search

Dasharo / open-source-firmware-validation

OSFV infrastructure with automated tests and scripts for managing test results
Apache License 2.0
9 stars 1 forks source link

V5xTNx laptops need a powernap after executing `efibootmgr` in CNB001.001 #431

Closed philipandag closed 1 month ago

philipandag commented 2 months ago

Device

V540TND, V560TNE

RTE version

v0.9.1-rc1

OSFV version

fix-novacustom-v5xtnx-configs-powerctrl

Affected component(s) or functionality

CNB001.001 test

Brief summary

The test fails and then the device is bricked for about 30 minutes

How reproducible

seems to appear every time

How to reproduce

Run CNB001.001

Expected behavior

The test should pass and the device should not become bricked

Actual behavior

The device enters a weir state where it seems it is bricked. The keyboard and control leds work fine but it never boots. Pressing power button causes the device to power off immediately suggesting it did not leave firmware

Link to screenshots or logs

V54 log: https://cloud.3mdeb.com/index.php/s/ba2tz85pH3XyTEC V56 log: https://cloud.3mdeb.com/index.php/s/TRYzeBPXc3LEXcd

V54 dasharo-security: https://cloud.3mdeb.com/index.php/s/GD9Gmni6XBLFg7Q V56 dasharo-security: https://cloud.3mdeb.com/index.php/s/WPtcyE6772dLXKG

Additional context

It is hard to reproduce because the laptops become unusable for about 30 minutes after the test. Both eventually managed to boot and work normally after performing multiple cmos resets and then leaving them on for about 30 minutes.

The described behavior also occured when performing dasharo-security.measured-boot, so it may not be connected to osfv. The V54 took the 'power nap' after MNE004.001 PASSed, the V56 survived for longer and become unresponsible after MNE002.001 test.

It may be worth mentioning that during this bricked-like state the laptops get pretty hot.

Solutions you've tried

Still haven't tried doing anything about it. Just finding out if it is reproducible takes a long time.

philipandag commented 2 months ago

@macpijan that is an interesting behavior that you might want to know about

mkopec commented 2 months ago

The cause has been narrowed down to commit https://github.com/Dasharo/open-source-firmware-validation/pull/455/commits/a539822bb98268ffd552f1dcc83709fd5d15d0f4 , not sure how to fix it correctly yet

philipandag commented 2 months ago

@mkopec This commit a539822 does not appear on the branch on which I had the issue. Are you sure it is the cause of the problems?

philipandag commented 1 month ago

I have been running tests on V540TND on v0.9.1-rc5 for half a day yesterday and the bricks doesn't seem to happen now, even during aggressive rebooting and stress testing. I will report if any bricks happen again.