If you search the release candidate threads, most people I've seen with WRT3200ACM are saying they've had good results, but there simply needs to be more people to test it with their specific configuration (please do). One feature that is said to be broken in mvebu is software flow offloading in the firewall, so don't turn that on.
1 Like
In the RC2 thread, @ne20002 , a user with a WRT3200ACM, is experiencing on RC2 what seems like a very similar bug to what we thought was fixed in 5.15 master, where it acts like a hub. Do you guys know what he should start testing to figure this out? He says:
I set the WRT3200ACM up with a new image, installed the packages I have usually installed then build a new image with AttendedSysUpgrade. After reboot I applied the backup from 22.03.
I think the switch started acting as a hub after I installed kmod-nft-bridge.
What Info do you need? How can I send it to you?
Start over by configuring from scratch, don't apply the backup from 22.03
1 Like
Just the other day I decided to take another crack at getting my linksys to a stable condition. It's running 23.05.0-rc2 now, but 2.4ghz has some kind of issue regardless of the different settings I've tried. Right now I get about 15mbps down at the absolute maximum on multiple devices, but upload is around 100mbps.
I've tried changing the country code, toggling 40mhz, removing the configured network and adding a fresh one leaving everything untouched aside from changing the SSID and wpa2 pass. When checking the logs nothing stands out as odd, at least not that I can tell... This is an access point, so firewall etc. etc. is turned off and disabled at startup.
I do live in the middle of nowhere, so interference from any other networks is a non issue.
Any ideas on what causes this or how I can fix it?? I still have a few 2.4ghz only devices and sometimes need the greater range when using my phone around the property, so it would be great to actually get this device functional on openwrt. If its bugs or back to the stock firmware Ill just put it back on the shelf I guess 
I have about the same problem. My connection is much slower than yours. I must be farther out in the country.
https://forum.openwrt.org/t/slow-wifi-on-linksys-wrt1900acs-v2-23-05-rc2/166481?u=donh
As now the flow-offloading is brocken I decided to drop the WRT3200ACM and get a BananaPi BPi R3. I'm currently setting this up.
Slightly off topic, but I thought I'd ask as many experts are around here.
I am trying to configure the wifi on my WRT1900ACS on mode 802.11s. So far I failed miserably, but before going deeper on troubleshooting, I just wanted to ask if anyone here had it working? Or is it a limitation of the router/mwlwifi?
Thanks, that explains the issues I faced. I should have looked at this table before.
I would refer to the info on the main doc page too: [https://openwrt.org/toh/linksys/wrt_ac_series#marvell_wifi] so yea several features like 802.11s (Mesh), 802.11w (for WPA3), and MU-MIMO were never implemented in the mwlwifi driver so will never work in OpenWrt for these routers.
2 Likes
Hey folks, how is it going with the mwlwifi test builds by @jbsky ? Do you still need more testers? I'd be up for helping if it's still needed, but can't tell since the last activity a couple of months ago.
Also, is the mwlwifi work by @jbsky currently going into the rc2/rc3 releases? I browsed the detailed changelogs for them and didn't see anything about mwlwifi, though there were changes to the kernel for mvebu. Does this mean the test builds are in their own fork and not going into the official 23.05 releases?
I have used the jbsky test drivers for something like 70+ days of uptime with zero issues. Very solid as far as stability goes. I believe there are a few reports of increased Wi-Fi latency. But there are also a number of bugs fixed.
I think jbsky has taken a bit of a break from mwlwifi for a little bit there.
From my understanding, approx. half of his commits have already made it into stable branch builds of OpenWrt. While the other approx. half of commits have remained in testing branches over at mwlwifi repo.
At the moment I’m not sure what the next step is until jbsky becomes active again on mwlwifi dev.
3 Likes
Ah, maybe I'll check out rc3 and report back then. In case those results are helpful too.
Great to hear you've had success with this!
1 Like
Interestingly some users noted lower latency again with the updated driver using kernel 6.1 (from Divested build thread), as always with mwlwifi, your mileage may vary.
1 Like
Yeah, I’ve been following that as well. Personally, I can’t really notice any difference in performance one way or another. Although I don’t know exactly how to properly measure Wi-Fi latency to compare.
I ended up doing some very basic wireless latency testing from laptop to router and also phone to router, to compare the older mwlwifi driver to the newer one.
I was actually expecting some differences, but it was basically identical. Not higher or lower. This was using 5.15 kernel.
It’s so strange how different users setups can produce such different results. I’m actually quite pleased with mwlwifi over the past year or two. The first bunch of years were a disaster though, for sure.
2 Likes
Is there a plan to fix NAT software offload in 23.05?
It's been broken with years with kernel 5.15, maybe even before that. I don't think it's fixed in kernel 6.1 either 
If you use SQM it's irrelevent since the features are incompatible, but if you want gigabit throughput and rely on software offload those days may be gone for this target (sorry to be pesimistic about it, it might take an incredibly skilled dev to find the issue).
One way to highlight it may be to file a bug upstream Linux kernel. Also contribute any log info to OpenWrt bug report. Maybe it's time for a DL-WRX36 if you need that performance, not sure.
DL-WRX36 are not imported in Europe. And whenever it is in my power, I do not support companies that do not bother to comply with EU standards and EU consumers' and workers' rights.
3 Likes