X-Authentication-Warning:	delorie.com: mail set sender to djgpp-bounces using -f
Date:	Tue, 01 Nov 2005 14:44:40 -0500
From:	Diego Cueva <dcueva AT todo1 DOT com>
Subject:	RE: Why the call instruction points 2 bytes before to the real start
	sub-routine ?
In-reply-to:	<001d01c5df15$d8384200$6401a8c0@SPRITELY>
To:	djgpp AT delorie DOT com
Message-id:	<003501c5df1c$b3f4a440$2432a8c0@dcueva>
MIME-version:	1.0
X-MIMEOLE:	Produced By Microsoft MimeOLE V6.00.2800.1409
X-Mailer:	Microsoft Outlook, Build 10.0.4024
Importance:	Normal
X-Priority:	3 (Normal)
X-MSMail-priority:	Normal
X-MIME-Autoconverted:	from QUOTED-PRINTABLE to 8bit by delorie.com id jA1Jkne9019086
Reply-To:	djgpp AT delorie DOT com
Errors-To:	nobody AT delorie DOT com
X-Mailing-List:	djgpp AT delorie DOT com
X-Unsubscribes-To:	listserv AT delorie DOT com

Thanks for your response.

Now I am using a 32 bit disassembler:

00000000  90                nop
00000001  E80B000000        call 0x11
00000006  66BE5C01          mov si,0x15c
0000000A  E834000000        call 0x43
0000000F  CD20              int 0x20
00000011  6650              push ax
00000013  6653              push bx
00000015  6651              push cx
00000017  6652              push dx
00000019  B406              mov ah,0x6
0000001B  B000              mov al,0x0
0000001D  B500              mov ch,0x0

This dis-assembled code is correct, but when I run the program does not
work fine. 

The program act as a 16 bit code, this mean that the 1st CALL
instruction, jump to 010F. In other words, this compiled program only
execute:
    100. - NOP
    101. - CALL    010F (jump really to 10F not to 0x11)
    10F. - INT 0x20   :(   That's all.

Exists any instruction to say to the processor that the next code is 32
bits?

Thanks a lot.

Att.
Diego Cueva.

PS. I do not know how I can do post via newsgroup.

-----Original Message-----
From: donthave [mailto:donthave AT noreply DOT bit] 
Sent: Tuesday, November 01, 2005 13:56
To: dcueva AT todo1 DOT com
Subject: RE: Why the call instruction points 2 bytes before to the real
start sub-routine ?

Your disassembler appears to be 16-bit.  It needs to be 32-bit. 

ndisasm -b32 ipl.s 

00000000  90                nop 
00000001  E80B000000        call 0x11 
00000006  66BE5C01          mov si,0x15c 
0000000A  E834000000        call 0x43 
0000000F  CD20              int 0x20 

Rod Pemberton 

PS. I attempted to post via newsgroup, but it didn't make it.
Could you post this to the group?  Thanks.


-----Original Message-----
From: dcueva AT todo1 DOT com
Sent: Tuesday, November 01, 2005 12:45
To: 'djgpp AT delorie DOT com'
Subject: Why the call instruction points 2 bytes before to the real
start sub-routine ?

Hi:
Please help me with the next issue:

In the next assembler program:
====================== ipl.s ================
.code32
.text
.global start
start:
        nop
        # Clear screen
        call clrscr
        # Print Hello world
        movw    $m1,    %si
        call    impstr
        # End
        int     $0x20
# Clear screen code using assembler
clrscr:
        pushw   %ax
        pushw   %bx
        pushw   %cx
        pushw   %dx
        movb    $6,     %ah
        movb    $0,     %al
        movb    $0,     %ch
        movb    $0,     %cl
        movb    $24,    %dh
        movb    $79,    %dl
        movb    $7,     %bh
        int     $0x10
        popw    %dx
        popw    %cx
        popw    %bx
        popw    %ax
        ret
# Putchar using assembler 
( ...more code here...) =============================================

I am using the compiler in this way (to generate a ipl.com file or pure
code machine file):

..\..\tools\as -o  ipl.o   ipl.s
..\..\tools\ld -m i386go32  -Ttext 0x0100 -s --oformat binary -o ipl.com
ipl.o

When I compile using ".code16" at the top of the code, the ipl.com file
works very fine, that I expected (clear the screen, print "Hello
world\n" and end).

But if I omit, or put explicit ".code32", at the top of the file, the
ipl.com file has the next problem:

1579:0100 90            NOP
1579:0101 E80B00        CALL    010F
1579:0104 0000          ADD     [BX+SI],AL
1579:0106 66            DB      66
1579:0107 BE5C01        MOV     SI,015C
1579:010A E83400        CALL    0141
1579:010D 0000          ADD     [BX+SI],AL
1579:010F CD20          INT     20

NOTE!!!
CALL    010F: Must point to the "clrscr" subroutine, but this subroutine
start at 0111.
CALL    0141: Must point to the "impstr" subroutine, but this subroutine
start at 0143.

Note 2 bytes of displacement in the CALL instruction.

Obviously the ipl.com file compiled with .code32 does not work fine.
010F points directly to INT 0x20 :(     :'(

Thanks a lot for your help and time.

Att.
Diego Cueva.

PD: I can accept a RTFM as a response, but say me the section or page.



TODO1 SERVICES, INC. -TODO1- es la propietaria de la información contenida
en este mensaje y sus anexos, que puede tener carácter confidencial y ha
sido enviada para uso exclusivo del destinatario intencional. Si usted ha
recibido este mensaje por error, por favor informe de ello al remitente a la
mayor brevedad y elimine este mensaje y sus anexos. Está prohibido su uso,
distribución, divulgación o almacenamiento, salvo autorización escrita de
TODO1. Todo mensaje enviado y/o recibido por TODO1 es monitoreado por
motivos operativos y actuando de acuerdo con prácticas de negocio y de
seguridad legítimas.

TODO1 SERVICES, INC. -TODO1- is the proprietary of the information contained
herein and its attachments, that may have confidential character and has
been sent to the intended recipient only. If you are not the intended
recipient, please contact the sender immediately by replying to this e-mail
and delete the message and its attachments. The use, forward, disclosure,
record or retention of this message for any purpose whatsoever, is strictly
prohibited, unless prior written authorization from TODO1. E-mails to and
from TODO1 are monitored for operational reasons and in accordance with
lawful business and security practices.

- Raw text -