On 04.12.2010 10:08, Freddie Chopin wrote:
So you're all about correctness and you don't reset halt the chip before flashing? How is that correct? Actually flashing does not work if you don't "reset halt" the chip. The need for halting is obvious. The need for reset is not, but think about what would happen if your code would have a timer (or any) interrupt enabled and running during your flashing. Actually you DO need to get the chip in some known (and specific!) condition to flash it.
I know what state the chip will be in - it is exactly the state which my startup code puts it into, so I can use the same state for flashing, without a need for an extra reset.
I'm not telling "slow down" - I'm saying "don't use this file that can speed things up because it also makes using OpenOCD a bit harder". Technically this file does not speed anything up, because there is no PLL setting anywhere in it - the new board config file (another patch) does speed things up and it can speed things up exactly the same way now or when lpc2478.cfg would use some default value. BTW why isn't the pll configuration function placed in target file?
The PLL settings *do* belong in a board config file. The code that sets the PLL is not board-specific but CPU-specific, so it can be shared.
If I'd have the microcontroller hooked up directly to JTAG and PSU with thin cables would that be considered a board too? It's just a simple microcontroller! It needs almost none configuration on OpenOCD's side - every board with LPC2478 was using the old file and it was fine.
The term "board" refers to a CPU plus environment - even if you replace the physical "board" part with air.
You need at least some decoupling capacitors, and you need to wire-up the JTAG connector, which can be done in different ways (regarding reset signals), so there are multiple possible setups that require different configurations.
cu Michael _______________________________________________ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development