Flint 2 (GL-MT6000 ) - bug reports - collective thread

I’ve been using version 4.5.2 for 2 days and so far the WiFi hasn’t dropped or anything serious.

I have adguard, tailscale, samba, dlna, fullconenat activated and it continues to work normally, today I will be connecting cable devices to the router to see if it reproduces the same error that I was having in firmware 4.5.3 and 4.5.4 and also in the mt3000.

People reporting WIFI issues on later versions make me not want to upgrade from 4.5.2 and is one of the reasons I continue to stay on it lol.

Apart from that weird “Wed Dec 31 16:00:00 1969 kern.emerg kernel:” which caused the system log to clear, I haven’t had any issues, that section of log has since repopulated with the usual stuff. I only have a couple devices connected to it and I don’t feel comfortable replacing my main router with this unit yet.

1 Like

Hi AlexWBaule. I need configuration GLiNet firmware.

Here is the command to generate a configuration file
---->sysupgrade -b /tmp/back.tar.gz
this command will get all the config of the router, then you can copy it in /tmp folder .
We are very appreciate to get that.

Sorry @alex_zheng , now im running a openwrt image. I Have a backup generated, before install openwrt firmware, thats work for you ?

I am on 4.5.4 and can also confirm that I have high packet loss with WiFi on both 5G and 2.4GHz. There does not appear to be any mention of problems in the log.

Ok all, I found another bug. Full Cone NAT is not working when using PPPOE. Tested with Nintendo Switch, when using DHCP, it works very well and the NAT type of Nintendo Switch is “A” with Full Cone NAT active, otherwise is “B”. With PPPoE is always “B”.

Using firmware 4.5.3 beta (21/11)

Please Share it to me. Our RD will check it.

I reproduced everything I had done in versions 4.5.3 and 4.5.4, both in the stable and in the betas.

The last step would be to connect devices via cable to check if my WiFi would have a problem and become disconnected and no longer be able to connect and it didn’t happen.

I was verifying that there are some differences between version 4.5.2 and the others,

The WED option is not active in this version, so I THINK it may be related to the problem we were facing with a dropped Wi-Fi connection.

only the bug related to fullconenat is only working in dhcp connection, if it is in pppoe it does not work as reported above by the colleague @raphamotta

@admins @alzhao @alex_zheng @Dipin @JerryZhao @hansome

1 Like

@alex_zheng i need some time. All backups that i have, is from openwrt images. To get a backup to you, i need to revert back to glinet image. I will do this tomorrow… OK ? and i will send it to you.

1 Like

I also tested full cone nat with mt3000 and it’s the same, works fine on DHCP, does not works on PPPoE.

I even connected MT6000 as PPPoE and MT3000 WAN to MT6000 LAN (DHCP) and full cone nat on MT3000 works (and of course, in MT6000 don’t works).

When using OpenWrt snapshots I’m able to have dozens of clients connected via WiFi and at least 1 device connected via an ethernet cable while WED is enabled.

I can see that one of the GL.iNet devs is working on a WED issue, although they haven’t submitted a pull request with the fix for the mt76 driver. And I was under the impression that GL.iNet already included some WED fixes in 4.5.3 or 4.5.4, but I guess maybe they didn’t merge everything?

With OpenWrt snapshots the only WED issue in the log seem to be this…

[    5.983584] mt798x-wmac 18000000.wifi: attaching wed device 0 version 2
[    6.025462] platform 15010000.wed: MTK WED WO Firmware Version: DEV_000000, Build Time: 20221012175005
[    6.034779] platform 15010000.wed: MTK WED WO Chip ID 00 Region 3
[    6.621890] mt798x-wmac 18000000.wifi: HW/SW Version: 0x8a108a10, Build Time: 20221012174648a
[    6.727387] mt798x-wmac 18000000.wifi: WM Firmware Version: ____000000, Build Time: 20221012174725
[    6.796534] mt798x-wmac 18000000.wifi: WA Firmware Version: DEV_000000, Build Time: 20221012174937
[    6.942544] mt798x-wmac 18000000.wifi: eeprom load fail, use default bin
[    6.949561] mt798x-wmac 18000000.wifi: registering led 'mt76-phy0'
[    6.956998] mt798x-wmac 18000000.wifi: registering led 'mt76-phy1'

And I’ve been seeing “ieee80211 phy0: WA: free done event” randomly in my log, but nothing’s stopped working.

Does anyone of you know how to apply a (kernel module) patch/.patch file to a openwrt snapshot source and compile all?

The normal openwrt compile instructions downloads the kernel (module) sources only when compiling it.

and how to restore back after upgrade?

OK, Thanks a lot. we will check it ASAP once we get it.

Maybe you can use uboot to restore the firmware.

If you use the “upgrade” image from openwrt, its easy.
just follow this guide

Somethink that i noted with the openwrt Firmware.

On 5Ghz. 160MHz

If i set 30dBm (1000mW), the radio freezes, and dont “start” (unkown signal) , and if i come back to channel 36 and 23dBm, its start. (after a reboot).

On GLiNet Image, i can set channel 36 and 30dBm on luci, and the radio starts, but i think thats is the problem, because there is a “scan” on start wifi, to find the correct channel to be 30dBm.

If you used the full OpenWrt one it will be more complicated since you need to use TFTP then.

Anyone having bad wifi signal for 5ghz?

Same location, same channel, my Flint 1 is doing better than Flint 2, not sure what is the problem.

SNR of 2.4ghz always more than 20 but for 5ghz the best I can get only around 10 to 15, is that normal case?

Thanks for your verification. WED is enabled in 4.5.3 and 4.5.4. We R&D are still doing more investigation.

2 Likes