flashforth-devel

[Flashforth-devel] Switching Endians

[craig b. <dab...@ya...>]

Well Mike, unless the Pic Wreg is a problem or there's some hidden instruction assembled that you haven't warned us about, no, there isn't.  Never made any rash claims about Atmels... Never met up with one.  8-bit Pic assembler is my playground for the last few decades, though and if you see something I've missed in that word, I'll be impressed (grin).

craig

 
...averaging may or may not help clarify data. 
   If Bill Gates walks into a bar, 
   the average patron becomes a millionaire...


That's great Craig, but can you spot the bug in your code ?

It can also be done with temp registers, since FF has alot of interrupt 
safe temp registers, Tp(fsr1) in this case.

[Flashforth-devel] Switching Endians

[craig b. <dab...@ya...>]

The way you wrote the initial stack allocation and the size limit you specified won't allow it to cross a page boundry, therefore there's no bug. (grin)  Thought of that when reverse engineering the locations the sp pointed at.  Re-read the source to be sure.   Actually, my first try had the FSR0 inc and dec reversed with amusing results...


craig

 
...averaging may or may not help clarify data. 
   If Bill Gates walks into a bar, 
   the average patron becomes a millionaire...


You cannot inc/dec the 12 -bit fsr pointer over a 256 byte page boundary :-)
It will stay on the same page.

You have to use the  special registers that change the whole pointer, or 
deal with the carry status.

Re: [Flashforth-devel] Switching Endians

[Mikael N. <mik...@pp...>]

It's  a bug in a FlashForth system.
All stack manipulating words are written so that the stack can be anywhere.

Let's say the word is used in one of many background tasks.
It's a bit difficult to predict if the word will be used on a page 
boundary or not.

If you don't use the word in a BG task you will in practice get away 
with it.
But somebody else may not and be hit with a really subtle bug.

Peace ! Mike

On 03/27/2014 09:57 PM, craig bair wrote:
> The way you wrote the initial stack allocation and the size limit you 
> specified won't allow it to cross a page boundry, therefore there's no 
> bug. (grin)  Thought of that when reverse engineering the locations 
> the sp pointed at. Re-read the source to be sure.   Actually, my first 
> try had the FSR0 inc and dec reversed with amusing results...
>

Re: [Flashforth-devel] Switching Endians

[Mikael N. <mik...@pp...>]

You cannot inc/dec the 12 -bit fsr pointer over a 256 byte page boundary :-)
It will stay on the same page.

You have to use the  special registers that change the whole pointer, or 
deal with the carry status.

My comment about Atmel was only that since FlashForth supports three
processor architectures I try to align the wordnames for those as much 
as possible.

BR Mike

On 03/27/2014 09:28 PM, craig bair wrote:
> Well Mike, unless the Pic Wreg is a problem or there's some hidden 
> instruction assembled that you haven't warned us about, no, there 
> isn't.  Never made any rash claims about Atmels... Never met up with 
> one.  8-bit Pic assembler is my playground for the last few decades, 
> though and if you see something I've missed in that word, I'll be 
> impressed (grin).
>