drivers/firmware/thead,th1520-aon.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
Record the error code before freeing "aon_chan" to avoid a
use after free.
Fixes: e4b3cbd840e5 ("firmware: thead: Add AON firmware protocol driver")
Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
---
drivers/firmware/thead,th1520-aon.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/drivers/firmware/thead,th1520-aon.c b/drivers/firmware/thead,th1520-aon.c
index 4416e9bbf854..38f812ac9920 100644
--- a/drivers/firmware/thead,th1520-aon.c
+++ b/drivers/firmware/thead,th1520-aon.c
@@ -203,6 +203,7 @@ struct th1520_aon_chan *th1520_aon_init(struct device *dev)
{
struct th1520_aon_chan *aon_chan;
struct mbox_client *cl;
+ int ret;
aon_chan = kzalloc(sizeof(*aon_chan), GFP_KERNEL);
if (!aon_chan)
@@ -217,8 +218,9 @@ struct th1520_aon_chan *th1520_aon_init(struct device *dev)
aon_chan->ch = mbox_request_channel_byname(cl, "aon");
if (IS_ERR(aon_chan->ch)) {
dev_err(dev, "Failed to request aon mbox chan\n");
+ ret = PTR_ERR(aon_chan->ch);
kfree(aon_chan);
- return ERR_CAST(aon_chan->ch);
+ return ERR_PTR(ret);
}
mutex_init(&aon_chan->transaction_lock);
--
2.47.2
On Sat, 15 Mar 2025 at 11:04, Dan Carpenter <dan.carpenter@linaro.org> wrote:
>
> Record the error code before freeing "aon_chan" to avoid a
> use after free.
>
> Fixes: e4b3cbd840e5 ("firmware: thead: Add AON firmware protocol driver")
> Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
Applied for next, thanks!
Kind regards
Uffe
> ---
> drivers/firmware/thead,th1520-aon.c | 4 +++-
> 1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/firmware/thead,th1520-aon.c b/drivers/firmware/thead,th1520-aon.c
> index 4416e9bbf854..38f812ac9920 100644
> --- a/drivers/firmware/thead,th1520-aon.c
> +++ b/drivers/firmware/thead,th1520-aon.c
> @@ -203,6 +203,7 @@ struct th1520_aon_chan *th1520_aon_init(struct device *dev)
> {
> struct th1520_aon_chan *aon_chan;
> struct mbox_client *cl;
> + int ret;
>
> aon_chan = kzalloc(sizeof(*aon_chan), GFP_KERNEL);
> if (!aon_chan)
> @@ -217,8 +218,9 @@ struct th1520_aon_chan *th1520_aon_init(struct device *dev)
> aon_chan->ch = mbox_request_channel_byname(cl, "aon");
> if (IS_ERR(aon_chan->ch)) {
> dev_err(dev, "Failed to request aon mbox chan\n");
> + ret = PTR_ERR(aon_chan->ch);
> kfree(aon_chan);
> - return ERR_CAST(aon_chan->ch);
> + return ERR_PTR(ret);
> }
>
> mutex_init(&aon_chan->transaction_lock);
> --
> 2.47.2
>
> Record the error code before freeing "aon_chan" to avoid a
> use after free.
Would it become helpful to mention which selection of source code
analysis tools detected such a questionable implementation detail?
…
> +++ b/drivers/firmware/thead,th1520-aon.c
> @@ -203,6 +203,7 @@ struct th1520_aon_chan *th1520_aon_init(struct device *dev)
> {
> struct th1520_aon_chan *aon_chan;
> struct mbox_client *cl;
> + int ret;
>
> aon_chan = kzalloc(sizeof(*aon_chan), GFP_KERNEL);
> if (!aon_chan)
> @@ -217,8 +218,9 @@ struct th1520_aon_chan *th1520_aon_init(struct device *dev)
> aon_chan->ch = mbox_request_channel_byname(cl, "aon");
> if (IS_ERR(aon_chan->ch)) {
> dev_err(dev, "Failed to request aon mbox chan\n");
> + ret = PTR_ERR(aon_chan->ch);
> kfree(aon_chan);
> - return ERR_CAST(aon_chan->ch);
> + return ERR_PTR(ret);
> }
>
> mutex_init(&aon_chan->transaction_lock);
May the additional variable (for an information) be defined only for
the affected if branch?
Would a smaller scope be more appropriate here?
Regards,
Markus
On Sun, Mar 16, 2025 at 01:05:08PM +0100, Markus Elfring wrote:
> > +++ b/drivers/firmware/thead,th1520-aon.c
> > @@ -203,6 +203,7 @@ struct th1520_aon_chan *th1520_aon_init(struct device *dev)
> > {
> > struct th1520_aon_chan *aon_chan;
> > struct mbox_client *cl;
> > + int ret;
> >
> > aon_chan = kzalloc(sizeof(*aon_chan), GFP_KERNEL);
> > if (!aon_chan)
> > @@ -217,8 +218,9 @@ struct th1520_aon_chan *th1520_aon_init(struct device *dev)
> > aon_chan->ch = mbox_request_channel_byname(cl, "aon");
> > if (IS_ERR(aon_chan->ch)) {
> > dev_err(dev, "Failed to request aon mbox chan\n");
> > + ret = PTR_ERR(aon_chan->ch);
> > kfree(aon_chan);
> > - return ERR_CAST(aon_chan->ch);
> > + return ERR_PTR(ret);
> > }
> >
> > mutex_init(&aon_chan->transaction_lock);
>
> May the additional variable (for an information) be defined only for
> the affected if branch?
> Would a smaller scope be more appropriate here?
There are some variables which should always be at function scope and
"int ret" is one of those.
regards,
dan carpenter
© 2016 - 2025 Red Hat, Inc.