Good, I respect your opinion, but I maintain mine. strong hug!
5 Likes
software… it was already corrected with the wifi closed in 21.02, it just needs polishing now…
2 Likes
Great news, thank you japa182.
A post was split to a new topic: Flint 2: OpenVPN AES CPU support missing?
bug related to power on each device, in 4.5.6 the information appears correctly in luci but in 4.5.7 it does not appear, I know it is a snapshot but follow the feedback for correction in a future beta or rc version.
4.5.6
4.5.7
As @admon already mentioned. Its still a snapshot and contains bugs by nature which the gl-inet team is probably already aware of but did not fix yet.
1 Like
It doesn’t hurt to reinforce, right?
Because if not, what would be the purpose of releasing snapshots so we don’t report bugs? Do they only release snapshots so we don’t install them and assume that the gl team knows? There’s no logic to that, right?
1 Like
*sigh*
The snapshots now are mostly for Wi-Fi testing. All other stuff is not important right now.
That’s just the way it is when you’re using the proprietary driver.
To view the stats and configure the driver GL.iNet would need to modify and include luci-app-mtwifi, which I previously linked to in another post.
https://github.com/Azexios/openwrt-r3p-mtk (less outdated)
https://github.com/Nossiac/mtk-openwrt-feeds (outdated)
I won’t use the GL.iNet snapshot version though since it’s a downgrade to an OpenWrt version that’ll no longer receive updates of any kind. And for comparison’s sake, imagine if Microsoft downgraded you from Windows 11 to Windows 8 because a driver performed better on Windows 8, even though Windows 8 no longer receives updates, it doesn’t support DX12 and both Steam and Chrome no longer support it.
5 Likes
I must be one of the few lucky ones, no crashes for over 10 days on 4.5.6 and I use it quite a bit (WG behind CGNAT for Plex, WG Transfers, IPv6, Adguard with Hagezi multi pro++ and many devices on lan/wifi)
The only issue I have is slowness on 2.4 ghz, slow loading times sometimes on instagram. I disabled 160mhz and went back to 80mhz since I don’t really need it for 5ghz.
I have not gone back to same ssid either, since separating them made the router more stable.
2 Likes
A post was split to a new topic: Flint 2 Problem with hardware acceleration
I believe that the combination of 80 MHz and not having the same SSID is the most stable one for the 5GHz channel on <=4.5.6, but it still doesn’t run flawlessly.
Even though gl-inet implemented a patch to bypass the kernel crashes caused by using the same SSID on both bands, it still affects the WiFi’s stability. But even running the router only on 5GHz is still causing problems.
The snapshot version 4.5.7 instead uses the closed source drivers and works perfectly with all kind of devices on both bands even with the same SSID.
I have it running now for almost 3 days and do not notice any WiFi speed degradation on the 5Ghz band. I didn’t test the 160MHZ bandwith yet, since its in my opinion a little overkill.
I really hope there is a way for GL-Inet to use the closed source drivers with OpenWRT 23.05.
Btw if you need a good theory introduction into WiFi 6 see:
5 Likes
@alex_zheng and @JerryZhao wouldn’t it be possible to build firmware with the latest version of OpenWRT, v23.05, and Mediatek closed source drivers?
1 Like
Yesterday i wanted to test my new philips airfryer over the 2.4 ghz wifi and it was a nightmare on the mt79 driver (not oem version, openwrt snapshot r24993-32b4498c15).
Since i live in a small appartment my kitchen is the farest away from all my 2.4ghz wifi.
First i thought it was a mtu issue, then the situation became worse and worse got errors like psk mismatch, and that my airfryer tried to associate before authentication.
The next day i checked it again, and then i figured like many others have said that the range is really bad, well i knew this but by measuring it with wifiman its not the same it gave me the impression my wifi would work atleast in my appartement but less than one elevation and far worse than flint 1, but it was even more worse, all these issues appeared from like 20 meter further with a plaster wall in between.
So im assuming it has to be a power issue and not reported correctly in the driver, it felt like <10 power, if you are testing close you won’t notice issues but the stability becomes extremely unstable by going further away.
Now I’m running 4.5.7 with mtk drivers and hurray, the range improved dramaticly for me, and my speed was 122 mb/s download on a Poco X6 pro, i haven’t tested further ranges but it looks alot better.
The only issue i found though and i hope it can get fixed later:
If i create advanced wireless settings via luci, all these mtk ssids spawn and you cannot sent your own, i was forced to use gl ui.
And the luci app for mtk is missing, i need to set country 
If i find something else i will report it here it looks very good so far.
2 Likes
I finally crashed 4.5.6 after 10+ days, basically lan/wan ports went down, I was able to connect via wifi but there was nothing interesting in the logs, other than the links are down and reset failed.
Post them anyway - maybe some others will find interesting parts.
So that version 4.5.6 is stable now? I got pop up upgrade frimware. Now beta is 4.5.7 
Packages in this new version are incompatible with the distributed packages, you cannot install sqm for example like the screenshots below.
and the mkt wifi package is also missing as there is no option to change the wifi region or the power of each connected client, as shown in the print below.
irq-smp-affinity is not working, after turning off hardware acceleration the entire process is played directly on core 1,
Even with the script playing exactly right, the script does not start with the router and does not accept starting either.
echo 2 > /proc/irq/130/smp_affinity
echo 8 > /proc/irq/131/smp_affinity
Edit: But the script is not correct, 130 and 131 are related to version 23.05 in 21.02 the directions are 120 and 121, tested here and working perfectly.
the script still doesn’t work even when changing the directions, just playing the commands to start with the router that works.
fullconenat was deleted in this version, did anything happen to the deletion? in the previous one it worked… but not in the pppoe connection, only in dhcp
1 Like
Are you going to sigh right now with the version being beta?