[PATCH] [SCSI] hpsa: fix return value check in start_controller_lockup_detector()

[PATCH] [SCSI] hpsa: fix return value check in start_controller_lockup_detector()

[Wei Yongjun]

From: Wei Yongjun <yongjun_wei@trendmicro.com.cn>

In case of error, the function kthread_run() returns ERR_PTR()
and never returns NULL. The NULL test in the return value check
should be replaced with IS_ERR().

Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
---
 drivers/scsi/hpsa.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/scsi/hpsa.c b/drivers/scsi/hpsa.c
index 891c86b..f413b14 100644
--- a/drivers/scsi/hpsa.c
+++ b/drivers/scsi/hpsa.c
@@ -4757,7 +4757,8 @@ static void start_controller_lockup_detector(struct ctlr_info *h)
 			kthread_run(detect_controller_lockup_thread,
 						NULL, HPSA);
 	}
-	if (!hpsa_lockup_detector) {
+	if (IS_ERR(hpsa_lockup_detector)) {
+		hpsa_lockup_detector = NULL;
 		dev_warn(&h->pdev->dev,
 			"Could not start lockup detector thread\n");
 		return;

Re: [PATCH] [SCSI] hpsa: fix return value check in start_controller_lockup_detector()

[scameron]

On Tue, Oct 29, 2013 at 02:43:56PM +0800, Wei Yongjun wrote:
> From: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
> 
> In case of error, the function kthread_run() returns ERR_PTR()
> and never returns NULL. The NULL test in the return value check
> should be replaced with IS_ERR().
> 
> Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
> ---
>  drivers/scsi/hpsa.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/scsi/hpsa.c b/drivers/scsi/hpsa.c
> index 891c86b..f413b14 100644
> --- a/drivers/scsi/hpsa.c
> +++ b/drivers/scsi/hpsa.c
> @@ -4757,7 +4757,8 @@ static void start_controller_lockup_detector(struct ctlr_info *h)
>  			kthread_run(detect_controller_lockup_thread,
>  						NULL, HPSA);
>  	}
> -	if (!hpsa_lockup_detector) {
> +	if (IS_ERR(hpsa_lockup_detector)) {
> +		hpsa_lockup_detector = NULL;
>  		dev_warn(&h->pdev->dev,
>  			"Could not start lockup detector thread\n");
>  		return;

Ack.

Thanks.

-- steve

Re: [PATCH] [SCSI] hpsa: fix return value check in start_controller_lockup_detector()

[James Bottomley]

On Tue, 2013-10-29 at 11:09 -0500, scameron@beardog.cce.hp.com wrote:
> On Tue, Oct 29, 2013 at 02:43:56PM +0800, Wei Yongjun wrote:
> > From: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
> > 
> > In case of error, the function kthread_run() returns ERR_PTR()
> > and never returns NULL. The NULL test in the return value check
> > should be replaced with IS_ERR().
> > 
> > Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
> > ---
> >  drivers/scsi/hpsa.c | 3 ++-
> >  1 file changed, 2 insertions(+), 1 deletion(-)
> > 
> > diff --git a/drivers/scsi/hpsa.c b/drivers/scsi/hpsa.c
> > index 891c86b..f413b14 100644
> > --- a/drivers/scsi/hpsa.c
> > +++ b/drivers/scsi/hpsa.c
> > @@ -4757,7 +4757,8 @@ static void start_controller_lockup_detector(struct ctlr_info *h)
> >  			kthread_run(detect_controller_lockup_thread,
> >  						NULL, HPSA);
> >  	}
> > -	if (!hpsa_lockup_detector) {
> > +	if (IS_ERR(hpsa_lockup_detector)) {
> > +		hpsa_lockup_detector = NULL;
> >  		dev_warn(&h->pdev->dev,
> >  			"Could not start lockup detector thread\n");
> >  		return;
> 
> Ack.

Hmm, this whole lockup thread start/stop thing is horribly racy with
hotplug (and bind and unbind).

Firstly, why do you need an actual thread, why not just use a workqueue?
That way you don't need a list or a lock, you can just schedule work for
each instance?  since they don't interfere, no global list and no lock.

If you insist on a thread, it should probably start at module insertion
and be shut down at module removal, so no races.  You can have it sleep
forever on list_empty(&hpsa_ctlr_list) and wake it when you add a
controller.

James

Re: [PATCH] [SCSI] hpsa: fix return value check in start_controller_lockup_detector()

[scameron]

On Tue, Oct 29, 2013 at 02:52:38PM -0700, James Bottomley wrote:
> On Tue, 2013-10-29 at 11:09 -0500, scameron@beardog.cce.hp.com wrote:
> > On Tue, Oct 29, 2013 at 02:43:56PM +0800, Wei Yongjun wrote:
> > > From: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
> > > 
> > > In case of error, the function kthread_run() returns ERR_PTR()
> > > and never returns NULL. The NULL test in the return value check
> > > should be replaced with IS_ERR().
> > > 
> > > Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
> > > ---
> > >  drivers/scsi/hpsa.c | 3 ++-
> > >  1 file changed, 2 insertions(+), 1 deletion(-)
> > > 
> > > diff --git a/drivers/scsi/hpsa.c b/drivers/scsi/hpsa.c
> > > index 891c86b..f413b14 100644
> > > --- a/drivers/scsi/hpsa.c
> > > +++ b/drivers/scsi/hpsa.c
> > > @@ -4757,7 +4757,8 @@ static void start_controller_lockup_detector(struct ctlr_info *h)
> > >  			kthread_run(detect_controller_lockup_thread,
> > >  						NULL, HPSA);
> > >  	}
> > > -	if (!hpsa_lockup_detector) {
> > > +	if (IS_ERR(hpsa_lockup_detector)) {
> > > +		hpsa_lockup_detector = NULL;
> > >  		dev_warn(&h->pdev->dev,
> > >  			"Could not start lockup detector thread\n");
> > >  		return;
> > 
> > Ack.
> 
> Hmm, this whole lockup thread start/stop thing is horribly racy with
> hotplug (and bind and unbind).
> 
> Firstly, why do you need an actual thread, why not just use a workqueue?
> That way you don't need a list or a lock, you can just schedule work for
> each instance?  since they don't interfere, no global list and no lock.
> 
> If you insist on a thread, it should probably start at module insertion
> and be shut down at module removal, so no races.  You can have it sleep
> forever on list_empty(&hpsa_ctlr_list) and wake it when you add a
> controller.

Alright, I'll work on it.  Thanks.

-- steve