Re: [alsa-devel] [PATCH V4 0/2] Add DT support for i2s
Hi Mark,
On Wed, Nov 28, 2012 at 4:17 PM, Padmavathi Venna padma.v@samsung.com wrote:
V4 patches are based on Mark Brown's for-next branch of "git://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound.git"
Changes since V3: - Added syntex of gpio specifier as suggested by Mark Brown in DT document file - Added the requirement for having I2S alias ids - Explained about gpio lines - Made a seperate patch to get the I2S rclk src clk from alias ID
Changes since V2: - Rebased on 3.7-rc3 - Custom DT bindings are prefixed with samsung - As generic device tree DMA helpers not yet mainlined I am still using custom dma bindings. So added a priliminary statement regarding the same. I will rework on my patch once generic DMA helpers are mainlined.
Chnages since V1: - Rebased on 3.6-rc6
Padmavathi Venna (2): ASoC: Samsung: Get I2S src_clk from clock alias id. ASoC: SAMSUNG: Add DT support for i2s
Any comments on DT support for i2s patch? If not when this patch can be pulled?
.../devicetree/bindings/sound/samsung-i2s.txt | 78 +++++++ include/linux/platform_data/asoc-s3c.h | 6 - sound/soc/samsung/dma.c | 1 + sound/soc/samsung/dma.h | 1 + sound/soc/samsung/i2s.c | 238 +++++++++++++++---- 5 files changed, 268 insertions(+), 56 deletions(-) create mode 100644 Documentation/devicetree/bindings/sound/samsung-i2s.txt
-- 1.7.4.4
Thanks Padma
On Thu, Dec 06, 2012 at 09:31:40AM +0530, Padma Venkat wrote:
On Wed, Nov 28, 2012 at 4:17 PM, Padmavathi Venna padma.v@samsung.com wrote:
Padmavathi Venna (2): ASoC: Samsung: Get I2S src_clk from clock alias id. ASoC: SAMSUNG: Add DT support for i2s
Any comments on DT support for i2s patch? If not when this patch can be pulled?
Don't send contentless pings.
I still don't really like what's going on with aliases here, I need to understand why things are structured the way they are to say something about that.
On Thu, 6 Dec 2012 13:11:42 +0900, Mark Brown broonie@opensource.wolfsonmicro.com wrote:
On Thu, Dec 06, 2012 at 09:31:40AM +0530, Padma Venkat wrote:
On Wed, Nov 28, 2012 at 4:17 PM, Padmavathi Venna padma.v@samsung.com wrote:
Padmavathi Venna (2): ASoC: Samsung: Get I2S src_clk from clock alias id. ASoC: SAMSUNG: Add DT support for i2s
Any comments on DT support for i2s patch? If not when this patch can be pulled?
Don't send contentless pings.
I still don't really like what's going on with aliases here, I need to understand why things are structured the way they are to say something about that.
Right. On a brief look, it appears that the binding is using aliases as a way to encode implementation specific details of the driver.
g.
Hi,
On Fri, Dec 7, 2012 at 7:21 PM, Grant Likely grant.likely@secretlab.ca wrote:
On Thu, 6 Dec 2012 13:11:42 +0900, Mark Brown broonie@opensource.wolfsonmicro.com wrote:
On Thu, Dec 06, 2012 at 09:31:40AM +0530, Padma Venkat wrote:
On Wed, Nov 28, 2012 at 4:17 PM, Padmavathi Venna padma.v@samsung.com wrote:
Padmavathi Venna (2): ASoC: Samsung: Get I2S src_clk from clock alias id. ASoC: SAMSUNG: Add DT support for i2s
Any comments on DT support for i2s patch? If not when this patch can be pulled?
Don't send contentless pings.
I still don't really like what's going on with aliases here, I need to understand why things are structured the way they are to say something about that.
Right. On a brief look, it appears that the binding is using aliases as a way to encode implementation specific details of the driver.
Samsung i2s driver registers the platform device twice one for the samsung-i2s.0,1 or 2 and two for samsung-i2s.4(which actually doesn't represent any H/W peripheral). The max number of I2S blocks on any Samsung SoC are 3, the secondary device registration starts with device number 4.The second time registration is for the secondary fifo interface of a CPU dai if any exists.
With DT support the device number in the driver is always -1, I used alias id for the above purpose.
g.
Thanks Padma
On Sat, Dec 08, 2012 at 10:07:54AM +0530, Padma Venkat wrote:
Samsung i2s driver registers the platform device twice one for the samsung-i2s.0,1 or 2 and two for samsung-i2s.4(which actually doesn't represent any H/W peripheral). The max number of I2S blocks on any Samsung SoC are 3, the secondary device registration starts with device number 4.The second time registration is for the secondary fifo interface of a CPU dai if any exists.
With DT support the device number in the driver is always -1, I used alias id for the above purpose.
So, the thing here is that while the numbers were totally OK for platform device use they really don't fit into DT at all well. I think the thing here is to either have a node for the second CPU side DAI or to have a single DT link automatically expanded into two DAI links. I think the second option is much better - there's only one link on the board, it just happens to get expanded into two the way things are currently implemented in ASoC.
This stuff probably also ought to be moved over to soc-pcm as well by the way.
Hi Mark,
On Sun, Dec 9, 2012 at 12:56 PM, Mark Brown broonie@opensource.wolfsonmicro.com wrote:
On Sat, Dec 08, 2012 at 10:07:54AM +0530, Padma Venkat wrote:
Samsung i2s driver registers the platform device twice one for the samsung-i2s.0,1 or 2 and two for samsung-i2s.4(which actually doesn't represent any H/W peripheral). The max number of I2S blocks on any Samsung SoC are 3, the secondary device registration starts with device number 4.The second time registration is for the secondary fifo interface of a CPU dai if any exists.
With DT support the device number in the driver is always -1, I used alias id for the above purpose.
So, the thing here is that while the numbers were totally OK for platform device use they really don't fit into DT at all well. I think the thing here is to either have a node for the second CPU side DAI or to have a single DT link automatically expanded into two DAI links. I think the second option is much better - there's only one link on the board, it just happens to get expanded into two the way things are currently implemented in ASoC.
Could you please explain me in more detail about "single DT link automatically expanded into two DAI links". How this can be done?
This stuff probably also ought to be moved over to soc-pcm as well by the way.
Thanks Padma
On Mon, Dec 10, 2012 at 04:30:50PM +0530, Padma Venkat wrote:
Could you please explain me in more detail about "single DT link automatically expanded into two DAI links". How this can be done?
The device tree should only represent the single physical link that exists and the handling of the DAIs should be done at runtime by some piece of software.
participants (3)
-
Grant Likely -
Mark Brown -
Padma Venkat