[PATCH V3] workqueue: fix double unlock bug

[PATCH V3] workqueue: fix double unlock bug

[Daeseok Youn]

Use default pwq when alloc_unbound_pwq() is failed.

And remove "if" condition for whether "pwq" is same as "wq->dfl_pwq"
when wq_calc_node_cpumask() returns false and just use "goto use_dfl_pwq"

Signed-off-by: Daeseok Youn <daeseok.youn@gmail.com>
---
V2: replace "if condition" with "goto" as Lai's comment.
V3: Use default pwq when alloc_unbound_pwq() is failed.

 kernel/workqueue.c |    8 +++-----
 1 files changed, 3 insertions(+), 5 deletions(-)

diff --git a/kernel/workqueue.c b/kernel/workqueue.c
index 0ee63af..0679854 100644
--- a/kernel/workqueue.c
+++ b/kernel/workqueue.c
@@ -4087,10 +4087,7 @@ static void wq_update_unbound_numa(struct workqueue_struct *wq, int cpu,
 		if (cpumask_equal(cpumask, pwq->pool->attrs->cpumask))
 			goto out_unlock;
 	} else {
-		if (pwq == wq->dfl_pwq)
-			goto out_unlock;
-		else
-			goto use_dfl_pwq;
+		goto use_dfl_pwq;
 	}
 
 	mutex_unlock(&wq->mutex);
@@ -4100,7 +4097,8 @@ static void wq_update_unbound_numa(struct workqueue_struct *wq, int cpu,
 	if (!pwq) {
 		pr_warning("workqueue: allocation failed while updating NUMA affinity of \"%s\"\n",
 			   wq->name);
-		goto out_unlock;
+		mutex_lock(&wq->mutex);
+		goto use_dfl_pwq;
 	}
 
 	/*
-- 
1.7.4.4

Re: [PATCH V3] workqueue: fix double unlock bug

[Lai Jiangshan]

Hi, Tejun

Acked-by: Lai Jiangshan <laijs@cn.fujitsu.com>
CC: stable@kernel.org

Thanks,
Lai

On 04/15/2014 07:17 AM, Daeseok Youn wrote:
> 
> Use default pwq when alloc_unbound_pwq() is failed.
> 
> And remove "if" condition for whether "pwq" is same as "wq->dfl_pwq"
> when wq_calc_node_cpumask() returns false and just use "goto use_dfl_pwq"
> 
> Signed-off-by: Daeseok Youn <daeseok.youn@gmail.com>
> ---
> V2: replace "if condition" with "goto" as Lai's comment.
> V3: Use default pwq when alloc_unbound_pwq() is failed.
> 
>  kernel/workqueue.c |    8 +++-----
>  1 files changed, 3 insertions(+), 5 deletions(-)
> 
> diff --git a/kernel/workqueue.c b/kernel/workqueue.c
> index 0ee63af..0679854 100644
> --- a/kernel/workqueue.c
> +++ b/kernel/workqueue.c
> @@ -4087,10 +4087,7 @@ static void wq_update_unbound_numa(struct workqueue_struct *wq, int cpu,
>  		if (cpumask_equal(cpumask, pwq->pool->attrs->cpumask))
>  			goto out_unlock;
>  	} else {
> -		if (pwq == wq->dfl_pwq)
> -			goto out_unlock;
> -		else
> -			goto use_dfl_pwq;
> +		goto use_dfl_pwq;
>  	}
>  
>  	mutex_unlock(&wq->mutex);
> @@ -4100,7 +4097,8 @@ static void wq_update_unbound_numa(struct workqueue_struct *wq, int cpu,
>  	if (!pwq) {
>  		pr_warning("workqueue: allocation failed while updating NUMA affinity of \"%s\"\n",
>  			   wq->name);
> -		goto out_unlock;
> +		mutex_lock(&wq->mutex);
> +		goto use_dfl_pwq;
>  	}
>  
>  	/*

Re: [PATCH V3] workqueue: fix double unlock bug

[Tejun Heo]

On Tue, Apr 15, 2014 at 08:17:43AM +0900, Daeseok Youn wrote:
> 
> Use default pwq when alloc_unbound_pwq() is failed.
> 
> And remove "if" condition for whether "pwq" is same as "wq->dfl_pwq"
> when wq_calc_node_cpumask() returns false and just use "goto use_dfl_pwq"
> 
> Signed-off-by: Daeseok Youn <daeseok.youn@gmail.com>

This is an awfully inadequate patch description.  Can you change it to
something like the following?

  workqueue: fix bugs in wq_update_unbound_numa() failure path

  wq_update_unbound_numa() failure path has the following two bugs.

  * alloc_unbound_pwq() is called without holding wq->mutex; however, if
    the allocation fails, it jumps to out_unlock which tries to unlock
    wq->mutex.

  * The function should switch to dfl_pwq on failure but didn't do so
    after alloc_unbound_pwq() failure.

  Fix it by regrabbing wq->mutex and jumping to use_dfl_pwq on
  alloc_unbound_pwq() failure.
      

> diff --git a/kernel/workqueue.c b/kernel/workqueue.c
> index 0ee63af..0679854 100644
> --- a/kernel/workqueue.c
> +++ b/kernel/workqueue.c
> @@ -4087,10 +4087,7 @@ static void wq_update_unbound_numa(struct workqueue_struct *wq, int cpu,
>  		if (cpumask_equal(cpumask, pwq->pool->attrs->cpumask))
>  			goto out_unlock;
>  	} else {
> -		if (pwq == wq->dfl_pwq)
> -			goto out_unlock;
> -		else
> -			goto use_dfl_pwq;
> +		goto use_dfl_pwq;

And please put this in a separate patch.  Patches which are to be
backported through -stable should be minimal.  Also, please update the
comment above to reflect the change.

Thanks.

-- 
tejun

Re: [PATCH V3] workqueue: fix double unlock bug

[DaeSeok Youn]

Hi, Tejun.

2014-04-15 23:43 GMT+09:00 Tejun Heo <tj@kernel.org>:
> On Tue, Apr 15, 2014 at 08:17:43AM +0900, Daeseok Youn wrote:
>>
>> Use default pwq when alloc_unbound_pwq() is failed.
>>
>> And remove "if" condition for whether "pwq" is same as "wq->dfl_pwq"
>> when wq_calc_node_cpumask() returns false and just use "goto use_dfl_pwq"
>>
>> Signed-off-by: Daeseok Youn <daeseok.youn@gmail.com>
>
> This is an awfully inadequate patch description.  Can you change it to
> something like the following?
OK. I will change the patch description as your comment.

>
>   workqueue: fix bugs in wq_update_unbound_numa() failure path
>
>   wq_update_unbound_numa() failure path has the following two bugs.
>
>   * alloc_unbound_pwq() is called without holding wq->mutex; however, if
>     the allocation fails, it jumps to out_unlock which tries to unlock
>     wq->mutex.
>
>   * The function should switch to dfl_pwq on failure but didn't do so
>     after alloc_unbound_pwq() failure.
>
>   Fix it by regrabbing wq->mutex and jumping to use_dfl_pwq on
>   alloc_unbound_pwq() failure.
>
>
>> diff --git a/kernel/workqueue.c b/kernel/workqueue.c
>> index 0ee63af..0679854 100644
>> --- a/kernel/workqueue.c
>> +++ b/kernel/workqueue.c
>> @@ -4087,10 +4087,7 @@ static void wq_update_unbound_numa(struct workqueue_struct *wq, int cpu,
>>               if (cpumask_equal(cpumask, pwq->pool->attrs->cpumask))
>>                       goto out_unlock;
>>       } else {
>> -             if (pwq == wq->dfl_pwq)
>> -                     goto out_unlock;
>> -             else
>> -                     goto use_dfl_pwq;
>> +             goto use_dfl_pwq;
>
> And please put this in a separate patch.  Patches which are to be
> backported through -stable should be minimal.  Also, please update the
> comment above to reflect the change.
Yes, I will do.

Thanks for review.

Regards,
Daeseok Youn.
>
> Thanks.
>
> --
> tejun