I see snapshot is sitting at kernel 6.1. I am hoping this is just a mediary to a newer kernel. There has been a lot of additions for newer wifi hardware and the groundwork for wifi 7 with kernel 6.2. Is 6.1 the target kernel for 24.05?
it's being considered https://lists.openwrt.org/pipermail/openwrt-devel/2024-February/042204.html
moving to 6.6 might however push the next release into 2025.
2 Likes
Awesome, 6.6 would be much better. Thanks
there are some drawbacks though, if you read the whole conversation 
1 Like
Yeah I see the size issue, but dropping devices due to these requirements is going to happen eventually anyways, it's inevitable. Devices that wouldn't have enough storage are probably the ones that really aren't worth carrying anymore. And someone mentioned dropping 4/32 devices, but I thought we were already past that point.
1 Like
Yea, it always sounds like this. And it is always a newer kernel released.
And it is always a new wifi standard ready. But no devices to use it with since we need 6GHz this time so completely new radios need to be built so that is a couple of years away.
2 Likes
Except that, as multiple devs pointed out, backporting stuff gets more complex the older your kernel gets. Yet people want new features. So...
If you inspect your 23.05 installation, you'll notice your wireless stack is from the 6.1 kernel despite the actual kernel being 5.15. As is the case with almost any modern OpenWrt release: wireless stack is more recent than the kernel.
Early adopters often end up being sorry when it comes to wireless, so don't jump on that wagon too soon. First gen hardware often still has teething problems (if you're getting first gen at all, and not just preproduction...).
1 Like
Yeah I just built 6.1 for a gl.inet a1300 and realized rtw for 8822 was available, but that isn't there until like 6.2.
In that case though, what is the purpose of backporting instead of just updating the kernel itself? I don't see the benefit of backporting? I'm not a kernel dev, I'm legitimately curious. To me it sounds like a waste of time, when that time could be used to build the newer kernel.
And it's not just the features that are important, a lot of security and bug fixes are included in newer kernels. My wireguard interface wouldn't work 50% of the time on 23.05 on my gl.inet slate plus, but with this custom 6.1 build I have no problems at all. I'm more interested in the increased stability and performance than I am the features.
switching to a newer kernel is not just a tickbox - it needs porting patches that are not in upstream + creating new patches and a lot of testing so let's say switching from 5.15 to 6.1 is easier than to go directly to 6.6 (small incremental steps)
2 Likes
The purpose is you can stick with a tested and stable LTS kernel - 4.14, 4.19, 5.4, 5.10, 5.15, 6.1, 6.6 - and still use a newer wireless stack. Just switching to the latest kernel because of the networking stack means you get all the new bugs too, not just those inside the networking stack 
.
Don't be mistaken, lots of relevant fixes (if not all) are backported to LTS kernels, that's why they are LTS after all. They are actively maintained.
3 Likes
Yesterday a shitload of commits introduced kernel 6.6 in main snapshot. A first target even set 6.6 for testing (mediatek). On this mailing list, devs seem to lead in favor of 6.6.
https://lists.openwrt.org/pipermail/openwrt-devel/2024-February/date.html
Remind us not to help you again.
4 Likes
This topic was automatically closed 10 days after the last reply. New replies are no longer allowed.