Re: [PATCH 0/2] driver core: platform: fix sw node use-after-free on device release

Next message: Rodrigo Alencar via B4 Relay: "[PATCH v3 01/10] iio: frequency: ad9910: powerdown fix"
Previous message: Alex Elder: "Re: [PATCH] staging: greybus: light: use sysfs_emit() instead of sprintf()"
In reply to: Danilo Krummrich: "Re: (subset) [PATCH 0/2] driver core: platform: fix sw node use-after-free on device release"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

From: Johan Hovold

Date: Tue Nov 03 2026 - 10:06:38 EST

On Mon, Nov 02, 2026 at 12:14:18PM +0100, Johan Hovold wrote:
> Sashiko noted a new use-after-free issue when freeing platform device
> which has a software node as its primary fw node.
>
> Turns out that can cause trouble also when drivers or subsystems remove
> software nodes themselves.

Hmm. The subtleties of the software node implementation got the better
of both the LLM and its operator it seems.

Clearing the secondary node when the device has only a software node
does indeed clear the fwnode pointer so there was never any UAF here.

I've dropped the first patch from the driver core testing branch.

Sorry for the noise.

Johan