mt76: fix system recovery routine for MT7915

[blocktrron]

This is a draft-PR to fix the longstanding "Message timed out" errors with mt7915 based radios.

I've had positive experience with this patchset over the last 3 weeks and appreciate feedback from communities which frequently face this problem.

Sometimes, the node itself might crash when attempting recovery. I'm not entirely sure where this is happening, but it seems to be somewhere else in the mac80211 stack. The node comes back online by itself, so i still consider this an improvement over the current situation.

Log from a recovered node:

[1029591.814363] br-client: received packet on bat0 with own address as source address (addr:b8:ec:a3:e1:75:4f, vlan:0)
[1029592.855657] br-client: received packet on bat0 with own address as source address (addr:b8:ec:a3:e1:75:4f, vlan:0)
[1029592.856005] br-client: received packet on bat0 with own address as source address (addr:b8:ec:a3:e1:75:4f, vlan:0)
[1031792.254655] mt7915e 0000:02:00.0: Retry message 0000aded (seq 7)
[1031794.334830] mt7915e 0000:02:00.0: Message 0000aded (seq 7) timeout
[1031794.479827] mt7915e 0000:02:00.0: HW/SW Version: 0x8a108a10, Build Time: 20240429200716a
[1031794.479827] 
[1031794.495189] mt7915e 0000:02:00.0: WM Firmware Version: ____000000, Build Time: 20240429200752
[1031794.518127] mt7915e 0000:02:00.0: WA Firmware Version: DEV_000000, Build Time: 20240429200812
[1031800.085604] ieee80211 phy0: Hardware restart was requested
[1031800.085669] ieee80211 phy1: Hardware restart was requested
[1206189.188440] br-client: received packet on bat0 with own address as source address (addr:b8:ec:a3:e1:75:4f, vlan:0)
[1206189.188719] br-client: received packet on bat0 with own address as source address (addr:b8:ec:a3:e1:75:4f, vlan:0)
[1211397.565504] br-client: received packet on bat0 with own address as source address (addr:b8:ec:a3:e1:75:4f, vlan:0)
[1211397.565790] br-client: received packet on bat0 with own address as source address (addr:b8:ec:a3:e1:75:4f, vlan:0)
[1275209.524711] Ignoring NSS change in VHT Operating Mode Notification from 90:9c:4a:ba:e4:a4 with invalid nss 3
[1278619.260856] br-client: port 6(owe1) entered disabled state
[1278619.298741] mt7915e 0000:02:00.0 owe1 (unregistering): left allmulticast mode
[1278619.298791] mt7915e 0000:02:00.0 owe1 (unregistering): left promiscuous mode

[Djfe]

small note for everyone building this:
If you have an existing build folder and don't increase the makefile variable PKG_RELEASE by one, then mt76 won't be rebuilt with the new patches.
https://github.com/openwrt/openwrt/blob/openwrt-24.10/package/kernel/mt76/Makefile#L4

[Djfe]

on d-link dap-x1860, covr-x1860 and tp-link archer ax23 (mt7621) I see
Sun Feb 2 07:46:59 2025 kern.err kernel: [ 24.279424] mt7915e 0000:02:00.0: Retry message 000021ed (seq 11)
early during boot now (early as in: before ntp was able to get a sync). Is this relevant? Or is one pretty much none since it doesn't act on it when it's just happening once?

It doesn't show up on a netgear wax202, but that could also be due to a different wireless config.
it shows up on a netgear wax206 (mt7622)
it doesn't show up on filogic (cudy wr3000, nwa50ax pro, netgear wax220)

[blocktrron]

You can ignore retries w/o timeouts, this is just requiring adjusting message type specific timeouts.

it doesn't show up on filogic (cudy wr3000, nwa50ax pro, netgear wax220)

This PR is not aimed at those.