I'm not sure when it was introduced. I've been trying quite some builds lately. Usually I check the throughput on wireless device, since usually the Ethernet is fine anyway, so it's been a while since I tested the Ethernet upload speed. It worked fine on the 0.6.2 build at least, so I guess I can just bisect between that and master HEAD (with those four commits taken out). Do you have a list of known bad commits so I don't try one that'll brick me?
We had broken builds between commit 8b4cba53a9402f44da2dced4a78ca065b15dca94 and commit 1b311aab3141bbea5667be4517e2599e6235cca0, so avoid building/flashing those as Ethernet will be dead. Even in this case there is no risk what-so-ever as you can just use the RESET button to boot into recovery firmware and then flash again from there, you won't even require the serial console (given you are using the device with U-Boot from UBI installer).
Known to be bad:
781f2f7300 Revert "kernel: backport MT7530 IRQ support"
2c929f8105 util-linux: package ipcs command
3f4301e123 kernel: backport MT7530 IRQ support
8b4cba53a9 kernel: backport MediaTek Ethernet PHY driver
All other commits should work.
1 Like
Yes exactly we are both rt3200 
thanks see you soon for your reeturn
1 Like
we have to consider, that my 3 AP are all situated with at least one wall or floor between, so the signal quality in best case is 55 db, but the weakest mesh connection is at 65 db, so with that said, I'll never expect 700Mbit. The question I initally wanted to answer was high latency and instable connections by @_FailSafe, who asked for someone with a usable 802.11s network, which I have, even I do not expect to win any price in performance races 
Best regards
just finished with the bisect
ee6ba216d8ba1b02154c287e64d709a8bc7b0054 is the first bad commit
commit ee6ba216d8ba1b02154c287e64d709a8bc7b0054
Author: DENG Qingfang <dqfext@gmail.com>
Date: Thu Feb 3 20:07:02 2022 +0800
kernel: backport MT7530 VLAN fix
Fix FDB learning bugs when VLAN filtering is enabled.
Signed-off-by: DENG Qingfang <dqfext@gmail.com>
Tested-by: Arınç ÜNAL <arinc.unal@arinc9.com>
Tested-by: Stijn Tintel <stijn@linux-ipv6.be>
2 Likes
@justinkb Thanks for investing all that effort to find that!
I guess @lga1150 @arinc9 and @stintel may know more about why commit ee6ba216d8ba could cause such a harsh performance regression....?
1 Like
right now I'm building with that backport patch set reverted, should have 100% confirmation in a minute or 5 whether that fixes the issue. as I've done a lot of speedtests during bisects, I've noticed it momentarily hits my max upload (30mbps roughly) for a few milliseconds or so, then drops off to about 10mbps. I would guess some kind of bottlenecking is taking place, based on that.
edit: confirmed
edit2: I also suspect the packet loss issue I mentioned earlier in the thread was caused by this. I can no longer reproduce it. I had stated earlier I thought my isp issues were to blame for that, but I've had it happen since those issues were fixed. Now with this patch set reverted it no longer happens
3 Likes
This post also reports a severe regression with those VLAN patches: MT7621 - 21.02/Master feedback firmware image test - IPV6 offload and disabled Flow Control - #84 by Vladdrako
This feedback was also given to the commit in question itself: https://github.com/openwrt/openwrt/commit/ee6ba216d8ba1b02154c287e64d709a8bc7b0054#commitcomment-66082744
1 Like
Good catch. Wish you posted that yesterday 
save me hours of bisecting
I was able to find it because of your post, because I looked into the commit that caused the regression. I wouldn't have found it otherwise, so thankfully your bisect wasn't wasted
I've tested the patch series on a private mt7621a board of a company I currently work with. The ports are 10/100M so I did not have a chance to test 1000M links. I've extensively tested this device, there is no packet loss and I get full 100M speed on the mt7530 switchports.
I'm not saying there isn't an issue with the commit as there is obviously something wrong. I suggest sending a mail to the openwrt-devel mailing list with Deng on the CC as well as other people from mediatek to make a fuss, so to speak.
2 Likes
Is VLAN filtering on or off?
could you please share your mesh config with me? Also a dump of what packages are installed? Specifically the wifi driver versions are interesting
2 Likes
Whatever the default is
Well, I tested speeds and bufferbloat and it seems that it is almost identical results. No difference when gaming from me. Is there test in specific that you would like me to perform?
Also, could you ask your colleague how is he getting 1gigabit speeds with qosify enabled? Maybe he is not shaping download and only on the upload.
1 Like
hello jimmy they put on their fiber 400 400 they have a fiber 1Go / 600m
two friends of mine said that packets steering wasn't going very well I can't verify, what games did you play? it seems that packet steering on high speeds is not so good but I'm not sure
1 Like
I usually play Overwatch, but I can also play GTA V Online to test or Apex Legends. When they say it doesn't go well for them, you mean they are lagging? I didn't experience any difference to be honest. But I can try some settings or tests if you want.
1 Like
not lagging behind but it sometimes takes bullshits that it shouldn't, can you connect to the same server in a game such as apex legends do you think and don't forget that once you check packets steering to restart the router to take into account I think, 
thank you
Starting yesterday, the 5 GHz/160 MHz channel width suddenly stopped working. I've reset the router, upgraded the snapshot, but nothing changed.
Looking over the logs, I noticed that even though I was using channel 36 [US], the router was trying to use DFS. Switching channel 149 [US], or 149 using several other country codes, resulted in this:
Mon Feb 14 19:32:04 2022 daemon.notice hostapd: Remove interface 'wlan1'
Mon Feb 14 19:32:04 2022 daemon.notice hostapd: wlan1: interface state DISABLED->DISABLED
Mon Feb 14 19:32:04 2022 daemon.notice hostapd: wlan1: AP-DISABLED
Mon Feb 14 19:32:04 2022 daemon.notice hostapd: wlan1: CTRL-EVENT-TERMINATING
Mon Feb 14 19:32:04 2022 daemon.err hostapd: hostapd_free_hapd_data: Interface wlan1 wasn't started
Mon Feb 14 19:32:04 2022 daemon.notice hostapd: nl80211: deinit ifname=wlan1 disabled_11b_rates=0
Mon Feb 14 19:32:04 2022 kern.info kernel: [ 381.409181] device wlan1 left promiscuous mode
Mon Feb 14 19:32:04 2022 kern.info kernel: [ 381.413725] br-lan: port 6(wlan1) entered disabled state
Mon Feb 14 19:32:04 2022 daemon.notice netifd: Wireless device 'radio1' is now down
Mon Feb 14 19:32:05 2022 daemon.notice hostapd: Configuration file: /var/run/hostapd-phy1.conf (phy wlan1) --> new PHY
Mon Feb 14 19:32:05 2022 kern.info kernel: [ 382.859782] br-lan: port 6(wlan1) entered blocking state
Mon Feb 14 19:32:05 2022 kern.info kernel: [ 382.865107] br-lan: port 6(wlan1) entered disabled state
Mon Feb 14 19:32:05 2022 kern.info kernel: [ 382.870717] device wlan1 entered promiscuous mode
Mon Feb 14 19:32:05 2022 kern.info kernel: [ 382.875554] br-lan: port 6(wlan1) entered blocking state
Mon Feb 14 19:32:05 2022 kern.info kernel: [ 382.880963] br-lan: port 6(wlan1) entered forwarding state
Mon Feb 14 19:32:05 2022 daemon.notice hostapd: wlan1: interface state UNINITIALIZED->COUNTRY_UPDATE
Mon Feb 14 19:32:05 2022 daemon.notice hostapd: wlan1: interface state COUNTRY_UPDATE->HT_SCAN
Mon Feb 14 19:32:05 2022 kern.info kernel: [ 382.887406] br-lan: port 6(wlan1) entered disabled state
Mon Feb 14 19:32:06 2022 daemon.notice hostapd: Switch own primary and secondary channel to get secondary channel with no Beacons from other BSSes
Mon Feb 14 19:32:06 2022 daemon.err hostapd: Interface initialization failed
Mon Feb 14 19:32:06 2022 daemon.notice hostapd: wlan1: interface state HT_SCAN->DISABLED
Mon Feb 14 19:32:06 2022 daemon.notice hostapd: wlan1: AP-DISABLED
Mon Feb 14 19:32:06 2022 daemon.notice netifd: Wireless device 'radio1' is now up
Mon Feb 14 19:32:51 2022 daemon.notice hostapd: Remove interface 'wlan1'
Mon Feb 14 19:32:51 2022 daemon.notice hostapd: wlan1: interface state DISABLED->DISABLED
Mon Feb 14 19:32:51 2022 daemon.notice hostapd: wlan1: AP-DISABLED
Mon Feb 14 19:32:51 2022 daemon.notice hostapd: wlan1: CTRL-EVENT-TERMINATING
Mon Feb 14 19:32:51 2022 daemon.err hostapd: hostapd_free_hapd_data: Interface wlan1 wasn't started
Mon Feb 14 19:32:51 2022 daemon.notice hostapd: nl80211: deinit ifname=wlan1 disabled_11b_rates=0
Mon Feb 14 19:32:51 2022 kern.info kernel: [ 428.199036] device wlan1 left promiscuous mode
Mon Feb 14 19:32:51 2022 kern.info kernel: [ 428.203580] br-lan: port 6(wlan1) entered disabled state
Mon Feb 14 19:32:51 2022 daemon.notice netifd: Wireless device 'radio1' is now down
Mon Feb 14 19:32:51 2022 daemon.notice hostapd: Configuration file: /var/run/hostapd-phy1.conf (phy wlan1) --> new PHY
Mon Feb 14 19:32:51 2022 kern.info kernel: [ 428.657183] br-lan: port 6(wlan1) entered blocking state
Mon Feb 14 19:32:51 2022 kern.info kernel: [ 428.662501] br-lan: port 6(wlan1) entered disabled state
Mon Feb 14 19:32:51 2022 kern.info kernel: [ 428.668151] device wlan1 entered promiscuous mode
Mon Feb 14 19:32:51 2022 kern.info kernel: [ 428.673003] br-lan: port 6(wlan1) entered blocking state
Mon Feb 14 19:32:51 2022 kern.info kernel: [ 428.678451] br-lan: port 6(wlan1) entered forwarding state
Mon Feb 14 19:32:51 2022 daemon.notice hostapd: wlan1: interface state UNINITIALIZED->COUNTRY_UPDATE
Mon Feb 14 19:32:51 2022 kern.info kernel: [ 428.684570] br-lan: port 6(wlan1) entered disabled state
Mon Feb 14 19:32:51 2022 daemon.notice hostapd: wlan1: interface state COUNTRY_UPDATE->HT_SCAN
Mon Feb 14 19:32:52 2022 daemon.err hostapd: Interface initialization failed
Mon Feb 14 19:32:52 2022 daemon.notice hostapd: wlan1: interface state HT_SCAN->DISABLED
Mon Feb 14 19:32:52 2022 daemon.notice hostapd: wlan1: AP-DISABLED
Mon Feb 14 19:32:52 2022 daemon.notice netifd: Wireless device 'radio1' is now up
Mon Feb 14 19:33:25 2022 daemon.notice hostapd: Remove interface 'wlan1'
Mon Feb 14 19:33:25 2022 daemon.notice hostapd: wlan1: interface state DISABLED->DISABLED
Mon Feb 14 19:33:25 2022 daemon.notice hostapd: wlan1: AP-DISABLED
Mon Feb 14 19:33:25 2022 daemon.notice hostapd: wlan1: CTRL-EVENT-TERMINATING
Mon Feb 14 19:33:25 2022 daemon.err hostapd: hostapd_free_hapd_data: Interface wlan1 wasn't started
Mon Feb 14 19:33:25 2022 daemon.notice hostapd: nl80211: deinit ifname=wlan1 disabled_11b_rates=0
Mon Feb 14 19:33:25 2022 kern.info kernel: [ 462.896363] device wlan1 left promiscuous mode
Mon Feb 14 19:33:25 2022 kern.info kernel: [ 462.900936] br-lan: port 6(wlan1) entered disabled state
Mon Feb 14 19:33:25 2022 daemon.notice netifd: Wireless device 'radio1' is now down
Changing the channel width to 80 GHz is the thing that worked.
Can anyone recommend a specific test clip for connecting to JTAG? I bought a 20-pin, but it didn't fit because it was blocked by the protruding aluminum piece that covers most of the chips.
1 Like
In my case I got myself a Belkin RT3200, promptly installed OpenWrt SNAPSHOT r18125-b764cb9e5b, and I'm seeing wireless performance issues. I enabled both Software flow offloading and Hardware flow offloading, however I'm pretty disappointed with the performance.
- When my 2 laptops are connected via Ethernet, iperf3 tests from one to the other laptop yield an overall bitrate of 948Mbps
- When one is connected via Ethernet and the other via wifi, iperf3 tests yield an overall bitrate of 275 to 433Mbps|
- When my 2 laptops are connected wirelessly via 5G as per snapshot (one indicated at 2.4Gbps RX / 1.8Gbps and the other 433Mbps RX / 867Mbps TX), iperf3 tests yield an overall bitrate of 139Mbps even though both laptops are sitting right by the RT3200
Am I missing a specific setting?