That's promising news. I can say the same on dl-wrx36, mx5300, mx4300. So now AX3600 added to that list. I think the only other boards that had concerns was the Redmi AX6. I will try reintroducing that patch.
Thanks, I've included it as well:
commit 0e52009e19f2d29c921b37d7d0104b7d3b8e76a8 (HEAD -> qualcommax-6.x-nss-wifi)
Author: AgustinLorenzo <agustin.lorenzo@thinco.es>
AuthorDate: Sun Oct 20 01:30:15 2024 +0200
Commit: Sean Khan <datapronix@protonmail.com>
CommitDate: Sat Oct 19 20:50:03 2024 -0400
qualcommax: ath10k-ct: bump patch for 6.7 to 6.10
The patches to get threaded NAPI for qca-nss-drv are independent of the wireless drivers. I think my branch naming could have been better. NAPI (New API) is just the event handling mechanism used by networking stack. The patches were meant to try and get the drivers in threaded mode. The wireless drivers have been fairly stable in threaded mode. However, NSS driver did not play nicely, so just reverted the whole branch to avoid confusion.
CONFIG_PACKAGE_ath10k-firmware-qca9984=y
CONFIG_PACKAGE_kmod-ath10k=y
# CONFIG_PACKAGE_ath10k-firmware-qca9984-ct is not set
# CONFIG_PACKAGE_ath10k-firmware-qca9984-ct-full-htt is not set
# CONFIG_PACKAGE_ath10k-firmware-qca9984-ct-htt is not set
# CONFIG_PACKAGE_kmod-ath10k-ct is not set
# CONFIG_PACKAGE_kmod-ath10k-ct-smallbuffers is not set
# CONFIG_PACKAGE_kmod-ath10k-smallbuffers is not set
Don't build with the -ct version. Use the regular version as that's the one with ethernet frame capability, and more performant when coupled with NSS offload for mac80211.
I did a reset via the LUCI interface. Then I did the following steps:
The wireless was still not working (with the same behaviour).
Run the following commands:
root@OpenWrt:~# uci set dhcp.lan.ignore=0 && uci commit && service dnsmasq restart
udhcpc: started, v1.36.1
udhcpc: broadcasting discover
udhcpc: no lease, failing
root@OpenWrt:~# logread -f
Sun Oct 20 09:21:59 2024 daemon.info hostapd: phy0-ap0: STA DEVICE_MAC_ADDRESS IEEE 802.11: authenticated
Sun Oct 20 09:21:59 2024 daemon.info hostapd: phy0-ap0: STA DEVICE_MAC_ADDRESS IEEE 802.11: associated (aid 1)
Sun Oct 20 09:21:59 2024 daemon.notice hostapd: phy0-ap0: AP-STA-CONNECTED DEVICE_MAC_ADDRESS auth_alg=open
Sun Oct 20 09:21:59 2024 daemon.info hostapd: phy0-ap0: STA DEVICE_MAC_ADDRESS RADIUS: starting accounting session 88831202481356AF
Sun Oct 20 09:22:17 2024 daemon.notice hostapd: phy0-ap0: AP-STA-DISCONNECTED DEVICE_MAC_ADDRESS
I rebooted, but got the same problem:
root@OpenWrt:~# logread -f
Sun Oct 20 09:21:10 2024 daemon.info hostapd: phy0-ap0: STA DEVICE_MAC_ADDRESS IEEE 802.11: authenticated
Sun Oct 20 09:21:10 2024 daemon.info hostapd: phy0-ap0: STA DEVICE_MAC_ADDRESS IEEE 802.11: associated (aid 1)
Sun Oct 20 09:21:10 2024 daemon.notice hostapd: phy0-ap0: AP-STA-CONNECTED DEVICE_MAC_ADDRESS auth_alg=open
Sun Oct 20 09:21:10 2024 daemon.info hostapd: phy0-ap0: STA DEVICE_MAC_ADDRESS RADIUS: starting accounting session C86CD4E357725B54
Sun Oct 20 09:24:15 2024 daemon.notice hostapd: phy0-ap0: AP-STA-DISCONNECTED DEVICE_MAC_ADDRESS
Attempted to update radio configurations:
root@OpenWrt:~# uci set wireless.default_radio0.dtim_period='1'
root@OpenWrt:~# uci set wireless.default_radio0.max_inactivity='86400'
root@OpenWrt:~# uci set wireless.default_radio0.disassoc_low_ack='0'
root@OpenWrt:~# uci set wireless.default_radio0.wnm_sleep_mode='1'
root@OpenWrt:~# uci set wireless.default_radio1.dtim_period='1'
root@OpenWrt:~# uci set wireless.default_radio1.max_inactivity='86400'
root@OpenWrt:~# uci set wireless.default_radio1.disassoc_low_ack='0'
root@OpenWrt:~# uci set wireless.default_radio1.wnm_sleep_mode='1'
root@OpenWrt:~# uci set wireless.default_radio2.dtim_period='1'
root@OpenWrt:~# uci set wireless.default_radio2.max_inactivity='86400'
root@OpenWrt:~# uci set wireless.default_radio2.disassoc_low_ack='0'
root@OpenWrt:~# uci set wireless.default_radio2.wnm_sleep_mode='1'
Test again and same problem:
root@OpenWrt:~# logread -f
Sun Oct 20 09:27:28 2024 daemon.info hostapd: phy0-ap0: STA DEVICE_MAC_ADDRESS IEEE 802.11: authenticated
Sun Oct 20 09:27:28 2024 daemon.info hostapd: phy0-ap0: STA DEVICE_MAC_ADDRESS IEEE 802.11: associated (aid 1)
Sun Oct 20 09:27:28 2024 daemon.notice hostapd: phy0-ap0: AP-STA-CONNECTED DEVICE_MAC_ADDRESS auth_alg=open
Sun Oct 20 09:27:28 2024 daemon.info hostapd: phy0-ap0: STA DEVICE_MAC_ADDRESS RADIUS: starting accounting session 05BCA808F23AE775
Sun Oct 20 09:27:46 2024 daemon.notice hostapd: phy0-ap0: AP-STA-DISCONNECTED DEVICE_MAC_ADDRESS
I rebooted and got the same problem:
root@OpenWrt:~# logread -f
Sun Oct 20 09:30:27 2024 daemon.info hostapd: phy0-ap0: STA DEVICE_MAC_ADDRESS IEEE 802.11: authenticated
Sun Oct 20 09:30:27 2024 daemon.info hostapd: phy0-ap0: STA DEVICE_MAC_ADDRESS IEEE 802.11: associated (aid 1)
Sun Oct 20 09:30:27 2024 daemon.notice hostapd: phy0-ap0: AP-STA-CONNECTED DEVICE_MAC_ADDRESS auth_alg=open
Sun Oct 20 09:30:27 2024 daemon.info hostapd: phy0-ap0: STA DEVICE_MAC_ADDRESS RADIUS: starting accounting session 174A32D78018BD68
Sun Oct 20 09:30:45 2024 daemon.notice hostapd: phy0-ap0: AP-STA-DISCONNECTED DEVICE_MAC_ADDRESS
Does the LUCI reset do a clean install, or should I try another approach?
Having to install the packages always after each upgrade is normal when running sysupgrade in openwrt, and the change of country code to US, is something that I set in each upgrade in the uci-defaults.
I am redo the repos over the latest commit from OpenWRT.
Changelog (include upstream of OpenWRT):
Updated NSS kernel and software to version 6.6.57.
Added the latest changes developed by @qosmio for NSS.
Updated the ath11k FW to version 2.9.0.1-02146.
Including the patch to eliminate the reduction to 2KB (originally it is at 4KB) in RX header, the RAM consumption could increase very slightly in some cases, in my case it stays in the range of 185-195 MB free (my device is AX3600).
Notes:
Compilation is performed with the following limits for ath11k (512M) and NSS (NSS_MEM_PROFILE_HIGH) and pbuf set to auto by default.
The build is compiled with GCC 13.3.
The default TCP congestion protocol is CUBIC.
By default, the radio0 (IoT) is off (you can enable in LUCI) and the radio 1 (2.4 GHz) y 2 (5 GHz) is on.
The default country code for the WiFi is USA to avoid problems with 160 MHZ because it does not work if it does not have a specified country code.
NOTE: Now I only compile the build for the following devices AX3600, AX6, AX9000, Dynalink DL-WRX36, Buffalo WXR-5950AX12, QNAP QHora-301W, Netgear RAX120v2, Arcadyan AW1000, Zyxel NBG7815, Linksys MX4200 (v1 and v2), Netgear SXR80, Netgear SXS80, Spectrum SAX1V1K and Netgear WAX218 if someone has used my versions with other devices (and it is supported), let me know and I will add it.
It would be good if you included Linksys MX4300 as well. There is a whole bunch of folks in another thread looking for regularly updated development builds.
@qosmio , in vlan setup mx4300 dumb ap,over wifi, the vlan that the client is connected cannot access other vlan management interface ip of same AP(vlans configured as wan.vlan1, wan.vlan2 etc and bridged as bridgevlan1,bridgevlan2 and subsequently trunked to router) .
i have dhcp client under 'network' on both interfaces for both bridges to get 2 management ips.
browser access for wifi connected vlan1 cannot access management ip of vlan 2 (router firewall access is correct and wired devices on vlan1 can connect mgmt url of vlan2). when i roam to another AP2 mx4300 wifi(same vlan as ap1), then i can connect to mgmt vlan 1 ip of AP1 but not its own ).
there seems to be issue with wifi connected device on vlan setup not being able to work correctly with other vlans. are there additional setups to be done?
I used version 2024-10-20-0126 (mesh version) and now I can connect to the wifi without security, when I tried to set up WPA2-PSK or WPA3-SAE it does not work
Also I tried to test the vlan's and it also does not work. My use case is for a dummy AP using vlan filtering, so I should disable the dnsmask, firewall and odhcpd. Regarding the vlan I get the following error:
DHCP packet received on br-lan.XXX which has no address
This sounds a lot like what I was experiencing in my attempts to bring VLANs to wifi over the weekend. When I have my SSID connected to br-lan then all assignments IP assignments came from the management VLAN.
I attached the individual VLAN interfaces and while I can see the VLAN assignments happening via RADIUS, nobody get an IP as of right now - WIP I guess.
What's interesting though, is that the system logs are spitting out some traces that might be useful for whatever is happening. I noticed these as well on the "working" setup (where everything worked as long as you really wanted the management vlan)
Of course, I'm also doing RADIUS assigned dynamic VLANs on a single SSID with WPA2-PSK, so I'm expecting to be banging up against the limits of what this thing will allow...
There were reports from a user of it causing kernel panics... but honestly could have (and probably was) an error on their part building with other patches I don't have in my repo.
You won't find typical methods work with NSS. NSS bypasses a lot of the network stack from the kernel, thus certain metrics won't be captured in many monitoring tools (nDPI for example). nlbwmon is about the only one I know, but that's not per client, and only per network protocol.
@qosmio
Hi,
I'm trying to porting a device (tp-link tl-er2260t), which using ipq8070a soc and qca8075 switch with 512M ram, my own build tested booting, but the switch does not working porperly.
kernel log:
[ 4.602336] GMAC1(ffffff8004048900) Invalid MAC@ - using fa:0e:0e:4e:17:bd
[ 4.605893] dp1: ppe offload disabled: 0 for macid 1
[ 4.612846] dp1: Switch attached to macid 1 status: 0
[ 4.809526] Qualcomm QCA8075 90000.mdio-1:00: attached PHY driver (mii_bus:phy_addr=90000.mdio-1:00, irq=POLL)
[ 4.810675] GMAC2(ffffff80042a2900) Invalid MAC@ - using 56:02:23:31:53:c2
[ 4.818421] dp2: ppe offload disabled: 0 for macid 2
[ 4.825293] dp2: Switch attached to macid 2 status: 0
[ 4.909091] Qualcomm QCA8075 90000.mdio-1:01: attached PHY driver (mii_bus:phy_addr=90000.mdio-1:01, irq=POLL)
[ 4.910173] GMAC3(ffffff80042a6900) Invalid MAC@ - using b6:ed:c7:62:61:c4
[ 4.917985] dp3: ppe offload disabled: 0 for macid 3
[ 4.924902] dp3: Switch attached to macid 3 status: 0
[ 4.989184] Qualcomm QCA8075 90000.mdio-1:02: attached PHY driver (mii_bus:phy_addr=90000.mdio-1:02, irq=POLL)
[ 4.990322] GMAC4(ffffff8003f14900) Invalid MAC@ - using 9a:ab:01:36:21:61
[ 4.998078] dp4: ppe offload disabled: 0 for macid 4
[ 5.004959] dp4: Switch attached to macid 4 status: 0
[ 5.089083] Qualcomm QCA8075 90000.mdio-1:03: attached PHY driver (mii_bus:phy_addr=90000.mdio-1:03, irq=POLL)
[ 5.090204] GMAC5(ffffff8003f10900) Invalid MAC@ - using e6:ee:38:33:a5:05
[ 5.097976] dp5-syn: ppe offload disabled: 0 for macid 5
[ 5.104860] dp5-syn: Switch attached to macid 5 status: 0
[ 5.110454] QCA SFP 90000.mdio-1:1d: attached PHY driver (mii_bus:phy_addr=90000.mdio-1:1d, irq=POLL)
[ 5.116673] GMAC6(ffffff8003d66900) Invalid MAC@ - using 2a:43:ff:e6:4e:21
[ 5.124812] dp6-syn: ppe offload disabled: 0 for macid 6
[ 5.131628] dp6-syn: Switch attached to macid 6 status: 0
[ 5.137179] QCA SFP 90000.mdio-1:1e: attached PHY driver (mii_bus:phy_addr=90000.mdio-1:1e, irq=POLL)
.....
[ 25.282689] br-lan: port 1(lan1) entered blocking state
[ 25.282749] br-lan: port 1(lan1) entered disabled state
[ 25.286776] nss-dp 3a001000.dp1 lan1: entered allmulticast mode
[ 25.292582] nss-dp 3a001000.dp1 lan1: entered promiscuous mode
[ 25.307458] br-lan: port 2(lan2) entered blocking state
[ 25.307524] br-lan: port 2(lan2) entered disabled state
[ 25.311617] nss-dp 3a001200.dp2 lan2: entered allmulticast mode
[ 25.317132] nss-dp 3a001200.dp2 lan2: entered promiscuous mode
[ 25.318997] ECM init complete
[ 25.330198] br-lan: port 3(lan3) entered blocking state
[ 25.331562] br-lan: port 3(lan3) entered disabled state
[ 25.336641] nss-dp 3a001400.dp3 lan3: entered allmulticast mode
[ 25.342498] nss-dp 3a001400.dp3 lan3: entered promiscuous mode
[ 25.353289] br-lan: port 4(lan4) entered blocking state
[ 25.353630] br-lan: port 4(lan4) entered disabled state
[ 25.358864] nss-dp 3a001600.dp4 lan4: entered allmulticast mode
[ 25.364470] nss-dp 3a001600.dp4 lan4: entered promiscuous mode
[ 27.244834] qca-nss 39000000.nss: Configuring additional NSS pbufs
[ 27.257534] qca-nss 39000000.nss: Additional pbufs of size 3100672 got added to NSS
[ 32.488742] l11: disabling
[ 42.968967] Qualcomm QCA8075 90000.mdio-1:00: Downshift occurred from negotiated speed 100Mbps to actual speed 10Mbps, check cabling!
[ 42.969037] nss-dp 3a001000.dp1 lan1: PHY Link up speed: 10
[ 42.980190] br-lan: port 1(lan1) entered blocking state
[ 42.985339] br-lan: port 1(lan1) entered forwarding state
as the log showed, the switch can only working at 10Mbs, and I found that the qsdk device tree has an option called cmnblk_clk = "external_50MHz", maybe this is the cause?
Hi, i'm coming from 28 sept. release of AugustinLorenzo ax3600 nss only and it was working well, i've now updated to this 20-oct release and my config stopped working:
the ax3600 should connect as a client to the main router one floor below using wpa2/3 on the 5ghz 80mhz band of the IPQ8074 (to have a faster backbone), and then as a router, provide internet on lan and on the 2.4ghz radio for the rest of my devices upstairs.
I also tried enabling the weaker QCA9887 on radio0 and that radio works, it connects as a client on 5ghz and i'm able to browse the internet, while the radio1 IPQ8074 connects but display a strange 6mbit/20mhz for both rx and tx, no ip and obviously no traffic.
Factory reset and forced AC and 80mhz (to have the same working conditions the QCA9887 has) but won't solve anything. One antenna in client mode does work, the other not.
