Hello, dear community, 

I write a program for Arduino UNO + Ethernet Shield. In my setup I run a TCP Client  which connects to a TCP Server. Server responds with one byte of information after connection establishment.  I use the Timer1 to measure how long i took for the server to respond after connection establishement. I need to use timer for maximal precision (counting ticks) 

I use Atmel Studio 6.2 + Visual Micro (Trial version) to debug my application. In my config I have chosen: 
Project Type: Arduino Sketch with C++ Class Example
Processor: ATmega328P
Board: Arduino Ethernet (there is either Arduino Uno or Arduino Ethernet possible to choose in the IDE, tried both, when switching Reload Tool Chain was done)
Project Properties->(Micro Debug): Full
Tools->Visual Micro checked: Automatic Debugging
Breakpoint Manager: Jump to Breakpoints
  
I have tried 2 types of debugging: 

1) Doing a live debug, i.e. I build project and start debugging right away. In this case if I don't use timer, then Client-Server communication works the way it should. However, my program hangs when I am attempting to set the TIMSK Register to enable the counter.
  
2) Debugging with a Simulator by opening an .elf file in IDE, switching from "No Tool" to "Simulator".  I am able to set the TIMSK and jump to my Interrupt Service Routine, when the timer overflows. However networking doesn't work. I track that with help of Wireshark.

I would very appreciate any help!

Here are some questions that I could not answer:
1) Which exactly board do I need to choose: Arduino UNO or Arduino Ethernet. In my case they are both combined.
2) Is the Simulator for those boards are designed to perform networking

But the most important for me is to answer why debugging with timer doesn't work. Here is a code snippet of my program:
ISR(TIMER1_OVF_vect){      sreg = SREG;

     digitalWrite(ledPin, digitalRead(ledPin)^ 1);      // invert the bit value on the ledPin      overflowCounter++;
     TIMSK1 &= ~(1 << TOIE1);      //disable interrupt on timer1
SREG = sreg; }

// the setup routine runs once when you press reset:
void setup() {                
  // initialize the digital pin as an output.
 
  Serial.begin(115200);
  pinMode(ledPin, OUTPUT);    
  
           //+Serial.println("\r\nSetup...");
           //+Serial.print("\r\nInitializing the w5100 chip and open a TCP socket...");
           W5100.init(); //
           W5100.writeSnMR(s, SnMR::TCP); // set type of socket in Mode Register

           W5100.setMACAddress(smac); // set own mac address
           W5100.setIPAddress(src_ip_addr); // set own ip address
           W5100.writeSnPORT(s, src_port); // set own port (for each socket different)
           W5100.setSubnetMask(subnet_mask);
           W5100.setGatewayIp(gateway_ip);

           W5100.writeSnDHAR(s, dst_mac); // set destination mac address
           W5100.writeSnDIPR(s, dst_ip_addr); // set destination ip
           W5100.writeSnDPORT(s, dst_port); // set destination port

           W5100.writeRMSR(0x03); // RX Memory Size Register. First two bits set to 1, this corresponds to assigning all 8 kByte of RX memory for the socket 0
           W5100.writeTMSR(0x03); // TX Memory Size Register. First two bits set to 1, this corresponds to assigning all 8 kByte of TX memory for the socket 0

           W5100.writeIMR(0x01); // Interrupt Mask Register is set to enable interrupts on socket 0.

           W5100.execCmdSn(s, Sock_OPEN);
                 Serial.print("\r\nDone opening socket");
                 Serial.print("\r\n");

           memset(data, 0, PAYLOAD_SIZE);
           strcpy(data, "A");
           len = strlen(data);
           
           cli();
           TCCR1B = 0;
           TCCR1A = 0;
           TCNT1 = 65300;
           // prescaler
           TCCR1B |= (1 << CS10);      // Time/Counter Control Register B no prescaler needed, since we count amount of ticks
           sei();
}

// the loop routine runs over and over again forever:
void loop() {
            // wait for a second
     
           if(measurementsCounter == 1000)
           while(true){}                  // endless loop after desired amount of measurements achieved

           W5100.execCmdSn(s, Sock_CONNECT);

           while ((W5100.readSnSR(s) & SnSR::ESTABLISHED) != SnSR::ESTABLISHED) { // check Status Register to find out if the connection was established

           }
           TCNT1 = 0; // reset counter. Note that the timer counts always, whether interrupt occurs decides bit in TIMSK
           TIMSK1 |= (1 << TOIE1); // enable timer overflow interrupt. Note we OR the current state of the TIMSK, so once set it is always set, if manually not changed

           while ((W5100.readSnIR(s) & SnIR::RECV) != SnIR::RECV) {}
           restTicks = TCNT1; // save rest ticks
           TIMSK1 &= ~(1 << TOIE1); // disable timer1.
           W5100.writeSnIR(s,SnIR::RECV);      // clear the interrupt bit

           time = (overflowCounter * 65535 + restTicks) * 62.5;
           measurementsCounter++;

           overflowCounter = 0;      // clear overflowCounter

           Serial.print("\r\nTerminate Connection");
           W5100.execCmdSn(s, Sock_DISCON);

           while ((W5100.readSnIR(s) & SnIR:ISCON) != SnIR:ISCON) {
           }
           W5100.writeSnIR(s,SnIR:ISCON);      // clear the interrupt bit
                 Serial.print("\r\nDisconnected.");

}

Hello, @Tim Leek,

thanks for such a quick reply. Here is a link to my UNO board :
http://www.pollin.de/shop/dt/NzE4OTgxOTk-/Bausaetze_Module/Entwicklerboards/ARDU...

and to ethernet shield:

http://www.pollin.de/shop/dt/MTA4OTgxOTk-/Bausaetze_Module/Entwicklerboards/ARDU...



I have tried to use breakpoints at different lines. In particular, i placed one against 

---> TIMSK1 |= (1 << TOIE1);

when breakpoint is hit, and I resume, then program hangs forever. As far as I know debugging uses Timer0, and I use Timer1, so debugging should not collide with my timing? Isn't that right?

> Visual Micro uses the millis() function to determine when a timed breakpoint should happen.
Are all breakpoints are timerd breakpoints?

In live debug modus (when flashing the board) Serial.print() works just fine. 

Thanks for help!

Best regards



ISR(TIMER1_OVF_vect){      sreg = SREG;

     digitalWrite(ledPin, digitalRead(ledPin)^ 1);      // invert the bit value on the ledPin      overflowCounter++;
     TIMSK1 &= ~(1 << TOIE1);      //disable interrupt on timer1
SREG = sreg; }

and

Sorry ignore my comment about millis and timer. I thought were using using timed breakpoints but I now see you were simply saying you had a breakpoint in your own timer function.

Question. does this work?

 TIMSK1 |= (1 << TOIE1);Serial.println("Timer function here"); 

I also suggest setting project property ThrottleEnabled = False

Switching off the throttle will prevent Visual Micro from altering timings with pauses. The throttle is for new users who put unconditional breakpoints in the loop() causing thousands of debug messages to be sent to the pc

Thanks

Hi,

Thanks

If you upload without any breakpoints does Serial.println("Timer function here"); print a message?

Thanks

Hi,

It is not possible do do much at all in an ISR. The best way is to save state as volatile variable then report on it back in normal code.

I suspect that a serial.print() in an ISR will also fail.

Thanks

I know that ISRs are for doing a short job, and I was not going to do serial communication within the ISR. As you see in the code above I do Serial output inside my loop() function. And I am curious why doesn't the serial output was incomplete. 

I generally don't understand how does the Serial output relate to the Timer and my ISR for my case. I would be grateful for the clarification! 

P.S.: I was previousely writing assembly code for ATmega8 Asuro in Atmel Studio (older version). And there I was able to enter the ISR and track what is going on there. Same thing I was expecting to do here.

Thanks for all your replies.

I never dig into serial code to see how it operates, but on occasion I had serial.print() instructions that never send anything to the window event though I know the program came across those instructions.

I recall Tim telling the serial buffers are small, could it be some data was erased before the instructions were executed ? It's one of those unsolved mysteries  .

@Tim Leek and @Gilles Plante, 

thank you very much for sharing your thoughts. It also was breaking my head trying to understand why I don't get messages printed. In fact it was because I was using Serial.print() instead of Serial.println().

Cheers