[PATCH] net: seeq: Fix use after free vulnerability in ether3 Driver Due to Race Condition

[PATCH] net: seeq: Fix use after free vulnerability in ether3 Driver Due to Race Condition

[Kaixin Wang]

In the ether3_probe function, a timer is initialized with a callback
function ether3_ledoff, bound to &prev(dev)->timer. Once the timer is
started, there is a risk of a race condition if the module or device
is removed, triggering the ether3_remove function to perform cleanup.
The sequence of operations that may lead to a UAF bug is as follows:

CPU0                                    CPU1

                      |  ether3_ledoff
ether3_remove         |
  free_netdev(dev);   |
  put_devic           |
  kfree(dev);         |
 |  ether3_outw(priv(dev)->regs.config2 |= CFG2_CTRLO, REG_CONFIG2);
                      | // use dev

Fix it by ensuring that the timer is canceled before proceeding with
the cleanup in ether3_remove.

Signed-off-by: Kaixin Wang <kxwang23@m.fudan.edu.cn>
---
 drivers/net/ethernet/seeq/ether3.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/net/ethernet/seeq/ether3.c b/drivers/net/ethernet/seeq/ether3.c
index c672f92d65e9..f9d27c9d6808 100644
--- a/drivers/net/ethernet/seeq/ether3.c
+++ b/drivers/net/ethernet/seeq/ether3.c
@@ -850,6 +850,7 @@ static void ether3_remove(struct expansion_card *ec)
 	ecard_set_drvdata(ec, NULL);
 
 	unregister_netdev(dev);
+	del_timer_sync(&priv(dev)->timer);
 	free_netdev(dev);
 	ecard_release_resources(ec);
 }
-- 
2.39.1.windows.1

Re: [PATCH] net: seeq: Fix use after free vulnerability in ether3 Driver Due to Race Condition

[Przemek Kitszel]

On 9/9/24 19:58, Kaixin Wang wrote:
> In the ether3_probe function, a timer is initialized with a callback
> function ether3_ledoff, bound to &prev(dev)->timer. Once the timer is
> started, there is a risk of a race condition if the module or device
> is removed, triggering the ether3_remove function to perform cleanup.
> The sequence of operations that may lead to a UAF bug is as follows:
> 
> CPU0                                    CPU1
> 
>                        |  ether3_ledoff
> ether3_remove         |
>    free_netdev(dev);   |
>    put_devic           |
>    kfree(dev);         |
>   |  ether3_outw(priv(dev)->regs.config2 |= CFG2_CTRLO, REG_CONFIG2);
>                        | // use dev
> 
> Fix it by ensuring that the timer is canceled before proceeding with
> the cleanup in ether3_remove.

this code change indeed prevents UAF bug
but as is, the CFG2_CTRLO flag of REG_CONFIG2 will be left in state "ON"

it would be better to first turn the LED off unconditionally

> 
> Signed-off-by: Kaixin Wang <kxwang23@m.fudan.edu.cn>
> ---
>   drivers/net/ethernet/seeq/ether3.c | 1 +
>   1 file changed, 1 insertion(+)
> 
> diff --git a/drivers/net/ethernet/seeq/ether3.c b/drivers/net/ethernet/seeq/ether3.c
> index c672f92d65e9..f9d27c9d6808 100644
> --- a/drivers/net/ethernet/seeq/ether3.c
> +++ b/drivers/net/ethernet/seeq/ether3.c
> @@ -850,6 +850,7 @@ static void ether3_remove(struct expansion_card *ec)
>   	ecard_set_drvdata(ec, NULL);
>   
>   	unregister_netdev(dev);
> +	del_timer_sync(&priv(dev)->timer);
>   	free_netdev(dev);
>   	ecard_release_resources(ec);
>   }

Re: [PATCH] net: seeq: Fix use after free vulnerability in ether3 Driver Due to Race Condition

[Kaixin Wang]

At 2024-09-11 17:29:44, "Przemek Kitszel" <przemyslaw.kitszel@intel.com> wrote:
>On 9/9/24 19:58, Kaixin Wang wrote:
>> In the ether3_probe function, a timer is initialized with a callback
>> function ether3_ledoff, bound to &prev(dev)->timer. Once the timer is
>> started, there is a risk of a race condition if the module or device
>> is removed, triggering the ether3_remove function to perform cleanup.
>> The sequence of operations that may lead to a UAF bug is as follows:
>> 
>> CPU0                                    CPU1
>> 
>>                        |  ether3_ledoff
>> ether3_remove         |
>>    free_netdev(dev);   |
>>    put_devic           |
>>    kfree(dev);         |
>>   |  ether3_outw(priv(dev)->regs.config2 |= CFG2_CTRLO, REG_CONFIG2);
>>                        | // use dev
>> 
>> Fix it by ensuring that the timer is canceled before proceeding with
>> the cleanup in ether3_remove.
>
>this code change indeed prevents UAF bug
>but as is, the CFG2_CTRLO flag of REG_CONFIG2 will be left in state "ON"
>
>it would be better to first turn the LED off unconditionally
>

I will fix it in the next version of patch.

>> 
>> Signed-off-by: Kaixin Wang <kxwang23@m.fudan.edu.cn>
>> ---
>>   drivers/net/ethernet/seeq/ether3.c | 1 +
>>   1 file changed, 1 insertion(+)
>> 
>> diff --git a/drivers/net/ethernet/seeq/ether3.c b/drivers/net/ethernet/seeq/ether3.c
>> index c672f92d65e9..f9d27c9d6808 100644
>> --- a/drivers/net/ethernet/seeq/ether3.c
>> +++ b/drivers/net/ethernet/seeq/ether3.c
>> @@ -850,6 +850,7 @@ static void ether3_remove(struct expansion_card *ec)
>>   	ecard_set_drvdata(ec, NULL);
>>   
>>   	unregister_netdev(dev);
>> +	del_timer_sync(&priv(dev)->timer);
>>   	free_netdev(dev);
>>   	ecard_release_resources(ec);
>>   }
>
>
Thanks for the review!

Best regards,
Kaixin Wang

Re: [PATCH] net: seeq: Fix use after free vulnerability in ether3 Driver Due to Race Condition

[Simon Horman]

On Tue, Sep 10, 2024 at 01:58:21AM +0800, Kaixin Wang wrote:
> In the ether3_probe function, a timer is initialized with a callback
> function ether3_ledoff, bound to &prev(dev)->timer. Once the timer is
> started, there is a risk of a race condition if the module or device
> is removed, triggering the ether3_remove function to perform cleanup.
> The sequence of operations that may lead to a UAF bug is as follows:
> 
> CPU0                                    CPU1
> 
>                       |  ether3_ledoff
> ether3_remove         |
>   free_netdev(dev);   |
>   put_devic           |
>   kfree(dev);         |
>  |  ether3_outw(priv(dev)->regs.config2 |= CFG2_CTRLO, REG_CONFIG2);
>                       | // use dev
> 
> Fix it by ensuring that the timer is canceled before proceeding with
> the cleanup in ether3_remove.
> 
> Signed-off-by: Kaixin Wang <kxwang23@m.fudan.edu.cn>

This seems like it is a bug fix to me.  If so, it should have a Fixes tag
(immediately above your Signed-off-by line, no blank line in between). And
be targeted at net.

	Subject: [PATCH net] ...

Unless the bug only exists in net-next.

	Subject: [PATCH net-next] ...

Link: https://docs.kernel.org/process/maintainer-netdev.html

...