compiler warning: "interrupts disabled"

[valemike]

I get a compiler warning that says:
"Line 430(0,1): Interrupts disabled during call to prevent re-entrancy."

However, my source code ends at line 429!

422 void panic()
423 {
424 while (1)
425 {
426 printf("PANIC!! ");
427 }
428 }
429

Why would I get a compiler warning with code that isn't even there? By the way, by merely adding this function in a 429-line program, I get this warning. The previous function didn't generate this warning.

Thanks,
Mike

[PCM programmer]

My guess is that you're calling printf from within an interrupt service
routine, somewhere else in your program.

If possible, avoid doing that. Instead, just set a global flag and
poll it in a while(1) loop, in main(). Then if the flag is set, clear it,
and do the printf there.

[valemike]

I'm not calling printf inside an ISR.

Like I said, the warning only shows up when i append that panic() function in there. Perhaps since it's in an infinite while(1) loop, then CCS proacts with that warning?

[PCM programmer]

I can't get the compiler to give me that error.
What compiler, version, and PIC are you using ?

[valemike]

I'm using a PIC16F876, with the latest corresponding CCS compiler, 3.186.

#int_ext
void extint_isr(void)
{
delay_ms(20);
if (!input(PIN_B0) && ready_flag == 1)
{
latch_asserted = 1;
}
}

I commented out that panic() function, and still found the warning, so disregard what i said about the panic() function.

In my ISR, i use a delay_ms(20) to debounce a switch for 20ms, as you see above. If i comment out that delay_ms(20), then CCS doesn't generate that warning. Is there something wrong with using delay_ms() in an ISR?

The strange thing is that the warning line number refers to one line after the end of the .c file.

[PCM programmer]

This is a common question. See this recent thread.
http://www.ccsinfo.com/forum/viewtopic.php?t=18356
At the end of the thread, Darren Rook has posted a link
to a work-around in the FAQ.

[valemike]

Okay, I read other posts concerning the use of delay_ms() in an ISR. Tell me if this conclusion is correct:

The compiling of delay_ms() usually does not lead to a disabling of interrupts. However, if I use delay_ms() in an isr, then it is possible that i was in the middle of a previous delay_ms() upon getting interrupted.

If delay_ms() is then called again in the isr, then it will screw up my previous delay_ms, and that is why CCS will then disable interrupts in EVERY delay_ms() call, once it finds out that delay_ms() is used in an isr. Correct?

Therefore if I do not want delay_ms() to disable interrupts, then I should find a different method of delaying for 20ms while in the ISR. Perhaps use delay_us or something, or use a software delay for-loop.
So as long as i don't use delay_ms in any isr, then i won't get the compiler warning, and i will not lock out a pending interrupt in delay_ms() calls. Correct?

[valemike]

Thanks for that link from Darren. (You posted it while i was typing a new msg, so it looks like i ignored it.)
-Mike

[Ttelmah]