On Wed, 29 Aug 2018, Gustavo A. R. Silva wrote:

> There is a NULL pointer dereference in case memory resources
> for *parse* are not successfully allocated.
> 
> Fix this by adding a new goto label and make the execution
> path jump to it in case vzalloc() fails.
> 
> Addresses-Coverity-ID: 1473081 ("Dereference after null check")
> Fixes: b2dd9f2e5a8a ("HID: core: fix memory leak on probe")
> Signed-off-by: Gustavo A. R. Silva <gust...@embeddedor.com>
> ---
>  drivers/hid/hid-core.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/hid/hid-core.c b/drivers/hid/hid-core.c
> index 4548dae..5bec924 100644
> --- a/drivers/hid/hid-core.c
> +++ b/drivers/hid/hid-core.c
> @@ -1000,7 +1000,7 @@ int hid_open_report(struct hid_device *device)
>       parser = vzalloc(sizeof(struct hid_parser));
>       if (!parser) {
>               ret = -ENOMEM;
> -             goto err;
> +             goto alloc_err;
>       }
>  
>       parser->device = device;
> @@ -1049,6 +1049,7 @@ int hid_open_report(struct hid_device *device)
>       hid_err(device, "item fetching failed at offset %d\n", (int)(end - 
> start));
>  err:
>       kfree(parser->collection_stack);
> +alloc_err:
>       vfree(parser);
>       hid_close_report(device);
>       return ret;

Queued in for-4.19/fixes. Thanks,

-- 
Jiri Kosina
SUSE Labs

Reply via email to