Is the pull request commit suitable for building?
you will probably want to add a ton of patches (unless included), but those are unfortunately only listed in the posts.
Thank you for the detailed instructions; managed to get one of these to work with fanboyās image!
Is there any trickery to get 6 GHz to work 320 MHz wide? iwinfo reports it is, though stations only see this AP only, if set to 160 MHz and not, if 320 MHz.
If anyoneās got one working on 320 MHz, would you share that config?
Thank you.
By the way, I've had similar issue with visibility of 320MHz at Meta Quest 3 as well after updating to fresher image from @OpenWRT-fanboy. I will update again to latest and retest, just to not confuse others.
a flat head screw driver works too, if it's small enough.
The flow offload fix by @tesf23 appears to work and be stable in standard configurations. But in more advanced situations, there may be some issues. Don't forget that this code has only been out in the wild for a few days.
To help debug the issues, I would start out in a basic config and then start adding complexity until it breaks. So far, it seems the problems materialize with VLAN and guest networks.
If someone is having this issue and going to buy something special for it, id suggest one of these. It's long enough and has many torx sizes for future endeavors! I got this one from harbor freight
Also to pry the case, I carefully used a sharp knife, it didn't marr the plastic or anything. Just be CAREFUL!
Iām working on a fix for it. The patch might need to skip the mac override when a interface is not belong to br-lan or it has a vlan tag that way it not interference with the layer3 offload. I will test another patch later this week. Under AP mode itās been work solid for me for couple days.
Thanks @tesf23 and @OpenWRT-fanboy for your work.
I'm wondering, are your modifications only effective in hardware offloading mode ?
I mean, if there is problems with Wifi and VLAN in router mode, is it tied to hardware offloading ?
When I set to 320MHz, my iPhone 17 Pro Max sees it, but connects at 160MHz (hardware limitation).
But at 320MHz, Windows laptop with BE200 does not see it at all and won't connect at 6G. I need to set to 160MHz for the BE200 to see and connect.
In practice, it seems like it is impossible to connect at 320MHz at the moment.
Thank you - at least now I know this is not due to me fudging up my configuration.
Fwiw, I managed to get a start popping the case in the corners opposite the ethernet connectors. I used guitar picks, mostly.
Iām thinking about getting one to use as an APā¦. just wondering if anyone has compared this to other āfamousā APs, like Unifiās U7 series? It should in theory perform better as it has more streamsā¦
U-boot has not changed during the install, so you absolutely can break in with serial and type.
I'm not able to break in at u-boot, at the fail safe prompt or press enter to get a console.
It's as though my keyboard isn't recognized for input. I'm working from the same keyboard now to type this note so it's not as simple as the keyboard being borked.
then your only option is to use the reset pin
Are you sure you have the tx-rx pins set on the correct locations and that tx-rx are not reversed? It SHOULD* work if everything is set up correctly.
@OpenWRT-fanboy Added the lan to wan and vlan safety check from netfilter patch. My test passed. https://github.com/rchen14b/OpenW1700k/commit/cb4f78d74252470857f824574f3ca217a0dfcb55
I can connect at 320MHz using a BE200 in a Linux laptop.
But the speeds (with routing) are limited by the high CPU usage : around 850Mbps with iperf3 TCP, and up to 1150Mbps with curl (not sure why itās faster).
In the new NPU status page, the flows are all shown as Unbound with 0 packets / bytes.
For IPv4 :
-
the Original Flow column lists something like REMOTE_IP:80 ā WAN_IP:port (I would expect the client IP somewhere, to accelerate the NAT)
-
the New Flow column shows weird public IPs
For IPv6 :
- the Original Flow column lists something like REMOTE_IP:80 ā CLIENT_IP:port
- the New Flow column is empty
Youāre going to have to get that to work. Maybe a different adapter? It does need to be 3.3V.
build 32197 (and later) has the @tesf23 offload bugfix.