[ranj063]

Revert these for now. Will be fixed with other multi-core fixes.

This reverts commit 928dba0.

Signed-off-by: Ranjani Sridharan ranjani.sridharan@linux.intel.com

[keyonjie]

@lgirdwood since the cAVS memory map refinement may take some time, so better to revert this now to rescue the CI first.

[lgirdwood]

@ranj063 can you create a ranj/topic/multicore branch on the sof repo for the multicore development. This branch will be the optimisation branch that can be merged in agian after @keyonjie has fixed the heap mappings.
@keyonjie can you explain why the heap refinement will take some time ? The heap dump is showing that we have lost of free space in certain blocks athat can be moved to other heaps or block ?

[keyonjie]

@ranj063 can you create a ranj/topic/multicore branch on the sof repo for the multicore development. This branch will be the optimisation branch that can be merged in agian after @keyonjie has fixed the heap mappings.
@keyonjie can you explain why the heap refinement will take some time ? The heap dump is showing that we have lost of free space in certain blocks athat can be moved to other heaps or block ?

Hi @lgirdwood I wanted to combine and reduce the type numbers of the mm_heap, e.g. only system, runtime, shared are needed, no system_runtime, buffer, system shared needed, that's why I said it will take some time.

But since that could be relative invasive change may impact IMX paltforms also, let me do smaller change to just make full use of our SRAM heap blocks only, this could be more easier and hopefully it will come within one or two days.

[keyonjie]

@ranj063 @lgirdwood now refinement available, we have much more memory to use on TGL/CNL/ICL platforms now.

#4445
thesofproject/rimage#51