Improve panic message reporting.

[fabricedesre]

That prints: Panic at example/blinky.rs:42 : assertion failed: (left == right) (left: 1, right: 2) when hitting assert_eq!(1, 2)

r? @dhylands

Will fix https://github.com/mozilla-sensorweb/sensorweb-firmware/issues/8

[dhylands]

There seems to be a std::panic::set_hook which would be a good addition.

[fabricedesre]

This works well when ./flash.sh-ing, but we get stuck when just ./run.sh-ing

[dhylands]

I added the following to board.c and it causes the reset function to issue a breakpoint and drop into the debugger rather than doing the reset.

typedef struct
{
  volatile uint32_t DHCSR;  /*!< Offset: 0x000 (R/W)  Debug Halting Control and Status Register */
  volatile uint32_t DCRSR;  /*!< Offset: 0x004 ( /W)  Debug Core Register Selector Register */
  volatile uint32_t DCRDR;  /*!< Offset: 0x008 (R/W)  Debug Core Register Data Register */
  volatile uint32_t DEMCR;  /*!< Offset: 0x00C (R/W)  Debug Exception and Monitor Control Register */
} CoreDebug_Type;

#define CoreDebug_BASE      (0xE000EDF0UL)                            /*!< Core Debug Base Address */
#define CoreDebug           ((CoreDebug_Type *)     CoreDebug_BASE)   /*!< Core Debug configuration struct */

int is_debugger_running(void) {
    return (CoreDebug->DHCSR & 1) != 0; // Bit 0 is DEBUGEN
}

void reset(void)
{
    if (is_debugger_running()) {
        __asm volatile ("bkpt #0");
    }
    PRCMMCUReset(1);
}

[dhylands]

The reset is actually working under the debugger, but because its doing a processor reset, its the same thing as pressing the reset button which will run the image from flash.