Re: [PATCH 2/5] ACPI: sysfs: use device lifecycle for _STR result

From: Thomas Weißschuh
Date: Thu Jun 27 2024 - 13:17:33 EST


Rafael:

Would you mind backing this series out for now?
I'll do some more experiments and resubmit.

On 2024-06-26 16:39:10+0000, Mark Brown wrote:
> On Tue, Jun 25, 2024 at 11:56:18PM +0200, Thomas Weißschuh wrote:
> > On 2024-06-25 21:57:13+0000, Mark Brown wrote:
>
> > > <2>[ 50.236703] acpi LNXTHERM:08: Resources present before probing
>
> > This does make sense, the device is not yet bound to a driver.
> > Which apparently precludes the usage of devres.
>
> Oh, yes - I really wouldn't expect that to work at all, devres is all
> about tying things to the device being bound so trying to use it outside
> of that is not something I'd expect to go well.
>
> > I'm also wondering why the _STR attribute behaved differently in the
> > first place.
> > Does the patch below work better?
>
> That patch applied on top of -next appears to resolve the issue.

Thanks for the confirmation