ramips-mt7621: dap-x1860 does not report airtime channel utilization

[maurerle]

Bug report

What is the problem?

The airtime graph shows inconsistent behavior, I did not see any other or user-affecting issue other than the wrong reporting.
Weird thing is, that this does only affect the ramips-mt7621 device dlink-dap-x1860 but not the wsm20 or the covr-x1860 which is about the same platform with the same (?) wifi chip.
https://grafana.ffac.rocks/d/000000002/node?orgId=1&var-node=a0a3f0489b70&from=1725290536782&to=1725294746982&var-select_hostname=ffac-neupforte-29-kueche&var-hostname=ffac-neupforte-29-kueche&var-saveinterval=1m&var-nodetolink=001a8cc101dd&viewPanel=5

What is the expected behaviour?
show correct graph output like before

Gluon Version:
Bisect log:

• v2023.2.3 - works
• c82206c - works
• 05b36ba - broken
• f473692 - broken
• v2023.2.x/main - broken

This leads to this diff: c82206c...f473692

While in this commit the mt76 related commits are suspected:
bfbd49f

(see also: openwrt/openwrt@e367ba9...0cdbbd8)

Site Configuration:
https://github.com/ffac/site

Custom patches:
none related, see repo

[rotanid]

next step would be to find out which of these commits it could be:
https://git.openwrt.org/?p=openwrt/openwrt.git;a=commit;h=f64576f367eb21f6f968af181001ab395a9c00d0
https://git.openwrt.org/?p=openwrt/openwrt.git;a=commit;h=50148a40d2b0de8adddd8f11b95d6436355c431e
and after that, which of the mt76 commits in this commit...

[maurerle]

It is exactly this Openwrt Commit:
openwrt/openwrt@50148a4

This works: openwrt/mt76@ce7ccc5

So I traced it now to:
openwrt/mt76@ce7ccc5...2135e20

these work:

• ce7ccc5401685f144531259fbd46a7e520a6bce5
• 4f681a8fbc91656f6384fdf6d71a6f5a9f79e65c

broken:

• b4a917417c856307fe19fb6a1a2819e3270df22e

openwrt/mt76@b4a9174

[maurerle]

@nbd168 do you have any idea what might cause the airtime channel utilization not to be reported correctly on dap-x1860 since this commit:
openwrt/mt76@b4a9174

I will build with mt76/master once more to verify (nope, just setting the mt76 commit did fail building the package..)

[maurerle]

The issue fixes itself after a few hours of runtime, but does reoccur on reboot:

So it does not have user impact, does fix itself after few hours, looks a little bad on the graph, but other than that it is not crucial.