Patch Set 3:

1M is really a lot and wouldn't fit in most ramstages. Is that much heap really needed or should the linker scripts be adapted?

It's lib/malloc.c, just print (uintptr_t)free_mem_end_ptr - (uinptr_t)free_mem_ptr before entry to payload to find out actual usage. Doing that in payload_run() after checkstack() would make some sense in any case.

While HEAP_SIZE is adjusted based on FLATTENED_DEVICE_TREE (and lib/device_tree.c), maybe majority of heap requirement comes from PAYLOAD_FIT_SUPPORT, unpack_fdt() in lib/fit_payload.c instead.

IMHO it is not particularly clean if payload does references back to ramstage heap.

View Change

To view, visit change 32392. To unsubscribe, or for help writing mail filters, visit settings.