fixing (potential) bugs found by static analyzers, but please slow down and (a) make sure that it's actually a bug, (b) ask if you're unsure, and (c) do your best to ensure that what you're sending is an overall improvement.

Totally agreed with you. Thank you so much for your insights on this. I will keep this into consideration moving forward.

Best Regards,
Hamza.