Netgear WAX220 support (almost complete)

For Developers
162

I dont think so. I have two WAX220 and no devices have ever had conncectivity issues. Then again i have no Apple devices. Only speed issues with the uplink.

163

Yesterday I decided to flash again openwrt on my WAX220.
The last vendor firmware started to exhibit some nasty glitches with WMM with my yamaha AVR.
Well it seems stable, however there are an inconsistency between the AP and the router reported link rates.
ethtool on the router (ER-X) says 1gb full duplex
ethtool on the AP (WAX220) says 1gb unknown duplex (ERR 255).
iperf3 test is consistent with a 1gb full duplex speed.

164

Have you tried with snapshot to compare the behavior?

I just flashed mine to latest snapshot but not yet tested, can you show me how to reproduce the problem?

165

It could not be reproducible, because It could be hardware dependant:

That's why I and others suggested a note in the wiki of nwa50ax pro and wax220...

166

OK I looked into that Github comment, so what they mean is, WAX220 might have more than 1 type of LAN PHY and Realtek PHY is known to have problem? I might take a look of mine tonight.

167

Nope, it says that there are various revisions of realtek phy and some of them are problematic (the oldest one).

168

I just tested and my WAX220 has the same problem as well :frowning:

169

You could try buying a 2.5gb switch...
Otherwise you have to revert to the vendor firmware, which has a severe bug with WMM and probably with multicast to unicast.

170

Is it possible to force duplex? And I just tested with 100M and it's even worse, unable to connect at all.

171

You can using ethtool and running

ethtool -s eth0 speed 1000 duplex full autoneg on but the results show that it doesn't exactly fix the issue. See my post above about the speed tests when I tried it.

172

If you are affected you can only buy a 2.5gb switch or revert to stock firmware or wait for a fix by openwrt developers

173

@Flole @daniel could this be related to the issues with 10/100/1000? The below is when connected to a 1000 switch

[    3.278578] mtk_soc_eth 15100000.ethernet eth0: PHY [mdio-bus:06] driver [RTL8221B-VB-CG 2.5Gbps PHY] (irq=61)
[    3.288667] mt7986b-pinctrl 1001f000.pinctrl: Can not find pin
[    3.294480] genirq: Failed to request resources for mdio-bus:06 (irq 61) on irqchip mt-eint
[    3.302828] RTL8221B-VB-CG 2.5Gbps PHY mdio-bus:06: Error -22 requesting IRQ 61, falling back to polling
[    3.316157] mtk_soc_eth 15100000.ethernet eth0: configuring for phy/2500base-x link mode
174

Does the new openwrt update fix the issue?

175

No, it did not.

176

Thanks for posting this patch.

Does this let you negotiate 1Gbps duplex on your hardware?

I haven't seen any follow ups to this patch on this thread yet.

177

no the patch i provided is for the apple disconnect issue.

1 Like
178

I'm a little confused about this device.
Netgear claims it has 4*4 5 GHz radio, but I can only see three antennas and so is claimed by the OpenWrt wiki.
Are they doing false advertising? Which one is true?

179

It's false advertising. The radios are 3x3 for the 5GHz and 2x2 for the 2.4GHz.

180

And can it still achieve 3600 Mbps theoretical speed over 5 GHz in WiFi 6 or is it false too? I've seen some MCS tables but I'm not sure at which column to look at.
Can it operate at 160 MHz in 3x3 mode? Now I own a wireless router that can either do 2x2 at 160 or 4x4 at 80.

181

I believe it can theoretically achieve 3.6 Gbps over the 5GHz with 160MHz channel width, at least according to my OpenWRT install. I ran a speed test using iPerf3 and achieved 1.5Gbps from laptop to AP, but the radio in my computer can only go up to 2.4Gbps theoretical speed. I don't know if either device's CPU can truly achieve 2.4Gbps or faster in a speedtest so my test was flawed. The 2.4GHz can only do 40MHz channel width though on a 2x2.