Re: [PATCH net] can: bcm: switch timer to HRTIMER_MODE_SOFT and remove hrtimer_tasklet

[Oliver Hartkopp <socketcan@hartkopp.net>]

On 11.01.22 03:02, Ziyang Xuan (William) wrote:
>> On Mon, Jan 10, 2022 at 09:23:22PM +0800, Ziyang Xuan wrote:
>>> From: Thomas Gleixner <tglx@linutronix.de>
>>>
>>> [ commit bf74aa86e111aa3b2fbb25db37e3a3fab71b5b68 upstream ]
>>>
>>> Stop tx/rx cycle rely on the active state of tasklet and hrtimer
>>> sequentially in bcm_remove_op(), the op object will be freed if they
>>> are all unactive. Assume the hrtimer timeout is short, the hrtimer
>>> cb has been excuted after tasklet conditional judgment which must be
>>> false after last round tasklet_kill() and before condition
>>> hrtimer_active(), it is false when execute to hrtimer_active(). Bug
>>> is triggerd, because the stopping action is end and the op object
>>> will be freed, but the tasklet is scheduled. The resources of the op
>>> object will occur UAF bug.
>>
>> That is not the changelog text of this commit.  Why modify it?
> 
> Above statement is the reason why I want to backport the patch to
> stable tree. Maybe I could give an extra cover-letter to explain
> the details of the problem, but modify the original changelog. Is it?
> 

If you backport the bcm HRTIMER_MODE_SOFT implementation to the 4.19 
stable tree the problem is not fixed for 4.14, 4.4, etc.

HRTIMER_MODE_SOFT has been introduced in 4.16

The issue of a race condition at bcm op removal has already been 
addressed before in commit a06393ed03167 ("can: bcm: fix hrtimer/tasklet 
termination in bcm op removal").

-       hrtimer_cancel(&op->timer);
-       hrtimer_cancel(&op->thrtimer);
-
-       if (op->tsklet.func)
-               tasklet_kill(&op->tsklet);
+       if (op->tsklet.func) {
+               while (test_bit(TASKLET_STATE_SCHED, &op->tsklet.state) ||
+                      test_bit(TASKLET_STATE_RUN, &op->tsklet.state) ||
+                      hrtimer_active(&op->timer)) {
+                       hrtimer_cancel(&op->timer);
+                       tasklet_kill(&op->tsklet);
+               }
+       }

IMO we should better try to improve this fix and enable it for older 
stable trees than fixing only the 4.19.

Best regards,
Oliver



>>
>>>
>>> ----------------------------------------------------------------------
>>>
>>> This patch switches the timer to HRTIMER_MODE_SOFT, which executed the
>>> timer callback in softirq context and removes the hrtimer_tasklet.
>>>
>>> Reported-by: syzbot+652023d5376450cc8516@syzkaller.appspotmail.com
> 
> This is the public problem reporter. Do I need to move it to cover-letter
> but here?
> 
>>> Cc: stable@vger.kernel.org # 4.19
> 
> I want to backport the patch to linux-4.19.y stable tree. How do I need to
> modify?
> 
>>> Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
>>> Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
>>> Acked-by: Oliver Hartkopp <socketcan@hartkopp.net>
>>> Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
>>> Signed-off-by: Ziyang Xuan <william.xuanziyang@huawei.com>
>>> ---
>>>   net/can/bcm.c | 156 +++++++++++++++++---------------------------------
>>>   1 file changed, 52 insertions(+), 104 deletions(-)
>>
>> What stable kernel tree(s) are you wanting this backported to?
>>
>> thanks,
>>
>> greg k-h
>> .
>>
> 
> Thank you for your patient guidance.
>