Re: lockups on lenovo p43s under current

On 5/11/20 4:21 PM, Pete Wright wrote:
>
>
> On 5/11/20 3:28 PM, Yuri Pankov wrote:
>> Pete Wright wrote:
>>> hello,
>>> i have a lenovo thinkpad P43s that exhibits lockups under CURRENT 
>>> but behaves fine when running STABLE.  i've tried to find a fully 
>>> reproducible situation to get this system to lockup but haven't 
>>> found anything yet.  i am starting to suspect that the changes 
>>> implemented in this review may be the issue though:
>>>
>>> https://reviews.freebsd.org/D23728
>>>
>>> my reasoning is that i've observed issues when:
>>> - removing AC power from the laptop, or inserting AC power
>>> - when the system display has gone to sleep
>>> - randomly hanging during boot with this as last line:
>>> battery0: battery enitialization start
>>>
>>> unfortunately while the above seem to be cases where this has 
>>> happened i haven't been able to %100 reproduce yet.
>>>
>>> so my first question is - would it be possible to just revert the 
>>> changes in that diff, or has too much time gone past to just back 
>>> out that single change.  alternatively, is there any debugging 
>>> information i can get on my end that might help figure out what the 
>>> root cause is?
>>
>> Not really what you are asking, but it's possible to disable ACPI 
>> subdevices, so you could check if disabling cmbat completely helps 
>> and it's indeed the suspect:
>>
>> debug.acpi.disabled="cmbat"
>
> Thanks Yuri,
> So I was able to boot my system once via batter with this set, but 
> unfortunately it crashed after I tried to suspend/resume. Realizing 
> that was a bit optimistic I attempted to reboot the system and wasn't 
> able to get it to fully boot after several attempts.
>
> I believe what the next step at this point is checkout the code right 
> before this commit and see if I can get it to successfully boot.  I'll 
> report back if I find anything after that test.
>

To follow-up on this I believe the updates in the above review may be 
the culprit.  What I have done is built a memstick.img set to the commit 
right before the changes in D23728 were merged.  running this image I 
can boot my system, disconnect and reconnect AC power without any issues.

i then booted from a memstick using the latest snapshot of current. i 
can disconnect AC power without issues, but reconnecting hangs the 
system immediately.

i've tested this a couple times and it seems pretty reproducible, not 
sure what the best next step would be though.  would someone here be 
willing to help me debug this, or would it be best to file a PR along 
with a dmesg and output from acpiconf?

cheers!
-pete

-- 
Pete Wright
pete_at_nomadlogic.org
_at_nomadlogicLA