Re: [PATCH 1/5] PCI: add ArrowLake-S PCI ID for Intel HDAudio subsystem.

[Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>]

On 8/2/23 11:34, Mark Brown wrote:
> On Wed, Aug 02, 2023 at 11:25:41AM -0500, Bjorn Helgaas wrote:
>> On Wed, Aug 02, 2023 at 11:07:36AM -0500, Pierre-Louis Bossart wrote:
> 
>>> I am not following. we just agreed a couple of weeks ago to record ALL
>>> Intel/HDaudio PCI IDs in the same pci_ids.h include file.
> 
>> I'm not sure who "we" is here.  If it included me and I signed up to
>> it, I apologize for forgetting, and go ahead and add my:
> 
>>   Acked-by: Bjorn Helgaas <bhelgaas@google.com>

This was the original thread for the record

https://lore.kernel.org/alsa-devel/20230717114511.484999-3-amadeuszx.slawinski@linux.intel.com/

>> I'm just pointing out the usual practice for pci_ids.h, as mentioned
>> in the file itself.

You're actually right that we didn't talk about the minimum criterion to
add a PCI ID to this file. To me it was a central place similar to the
cpu ids, etc., if it wasn't clear to everyone than it's good to agree on
this second point.

> I think the thing with these drivers is that we know they will become
> shared in fairly short order so it just becomes overhead to add then
> move the identifier and update.

Indeed, the sharing part is not always predictable and is subject to
roadmap changes made above my pay grade.

The intended use of the devices can vary as well, some PCI IDs for
desktops are intended to be used only by snd-hda-intel, but if one OEM
starts adding digital microphones then the SOF driver becomes required.

So rather than force everyone to follow changes at Intel or Intel
customers it's simpler to just add PCI IDs in pci_ids.h. We typically
deal with 3-4 PCI IDS per year