[v2] staging: gpib: Fix error handling paths in cb_gpib_probe()

[PATCH v2] staging: gpib: Fix error handling paths in cb_gpib_probe()

Posted by Christophe JAILLET 3 months ago

If cb_gpib_config() fails, 'info' needs to be freed, as already done in the
remove function.

While at it, remove a pointless comment related to gpib_attach().

Fixes: e9dc69956d4d ("staging: gpib: Add Computer Boards GPIB driver")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
---
Changes in v2:
  - Fix the Fixes tag   [Thomas Richard]
  - Synch with latest -next
  - no compile tested. I think, thanks to commit 79d2e1919a27
    ("staging: gpib: fix Makefiles")

v1: https://lore.kernel.org/all/459c267de8c9bf48fcb555364930ae7e3cdc798b.1729940596.git.christophe.jaillet@wanadoo.fr/

Compile tested only.
---
 drivers/staging/gpib/cb7210/cb7210.c | 15 +++++++++++----
 1 file changed, 11 insertions(+), 4 deletions(-)

diff --git a/drivers/staging/gpib/cb7210/cb7210.c b/drivers/staging/gpib/cb7210/cb7210.c
index 298ed306189d..3e2397898a9b 100644
--- a/drivers/staging/gpib/cb7210/cb7210.c
+++ b/drivers/staging/gpib/cb7210/cb7210.c
@@ -1184,8 +1184,7 @@ struct local_info {
 static int cb_gpib_probe(struct pcmcia_device *link)
 {
 	struct local_info *info;
-
-//	int ret, i;
+	int ret;
 
 	/* Allocate space for private device-specific data */
 	info = kzalloc(sizeof(*info), GFP_KERNEL);
@@ -1211,8 +1210,16 @@ static int cb_gpib_probe(struct pcmcia_device *link)
 
 	/* Register with Card Services */
 	curr_dev = link;
-	return cb_gpib_config(link);
-} /* gpib_attach */
+	ret = cb_gpib_config(link);
+	if (ret)
+		goto free_info;
+
+	return 0;
+
+free_info:
+	kfree(info);
+	return ret;
+}
 
 /*
  *   This deletes a driver "instance".  The device is de-registered
-- 
2.50.0

Re: [PATCH v2] staging: gpib: Fix error handling paths in cb_gpib_probe()

Posted by Thomas Richard 3 months ago

Hi Christophe,

On 7/5/25 11:52 AM, Christophe JAILLET wrote:
> If cb_gpib_config() fails, 'info' needs to be freed, as already done in the
> remove function.
> 
> While at it, remove a pointless comment related to gpib_attach().
> 
> Fixes: e9dc69956d4d ("staging: gpib: Add Computer Boards GPIB driver")
> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
> ---
> Changes in v2:
>   - Fix the Fixes tag   [Thomas Richard]
>   - Synch with latest -next
>   - no compile tested. I think, thanks to commit 79d2e1919a27
>     ("staging: gpib: fix Makefiles")
> 
> v1: https://lore.kernel.org/all/459c267de8c9bf48fcb555364930ae7e3cdc798b.1729940596.git.christophe.jaillet@wanadoo.fr/
> 
> Compile tested only.
> ---
>  drivers/staging/gpib/cb7210/cb7210.c | 15 +++++++++++----
>  1 file changed, 11 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/staging/gpib/cb7210/cb7210.c b/drivers/staging/gpib/cb7210/cb7210.c
> index 298ed306189d..3e2397898a9b 100644
> --- a/drivers/staging/gpib/cb7210/cb7210.c
> +++ b/drivers/staging/gpib/cb7210/cb7210.c
> @@ -1184,8 +1184,7 @@ struct local_info {
>  static int cb_gpib_probe(struct pcmcia_device *link)
>  {
>  	struct local_info *info;
> -
> -//	int ret, i;
> +	int ret;
>  
>  	/* Allocate space for private device-specific data */
>  	info = kzalloc(sizeof(*info), GFP_KERNEL);

You should use devm_kzalloc(). The memory will be automatically freed by
the core. So no need to call kfree() in case of error during the probe.
And you can remove the kfree() in cb_gpib_remove().

Regards,

Thomas

Re: [PATCH v2] staging: gpib: Fix error handling paths in cb_gpib_probe()

Posted by Greg Kroah-Hartman 3 months ago

On Sat, Jul 05, 2025 at 04:17:35PM +0200, Thomas Richard wrote:
> Hi Christophe,
> 
> On 7/5/25 11:52 AM, Christophe JAILLET wrote:
> > If cb_gpib_config() fails, 'info' needs to be freed, as already done in the
> > remove function.
> > 
> > While at it, remove a pointless comment related to gpib_attach().
> > 
> > Fixes: e9dc69956d4d ("staging: gpib: Add Computer Boards GPIB driver")
> > Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
> > ---
> > Changes in v2:
> >   - Fix the Fixes tag   [Thomas Richard]
> >   - Synch with latest -next
> >   - no compile tested. I think, thanks to commit 79d2e1919a27
> >     ("staging: gpib: fix Makefiles")
> > 
> > v1: https://lore.kernel.org/all/459c267de8c9bf48fcb555364930ae7e3cdc798b.1729940596.git.christophe.jaillet@wanadoo.fr/
> > 
> > Compile tested only.
> > ---
> >  drivers/staging/gpib/cb7210/cb7210.c | 15 +++++++++++----
> >  1 file changed, 11 insertions(+), 4 deletions(-)
> > 
> > diff --git a/drivers/staging/gpib/cb7210/cb7210.c b/drivers/staging/gpib/cb7210/cb7210.c
> > index 298ed306189d..3e2397898a9b 100644
> > --- a/drivers/staging/gpib/cb7210/cb7210.c
> > +++ b/drivers/staging/gpib/cb7210/cb7210.c
> > @@ -1184,8 +1184,7 @@ struct local_info {
> >  static int cb_gpib_probe(struct pcmcia_device *link)
> >  {
> >  	struct local_info *info;
> > -
> > -//	int ret, i;
> > +	int ret;
> >  
> >  	/* Allocate space for private device-specific data */
> >  	info = kzalloc(sizeof(*info), GFP_KERNEL);
> 
> You should use devm_kzalloc(). The memory will be automatically freed by
> the core. So no need to call kfree() in case of error during the probe.
> And you can remove the kfree() in cb_gpib_remove().

Some people, myself include, hate the devm_*() apis, so no need to force
them on anyone please :)

thanks,

greg k-h