search

greearb / ath10k-ct

Stand-alone ath10k driver based on Candela Technologies Linux kernel.
111 stars 41 forks source link

9980 crash in scheduler (same as 54 but for 9980) #56

Closed greearb closed 5 years ago

greearb commented 5 years ago

Reported in OpenWRT forum by escalade.

Same root cause as bug 54, but for chipset 9980.

greearb commented 5 years ago

Please try this firmware. And, please run with debug-level of 0xc0000020 and send me 'dmesg' output after the system has been running for a bit even if it doesn't crash or have obvious issues.

firmware-5-full-community.bin.gz

ghost commented 5 years ago

I see some updates on #54 for 9984 regarding this bug but not here. If I'd like to test for 9980 (r7500v2) is the firmware above still relevant or is there a newer version? Do you still want dmesg output (log level 0xc0000020) after an hour or two up?

FWIW, similar to other reports I got 10+ days using 10.4b-ct-9980-fH-012-81e1edd. The router just crashed and reset (sorry no dmesg/log output due to the router reboot).

greearb commented 5 years ago

Let me try to get this fixed in bug #54, and when it seems to work there, I'll update this bug with a new firmware to test with.

greearb commented 5 years ago

Ok, I think I made some progress on bug 54, please try this image and send me logs if you see crashes or other problems.

firmware-5-full-community.bin.gz

ghost commented 5 years ago

I only got about an hour and half with 10.4b-ct-9980-fW-012-4651c335 before the router crashed and rebooted. I was remote logging on a wired connection (debug level 0xc0000020) and that file is here.

There is no indication of the crash that I can see.

If it helps, there is a dmesg and logread output I took before the crash right after booting here.

greearb commented 5 years ago

I do not see any obvious errors in the logs. The log messages are almost all about ERP being disabled and about short pre-amble being enable and disabled. Looks like the driver just sending commands to do these things, and probably it is no big deal. Perhaps the crash was not related to the firmware, or maybe the log captures just missed the important event.

ghost commented 5 years ago

thank you for responding. I'll try again with it and report back; however, I don't have a serial connection so if I keep getting unexplained crashes I'll stop testing for now.

ghost commented 5 years ago

a little over two days up now and no issues this time. I did run most of that time at the default debug level and intermittent remote logging. The last 3 hours or so, I used debug level 0xc0000020 with remote logging in the off chance that doing so precipitated the afore mentioned crash - no issues tho.

Unless there is a newer version you like to have tested, I'll let this go through the weekend when traffic is heavier and see how it does.

ghost commented 5 years ago

6+ days up now, no issues

greearb commented 5 years ago

Ok, I'm going to consider this one fixed. Please open a new bug if you see more problems.