新增網頁8

\ Lesson 7 - Code Words and DOS I/O

\ The Forth Course

\ by Richard E. Haskell

\ Dept. of Computer Science and Engineering

\ Oakland University, Rochester, MI 48309

comment:

Lesson 7

CODE WORDS AND DOS I/O

7.1 CODE WORDS ........................................................................ 7-2

7.2 CODE CONDITIONALS ............................................................ 7-5

7.3 LONG MEMORY WORDS ........................................................ 7-6

7.4 DOS WORDS ............................................................................. 7-7

7.5 BASIC FILE I/O ......................................................................... 7-9

7.6 READING NUMBERS AND STRINGS ...................................... 7-14

7.7 WRITING NUMBERS AND STRINGS ....................................... 7-20

7.1 CODE WORDS

Assembly language instructions can be used to define Forth words when the maximum speed of execution is needed or when direct access to the computers hardware is required. This is accomplished by using the CODE word to define a Forth word. The general form of the CODE word is as follows:

CODE <name>

END-CODE

The word CODE takes the place of the colon in a colon definition and builds a header for the name of the Forth word <name>. The word END-CODE takes the place of the semi-colon and ends the code word definition.

The <assembly commands> can be written in either POSTFIX or PREFIX notation. We recommend PREFIX which makes the assembly language look very much like standard 8086/8088 assembly language. The Forth word PREFIX needs to be executed before the CODE word is compiled.

The <return command> can be any of the following:

NEXT JMP >NEXT ( jumps to the inner interpreter >NEXT )

1PUSH PUSH AX

JMP >NEXT ( pushes ax on the stack and jumps to >NEXT )

2PUSH PUSH DX

PUSH AX ( pushes dx and ax on the stack

JMP >NEXT and then jumps to >NEXT )

Debugging CODE words is made easier using the 8088 Tutor monitor that is included with this Forth Course. A complete description of how to use the Tutor monitor in the process of learning 8088/8086 assembly language is given in the book

"IBM PC - 8088 Assembly Language Programming" by Richard E. Haskell.

Instructions for ordering the book are given when you run the program.

As an example of using the Tutor monitor to disassemble and single step through a CODE word consider the following F-PC 3.5 definition of the Forth word CMOVE that moves a string of <count> bytes from the address <source> to the address <dest>. It is assumed that the direction flag in the status register is zero (by having executed CLD) so that the string primitive MOVSB will automatically increment the values of SI and DI.

CODE CMOVE ( source dest count -- )

MOV BX, SI \ save SI (IP)

MOV AX, DS \ copy DS for setting ES

POP CX \ cx = count

POP DI \ di = destination address

POP SI \ si = source address

MOV DX, ES \ save es in dx

MOV ES, AX \ point es to code segment

REPNZ \ repeat until count is zero

MOVSB \ copy DS:SI to ES:DI

MOV SI, BX \ restore si

MOV ES, DX \ restore es

NEXT \ done, jmp to >NEXT

END-CODE

When you FLOAD this lesson the following Forth code will store the hex values 11 22 33 44 55 at the offset address "source.addr" in the code segment. The actual value of the code segment is given by the Forth word ?CS: and will be printed on the screen when you type the word "show.addrs".

A five byte space is reserved at the offset address "dest.addr". The offset addresses for "source.addr", "dest.addr", the top of the stack, and the CFA of CMOVE will also be printed on the screen when you type "show.addrs".

comment;

HEX

CREATE source.addr 11 C, 22 C, 33 C, 44 C, 55 C,

CREATE dest.addr 5 ALLOT

5 CONSTANT #bytes

: test ( -- )

source.addr dest.addr #bytes CMOVE ;

: show.addrs ( -- )

HEX

CR ." code segment = " ?cs: u.

CR ." source addr = " source.addr u.

CR ." dest addr = " dest.addr u.

CR ." top of stack = " SP0 @ U.

CR ." address of CMOVE = " [ ' CMOVE ] LITERAL U.

CR DECIMAL ;

comment:

The words [, ] and LITERAL will be discussed in Lesson 9.

Assume the values printed when you type "show.addrs" are the following:

code segment = 1091

source addr = 74E0

dest addr = 74E8

top of stack = FFE2

address of CMOVE = 477

Your values may be different. If they are, use your corresponding values in the following exercise.

Type debug test.

Type HEX

Type test.

Step through the first three word which will put the following

values on the stack:

74E0 74E8 5

Press F to go to Forth.

Type SYS TUTOR - This will execute the TUTOR program

From the TUTOR memory display

Type >S1091 to display the code segment.

Type /GS1091 to display the data segment = code segment.

Type /GOFEDC to display the stack starting at the top of the

stack (FEE2) minus 6 in the data segment region. The

value 5 (05 00) should be on top of the stack, followed

by the "source addr" 74E0 (E0 74) and the "dest addr"

74E8 (E8 74).

Type /GO74E0 to display the "source addr" in the data segment.

Note that 11 22 33 44 55 is displayed.

Type >O477 to go to the start of the CMOVE code.

Single step the first two instructions by pressing key F1 twice. Note that the value of SI was moved to BX and the value of DS was moved to AX. The next instruction is POP CX which is suppose to pop the value of #bytes (5) from the top of the stack into CX. However, Tutor's stack pointer and stack segment register are not pointing to these values that we saw are really at 1091:FEDC. You could change the values of SS and SP to these values by typing /RSS1091 to make the stack segment the same as the code segment and then typing /RPSFEDC to set the stack pointer equal to the top of stack (FFE2) minus 6. You could then execute POP CX by pressing F1. However, you will then have trouble getting back to F-PC. When you exit Tutor you may go back to DOS or your computer may hang up. An alternative is to load CX with the proper value of 5 by hand. To do this, type /RGC5. Then skip over the instruction POP CX by just pressing the right cursor arrow.

Skip over the next instruction POP DI the same way and load the value of "dest addr" by hand by typing /RID74E8.

Skip over the next instruction POP SI the same way and load the value of "source addr" by hand by typing /RIS74E0.

You can execute the next two instructions by pressing F1 twice.

You should now be at the REP instruction. Press F1. Note that the value 11 was copied from address 74E0 in the data segment to address 74E8 in the extra segment (the same as the data segment) and SI and DI were incremented by 1. This is what the instruction MOVSB does -- and it just did it once. Note also that CX was decrement from 5 to 4.

Press F1 again. Note that the 22 was moved from where SI is pointing in the data segment (74E1) to where DI is pointing in the extra segment (74E9). The value of CX is also decremented to 3.

Press F1 three more times and watch the 33 44 and 55 get moved. Note that when CX gets to zero the REP loop is terminated and the next instruction MOV SI, BX is about to be executed.

Press F1 two more times to execute the next two instructions. The next instruction is a JMP instruction. This instruction jumps to >NEXT.

To exit TUTOR, type /QD. This should take you back to Forth where you had typed "sys tutor". Press <Enter> to get back to the debug mode and then press the space bar until you get back to Forth.

The Forth word CMOVE> ( source dest count -- ) is similar to CMOVE except that the bytes are moved in the opposite direction. That is, the highest address byte is moved first. It is necessary to use this word if you are moving a string up in memory where the destination string may overlap the source string. The use of CMOVE will cause the overlapped portion of the source string to be destroyed before it has a chance to be moved.

7.2 CODE CONDITIONALS

When using the Forth assembler jump instructions are achieved by using the Forth words IF...ELSE...THEN, BEGIN...WHILE...REPEAT, and BEGIN...UNTIL together with the following code conditionals:

Forth Assembled Code

0= JNE/JNZ

0<> JE/JZ

0< JNS

0>= JS

< JNL/JGE

>= JL/JNGE

<= JNLE/JG

> JLE/JNG

U< JNB/JAE/JNC

U>= JB/JNAE/JC

U<= JNBE/JA

U> JBE/JNA

OV JNO

CX<>0 JCX0

As an example, consider the definition of the Forth word ?DUP which duplicates the value on top of the stack only if the value is non-zero.

CODE ?DUP ( n -- n n | 0 )

MOV DI, SP

MOV CX, 0 [DI]

CX<>0

PUSH CX

THEN

END-CODE

Note that when this definition gets assembled into machine code the statement CX<>0 is assembled as JCX0 to the instruction following THEN.

7.3 LONG MEMORY WORDS

The following long memory words are useful for accessing data in segments other than the code segment.

CODE @L ( seg off -- n ) \ Fetch 16-bit value from seg:off

POP BX \ BX = offset address

POP DS \ DS = segment address

MOV AX, 0 [BX] \ AX = data at DS:BX

MOV BX, CS \ Restore DS to CS value

MOV DS, BX

1PUSH \ push value on stack

END-CODE

CODE !L ( n seg off -- ) \ Store 16-bit value at seg:off

POP BX \ BX = offset address

POP DS \ DS = segment address

POP AX \ AX = n

MOV 0 [BX],AX \ Store n at DS:BX

MOV BX, CS \ Restore DS to CS value

MOV DS, BX

END-CODE

The following are other useful long memory words:

C@L ( seg off -- byte ) \ Fetch 8-bit byte from seg:off

C!L ( byte seg off -- ) \ Store 8-bit byte at seg:off

CMOVEL ( sseg soff dseg doff count )

\ move a block of count bytes from sseg:soff to dseg:doff

CMOVEL> ( sseg soff dseg doff count )

\ move a block of count bytes from sseg:soff to dseg:doff

\ moves last byte first to avoid overwriting moved data

7.4 DOS WORDS

F-PC has a large number of Forth words for handling DOS file I/O. These words are defined in the source files HANDLES.SEQ and SEQREAD.SEQ. In this and the next section we will develop a set of file I/O words that you can use and extend to handle a variety of file I/O and other DOS operations. These words can be used in place of, or in conjunction with, the F-PC DOS and file I/O words.

comment;

VARIABLE ITEMS \ used to record stack depth

VARIABLE handl \ file handle

VARIABLE eof \ TRUE if end-of-file was read

CREATE fname 80 ALLOT \ 80 byte buffer containing ASCII filename

: {{ ( -- )

DEPTH ITEMS ! ;

: }} ( -- c )

DEPTH ITEMS @ - ;

comment:

{{ . . . }} Used to keep track of the number of elements

put on the stack. For example,

will leave the following values

on the top of the stack:

5 2 8 3

The 3 on top of the stack is the number of

items entered between {{ and }}.

comment;

: $>asciiz ( addr1 -- addr2 ) \ change counted string to ASCIIZ string

DUP C@ SWAP 1+

TUCK +

0 SWAP C! ;

\ DOS 2.0+ disk I/O functions

comment: ----------------------------------------------------------

2fdos calls the DOS INT 21H function with ax=ah:al,

bx, cx and dx on the stack. It returns ax, dx

and an error flag on the stack. If the error flag

is TRUE, the error code is in ax (3rd element on the

stack). If the error flag is FALSE, then ax and dx

will have values that depend on the function call.

fdos is similar to 2fdos, but does not return an error

flag. It should be used for DOS INT 21H calls that

do not use the carry flag to indicate an error.

*******************************************************************

comment;

PREFIX

HEX

CODE 2fdos ( ax bx cx dx -- ax dx f )

POP DX

POP CX

POP BX

POP AX

INT 21 \ DOS function call

U>=

IF \ if carry = 0

MOV BX, # FALSE \ set error flag to false

ELSE \ else

MOV BX, # TRUE \ set error flag to true

THEN

PUSH AX

PUSH DX

PUSH BX

END-CODE

CODE fdos ( ax bx cx dx -- ax dx )

POP DX

POP CX

POP BX

POP AX

INT 21 \ DOS function call

PUSH AX

PUSH DX

END-CODE

DECIMAL

comment:

7.5 BASIC FILE I/O

The following words can be used for basic file I/O operations such as opening, creating, closing and deleting files, as well as reading and writing bytes from and to the disk file.

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

open.file ( addr -- handle ff | error.code tf )

Opens a file. Returns handle under a false flag

or returns error code under a true flag.

addr points to an asciiz string.

Access code is set to 2 to open for reading and writing.

comment;

HEX

: open.file ( addr -- handle ff | error.code tf )

3D02 \ ah = 3D; al = access.code=2

0 ROT 0 SWAP \ 3D02 0 0 addr

2fdos \ DOS function call

NIP ; \ nip dx

comment: -----------------------------------------------------

close.file Closes file whose handle is on the stack.

Prints error message if unable to close.

comment;

: close.file ( handle -- )

3E00 \ ah = 3E

SWAP 0 0 \ bx = handle

2fdos

NIP \ nip dx

." Close error number " . ABORT

THEN

DROP ;

comment: -----------------------------------------------------

create.file Creates file -- returns values as in open.file

addr points to an asciiz string

attr is the file attribute: 0 - normal file

01H - read only 02H - hidden

04H - system 08H - volume label

10H - subdirectory 20H - archive

comment;

: create.file ( addr attr -- handle ff | error.code tf )

3C00 \ ah = 3C

0 2SWAP SWAP \ 3C00 0 attr addr

2fdos

NIP ; \ nip dx

comment: ------------------------------------------------------

open/create Opens a file if it exists,

otherwise creates a new normal file.

"addr" points to an asciiz string.

Returns a handle for the opened file.

Prints error messages if unable to open.

comment;

: open/create ( addr -- handle )

DUP open.file

DUP 2 =

DROP 0 create.file

IF ." Create error no. " . ABORT

THEN

ELSE

." Open error no. " . DROP ABORT

THEN

ELSE

NIP

THEN ;

: delete.file ( addr -- ax ff | error.code tf )

4100

0 ROT 0 SWAP

2fdos

NIP ;

: erase.file ( $addr -- ) \ erase file with counted string at $addr

$>asciiz

delete.file

CR ." Delete file error no. " .

ELSE

DROP

THEN ;

comment: -----------------------------------------------------

read.file Reads '#bytes' bytes from file with 'handle'

into buffer at 'buff.addr'. Returns #bytes

actually read. If this value is 0 then the

end of file was read. Prints error message

if unsuccessful.

comment;

: read.file ( handle #bytes buff.addr -- #bytes )

>R 3F00 \ handle #bytes 3F00

-ROT R> \ 3F00 handle #bytes addr

2fdos

NIP \ nip dx

." Read error no. " . ABORT

THEN ;

comment: -----------------------------------------------------

write.file Writes '#bytes' bytes from buffer at 'buff.addr'

to file with 'handle'. Prints error message

if unsuccessful.

comment;

: write.file ( handle #bytes buff.addr -- )

>R 4000 \ handle #bytes 4000

-ROT R> \ 4000 handle #bytes addr

2fdos

NIP \ nip dx

." Write error no. " . ABORT

ELSE

DROP

THEN ;

comment: -------------------------------------------------------

mov.ptr Moves the file pointer of the file with 'handle'.

doffset is a double number (32-bit) offset

code is the method code:

0 - move pointer to start of file + offset

1 - increase pointer by offset

2 - move pointer to end of file + offset

comment;

: mov.ptr ( handle doffset code -- dptr )

42 FLIP + \ hndl offL offH 42cd

ROT >R \ hndl offH 42cd

-ROT R> \ 42cd hndl offH offL

2fdos

DROP ." Move pointer error no. " . ABORT

THEN ;

comment: -------------------------------------------------------

rewind.file Moves the pointer of file with 'handle'

to the start of file.

comment;

: rewind.file ( handle -- )

0 0 0 mov.ptr 2DROP ;

comment: -------------------------------------------------------

get.length Returns the 32-bit length of the file with

'handle'.

comment;

: get.length ( handle -- dlength )

0 0 2 mov.ptr ;

comment: -------------------------------------------------------

read.file.L Reads the next "#bytes" bytes from the opened file

with handle "handle" and stores these bytes in

extended memory at seg:offset.

comment;

CODE read.file.L ( handle #bytes seg offset -- ax f )

POP DX

POP DS

POP CX

POP BX

MOV AH, # 3F

INT 21

U>=

MOV BX, # FALSE

ELSE

MOV BX, # TRUE

THEN

MOV CX, CS \ restore DS

MOV DS, CX

PUSH AX

PUSH BX

END-CODE

comment: -------------------------------------------------------

write.file.L Writes "#bytes" bytes from extended memory at

seg:offset to the opened file with handle "handle".

comment;

CODE write.file.L ( handle #bytes seg offset -- ax f )

POP DX

POP DS

POP CX

POP BX

MOV AH, # 40

INT 21

U>=

MOV BX, # FALSE

ELSE

MOV BX, # TRUE

THEN

MOV CX, CS \ restore DS

MOV DS, CX

PUSH AX

PUSH BX

END-CODE

comment: -------------------------------------------------------

findfirst.dir Search the directory for the first match of the

file specified by the asciiz string at "addr".

comment;

CODE findfirst.dir ( addr -- f ) \ search directory for first match

POP DX \ dx = addr of asciiz string

PUSH DS \ save ds

MOV AX, CS

MOV DS, AX \ ds = cs

MOV CX, # 10 \ attr includes subdirectories

MOV AX, # 4E00 \ ah = 4E

INT 21 \ DOS function call

JC 1 $ \ if no error

MOV AX, # FF \ flag = TRUE

JMP 2 $ \ else

1 $: MOV AX, # 0 \ flag = FALSE

2 $: POP DS \ restore ds

PUSH AX \ push flag on stack

END-CODE

comment: -------------------------------------------------------

findnext.dir Search the directory for the next match of the

file specified by the asciiz string at "addr".

comment;

CODE findnext.dir ( -- f ) \ search directory for next match

PUSH DS \ save ds

MOV AX, CS

MOV DS, AX \ ds = cs

MOV AX, # 4F00 \ ah = 4F

INT 21 \ DOS function call

JC 1 $ \ if no error

MOV AX, # FF \ flag = TRUE

JMP 2 $ \ else

1 $: MOV AX, # 0 \ flag = FALSE

2 $: POP DS \ restore ds

PUSH AX \ push flag on stack

END-CODE

comment: -------------------------------------------------------

set-dta.dir Set the disk transfer area address.

comment;

CODE set-dta.dir ( addr -- ) \ set disk transfer area address

POP DX \ dx = dta address

PUSH DS \ save ds

MOV AX, CS

MOV DS, AX \ ds = cs

MOV AX, # 1A00 \ ah = 1A