Re: [PATCH] ACPI: battery: Fix bogus cycle count values on systems without _BIX

[Armin Wolf]

Am 09.09.26 um 16:27 schrieb Rafael J. Wysocki (Intel):

> On Tue, Sep 1, 2026 at 10:18 PM Armin Wolf <W_Armin@xxxxxx> wrote:
> > Am 01.09.26 um 22:16 schrieb Rafael J. Wysocki (Intel):
> >
> > > On Tue, Sep 1, 2026 at 10:11 PM Armin Wolf <W_Armin@xxxxxx> wrote:
> > > > Am 07.08.26 um 01:38 schrieb Armin Wolf:
> > > >
> > > > > Devices supporting only the legacy _BIF ACPI control method cannot
> > > > > know the cycle count of the associated battery, so cycle_count always
> > > > > remains 0. Claiming support for POWER_SUPPLY_PROP_CYCLE_COUNT on
> > > > > such devices can thus mislead userspace applications and prevents
> > > > > power supply extensions from supplying correct cycle count values.
> > > > >
> > > > > Fix this by not claiming support for POWER_SUPPLY_PROP_CYCLE_COUNT
> > > > > on devices without _BIX.
> > > > Any thoughts on this? I tested it on my device, and so far everything seems
> > > > to work flawlessly.
> > > It looks good overall, but it is quite substantial, which is why it
> > > didn't make it into 7.3-rc1.
> > >
> > > I'm planning to apply it for 7.4.
> > Alright, that works for me.
> And so applied.
>
> Sashiko has a comment on the _BIX failing case:
>
> https://sashiko.dev/#/patchset/20260806233817.622975-1-W_Armin%40gmx.de
>
> but this can be addressed on top of the current patch IMV.
>
> If you think that it's worth addressing, please send a follow-up one for that.
>
> Thanks!

Thank you :)

I do not think that Sashiko is right in this case, as failures inside _BIX
should _normally_ only be temporary. So i think the current strategy of temporally
falling-back to _BIF if _BIX fails is fine.

Thanks,
Armin Wolf