mt7915e: joining ac device crashes complete mesh between 2 mt7915e ax devices

[PolynomialDivision]

I have following mesh setup:

       [TP-Link EAP 225 Outdoor (QCA9886 ath10k)]
         / \                             / \
          |                               |
         \ /                             \ /
[Netgear WAX202 (mt7915e)]  <-------> [ZyXEL NWA55AXE (mt7915e)]

Configuration:

• Netgear WAX202: HE20
• ZyXEL NWA55AXE: HE20
• TP-Link EAP 225 Outdoor: VHT20

If I start only the Netgear WAX202 and ZyXEL NWA55AXE everything is fine. Mesh is working and also the bitrate is correct:

$ (iw dev wlan5-mesh station dump && iw dev wlan2-mesh station dump) | grep -E '(Station|(tx|rx) bitrate)'
Station 34:98:b5:0f:37:62 (on wlan5-mesh)
	tx bitrate:	32.4 MBit/s HE-MCS 1 HE-NSS 2 HE-GI 1 HE-DCM 0
	rx bitrate:	68.8 MBit/s HE-MCS 3 HE-NSS 2 HE-GI 0 HE-DCM 0

However, when I start the TP-Link EAP 225 nothing works anymore. Sometimes the complete mesh communication breaks down and sometimes only partial. In all cases, the communication between the two ax devices does not work at all. Only the communication to the TP-Link EAP 225 might be possible.

Station b8:ec:a3:e1:22:05 (on wlan5-mesh)
	tx bitrate:	6.0 MBit/s
	rx bitrate:	68.8 MBit/s HE-MCS 3 HE-NSS 2 HE-GI 0 HE-DCM 0
Station d8:47:32:3f:70:19 (on wlan5-mesh)
	tx bitrate:	28.9 MBit/s VHT-MCS 1 short GI VHT-NSS 2
	rx bitrate:	86.7 MBit/s VHT-MCS 4 short GI VHT-NSS 2

As you can see the tx bitrate: 6.0 MBit/s breaks down.

[DragonBluep]

It seems like that low tx bitrate issue is related to openwrt/openwrt@2f96580

[PolynomialDivision]

After some time the mesh between two ax station again broke. Now it is in a weird state. If I do a ping ff02::1%wlan5-mesh, the request arrive at the other router, but they don't respond to that echo request. However, both can now communicate to the ath10k (tp link eap225) router and this router answers and both router can see the echo reply.

It seems that the mt76 devices just throws away the echo request from the other mt76 device.

[PolynomialDivision]

Here is a situation in which none of the routers answers anymore:

I do a multicast ipv6 ping on the mesh interface. Both ax devices see the ping (on monitor mode interface) but none responds.

You don't need a monitor interface to see the multicast ping on the ax device. However, I wanted to be sure that I see some reply if some station would reply (but they don't).

[PolynomialDivision]

Okay, if I enable mesh_fwding on the mesh interface, it does not crash anymore. It must have something to do with forwarding.

[ryderlee1110]

Do you still observe mt7915 mesh related issue?

[PolynomialDivision]

Do you still observe mt7915 mesh related issue?

Yes. Mesh communication still freezing after some time.

[ryderlee1110]

Could you please upgrade to the latest openwrt?

[PolynomialDivision]

Could you please upgrade to the latest openwrt?

Did not help. :/

[ryderlee1110]

Only happened on AC device?

[PolynomialDivision]

Only happened on AC device?

All devices are freezing. I am not sure if it would also crash if I have 3 ax devices, or some other combination. I need to organize me some other wireless equipment to test a bit further. But especially this 2 x mt7915e and the ath10k is always freezing after some time.

[stweil]

I get crashes in a wireless mesh with 3 NWA55AX. Upgrading to latest OpenWrt / mt76 with gluon's next branch did not fix it. Routers with higher load obviously crash more often. Related issue with more information: freifunk-gluon/gluon#2822.

[PolynomialDivision]

Seems like latest OpenWrt main is fixing the issue. Not sure which commit exactly fixed it. I guess it's one of @nbd168 commits:

• openwrt/openwrt@86ebaef
• openwrt/openwrt@c896358

[Djfe]

@PolynomialDivision This should be closed by now. I'm fairy certain, that this was fixed in/before release OpenWrt 23.05