[PATCH] x86: KVM, fix lock imbalance

[PATCH] x86: KVM, fix lock imbalance

[Jiri Slaby]

Stanse found that there is an omitted unlock in kvm_create_pit in one fail
path. Add proper unlock there.

Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Cc: Avi Kivity <avi@redhat.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: x86@kernel.org
Cc: Gleb Natapov <gleb@redhat.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Gregory Haskins <ghaskins@novell.com>
Cc: kvm@vger.kernel.org
---
 arch/x86/kvm/i8254.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/arch/x86/kvm/i8254.c b/arch/x86/kvm/i8254.c
index 467cc47..70db4d4 100644
--- a/arch/x86/kvm/i8254.c
+++ b/arch/x86/kvm/i8254.c
@@ -696,6 +696,7 @@ struct kvm_pit *kvm_create_pit(struct kvm *kvm, u32 flags)
 
 	pit->wq = create_singlethread_workqueue("kvm-pit-wq");
 	if (!pit->wq) {
+		mutex_unlock(&pit->pit_state.lock);
 		kfree(pit);
 		return NULL;
 	}
-- 
1.7.1

Re: [PATCH] x86: KVM, fix lock imbalance

[Ingo Molnar]

* Jiri Slaby <jirislaby@gmail.com> wrote:

> Stanse found that there is an omitted unlock in kvm_create_pit in one fail
> path. Add proper unlock there.
> 
> Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
> Cc: Avi Kivity <avi@redhat.com>
> Cc: Marcelo Tosatti <mtosatti@redhat.com>
> Cc: Thomas Gleixner <tglx@linutronix.de>
> Cc: Ingo Molnar <mingo@redhat.com>
> Cc: "H. Peter Anvin" <hpa@zytor.com>
> Cc: x86@kernel.org
> Cc: Gleb Natapov <gleb@redhat.com>
> Cc: "Michael S. Tsirkin" <mst@redhat.com>
> Cc: Gregory Haskins <ghaskins@novell.com>
> Cc: kvm@vger.kernel.org
> ---
>  arch/x86/kvm/i8254.c |    1 +
>  1 files changed, 1 insertions(+), 0 deletions(-)
> 
> diff --git a/arch/x86/kvm/i8254.c b/arch/x86/kvm/i8254.c
> index 467cc47..70db4d4 100644
> --- a/arch/x86/kvm/i8254.c
> +++ b/arch/x86/kvm/i8254.c
> @@ -696,6 +696,7 @@ struct kvm_pit *kvm_create_pit(struct kvm *kvm, u32 flags)
>  
>  	pit->wq = create_singlethread_workqueue("kvm-pit-wq");
>  	if (!pit->wq) {
> +		mutex_unlock(&pit->pit_state.lock);
>  		kfree(pit);
>  		return NULL;
>  	}

A cleanliness comment: why is that tear-down/dealloc sequence open-coded? It 
should be at the end of the function, with goto labels, like we do it in 
similar cases.

Thanks,

	Ingo

Re: [PATCH] x86: KVM, fix lock imbalance

[Jiri Slaby]

On 07/07/2010 03:05 PM, Ingo Molnar wrote:
> 
> * Jiri Slaby <jirislaby@gmail.com> wrote:
> 
>> Stanse found that there is an omitted unlock in kvm_create_pit in one fail
>> path. Add proper unlock there.
>>
>> Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
>> Cc: Avi Kivity <avi@redhat.com>
>> Cc: Marcelo Tosatti <mtosatti@redhat.com>
>> Cc: Thomas Gleixner <tglx@linutronix.de>
>> Cc: Ingo Molnar <mingo@redhat.com>
>> Cc: "H. Peter Anvin" <hpa@zytor.com>
>> Cc: x86@kernel.org
>> Cc: Gleb Natapov <gleb@redhat.com>
>> Cc: "Michael S. Tsirkin" <mst@redhat.com>
>> Cc: Gregory Haskins <ghaskins@novell.com>
>> Cc: kvm@vger.kernel.org
>> ---
>>  arch/x86/kvm/i8254.c |    1 +
>>  1 files changed, 1 insertions(+), 0 deletions(-)
>>
>> diff --git a/arch/x86/kvm/i8254.c b/arch/x86/kvm/i8254.c
>> index 467cc47..70db4d4 100644
>> --- a/arch/x86/kvm/i8254.c
>> +++ b/arch/x86/kvm/i8254.c
>> @@ -696,6 +696,7 @@ struct kvm_pit *kvm_create_pit(struct kvm *kvm, u32 flags)
>>  
>>  	pit->wq = create_singlethread_workqueue("kvm-pit-wq");
>>  	if (!pit->wq) {
>> +		mutex_unlock(&pit->pit_state.lock);
>>  		kfree(pit);
>>  		return NULL;
>>  	}
> 
> A cleanliness comment: why is that tear-down/dealloc sequence open-coded? It 
> should be at the end of the function, with goto labels, like we do it in 
> similar cases.

Because the lock is around a block only. I usually don't create a goto
fail-paths in these cases. Do you want one?

thanks,
-- 
js

Re: [PATCH] x86: KVM, fix lock imbalance

[Jiri Slaby]

On 07/07/2010 03:07 PM, Jiri Slaby wrote:
>>> --- a/arch/x86/kvm/i8254.c
>>> +++ b/arch/x86/kvm/i8254.c
>>> @@ -696,6 +696,7 @@ struct kvm_pit *kvm_create_pit(struct kvm *kvm, u32 flags)
>>>  
>>>  	pit->wq = create_singlethread_workqueue("kvm-pit-wq");
>>>  	if (!pit->wq) {
>>> +		mutex_unlock(&pit->pit_state.lock);
>>>  		kfree(pit);
>>>  		return NULL;
>>>  	}
>>
>> A cleanliness comment: why is that tear-down/dealloc sequence open-coded? It 
>> should be at the end of the function, with goto labels, like we do it in 
>> similar cases.
> 
> Because the lock is around a block only. I usually don't create a goto
> fail-paths in these cases.

To be more precise what I mean by that:
if ()
  return;

lock();
...
if () { [single if inside the crit section]
  unlock();
  return;
}
...
unlock()

...
if ()
  return;
...
if ()
  return;

-- 
js

Re: [PATCH] x86: KVM, fix lock imbalance

[Avi Kivity]

On 07/07/2010 04:07 PM, Jiri Slaby wrote:
>
>> A cleanliness comment: why is that tear-down/dealloc sequence open-coded? It
>> should be at the end of the function, with goto labels, like we do it in
>> similar cases.
>>      
> Because the lock is around a block only. I usually don't create a goto
> fail-paths in these cases. Do you want one?
>    

In any case, the patch is a minimal fix, so I'll apply it.  Any clean 
ups can go on top.

-- 
I have a truly marvellous patch that fixes the bug which this
signature is too narrow to contain.

Re: [PATCH] x86: KVM, fix lock imbalance

[Ingo Molnar]

* Avi Kivity <avi@redhat.com> wrote:

> On 07/07/2010 04:07 PM, Jiri Slaby wrote:
> >
> >>A cleanliness comment: why is that tear-down/dealloc sequence open-coded? It
> >>should be at the end of the function, with goto labels, like we do it in
> >>similar cases.
> >Because the lock is around a block only. I usually don't create a goto
> >fail-paths in these cases. Do you want one?
> 
> In any case, the patch is a minimal fix, so I'll apply it.  Any clean ups 
> can go on top.

The reason the pattern caught my attention is that it is one of the typical 
cases where goto labels help prevent similar bugs. I.e. had it been clean to 
begin with the bug might not have happened.

Thanks,

	Ingo

Re: [PATCH] x86: KVM, fix lock imbalance

[Avi Kivity]

On 07/07/2010 05:10 PM, Ingo Molnar wrote:
>
>> In any case, the patch is a minimal fix, so I'll apply it.  Any clean ups
>> can go on top.
>>      
> The reason the pattern caught my attention is that it is one of the typical
> cases where goto labels help prevent similar bugs. I.e. had it been clean to
> begin with the bug might not have happened.
>    

Agreed.

-- 
I have a truly marvellous patch that fixes the bug which this
signature is too narrow to contain.