[alsa-devel] [PATCH] ASoC: kirkwood: change the compatible string of the kirkwood-i2s driver
The compatible string of the kirkwood-i2s driver was chosen as "marvell,mvebu-audio". Using such a compatible string is not a good idea, since "mvebu" is the name of a large family of SOCs, in which new, unknown SOCs will be coming in the future. It is therefore impossible to know what will be evolutions of this hardware block in the next generations of the SOCs. For this reason, the recommandation for compatible strings of on-SOCs devices has always been to use the name of the oldest SOC that has the hardware block. New SOCs that have an exactly compatible hardware block can reference it using the same compatible string. See [1], [2] and [3] for various cases were this suggestion was made, including from Rob Herring, a Device Tree binding maintainer.
As an example, there are already small differences between current generations:
* On Kirkwood, only one interrupt is used for audio. * On Dove, two interrupts are used, one for audio data and one for error reporting.
In the near future, I'll be adding audio support to Armada 370, which allows has the same hardware block (but maybe with minor variants).
Therefore, this patch changes the driver to accept "marvell,kirkwood-audio" and "marvell,dove-audio" as compatible strings instead of the too-generic "marvell,mvebu-audio". The reason for the two different compatible strings is the difference in the number of interrupts used by the two SOCs for audio.
This Device Tree binding has never been part of a Linux kernel stable release so far, so it can be changed now without breaking backward compatibility.
[1] http://lists.infradead.org/pipermail/linux-mtd/2012-March/040417.html [2] http://lists.infradead.org/pipermail/linux-arm-kernel/2013-April/161065.html [3] http://lists.infradead.org/pipermail/linux-arm-kernel/2012-March/087702.html
Signed-off-by: Thomas Petazzoni thomas.petazzoni@free-electrons.com --- Documentation/devicetree/bindings/sound/mvebu-audio.txt | 12 ++++++++---- sound/soc/kirkwood/kirkwood-i2s.c | 3 ++- 2 files changed, 10 insertions(+), 5 deletions(-)
diff --git a/Documentation/devicetree/bindings/sound/mvebu-audio.txt b/Documentation/devicetree/bindings/sound/mvebu-audio.txt index 7e5fd37..f0062c5 100644 --- a/Documentation/devicetree/bindings/sound/mvebu-audio.txt +++ b/Documentation/devicetree/bindings/sound/mvebu-audio.txt @@ -2,13 +2,17 @@
Required properties:
-- compatible: "marvell,mvebu-audio" +- compatible: + "marvell,kirkwood-audio" for Kirkwood platforms + "marvell,dove-audio" for Dove platforms
- reg: physical base address of the controller and length of memory mapped region.
-- interrupts: list of two irq numbers. - The first irq is used for data flow and the second one is used for errors. +- interrupts: + with "marvell,kirkwood-audio", the audio interrupt + with "marvell,dove-audio", a list of two interrupts, the first for + the data flow, and the second for errors.
- clocks: one or two phandles. The first one is mandatory and defines the internal clock. @@ -21,7 +25,7 @@ Required properties: Example:
i2s1: audio-controller@b4000 { - compatible = "marvell,mvebu-audio"; + compatible = "marvell,dove-audio"; reg = <0xb4000 0x2210>; interrupts = <21>, <22>; clocks = <&gate_clk 13>; diff --git a/sound/soc/kirkwood/kirkwood-i2s.c b/sound/soc/kirkwood/kirkwood-i2s.c index 7fce340..0f3d73d 100644 --- a/sound/soc/kirkwood/kirkwood-i2s.c +++ b/sound/soc/kirkwood/kirkwood-i2s.c @@ -559,7 +559,8 @@ static int kirkwood_i2s_dev_remove(struct platform_device *pdev)
#ifdef CONFIG_OF static struct of_device_id mvebu_audio_of_match[] = { - { .compatible = "marvell,mvebu-audio" }, + { .compatible = "marvell,kirkwood-audio" }, + { .compatible = "marvell,dove-audio" }, { } }; MODULE_DEVICE_TABLE(of, mvebu_audio_of_match);
On Thu, 5 Sep 2013 14:56:31 +0200 Thomas Petazzoni thomas.petazzoni@free-electrons.com wrote:
The compatible string of the kirkwood-i2s driver was chosen as "marvell,mvebu-audio". Using such a compatible string is not a good idea, since "mvebu" is the name of a large family of SOCs, in which new, unknown SOCs will be coming in the future. It is therefore impossible to know what will be evolutions of this hardware block in the next generations of the SOCs. For this reason, the recommandation for compatible strings of on-SOCs devices has always been to use the name of the oldest SOC that has the hardware block. New SOCs that have an exactly compatible hardware block can reference it using the same compatible string. See [1], [2] and [3] for various cases were this suggestion was made, including from Rob Herring, a Device Tree binding maintainer.
As an example, there are already small differences between current generations:
- On Kirkwood, only one interrupt is used for audio.
- On Dove, two interrupts are used, one for audio data and one for error reporting.
In the near future, I'll be adding audio support to Armada 370, which allows has the same hardware block (but maybe with minor variants).
Therefore, this patch changes the driver to accept "marvell,kirkwood-audio" and "marvell,dove-audio" as compatible strings instead of the too-generic "marvell,mvebu-audio". The reason for the two different compatible strings is the difference in the number of interrupts used by the two SOCs for audio.
This Device Tree binding has never been part of a Linux kernel stable release so far, so it can be changed now without breaking backward compatibility.
[1] http://lists.infradead.org/pipermail/linux-mtd/2012-March/040417.html [2] http://lists.infradead.org/pipermail/linux-arm-kernel/2013-April/161065.html [3] http://lists.infradead.org/pipermail/linux-arm-kernel/2012-March/087702.html
Signed-off-by: Thomas Petazzoni thomas.petazzoni@free-electrons.com
The name of the kirwood audio driver has been changed from "kirkwood-i2s" to "mvebu-audio" by Russell King (http://www.spinics.net/lists/arm-kernel/msg264185.html). I have just used this name in the DT.
Then, the second interrupt in Dove is not used in the driver. If it should, it would be present in the DTs of the dove boards only, so, actually, there is not need to have different compatibles.
Eventually, you are talking about some other mvebu SoCs. Before adding any new compatible, it would be nice to know if, for example, some changes are needed in the kirkwood/dove audio driver for the Armada 370...
Dear Jean-Francois Moine,
On Thu, 5 Sep 2013 18:54:56 +0200, Jean-Francois Moine wrote:
The name of the kirwood audio driver has been changed from "kirkwood-i2s" to "mvebu-audio" by Russell King (http://www.spinics.net/lists/arm-kernel/msg264185.html). I have just used this name in the DT.
I don't care about the driver name: it can be changed whenever you want.
On the opposite, the DT compatible string becomes part of the kernel ABI, and therefore cannot be changed once it has been defined. This is why I worry about this one.
And the compatible string doesn't necessarily need to match the driver name.
Then, the second interrupt in Dove is not used in the driver. If it should, it would be present in the DTs of the dove boards only, so, actually, there is not need to have different compatibles.
Indeed the second interrupt is not used in the driver, but the binding still specifies that two interrupts have to be mentioned, which is incorrect for Kirkwood. The fact that on Kirkwood there is one interrupt, while Dove has two, is an indication that the IP block is not exactly the same. Should this error interrupt become useful in the driver, it should be done only for Dove, so it makes sense to have two different compatible strings.
Eventually, you are talking about some other mvebu SoCs. Before adding any new compatible, it would be nice to know if, for example, some changes are needed in the kirkwood/dove audio driver for the Armada 370...
So far, it looks like no changes should be needed. I'm sure this will bring you to the conclusion that 'mvebu-audio' is a good compatible string: it is not because while we know what Armada 370 looks today, we have absolutely no idea what the next Armada <something> will look like. Maybe it will use the same audio IP block, with just a few little variations.
Again, please read the mails from Rob Herring and Arnd Bergmann. It has been said numerous times that using the name of the oldest SoC that has the compatible IP block was the right choice for the compatible string.
Best regards,
Thomas
On Thu, Sep 05, 2013 at 07:03:41PM +0200, Thomas Petazzoni wrote:
not exactly the same. Should this error interrupt become useful in the driver, it should be done only for Dove, so it makes sense to have two different compatible strings.
In general it's reasonable to add compatible strings for the SoCs anyway since even if we don't know of any reason why we'd want to distingush between them right now it's possible that in the future some reason might be discovered (such as an erratum being found). Worst case the kernel just treats both compatible strings identically which isn't a great loss. It's not essential though.
On Thu, 5 Sep 2013 19:03:41 +0200 Thomas Petazzoni thomas.petazzoni@free-electrons.com wrote:
Again, please read the mails from Rob Herring and Arnd Bergmann. It has been said numerous times that using the name of the oldest SoC that has the compatible IP block was the right choice for the compatible string.
OK. Thanks.
On Thu, Sep 05, 2013 at 02:56:31PM +0200, Thomas Petazzoni wrote:
The compatible string of the kirkwood-i2s driver was chosen as "marvell,mvebu-audio". Using such a compatible string is not a good idea, since "mvebu" is the name of a large family of SOCs, in which new, unknown SOCs will be coming in the future. It is therefore
Applied, thanks.
participants (3)
-
Jean-Francois Moine -
Mark Brown -
Thomas Petazzoni