[PATCH net] net_sched: sch_sfq: reject invalid perturb period

[PATCH net] net_sched: sch_sfq: reject invalid perturb period

[Eric Dumazet]

Gerrard Tai reported that SFQ perturb_period has no range check yet,
and this can be used to trigger a race condition fixed in a separate patch.

We want to make sure ctl->perturb_period * HZ will not overflow
and is positive.

Tested:

tc qd add dev lo root sfq perturb -10   # negative value : error
Error: sch_sfq: invalid perturb period.

tc qd add dev lo root sfq perturb 1000000000 # too big : error
Error: sch_sfq: invalid perturb period.

tc qd add dev lo root sfq perturb 2000000 # acceptable value
tc -s -d qd sh dev lo
qdisc sfq 8005: root refcnt 2 limit 127p quantum 64Kb depth 127 flows 128 divisor 1024 perturb 2000000sec
 Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
 backlog 0b 0p requeues 0

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Reported-by: Gerrard Tai <gerrard.tai@starlabs.sg>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: stable@vger.kernel.org
---
 net/sched/sch_sfq.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/net/sched/sch_sfq.c b/net/sched/sch_sfq.c
index 77fa02f2bfcd56a36815199aa2e7987943ea226f..a8cca549b5a2eb2407949560c2b6b658fb7a581f 100644
--- a/net/sched/sch_sfq.c
+++ b/net/sched/sch_sfq.c
@@ -656,6 +656,14 @@ static int sfq_change(struct Qdisc *sch, struct nlattr *opt,
 		NL_SET_ERR_MSG_MOD(extack, "invalid quantum");
 		return -EINVAL;
 	}
+
+	if (ctl->perturb_period < 0 ||
+	    ctl->perturb_period > INT_MAX / HZ) {
+		NL_SET_ERR_MSG_MOD(extack, "invalid perturb period");
+		return -EINVAL;
+	}
+	perturb_period = ctl->perturb_period * HZ;
+
 	if (ctl_v1 && !red_check_params(ctl_v1->qth_min, ctl_v1->qth_max,
 					ctl_v1->Wlog, ctl_v1->Scell_log, NULL))
 		return -EINVAL;
@@ -672,14 +680,12 @@ static int sfq_change(struct Qdisc *sch, struct nlattr *opt,
 	headdrop = q->headdrop;
 	maxdepth = q->maxdepth;
 	maxflows = q->maxflows;
-	perturb_period = q->perturb_period;
 	quantum = q->quantum;
 	flags = q->flags;
 
 	/* update and validate configuration */
 	if (ctl->quantum)
 		quantum = ctl->quantum;
-	perturb_period = ctl->perturb_period * HZ;
 	if (ctl->flows)
 		maxflows = min_t(u32, ctl->flows, SFQ_MAX_FLOWS);
 	if (ctl->divisor) {
-- 
2.50.0.rc0.642.g800a2b2222-goog

Re: [PATCH net] net_sched: sch_sfq: reject invalid perturb period

[patchwork-bot+netdevbpf]

Hello:

This patch was applied to netdev/net.git (main)
by Jakub Kicinski <kuba@kernel.org>:

On Wed, 11 Jun 2025 08:35:01 +0000 you wrote:
> Gerrard Tai reported that SFQ perturb_period has no range check yet,
> and this can be used to trigger a race condition fixed in a separate patch.
> 
> We want to make sure ctl->perturb_period * HZ will not overflow
> and is positive.
> 
> Tested:
> 
> [...]

Here is the summary with links:
  - [net] net_sched: sch_sfq: reject invalid perturb period
    https://git.kernel.org/netdev/net/c/7ca52541c05c

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html

Re: [PATCH net] net_sched: sch_sfq: reject invalid perturb period

[Cong Wang]

On Wed, Jun 11, 2025 at 08:35:01AM +0000, Eric Dumazet wrote:
> Gerrard Tai reported that SFQ perturb_period has no range check yet,
> and this can be used to trigger a race condition fixed in a separate patch.
> 
> We want to make sure ctl->perturb_period * HZ will not overflow
> and is positive.
> 
> Tested:
> 
> tc qd add dev lo root sfq perturb -10   # negative value : error
> Error: sch_sfq: invalid perturb period.
> 
> tc qd add dev lo root sfq perturb 1000000000 # too big : error
> Error: sch_sfq: invalid perturb period.
> 
> tc qd add dev lo root sfq perturb 2000000 # acceptable value
> tc -s -d qd sh dev lo
> qdisc sfq 8005: root refcnt 2 limit 127p quantum 64Kb depth 127 flows 128 divisor 1024 perturb 2000000sec
>  Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
>  backlog 0b 0p requeues 0

Please kindly provide a selftest (as a separate patch) since it looks
fairly easy to reproduce. With AI copilot today, this becomes much
easier, so hopefully it won't bring you much burden. :)

Thanks.