Re: [PATCH] media: atomisp: Fix resource leak in atomisp_pci_probe()

[Dan Carpenter]

On Mon, Jun 08, 2026 at 04:27:06PM +0800, Dawei Feng wrote:
> During atomisp_pci_probe(), the ISP subdev is initialized via
> atomisp_initialize_modules() prior to entity registration. If
> atomisp_register_entities() fails, the current error path only
> uninitializes the CSI2 modules. This leaks the subdev entity and control
> handler that were previously set up by atomisp_subdev_init().
> 
> Fix this by calling atomisp_subdev_unregister_entities() to properly
> release the subdev state on this specific error path. Later error paths
> remain unchanged, as they correctly use atomisp_unregister_entities() to
> handle broader cleanup after successful registration.
> 
> The bug was first flagged by an experimental analysis tool we are
> developing for kernel memory-management bugs while analyzing v6.13-rc1.
> The tool is still under development and is not yet publicly available.
> Manual inspection confirms that the bug is still present in v7.1-rc5.
> 
> An x86_64 allyesconfig build showed no new warnings. As we do not have an
> Intel Atom ISP platform to test with, no runtime testing was able to be
> performed.
> 
> Fixes: 9d4fa1a16b28 ("media: atomisp: cleanup directory hierarchy")
> Signed-off-by: Zilin Guan <zilin@xxxxxxxxxx>
> Signed-off-by: Dawei Feng <dawei.feng@xxxxxxxxxx>
> ---

The code is buggy, but this isn't the right fix.

Here is generally the standard way to do error handling.
https://staticthinking.wordpress.com/2022/04/28/free-the-last-thing-style/

1. An allocation should clean up it's own partial allocations.  That
should not be handled in the caller.  2.  Every allocation function should
have a mirror cleanup function.

The atomisp_uninitialize_modules() function is just a dummy and was never
actually implemented.  The correct thing is to implement it.

regards,
dan carpenter