Belkin RT3200/Linksys E8450 WiFi AX discussion

Already present at 22.03.0-rc1 release candidate.

2 Likes

Is anyone has a same problem kn 22.03 rc1 wireless 2.4 disconnect intermittentence my Brother has a problèm but mine no thanks

1 Like

I am about to buy this router, is the snapshot version stable for 2.4ghz and 5ghz wifi? I am going to use OpenWrt especially for SQM

Yup, I haven't run into any stability problems yet and run the latest snapshot. Tend to update it approximately every week.

1 Like

The most famous mass problem on this band is described here.

1 Like

yes i has write but i didi'nt resolve again the problem

the all device disconnect to 2.4

camera iphone portable and tv ... thaks for your help

Made a thermal fix patch for our CPU if you build from source:

--- a/arch/arm64/boot/dts/mediatek/mt7622.dtsi
+++ b/arch/arm64/boot/dts/mediatek/mt7622.dtsi
@@ -156,18 +156,6 @@
 			thermal-sensors = <&thermal 0>;
 
 			trips {
-				cpu_passive: cpu-passive {
-					temperature = <47000>;
-					hysteresis = <2000>;
-					type = "passive";
-				};
-
-				cpu_active: cpu-active {
-					temperature = <67000>;
-					hysteresis = <2000>;
-					type = "active";
-				};
-
 				cpu_hot: cpu-hot {
 					temperature = <87000>;
 					hysteresis = <2000>;
@@ -183,18 +171,6 @@
 
 			cooling-maps {
 				map0 {
-					trip = <&cpu_passive>;
-					cooling-device = <&cpu0 THERMAL_NO_LIMIT THERMAL_NO_LIMIT>,
-							 <&cpu1 THERMAL_NO_LIMIT THERMAL_NO_LIMIT>;
-				};
-
-				map1 {
-					trip = <&cpu_active>;
-					cooling-device = <&cpu0 THERMAL_NO_LIMIT THERMAL_NO_LIMIT>,
-							 <&cpu1 THERMAL_NO_LIMIT THERMAL_NO_LIMIT>;
-				};
-
-				map2 {
 					trip = <&cpu_hot>;
 					cooling-device = <&cpu0 THERMAL_NO_LIMIT THERMAL_NO_LIMIT>,
 							 <&cpu1 THERMAL_NO_LIMIT THERMAL_NO_LIMIT>;
4 Likes

For those having their 5GHz network disappear occasionally, making the 5GHz band only WPA3-SAE rather than mixed has solved all my issues (so far).

I have not had the 5GHz network disappear since making it only WPA3.

1 Like

good news thanks the mine don't missed but not connection internet

you think i can try to change to wpa3 with 2.4 ?

@nbd If it helps, I made today more tests to detect the cause of non-functional wed in the case of Belkin RT3200/Linksys E8450. I compiled a new master build starting with the default settings then added bridger, luci-ssl, wpad-wolfssl (instead of wpad-basic-wolfssl), dnsmasq-full (instead of dnsmasq + odhcpd-ipv6only), luci-upnp, irqbalance. After configuring the PPPoE connection, wireless, wed_enable etc, the tests (basically watching if ppe is active with cat /sys/kernel/debug/mtk_ppe/bind and monitoring CPU usage) showed that:

  • PPPoE HW offloading works fine for Ethernet clients: CPU usage is close to 0% during NAT (internet) speed tests.
  • wed is working between an Ethernet client and a WLAN client, so bridger works with wed during file transfers between clients.
  • wed is not working between internet and a WLAN client, meaning no HW offloading when doing NAT (internet) speed tests with WLAN clients (no ppe bindings and CPU usage over 50-60%).
    Note that I only have PPPoE WAN connection, so I cannot test if wed for NAT can work with other types of connections like DHCP.

Then I made a new build with my usual config, but without the 726ef8ba2dbe4d4a693c4d9300bc69e234e6d67d commit (basically deleting the 704-01, 704-02 and 704-03 patches and replacing the 650 patch with the previous version) then I applied the patch from https://github.com/MeIsReallyBa/openwrt/blob/master/target/linux/mediatek/patches-5.10/999-fix-pppoe-flowoffload.patch adapted to the 5.15 kernel. I also used firewall3 instead of firewall4 until miniupnpd work will be finalized but I suppose this is not making any difference to the offloading path. The results showed that wed is working now between internet and WLAN clients, so I think the conclusion is that the 726ef8ba2dbe4d4a693c4d9300bc69e234e6d67d commit breaks wed for at least PPPoE NAT connections.

5 Likes

To enable wed do you need to only install bridger and set the wed enable flag or is some config for bridger needed?

I have edited the file /etc/config/bridger enabling the commented line, so I have just list blacklist eth0. Wed can be enabled by editing /etc/modules.d/mt7915e file and append wed_enable=Y to the existing "mt7915e" text, then reboot. To check if hw offloading works, just cat /sys/kernel/debug/mtk_ppe/bind during transfers. Following the MAC addresses you can precisely say which path is accelerated, so when wed works you'll see MACs from WLAN clients.

2 Likes

I have not actually ever had the 2.4GHz network disappear or drop.
I have it in WPA2 PSK mode (not mixed)

I do have software/hardware offloading off at the moment as it was unstable for me.

Just updated to RC1 build. Just to confirm, does this build have the fixes for working hardware offloading? Or do we still need to use a more recent snapshot?

It does not. You can see the commit history for the 22.03 branch here: https://git.openwrt.org/?p=openwrt/openwrt.git;a=shortlog;h=refs/heads/openwrt-22.03

As you can see, 22.03RC1 was tagged before the HW offload fixes.

3 Likes

I noticed in the link @Mushoz provided that the latest commit has been tagged with openwrt 22.03 in green, would this mean that the stable release has been forked?

1 Like

No, that is just the Git Head (latest commit) of the 22.03 branch. It's not a tag.

1 Like

Not quite sure of your terminology, but yes, the stable openwrt-22.03 branch has been created/forked from "master" branch in March. 22.03.x releases will be made from that branch.

But like Mushoz says, the green label indicates a branch head, not a tag. Tags are applied just to the actual fixed release commits, so there is a tag for the 22.03.0-rc1 on 20 April in the branch history.

Note that development master has already progressed forward, so 22.03 and master are deviating.

2 Likes

@Gingernut: What is your "themal fix patch" fixing? Isnt "cpu-hot" the default state after you removed "active" & "passive"? Doesnt it make sense to slow down the cpu in idle so its not so hot (my patches abouve are for 5.10)
Beside that, is kernel 5.15 running okay? Ihave still not updated from 5.10

I’m not sure if you’ve read this discussion or not but it explains it. I just ported the patch to our SOC.

https://github.com/openwrt/openwrt/issues/9396