we are using a P2010 without any directly connected non-volatile storage. So, we need to load the program code (not a Linux kernel, just bare-metal, without OS) to the external SRAM, connected to the P2010 local bus, right after power on. Currently, gdb can do the basic initializations and load the program code to the SRAM utilizing a CW-TAP as bridge from PC to the P2010 JTAG interface. We want to implement this basic functionality (without debug capabilities) to our end product, so that our embedded Linux processor (probably ARM) can do this autonomously without PC. Freescale offers the binaries (gdb proxy & ccs) that run on CW-TAP ARM processor, which translate the gdb operations to JTAG interface. CCS seems to have an API. Are they somehow usable? How difficult would it be to implement this basic functionality over the debug channel? Should I invest time in this task or should I try simple & slow boundary scan bit banging to load code to SRAM?