PCM 2.734 Problems with Main() in bank 1

Peter Pitzeier · Joined: 24 Apr 2005 Posts: 6 Location: Austria

Hello *,

As long as Main() resides in bank 0 my programs work. But as my functions get bigger PCM decides at some point to move Main() to 0x0800 and from there it seems that Main() don't even start.

Someone experienced similiar trouble? Is it a bug known to this compiler release? Workaround? How to track down the problem? I'm not too familiar with assembler, the reason why I use a C-compiler.

Any help appreciated!

TIA,

Peter.

PCM programmer · Joined: 06 Sep 2003 Posts: 21708

What PIC are you using ?

Are you doing anything unusual, such as:
1. Mixing #asm and C code.
2. Using lots of floating point.
3. Using #org statements.
4. Using #separate statements.

Peter Pitzeier · Joined: 24 Apr 2005 Posts: 6 Location: Austria

Device is 16F873. (Sorry, I didn't mention)
No #asm.
No floating point.
No #org.
No #separate.

Problems started when code size exceeded 50% (2K). Rom usage ist ~57% now.

I've played around with #org and #separate and sometimes it helps until code size increases again.
'
I'll have a LED blinking in main() to see, where it hangs. Also I'll post the LST-file of a non-working version.

tx

Mark · Joined: 07 Sep 2003 Posts: 2838 Location: Atlanta, GA

Not exactly your version but it works just fine:

danielefilisetti · Joined: 27 Apr 2005 Posts: 3 Location: Italy

Hi Peter,

I got the same problem on PIC16F913.

When program become longher than 50% the compiler moves main() to 0x800, and the program doesn't run correctly?

CCS compiler bug???? Twisted Evil

How to force main() in the bank0? Question

Daniele

Ttelmah · Guest

Any possibility you are using ICD?. If so, then the problem is that the 'main' program has got long enough that it no longer fits in bank0, because of the other overheads in this bank (interrupt handlers etc.), but is also long enough that it overwrites the upper memory area used by the ICD system, when loaded in bank1. Shortening the size of the 'main', should allow this to be avoided.

Best Wishes

Guest · Posted: Fri Apr 29, 2005 1:53 am

hello *,

found out that the problem isn't just main() in the upper bank. it's almost always when the ISR (T0 overflow in this case) resides in the other bank. I've looked at the ISR in the LST-file and really don't understand what manipulation of PCLATH (0x0A) is all about.

Guest · Posted: Fri Apr 29, 2005 1:56 am

No, I haven't got an ICD.

danielefilisetti · Joined: 27 Apr 2005 Posts: 3 Location: Italy

Interrupt logic should keep PCl and PCH in the stack and jump to vector 004 (PCH=0) automatically, while return from interrupt instruction should recover PCL and PCH.
So, different bank locations should not be a problem. or not Question

Daniele