Repository navigation
Change the naming convention for the ALH DAI copiers in topology - #7640
Conversation
There was a problem hiding this comment.
we already have a dai_index on line 20.
a) do we need the duplication?
b) if yes, should we use a macro to make sure there's always a matching value in both locations?
There was a problem hiding this comment.
is this one right? I don't recall us using a SoundWire DAI '10'?
There was a problem hiding this comment.
is this one right? I don't recall us using a SoundWire DAI '10'?
Yes, this follows the rule from b407a1a
Sdw dai index is not used by node_id. The only rule of sdw dai index is don't be duplicated.
For example, we can't have 2 dai_index 10.
There was a problem hiding this comment.
@plbossart @bardliao please have a look at this PR again. I have removed those artificial DAI index that we proviously used and replaced it with the actual index of the DAI in an aggregated group. If there is no aggregation the dai_index will always be 0
There was a problem hiding this comment.
here too. I must admit I don't follow what this dai_index is. It clearly makes no sense if this is used to name a DAI.
There was a problem hiding this comment.
here too. I must admit I don't follow what this dai_index is. It clearly makes no sense if this is used to name a DAI.
20 means pcm 2, index 0.
We include pcm id in the dai_index because the dais are defined in different files and sdw-xxx-generic.conf can be included by different sdw topologys. The dai index could be conflict if they are 0,1,2,3... We don't know what index can we use when we add a new sdw topology unless we check all included .confs.
For the reason above, I think use different dai index group for differnt pcm could be easier to decide a new dai index.
46a8a36 to
81169c7
Compare
ranj063
changed the title
There was a problem hiding this comment.
Yes, we do need the dai_index. When there's aggregation, the dai_index would be set based on the index of the DAI in the aggregated group. When there' sno aggregation, it will just be 0
|
Device test with TGLU_RVP_SDW_IPC4ZPH failed because my test kept running after my reservation was expired. This is my bad, I reserved the device too tight. Sorry about that. Public link: https://sof-ci.01.org/sofpr/PR7640/build8284/devicetest/index.html |
There was a problem hiding this comment.
commit title: topology2: componenets:
…opiers Create a new class for ALH because they are different from the other DAI types with DAI index being irrelevant. So, it would need a new naming scheme that's more meaningful. For ex: dai-copier-ALH.SDW1-Playback.0 Signed-off-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
This is specifically meant to be used with ALH DAI copiers. Signed-off-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Replace the use of the generic dai-copier class with the ALH specific DAI copier and pipeline class. Signed-off-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Remove ALH support in the generic dai-copier class as it has its own class now. Signed-off-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
When there are aggregated amps, the topology includes a DAI copier for each aggregated DAI. But, since the second DAI is not connected to anything in topology, it doesn't show up in the graph. So, add a virtual widget and connect it to the aggregated DAI and the gain module to show its existence in the graph. When parsing the topology, the kernel ignores all routes that contain a virtual widget at any end of the route. Signed-off-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
|
Usual failures with nocodec. Merging. |
|
Regression reported in Most tests are failing on ADLP_SKU0B00_SDCA_IPC4ZPH |
For ALH DAI, the new name is changed to reflect the link ID and index in an aggregated group like: dai-copier-ALH.SDW1-Playback.0 or alh-copier.SDW1-Playback.1
Also, today we do not show the aggregated DAI copiers in the topology graph. So add a route via a virtual widget to show its existence as below
