[PATCH v2 resend] pinctrl: devicetree: fix refcount leak in pinctrl_dt_to_map()

Zeng Heng posted 1 patch 2 weeks, 1 day ago
drivers/pinctrl/devicetree.c | 11 +++++++----
1 file changed, 7 insertions(+), 4 deletions(-)
[PATCH v2 resend] pinctrl: devicetree: fix refcount leak in pinctrl_dt_to_map()
Posted by Zeng Heng 2 weeks, 1 day ago
If we fail to allocate propname buffer, we need to drop the reference
count we just took, otherwise it will lead reference leak. Here the
error exit path is modified to jump to the err label and call
pinctrl_dt_free_maps() which would drop the counter.

In the meantime, if it is found that the property 'pinctrl-0' is not
present, ENODEV is returned and also jump to the err label and call the
free function, in case the Smatch tool complains.

Fixes: 91d5c5060ee2 ("pinctrl: devicetree: fix null pointer dereferencing in pinctrl_dt_to_map")
Suggested-by: Dan Carpenter <dan.carpenter@linaro.org>
Signed-off-by: Zeng Heng <zengheng4@huawei.com>
---
v2: add a comment and modify the commit message, without any logical
    changes.
---
 drivers/pinctrl/devicetree.c | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/drivers/pinctrl/devicetree.c b/drivers/pinctrl/devicetree.c
index df1efc2e5202..37069e40af2b 100644
--- a/drivers/pinctrl/devicetree.c
+++ b/drivers/pinctrl/devicetree.c
@@ -220,14 +220,17 @@ int pinctrl_dt_to_map(struct pinctrl *p, struct pinctrl_dev *pctldev)
 	for (state = 0; ; state++) {
 		/* Retrieve the pinctrl-* property */
 		propname = kasprintf(GFP_KERNEL, "pinctrl-%d", state);
-		if (!propname)
-			return -ENOMEM;
+		if (!propname) {
+			ret = -ENOMEM;
+			goto err;
+		}
 		prop = of_find_property(np, propname, &size);
 		kfree(propname);
 		if (!prop) {
 			if (state == 0) {
-				of_node_put(np);
-				return -ENODEV;
+				/* Return -ENODEV if the property 'pinctrl-0' is not present. */
+				ret = -ENODEV;
+				goto err;
 			}
 			break;
 		}
--
2.25.1
Re: [PATCH v2 resend] pinctrl: devicetree: fix refcount leak in pinctrl_dt_to_map()
Posted by Andy Shevchenko 2 weeks, 1 day ago
On Thu, Apr 18, 2024 at 07:58:13PM +0800, Zeng Heng wrote:
> If we fail to allocate propname buffer, we need to drop the reference
> count we just took, otherwise it will lead reference leak. Here the
> error exit path is modified to jump to the err label and call
> pinctrl_dt_free_maps() which would drop the counter.
> 
> In the meantime, if it is found that the property 'pinctrl-0' is not
> present, ENODEV is returned and also jump to the err label and call the
> free function, in case the Smatch tool complains.

...

> ---
> v2: add a comment and modify the commit message, without any logical
>     changes.

You probably didn't get what I was saying...
Linus already applied your version, what is missing is the comment.
This should be done as a followup patch.

-- 
With Best Regards,
Andy Shevchenko