UDOO Neo A9/M4 serial communication on /dev/ttyMCC buggy

[ddewaele]

I'm having issues with the serial communication between the A9 and M4 on my UDOO Neo

When reading data from the serial line it eithers

• hangs completely
• runs very slow
• contains corrupt data

Any ideas on how to fix this ?

[ektor5]

Hi there,

Can you please supply some sketches in order to reproduce the issue? Also some generic info like os, kernel version, branch, etc...

Thanks! Ek5 On 8 Dec 2015 16:11, "Davy De Waele" notifications@github.com wrote:

I'm having issues with the serial communication between the A9 and M4.

When reading data from the serial line it eithers

• hangs completely
• runs very slow
• contains corrupt data

Any ideas on how to fix this ?

— Reply to this email directly or view it on GitHub https://github.com/UDOOboard/linux_kernel/issues/14.

[ddewaele]

It's any sketch basically that sends out some data on the serial port.

The sketch that makes it crash here :

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

void loop()
{
  int lastReading = 0;
  Serial.print("{");
  for(int i=0;i<6;i++) {
     if(i>0) {
       Serial.print(",");
     }
     Serial.print("\"a");
     Serial.print(i);
     Serial.print("\":");
    lastReading = analogRead(0);
     Serial.print(analogRead(i));
  }
  Serial.print(",\"volts\":\"");
  Serial.print(lastReading * (3.3 / 4095.0));
  Serial.print("v\"");

  Serial.println("}");
  delay(500);
}

I'm using a stock UDOO Neo and only enabled tunneling support on the kernel

Linux udooneo 3.14.28 #1 SMP PREEMPT Fri Dec 4 20:50:04 JST 2015 armv7l armv7l armv7l GNU/Linux

Here's the output from the serial line

ddewaele@udooneo:~$ sudo cat /dev/ttyMCC 
{"a0":1,"a1":756

{"a0":1,"a1":880,"a2":887,"a3":951,"a4":1089,"a5":2889,"volts":"0.00v"}

{"a0":1,"a1":796874,"a3":1203,"a5":3000,"volts":"0.00v"}
{"a0":1,"a1":899870,"a3":1071,"a5":0.00v"}

{"a0":1,"a1":894,"a3":1009,"a43000,"volts":"0.00v"}

<-----  it basically halts here. Also notice how the data is corrupted (empty lines, missing elements in the json , ....) 

After a couple of lines the serial line goes dead, and you need a board reboot to get it up and running again.

Also, when I use nodejs libraries like https://github.com/voodootikigod/node-serialport i also get corrupted data (using a different sketch that only outputs 1 analogReading)

data received: 1963

data received: 1962

data received: 1961
data received: 

data received: 1883
data received: 

data received: 1623
data received: 

data received: 1383
�������������΅~�~�t������������~������w������s[����������k������������5����������t_�|y�Wߚ�����������������������M��������������7����ݯ����������_��������o����~������ߟ������������������}Y߿����{߿w��~����}O������s��w��}����������|������.��=���}����}�z�W_}���_��~~g��oO�v����#��}����O��������V|�����������s������~��_���?����������[�������������߿���9}�=_���{���wu��?�1�߿z>�y��?��~���~{��ܿ���{����M�y���g�_�o��?��_����?��������߷}��������?�����wן��f�?���ҷ��������y{���g���c�����������������.�����[����}?��������o���������������}�������}/��������������������}�u��;��u��~|�%��K������[�����^�}������[}��m������Y��������_���������[}�T}������w������Gk��˙s�������d;������{�����������o��o�p�{���}����kw�s���>���������s������������>q��w~������^�=��������۫���}]���s�o�����s������W����y�?�����̟�w���������������~�����������澼����?>���}�����

[ddewaele]

I have the impression it's caused by the analogRead calls.

Also not really sure if this below is related (or should even be logged as an issue here) , but if I take the traditional blink sketch and I add some Serial output

void setup() {
  Serial.begin(9600);
  pinMode(13, OUTPUT);
}

// the loop function runs over and over again forever
void loop() {
  Serial.println("LED ON");
  digitalWrite(13, HIGH);   // turn the LED on (HIGH is the voltage level)
  delay(1000);              // wait for a second
  Serial.println("LED OFF");
  digitalWrite(13, LOW);    // turn the LED off by making the voltage LOW
  delay(1000);              // wait for a second
}

It actually blinks the LED with 4 second intervals (as opposed to 1 second intervals)

As soon as I start reading from the Serial (using sudo cat /dev/ttyMCC) it blinks with 1 second intervals. If I exit the cat it again blinks with 4 second intervals. If I remove the Serial code it behaves normally.

[tcmichals]

One issue is with the imx6sx-udoo-neo-m4.dtsi and imx6sx-udoo-neo.dtsi:

imx6sx-udoo-neo: memory { linux,usable-memory = <0x80000000 0x3f800000>; <----- this should be 0x3f700000 reg = <0x80000000 0x40000000>; }; 0x80000000 - 0x800000 (M4) - 0x100000 (MCC)

imx6sx-udoo-neo-m4.dtsi already has this memory for the M4 so the above can be the full range...

reserved-memory {

    #address-cells = <1>;
    #size-cells = <1>;
    ranges;

    /* Reserved 8Mb for M4 */
    m4_memory:  m4@0x84000000 {
        no-map;
        reg = <0x84000000 0x00800000>;
    };

    /* Reserved 1Mb for MCC */
    mcc_memory:  m4@0xBFF00000 {
        no-map;
        reg = <0xBFF00000 0x00100000>;
    };
};

Also, for the basicks or basic neo with only 512MB of memory; MCC memory needs to be mapped to 0x9FF00000 . I'm still in the process of testing the basicks. I don't have neo with 1GB of memory.

Just need to choose which way is better reserved-memory or memory to define where reserved memory is used, but not both

[ektor5]

Commit abdce68f40037226fc0ae3300ca75252ccb5ac02 should fix most of the issues. Please try it out and feedback.

Thanks, ek5

[graugans]

The link seems to be broken. The fixed one: abdce68f40037226fc0ae3300ca75252ccb5ac02

[ektor5]

Thanks, fixed :+1:

[tcmichals]

• if (num_of_received_bytes > 0) {
  • /* flush the recv-ed data to tty node */
  • tty_msg.data[num_of_received_bytes++] = 0;

What happens if num_of_received_bytes is == (MCC_ATTR_BUFFER_SIZE_IN_BYTES - 24)? there is a int16_t after the data member in the data structure, that would be used to old the NULL byte.

[waltervl]

void setup() { Serial.begin(9600); pinMode(13, OUTPUT); }

// the loop function runs over and over again forever void loop() { Serial.println("LED ON"); digitalWrite(13, HIGH); // turn the LED on (HIGH is the voltage level) delay(1000); // wait for a second Serial.println("LED OFF"); digitalWrite(13, LOW); // turn the LED off by making the voltage LOW delay(1000); // wait for a second }

It actually blinks the LED with 4 second intervals (as opposed to 1 second intervals) As soon as I start reading from the Serial (using sudo cat /dev/ttyMCC ) it blinks with 1 second intervals. If I exit the cat it again blinks with 4 second intervals. If I remove the Serial code it behaves normally.

This issue that the not reading the /dev/ttyMCC delays the execution of the program with seconds is still there. So not solved.

I also still see missing EOL in the Arduino IDE Serial Monitor

[FrancescoFerraro]

If M4 binary works with MCC (Serial), on A9 side must be opened the endpoint for receive (cat/dev/ttyMCC), otherwise it will time out every Serial.println().

[gefla]

Still crashes for me with abdce68f40037226fc0ae3300ca75252ccb5ac02. Strangely, I'm not getting a crash dump even though I set CONFIG_CRASH_DUMP=y, just "Kernel panic - not syncing: Aiee, killing interrupt handler!" on the serial consiole.

config.txt

[marcdraco]

In my experience (YMMV) reliability is directly correlated to the load on the A9 - and LXDE/XOrg are the main problems there. Higher level languages like Python probably take a toll too.

[waltervl]

A lot has improved lately. No more missing lines or characters. Serial.println() still time outs if not read on A9 side though. Will these be fixed or is this as designed?

[Tengoles]

I've been having issues with the serial comm too, if I update my udoobuntu with this: https://github.com/UDOOboard/linux_kernel/commit/abdce68f40037226fc0ae3300ca75252ccb5ac02

Is it supposed to improve things? how do I update my udoobuntu with it? sorry for noob questions I'm new in a lot of this things.

[waltervl]

@Tengoles you better ask these kind of things on the forum. To update Udoobuntu see the documentation: http://www.udoo.org/docs-neo/Software_&_Operating_Systems/UDOObuntu.html