OPTEE panic in core_init_mmu_map

[dkal]

Hi, I have a custom board with a stm32mp153.  I was able to get past the trusted bootloader and into optee.  It builds the mmu map and then panics when checking it.  I am able to partially step through building the table and the memory starts to get corrupted but I can't see why.  I also have a 157-ev which works fine when I build for that on the same dev machine.  I am looking for possible clues on where to go from here.

[dkal]

I've tracked the issue down to memmove not working properly, why would it be fine in stm32mp157 and not fine in stm32mp153?

[dkal]

More specifically, the memmove is filling the destination with garbage even though the source and destination pointers are valid.

[dkal]

It seems that single byte copies are the issue, when it can do a more optimized copy with longs, that memory is correct.  

[PatrickF]

Hi @dkal 

if you are writing to DDR, maybe double check the correct DM Vs DQ/DQS wirings on your schematics. We had seen similar behavior when customers swapped Data Mask and Bytes which mean x32-bits write works but not bytes.

Regards.

[dkal]

Thank you, we will check.

[dkal]

I am going to assume that this is a RAM issue and am now trying the DDR utility.