Use arrays for objects in top-level conf files and included conf files #7113
Conversation
ranj063
requested review from
dbaluta,
lbetlej,
lgirdwood,
mmaka1 and
plbossart
as code owners
I have verified that the generated binaries with this change are identical to the ones we have today except for these three files and I think these are because of some module names being different due to the change in instance IDs. |
kv2019i
left a comment
kv2019i
left a comment
There was a problem hiding this comment.
Seems much cleaner without the additional ids (that needed to be unique), thanks!
ranj063
force-pushed
the
fix/use_arrays
branch
from
2181f5b to
b8011c4
Compare
Where there more than 1 object of the same type defined at the same node
level, use arrays to define the object. For example, we can define 2
routes as follows:
Object.Base.route [
{
source "smart_amp.2.1"
sink "copier.SSP.2.1"
}
{
source "mixin.1.1"
sink "mixout.2.1"
}
]
This allows us to merge 2 arrays from different conf files without
needing to make their instance ID's unique.
Assume we add another route in a separate conf file like below:
Object.Base.route [
{
source "gain.5.1"
sink "copier.5.1"
}
]
The alsatplg compiler will merge them as follows resulting in 3 route
objects.
Object.Base.route [
{
source "smart_amp.2.1"
sink "copier.SSP.2.1"
}
{
source "mixin.1.1"
sink "mixout.2.1"
}
{
source "gain.5.1"
sink "copier.5.1"
}
]
Signed-off-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
These are already in the common includes. Signed-off-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
ranj063
force-pushed
the
fix/use_arrays
branch
from
b8011c4 to
533b303
Compare
jsarha
left a comment
jsarha
left a comment
There was a problem hiding this comment.
All in all I think I understand the logic behind these changes, and yes the result look much cleaner.
| stream_name 'SSP0 Playback' | ||
| } | ||
| } | ||
| passthrough-be [ |
There was a problem hiding this comment.
The changes to these new pipeline object instantiations with number of pipeline arrays were hardest to follow. The new jumping indexes do not look the most logical choices, but then again I can see that they come from history. In the end the indexes do not matter for anything else but they are needed when referring to these objects from outside, do they?
There was a problem hiding this comment.
yes, the index has no meaning in most cases except for the widget names. But I havent touched those obects within pipeline class definitions yet. That will be the next target after this one is merged.
|
qemu-boot for icl/cnl fail and the mtl-zephyr-sparse are known issues, other checks pass so proceeding with merge. |
4 participants
This looks like a lot of changes for one patch but in reality the changes are quite monotonous to use arrays in object declarations. This fixes the problem of the topology writer having to come up with obscenely large instance ID's for objects just to make them exclusive when including conf files programatically.