[sdcc-devel] [ sdcc-Bugs-845089 ] error in produced assembly code

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

Bugs item #845089, was opened at 2003-11-19 13:36
Message generated for change (Comment added) made by kflittner
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=100599&aid=845089&group_id=599

Category: Live range problems
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Nobody/Anonymous (nobody)
>Assigned to: Klaus Flittner (kflittner)
Summary: error in produced assembly code

Initial Comment:
reply to omer.enbar [at] ceibo [dot] com
(ceibo.com)
hello,

command line:
sdcc.exe err_sample.c --model-large -Lc:\sdcc\lib\large

version
SDCC : 
mcs51/gbz80/z80/avr/ds390/pic14/pic16/TININative/xa51
/ds400/hc08 2.3.6 (N
ov  3 2003) (MINGW32)



the incorrect output:
a = 5;
b = 6;
swap(&a, &b);

should be: a=6; b=5; 
but gets a=6; b=6;

I tracked down the error in the assembly code, if needed.
also, the bug only occurs when the function is reentrant.
I also tried to remove any optimizations (--noinduction, -
-no..., etc...) but it still doesn't work.

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

>Comment By: Klaus Flittner (kflittner)
Date: 2003-11-21 20:53

Message:
Logged In: YES 
user_id=879538

Yes, this seems to be the problem.
The best solution IMO would be, to do the LR computation
after the 
register packing and only the really needed parts before.
If there are no objections i will do this.

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

Comment By: Erik Petrich (epetrich)
Date: 2003-11-19 19:55

Message:
Logged In: YES 
user_id=635249

Looks like packRegsForAssign is substituting iTemp3 for
iTemp1, but not transfering the clash between iTemp1 &
iTemp6 to iTemp3?

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

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=100599&aid=845089&group_id=599