[sdcc-devel] [ sdcc-Bugs-585901 ] for loop bug

[SourceForge.net <no...@so...>]

Bugs item #585901, was opened at 2002-07-24 14:05
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=100599&aid=585901&group_id=599

Category: None
Group: unreproducable
Status: Closed
Resolution: None
Priority: 2
Submitted By: Nobody/Anonymous (nobody)
Assigned to: Nobody/Anonymous (nobody)
Summary: for loop bug

Initial Comment:
When this is compiled i (R2) is initialised to 8 and is 
decremented instead of 0 and then incremented.
See the code and assembler below.
 
unsigned char CRC8_Serial(void)
{
 unsigned char CRC = 0, i;
 for(i = 0; i < 8; i++)
 {
  CRC = CRCTABLE[CRC ^ RomTemp[i]];
 }
 return CRC;
}

;------------------------------------------------------------
;Allocation info for local variables in 
function 'CRC8_Serial'
;------------------------------------------------------------
; C:\SportsSc\sports.c 708
; -----------------------------------------
;  function CRC8_Serial
; -----------------------------------------
_CRC8_Serial:
; C:\SportsSc\sports.c 710
 mov r2,#0x00
; C:\SportsSc\sports.c 715
 mov r3,#0x08  <-----------------THIS IS THE PROBLEM
00103$:
; C:\SportsSc\sports.c 713
 mov a,r3
 dec a
; Peephole 105   removed redundant mov
 mov  r4,a
 add a,#_RomTemp
 mov r0,a
 mov ar5,@r0
 mov a,r2
 xrl ar5,a
 mov a,r5
; Peephole 180   changed mov to clr
; Peephole 186   optimized movc sequence
 mov  dptr,#_CRCTABLE
 movc a,@a+dptr
 mov r2,a
 mov ar3,r4
 mov a,r3
; Peephole 109   removed ljmp by inverse jump logic
 jnz  00103$
00108$:
; C:\SportsSc\sports.c 715
 mov dpl,r2
00104$:
 ret
 

----------------------------------------------------------------------

Comment By: Bernhard Held (bernhardheld)
Date: 2002-07-24 20:58

Message:
Logged In: YES 
user_id=203539

Could you please show us your definition for CRCTABLE and    
RomTemp? What sdcc version?  
  
Using 2.3.2 and    
   unsigned char code CRCTABLE[256];    
   unsigned char RomTemp[256];    
doesn't reproduce the problem:   
> mov r3,#0x00  

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=100599&aid=585901&group_id=599