[PATCH] siox: fix possible memory leak in siox_device_add()

Yang Yingliang posted 1 patch 3 years, 5 months ago
drivers/siox/siox-core.c | 2 ++
1 file changed, 2 insertions(+)
[PATCH] siox: fix possible memory leak in siox_device_add()
Posted by Yang Yingliang 3 years, 5 months ago
If device_register() returns error in siox_device_add(),
the name allocated by dev_set_name() need be freed. As
comment of device_register() says, it should use put_device()
to give up the reference in the error path. So fix this
by calling put_device(), then the name can be freed in
kobject_cleanup(), and sdevice is freed in siox_device_release(),
set it to null in error path.

Fixes: bbecb07fa0af ("siox: new driver framework for eckelmann SIOX")
Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
---
 drivers/siox/siox-core.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/siox/siox-core.c b/drivers/siox/siox-core.c
index 7c4f32d76966..561408583b2b 100644
--- a/drivers/siox/siox-core.c
+++ b/drivers/siox/siox-core.c
@@ -839,6 +839,8 @@ static struct siox_device *siox_device_add(struct siox_master *smaster,
 
 err_device_register:
 	/* don't care to make the buffer smaller again */
+	put_device(&sdevice->dev);
+	sdevice = NULL;
 
 err_buf_alloc:
 	siox_master_unlock(smaster);
-- 
2.25.1
Re: [PATCH] siox: fix possible memory leak in siox_device_add()
Posted by Uwe Kleine-König 3 years, 5 months ago
Hello,

On Fri, Nov 04, 2022 at 10:13:34AM +0800, Yang Yingliang wrote:
> If device_register() returns error in siox_device_add(),
> the name allocated by dev_set_name() need be freed. As
> comment of device_register() says, it should use put_device()
> to give up the reference in the error path. So fix this
> by calling put_device(), then the name can be freed in
> kobject_cleanup(), and sdevice is freed in siox_device_release(),
> set it to null in error path.
> 
> Fixes: bbecb07fa0af ("siox: new driver framework for eckelmann SIOX")
> Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
> ---
>  drivers/siox/siox-core.c | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/drivers/siox/siox-core.c b/drivers/siox/siox-core.c
> index 7c4f32d76966..561408583b2b 100644
> --- a/drivers/siox/siox-core.c
> +++ b/drivers/siox/siox-core.c
> @@ -839,6 +839,8 @@ static struct siox_device *siox_device_add(struct siox_master *smaster,
>  
>  err_device_register:
>  	/* don't care to make the buffer smaller again */
> +	put_device(&sdevice->dev);
> +	sdevice = NULL;

The problem is real and the fix works as intended. I don't like the
necessity to set sdevice to NULL, but that's the easiest approach here,
so:

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

>  err_buf_alloc:
>  	siox_master_unlock(smaster);

@gregkh: Could you pick up this patch please[1]? While I don't consider it
very urgent, it's technically a fix. I don't care much if it goes in now
or during the next merge window.

Best regards
Uwe

[1] the original Message-Id is
    20221104021334.618189-1-yangyingliang@huawei.com. Maybe this helps
    you to apply the patch even if you didn't get a Cc.

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