Re: [PATCH] PNP: Detach device after resource transition failure
From: Rafael J. Wysocki (Intel)
Date: Mon Sep 14 2026 - 14:24:39 EST
On Thu, Sep 10, 2026 at 10:37 PM Myeonghun Pak <mhun512@xxxxxxxxx> wrote:
>
> pnp_device_probe() attaches the PNP device before it activates or
> disables its resources. The attach changes the status from PNP_READY
> to PNP_ATTACHED, but errors from either resource transition return
> directly and leave that status behind. A later bind or manual PNP
> configuration attempt then sees a device that is still marked in use.
>
> Route both errors through the existing failure path so it restores the
> status to PNP_READY. Do not disable the device or clean its resources:
> pnp_activate_dev() only marks it active after a successful start, while
> pnp_disable_dev() preserves the active state and resource table when
> stopping fails.
>
> This issue was identified during our ongoing static-analysis research while
> reviewing kernel code.
>
> Co-developed-by: Ijae Kim <ae878000@xxxxxxxxx>
> Signed-off-by: Ijae Kim <ae878000@xxxxxxxxx>
> Signed-off-by: Myeonghun Pak <mhun512@xxxxxxxxx>
> ---
> drivers/pnp/driver.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/pnp/driver.c b/drivers/pnp/driver.c
> index 6d58b1465081..80c2a0f15b80 100644
> --- a/drivers/pnp/driver.c
> +++ b/drivers/pnp/driver.c
> @@ -96,13 +96,13 @@ static int pnp_device_probe(struct device *dev)
> if (!(pnp_drv->flags & PNP_DRIVER_RES_DO_NOT_CHANGE)) {
> error = pnp_activate_dev(pnp_dev);
> if (error < 0)
> - return error;
> + goto fail;
> }
> } else if ((pnp_drv->flags & PNP_DRIVER_RES_DISABLE)
> == PNP_DRIVER_RES_DISABLE) {
> error = pnp_disable_dev(pnp_dev);
> if (error < 0)
> - return error;
> + goto fail;
> }
> error = 0;
> if (pnp_drv->probe) {
> --
Applied as 7.4 material, thanks!