Port 5.20 Free RTOS in Cortex

spyhabs wrote on Thursday, May 07, 2009:


I’d like to know if you can help me.  I port the new free rtos version in a new Cortex-M3 project in IAR.  There is no call in my created task and it is stalled in vPortStartFirstTask()

    /* Use the NVIC offset register to locate the stack. */
    ldr r0, =0xE000ED08
    ldr r0, [r0]
    ldr r0, [r0]
    /* Set the msp back to the start of the stack. */
    msr msp, r0
    /* Call SVC to start the first task. */
    svc 0   <<<<<<<<<<<<<<<<<<< STOP HERE

What does it mean? 


edwards3 wrote on Thursday, May 07, 2009:

Check the vector table, do you have vPortSVCHandler installed as the SVC handler? Look at the file startup_ewarm.c in Demo\CORTEX_LM3Sxxxx_IAR_Keil for an example.

spyhabs wrote on Thursday, May 07, 2009:

Yes I have this handler in the vector.  I wonder if global interrupts are disabled!


edwards3 wrote on Thursday, May 07, 2009:

Interrupts should be disabled at that point, the svc instruction is synchronous and will execute even when interrupts are disabled.

comcle wrote on Wednesday, June 10, 2009:

I got the same problem with 5.30 of Free RTOS. The vector table looks like this:

__attribute__ ((section(".isr_vector")))
void (* const g_pfnVectors[])(void) =
    (void (*)(void))((unsigned long)pulStack + sizeof(pulStack)),
                                            // The initial stack pointer
    ResetISR,                               // The reset handler
    NmiSR,                                  // The NMI handler
    FaultISR,                               // The hard fault handler
    IntDefaultHandler,                      // The MPU fault handler
    IntDefaultHandler,                      // The bus fault handler
    IntDefaultHandler,                      // The usage fault handler
    0,                                      // Reserved
    0,                                      // Reserved
    0,                                      // Reserved
    0,                                      // Reserved
    vPortSVCHandler,                        // SVCall handler
    IntDefaultHandler,                      // Debug monitor handler
    0,                                      // Reserved
    xPortPendSVHandler,                     // The PendSV handler
    xPortSysTickHandler,                    // The SysTick handler
    IntDefaultHandler,                      // GPIO Port A

It should work, but cpu stucks at svc command in vPortStartFirstTask.
Can anyone help?

comcle wrote on Wednesday, June 17, 2009:

OK, i found the Problem with SVC Command in the Cortex-M3.

It is a general problem while debugging with GDB and openocd. During Step-wise debugging, interrupts are disabled. SVC command causes an interrupt, so the command doesn’t work during stepping.

I found several suggestion in openOCD forums that i should enable interrupts again with “monitor cortex_m3 maskisr off” in the gdb console. This didn’t work.

After 2 days of searching a solution, it has shown that the problem was a wrong openOCD and GDB configuration. (It seems that 99% all problems are openOCD config problems)

Now i use openocd from the YAGARTO Homepage ( r1888+patch) with the following openocd Configuration files: interface/luminary.cfg target/lm3s6965.cfg, and following Startup code for GDB:

target extended-remote localhost:3333
monitor soft_reset_halt
break main

Every works now. After startup troubles, the Luminary Cortex-M3 µCs + Codesourcery Lite + FreeRTOS + openOCD + Eclipse works great.