From: Johannes Berg <[email protected]>
Thomas explained in https://lore.kernel.org/r/87mtoeb4hb.ffs@tglx
that our handling of the hrtimer here is wrong: If the timer fires
late (e.g. due to vCPU scheduling, as reported by Dmitry/syzbot)
then it tries to actually rearm the timer at the next deadline,
which might be in the past already:
1 2 3 N N+1
| | | ... | |
^ intended to fire here (1)
^ next deadline here (2)
^ actually fired here
The next time it fires, it's later, but will still try to schedule
for the next deadline (now 3), etc. until it catches up with N,
but that might take a long time, causing stalls etc.
Now, all of this is simulation, so we just have to fix it, but
note that the behaviour is wrong even per spec, since there's no
value then in sending all those beacons unaligned - they should be
aligned to the TBTT (1, 2, 3, ... in the picture), and if we're a
bit (or a lot) late, then just resume at that point.
Therefore, change the code to use hrtimer_forward_now() which will
ensure that the next firing of the timer would be at N+1 (in the
picture), i.e. the next interval point after the current time.
Suggested-by: Thomas Gleixner <[email protected]>
Reported-by: Dmitry Vyukov <[email protected]>
Reported-by: [email protected]
Fixes: 01e59e467ecf ("mac80211_hwsim: hrtimer beacon")
Signed-off-by: Johannes Berg <[email protected]>
---
v2: add fixes tag - it's kind of old and the patch won't apply,
but even the original hrtimer code here had this problem
---
drivers/net/wireless/mac80211_hwsim.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/net/wireless/mac80211_hwsim.c b/drivers/net/wireless/mac80211_hwsim.c
index ffa894f7312a..0adae76eb8df 100644
--- a/drivers/net/wireless/mac80211_hwsim.c
+++ b/drivers/net/wireless/mac80211_hwsim.c
@@ -1867,8 +1867,8 @@ mac80211_hwsim_beacon(struct hrtimer *timer)
bcn_int -= data->bcn_delta;
data->bcn_delta = 0;
}
- hrtimer_forward(&data->beacon_timer, hrtimer_get_expires(timer),
- ns_to_ktime(bcn_int * NSEC_PER_USEC));
+ hrtimer_forward_now(&data->beacon_timer,
+ ns_to_ktime(bcn_int * NSEC_PER_USEC));
return HRTIMER_RESTART;
}
--
2.31.1
On Wed, Sep 15 2021 at 11:29, Johannes Berg wrote:
> From: Johannes Berg <[email protected]>
>
> Thomas explained in https://lore.kernel.org/r/87mtoeb4hb.ffs@tglx
> that our handling of the hrtimer here is wrong: If the timer fires
> late (e.g. due to vCPU scheduling, as reported by Dmitry/syzbot)
> then it tries to actually rearm the timer at the next deadline,
> which might be in the past already:
>
> 1 2 3 N N+1
> | | | ... | |
>
> ^ intended to fire here (1)
> ^ next deadline here (2)
> ^ actually fired here
>
> The next time it fires, it's later, but will still try to schedule
> for the next deadline (now 3), etc. until it catches up with N,
> but that might take a long time, causing stalls etc.
>
> Now, all of this is simulation, so we just have to fix it, but
> note that the behaviour is wrong even per spec, since there's no
> value then in sending all those beacons unaligned - they should be
> aligned to the TBTT (1, 2, 3, ... in the picture), and if we're a
> bit (or a lot) late, then just resume at that point.
Well done changelog!
> Therefore, change the code to use hrtimer_forward_now() which will
> ensure that the next firing of the timer would be at N+1 (in the
> picture), i.e. the next interval point after the current time.
>
> Suggested-by: Thomas Gleixner <[email protected]>
> Reported-by: Dmitry Vyukov <[email protected]>
> Reported-by: [email protected]
> Fixes: 01e59e467ecf ("mac80211_hwsim: hrtimer beacon")
> Signed-off-by: Johannes Berg <[email protected]>
Reviewed-by: Thomas Gleixner <[email protected]>