MT76: MT7613BE crashs when multiple stations

kpoman · June 11, 2021, 2:29pm

Hello everyone,
I am testing a snapshot compatible with my device (TP Link Archer C6 V3.2). Everything works well, but as soon as more than 1 station connects on 5GHz, it crashs the connection and only way to get it back is rebooting the router. The device seems to still work as it is able to do some iw wlan1 scan, etc. I suppose maybe some daemon behind it, which handles new connections and advertisement of the SSID, does crash, I am speculating (I tried restarting manually hostapd without luck).
Seems like it enters some kind of loop as can be seen by below logs:

Fri Jun 11 10:50:19 2021 daemon.info hostapd: wlan1: STA a4:34:d9:29:e9:56 RADIUS: starting accounting session 5993C3031F05B1A8
Fri Jun 11 10:50:19 2021 daemon.info hostapd: wlan1: STA a4:34:d9:29:e9:56 WPA: pairwise key handshake completed (RSN)
Fri Jun 11 10:50:19 2021 daemon.notice hostapd: wlan1: EAPOL-4WAY-HS-COMPLETED a4:34:d9:29:e9:56
Fri Jun 11 10:50:19 2021 daemon.notice hostapd: wlan1: AP-STA-DISCONNECTED a4:34:d9:29:e9:56
Fri Jun 11 10:50:22 2021 daemon.info hostapd: wlan1: STA a4:34:d9:29:e9:56 IEEE 802.11: authenticated
Fri Jun 11 10:50:22 2021 daemon.info hostapd: wlan1: STA a4:34:d9:29:e9:56 IEEE 802.11: associated (aid 2)
Fri Jun 11 10:50:22 2021 daemon.notice hostapd: wlan1: AP-STA-CONNECTED a4:34:d9:29:e9:56
Fri Jun 11 10:50:22 2021 daemon.info hostapd: wlan1: STA a4:34:d9:29:e9:56 RADIUS: starting accounting session C5BF4BBE4E312C73
Fri Jun 11 10:50:22 2021 daemon.info hostapd: wlan1: STA a4:34:d9:29:e9:56 WPA: pairwise key handshake completed (RSN)
Fri Jun 11 10:50:22 2021 daemon.notice hostapd: wlan1: EAPOL-4WAY-HS-COMPLETED a4:34:d9:29:e9:56
Fri Jun 11 10:50:23 2021 daemon.notice hostapd: wlan1: AP-STA-DISCONNECTED a4:34:d9:29:e9:56
Fri Jun 11 10:50:24 2021 daemon.info hostapd: wlan1: STA a4:34:d9:29:e9:56 IEEE 802.11: authenticated
Fri Jun 11 10:50:24 2021 daemon.info hostapd: wlan1: STA a4:34:d9:29:e9:56 IEEE 802.11: associated (aid 2)
Fri Jun 11 10:50:25 2021 daemon.notice hostapd: wlan1: AP-STA-CONNECTED a4:34:d9:29:e9:56
Fri Jun 11 10:50:25 2021 daemon.info hostapd: wlan1: STA a4:34:d9:29:e9:56 RADIUS: starting accounting session 7B03AADDC39AD569
Fri Jun 11 10:50:25 2021 daemon.info hostapd: wlan1: STA a4:34:d9:29:e9:56 WPA: pairwise key handshake completed (RSN)
Fri Jun 11 10:50:25 2021 daemon.notice hostapd: wlan1: EAPOL-4WAY-HS-COMPLETED a4:34:d9:29:e9:56
Fri Jun 11 10:50:25 2021 daemon.notice hostapd: wlan1: AP-STA-DISCONNECTED a4:34:d9:29:e9:56
Fri Jun 11 10:50:27 2021 daemon.info hostapd: wlan1: STA a4:34:d9:29:e9:56 IEEE 802.11: authenticated
Fri Jun 11 10:50:27 2021 daemon.info hostapd: wlan1: STA a4:34:d9:29:e9:56 IEEE 802.11: associated (aid 2)
Fri Jun 11 10:50:27 2021 daemon.notice hostapd: wlan1: AP-STA-CONNECTED a4:34:d9:29:e9:56
Fri Jun 11 10:50:27 2021 daemon.info hostapd: wlan1: STA a4:34:d9:29:e9:56 RADIUS: starting accounting session 0FDC9DD9A83822F2
Fri Jun 11 10:50:27 2021 daemon.info hostapd: wlan1: STA a4:34:d9:29:e9:56 WPA: pairwise key handshake completed (RSN)
Fri Jun 11 10:50:27 2021 daemon.notice hostapd: wlan1: EAPOL-4WAY-HS-COMPLETED a4:34:d9:29:e9:56
Fri Jun 11 10:50:27 2021 daemon.notice hostapd: wlan1: AP-STA-DISCONNECTED a4:34:d9:29:e9:56
Fri Jun 11 10:50:29 2021 daemon.info hostapd: wlan1: STA a4:34:d9:29:e9:56 IEEE 802.11: authenticated
Fri Jun 11 10:50:29 2021 daemon.info hostapd: wlan1: STA a4:34:d9:29:e9:56 IEEE 802.11: associated (aid 2)
Fri Jun 11 10:50:29 2021 daemon.notice hostapd: wlan1: AP-STA-CONNECTED a4:34:d9:29:e9:56
Fri Jun 11 10:50:29 2021 daemon.info hostapd: wlan1: STA a4:34:d9:29:e9:56 RADIUS: starting accounting session 1DF8145CA24CA839
Fri Jun 11 10:50:29 2021 daemon.info hostapd: wlan1: STA a4:34:d9:29:e9:56 WPA: pairwise key handshake completed (RSN)
Fri Jun 11 10:50:29 2021 daemon.notice hostapd: wlan1: EAPOL-4WAY-HS-COMPLETED a4:34:d9:29:e9:56
Fri Jun 11 10:50:30 2021 daemon.notice hostapd: wlan1: AP-STA-DISCONNECTED a4:34:d9:29:e9:56
Fri Jun 11 10:50:31 2021 daemon.info hostapd: wlan1: STA a4:34:d9:29:e9:56 IEEE 802.11: authenticated
Fri Jun 11 10:50:31 2021 daemon.info hostapd: wlan1: STA a4:34:d9:29:e9:56 IEEE 802.11: associated (aid 2)

Tried turning off and on that radio, still no luck:

root@openwrt-c6-v3:~# wifi down radio1
root@openwrt-c6-v3:~# wifi up radio1
root@openwrt-c6-v3:~# logread | tail -n 20
Fri Jun 11 11:13:14 2021 daemon.notice hostapd: Remove interface 'wlan1'
Fri Jun 11 11:13:14 2021 daemon.notice hostapd: wlan1: interface state ENABLED->DISABLED
Fri Jun 11 11:13:14 2021 daemon.notice hostapd: wlan1: AP-DISABLED
Fri Jun 11 11:13:14 2021 daemon.notice hostapd: wlan1: CTRL-EVENT-TERMINATING
Fri Jun 11 11:13:14 2021 daemon.notice hostapd: nl80211: deinit ifname=wlan1 disabled_11b_rates=0
Fri Jun 11 11:13:14 2021 kern.info kernel: [ 2760.451371] device wlan1 left promiscuous mode
Fri Jun 11 11:13:14 2021 kern.info kernel: [ 2760.456095] br-lan: port 6(wlan1) entered disabled state
Fri Jun 11 11:13:14 2021 daemon.notice netifd: Network device 'wlan1' link is down
Fri Jun 11 11:13:28 2021 daemon.notice hostapd: Configuration file: /var/run/hostapd-phy1.conf (phy wlan1) --> new PHY
Fri Jun 11 11:13:28 2021 kern.info kernel: [ 2775.196795] br-lan: port 6(wlan1) entered blocking state
Fri Jun 11 11:13:28 2021 kern.info kernel: [ 2775.202163] br-lan: port 6(wlan1) entered disabled state
Fri Jun 11 11:13:28 2021 kern.info kernel: [ 2775.208259] device wlan1 entered promiscuous mode
Fri Jun 11 11:13:28 2021 daemon.notice hostapd: wlan1: interface state UNINITIALIZED->COUNTRY_UPDATE
Fri Jun 11 11:13:28 2021 daemon.notice hostapd: wlan1: interface state COUNTRY_UPDATE->HT_SCAN
Fri Jun 11 11:13:29 2021 kern.info kernel: [ 2775.957760] IPv6: ADDRCONF(NETDEV_CHANGE): wlan1: link becomes ready
Fri Jun 11 11:13:29 2021 kern.info kernel: [ 2775.964482] br-lan: port 6(wlan1) entered blocking state
Fri Jun 11 11:13:29 2021 kern.info kernel: [ 2775.969797] br-lan: port 6(wlan1) entered forwarding state
Fri Jun 11 11:13:29 2021 daemon.notice hostapd: wlan1: interface state HT_SCAN->ENABLED
Fri Jun 11 11:13:29 2021 daemon.notice hostapd: wlan1: AP-ENABLED
Fri Jun 11 11:13:30 2021 daemon.notice netifd: Network device 'wlan1' link is up
root@openwrt-c6-v3:~#

At the beginning, the peer trying to connect even appears via Luci and gets kicked of the SSID, then after some trials, no more advertisement or anything.

123serge123 · June 11, 2021, 4:57pm

It's a known problem
But there exist working workaround

intdev · September 9, 2021, 5:13am

I bought Linksys E5600 and tested 5GHz WiFi on OpenWrt.
5GHz WLAN is MT7613E. (Same as TP-Link Archer C6 / A6 / EAP235)

21.02.0-rc3: Initial support build of E5600. 5GHz is very unstable. Rapidly disconnected, connection fails, weird behavior occurs.

21.02.0-rc4: In despite of short period testing, at least it seems pretty usable.

21.02.0(Final Release): Same as 21.02.0-rc3.

Recent Snapshot(20210908): No rapid disconnection. No connection fail. Web browsing is okay. But, under heavy WiFi load(iperf3 multi, Speedtest.net Multi speed test), 5GHz WiFi suddenly dies. No more WiFi works. And LuCi response is getting very slow. But CPU usage is idle. (I didn't check memory usage.) Only resolution is rebooting the router. ('reboot' command via CLI also slow.)

I searched OpenWrt forum and found MT7613E (MediaTek 5GHz WLAN Chip) devices have problems like me.

My question is:

RC3 was bad. RC4 is OK. Maybe it's fixed? If yes, what was root cause? And what packages,codes are changed? Is there anybody knows this? Desperate!
Why this problems respawned to final release? (RC3 (Not OK) -> RC4 (OK) -> Final Release(Not OK)
It's headache!

I confirmed mt76 drivers,firmwares are different(each builds).

My wish to TP-Link Archer C6 / A6 / EAP235 users:

Could you test RC3(optional), RC4(essential), Final Release(essential ?), Recent Snapshot(essential) and share 5Ghz WiFi stability test results to here?

Thanks.

intdev · September 9, 2021, 11:39am

ADD:

In rc4 firmware, If I change WiFi settings few times, then WiFi connection goes unstable state. Reboot required. Very weird. RC4 also not perfect. But until/unless changing setting, it works well.

Moreover, I reflashed firmware to 21.02.0 Final Release once again, setting WiFi and checked poor WiFi state. But, after reboot, 21.02.0 Final also works well!

To sum it up,

RC4 ~ Final Release: 5GHz WiFi is OK. Unless touching the wireless settings.
Recent Snapshot: Definitively, there is another problem exists.

I'll keep going testing.

dsouza · November 6, 2021, 1:23pm

Hi, just a follow-up. I've been using 4 Archer C6 v3.2 (one as router + 3 as access points) and while sometime ago it had some issues with 5Ghz stability, in the latest snapshot builds it improved a lot.

While I still have a crontab task to reboot every night, the overall 5Ghz stability has really improved. I use to do my own builds, and now running a build from Oct 31st (r17900-e06aaba4e3) and it seems very usable.

elcovv · November 25, 2021, 11:34am

A possible fix has been made for this issue and the developer has asked to test.

Does anyone know how to build a snapshot using the latest mt76 feed?

elcovv · November 25, 2021, 7:37pm

I got the answers on how to build the latest mt76 feed in the issue thread from here onwards.