I've been successfully using the new CGT15 feature that automatically copies ramfuncs at boot using BINIT table. This is a great feature, and I was wondering if it was possible for CLA-related things. For now, I am still copying manually using memcpy (CLA program, and CLA math library).
Right now, I have the following definition in my cmd file for ramfuncs
.TI.ramfunc : {
-lFlash2803x_API_V100.lib(.econst)
-lFlash2803x_API_V100.lib(.text)
}
LOAD = FLASH_A_H,
RUN = RAML0,
TABLE(BINIT)
which automatically copies FlashAPI functions and all other ramfunc-attributed functions successfully from flash to RAM.
Is it possible somehow to have several .TI.ramfunc sections so that they can have different RUN locations? My CLA math lib runs on CLARAM1 and CLAprog runs in RAML3. I was thinking I could just create another .TI.ramfunc section but with different RUN location (same section name). Or, if I can't have the same name, just create another section and include TABLE(BINIT).
For the CLA math library, I have tried converting the old section
CLA1mathTables : LOAD = FLASH_A_H, PAGE = 0,
RUN = CLARAM1, PAGE = 1,
LOAD_START(_Cla1mathTablesLoadStart),
LOAD_SIZE(_Cla1mathTablesLoadSize),
RUN_START(_Cla1mathTablesRunStart)
to another .TI.ramfunc section
.TI.ramfunc.2 : {
-lcla0_math_library.lib
}
LOAD = FLASH_A_H, PAGE = 0,
RUN = CLARAM1, PAGE = 1,
TABLE(BINIT)
but it complains about mixing CLA and non-CLA sections. What is different about .TI.ramfunc.2 vs CLA1mathTables that it has CLA and non-CLA sections? Both have LOAD in FLASH_A_H and both RUN from CLARAM1? Anyway, I feel like I'm trying to do something that's not really supported at the moment, but thought maybe there is a clever workaround.
