• State
  • Locked Locked
  • Replies 10 replies
  • Subscribers 85 subscribers
  • Views 527 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

GPIO compiler bugs in setting consecutive IOs

anusheel nahar
Prodigy 210 points
I need to set GPIO2 and 3 , 1 and 0 respectively. However if I write them in consecutive statements GPIO2 never sets. On the scope I see a narrow pulse on GPIO2 of about 250 us .
If I add a dummy i++ as shown below it works.
GpioDataRegs.GPADAT.bit.GPIO2= 1;
i++;
i++;
i++;
GpioDataRegs.GPADAT.bit.GPIO3 = 0;
It happens with GPIOs10 and 11 also.  (consecutive Ios). Its a showstopper as I need to set and reset Ios hundreds of places in my application and my confidence is low on it. I cannot possibly put i++ everywhere.
the GPIOtoggle command works perfectly as I suspect its a multiple assembly instructions.
I suspect its a compiler bug. pls help. I am using piccolo 28069
  • 0
    Guru 192875 points

    Hi Anusheel,

    Can you be a bit precise on what exactly are you looking for? In short what kind of waveform on which GPIOs?

    anusheel nahar said:
    I need to set GPIO2 and 3 , 1 and 0 respectively.

    Is it PWM kind of waveform you're looking for?

    Regards,

    Gautam

  • 0 in reply to Gautam Iyer
    Prodigy 210 points

    i just need GPIO2 high and GPIO3 low. If I put those statements consecutively ,GPIO2 behaves erratically. instead of remaining hi , there is a pulse of 250 us. I had to add dummy i++ in between to make it work. same thing with GPIO10 and 11. seems to be compiler stupidity.

  • 0 in reply to anusheel nahar
    Guru 192875 points

    anusheel nahar said:
    i just need GPIO2 high and GPIO3 low.

    Try this simpe code out:

    EALLOW;
   //GPIO
   GpioCtrlRegs.GPAMUX1.bit.GPIO2 = 1;  // GPIO2 = GPIO2
   GpioCtrlRegs.GPADIR.bit.GPIO2 = 1;   // GPIO2 = output
   
   GpioCtrlRegs.GPAMUX1.bit.GPIO3 = 1;  // GPIO3 = GPIO3
   GpioCtrlRegs.GPADIR.bit.GPIO3 = 1;   // GPIO3 = output
EDIS;
   
   
   GpioDataRegs.GPASET.bit.GPIO2 = 1;  
   GpioDataRegs.GPACLEAR.bit.GPIO3 = 1;

    Regards,

    Gautam

  • 0 in reply to Gautam Iyer
    Prodigy 210 points

    All that settings is correct.

    GpioDataRegs.GPASET.bit.GPIO2 = 1;  
   GpioDataRegs.GPACLEAR.bit.GPIO3 = 0; 

    if I write this GPIO2 is always 0 except for very narrow pulse of 500 us . I put them in for loop with delay and I see pulses periodically



    if I write this
     GpioDataRegs.GPASET.bit.GPIO2 = 1;  
    i++;
   GpioDataRegs.GPACLEAR.bit.GPIO3 = 0; 

    where i is a dummy , it behaves correctly.  The code you sent wont/isnt working properly. 
    i can always do this  way but it reduces my confidence in compiler /TI things in general and is very irritating . Whenever I toggle some io , i dont know whether the next (unrelated) statement will mess it up . I think its a serious show stopper and TI shud investigate it.



  • 0 in reply to anusheel nahar
    Guru 192875 points

    That's very unusual, Anusheel. Its working perfectly on my controlStick!

    Please mention your CCs and compiler version so that a TIer can take care of you query. I'm also looping this to a CCS expert.

    Regards,

    Gautam

  • 0 in reply to anusheel nahar
    Guru 30330 points

    Anusheel,

    This is a well known behavior of GPIOs on C2000.  

    Please refer to the following pages for more information.

    http://processors.wiki.ti.com/index.php/General_Purpose_IO_(GPIO)_FAQ_for_C2000#Q:_Back-to-back_DAT_register_writes_do_not_work_as_expected

    http://e2e.ti.com/support/microcontrollers/c2000/f/171/t/319643.aspx

  • 0 in reply to Mark Labbato
    Guru 192875 points

    Thanks for the info, Mark! In short one will have to use a NOP instruction in between.

    Regards,

    Gautam

  • 0 in reply to Mark Labbato
    Prodigy 40 points

    Mark,

    The way I read the first article shows in its example code that the problem is with GPADAT. Then it goes on to say:

    "One solution is to use CLEAR/SET registers for this example."

    Anusheel's code is already using GPACLEAR/GPASET.

  • 0 in reply to Gene Morgan1
    Expert 1806 points

    Gene Morgan1 said:
    Anusheel's code is already using GPACLEAR/GPASET.

    Yes, but in the wrong way.

    The statement:

    GpioDataRegs.GPACLEAR.bit.GPIO3 = 0;

    has no effect. Only writes of "1" will clear the GPIO.

    Regards, Andy

  • 0 in reply to AndyP
    Prodigy 40 points

    Ah, just using the correct regs is a necessary though not sufficient condition.

    Thanks for the catch.