linux-kernel - Re: [PATCH v2] iio: adc: at91-sama5d2_adc: Fix potential use-after-free in sama5d2

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20251027134908.36d63b9f@jic23-huawei>
Date: Mon, 27 Oct 2025 13:49:08 +0000
From: Jonathan Cameron <jic23@...nel.org>
To: Pei Xiao <xiaopei01@...inos.cn>
Cc: eugen.hristev@...aro.org, alexandre.belloni@...tlin.com,
 lars@...afoo.de, linux-arm-kernel@...ts.infradead.org,
 linux-iio@...r.kernel.org, linux-kernel@...r.kernel.org,
 nicolas.ferre@...rochip.com
Subject: Re: [PATCH v2] iio: adc: at91-sama5d2_adc: Fix potential
 use-after-free in sama5d2_adc driver

On Mon, 20 Oct 2025 15:49:25 +0800
Pei Xiao <xiaopei01@...inos.cn> wrote:

> at91_adc_interrupt can call at91_adc_touch_data_handler function
> to start the work by schedule_work(&st->touch_st.workq).
> 
> If we remove the module which will call at91_adc_remove to
> make cleanup, it will free indio_dev through iio_device_unregister
> while the work mentioned above will be used. The sequence of operations
> that may lead to a UAF bug is as follows:
> 
> CPU0                                      CPU1
> 
>                                      | at91_adc_workq_handler
> at91_adc_remove                      |
> iio_device_unregister(indio_dev)     |
> //free indio_dev                     |
>                                      | iio_push_to_buffers(indio_dev)
>                                      | //use indio_dev

Hi,

I'm not completely following your description here.
The free doesn't happen in iio_device_unregister() but quite a bit later.
So either the problem you are seeing is actually devm_ tear down that
will do the free, or it's a more specific action in iio_device_unregister()
though I'm not sure what it might be. Possibly a specific buffer mask
getting torn down?  I haven't analysed it closely enough to figure out if
there is a race there but it's the only thing I can immediately spot that
would even be of interest to a work item in a driver via some core interfaces.

Other than working out exact cause for anyone looking at this later, I'm
also not sure you don't leave a potential race where a fresh request comes in
between that cancel_work_sync() and the iio_device_unregister() call as it
is only when iio_device_unregister() is complete that all interfaces are torn
down that could start a fresh capture.

So were the cancel_work_sync() one line later I would have been happy but
from your description I'm not sure that fixes the bug you are seeing!

Jonathan

> 
> Fix it by ensuring that the work is canceled before proceeding with
> the cleanup in at91_adc_remove.
> 
> Fixes: 3ec2774f1cc ("iio: adc: at91-sama5d2_adc: add support for position and pressure channels")
> Signed-off-by: Pei Xiao <xiaopei01@...inos.cn>
> ---
>  drivers/iio/adc/at91-sama5d2_adc.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/drivers/iio/adc/at91-sama5d2_adc.c b/drivers/iio/adc/at91-sama5d2_adc.c
> index b4c36e6a7490..1cd6ce61cf17 100644
> --- a/drivers/iio/adc/at91-sama5d2_adc.c
> +++ b/drivers/iio/adc/at91-sama5d2_adc.c
> @@ -2480,6 +2480,7 @@ static void at91_adc_remove(struct platform_device *pdev)
>  	struct iio_dev *indio_dev = platform_get_drvdata(pdev);
>  	struct at91_adc_state *st = iio_priv(indio_dev);
>  
> +	cancel_work_sync(&st->touch_st.workq);
>  	iio_device_unregister(indio_dev);
>  
>  	at91_adc_dma_disable(st);