Well, there is no NSS in my build (which this thread is about).
My bad... I'm exploring a few options and just hit the "quote & reply" button. Sorry about that.
1 Like
Hi,
please update kernel to latest lts versions of both builds.
reason:
https://lwn.net/ml/oss-security/20221013101046.GB20615@suse.de/
thank you for ur hard work.
You have overlooked that those fixes were backported to OpenWrt already 7 days ago. The kernel version bump has no relevance here (as we use a separately handled wifi driver infra).
E.g. 22.03: https://git.openwrt.org/?p=openwrt/openwrt.git;a=commit;h=f1de43d0a045a154c74281bc60bf1c44c990071b
Both my latest builds already have the fix.
2 Likes
Iam sorry, my bad.
thank you & have a nice day!
I know this is for r7800 only but i wonder if someone here have a nbg with a serial cable attached... currently master bootloops on such device and i desperately need a log to understand what is wrong with them.
Can I safely flash between this build and others using the GUI, or do I need to go back to a TFTP flash like I used when converting from OEM firmware? Do I need to start with the factory image or can I go straight to sysupgrade since I'm on OpenWRT already?
I am on a Gigabit connection and would like to test speed difference between this build, official, and the Ipq806x NSS build from ACwifidude.
You should be able to use normal sysupgrade.
1 Like
The current kernel 5.15 in master might cause some problems for R7800.
In the last few weeks, I have twice experienced a lockup, where R7800 gets unresponsive for both wired and wifi.
Only cold boot / power-off helps.
Even the serial log gets stuck.
On the first time I managed to get this log from serial:
[313128.262820] rcu: INFO: rcu_sched detected stalls on CPUs/tasks:
[313128.262867] rcu: 0-...0: (3 ticks this GP) idle=103/1/0x40000000 softirq=850534/850535 fqs=1635235
[313128.267912] (detected by 1, t=3904897 jiffies, g=1677093, q=158971)
[313128.277022] Sending NMI from CPU 1 to CPUs 0:
I have documented the mishap to the ipq806x 5.15 fix PR (https://github.com/openwrt/openwrt/pull/11032#issuecomment-1294581828), but I mention it also here in case somebody else noticed the same and could give some new debug info.
(I have not seen similar stall/lockup in the previous 6 years with R7800, so this is something new)
7 Likes
I use the default ondemand governor, like I have been doing since 2016 
(I try to keep near the defaults in general)
3 Likes
Thanks. How much does it load the CPU?
Dear @hnyman Can I use this build for Zyxel Armor Z2 ?
Thank you
I only build for R7800.
I have no idea about Zyxel Armor Z2.
@hnyman I like the new version with the stability patch, but I'm curious, what is this patch about?
1 Like
Tentative patch for ipq806x kernel 5.15 from @ansuel, PR11173
I have tested a bit 5.15 build with and without patch and stability might not be the only problem. Namely I can see degrading WiFi performance significantly:
- 5.10 build with WiFI iperf3 rate at 655 Mbits/sec (System load: 0.11, 0.08, 0.13)
- 5.15 build - iperf3 over WiFI rate falls down to 177 Mbits/sec (System load: 0.55, 0.56, 0.39)
Tested WiFi (5GHz) to copper. Copper to copper performance seems not to be affected.
Getting at least one random reboot a day with the latest builds, including the stable 22.0 r19863. I tried to see if there's anything in logs (I gather all logs with graylog, but there's nothing, last log before the crash was from ntpd notice about stratum change, then 20mins later the fresh startup).
Any idea why could this be? I update your build about once a month and it's been rock solid last year+
In master the 5.15 had new CPU frequency (cache) scaling that apparently may cause instability.
But that has no relevance to 22.03 which has not seen major changes for ages.
The logging likely gets broken in a crash. But you should check the kernel crash log feature "pstore" where crash log in RAM survives a reboot. (But gets cleared in a power-off or by deleting files manually)
Check the possible logs in /sys/fs/pstore
1 Like