[PATCH] counter: fix an IS_ERR() vs NULL bug

[PATCH] counter: fix an IS_ERR() vs NULL bug

[Dan Carpenter]

There are 8 callers for devm_counter_alloc() and they all check for NULL
instead of error pointers.  I think NULL is the better thing to return
for allocation functions so update counter_alloc() and devm_counter_alloc()
to return NULL instead of error pointers.

Fixes: c18e2760308e ("counter: Provide alternative counter registration functions")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
---
 drivers/counter/counter-core.c | 12 +++++-------
 1 file changed, 5 insertions(+), 7 deletions(-)

diff --git a/drivers/counter/counter-core.c b/drivers/counter/counter-core.c
index 7e0957eea094..0abb28cd3d16 100644
--- a/drivers/counter/counter-core.c
+++ b/drivers/counter/counter-core.c
@@ -90,10 +90,8 @@ struct counter_device *counter_alloc(size_t sizeof_priv)
 	int err;
 
 	ch = kzalloc(sizeof(*ch) + sizeof_priv, GFP_KERNEL);
-	if (!ch) {
-		err = -ENOMEM;
+	if (!ch)
 		goto err_alloc_ch;
-	}
 
 	counter = &ch->counter;
 	dev = &counter->dev;
@@ -125,7 +123,7 @@ struct counter_device *counter_alloc(size_t sizeof_priv)
 	kfree(ch);
 err_alloc_ch:
 
-	return ERR_PTR(err);
+	return NULL;
 }
 EXPORT_SYMBOL_GPL(counter_alloc);
 
@@ -208,12 +206,12 @@ struct counter_device *devm_counter_alloc(struct device *dev, size_t sizeof_priv
 	int err;
 
 	counter = counter_alloc(sizeof_priv);
-	if (IS_ERR(counter))
-		return counter;
+	if (!counter)
+		return NULL;
 
 	err = devm_add_action_or_reset(dev, devm_counter_put, counter);
 	if (err < 0)
-		return ERR_PTR(err);
+		return NULL;
 
 	return counter;
 }
-- 
2.20.1

Re: [PATCH] counter: fix an IS_ERR() vs NULL bug

[Uwe Kleine-König]

[-- Attachment #1: Type: text/plain, Size: 1017 bytes --]

On Tue, Jan 11, 2022 at 10:25:00AM +0300, Dan Carpenter wrote:
> There are 8 callers for devm_counter_alloc() and they all check for NULL
> instead of error pointers.  I think NULL is the better thing to return
> for allocation functions so update counter_alloc() and devm_counter_alloc()
> to return NULL instead of error pointers.
> 
> Fixes: c18e2760308e ("counter: Provide alternative counter registration functions")
> Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>

Oh indeed. I wonder how you found that problem.

There is a possibility that counter_alloc() (as it's implemented now in
next) returns an error value != -ENOMEM. (ida_alloc() can return
-ENOSPC, counter_chrdev_add() can return -EINVAL). Still returning NULL
on error looks sane.

Reviewed-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>

Thanks
Uwe


-- 
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | https://www.pengutronix.de/ |

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: [PATCH] counter: fix an IS_ERR() vs NULL bug

[Dan Carpenter]

On Tue, Jan 11, 2022 at 09:20:55AM +0100, Uwe Kleine-König wrote:
> On Tue, Jan 11, 2022 at 10:25:00AM +0300, Dan Carpenter wrote:
> > There are 8 callers for devm_counter_alloc() and they all check for NULL
> > instead of error pointers.  I think NULL is the better thing to return
> > for allocation functions so update counter_alloc() and devm_counter_alloc()
> > to return NULL instead of error pointers.
> > 
> > Fixes: c18e2760308e ("counter: Provide alternative counter registration functions")
> > Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
> 
> Oh indeed. I wonder how you found that problem.

Of course from static analysis.  Each caller triggers a "this NULL
check should be an IS_ERR() check warning" followed by a "potential
error pointer dereference" warning.

regards,
dan carpenter

Re: [PATCH] counter: fix an IS_ERR() vs NULL bug

[William Breathitt Gray]

[-- Attachment #1: Type: text/plain, Size: 2103 bytes --]

On Tue, Jan 11, 2022 at 10:25:00AM +0300, Dan Carpenter wrote:
> There are 8 callers for devm_counter_alloc() and they all check for NULL
> instead of error pointers.  I think NULL is the better thing to return
> for allocation functions so update counter_alloc() and devm_counter_alloc()
> to return NULL instead of error pointers.
> 
> Fixes: c18e2760308e ("counter: Provide alternative counter registration functions")
> Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>

Hi Dan,

Thank you for catching this bug. I do request a minor adjustment to this
patch for the sake of matching the format of the rest of the code in
this file: please remove the goto err_alloc_ch and simply exit early
with a return NULL.

> ---
>  drivers/counter/counter-core.c | 12 +++++-------
>  1 file changed, 5 insertions(+), 7 deletions(-)
> 
> diff --git a/drivers/counter/counter-core.c b/drivers/counter/counter-core.c
> index 7e0957eea094..0abb28cd3d16 100644
> --- a/drivers/counter/counter-core.c
> +++ b/drivers/counter/counter-core.c
> @@ -90,10 +90,8 @@ struct counter_device *counter_alloc(size_t sizeof_priv)
>  	int err;
>  
>  	ch = kzalloc(sizeof(*ch) + sizeof_priv, GFP_KERNEL);
> -	if (!ch) {
> -		err = -ENOMEM;
> +	if (!ch)
>  		goto err_alloc_ch;

So return NULL here.

> -	}
>  
>  	counter = &ch->counter;
>  	dev = &counter->dev;
> @@ -125,7 +123,7 @@ struct counter_device *counter_alloc(size_t sizeof_priv)
>  	kfree(ch);
>  err_alloc_ch:

Remove this label.

Thank you,

William Breathitt Gray

>  
> -	return ERR_PTR(err);
> +	return NULL;
>  }
>  EXPORT_SYMBOL_GPL(counter_alloc);
>  
> @@ -208,12 +206,12 @@ struct counter_device *devm_counter_alloc(struct device *dev, size_t sizeof_priv
>  	int err;
>  
>  	counter = counter_alloc(sizeof_priv);
> -	if (IS_ERR(counter))
> -		return counter;
> +	if (!counter)
> +		return NULL;
>  
>  	err = devm_add_action_or_reset(dev, devm_counter_put, counter);
>  	if (err < 0)
> -		return ERR_PTR(err);
> +		return NULL;
>  
>  	return counter;
>  }
> -- 
> 2.20.1
> 

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]