Closed DragonBluep closed 2 years ago
im getting this also, but on all 100+ channels
i guess it has something to do with missing eeprom data
mtkeepmgr shows nothing rrom@phoenixhawk:~/eeprom/mtkeepmgr# ./mtkeepmgr -F RE200.mtd6.bin [EEPROM identification] ChipID : 7628h Version : 2.0 Chip : MT7628
[Device identification] MacAddr : 00:0c:43:e1:76:90
i guess we have to fill values from other eeproms or do some code magic if no values are found, i feared using other eeprom since i cant change the mac in file fo fit mine.
as far as i remember stock firmware did only channel 36 to 58, so they simply didnt implement it
i guess it has something to do with missing eeprom data
This is due to mtkeepmgr not supporting mt7628. Contrary to yours, my device factory firmware only supports 149-161 channels, but now only 36-64 channels work.
i feared using other eeprom since i cant change the mac in file fo fit mine.
@phil2sat I just checked the source code, the MAC of the RE200 is not stored in the eeprom partition, so you don't have to worry about it. By the way, I guess your MT7628 is also not working properly. Sometimes the signal disappears suddenly. This is a problem that has existed for a long time.
but if its not stored in eeprom, why mtkeepmgr shows mac from eeprom? also the mac error on non supported channels, seems like some parts in eeprom missing, as at least you can use 100+ channels, for me its only the 36-58 range.
and since my hack, the 5ghz mesh network runs stable, without any reboot.
root@RE200:~# uptime 11:09:44 up 8 days, 19:09, load average: 0.05, 0.04, 0.00
Most devices store the mac in the eeprom, usually located at eeprom@0x04. The mac you see from mtkeepmgr is not the real value, for the RE200 it is in the "config" partition, 0x2008. nvmem-cells = <&macaddr_config_2008>;
The "MAC" in the log "MAC error detected" is not mean "MAC address", they are two different things.
Before this commit 20d1fed838b958701ed5d030b663ba9227477ab1, there is a chance that channel 100+ will work properly. This commit seems to have added more stringent checks, which caused MT7610 to fail to start. I have looked around at the GPL source code and there seems to be no check for 0x10f4 register in MT7610. I am not sure if this register is only available in MT7612. https://github.com/openwrt/mt76/blob/747c70fc6c890c1f0e4f33d4eff20bb6498dc662/mt76x02_mac.c#L1052
[70516.605224] mt76x2e 0000:01:00.0: MAC error detected
Sun May 8 06:04:23 2022 kern.err kernel: [70516.605224] mt76x2e 0000:01:00.0: MAC error detected
Using OpenWRT snapshot build with the latest mt76 driver, D-Link DIR-860L B1.
Edit: The error logs have been very rare now, and it used to appear very frequently. The log error above appeared after running for over 1 day.
Same problem on Xiaomi Mi Router 3G with channel 36
@phil2sat MT7628 is the 802.11bgn SoC (CPU + WLAN) so it does not support 5 GHz. MT7610E is the 5 GHz WLAN card. The onboard EEPROM has 64Kbytes, the first 32K (0-32767) are for MT7628, the second 32K are for MT7610E.
Use the following command to strip the mtd6.bin:
dd if=mtd6.bin of=mt7628eep.bin bs=512 count=64
dd if=mtd6.bin of=mt7610eep.bin bs=512 skip=64
Then use mtkeemgr
to check mt7610eep.bin
to print out the EEPROM content. The GPL tarball released by TP-Link seems to have an SKUdata.dat file so I highly suspect that all TP-Link's MediaTek-based products use external files to determine TX power and do not rely on EEPROM content. So if I guess right, you can to either manually write target power values to the EEPROM, or use my module parameter hack approach in #633.
Please try the latest version
"MAC error detected" problem was fixed. Although channel 149-161 still doesn't work, it might be a hardware bug, so maybe we can close this issue.
There're still similar error logs on my D-Link DIR-860L B1 running openwrt snapshot.
[257938.974472] mt76x2e 0000:01:00.0: MAC error detected
[258658.955640] mt76x2e 0000:01:00.0: MAC error detected
On my ASUS RT-N56U B1 with the release OpenWrt 22.03.3 also has these errors
[42192.663667] mt76x2e 0000:01:00.0: MAC error detected [42272.221710] mt76x2e 0000:01:00.0: MAC error detected [42298.641011] mt76x2e 0000:01:00.0: MAC error detected [43173.059364] mt76x2e 0000:01:00.0: MAC error detected [43191.118853] mt76x2e 0000:01:00.0: MAC error detected [43203.128558] mt76x2e 0000:01:00.0: MAC error detected [43372.434374] mt76x2e 0000:01:00.0: MAC error detected [43388.733976] mt76x2e 0000:01:00.0: MAC error detected
Same issue on the Asus RT-AC51U
kern.err kernel: [649615.261728] mt76x2e 0000:01:00.0: MAC error detected
Appears on my Xiaomi Mi 4A Gigabit Edition too. Using latest snapshot. 5 GHzs is set to channel 36 - 80 MHz.
The AP also hangs when watching Twitch.tv using iPhone SE 2020. Shows no other errors on log. 2.4 GHz works fine.
@nbd168 @DragonBluep
config wifi-device 'radio1'
option type 'mac80211'
option path '1e140000.pcie/pci0000:00/0000:00:00.0/0000:01:00.0'
option channel '36'
option band '5g'
option country 'TR'
option cell_density '0'
option beacon_int '100'
option htmode 'VHT80'
config wifi-iface 'default_radio1'
option device 'radio1'
option network 'lan'
option mode 'ap'
option ssid '*'
option dtim_period '3'
option disassoc_low_ack '0'
option encryption 'psk2+ccmp'
option key '*'
Sat Mar 18 13:09:23 2023 kern.err kernel: [825180.433285] mt76x2e 0000:01:00.0: MAC error detected
Sat Mar 18 13:11:09 2023 kern.err kernel: [825285.483278] mt76x2e 0000:01:00.0: MAC error detected
Sat Mar 18 13:11:09 2023 kern.err kernel: [825285.703219] mt76x2e 0000:01:00.0: MAC stop failed
Sat Mar 18 13:11:09 2023 kern.info kernel: [825285.754461] mt76x2e 0000:01:00.0: Firmware Version: 0.0.00
Sat Mar 18 13:11:09 2023 kern.info kernel: [825285.760052] mt76x2e 0000:01:00.0: Build: 1
Sat Mar 18 13:11:09 2023 kern.info kernel: [825285.764280] mt76x2e 0000:01:00.0: Build Time: 201607111443____
Sat Mar 18 13:11:09 2023 kern.info kernel: [825285.793272] mt76x2e 0000:01:00.0: Firmware running!
Sat Mar 18 13:11:09 2023 kern.info kernel: [825285.803381] ieee80211 phy1: Hardware restart was requested
Sat Mar 18 13:16:48 2023 kern.err kernel: [825624.443228] mt76x2e 0000:01:00.0: MAC error detected
Sat Mar 18 13:17:43 2023 kern.err kernel: [825679.793243] mt76x2e 0000:01:00.0: MAC error detected
Sat Mar 18 13:23:21 2023 kern.err kernel: [826018.233192] mt76x2e 0000:01:00.0: MAC error detected
Today this happened while using a MacBook Air 2020 (Intel).
The things I've tried so far: Trying WPA2 and WPA3 (with 802.11w), using DTIM Interval values of 1, 2, and 3. Both combinations of Disable Inactivity Polling / Disassociate On Low Acknowledgement. Changing channel width to 20, 40 and 80 MHz.
Now I'll try reducing maximum transmit power starting from 22dBm down to 20dBm.
same with UniElec U7621-06 (16M flash) OpenWrt 22.03.3 r20028-43d71ad93e
Wed Jul 12 16:08:30 2023 kern.err kernel: [ 2156.969937] mt76x2e 0000:03:00.0: MAC error detected
I had the same issue with MAC error detected
.
I solved by removing external connected USB SSD, so probably not enough power from power supply is the root cause in my case
Can anyone confirm if whoever is having this issue has RF shield or not on the wifi chips on their router?
I tried every fix mentioned and nothing worked. I tried the snapshot release, 23.X, 22.X. I tried using a higher quality power adapter, made sure no usb peripheral were connected, installed heatsinks and even a fan to rule out overheating issues but only after I made the RF Shields using copper and soldered them on the problem went away. The router has been sooo stable ever since I did that mod. Here's how it turned out:
Ig if anyone is still facing this problem, it's worth a shot at least. (I recently came to know that Mi cheaped out on the later revisions of my model [Mi Router 4a] and removed the shielding)
Anyway, copper is so easy to solder on that it was was a breeze, the only time consuming part was bending the copper sheet into the correct size. I also installed thermal pads inside so that it also acts like a heatsink. I used a 0.3mm copper sheet btw.
10/10 would recommend :p
MT7610E only channel 36-64 work, after setting the channel to 149-161, I get an error:
Associated with #664 #593