Query FW config to reload library #4680
Conversation
RanderWang
requested review from
bardliao,
lyakh,
lgirdwood,
plbossart,
ranj063,
kv2019i,
dbaluta,
ujfalusi,
vijendarmukunda,
kuanhsuncheng and
yaochunhung
as code owners
RanderWang
force-pushed
the
query_lib_reload
branch
from
4f3b4f1 to
896daba
Compare
RanderWang
force-pushed
the
query_lib_reload
branch
from
896daba to
7a998dd
Compare
RanderWang
force-pushed
the
query_lib_reload
branch
2 times, most recently
from
4964196 to
f0e5f81
Compare
| /* IMR booting will restore the libraries as well, skip the loading */ | ||
| if (reload && hda->booted_from_imr) | ||
| /* if IMR booting is enabled and fw context is saved for D3 state, skip the loading */ | ||
| if (reload && hda->booted_from_imr && ipc4_data->fw_context_save) |
There was a problem hiding this comment.
it's a bit odd to have different structures to stored 'booted_from_imr' and 'fw_context_save'. Can we put all the information in the same place?
There was a problem hiding this comment.
@plbossart booted_from_imr is a intel feature which is set in sof/intel layer. fw_context_save is a common feature which is set in sof core layer. So I don't prefer to combine them in hda structure.
include/sound/sof/ipc4/header.h
Outdated
| SOF_IPC4_BUS_HARDWARE_ID, | ||
| SOF_IPC4_DEBUG_WINDOW_SIZE, | ||
| SOF_IPC4_KPB_RT_SINK_COUNT, | ||
| SOF_IPC4_DMI_FORCE_L1_EXIT, |
There was a problem hiding this comment.
do we need to expose all of the new definitions? Should we have a comment that says which ones are actually used by the driver? I don't know e.g. what the "BUS_HARDWARE_ID" might refer to.
There was a problem hiding this comment.
These config id are used by ref fw. I will change them to SOF_IPC4_RESERVED_xxx
Update fw_config_params in driver. Signed-off-by: Rander Wang <[email protected]>
Driver queries fw_context_save feature when fw is ready and can skip library reload with this feature since library is saved in persistent memory. The default value of fw_context_save is true unless fw reports false. Signed-off-by: Rander Wang <[email protected]>
If fw_context_save is defined by fw, driver can skip library reload on d3 exit or reload library. Signed-off-by: Rander Wang <[email protected]>
RanderWang
force-pushed
the
query_lib_reload
branch
from
f0e5f81 to
7902f64
Compare
|
SOFCI TEST |
|
@plbossart @ujfalusi the fw side PR thesofproject/sof#8428 was merged, please help to review kernel side. Thanks! we are preparing release for customer. |
| @@ -139,6 +139,8 @@ int sof_lnl_ops_init(struct snd_sof_dev *sdev) | |||
|
|
|||
| ipc4_data->mtrace_type = SOF_IPC4_MTRACE_INTEL_CAVS_2; | |||
|
|
|||
| ipc4_data->fw_context_save = true; | |||
There was a problem hiding this comment.
why do we need the default value to be true?
There was a problem hiding this comment.
It is for ref fw which supports it by default. Thanks @ujfalusi mentioned it for several times :)
|
We have 3 approvals and the test result is positive. @RanderWang has answered @plbossart's questions and the PR is urgent. I will go ahead to merge it to let CI run daily test with this PR. We can submit a follow up PR if there is any change needed. |
We got a final conclusion that driver should query FW config to reload library
fw pr: thesofproject/sof#8428