Disable Traffic Control in LEDE

I have LEDE OS with Linux Kernel 4.4.61 running on my router. I want to disable traffic control on my router on a certain interface. When I check the traffic control on my wireless interface, I see that FQ-Codel is enabled by default as following:

qdisc fq_codel 0: dev wlan2 root refcnt 2 limit 10240p flows 1024 quantum 1514 target 5.0ms interval 100.0ms ecn
Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
maxpacket 0 drop_overlimit 0 new_flow_count 0 ecn_mark 0
new_flows_len 0 old_flows_len 0

I want to enable pfifio_fast which is the default AQM in all OSs or even disable traffic control at all in my system. Is that possible in LEDE?

Using pfifo_fast is currently impossible, as it has been completely patched out.

I removed this patch for the 4.12 kernel in my staging tree, in favor of using NET_SCH_DEFAULT, but this option was only added in 4.12.

Is it? As far as I can tell pfifo_fast is exclusive to Linux certainly not all OSs. But may I ask, why do you need pfifo_fast at all?
Technically pfifo_fast and fq_codel are just qdiscs (queueing disciplines) that by them selves do not make an AQM...

This patch affects kernel 4.9, while the OP uses [quote="ha87, post:1, topic:5010"]
Kernel 4.4.61
[/quote]...

That patch also exists for 4.4 and 3.18.

Ah, thanks. That still leaves the question why the OP would prefer pfifo_fast over fq_codel?

I disabled the traffic control more or less by using pfifo only. I think this solution is similar to not compiling traffic control.

But pfifo is just a qdisc, that is not really traffic control... So may I ask, why did you switch to pfifo, or put differently what was wrong with fq_codel?

Best Regards

Codel is a QDisc and it implements an algorithm to tackle buffer bloat problem in the router due to the extra large buffers. pfifo is again QDisc but it does not implement an algorithm to monitor the buffer and the latency of the communication. So by replacing fq_codel which is the default QDisc in LEDE by pfifo I would be able to benchmark the performance with and without AQM algorithms.

Best Regards,
Hany Assasa