After a lengthy support back and forth with the fine folks at Rowley Crossworks, we are at a stand still … hoping somebody can give us an "AH HA" moment! I can send the project files if somebody with Crossworks wants to look.
Anyway, we are using the NXP/Phillips USB Secondary Bootloader (AN10759.pdf on their site). It works fine with their samples built with the Keil toolchain.
The FreeRTOS Rowley demo works great "out of the box", but when we make the settings changes in Crossworks to rebuild the demo to use 0x2000 as the starting address (and remap the interrupts), it doesn’t work.
The Rowley folks used a JTAG debugger (we’re evaluating their tool, so we don’t have a compatible JTAG device yet) and the result was:
<B>If I debug your FreeRTOS app it starts up (again under debugger control) and dies on the call to vTaskStartScheduler. Interrupts seem to be routed OK.,</b>
Is this an ARM7 device? If so dyeing on vTaskStartScheduler() probably means the chip is not in Supervisor mode. Put a break point on vTaskStartScheduler() then check the mode.
The start up assembly file will switch into each mode, set up a stack for that mode, before switching into the next mode (modes being IRQ, FIQ, Supervisor, User, etc). You have to ensure that before calling main() you leave the processor in Supervisor mode. Also never go into User mode but instead go into System mode, otherwise you will not be able to get out of User mode without making a system call.
The same will be true before you call the FreeRTOS application from your boot loader - make sure you are in Supervisor mode.
We are using the supplied .s file (with the demo) except for the offset of the Memory Map to start at 0x2000 … the "SUPERVISOR" define is defined in the .s file …