Re: [PATCH] ACPI: battery: Re-read battery information in the complete callback

[johannes . goede]

Hi,

On 13-Oct-26 23:05, Ali Ahmet Memiş wrote:
> On Tue, Oct 13, 2026 at 07:20:55PM +0200, Rafael J. Wysocki (Intel) wrote:
>> Also, we don't actually know how many "bad" cases there are and so far
>> we have identified a few.
> 
> I wrote a script that extracts the ACPI tables from all 208 BIOS
> packages in Lenovo's ThinkPad catalog (370 models) and flags methods
> that switch the EC page selector (HIID) without BATM while other methods
> use it under BATM, and checked the flagged ones by hand.  Two BIOSes do
> it both with \GTST from _TMP: P1 Gen2 / X1 Extreme Gen2 and P53 / P73.
> Apart from those, NVPS on T450 / T450s does it from the NVIDIA GPS
> _DSM, which only the out-of-tree NVIDIA driver evaluates.  This doesn't
> cover races like the UCSI one in bko221065.
> 
> FWIW, the exclusive variant is done.  With it the emulated \GTST breaks
> neither _BIX after resume nor _BST at run time (0 of 5 and 0 of 1074),
> and with only the _TMP side converted both still break (5 of 5 and 421
> of 424).

As I just agreed in my reply to Rafael, it looks like we want to go
with the exclusive variant for now.

Can you post a patch series for the code you've for this now ?

Note I guess you're using LLM to help you with this, that is fine,
but can you please manually shorten the commit messages of the patches
a bit if they were written by LLM? OR instruct it to write "short and
concise" commit messages?

LLM's tend to write overly verbose commit messages. Note a good commit
message is important, but lets try to keep them compact.

And the same for the cover-letter.

Regards,

Hans