oh you have no idea how we would love to improve this thing with correct documentation... the soc is really good problem is sw
again with wifi offload and correct settings i reached 3.1 gbps of iperf on wifi... just to give you some numbers of the amount of perf this thing can reach
If someone wants to help, try and find a device with IPQ8074 and QFP4550 FEMs and let us know. The AX6 is still crashing and we are reaching a point where there are no more ideas about what to try. So we need some examples.
Thank you, I was able to flash the latest OpenWRT image on the AX3600.
What client device is it (old/new, android, iphone, smartv)? How is it connected (ax/ac/bgn)? Is the connection stable if you don't have roaming enabled? Is your device changing AP's if they don't run with 802.11r? Is roaming working for other devices (old/new)? Do the logs show anything usefull? Are your ax3600 devices running stable (crash/reboots)?
Impossible to give any hint without input. Generally:
I can recommend an Android App called Fritz!App WLAN for analyzing wireless networks. It is very helpful for channel selection and router placement/dbm adjustment and its free.
EDIT: Found the bookmark what I would like to link/share, about roaming (its a quite good overview for beginners IMO): https://datto.engineering/post/wi-fi-roaming
Hello
Is there any prebuilt image for AX3600 including HW acceleration (with what is working so far) ?
Regards
I don't think so... Up to this point; almost all efforts were to make it work with OpenWRT + make changes upstream-able (so changes could be merged into the official OpenWRT repo).
Unfortunately, NSS/HW acceleration requires enormous changes (due to how Qualcomm designed & implemented it).
So even for devices that OpenWRT developers were able to add NSS/HW support, these changes are still not upstreamed (IPQ806x + NSS) and have to be kept in separate repos (i.e. https://github.com/ACwifidude/openwrt/tree/openwrt-22.03-nss-qsdk10.0)
Having said that, IPQ807x devices seem powerful enough to cope with typical traffic without additional NSS/HW acceleration, so the need for HW acceleration is less vital than for older devices like Netgear R7800.
Thank you for the information.
I will observe the issue and open a new thread.
I am gonna to test 802.11s. Many tutorials say, I have to remove wpad-basicand wpad-mini, then I have to install the package 'wpad-mesh-openssl`. Cant find anything like that, but in the wifi network configuration I can select "802.11s". Does it work "out of the box" or did I have to install further packages?
Thank you!
The robimarko images come with wpad out of the box, so 802.11s works directly.
ath11k WDS AP seems broken in a new way on the build I made today (f099f79, with ath11k: enable ethernet offload by default removed).
It kind of works, but seems to only support one client at a time? I have two WDS clients that attempt to connect. Client A will get a good link, while Client B re-associates in a loop. Pulling the plug on Client A and rebooting the AP gets a good link for Client B, while Client A will then fail to associate.
All three devices are AX3600s. Reverting the AP to a -pr build from 1 month ago fixes the situation, while the two clients are now working OK today's f099f79.
The view from the AP is this sequence over-and-over:
Mon Oct 24 01:10:22 2022 daemon.info hostapd: phy1-ap1: STA [STA_mac_here] IEEE 802.11: authenticated
Mon Oct 24 01:10:22 2022 daemon.info hostapd: phy1-ap1: STA [STA_mac_here] IEEE 802.11: associated (aid 2)
Mon Oct 24 01:10:22 2022 daemon.notice hostapd: phy1-ap1: WDS-STA-INTERFACE-ADDED ifname=phy1-ap1.sta2 sta_addr=[STA_mac_here]
Mon Oct 24 01:10:22 2022 daemon.err hostapd: Could not set interface phy1-ap1.sta2 flags (UP): Not supported
Mon Oct 24 01:10:22 2022 daemon.err hostapd: nl80211: Failed to set WDS STA interface phy1-ap1.sta2 up
Mon Oct 24 01:10:22 2022 daemon.err hostapd: nl80211: NL80211_ATTR_STA_VLAN (addr=[STA_mac_here] ifname=phy1-ap1.sta2 vlan_id=0) failed: -100 (Network is down)
Mon Oct 24 01:10:22 2022 kern.info kernel: [ 1457.706292] br-lan: port 10(phy1-ap1.sta2) entered blocking state
Mon Oct 24 01:10:22 2022 kern.info kernel: [ 1457.706336] br-lan: port 10(phy1-ap1.sta2) entered disabled state
Mon Oct 24 01:10:22 2022 kern.info kernel: [ 1457.711769] device phy1-ap1.sta2 entered promiscuous mode
Mon Oct 24 01:10:31 2022 daemon.info hostapd: phy1-ap1: STA [STA_mac_here] IEEE 802.11: deauthenticated due to local deauth request
Mon Oct 24 01:10:31 2022 daemon.err hostapd: nl80211: NL80211_ATTR_STA_VLAN (addr=[STA_mac_here] ifname=phy1-ap1 vlan_id=0) failed: -2 (No such file or directory)
Mon Oct 24 01:10:31 2022 kern.info kernel: [ 1466.720385] device phy1-ap1.sta2 left promiscuous mode
Mon Oct 24 01:10:31 2022 kern.info kernel: [ 1466.720449] br-lan: port 10(phy1-ap1.sta2) entered disabled state
Mon Oct 24 01:10:31 2022 daemon.notice hostapd: phy1-ap1: WDS-STA-INTERFACE-REMOVED ifname=phy1-ap1.sta2 sta_addr=[STA_mac_here]
This is the view from the "locked out" STA: a repeating sequence of fail count increments, timed back-off, and retry. The pre-shared key is correct.
Thu Oct 20 11:02:53 2022 daemon.notice wpa_supplicant[2598]: phy1-sta0: CTRL-EVENT-SSID-REENABLED id=6 ssid="WDS_AP_name"
Thu Oct 20 11:02:53 2022 daemon.notice wpa_supplicant[2598]: phy1-sta0: SME: Trying to authenticate with [AP_mac_here] (SSID='WDS_AP_name' freq=5180 MHz)
Thu Oct 20 11:02:53 2022 kern.info kernel: [ 862.400666] phy1-sta0: authenticate with [AP_mac_here]
Thu Oct 20 11:02:53 2022 kern.info kernel: [ 862.416157] phy1-sta0: send auth to [AP_mac_here] (try 1/3)
Thu Oct 20 11:02:53 2022 daemon.notice wpa_supplicant[2598]: phy1-sta0: Trying to associate with [AP_mac_here] (SSID='WDS_AP_name' freq=5180 MHz)
Thu Oct 20 11:02:53 2022 kern.info kernel: [ 862.419998] phy1-sta0: authenticated
Thu Oct 20 11:02:53 2022 kern.info kernel: [ 862.429671] phy1-sta0: associate with [AP_mac_here] (try 1/3)
Thu Oct 20 11:02:53 2022 kern.info kernel: [ 862.434061] phy1-sta0: RX AssocResp from [AP_mac_here] (capab=0x11 status=0 aid=2)
Thu Oct 20 11:02:53 2022 kern.info kernel: [ 862.436135] phy1-sta0: associated
Thu Oct 20 11:02:53 2022 daemon.notice netifd: Network device 'phy1-sta0' link is up
Thu Oct 20 11:02:53 2022 daemon.notice wpa_supplicant[2598]: phy1-sta0: Associated with [AP_mac_here]
Thu Oct 20 11:02:53 2022 daemon.notice wpa_supplicant[2598]: phy1-sta0: CTRL-EVENT-SUBNET-STATUS-UPDATE status=0
Thu Oct 20 11:02:53 2022 daemon.notice wpa_supplicant[2598]: phy1-sta0: Unknown event 37
Thu Oct 20 11:02:54 2022 daemon.notice wpa_supplicant[2598]: phy1-sta0: Unknown event 37
Thu Oct 20 11:02:55 2022 daemon.info hostapd: phy2-ap0: STA [STA_mac_here] IEEE 802.11: authenticated
Thu Oct 20 11:02:55 2022 daemon.info hostapd: phy2-ap0: STA [STA_mac_here] IEEE 802.11: associated (aid 2)
Thu Oct 20 11:02:55 2022 daemon.notice wpa_supplicant[2598]: phy1-sta0: Unknown event 37
Thu Oct 20 11:02:55 2022 daemon.notice hostapd: phy2-ap0: AP-STA-CONNECTED [STA_mac_here] auth_alg=open
Thu Oct 20 11:02:55 2022 daemon.info hostapd: phy2-ap0: STA [STA_mac_here] WPA: pairwise key handshake completed (RSN)
Thu Oct 20 11:02:55 2022 daemon.notice hostapd: phy2-ap0: EAPOL-4WAY-HS-COMPLETED [STA_mac_here]
Thu Oct 20 11:02:56 2022 daemon.notice wpa_supplicant[2598]: phy1-sta0: Unknown event 37
Thu Oct 20 11:02:57 2022 kern.info kernel: [ 866.439298] phy1-sta0: deauthenticated from [AP_mac_here] (Reason: 15=4WAY_HANDSHAKE_TIMEOUT)
Thu Oct 20 11:02:57 2022 daemon.notice netifd: Network device 'phy1-sta0' link is down
Thu Oct 20 11:02:57 2022 daemon.notice wpa_supplicant[2598]: phy1-sta0: CTRL-EVENT-DISCONNECTED bssid=[AP_mac_here] reason=15
Thu Oct 20 11:02:57 2022 daemon.notice wpa_supplicant[2598]: phy1-sta0: WPA: 4-Way Handshake failed - pre-shared key may be incorrect
Thu Oct 20 11:02:57 2022 daemon.notice wpa_supplicant[2598]: phy1-sta0: CTRL-EVENT-SSID-TEMP-DISABLED id=6 ssid="WDS_AP_name" auth_failures=2 duration=20 reason=WRONG_KEY
Thu Oct 20 11:02:57 2022 daemon.notice wpa_supplicant[2598]: BSSID [AP_mac_here] ignore list count incremented to 2, ignoring for 10 seconds
Thu Oct 20 11:03:19 2022 daemon.notice wpa_supplicant[2598]: phy1-sta0: CTRL-EVENT-SSID-REENABLED id=6 ssid="WDS_AP_name"
I have no other problem apart from this. Thanks as always...
Don't mix up 802.11r and 802.11s. That are two completely different things. 802.11s is a "selfmanged" layer 2 wifi-network. It does not manage client roaming. As I assume that both routers are connected by wire to your main/isp router, mesh would do nothing for you (wired is always better). First I would ensure that basic roaming is working properly (e. g. clients are changing from 5G to 2G and from router 1 to router 2). If this is working then you can dive into 802.11r aka "fast" roaming.
@robimarko i noticed your 6.1 branch a couple of things and thank you for sharing
thank you
I will stop even pushing development stuff as this is getting annoying.
Its only for development, no for use at all, so I dont see why somebody always has to try and then report that it does not work.
And yeah, 6.1 should be the next LTS as there is no chance of 6.2 being released this year
ok ... i will stop reporting but just keep pushing it ...let me know if you need a tester of 6.1
Indeed Robi - don't stop pushing your work.
Maybe instead add a small description to the readme file explaining what's your intention for that particular work (i.e. kernel 6.1 initial investigations, don't even bother reporting issues)?
Why should he/she comment his/her work for everybody? Maybe on a daily base!? Just a waste of time like the hunting for the latest stuff (imo). ;D
Yesterday I disabled ipv6 and I started getting' general failure' when pinging and errors related to MAC layer. My client devices are not using ipv6 yet having it enabled on the router made a difference (it could be that turning off ipv6 via luci had turned off something else)
Maybe you should read what Robi wrote and stop being so provocative?
I never requested daily/weekly commits from Robi 
He was annoyed that people were testing / provide feedback to his commits to branches that he was not expecting feedback (i.e. 6.1 kernel)
So it was his choice to commit it and I only suggested a relatively simply resolution how this could be resolved/improved.
I think a lot of people learn from Robi here, and even his investigation code could be useful
@Ansuel
Hi,
I have noticed some problem with your commits to increase memory on top of newest robimarko's -pr- repo.
Everytime I try to boot from memory this error occurs:
[ 1.742189] ubi0: attaching mtd12
[ 1.748390] ubi0 error: ubi_attach: bad image sequence number 155267035 in PEB 1, expected 1696626347
[ 1.751103] Erase counter header dump:
[ 1.760260] magic 0x55424923
[ 1.763892] version 1
[ 1.767623] ec 13
[ 1.770677] vid_hdr_offset 2048
[ 1.773614] data_offset 4096
[ 1.776999] image_seq 155267035
[ 1.780218] hdr_crc 0x4239f487
[ 1.783771] erase counter header hexdump:
[ 1.799639] Waiting for root device /dev/ubiblock0_0...ttach mtd12, error -22
Bootm initframs is without problems with free flash mem ~178 MB
Can you please look into it or hint me what's going on ?
ok if things are broken on that level.... i will get crazy....