MT6000 custom build with LuCi and some optimization - kernel 6.6.x

pesa1234 · 2024-12-24T09:53:32.591Z

Done what you suggest but I didn't solve

root@MT6000:~# cat /etc/modules.d/mt7915e 
root@MT6000:~#

It seems that this if doesn't work...

if (request?.args?.enabled == true) {

pesa1234 · 2024-12-24T10:02:22.933Z

Where can I find that patch for rpi4 from mtk?

di_Niko · 2024-12-24T11:26:17.939Z

In few threads i found next info:

https://git01.mediatek.com/plugins/gitiles/openwrt/feeds/mtk-openwrt-feeds/+/c9a7a0b225223282b6167154ccd80a15bb8dfa52

And

https://git01.mediatek.com/plugins/gitiles/openwrt/feeds/mtk-openwrt-feeds/+/fb32b5121a18b2f4fbe5cbe8cb267f388a985538

Idle cpus + good bufferbloat results with hqos enabled reports one of the users in rp4 thread

Extra info:

Scribd

MT7986 Hardware QoS Application Note - 2022-01-20 - v1.0 | PDF | Scheduling...

MT7986_HW_QoS_Application Note_2022-01-20_v1.0

pesa1234 · 2024-12-24T11:31:10.853Z

Hi, in this way I can write the file... but it stay off...

const test = request?.args?.enabled;
				if (test) {
					fw.write('Y\n');
					fm.write('mt7915e wed_enable=1\n');
				} else if (!test) {
					fw.write('N\n');
					fm.write('mt7915e wed_enable=0\n');
				}

edit @systemcrash

by divide the set and get function it is ok, please have a look at the fix. It is still not clean and perfect, but it works.

github.com/openwrt/luci

luci-app-firewall: fix WED setting

committed 03:31PM - 25 Dec 24 UTC

pesa1234

+36 -20

Split setWED and getWED. Signed-off-by: Andrea Pesaresi <andreapesaresi82@gmail….com>

pesa1234 · 2024-12-25T19:13:00.483Z

Sorry, where you get the info that this function is ok for rpi4?

di_Niko · 2024-12-25T20:31:28.410Z

There are two outputs:

This one [Banana BPI-R4] Wifi7 status - #1036 by glassdoor

And

[Banana BPI-R4] Wifi7 status For Developers

all the mtk kernel 6.6 patches apply cleanly on current openwrt 24.10. patches mtk_ppe as per upstream linux and does not use mtk_hnat as per kernel 5.4 from my limited testing, the bridge offload without using bridger works. zero cpu load when i bridge sfp-lan with the switch lan ports. similar to when bridger is used. as for the Qos. from limited testing on waveform buffer bloat test 0 (disable) - A score, D/L latency from +7-12ms 1 (HQos) - A+ on every run, D/lL latency +3ms ( over multi…

Cpu used in mt6000 also has hqos support

pesa1234 · 2024-12-26T08:54:35.548Z

Ahhh it isn't a raspberry pi 4... Okok....

di_Niko · 2024-12-26T09:41:30.002Z

My bad…not sure why i was thinking on rpi4 if you are going to play with these patches, please, let us know, im sure few of us will be interested to test them and compile from your repo

pesa1234 · 2024-12-26T15:06:33.065Z

On my repo... branch next-r4.5.0.mtk have HQoS but at the moment I'm not able to test it...
If somebody would like to test it

From 29/12/2024 I'll test it...

from make menuconfig under Network you will find a new package called mtkhnat_util

Compile it and run it

romanovj · 2024-12-26T15:43:53.112Z

Do we need
https://git01.mediatek.com/plugins/gitiles/openwrt/feeds/mtk-openwrt-feeds/+/refs/heads/master/autobuild/autobuild_5.4_mac80211_release/package/kernel/mt76/patches/0023-wifi-mt76-mt7915-fix-add-ba-issue-on-tid-not-equal-t.patch

after
https://git01.mediatek.com/plugins/gitiles/openwrt/feeds/mtk-openwrt-feeds/+/refs/heads/master/autobuild/unified/filogic/mac80211/24.10/files/target/linux/mediatek/patches-6.6/999-3018-mtk-wed-add-hwpath-wmm-support.patch

Also, I don't think that it could work without the rest of the patches.
Especially 999-2705-net-ethernet-mtk_eth_soc-support-proprietary-debugfs.patch
https://git01.mediatek.com/plugins/gitiles/openwrt/feeds/mtk-openwrt-feeds/+/refs/heads/master/autobuild/unified/global/24.10/files/target/linux/mediatek/patches-6.6/

and files from here https://git01.mediatek.com/plugins/gitiles/openwrt/feeds/mtk-openwrt-feeds/+/refs/heads/master/autobuild/unified/global/24.10/files/target/linux/mediatek/files-6.6/drivers/net/ethernet/mediatek

pesa1234 · 2024-12-26T15:54:32.114Z

Do you check mtk patch present on next-r4.5.0.mtk branch?

Any feedback is appreciated.

di_Niko · 2024-12-26T17:34:37.389Z

Thank you! Compiling...

Nit3H8wk · 2024-12-26T19:09:20.241Z

I tried compiling with hqos and it is failing.
Screenshot 2024-12-26 140650

di_Niko · 2024-12-26T19:23:05.617Z

same issue in my case, probably we are missing something here.

Update, I have same issue even without mtkhnat_util

pesa1234 · 2024-12-26T20:17:31.912Z

I was able to compile...

It is strange....

Can you please send me the diffconfig, in the next days I'll check it.

Please remember that I can not test it until Sunday, so, feedback are welcome as usual.

di_Niko · 2024-12-26T20:50:15.742Z

Could be probably related to this warning:

WARNING: Makefile 'package/kernel/linux/Makefile' has a dependency on 'kmod-nf-conntrack-timeout', which does not exist

i have excluded some nft-conntrackapks which im not using, its compiling right now

Edit: build successful, going to test tomorrow

Edit 2:

The startup and config files are there but probably im missing some dependencies:

Summary

root@OpenWrt:~# sh /sbin/mtkhnat
/sbin/mtkhnat: line 17: can't create /proc/sys/net/bridge/bridge-nf-call-arptables: nonexistent directory
/sbin/mtkhnat: line 18: can't create /proc/sys/net/bridge/bridge-nf-call-iptables: nonexistent directory
/sbin/mtkhnat: line 19: can't create /proc/sys/net/bridge/bridge-nf-call-ip6tables: nonexistent directory
/sbin/mtkhnat: line 20: can't create /proc/sys/net/bridge/bridge-nf-filter-pppoe-tagged: nonexistent directory
/sbin/mtkhnat: line 21: can't create /proc/sys/net/bridge/bridge-nf-filter-vlan-tagged: nonexistent directory
/sbin/mtkhnat: line 22: can't create /proc/sys/net/bridge/bridge-nf-pass-vlan-input-dev: nonexistent directory
/sbin/mtkhnat: line 32: can't create /proc/sys/net/bridge/bridge-nf-call-iptables: nonexistent directory
/sbin/mtkhnat: line 33: can't create /proc/sys/net/bridge/bridge-nf-call-ip6tables: nonexistent directory
Failed to find mtkhnat. Maybe it is a built in module ?
/sbin/mtkhnat: line 77: can't create /sys/kernel/debug/mtk_ppe/qdma_sch0: nonexistent directory
/sbin/mtkhnat: line 77: can't create /sys/kernel/debug/mtk_ppe/qdma_sch1: nonexistent directory
/sbin/mtkhnat: line 81: can't create /proc/sys/net/bridge/bridge-nf-call-iptables: nonexistent directory
/sbin/mtkhnat: line 82: can't create /proc/sys/net/bridge/bridge-nf-call-ip6tables: nonexistent directory
/sbin/mtkhnat: eval: line 118: can't create /sys/kernel/debug/mtk_ppe/qdma_txq0: nonexistent directory
/sbin/mtkhnat: eval: line 118: can't create /sys/kernel/debug/mtk_ppe/qdma_txq1: nonexistent directory
/sbin/mtkhnat: eval: line 118: can't create /sys/kernel/debug/mtk_ppe/qdma_txq2: nonexistent directory
/sbin/mtkhnat: eval: line 118: can't create /sys/kernel/debug/mtk_ppe/qdma_txq3: nonexistent directory
/sbin/mtkhnat: eval: line 118: can't create /sys/kernel/debug/mtk_ppe/qdma_txq4: nonexistent directory
/sbin/mtkhnat: eval: line 118: can't create /sys/kernel/debug/mtk_ppe/qdma_txq5: nonexistent directory
/sbin/mtkhnat: eval: line 118: can't create /sys/kernel/debug/mtk_ppe/qdma_txq6: nonexistent directory
/sbin/mtkhnat: eval: line 118: can't create /sys/kernel/debug/mtk_ppe/qdma_txq7: nonexistent directory
/sbin/mtkhnat: eval: line 118: can't create /sys/kernel/debug/mtk_ppe/qdma_txq8: nonexistent directory
/sbin/mtkhnat: eval: line 118: can't create /sys/kernel/debug/mtk_ppe/qdma_txq9: nonexistent directory
/sbin/mtkhnat: eval: line 118: can't create /sys/kernel/debug/mtk_ppe/qdma_txq10: nonexistent directory
/sbin/mtkhnat: eval: line 118: can't create /sys/kernel/debug/mtk_ppe/qdma_txq11: nonexistent directory
/sbin/mtkhnat: eval: line 118: can't create /sys/kernel/debug/mtk_ppe/qdma_txq12: nonexistent directory
/sbin/mtkhnat: eval: line 118: can't create /sys/kernel/debug/mtk_ppe/qdma_txq13: nonexistent directory
/sbin/mtkhnat: eval: line 118: can't create /sys/kernel/debug/mtk_ppe/qdma_txq14: nonexistent directory
/sbin/mtkhnat: eval: line 118: can't create /sys/kernel/debug/mtk_ppe/qdma_txq15: nonexistent directory

Nit3H8wk · 2024-12-26T23:11:01.899Z

di_Niko:

km

You got further than I did. I can't figure out how to uncheck or remove kmod-nft-conntrackpacks in make menuconfig. If you are able to get hqos working I would like to know how as I would like to remove the need for x86 doing sqm for 1 gig up and down.

romanovj · 2024-12-27T01:55:43.994Z

di_Niko:

The startup and config files are there but probably im missing some dependencies

CONFIG_PACKAGE_kmod-br-netfilter=y

And we also need patches and files for proprietary debugfs MT6000 custom build with LuCi and some optimization - kernel 6.6.x - #1480 by romanovj

pesa1234 · 2024-12-27T22:12:29.953Z

Thanks for feedback. As soon as I come back I'll check it. Thanks

BoltonUK · 2024-12-28T23:40:26.392Z

What’s the latest stable firmware?

I’ve just tried your latest build but unfortunately the wireless is buggy and won’t save any settings