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]
Date: Wed, 27 Mar 2024 19:10:25 +0000
From: Simon Horman <horms@...nel.org>
To: Duoming Zhou <duoming@....edu.cn>
Cc: netdev@...r.kernel.org, linux-kernel@...r.kernel.org,
	linux-hams@...r.kernel.org, pabeni@...hat.com, kuba@...nel.org,
	edumazet@...gle.com, davem@...emloft.net, jreuter@...na.de
Subject: Re: [PATCH net] ax25: fix use-after-free bugs caused by
 ax25_ds_del_timer

On Tue, Mar 26, 2024 at 10:25:42PM +0800, Duoming Zhou wrote:
> When the ax25 device is detaching, the ax25_dev_device_down()
> calls ax25_ds_del_timer() to cleanup the slave_timer. When
> the timer handler is running, the ax25_ds_del_timer() that
> calls del_timer() in it will return directly. As a result,
> the use-after-free bugs could happen, one of the scenarios
> is shown below:
> 
>       (Thread 1)          |      (Thread 2)
>                           | ax25_ds_timeout()
> ax25_dev_device_down()    |
>   ax25_ds_del_timer()     |
>     del_timer()           |
>   ax25_dev_put() //FREE   |
>                           |  ax25_dev-> //USE
> 
> In order to mitigate bugs, when the device is detaching, use
> timer_shutdown_sync() to stop the timer.

FWIIW, in my reading of things there is another failure mode whereby
ax25_ds_timeout may rearm the timer after the call to del_timer() but
before the call to ax25_dev_put().

> Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
> Signed-off-by: Duoming Zhou <duoming@....edu.cn>
> ---
>  net/ax25/ax25_ds_timer.c | 7 ++++++-
>  1 file changed, 6 insertions(+), 1 deletion(-)
> 
> diff --git a/net/ax25/ax25_ds_timer.c b/net/ax25/ax25_ds_timer.c
> index c4f8adbf814..5624c0d174c 100644
> --- a/net/ax25/ax25_ds_timer.c
> +++ b/net/ax25/ax25_ds_timer.c
> @@ -43,7 +43,12 @@ void ax25_ds_setup_timer(ax25_dev *ax25_dev)
>  
>  void ax25_ds_del_timer(ax25_dev *ax25_dev)
>  {
> -	if (ax25_dev)
> +	if (!ax25_dev)
> +		return;
> +
> +	if (!ax25_dev->device_up)
> +		timer_shutdown_sync(&ax25_dev->dama.slave_timer);
> +	else
>  		del_timer(&ax25_dev->dama.slave_timer);
>  }

I think that a) it is always correct to call timer_shutdown_sync,
and b) ax25_dev->device_up is always true. So a call to
timer_shutdown_sync can simply replace the call to del_timer.

Also, not strictly related, I think ax25_dev cannot be NULL,
so that check could be dropped. But perhaps that is better left alone.


Zooming out a bit, has removal of ax25 been considered.
I didn't check the logs thoroughly, but I'm not convinced it's been
maintained - other than clean-ups and by-inspection bug fixes - since git
history began.

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ