confirmed. this is the best setup for me now.
There is already a new firmware-5.bin_10.4-3.9.0.2-00139 version, however with this firmware the 5ghz dont come up.
With firmware-5.bin_10.4-3.9.0.2-00135 all my network devices work (also the ones that used to cause problems in the wifi again and again)
Thanks
I just tried it, it seems OK on my R7800.
If anyone is interested in comparing what "features" each wifi firmware provides, I sampled a few, as reported by
dmesg | grep 'firmware ver'
Mainline/old:
firmware ver 10.4-3.9.0.2-00131 api 5 features no-p2p,mfp,peer-flow-ctrl,btcoex-param,allows-mesh-bcast,no-ps,peer-fixed-rate crc32 23bd9e43
firmware ver 10.4-3.9.0.2-00139 api 5 features no-p2p,mfp,peer-flow-ctrl,btcoex-param,allows-mesh-bcast,no-ps,peer-fixed-rate,bit22 crc32 84f36f58
firmware ver 10.4-3.10-00047 api 5 features no-p2p,mfp,peer-flow-ctrl,btcoex-param,allows-mesh-bcast,no-ps crc32 19ca6df2
CT:
firmware ver 10.4b-ct-9984-fW-13-5ae337bb1 api 5 features mfp,peer-flow-ctrl,txstatus-noack,wmi-10.x-CT,ratemask-CT,regdump-CT,txrate-CT,flush-all-CT,pingpong-CT,ch-regs-CT,nop-CT,set-special-CT,tx-rc-CT,cust-stats-CT,txrate2-CT,beacon-cb-CT,wmi-block-ack-CT,wmi-bcn-rc-CT crc32 7ea63dc5
I also ran some iperf3 tests, one side is a wired Linux server, the other side is a 2019 MacBook Pro (5GHz radio). Same build (master-r15759-ce4cb8e51d-20210214), same location, the only change is the firmware:
Mainline/old firmware 10.4-3.9.0.2-00139:
[ ID] Interval Transfer Bitrate Retr
[ 7] 0.00-10.01 sec 571 MBytes 478 Mbits/sec 110 sender
[ 7] 0.00-10.00 sec 568 MBytes 476 Mbits/sec receiver
CT:
[ ID] Interval Transfer Bitrate Retr
[ 7] 0.00-10.00 sec 205 MBytes 172 Mbits/sec 250 sender
[ 7] 0.00-10.00 sec 202 MBytes 170 Mbits/sec receiver
Stability, performance, mainline looks by far much better for me than CT.
FWIW, similar reports here: https://github.com/greearb/ath10k-ct/issues/138
It appears Rx rate is always 0 in build r15759.
This commit might be the fix:
The discussion on comparison of ath10k-ct to mainline ath10k features & performance might be taken somewhere else, as it has nothing to do specially with my build.
Understood. I figured it would be useful for someone trying to decide which build to use as the question does come up occasionally. It was more to mean "try both and decide which to use".
Did you try comparing a recent master build since the fix is applied there?
Btw,
both master and 21.02 branches have just been updated to use the kernel 5.10 based -ct drivers, matching the recently upgraded mac80211 that is also 5.10 backports.
Looks like kernel 5.10 is coming to mainline too. But 21.02 will use 5.4
currently porting 5.10 to ipq8064... i need to do some test to check if we can switch and use dsa
This seems to be fixed in master-r15838-d2d32dcd5f-20210218.
161 is being properly selected
SyntaxError
illegal character
in https://192.168.20.1/luci-static/resources/view/ddns/overview.js?v=git-21.044.37257-ccdca27:852
at compileClass/< (https://192.168.20.1/luci-static/resources/luci.js?v=git-21.044.37257-ccdca27:2516:7)
i get this in master-r15838-d2d32dcd5f-20210218.
Possibly something strange in your DDNS config.
I can see the main overview page and the ddns page normally, so this might be something related to your setup.
stable openwrt-21.02
owrt2102-r15841-60823c67cb-20210220 (opkg package download does not work, yet)
I have built the first 21.02 build. It is almost identical to master at the moment, so this is mostly for people who want to initialize their own 21.02 build environments.
Note that the 21.02 buildbot has not been signing the package lists properly, so opkg is currently unable to download and verify package lists. This has already been fixed, but it will take 1-2 days to propagate.
(The package download dirs are already there, so manual downloads are possible: https://downloads.openwrt.org/releases/packages-21.02/ )
I will build 21.02 irregularly for now, and will switch the regular "stable build" to 21.02 latest when the first official rc version is released. I will then drop the 19.07 build soon after that.
Thanks a lot hnyman!
Still too complicated without the actual commands needed.
Just read the message #2 of this thread:
Any plans for 21.02 builds using the mainline ath10k driver/firmware? Or with these builds can we just opkg remove ct and opkg install mainline?
I will mainly build with the OpenWrt defaults, currently -ct, and you can manually change it in your .config if you want.
Installing ath10k firmware in a live router is possible, but the kernel driver requires build so that need your own compilation.
I might sometimes compile with mainline, as it only requires minor changes to .config.init
But not a regular thing.
Is it possible to always build the mainline driver even for the -ct builds and have the pkg available in the image in some directory but not installed? Then one would be able to just grab the ct build and install the mainline driver. Or have it available as a separate download from your dropbox? Not sure whether the build system supports something like this. It would be nice to have an easy way to switch between the two and run tests.