RE: [avr-gcc-list] Is this a Stack problem?

[Brian Cuthie]

Basically, you need to read the Atmel data sheet very carefully. But your
interrupt handler will look something like this (in pseudo code):

// transmitter interrupt
// BTW, we're not interested in the TXC interrupt.
UDRE_Interrupt() {
        
        IntProlog();            // you really need this

        if (xmitbuffer count > 0) {
                UDR = *(xmitbuffer pointer)    // this clears the interrupt
                increment xmitbuffer pointer
                decrement xmitbuffer count
        }
        else {
                disable interrupt by writing UDRIE to 0
                notify outer (non-interrupt) task that send is complete
        }

        Epilog();
}

// receiver interrupt
RXC_Interrupt() {

        IntProlog();    // don't ever forget this!

        if (recvbuffer size > 0) {
                *(recvbuffer pointer) = UDR
                increment recvbuffer pointer
                decrement recvbuffer size
        }

        if (recvbuffer <= 0) {
                disable receiver interrupt by writing 0 to RXCIE
                notify outer (non-interrupt) task that receive is complete
        }

        Epilog();
}

I typically use AvrXIntSetSemaphore to notify the task that started the I/O
that it's compelete.

I doubt the interrupt is getting enabled inside his interrupt handler. But I
suspsect that he's not clearing the interrupt in some cases.

-brian


> -----Original Message-----
> From: address@hidden 
> [mailto:address@hidden On Behalf Of Lorne Gutz
> Sent: Tuesday, January 20, 2004 9:42 AM
> To: Eric Fu; address@hidden
> Subject: Re: [avr-gcc-list] Is this a Stack problem?
> 
> If you check you will find that there are 2 interrupts 
> involved with the UART transmitt.
> Basicly you must use one to enable the other when the UART is 
> ready to except another character.  
> From what you have stated below I believe that your code has 
> rammed 11 characters into the UART before anything
> ever gets sent.    You could easly have a stack overflow if
> your code is reenterand, and interrupt gets enabled inside your ISR.
> 
> cheers
> Lorne
> 
> 
> 
> On Monday 19 January 2004 16:41, Eric Fu wrote:
> > Hi All,
> >
> > I'm writing a test code to implement a SW half duplex 
> interrupt UART 
> > with
> > ATmega16 / STK500. The code passes the compiler (modified from a 
> > successful HW UART code). However, it doesn't work. It couldn't 
> > transmit anything, nothing happens at the transmit pin. The 
> degugging 
> > I did shows the
> > following: a.. The timer interrupt routine has been entered 
> 22 times 
> > before it stops. And it is repeatable. Note: the timer period is 
> > defined same as bit rate. It supposed to transmit 11 ASCII code, 
> > equivalent to 220 timer interrupt entries. b.. If I run the 
> Simulator 
> > in AVR Studio 4.08 in step mode, it could transmit the all the 11 
> > characters, at least this is what I saw in the Simulator, and the 
> > transmit pin behaves as expected (in the
> > Simulator) c.. If I run in continuos mode with a breakpoint set 
> > immediately after the 11 characters, it took 28 Secs to complete, 
> > however, it seems it does it connectedly. As I don't have a 
> emulator 
> > at the moment, it is hard for me to find the fault.
> >
> > Could it be Stack problem? If it is, how to fix it?
> > The compile message for code size is
> >
> > Size after:
> > RF001.elf  :
> > section     size      addr
> > .text       1622         0
> > .data         12   8388704
> > .bss         137   8388716
> > .noinit        0   8388853
> > .eeprom        0   8454144
> > .stab       7212         0
> > .stabstr    3158         0
> > Total      12141
> >
> > I'm not sure what stab and stabstr are. Are they supposed 
> to be like that?
> >
> > Thanks
> >
> > Eric Fu
> >
> > _______________________________________________
> > avr-gcc-list mailing list
> > address@hidden
> > http://www.avr1.org/mailman/listinfo/avr-gcc-list
> 
> 
> _______________________________________________
> avr-gcc-list mailing list
> address@hidden
> http://www.avr1.org/mailman/listinfo/avr-gcc-list
>