OpenWrt 22.03.0 first stable release

Apologies is this is a FAQ, I searched quite a bit but wasn't able to figure this out.

It's been a long while since I last upgraded between versions where sysupgrade wasn't explicitly supported.

I want to upgrade a Netgear R7800 from 19.07.10 to 22.03.0. The router also acts as AP for several WLANs, two of which are configured in WPA3/SAE mode.

So, two questions:

  • I've built an image with imagebuilder, specifying PACKAGES="hostapd-openssl -wpad-basic-wolfssl" to ensure WPA3/SAE AP will be supported, based on what I found in the docs. As of 22.03.0, is that still required for this use case, or can I safely use the default images from the downloads?

  • I understand that keeping configuration won't be supported, but I'm not clear on which image I should flash: Is it still the sysupgrade one or should I use the factory one this time?

:grin: What? Have never done anything like that before.

Maybe you are happy with your setup. Do you have 5GHz Band un use, also WPA3? Also Unbound, Adblock and couple of Wifi Stations set up?
For me, I reverted back yesterday, since the new version was simply unusable.

The same config, the same config.init is build now for 21.02 and running stable.

Yes I have wpa3 sae, and use 5Ghz as well, run wireguard and run 3 different ssids on both bands per r7800. I also have fast roaming setup etc.

I do not have unbound or adblock. Maybe try disabling those?

But if you did not set the frequencies or governor I would try this:

echo ondemand > /sys/devices/system/cpu/cpufreq/policy0/scaling_governor
echo ondemand > /sys/devices/system/cpu/cpufreq/policy1/scaling_governor
echo 800000 > /sys/devices/system/cpu/cpufreq/policy0/scaling_min_freq
echo 800000 > /sys/devices/system/cpu/cpufreq/policy1/scaling_min_freq
echo 25 > /sys/devices/system/cpu/cpufreq/ondemand/up_threshold
echo 10 > /sys/devices/system/cpu/cpufreq/ondemand/sampling_down_factor

echo 800000 > /sys/devices/system/cpu/cpu0/cpufreq/scaling_min_freq
echo 800000 > /sys/devices/system/cpu/cpu1/cpufreq/scaling_min_freq

exit 0

If that works ok then just add it to the startup script (like I did).

Note I also had random reboots in the past (before 22.03) but went away with this startup script. I think more ppl suffer from that.

1 Like

I noticed the latency issues before I discovered the fix others had implemented. Latency was a huge issue for me on 5GHz especially with Apple devices requiring disabling and re-enabling device wifi and/or restarting the router. The fix was a huge breakthrough for my end-user experience. But I have maintained this fix in my configuration through many version upgrades and haven't attempted to run without in the last few years.

1 Like

I have never had random reboots.

And honestly: I dint understand how the settings for frequencies or governor are related to my "outages".
Moreover: I don't understand why they should be required just for 22.03 - and never before.

Another r7800 running 22.03 here with perfect stability... And using DNSmasq with the full oisd Blocklist on router.
I set CPU governer to 1.4ghz fixed.

I think that's just a weird error message. I don't use adguard but I think the instructions you linked are missing some details. There are no (AFAIK) default tables or chains in nftables, and without a hook those chains wouldn't get processed unless you jump/goto them. I would try something like:

nft add table nat
nft 'add chain nat prerouting { type nat hook prerouting priority -100; }'
nft 'add rule nat prerouting udp dport 53 ip saddr 192.168.1.0/24 dnat 192.168.1.1:53'

And why did you set this?

Both fw3 and firewall4 can create the correct rule using the UCI redirect provided in the wiki article. No need for native custom rules in this case.

The one in the section above it? It might be worth editing the article so it says it's for fw3 and fw4 and separating that from the iptables and nftables specific information. Presumably the user I was replying to was trying to use nftables specific rules because of that.

If I don't configure firewall rules, I can still use AdGuard Home normally, but there is only 192.168.1.1 client in the web management page of AdGuard Home, and no other client records are recorded, even if other clients surf the Internet normally and have advertisement filtering function.

My guess is that depending on the exact copy of the r7800 you have and depending on exactly how much cpu load there is, the cpu frequencies get throttled back too much which is causing bits to fall over then error detection inside the cpu panics, as it has no way to continue executing code and the only thing it can do is reboots the device.

So yours with the new release has maybe slightly less load leading to the observed behavior.

Regardless, you can do 5 things, which may stop the random reboots:

  1. revert back to the old version
  2. reboot every day, that may or may not stop all random reboots
  3. try what I posted above
  4. buy a new r7800 or different model all together and see if that works
  5. go back to stock

My suggestion is try number 3 as this is by far the most elegant solution and has helped many people before. Your mileage may vary though.

1 Like

I have realised that remote logging also sends the kernel logs to the remote server :slight_smile: funny.
This might be pure r7800 thread related... but we just started here, so I post the suspicious kernel logs here.
They occur every time, when the device was frozen:

Sep 30 21:52:35 router kernel: [   79.865706] br-lan: port 3(wlan0) entered blocking state
Sep 30 21:52:35 router kernel: [   79.870873] br-lan: port 3(wlan0) entered forwarding state
Sep 30 21:53:57 router kernel: [  162.097500] ath10k_pci 0001:01:00.0: Invalid VHT mcs 15 peer stats

I am building with ath10k-ct drivers.

it crashes already 162 seconds after startup?

Does someone know how to make DNS-over-HTTPS to make it work?? DNS are not resolving, tracert doesn't work. I had to remove the package completely in order to make it work. Is there any workaround for this beside disabling it?

Ok, it seems like these settings are doing the magic.
The devices are up and running since 2 days!

Shouldn't this be mentioned somewhere in the Wiki? Or even in the sources by default?

22.03 was useless - not usable at all. You need to know that I am running the same config baked into the image. The devices firmware is just upversioned / upgraded images. No new features.

21.02 was working fine, 22.0x was really trash, useless, without these settings!

echo ondemand > /sys/devices/system/cpu/cpufreq/policy0/scaling_governor
echo ondemand > /sys/devices/system/cpu/cpufreq/policy1/scaling_governor
echo 800000 > /sys/devices/system/cpu/cpufreq/policy0/scaling_min_freq
echo 800000 > /sys/devices/system/cpu/cpufreq/policy1/scaling_min_freq
echo 25 > /sys/devices/system/cpu/cpufreq/ondemand/up_threshold
echo 10 > /sys/devices/system/cpu/cpufreq/ondemand/sampling_down_factor

echo 800000 > /sys/devices/system/cpu/cpu0/cpufreq/scaling_min_freq
echo 800000 > /sys/devices/system/cpu/cpu1/cpufreq/scaling_min_freq

exit 0

Changing governor settings should have no impact whatsoever on stability unless your device is overheating with default settings.

1 Like

My guess is if you have a too low clock then cpu registers (or some memory thing) get unstable. I personally never tested it out though. I just changed it to what I have above, then never looked back as it did the magic (was with release 19)

22.03.1 has been released.

2 Likes