Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <bbe9f772-e049-4ad3-18aa-cca0b793439f@linux.intel.com>
Date:   Wed, 28 Jun 2023 17:57:56 +0200
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Firefox/102.0 Thunderbird/102.11.0
Subject: Re: [RFC PATCH 0/8] PCI: Define Intel PCI IDs and use them in drivers
Content-Language: en-US
To:     Andy Shevchenko <andriy.shevchenko@intel.com>,
        =?UTF-8?Q?Amadeusz_S=c5=82awi=c5=84ski?= 
        <amadeuszx.slawinski@linux.intel.com>
Cc:     Bjorn Helgaas <bhelgaas@google.com>,
        Mark Brown <broonie@kernel.org>, Takashi Iwai <tiwai@suse.com>,
        alsa-devel@alsa-project.org, linux-pci@vger.kernel.org,
        linux-kernel@vger.kernel.org,
        Cezary Rojewski <cezary.rojewski@intel.com>,
        Kai Vehmanen <kai.vehmanen@linux.intel.com>
References: <20230628205135.517241-1-amadeuszx.slawinski@linux.intel.com>
 <ZJxIZGV4+5Al0CpW@smile.fi.intel.com>
From:   Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
In-Reply-To: <ZJxIZGV4+5Al0CpW@smile.fi.intel.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Precedence: bulk


On 6/28/23 16:49, Andy Shevchenko wrote:
> On Wed, Jun 28, 2023 at 10:51:27PM +0200, Amadeusz Sławiński wrote:
>> PCI IDs for Intel HDA are duplicated across quite a few drivers, due to
>> various configurations and historical reasons. Currently almost all uses
>> of HDA PCI IDs have corresponding comment telling which platform it is.
>> Additionally there are some inconsistencies between drivers about which
>> ID corresponds to which device.
>>
>> Simplify things, by adding PCI IDs to global header and make use of them
>> in drivers. This allows for removal of comments by having IDs themselves
>> being self explanatory. Additionally it allows for removal of existing
>> inconsistencies by having one source of truth.
> 
> I'm in favour of this series. It allows to use PCI_DEVICE_DATA() in many places.
> With that said, I think you can also add some more definitions to PCI IDs header
> for the sake of being able to use that macro.

I don't have any objections on the change.

The big open is how we add new definitions without a 3-way deadlock
between PCI, sound and ASoC trees, and how those definitions can be
added to the -stable trees.

This isn't an hypothetical case, we have 2 pending submissions for
LunarLake [1] and ArrowLake [2] which will be provided as soon as the
merge window closes.

It's not clear to me if Bjorn is ok to let those audio-specific PCI IDs
go the audio trees, and how things would work between Mark and Takashi.

[1] https://github.com/thesofproject/linux/pull/4425
[2] https://github.com/thesofproject/linux/pull/4437