2021-12-10 09:37:41

by Jiapeng Chong

Subject: [PATCH v2] staging: r8188eu: Use strndup_user instead of kmalloc/copy_from_user

This code assumes that the user is going to give us a NULL terminated
string which is not necessarily true. The original code was buggy too.

Fix following coccicheck warning:

./drivers/staging/r8188eu/os_dep/ioctl_linux.c:4253:8-15: WARNING
opportunity for memdup_user.

Reported-by: Abaci Robot <[email protected]>
Fixes: 7bdedfef085b ("staging: r8188eu: Remove mp, a.k.a. manufacturing process, code")
Signed-off-by: Jiapeng Chong <[email protected]>
---
Changes in v2:
-For the follow advice:https://lore.kernel.org/lkml/20211210083010.GM1956@kadam/

drivers/staging/r8188eu/os_dep/ioctl_linux.c | 13 ++++---------
1 file changed, 4 insertions(+), 9 deletions(-)

diff --git a/drivers/staging/r8188eu/os_dep/ioctl_linux.c b/drivers/staging/r8188eu/os_dep/ioctl_linux.c
index 56adfe4087a8..5d26dffe5c0d 100644
--- a/drivers/staging/r8188eu/os_dep/ioctl_linux.c
+++ b/drivers/staging/r8188eu/os_dep/ioctl_linux.c
@@ -4250,17 +4250,12 @@ static int rtw_test(
DBG_88E("+%s\n", __func__);
len = wrqu->data.length;

- pbuf = kzalloc(len, GFP_KERNEL);
- if (!pbuf) {
- DBG_88E("%s: no memory!\n", __func__);
- return -ENOMEM;
- }
-
- if (copy_from_user(pbuf, wrqu->data.pointer, len)) {
- kfree(pbuf);
+ pbuf = strndup_user(wrqu->data.pointer, len);
+ if (IS_ERR(pbuf)) {
DBG_88E("%s: copy from user fail!\n", __func__);
- return -EFAULT;
+ return PTR_ERR(pbuf);
}
+
DBG_88E("%s: string =\"%s\"\n", __func__, pbuf);

ptmp = (char *)pbuf;
--
2.20.1.7.g153144c

2021-12-10 10:05:30

by Dan Carpenter

Subject: Re: [PATCH v2] staging: r8188eu: Use strndup_user instead of kmalloc/copy_from_user

On Fri, Dec 10, 2021 at 05:37:01PM +0800, Jiapeng Chong wrote:
> This code assumes that the user is going to give us a NULL terminated

I wouldn't comment on this except that the Fixes tag is wrong so it
needs to be resent anyway. NULL is a pointer. When it's a character
'\0'; then it's just one L "NUL terminated".

> string which is not necessarily true. The original code was buggy too.
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Leave this out.

>
> Fix following coccicheck warning:
>
> ./drivers/staging/r8188eu/os_dep/ioctl_linux.c:4253:8-15: WARNING
> opportunity for memdup_user.
>
> Reported-by: Abaci Robot <[email protected]>
> Fixes: 7bdedfef085b ("staging: r8188eu: Remove mp, a.k.a. manufacturing process, code")

The fixes tag should be:

Fixes: 2b42bd58b321 ("staging: r8188eu: introduce new os_dep dir for RTL8188eu driver")


> Signed-off-by: Jiapeng Chong <[email protected]>
> ---
> Changes in v2:
> -For the follow advice:https://lore.kernel.org/lkml/20211210083010.GM1956@kadam/

Don't put a link in the email. Just say what the difference is. No
one wants to have to click on a link.

>
> drivers/staging/r8188eu/os_dep/ioctl_linux.c | 13 ++++---------
> 1 file changed, 4 insertions(+), 9 deletions(-)
>
> diff --git a/drivers/staging/r8188eu/os_dep/ioctl_linux.c b/drivers/staging/r8188eu/os_dep/ioctl_linux.c
> index 56adfe4087a8..5d26dffe5c0d 100644
> --- a/drivers/staging/r8188eu/os_dep/ioctl_linux.c
> +++ b/drivers/staging/r8188eu/os_dep/ioctl_linux.c
> @@ -4250,17 +4250,12 @@ static int rtw_test(
> DBG_88E("+%s\n", __func__);
> len = wrqu->data.length;
>
> - pbuf = kzalloc(len, GFP_KERNEL);
> - if (!pbuf) {
> - DBG_88E("%s: no memory!\n", __func__);
> - return -ENOMEM;
> - }
> -
> - if (copy_from_user(pbuf, wrqu->data.pointer, len)) {
> - kfree(pbuf);
> + pbuf = strndup_user(wrqu->data.pointer, len);
> + if (IS_ERR(pbuf)) {
> DBG_88E("%s: copy from user fail!\n", __func__);

This printk is out of date now that we're not using the copy_from_user()
function. It doesn't add useful information. Just delete it.

regards,
dan carpenter