Re: [PATCH] watchdog: ixp4xx: fix reference leak on platform_device_register() failure

[Guenter Roeck]

On 4/19/26 13:22, Linus Walleij wrote:
> Hi Guangshuo,
>
> thanks for your patch!
>
> On Mon, Apr 13, 2026 at 5:47 PM Guangshuo Li <lgs201920130244@xxxxxxxxx> wrote:
>
> > ixp4xx_timer_probe() directly returns the result of
> > platform_device_register(&ixp4xx_watchdog_device). When registration
> > fails, the embedded struct device in ixp4xx_watchdog_device has already
> > been initialized by device_initialize(), but the failure path does not
> > drop the device reference, leading to a reference leak.
> (...)
>
> > -       return platform_device_register(&ixp4xx_watchdog_device);
> > +       ret = platform_device_register(&ixp4xx_watchdog_device);
> > +       if (ret)
> > +               platform_device_put(&ixp4xx_watchdog_device);
>
> If the problem in the description is indeed there, it seems the bug
> is inside platform_device_register(), surely a function returning an
> error code is supposed to clean up any resources it takes before
> returning an error. It seems wrong to try to fix this in all the
> consumers.

From platform_device_register():

/**
 * platform_device_register - add a platform-level device
 * @pdev: platform device we're adding
 *
 * NOTE: _Never_ directly free @pdev after calling this function, even if it
 * returned an error! Always use platform_device_put() to give up the
 * reference initialised in this function instead.
 */

Not that any code actually does that as far as I can see, but isn't
the above doing exactly what the comment suggests ?

Thanks,
Guenter