Identifying source of restart

[palmerr23]

Earle,

I have some code that can identify several sources of restart.

Is it worth adding to the RP2040 helper class, or will I just turn it into a little library?

Richard

/*
Test Reset reason code
*/

#include "hardware/regs/vreg_and_chip_reset.h"
#include "hardware/watchdog.h"
#include "hardware/structs/watchdog.h"
/*

restart after 

  Upload, Watchdog, soft reboot:  (0x100 or 0x10000) + watchdog

  power on 0x100: no watchdog

  run switch: 0x10000 + no watchdog

  reset_reason does not reset after Run switch, but WDT is T for subsequent (soft) restarts

*/
enum reset_reason {UNKNOWN_RESET, PWRON_RESET, RUN_RESET, SOFT_RESET, DEBUG_RESET, DEEPSLEEP_RESET, PSM_RESET};
reset_reason getResetReason(void)
{    
  if(watchdog_caused_reboot()) // all soft restarts. This test first, as the VREG bits don't seem to reset on RUN restarts
    return SOFT_RESET; //3

  // POR/BOR, RUN , DEBUG
  io_rw_32 *rrp = (io_rw_32 *) (VREG_AND_CHIP_RESET_BASE + VREG_AND_CHIP_RESET_CHIP_RESET_OFFSET);
  uint32_t reset_reason = *rrp;
  reset_reason =  reset_reason & VREG_AND_CHIP_RESET_CHIP_RESET_BITS;
  
  if (reset_reason & VREG_AND_CHIP_RESET_CHIP_RESET_HAD_POR_BITS)  // POR: Last reset was from the power-on reset or brown-out detection
    return PWRON_RESET;   // 1

  if (reset_reason & VREG_AND_CHIP_RESET_CHIP_RESET_HAD_RUN_BITS)  // RUN: Last reset was from the RUN pin, not watchdog
    return RUN_RESET; //2
  
  // 4 test for debug restart unknown

  // 5 test for sleep restart unknown

   if (reset_reason & VREG_AND_CHIP_RESET_CHIP_RESET_HAD_PSM_RESTART_BITS)  // DEBUG: Last reset was from the debug port    
    return PSM_RESET; // 6 **** untested ****

   return UNKNOWN_RESET;
}

void setup() {
  pinMode(LED_BUILTIN, OUTPUT);
  Serial.begin(115200);
  while (!Serial && millis() < 5000)
    delay(10);
  Serial.println("Reset reason test");
  //Serial.printf("Reset reason %i\n", getResetReason());
  switch(getResetReason())
  {
    case PWRON_RESET :
      Serial.println("Power on reset");
      break;
    case RUN_RESET :
      Serial.println("Run button reset");
      break;
    case SOFT_RESET :
      Serial.println("Soft reset: upload, reboot or watchdog");
      break;  
      default:    
      Serial.println("Reset: unknown cause");
  }
  //rp2040.wdt_begin(5000); // test watchdog timeout after 5 secs
}

bool blink;
#define RESTART_AFTER 10000
void loop()
{
  digitalWrite(LED_BUILTIN, HIGH); // blink, as Serial sometimes doesn't restart
  delay(500);

  digitalWrite(LED_BUILTIN, LOW);
  delay(500);

  if(millis() > RESTART_AFTER)
  {
    ;
    // rp2040.restart(); // test soft restart (doesn't use watchdog, but Serial won't restart)
    // p2040.reboot(); // test soft reboot (uses watchdog)
  }
}

[earlephilhower]

Sorry, this looks like it got lost in the holiday weekend noise.

I think this would be fine to add to the RP2040 helper class, since it's focusing on the chip internals (like most of the other helpers there). @kendo55 also had a suggestion for the same here, so it's evidently got interest!

[palmerr23]

Earle,

I'll write the code, create an associated enum, and issue a pull request for int rp2040.getResetReason()

It might take a week or two, as I'm in the middle of something that needed this particular functionality - so it's not fully baked yet!

Thanks for the @kendo55 reference, there's an additional code line there that helps unpick the different flavours of wdt resets.

BTW, I'll add the currently unidentifiable (eg sleep) causes to the enum for forward compatibility, and make a note in the doco entry.

[palmerr23]

Here's the final code and a test harness.

Currently there are four discernable reset causes.

• The RP2040 doco suggests that the debug port reset causes the Pico to reboot to the bootloader (untested), so the next reset will be a software reset.
• Resume from sleep doesn't cause a reset.

I'll write up some doco and submit a PR.

Richard

rp2040.getResetReason( ):

#include "hardware/regs/vreg_and_chip_reset.h"
#include "hardware/watchdog.h"
#include "hardware/structs/watchdog.h"

enum resetReason_t {UNKNOWN_RESET, PWRON_RESET, RUN_PIN_RESET, SOFT_RESET, WDT_RESET};

resetReason_t getResetReason(void)
{    
  io_rw_32 *WD_reason_reg = (io_rw_32 *) (WATCHDOG_BASE + WATCHDOG_REASON_OFFSET);	
  io_rw_32 *rrp = (io_rw_32 *) (VREG_AND_CHIP_RESET_BASE + VREG_AND_CHIP_RESET_CHIP_RESET_OFFSET);
  
  if(watchdog_caused_reboot() && watchdog_enable_caused_reboot()) // WDT timer
    return WDT_RESET;

  if (*WD_reason_reg & WATCHDOG_REASON_TIMER_BITS) // soft reset() or reboot() 
    return SOFT_RESET;

  if (*rrp & VREG_AND_CHIP_RESET_CHIP_RESET_HAD_POR_BITS)  // POR: power-on reset or brown-out detection
    return PWRON_RESET;  

  if (*rrp & VREG_AND_CHIP_RESET_CHIP_RESET_HAD_RUN_BITS)  // RUN pin
    return RUN_PIN_RESET; 

   if (*rrp & VREG_AND_CHIP_RESET_CHIP_RESET_HAD_PSM_RESTART_BITS)  // DEBUG port
    return DEBUG_RESET; // **** untested **** debug reset may just cause a rebootToBootloader()

   return UNKNOWN_RESET;
}

test harness:

#include <EEPROM.h>
#define NTESTS 5 // Debug port test not enabled
char resetReasonText[][24] ={"Unknown", "Power On / Brownout", "Run pin", "Software", "Watchdog Timer"};
char testText[][128] ={"Run pin\n Short Run pin to ground", "Power On / Brownout\nDisconnect / reconnect power (i.e. USB)", "Software reBOOT in 5 Secs", "Watchdog Timer in 7 Secs", "Software reSTART in 5 Secs\n USB Serial does not always restart", "Debug port - do something!"};

void delayCount(void)
{
  int t = 1;
  while(1)
  {
    Serial.print(".");
    if (t % 50 == 0)
      Serial.println();
    delay(1000);
  }
}

void setup() {
  pinMode(LED_BUILTIN, OUTPUT);
  Serial.begin(115200);
  while (!Serial)
    delay(10);
  Serial.println("Reset reason:");

  resetReason_t rr = getResetReason();
  Serial.printf("RR %i %s\n\n", rr, resetReasonText[rr]);

  EEPROM.begin(512);
  byte test = EEPROM.read(0);
  if(test >= NTESTS) // un-initialised EEPROM read is random
    test = -1;
  test = ++test % NTESTS;
  Serial.printf("Test %i: %s\n", test, testText[test]); 
  EEPROM.write(0, test);
  EEPROM.commit();
  delay(5000);
  switch (test)
  {
    case 0: //  Run pin
      delayCount();
    case 1: // Power On
      delayCount();
    case 2: // reboot comand
      rp2040.reboot();
      delayCount();
    case 3: // watchdog
      rp2040.wdt_begin(2000);
      rp2040.wdt_reset();
      delayCount();  
    case 4: // restart command
      rp2040.restart();
      delayCount();
  }
}

// never gets here
void loop()
{
  delay(100);
}

[earlephilhower]

Nice test! Should be added as an example in the libraries/rp2040/examples directory.

[palmerr23]

OK, will do.

R

[retsifp]

Hi @palmerr23: Are you still planning to submit that PR? :)
I'd really like to see that feature, because I want to test how often the watchdog resets my device (and fix that) before OTA updating the end application...

If I can help, just let me know. :)

[palmerr23]

I've been working on other stuff.

I'll get back to it shortly.

[retsifp]

I found some time today and tested it, fixed some small issues and submitted it. PR #2516.
Hope that's fine for you. :)

[palmerr23]

Beautiful!

Thank you.