Qualcommax NSS Build

[bgcngm]

@qosmio can you tell if it’s worth to use NSS FW 12.x instead of 11.4 if mesh is not needed and disabled in my builds?

[hingbong]

Qualcommax NSS Build

Yup, it pretty much means: allow this node to forward client traffic from other nodes. You would also need to set mesh_ttl and mesh_element_ttl to 1 which makes the node to only communicate via direct connections. So basically you're looking at a star topology with 1-to-1 communication, which is what I have for my setup. This is currently how I have mine setup Root AP (Primary) config wifi-iface 'mesh' option device 'radio2' option encryption 'sae' option key 'your_password' …

I asked before

[sppmaster]

@qosmio

My ugly hack:

--- /rom/www/luci-static/resources/view/status/include/10_system.js	2024-02-27 17:00:31.000000000 -0500
+++ /www/luci-static/resources/view/status/include/10_system.js	2024-02-28 02:14:33.000000000 -0500
@@ -25,14 +25,16 @@
 		return Promise.all([
 			L.resolveDefault(callSystemBoard(), {}),
 			L.resolveDefault(callSystemInfo(), {}),
-			L.resolveDefault(callLuciVersion(), { revision: _('unknown version'), branch: 'LuCI' })
+			L.resolveDefault(callLuciVersion(), { revision: _('unknown version'), branch: 'LuCI' }),
+			fs.exec_direct('awk',[ '/%/{printf "Avg: %s Max: %s",$2,$3}', '/sys/kernel/debug/qca-nss-drv/stats/cpu_load_ubi'])
 		]);
 	},

 	render: function(data) {
 		var boardinfo   = data[0],
 		    systeminfo  = data[1],
-		    luciversion = data[2];
+		    luciversion = data[2],
+		    nssinfo     = data[3];

 		luciversion = luciversion.branch + ' ' + luciversion.revision;

@@ -64,7 +66,8 @@
 				systeminfo.load[0] / 65535.0,
 				systeminfo.load[1] / 65535.0,
 				systeminfo.load[2] / 65535.0
-			) : null
+			) : null,
+			_('NSS Load'),         nssinfo
 		];

 		var table = E('table', { 'class': 'table' });
--- /rom/usr/share/rpcd/acl.d/luci-mod-status.json	2024-02-27 17:00:31.000000000 -0500
+++ /usr/share/rpcd/acl.d/luci-mod-status.json	2024-02-28 02:07:55.000000000 -0500
@@ -31,6 +31,7 @@
 		"read": {
 			"cgi-io": [ "exec" ],
 			"file": {
+				"/usr/bin/awk": [ "exec" ],
 				"/bin/dmesg -r": [ "exec" ],
 				"/sbin/logread": [ "stat" ],
 				"/sbin/logread -e ^": [ "exec" ],

@qosmio Hi.
If you have spare time to look at these, for another "pretty" hack. The files have changed and this no longer works.
Thanks for your awesome work!

[wolfxr]

Can you guys confirm that ipv6 is working in main-nss?

[hingbong]

yes, using it

[joq3]

I’m running a NSS regular build from @AgustinLorenzo from a while back.
This is the version: LuCI Master (25.009.47396~b0f8155) / OpenWrt SNAPSHOT (r0-a279290) Kernel Version 6.6.69

Maybe the build is 6-8 months old or something. I really want to update to maybe fix some wifi issues I have with devices disconnecting and never connecting again until a Router-reboot.

Is it safe to update from this version to this: https://github.com/AgustinLorenzo/openwrt/releases/tag/ipq807x-nsswifi-2025-10-12-1405

Also I don’t remember if I have single boot or not.

Is it possible to make a complete backup of the router which I can restore 1 to 1?
Will the update remove any packages I might have installed.

[JuliusBairaktaris]

@qosmio as always thank you so much for your great work. I wanted to try running SElinux with nss enabled but firmware loading is blocked by the current policies. Could you add the policies for it?

[ 12.296182] qca-nss 39000000.nss: probe with driver qca-nss failed with error -12

\[ 12.236353\] audit: type=1400 ... avc: denied { create } for pid=991 comm="ln" name="qca-nss0.bin" ... tclass=lnk_file permissive=0

[ 12.525691] audit: type=1400 ... avc: denied { net_admin } for pid=959 comm="kmodloader" capability=12 ... tclass=capability permissive=0


[ 14.322673] ath11k c000000.wifi: failed to initialize qmi handle: -517

[davidkrocks]

I’m having trouble. Which is the best version to use for a stock firmware mx4300 to convert to openwrt with NSS?

[JoJo_N]

@OkieDoke

melessly using ChatGPT to help build a working config based on the documentation didn’t get me anywhere near a working setup, hence why I gave up on trying to use mesh11sd. In my mind, anything that package claims to be able to achieve must surely be achievable with tweaks without the need for a dedicated package, but that’s where things fell short -

Same here, tried to run mesh11sd following instructions and it totally messed up the entire working mesh network, where each node went dead. I had to reset each one, wasting a couple hours in the process.

[JoJo_N]

You might try other AI's, they might offer other helpful suggestions.

[qosmio]

@bgcngm

@qosmio can you tell if it’s worth to use NSS FW 12.x instead of 11.4 if mesh is not needed and disabled in my builds?

Personally, I would stick with 11.4 for stability not just for feature completeness. You can experiment with upgrading to newer versions but I would not recommend it to tbh.

@sppmaster

@qosmio Hi.
If you have spare time to look at these, for another "pretty" hack. The files have changed and this no longer works.
Thanks for your awesome work!

I'd forgot about this, was this the version I posted?

@JuliusBairaktaris

@qosmio as always thank you so much for your great work. I wanted to try running SElinux with nss enabled but firmware loading is blocked by the current policies. Could you add the policies for it?

SELinux is the bane of my existence ... I work (fight) with it almost daily for work. I had not even considered it for personal single user use like routers.

[sppmaster]

@qosmio

I'd forgot about this, was this the version I posted?

Wow your current Luci status page looks really good and polished.
The last code I saw/know of (I may have missed your newer code though) was in this post

@qosmio

My ugly hack:

--- /rom/www/luci-static/resources/view/status/include/10_system.js	2024-02-27 17:00:31.000000000 -0500
+++ /www/luci-static/resources/view/status/include/10_system.js	2024-02-28 02:14:33.000000000 -0500
@@ -25,14 +25,16 @@
 		return Promise.all([
 			L.resolveDefault(callSystemBoard(), {}),
 			L.resolveDefault(callSystemInfo(), {}),
-			L.resolveDefault(callLuciVersion(), { revision: _('unknown version'), branch: 'LuCI' })
+			L.resolveDefault(callLuciVersion(), { revision: _('unknown version'), branch: 'LuCI' }),
+			fs.exec_direct('awk',[ '/%/{printf "Avg: %s Max: %s",$2,$3}', '/sys/kernel/debug/qca-nss-drv/stats/cpu_load_ubi'])
 		]);
 	},

 	render: function(data) {
 		var boardinfo   = data[0],
 		    systeminfo  = data[1],
-		    luciversion = data[2];
+		    luciversion = data[2],
+		    nssinfo     = data[3];

 		luciversion = luciversion.branch + ' ' + luciversion.revision;

@@ -64,7 +66,8 @@
 				systeminfo.load[0] / 65535.0,
 				systeminfo.load[1] / 65535.0,
 				systeminfo.load[2] / 65535.0
-			) : null
+			) : null,
+			_('NSS Load'),         nssinfo
 		];

 		var table = E('table', { 'class': 'table' });
--- /rom/usr/share/rpcd/acl.d/luci-mod-status.json	2024-02-27 17:00:31.000000000 -0500
+++ /usr/share/rpcd/acl.d/luci-mod-status.json	2024-02-28 02:07:55.000000000 -0500
@@ -31,6 +31,7 @@
 		"read": {
 			"cgi-io": [ "exec" ],
 			"file": {
+				"/usr/bin/awk": [ "exec" ],
 				"/bin/dmesg -r": [ "exec" ],
 				"/sbin/logread": [ "stat" ],
 				"/sbin/logread -e ^": [ "exec" ],

If you really wanna get adventurous, /sys/devices/virtual/thermal has all the hardware temps. Didn't really bother monkeypatching these in. Ideally would need to properly average CPU 0-3, PHYA/B 0-1

➤ for i in /sys/devices/virtual/thermal/ther*/type; do awk -v temp="$(cat ${i%/*}/temp)" '{gsub(/(wcss-|-thermal)/,"");printf("%8s %.1f°C \n", $1, temp / 1000)}' $i; done|sort
    cpu0 53.2°C
    cpu1 53.9°C
    cpu2 53.6°C
    cpu3 53.6°C
    nss0 54.9°C
    nss1 53.9°C
   phya0 52.6°C
   phya1 51.6°C
   phyb0 52.9°C
   phyb1 52.9°C
 cluster 54.5°C
 nss-top 52.9°C

This has just been merged. Although I don't see those errors on any NSS build with arh11k. With ath10k they were pretty frequent though.

ath11k: fix transmit queue flushing with flush_sta implementation

[JuliusBairaktaris]

Hahahaha, it came to hunt you in your personal life as well

[vochong]

@qosmio murmuring…. “I tried to fix SELinux once.”
“What happened?”
“Now it’s fixing me.”

Cheers!

[vochong]

Yes, I have only seen these errors with ath10k in the past (which were fixed in a recent commit), but never with ath11k.

[reflector]

your status page looks pretty mind sharing the changes you did for that?

[davidkrocks]

Quick Questions.

1. Can I keep one partition with Linksys stock image and then have openwrt for the second?
2. If the router is still stock (on both partitions) Do I have to install openwrt vanilla first and then sys upgrade it to qosmio image?
3. What software should I use for a windows PC to compile the qosmio needed to build the image?

[LS3434]

1. That will work if you want nss on snapshot, on stable i think there may be problems.
2. No, you can flash nss firmware from the start using factory image.
3. You can’t do it on Windows, you need linux for that.

[reflector]

re: 3, on windows you can use WSL.

[OpenWrt Wiki] Build system setup WSL

or install a virtualization software and run a linux distro to compile.

[Sinbios]

Yeah there's something strange going on with the libubus package that's breaking a lot of packages including Dawn. Hope it's fixed with the next build.