Re: [PATCH 1/1] ACPI: property: Fix REF STR... reference parsing

[Sakari Ailus <sakari.ailus@linux.intel.com>]

Hi Rafael,

On Mon, Mar 31, 2025 at 05:35:50PM +0200, Rafael J. Wysocki wrote:
> Hi Sakari,
> 
> On Mon, Mar 31, 2025 at 5:10 PM Sakari Ailus
> <sakari.ailus@linux.intel.com> wrote:
> >
> > Hi Rafael,
> >
> > On Mon, Mar 31, 2025 at 05:00:52PM +0200, Rafael J. Wysocki wrote:
> > > Hi Sakari,
> > >
> > > On Mon, Mar 31, 2025 at 3:35 PM Sakari Ailus
> > > <sakari.ailus@linux.intel.com> wrote:
> > > >
> > > > Hi Rafael,
> > > >
> > > > On Mon, Mar 31, 2025 at 03:19:07PM +0200, Rafael J. Wysocki wrote:
> > > > > On Mon, Mar 31, 2025 at 2:18 PM Sakari Ailus
> > > > > <sakari.ailus@linux.intel.com> wrote:
> > > > > >
> > > > > > Restore parsing of ACPI data node references consisting of a device node
> > > > > > reference followed by one or more child data node names.
> > > > > >
> > > > > > Fixes: 9880702d123f ("ACPI: property: Support using strings in reference properties")
> > > > > > Cc: stable@vger.kernel.org # for 6.8 and later
> > > > > > Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
> > > > > > ---
> > > > > > Hi Rafael,
> > > > > >
> > > > > > It seems that support for REF STR... references got accidentally removed
> > > > > > when pure STR reference were added. The former are documented in
> > > > > > Documentation/firmware-guide/acpi/dsd/graph.rst .
> > > > >
> > > > > It would be good to provide an ASL example that is not parsed as
> > > > > expected before the change and will be parsed correctly after it.
> > > > >
> > > > > Admittedly, I can't quite recall the history leading to the above
> > > > > commit, but this paragraphs is present in its changelog:
> > > > >
> > > > > "Also remove the mechanism allowing data nodes to be referenced
> > > > >  indirectly, with the help of an object reference pointing to the
> > > > >  "ancestor" device and a path relative to it (this mechanism is not
> > > > >  expected to be in use in any production platform firmware in the field)."
> > > > >
> > > > > so the change in question appears to be intentional rather than accidental.
> > > >
> > > > Right. This still continues to be documented in
> > > > Documentation/firmware-guide/acpi/dsd/graph.rst as noted in my previous
> > > > e-mail. A sample ASL snippet is can also be found in the same file. I only
> > > > noticed this when I tried to use such references.
> > > >
> > > > The other option indeed is to change the documentation and hope no-one
> > > > depends on it.
> > >
> > > I thought that this was the original plan.
> >
> > I wasn't aware of the plan. But given that no-one has complained yet,
> > there's a fair chance no-one will. I can post patches to switch to
> > string-only references.
> 
> I think that it should be sufficient to update the documentation so it
> does not describe the indirect mechanism mentioned in the changelog of
> commit 9880702d123f, shouldn't it?

I'd think so.

The REF+STR data node references actually go back to
Documentation/firmware-guide/acpi/dsd/data-node-references.rst merged in
2018 and their documented use includes LEDs (besides graphs) as well. Just
FYI though.

-- 
Sakari Ailus