Similar to device_for_each_child, device_find_child should not ooops
if parent->p is NULL. While fuzzing the Bluetooth subsystem, I got a null
pointer dereference in next_device, which is prevented by this change.
Buglink: https://bugzilla.kernel.org/show_bug.cgi?id=215497
Signed-off-by: Soenke Huster <[email protected]>
---
I found this while fuzzing the bluetooth subsystem and reported this bug
in bugzilla. As I feel more like a kernelnewbie, I am not sure whether
this needs to be patched in driver core, or is just an issue of the
bluetooth subsystem itself. If the latter is the case: sorry for the
spam!
A similar fix was applied in 014c90dbb9b6 ("driver core: prevent
device_for_each_child from oopsing") for device_for_each_child, which is
mentioned in the docstring of device_find_child as similar function.
drivers/base/core.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/base/core.c b/drivers/base/core.c
index 7bb957b11861..654d904631f9 100644
--- a/drivers/base/core.c
+++ b/drivers/base/core.c
@@ -3780,7 +3780,7 @@ struct device *device_find_child(struct device *parent, void *data,
struct klist_iter i;
struct device *child;
- if (!parent)
+ if (!parent || !parent->p)
return NULL;
klist_iter_init(&parent->p->klist_children, &i);
--
2.34.1
On Thu, Jan 20, 2022 at 04:02:46PM +0100, Soenke Huster wrote:
> Similar to device_for_each_child, device_find_child should not ooops
> if parent->p is NULL. While fuzzing the Bluetooth subsystem, I got a null
> pointer dereference in next_device, which is prevented by this change.
>
> Buglink: https://bugzilla.kernel.org/show_bug.cgi?id=215497
> Signed-off-by: Soenke Huster <[email protected]>
> ---
> I found this while fuzzing the bluetooth subsystem and reported this bug
> in bugzilla. As I feel more like a kernelnewbie, I am not sure whether
> this needs to be patched in driver core, or is just an issue of the
> bluetooth subsystem itself. If the latter is the case: sorry for the
> spam!
bluetooth should be fixed, this is an indication that something is using
a stale pointer and this would be the least of the problems here.
thanks,
greg k-h