Adds noInterrupt() and interrupt() functionality

[SuGlider]

Description of Change

Adds noInterrupt() and interrupt() to Arduino.
This must be used carefully because ESP32 Arduino runs under FreeRTOS and it is easy to get Watchdog Timeout error if it is no correctly used.

Tests scenarios

Tested with this sketch (as suggested in the issue)

//  To run the example sketch to demonstrate the issue, 
//  two digital pins need to be connected together somehow. 
//  This was done on a breadboard using a jumper wire.

#define OUT_PIN 14
#define IN_PIN 32

int count = 0;

void counter() {
  count++;
}

void setup() {
  Serial.begin(115200);

  pinMode(OUT_PIN, OUTPUT);
  digitalWrite(OUT_PIN, LOW);
  pinMode(IN_PIN, INPUT);

  attachInterrupt(digitalPinToInterrupt(IN_PIN), counter, RISING);

  for (int i=0; i<50; i++) {
    digitalWrite(OUT_PIN, HIGH);
    digitalWrite(OUT_PIN, LOW);
  }

  noInterrupts();
  
  for (int i=0; i<50; i++) {
    digitalWrite(OUT_PIN, HIGH);
    digitalWrite(OUT_PIN, LOW);
  }
  
  interrupts();

  Serial.print("count = "); Serial.println(count);
}

void loop() {
}

Expected output:

ets Jun  8 2016 00:22:57

rst:0x1 (POWERON_RESET),boot:0x13 (SPI_FAST_FLASH_BOOT)
configsip: 0, SPIWP:0xee
clk_drv:0x00,q_drv:0x00,d_drv:0x00,cs0_drv:0x00,hd_drv:0x00,wp_drv:0x00
mode:DIO, clock div:1
load:0x3fff0030,len:1344
load:0x40078000,len:13864
load:0x40080400,len:3608
entry 0x400805f0
count = 50

wrong output (before this PR) says count = 100

Related links

Fixes #7211

[PilnyTomas]

Tested on S3 and I'm getting 51 instead of 50. Tried various pins:
OUT | IN
14 | 13
14 | 47
40 | 47

[caternuson]

Thanks. This seems to resolve the issue with the library here:
adafruit/Adafruit_VS1053_Library#45

However, I'm also seeing the extra count show up in the simple test sketch. Revised test sketch:

#define OUT_PIN 14
#define IN_PIN 32

int count = 0;

void counter() {
  count++;
}

void setup() {
  Serial.begin(9600);

  pinMode(OUT_PIN, OUTPUT);
  digitalWrite(OUT_PIN, LOW);
  pinMode(IN_PIN, INPUT);

  attachInterrupt(digitalPinToInterrupt(IN_PIN), counter, RISING);

  for (int i=0; i<50; i++) {
    digitalWrite(OUT_PIN, HIGH);
    digitalWrite(OUT_PIN, LOW);
  }

  Serial.print("count = "); Serial.println(count);

  noInterrupts();

  for (int i=0; i<50; i++) {
    digitalWrite(OUT_PIN, HIGH);
    digitalWrite(OUT_PIN, LOW);
  }

  interrupts();

  Serial.print("count = "); Serial.println(count);
}

void loop() {
}

Outputs:

count = 50
count = 51

[P-R-O-C-H-Y]

I am seeing just same result as guys mentioned above, tested on ESP32.
I think it may be due to interrupt flag is still set in registers after noInterrupts();, but its triggered just after interrupts(); is called.

[P-R-O-C-H-Y]

I have edited the sketch @caternuson posted to demonstrate, that my previous comment really happening.
I have added next print of count right before interrupts() is called.

#define OUT_PIN 14
#define IN_PIN 32

int count = 0;

void counter() {
  count++;
}

void setup() {
  Serial.begin(115200);

  pinMode(OUT_PIN, OUTPUT);
  digitalWrite(OUT_PIN, LOW);
  pinMode(IN_PIN, INPUT);

  attachInterrupt(digitalPinToInterrupt(IN_PIN), counter, RISING);

  for (int i=0; i<50; i++) {
    digitalWrite(OUT_PIN, HIGH);
    digitalWrite(OUT_PIN, LOW);
  }

  Serial.print("count = "); Serial.println(count);

  noInterrupts();

  for (int i=0; i<50; i++) {
    digitalWrite(OUT_PIN, HIGH);
    digitalWrite(OUT_PIN, LOW);
  }
  
  Serial.print("count = "); Serial.println(count);
  
  interrupts();

  Serial.print("count = "); Serial.println(count);
}

void loop() {
}

Output is now:

count = 50
count = 50
count = 51

[P-R-O-C-H-Y]

From my point of view it works as expected, there is no easy way to clear all coming interrupts before Interrupts() is called. As suggested you must be careful when using noInterrupts(), and its better to use detachInterrupt().

[me-no-dev]

You could have define sei and cli instead for backwards compatibility with older code. Any reason you didn't @SuGlider?

[SuGlider]

As far as I know, SEI and CLI are ASM instructions used in ATMEL chips.
Arduino official API is interrupts() and noInterrupts()
I see no reason for defining those ATMEL specific functions.

//interrupts is defined as:

#define interrupts() sei()
//and sei is defined as for the UNO/328p:

# define sei()  __asm__ __volatile__ ("sei" ::)

[SuGlider]

Looking ESP8266 Arduino.h I see that it also doesn't define sei() nor cli().
My vote is for keeing it as defined in this PR.

[PilnyTomas]

interrupts() and noInterrupts() are described in Arduino Language Reference so I vote for this too.

[SuGlider]

OK, added back.

[PilnyTomas]

LGTM

[me-no-dev]

sei() and cli() are AVR functions present in lots of old Arduino code. I see no fault to have them there

[PilnyTomas]

sei() and cli() are AVR functions present in lots of old Arduino code. I see no fault to have them there

Having both variants should not hurt...

[me-no-dev]

@SuGlider will merge as soon as you re-add sei and cli :)