On 10/28/24 9:58 PM, yiyang (D) wrote:
On 2024/10/28 20:46, Alex Elder wrote:
On 10/26/24 5:50 AM, Dan Carpenter wrote:
On Sat, Oct 26, 2024 at 08:11:53AM +0000, Yi Yang wrote:
The gb_audio_manager_get_module() is EXPORT_SYMBOL, and will return NULL when incoming parameter id < 0, fix possible null-ptr-deref by add check for return value.
Fixes: 8db00736d365 ("greybus: audio: Add Audio Manager") Signed-off-by: Yi Yang yiyang13@huawei.com
Where is gb_audio_manager_get_module() called from? So far as I can see it's never used. Why not just delete it?
regards, dan carpenter
I agree with this. I suspected all callers might have guaranteed that the "id" value passed would be always valid, but... there are no callers.
It is a simple function, and could be added back again if it is needed in the future (possibly even by reverting the commit that removes it).
If you do this, please remove gb_audio_put_module() in the same patch. It too has no callers.
Thank you.
-Alex
.
I tried to find the caller before I modified it, but unfortunately I didn't find the caller, so I suspect some non-kernel driver code will try to call this functions.
I just found this problem while reading the code and I'm not sure if I should remove the function.
The Linux kernel is a unified whole. This means that if you have the current version of the kernel source code, it will incorporate *all* possible callers of a given function (or more generally, references to or users of a given symbol). The only possible exception might be out-of-tree code, but that is not our concern.
If you can't find any callers, none exist. It's fine to propose removing the function. And if the maintainer accepts it, it's a good change.
-Alex
regards, Yiyang