The fix is in now!
And this commit maybe could alleviate the problem on AX160 with AC160 clients?
Thank you OpenWrt community for raising awareness of the issue and fix and thank you devs for listening! Should be plenty of time to get this tested out in snapshot and backed into 23.05 now. This is great news.
7 Likes
I see that we are looking at the functionnality on of the mt76xx driver
Can't they also look to make it work in AP/STA bridge mode like with the MTK driver ?
What is AP/STA bridge mode? Is it Station WDS/AP WDS, and if so, this already works... I'm using it on a couple of devices with different drivers: mt76, rt2800, ath9k, ath10k, brcmsmac. If it is a proprietary "universal" implementation that uses MAC NAT in 1:1 or 1:n modes, I don't think it will be implemented any time soon, and even then, it will not be implemented as a mt76 feature but a feature of mac80211/cfg80211.
I have issues reading the code.
So IEEE80211_VHT_CAP_SUPP_CHAN_WIDTH_160MHZ got removed for mt7915, but it seems to still be present for mt7916. The latter is used on DBDC devices AFAIK:
root@OpenWrt:~# lspci
00:00.0 PCI bridge: Device 0e8d:0801 (rev 01)
00:01.0 PCI bridge: Device 0e8d:0801 (rev 01)
01:00.0 Unclassified device [0002]: MEDIATEK Corp. Device 7916
02:00.0 Unclassified device [0002]: MEDIATEK Corp. MT7915E 802.11ax PCI Express Wireless Network Adapter
looks like I'll need to run tests on my iPhone. Are you guys doing it with iperf?
This is great news. Is it expected to fix the Xiaomi "Redmi AX6000" also? The patched file, mt7915/init.c, includes a great number of references to MT7986 which is used by the Redmi AX6000.
Yes, iperf3 upload test.
When the uplink connection from an AX capable iPhone or iPad drops to 32 Mbit/s HE MCS0 (zero) with only one stream with obstacles in line of sight in AX mode and in AC mode you get hundreds of Mbit/s uplink connections it’s probably the issue.
On 80 MHz channels.
1 Like
Great news! It also has been already committed to the OpenWrt master branch (so it will be included in the upcoming snapshot builds)!
https://git.openwrt.org/?p=openwrt/openwrt.git;a=commit;h=063641f8cf7990731919b950c1bf9eb15d1e74c9
Hopefully it will also be added to the 23.05 branch (but not yet so far).
2 Likes
Absolutely fantastic news. Fingers crossed for a push to RC3, not keen on using snapshots for production
Does this just bring AX on a par with AC or is AX now worth using? And does it all work in WDS mode?
I've just been using AC on my 3x RT3200s connected via WDS for ages now and it has been working just fine.
I believe SNAPSHOT r23566-37ff916af7 includes these commits, but unfortunately AC160 capable client again connects at n40 when router is setup with AX160.
PS: someone mentioned that the laptop drivers are outdated- it could be also a reason. However with Netgear R6800, same laptop do not have problems withac160 or ac80. Unfortunaly I can nou update the Wlan drivers on the laptop.
80 MHz AX channel solved the issue for your case: 866 Mbit/s AC instead of 300 Mbit/s N connection mode with your device, OS and driver setup.
In any case, a 300 Mbit/s N connection is about 10 times faster what iPhone and iPad users had in AX mode with the 32 Mbit/s uplink rate and down to 14 Mbit/s upload speeds in good conditions before the patches.
I Know, I know. I am not trying to hijack the discussion, but to help with a look at all dimensions and report any side effects.
Yes. To exclude possible issue causes from driver versions or hardware limitations your case should be verified from more users of the Intel Wireless AC 9560 card with newer drivers and maybe also with Linux.
1 Like
I built the mt76 patch against the latest on 23.05 branch a day or two ago. The patch greatly improves wifi upload speeds and stability for my iphone when connected with AX. Kudos to @Anteus!
I configured the AP for 80Mhz. Is setting 160Mhz mode a prerequisite for reproducing a problem? I honestly don't see the point of 160Mhz if the manufacturer says the HW doesn't fully support it.
I tried to reproduce @Underworld's issue. Unsure if he is using the patch or the script workaround. I don't have an intel ac-9560 to test with but today forced an intel ax210 to AC mode by loading the module (in linux) without 802.11ax support. I get the following iperf3 speeds and I verified that it was connected with VHT 80Mhz modulation. I also tried an AC connection from a windows intel AC-8265 client and it connected at the max VHT modulation as well.
AX210 AC 80Mhz Upload Speeds:
[ ID] Interval Transfer Bitrate Retr
[ 5] 0.00-10.00 sec 827 MBytes 694 Mbits/sec 0 sender
AX210 AC 80Mhz Download Speeds:
[ ID] Interval Transfer Bitrate Retr
[ 5] 0.00-10.04 sec 745 MBytes 623 Mbits/sec 4 sender
1 Like
Completely agree. If it was running to spec this would never have been an issue to begin with 
This problem is unfortunately not solved on the Xiaomi "Redmi AX6000"
- Firmware Version: OpenWrt SNAPSHOT r23566-37ff916af7 / LuCI Master git-23.158.78004-23a246e (
37ff916is later than the commit that includes the mt76 version bump) - MediaTek MT7986 802.11ac/ax/n
- 802.11ax 80Mhz on channel 149
As you can see, from my wife's MacBook Pro, the 802.11ax speeds are still terrible in one direction when a wall is between router and MacBook. (Other devices in same location are perfectly fine, and this device is totally fine on 802.11ac 80Mhz channel 149.)
1 Like
The patch only removes 160 MHz AC channel capability for MT7915 wireless controller, that never was advertised with 160 MHz support from the vendor.
MT7986 is Filogic 830 which is advertised with 160 MHz channel support:
it’s integrated dual Wi-Fi 6 baseband supports up to 1024QAM, 24 user OFDMA, BW160 and 8 spatial streams (SS).
If the AX capable Apple iOS and iPadOS AX uplink performance issue was inherited to Filogic 830 MT7986 with official 160 MHz channel support I think it could be a different discussion.
Could you try the suggested mac80211.sh workaround first?
And please run
iwinfo <device> assoclist
or check the Luci wireless network page to see which wifi connection rates are negotiated with the Apple device while running the iperf3 test. In my case the uplink connection drops to the slowest AX mode of 32 Mbit/s HE MCS0 single stream, which explains the speed issue.
2 Likes
Just a quick note that this mt76 patch also seems, so far, to be fixing a tangential issue I had been experiencing with a WAX206 in AX mode only.
Editing to update: the patch didn't fix this after all. Seems to be a long term problem, trying possible workaround. Details in the same thread.
2 Likes
OK so I’m getting > 300mbps without doing anything on my iPhone. This is DBDC mt7915. This is probably fine.