Re: [PATCH] PM: Use hibernate flows for system power off

[Mario Limonciello <superm1@kernel.org>]

On 5/18/2025 3:24 AM, Martin Steigerwald wrote:
> Hi.
> 
> Thanks for your reply, Mario.
> 
> Mario Limonciello - 14.05.25, 19:06:03 CEST:
>>> I will eventually see I bet.
>>>
>>> Currently I only compile my own kernel for my current ThinkPad which
>>> is not affected by this issue.
> 
>> As a debugging tactic for your problem you can try to save your shutdown
>> log to the EFI pstore by adding this to your kernel command line for a
>> boot.
>>
>> efi_pstore.pstore_disable=N printk.always_kmsg_dump=Y
>>
>> Then the next boot if you have the systemd-pstore service enabled it
>> will move the log into /var/lib/systemd/pstore.
>>
>> If you don't have it enabled you can run this to manually do it one
>> time.
>>
>> sudo systemctl start systemd-pstore.service
>>
>> Hopefully that log will be helpful in identifying your problem.
> 
> Thanks for the hint on how to debug things like this without some serial
> console or so.
> 
> As I use Devuan and thus do not have Systemd installed, I will need to
> find a different way to obtain the log from the pstore. But according to
> kernel documentation this seems to be easy enough¹:
> 
> mount -t pstore pstore /sys/fs/pstore
> 
> I can indeed mount it and there are a lot of "dmesg-efi_pstore-
> [timestamp]" files where time stamp seems to be in seconds since beginning
> of 1970.
> 
> [1] https://docs.kernel.org/admin-guide/pstore-blk.html
> 
> Thanks,

Great.  As a word of warning - you will want to clear these out manually 
while debugging if you don't have something like systemd to do it for you.

Now that you can debug, does this patch series improve anything?  Or 
it's a completely separate problem you're seeing?