[sdcc-devel] [ sdcc-Bugs-2944399 ] short or truncated to 8-bit

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

Bugs item #2944399, was opened at 2010-02-02 06:39
Message generated for change (Comment added) made by maartenbrock
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=100599&aid=2944399&group_id=599

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: mcs51(8051) target
>Group: fixed
>Status: Closed
>Resolution: Fixed
Priority: 5
Private: No
Submitted By: dfulab (dfulab)
>Assigned to: Maarten Brock (maartenbrock)
Summary: short or truncated to 8-bit

Initial Comment:
Compiler Version: SDCC : mcs51/gbz80/z80/avr/ds390/pic16/pic14/TININative/xa51/ds400/hc08 2.9.1 #5471 (Jul  2 2009) (MINGW32)
SDCCCFLAGS = -mmcs51 --model-small

Short source code that shows the "bug":
-----------------------------------------------------------
unsigned short Addr1(void)
{
  return (0x1234);
 }
 
void func16(unsigned short Dummy)
{
  Dummy;
 }

void Test(void)
{ unsigned short a, b;

  a = 0x01 | Addr1();
  func16(a);
  
  b = Addr1();
  b |= 0x01;
  func16(b);
 }

"a" and "b" logically should have the same results that is passed into their func16( ) call.
In reality, a has been truncated and wrong values passed while b has been treated correctly as full 16-bit "or".

133 _Test:
134 ;	test.c:14: a = 0x01 | Addr1();
135 	lcall	_Addr1
136 	mov	b,dpl
137 	mov	a,#0x01
138 	orl	a,b
139 	mov	dpl,a
140 	mov	dph,b                             ; Problem is right here!!!   b has the lower 8-bit value and it over-written the call parameter to func16(unsigned short)!
141 ;	test.c:15: func16(a);
142 	lcall	_func16
 
Contrast this with the code for b
143 ;	test.c:17: b = Addr1();
144 	lcall	_Addr1
145 	mov	r2,dpl
146 	mov	r3,dph
147 ;	test.c:18: b |= 0x01;
148 	orl	ar2,#0x01
149 ;	test.c:19: func16(b);
150 	mov	dpl,r2
151 	mov	dph,r3                               ; The upper 8-bit has been preserved correctly
152 	ljmp	_func16


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

>Comment By: Maarten Brock (maartenbrock)
Date: 2010-08-23 22:26

Message:
Fixed in SDCC 2.9.7 #5944.

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

Comment By: Philipp Krause (spth)
Date: 2010-05-21 15:14

Message:
Looks like this really is a mcs51 bug. The Z80 port generates correct code
for both a and b.

Philipp


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

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=100599&aid=2944399&group_id=599