Apple Assembly Line Volume 1 -- Issue 12 September, 1981 In This Issue... ---------------- Field Input Routine for Applesoft . . . . . . . . . . . . 2 CHRGET and CHRGOT in Applesoft . . . . . . . . . . . . . . 8 Leaving the S-C Assembler II . . . . . . . . . . . . . . . 11 A New, Fancier .AS Directive . . . . . . . . . . . . . . . 12 Commented Listing of DOS 3.3 RWTS . . . . . . . . . . . . 16 Quarterly Disk #4 ----------------- The fourth Quarterly Disk is now ready, containing all the source code from issues 10 through 12. The cost is only $15, and it will save you a lot of typing and possible searching for typos. All previous Quarterly Disks are still available, at the same price. Renewing subscriptions ---------------------- The 4-digit number in the upper right corner of your mailing label is the expiration date of your subscription. The first two digits are the year, and the last two digits are the month of the last issue you have paid for. If it says "8109", this is your last issue. Unless, of course, I receive your renewal check for $12. If your label says 8111 or less, now is the time to renew! More about the Firmware Card in Slot 4 -------------------------------------- Michael Sanders' DOS patch for using the Firmware card in slot 4 is really nice. A lot of you have written or called about it, and I use it myself now. In fact, I have changed my HELLO programs to do the patch. All it takes is two POKEs: 10 POKE 42424,192 : POKE 42432,193 I like doing it this way a lot better than INITting a disk with a modified DOS. If you want to test for the presence of a card before patching, you can do it like this: 10 FOR I = 768 TO 817: READ A: POKE I,A: NEXT: CALL 768 20 DATA 173,192,192,162,2,189,0,224,221,44,3,208,16,202,16,245, 162,192,142,184,165,232,142,192,165,173,193,192,96,162,2,189, 0,224,221,47,3,208,242,202,16,245,48,228,32,0,240,78,40,241 30 TEXT: HOME: PRINT CHR$(4)"CATALOG Field Input Routine for Applesoft....................Bob Potts -------------------------------------------------------------- Inputting strings to an Applesoft program is normally a simple task. What could be easier than "INPUT A$"? But, that method will not allow commas or colons. Another easy way is to use GET C$ for each character, and append them to a string using A$=A$+C$. But, by the time you add the testing for each input character to find the end of input and other possible control characters, the routine can be terribly slow. Furthermore, it eats up string space like crazy; eventually Applesoft garbage collection starts, and the program dies for a while. Here is the kind of loop I am talking about: 10 A$="" 20 GET C$ 30 40 A$=A$+C$:PRINT C$; 50 GO TO 20 As the string increases in length, the speed decreases dramatically. In fact, some characters may be lost if you are a fast typist. One way to correct this is to use a machine language routine to input each keystroke, test it, and build a string for the Applesoft program. Such a routine was printed in "Apple Assembly Line" issue #7 (April, 1981), pages 6-8. But that routine used the monitor's RDLINE subroutine to input the string. I needed a routine more adapted to inputting a series of fields, using the screen in a "fill-in-the-blanks" mode. The following program was designed for use in the various branches of the Bank of Louisville. The Apple is used to calculate loans, print the installment notes, and to enter loan applications. A loan application involves filling in the blanks on several screens full of prompts. To use the input routine, you first position the cursor to the start of field using VTAB and HTAB; then set a field length using the SCALE= statement, and a field code using the ROT= statement. The actual call to the input routine is done with "&INPUT" and the name of your string. Here is an example for inputting a 5-character field starting in column 10 of line 7: 10 VTAB 7 : HTAB 10 : SCALE=5 : ROT = 0 : &INPUT A$ The input routine allows skipping from field to field, either forward or backward through a form. Backspace and copy (right arrow) are supported. Filling up a field, or hitting RETURN within a field, finish that field and return the value to Applesoft. An EXIT CODE tells the Applesoft program whether a value was returned in the string or some other exit was chosen. You access the exit code with a PEEK(224). Here are the four exit codes and their meanings: EXIT CODE = 0 Field was filled or RETURN typed. = 1 ESCAPE was typed at beginning of field. = 2 CTRL-F was typed at beginning of field. = 3 Left Arrow (backspace) was typed at beginning of field. If the exit code is zero, then the field data you typed is in your string. Otherwise, the string's value is not changed. Finishing a field by either filling it up or hitting RETURN puts the field data into your string, and I then advance to the next field on the form. I use an exit code of 3 (backspace at beginning of field) to mean that the Applesoft program should go back to the previous field on the current form. How you use the exit codes of 1 and 2 is up to you. You might use an ESCAPE (exit code = 1) to abort the form-filling and return to a main menu. The ESCAPE is now only recognized if you are at the beginning of the field and the field code is non-zero. Of course, you could change that. You might use the control-F to mean you are finished with the current form. How Does It Work? Line 1110 sets the origin to $0300. If you already have something else in page 3, you can change the origin to whatever suits your fancy. Just remember to set the correct values for HIMEM and LOMEM to protect it from Applesoft, and vice versa. Lines 1380-1440 install the ampersand vector. If you BRUN the program, this code is executed. If you BLOAD it, then CALL 768 will execute it. You only have to execute this once in your program. Once done, any occurrence of an ampersand statement in your program will branch to INPUT.FIELD, at line 1460. Lines 1460-1500 check for the keyword "INPUT", and a string variable name. The three routines (and others used in this program) starting with "AS." are in the Applesoft ROMs. AS.SYNCHR compares the current character with what is in the A-register; if different you get SYNTAX ERROR, and if the same the character pointer is advanced. AS.PTRGET scans a variable name and finds its descriptor in memory. AS.CHKSTR makes sure that the variable is a string (if not you get TYPE MISMATCH). At this point the address of the string descriptor is in $83,84. The address in $83,84 points to 3 bytes which tell the length and address of the string's contents. Lines 1520-1690 test the input character and branch accordingly. I use MON.RDKEY to read the character, which means that the data could come from any I/O slot as well as the normal Apple Keyboard. You could add more tests here, or remove some. If it is a printing character, we fall into lines 1730-1810 to store the character in the input buffer and on the screen. If the filed is now full, line 1810 jumps to the routine which passes the data to Applesoft. Note that characters stored in the input buffer have the high-bit equal to zero (Applesoft likes them that way). Characters written on the screen have the high-bit set to one, so that they print in NORMAL video. Lines 1920-1990 handle the backspace character. If you are at the beginning of a field, the routine will return with an exit code of 3. Otherwise, the current character will be replace on the screen with an underline character, and the cursor will be backed up. Lines 2030-2050 handle the right arrow. Normally this just copies over a character on the screen. Characters are picked up from the screen image, and the treated just as though they came from the keyboard. Note that the right arrow will not advance over an underline character. Lines 2090-2140 handle ESCAPE. As I mentioned earlier, ESCAPE is ignored unless it is typed when the cursor is at the beginning of the field, and the field code is non-zero. This is the only use for the field code in the input routine presented here, but you might think of many more uses and make your own modifications. Lines 2180-2190 make Applesoft allocate some space for the string in the normal string data space. Then lines 2200-2270 set up the string variable's descriptor to point to this space. Lines 2280-2310 move the string data from the input buffer up to the new place. This code was copied from the "Fast String Input Routine" in AAL #7. The input routine is presented here in a very simple form; I leave it up to you to modify it to suit your most demanding applications. 1000 *-------------------------------- 1010 * FIELD INPUT SUBROUTINE 1020 * ---------------------- 1030 * BY ROBERT W. POTTS 1040 * BANK OF LOUISVILLE 1050 * P. O. BOX 1101 1060 * LOUISVILLE, KY 40201 1070 * 1080 * MODIFIED BY BOB SANDER-CEDERLOF 1090 * FOR THE "APPLE ASSEMBLY LINE" 1100 *--------------------------------- 1110 .OR $300 1120 *-------------------------------- 1130 MON.CH .EQ $24 MONITOR HORIZONTAL 1140 MON.BASL .EQ $28 1150 SPC.PNTR .EQ $71,72 1160 STR.PNTR .EQ $83,84 1170 *--------------------------------- 1180 CT .EQ $E1 CHARACTER COUNT 1190 FL .EQ $E7 FIELD LENGTH (SET BY "SCALE=FL") 1200 FLDCOD .EQ $F9 FIELD CODE (SET BY "ROT=FC") 1210 EXITCODE .EQ $E0 PEEK (224) TO SEE EXIT CODE 1220 *--------------------------------- 1230 INPUT.BUFFER .EQ $0200 1240 AMPER.VECTOR .EQ $03F5 1250 *--------------------------------- 1260 MON.RDKEY .EQ $FD0C MONITOR CHAR INPUT 1270 MON.COUT .EQ $FDED 1280 MON.BS .EQ $FC10 MONITOR BACKSPACE 1290 *--------------------------------- 1300 AS.CHKSTR .EQ $DD6C 1310 AS.SYNCHR .EQ $DEC0 1320 AS.PTRGET .EQ $DFE3 1330 AS.GETSPA .EQ $E452 1340 AS.MOVSTR .EQ $E5E2 1350 *--------------------------------- 1360 * SET UP AMPERSAND VECTOR 1370 *-------------------------------- 1380 SETUP LDA #$4C JMP OPCODE 1390 STA AMPER.VECTOR 1400 LDA #INPUT.FIELD 1410 STA AMPER.VECTOR+1 1420 LDA /INPUT.FIELD 1430 STA AMPER.VECTOR+2 1440 RTS 1450 *--------------------------------- 1460 INPUT.FIELD 1470 LDA #$84 "INPUT" TOKEN 1480 JSR AS.SYNCHR REQUIRE "INPUT" OR SYNTAX ERROR 1490 JSR AS.PTRGET GET STRING VARIABLE 1500 JSR AS.CHKSTR REQUIRE STRING OR MISMATCH 1510 *--------------------------------- 1520 LDA #0 ZERO OUT CHARACTER COUNT 1530 STA CT 1540 .1 JSR MON.RDKEY GET CHARACTER 1550 .2 AND #$7F APPLESOFT STYLE 1560 CMP #$06 CONTROL-F? 1570 BEQ .3 YES 1580 CMP #$08 BACKSPACE? 1590 BEQ .4 YES 1600 CMP #$0D RETURN? 1610 BEQ .7 YES, END OF FIELD 1620 CMP #$15 RIGHT ARROW? 1630 BEQ .5 YES 1640 CMP #$1B ESCAPE? 1650 BEQ .6 YES 1660 CMP #$20 SOME OTHER CONTROL CHARACTER? 1670 BCC .1 YES, IGNORE IT 1680 CMP #$5B ACCEPTABLE PRINTING CHARACTER? 1690 BCS .1 NO, IGNORE IT 1700 *-------------------------------- 1710 * GOT PRINTING CHARACTER - STORE IT 1720 *-------------------------------- 1730 LDY CT CHARACTER COUNTER 1740 STA INPUT.BUFFER,Y STORE IN STRING 1750 ORA #$80 TURN ON HIGH BIT 1760 JSR MON.COUT PRINT CHARACTER 1770 INC CT INCREMENT CHARACTER COUNT 1780 LDA CT 1790 CMP FL IS FIELD FILLED UP? 1800 BNE .1 NO, GET ANOTHER CHARACTER 1810 BEQ .7 ...ALWAYS 1820 *--------------------------------- 1830 * HANDLE CONTROL-F 1840 *--------------------------------- 1850 .3 LDA CT ON FIRST CHARACTER? 1860 BNE .1 NO, GET ANOTHER CHARACTER 1870 LDA #2 EXIT CODE = 2 1880 BNE .8 ...ALWAYS 1890 *--------------------------------- 1900 * HANDLE BACKSPACE 1910 *--------------------------------- 1920 .4 LDA #3 EXIT CODE = 3 IF IN 1ST CHAR 1930 DEC CT DECREMENT CHARACTER COUNTER 1940 BMI .8 ON FIRST POSITION 1950 JSR MON.BS BACKSPACE 1960 LDA #$DF UNDERLINE 1970 JSR MON.COUT PRINT IT 1980 JSR MON.BS BACKSPACE AGAIN 1990 JMP .1 DO AGAIN 2000 *--------------------------------- 2010 * HANDLE RIGHT ARROW 2020 *--------------------------------- 2030 .5 LDY MON.CH YES, GET NEXT CHARACTER FROM SCREEN 2040 LDA (MON.BASL),Y 2050 JMP .2 2060 *--------------------------------- 2070 * HANDLE ESCAPE 2080 *--------------------------------- 2090 .6 LDA FLDCOD FIELD CODE = 0? 2100 BEQ .1 YES, GET ANOTHER CHARACTER 2110 LDA CT 2120 BNE .1 NO, GET ANOTHER CHARACTER 2130 LDA #1 EXIT CODE = 1 2140 BNE .8 ...ALWAYS 2150 *-------------------------------- 2160 * STORE THE INPUT DATA IN THE STRING 2170 *-------------------------------- 2180 .7 LDA CT STRING LENGTH 2190 JSR AS.GETSPA GET SPACE IN STRING AREA 2200 LDY #0 MOVE DATA INTO VARIABLE 2210 STA (STR.PNTR),Y LENGTH 2220 LDA SPC.PNTR 2230 INY 2240 STA (STR.PNTR),Y LO-BYTE OF ADDRESS 2250 LDA SPC.PNTR+1 2260 INY 2270 STA (STR.PNTR),Y HI-BYTE OF ADDRESS 2280 LDX #INPUT.BUFFER 2290 LDY /INPUT.BUFFER 2300 LDA CT LENGTH 2310 JSR AS.MOVSTR 2320 LDA #0 EXIT CODE = 0 2330 .8 STA EXITCODE 2340 RTS Here is a brief sample showing how you might use the input routine to fill in five fields: 10 PRINT CHR$(4)"BRUN B.INPUT ROUTINE" 20 DIM V(5),H(5),L(5),T$(5),A$(5) 30 FOR I = 1 TO 5: READ V(I),H(I),L(I),T$(I): NEXT 40 DATA 5,7,30,NAME 50 DATA 7,7,3,AGE 60 DATA 7,27,3,WEIGHT 70 DATA 9,7,12,STATE 80 DATA 9,27,5,ZIP 90 TEXT: HOME: VTAB 23: INVERSE: PRINT " TYPE CTRL-F WHEN FORM FINISHED ": NORMAL 100 FOR I = 1 TO 5: VTAB V(I): HTAB H(I)-LEN(T$(I))-1: PRINT T$(I)" ";: FOR J=1 TO L(I): PRINT CHR$(95);: NEXT: NEXT 110 I = 1 120 VTAB V(I): HTAB H(I): SCAL=L(I): ROT=0 130 & INPUT A$(I): XC=PEEK(224) 140 ON XC+1 GOTO 200,300,400,500 200 I=I+1: IF I>5 THEN 110 210 GOTO 120 300 END: REM ESCAPE 400 REM CONTROL-F 410 HOME: FOR I=1 TO 5: PRINT A$(I): NEXT: END 500 REM BACKSPACE 510 I=I-1: IF I=0 THEN I=5 520 GOTO 120 CHRGET and CHRGOT in Applesoft ------------------------------ On pages 13 and 14 of the September 1981 Kilobaud Microcomputing (Robert Baker's Pet-Pourri column) there is a good description of the CHRGET/CHRGOT duo. These two subroutines (really two entry points into one routine) seem to be common to the Microsoft Basics, at least the 6502 versions. What are they? When Applesoft initializes itself one of the tasks is to copy a short subroutine into page zero, from $00B1 through $00C8. There is no difference between the PET and the Apple versions, except that the PET version is copied into $0070-0087. Here is the code: 1000 *--------------------------------- 1010 * APPLESOFT CHRGET/CHRGOT SUBROUTINES 1020 *--------------------------------- 1030 .OR $00B1 1040 *--------------------------------- 1050 TXTPTR .EQ $B8 INSIDE 'LDA' INSTRUCTION 1060 *--------------------------------- 1070 CHRGET INC TXTPTR INCREMENT ADDRESS OF NEXT CHARACTER 1080 BNE CHRGOT 1090 INC TXTPTR+1 1095 *--------------------------------- 1100 CHRGOT LDA $8888 PICK UP THE NEXT CHARACTER 1110 CMP #$3A TEST IF COLON 1120 BCS .1 YES, Z AND C SET, RETURN 1130 CMP #$20 TEST IF BLANK 1140 BEQ CHRGET YES, IGNORE IT 1150 SEC DO DIGIT TEST 1160 SBC #$30 1170 SEC SET Z IF VALUE WAS $00 (EOL TOKEN) 1180 SBC #$D0 AND CLEAR CARRY IF DIGIT ($30-39) 1190 .1 RTS Almost every time Applesoft wants to look at a character from your program or even from the input buffer, it does so by calling this subroutine. The CHRGET entry increments the address used to pick up the next character, and then falls into CHRGOT. In either case, the character is picked up and several tests are performed. Blanks are passed over, ignored. Colon (end of statement) and $00 (end of line) set the Z status bit. Digits clear CARRY, non-digits set CARRY. The calling program can use these status bits. For example: JSR CHRGET BEQ END BRANCH IF COLON OR END-OF-LINE BCC DIGIT BRANCH IF CHAR IS DIGIT (0-9) The article in Kilobaud suggests patching this routine at $00BA to jump to your own code. Your program can trap certain characters for special functions, in much the same way as the "&" is now handled by Applesoft. You just have to be sure that you execute the instructions your JMP overlayed before returning to the remainder of CHRGET. It appears that many of the enhancement packages available for PET Basic use this scheme. Why use this patching scheme instead of the "&" for special functions? Because your special functions can be made to appear an integral part of the language, without the telltale ampersand. Because even special codes inside expressions or other statements can be trapped. Because you want to encode or otherwise obfuscate your program for security. Because you just want to be different. Of course, the disadvantage is that the entire operation of Applesoft is slowed down by the amount of time your extra testing takes, since every character retrieved by the interpreter will go through your routine as well as the standard CHRGET. 1000 *--------------------------------- 1010 * SAMPLE APPLESOFT FILTER PROGRAM 1020 *--------------------------------- 1030 .OR $BA 1040 JMP FILTER 1050 *--------------------------------- 1060 .OR $300 1070 FILTER CMP #'# CHECK FOR "#" CHARACTER 1080 BNE .1 NO, PASS UNMOLESTED 1090 JSR WHATEVER.YOU.WANT 1100 JMP $B1 1110 .1 CMP #$3A CHECK FOR COLON 1120 BCS .2 YES, RETURN JUST CHRGET WOULD 1130 JMP $BE NO, RECONNECT WITH CHRGET 1140 .2 RTS 1150 *--------------------------------- 1160 WHATEVER.YOU.WANT 1170 JSR $FBE2 RING BELL 1180 RTS Here is a sample patch program, just show how it is done. Any time the patch discovers a "#" character, it will ring the Apple's bell. The sample Applesoft lines show what I mean. If you want to try out the patch, assemble it and then call Applesoft. Then get to the monitor and patch CHRGET like this: ]CALL -151 *BA:4C 00 03 *3D0G Then enter some Applesoft lines with embedded "#" characters, and RUN. ]LIST 10 PRINT #1;#2;#3;#4;#5;#6;#7 20 A#=3: B#=4: PRINT A+B If you think of some really practical ways to use patches like this, let me know about them. Leaving the S-C Assembler II ---------------------------- How do you get out of the assembler? I suppose I could have made a QUIT or EXIT command, but I didn't. If you want to go to Applesoft or Integer BASIC, type FP or INT. You will then be instantly in the version of Basic you wanted. However, you will still be hooked into the Assembler's output subroutine. If you load a small program and LIST it, you will find that tapping the space bar will stop the listing and restart it, just as inside the assembler. Notice I said a "small" program; a large program might over-write part of the assembler, causing the computer to hang up. What you must do is type FP or INT, and then PR#0. The PR#0 unhooks the assembler output routine, and you are free. Now, if you are sure that you have not over-written the assembler with your Applesoft or Integer BASIC program, and you want to return to the assembler, you can do so by typing CALL 4096. I use this for going back and forth rapidly when I am testing &-routines and the like. What if you want to leave the assembler to go to the monitor? First of all, remember that you can use all of the monitor commands without ever leaving the assembler, by typing a dollar sign and then the monitor command. But if you really want out, how do you get there? If you have an old monitor ROM (not AUTOSTART), hitting RESET will get you to the monitor. With the Autostart ROM, you can type $FF59G or $FF69G. The first will unhook DOS, while the second will leave DOS hooked in. (The second is the same as the Basic command CALL-151.) Still another way is to patch the Autostart ROM RESET vector at $3F2 (type "$3F2:69 FF 5A"), so that RESET enters the monitor. And how do you get back to the assembler from the monitor, without disturbing or losing your source code? Simply type "1003G" and you will be there. If you type "1000G" you will also get to the assembler, but all your source code will be gone, just as though you had typed the "NEW" command. A New, Fancier .AS Directive ---------------------------- Many times I write text printing loops that depend on the sign bit of each byte to indicate the end of text. I might set up the text this way: .AS /THIS IS THE TEXT I WANT TO PRIN .AS -/T/ This assembles with the sign bits off (0) on all the characters of the text except the last one. I can terminate my printing loop by testing that bit. A little later, I will show you an example of just such a loop. But when there are many messages, I get tired of using separate lines for the last character of each message! Why not have an assembler directive which automatically sets the sign bit of the last character the opposite of the sign bits of the rest of the message? Since Version 4.0 of the S-C Assembler II has a .US directive for me, the user, to program.... The only problem is that how to program for the .US directive has never been revealed. Until now. The following little program will implement just the directive I want, and install it as the .US directive. It uses five programs inside the assembler (see lines 1100-1140). The code is patterned directly after the code for the .AS directive, which starts at $203C in most copies of Version 4.0. NOTE: You should check your assembler to make sure that the four bytes starting at $203C are "A0 00 84 04"; if they are, you can use the same addresses for the five routines as I have shown here. (If not, send me your original Version 4.0 disk for a free update. Be sure to adequately protect the disk for shipping, because your new copy will come back on the same disk.) Line 1000 sets the origin of the code to $0F00. You could use some other origin, like $0300, if you wish. Just be sure it is an area of memory that you will not be using for some other purpose wile you are assembling. Line 1010 directs the object code to a BRUNnable file named B.US.DIRECTIVE. The code from 1160 to 1210 is executed when you BRUN B.US.DIRECTIVE. It stores the address of DIR.US in the .US vector at the beginning of the assembler. You can read a little about this on page 15 of the Version 4.0 update manual. Lines 1030-1050 define a few variables. WBUF is the line buffer the assembler uses, starting at $0200. The assembler unpacks a line from the source code into this buffer, and then proceeds to analyze it. DLIM and HIBIT are temporary locations in page zero where I will save the delimiter character and the high-bit setting. The meat of the directive is in lines 1230-1510. If you disassemble the code at $203C in the S-C Assembler II, you will see a marked similarity here. You might also try disassembling the code for the GNNB and GNC subroutines. GNC retrieves the next character from WBUF and increments the pointer. The character is tested. Carry status is set if the end-of-line token was picked up. Equal status is set if a blank or end-of-line token was picked up. GNNB calls on GNC until a non-blank character is found. GNC returns with the character in the A-register, and the pointer to the next character in the Y-register. Lines 1240-1310 scan from the end of the opcode field to try to find the delimiter. If no non-blank character is found after the opcode field, you will get the "BAD ADDRESS ERROR". If a minus sign is found, $80 is stored in HIBIT instead of $00. This value will be merged with every character between the delimiters, to set or clear the high-bit of each byte. When the delimiter is found, it is stored in DLIM. Lines 1320-1350 check to make sure that there are some characters after the delimiter before the next occurrence of the delimiter. For example, if you write ".US //", I want to assemble no bytes and go on. If I find the end-of-line token, you will get the error message. Lines 1360-1430 are a loop to output the bytes one by one. I have to look ahead to see if the next character is the delimiter again. If not, then I will output the current character (by now accessed with "LDA WBUF-2,Y", because Y has been advanced). If the next one is the delimiter, then the current one is the last character of the string; I will have to go to ".3", to handle the last character. Lines 1450-1490 handle the last character of the string between the delimiters. The high-bit is first set just like all the rest of the bytes at line 1460, and then reversed with the EOR #$80 at line 1470. There is no end to the detail we could get into by describing how EMIT, CMNT, and ERBA work. I will leave them for you to puzzle over at your leisure. (Can't give away the whole plot in chapter 1!) 1000 .OR $F00 1010 .TF B.US.DIRECTIVE 1020 *--------------------------------- 1030 WBUF .EQ $0200 1040 DLIM .EQ $DA 1050 HIBIT .EQ $04 1060 *--------------------------------- 1070 * THE FOLLOWING VALUES ARE FOR VERSION 4.0 1080 * OF S-C ASSEMBLER II (DISK) 1090 *--------------------------------- 1100 GNNB .EQ $1283 GET NEXT NON-BLANK CHAR 1110 GNC .EQ $128B GET NEXT CHAR 1120 CMNT .EQ $188E FINISH THE LINE 1130 ERBA .EQ $1932 ERROR: BAD ADDRESS 1140 EMIT .EQ $19FA EMIT A BYTE OF OBJECT CODE 1150 *--------------------------------- 1160 ACTIVATE.US 1170 LDA #DIR.US STORE ADDRESS IN .US VECTOR 1180 STA $100D INSIDE S-C ASSEMBLER II VER 1190 LDA /DIR.US DISK VERSION 4.0 1200 STA $100E 1210 RTS 1220 *--------------------------------- 1230 DIR.US 1240 LDY #0 START WITH HI-BIT EQUAL TO ZERO 1250 .1 STY HIBIT SET HI-BIT ZERO OR ONE 1260 JSR GNNB GET NEXT NON-BLANK AFTER OPCODE 1270 BCS ERBA2 END OF LINE IS BAD NEWS 1280 LDY #$80 IN CASE WE NEED HI-BIT OF ONE 1290 CMP #$2D CHECK FOR MINUS SIGN 1300 BEQ .1 YES, WE NEED HI-BIT OF ONE 1310 STA DLIM NOT MINUS, MUST BE DELIMITER 1320 JSR GNC GET NEXT CHARACTER 1330 BCS ERBA2 END OF LINE IS BAD NEWS 1340 CMP DLIM SEE IF DELIMITER ALREADY 1350 BEQ .4 YES, NO STRING IN BETWEEN 1360 .2 JSR GNC GET NEXT CHARACTER 1370 BCS ERBA2 END OF LINE IS BAD NEWS 1380 CMP DLIM SEE IF DELIMITER YET 1390 BEQ .3 YES, FINISH UP AND RETURN 1400 LDA WBUF-2,Y NO, GET PREVIOUS CHAR 1410 ORA HIBIT MERGE WITH SELECTED HI-BIT 1420 JSR EMIT EMIT THE OBJECT CODE BYTE 1430 JMP .2 GO FOR ANOTHER ONE 1440 *--------------------------------- 1450 .3 LDA WBUF-2,Y GET PREVIOUS CHAR 1460 ORA HIBIT MERGE WITH SELECTED HI-BIT 1470 EOR #$80 TOGGLE HI-BIT SINCE LAST CHAR 1480 JSR EMIT EMIT THE OBJECT CODE BYTE 1490 .4 JMP CMNT FINISH PROCESSING THE LINE 1500 *--------------------------------- 1510 ERBA2 JMP ERBA BAD ADDRESS ERROR The following program shows how I might use the new .US directive I have just built. It prints the line of text from line 1230 ten times on the screen. The .US directive assures that I can tell when I am at the end of the text string by looking at the sign bit. That is just what the BMI opcode at line 1110 is doing. Lines 1070, 1080, 1190, and 1200 are the looping code to make ten copies of the line. Lines 1090-1150 print the message except for the last character; lines 1170-1180 print that last character and a carriage return. 1000 *--------------------------------- 1010 * DEMONSTRATE USE OF .US DIRECTIVE 1020 *--------------------------------- 1030 MON.COUT .EQ $FDED 1040 MON.CROUT .EQ $FD8E 1050 *--------------------------------- 1060 DEMO.US 1070 LDA #10 DO 10 LINES 1080 STA LINE.COUNT 1090 .3 LDY #0 1100 .1 LDA TEXT,Y GET CHAR FROM TEXT STRING 1110 BMI .2 1120 ORA #$80 MAKE NORMAL VIDEO 1130 JSR MON.COUT 1140 INY NEXT CHARACTER 1150 BNE .1 ...ALWAYS 1160 *--------------------------------- 1170 .2 JSR MON.COUT 1180 JSR MON.CROUT 1190 DEC LINE.COUNT 1200 BNE .3 1210 RTS 1220 *--------------------------------- 1230 TEXT .US /THIS IS MY MESSAGE/ 1240 LINE.COUNT .BS 1 1250 *--------------------------------- Commented Listing of DOS 3.3 RWTS --------------------------------- Last March I started out this series of DOS listings with the RWTS portion of DOS 3.2.1. Since then I have printed all of DOS 3.2.1 and DOS 3.3 from $B800 thru $BFFF, except for DOS 3.3 RWTS. Somehow it almost was overlooked, but here it is now. There are minor differences between the two versions of RWTS, which you can find by comparing the listing from the March 1981 issue of AAL and this one. The differences start at line 1810. I suppose the changes are meant to be improvements, but most of them seem to make very little difference. One critical major difference: DOS 3.2.1 and previous versions use sector numbers which are actually written in the headers. DOS 3.3 uses two different sets of sector numbers: physical and logical. The physical sector numbers are recorded in the sector header blocks; logical sector numbers are used in RWTS calls and File Manager calls. The translation is performed using the table at line 4280, which I have called the PHYSICAL.SECTOR.VECTOR. This table is accessed at line 3310: the logical sector number is in the Y-register, and indexes into the physical sector vector to pick up a physical sector number. 1000 *--------------------------------- 1010 * DOS 3.3 DISASSEMBLY $BD00-BEAE 1020 * BOB SANDER-CEDERLOF 3-3-81 1030 *--------------------------------- 1040 CURRENT.TRACK .EQ $478 1050 DRIVE.1.TRACK .EQ $478 THRU 47F (INDEX BY SLOT) 1060 DRIVE.2.TRACK .EQ $4F8 THRU 4FF (INDEX BY SLOT) 1070 SEARCH.COUNT .EQ $4F8 1080 RETRY.COUNT .EQ $578 1090 SLOT .EQ $5F8 1100 SEEK.COUNT .EQ $6F8 1110 *--------------------------------- 1120 PHASE.OFF .EQ $C080 1130 PHASE.ON .EQ $C081 1140 MOTOR.OFF .EQ $C088 1150 MOTOR.ON .EQ $C089 1160 ENABLE.DRIVE.1 .EQ $C08A 1170 ENABLE.DRIVE.2 .EQ $C08B 1180 Q6L .EQ $C08C 1190 Q6H .EQ $C08D 1200 Q7L .EQ $C08E 1210 Q7H .EQ $C08F 1220 *--------------------------------- 1230 SECTOR .EQ $2D 1240 TRACK .EQ $2A 1250 VOLUME .EQ $2F 1260 DRIVE.NO .EQ $35 1270 DCT.PNTR .EQ $3C,3D 1280 BUF.PNTR .EQ $3E,3F 1290 MOTOR.TIME .EQ $46,47 1300 IOB.PNTR .EQ $48,49 1310 *--------------------------------- 1320 PRE.NYBBLE .EQ $B800 1330 WRITE.SECTOR .EQ $B82A 1340 READ.SECTOR .EQ $B8DC 1350 READ.ADDRESS .EQ $B944 1360 POST.NYBBLE .EQ $B8C2 1370 SEEK.TRACK.ABSOLUTE .EQ $B9A0 1380 DELAY.LOOP .EQ $BA00 1390 *--------------------------------- 1400 ERR.WRITE.PROTECT .EQ $10 1410 ERR.WRONG.VOLUME .EQ $20 1420 ERR.BAD.DRIVE .EQ $40 1430 *--------------------------------- 1440 .OR $BD00 1450 .TA $800 1460 *--------------------------------- 1470 RWTS STY IOB.PNTR SAVE ADDRESS OF IOB 1480 STA IOB.PNTR+1 1490 LDY #2 1500 STY SEEK.COUNT UP TO 2 RE-CALIBRATIONS 1510 LDY #4 1520 STY SEARCH.COUNT 1530 LDY #1 POINT AT SLOT# IN IOB 1540 LDA (IOB.PNTR),Y SLOT# FOR THIS OPERATION 1550 TAX 1560 LDY #15 POINT AT PREVIOUS SLOT# 1570 CMP (IOB.PNTR),Y SAME SLOT? 1580 BEQ .3 YES 1590 TXA SAVE NEW SLOT ON STACK 1600 PHA 1610 LDA (IOB.PNTR),Y GET OLD SLOT# 1620 TAX 1630 PLA STORE NEW SLOT # 1640 PHA INTO OLD SLOT# SPOT 1650 STA (IOB.PNTR),Y 1660 *--------------------------------- 1670 * SEE IF OLD MOTOR STILL SPINNING 1680 *--------------------------------- 1690 LDA Q7L,X GO INTO READ MODE 1700 .1 LDY #8 IF DATA DOES NOT CHANGE 1710 LDA Q6L,X FOR 96 MICROSECONDS, 1720 .2 CMP Q6L,X THEN THE DRIVE IS STOPPED 1730 BNE .1 WOOPS! IT CHANGED! 1740 DEY TIME UP YET? 1750 BNE .2 NO, KEEP CHECKING 1760 PLA GET NEW SLOT # AGAIN 1770 TAX 1780 *--------------------------------- 1790 .3 LDA Q7L,X SET UP TO READ 1800 LDA Q6L,X 1810 LDY #8 1820 .31 LDA Q6L,X GET CURRENT DATA 1830 PHA 7 CYCLE DELAY 1840 PLA 1850 PHA 7 CYCLE DELAY 1860 PLA 1870 STX SLOT 1880 CMP Q6L,X SEE IF DATA CHANGED 1890 BNE .32 YES, IT CHANGED 1900 DEY 1910 BNE .31 KEEP WAITING 1920 .32 PHP SAVE ANSWER ON STACK 1930 LDA MOTOR.ON,X TURN ON MOTOR 1940 LDY #6 COPY POINTERS INTO PAGE ZERO 1950 .4 LDA (IOB.PNTR),Y 1960 STA DCT.PNTR-6,Y 1970 INY DCT.PNTR .EQ $3C,3D 1980 CPY #10 BUF.PNTR .EQ $3E,3F 1990 BNE .4 2000 LDY #3 GET MOTOR ON TIME FROM DCT 2010 LDA (DCT.PNTR),Y 2020 STA MOTOR.TIME+1 HIGH BYTE ONLY 2030 LDY #2 GET DRIVE # 2040 LDA (IOB.PNTR),Y 2050 LDY #16 SEE IF SAME AS OLD DRIVE# 2060 CMP (IOB.PNTR),Y 2070 BEQ .5 YES 2080 STA (IOB.PNTR),Y UPDATE OLD DRIVE # 2090 PLP SET Z STATUS 2100 LDY #0 TO FLAG MOTOR OFF 2110 PHP 2120 .5 ROR CHECK LSB OF DRIVE # 2130 BCC .6 DRIVE 2 2140 LDA ENABLE.DRIVE.1,X 2150 BCS .7 ...ALWAYS 2160 .6 LDA ENABLE.DRIVE.2,X 2170 .7 ROR DRIVE.NO SET SIGN BIT IF DRIVE 1 2180 PLP WAS MOTOR PROBABLY OFF? 2190 PHP 2200 BNE .9 NO, DEFINITELY ON 2210 *--------------------------------- 2220 * DELAY FROM 150 TO 180 MILLISECONDS, 2230 * DEPENDING ON WHAT GARBAGE IS IN A-REG 2240 *--------------------------------- 2250 LDY #7 YES, WAIT A WHILE 2260 .8 JSR DELAY.LOOP 2270 DEY BUT IT WORKS ANYWAY.... 2280 BNE .8 2290 LDX SLOT RESTORE SLOT# 2300 *--------------------------------- 2310 .9 LDY #4 GET TRACK # 2320 LDA (IOB.PNTR),Y 2330 JSR SEEK.TRACK 2340 PLP WAS MOTOR DEFINITELY ON? 2350 BNE PROCESS.COMMAND YES, MOTOR ON 2360 LDY MOTOR.TIME+1 SEE IF NEED TO WAIT 2370 BPL PROCESS.COMMAND NO 2380 *--------------------------------- 2390 * MOTOR WAS OFF, SO WAIT REST OF MOTOR ON TIME 2400 * FOR APPLE DISK II, MOTOR ON TIME IS 1 SECOND. 2410 * PART OF THIS TIME IS COUNTED DOWN WHILE SEEKING 2420 * FOR THE TRACK. 2430 *--------------------------------- 2440 .10 LDY #18 ABOUT 100 MICROSECONDS PER TRIP 2450 .11 DEY 2460 BNE .11 2470 INC MOTOR.TIME 2480 BNE .10 2490 INC MOTOR.TIME+1 2500 BNE .10 2510 *--------------------------------- 2520 * MOTOR ON AND UP TO SPEED, SO LET'S 2530 * FIND OUT WHAT THE COMMAND IS AND DO IT! 2540 *--------------------------------- 2550 PROCESS.COMMAND 2560 LDY #12 GET COMMAND 2570 LDA (IOB.PNTR),Y 2580 BEQ .8 NULL COMMAND, LET'S LEAVE 2590 CMP #4 FORMAT? 2600 BEQ .9 YES 2610 ROR SET CARRY=1 IF READ, =0 IF WRITE 2620 PHP SAVE ON STACK 2630 BCS .1 READ 2640 JSR PRE.NYBBLE WRITE 2650 .1 LDY #48 UP TO 48 RETRIES 2660 STY RETRY.COUNT 2670 .2 LDX SLOT GET SLOT NUMBER AGAIN 2680 JSR READ.ADDRESS 2690 BCC .5 GOOD ADDRESS READ 2700 .21 DEC RETRY.COUNT 2710 BPL .2 KEEP TRYING 2720 .3 LDA CURRENT.TRACK GET TRACK WE WANTED 2730 PHA SAVE IT 2740 LDA #96 PRETEND TO BE ON TRACK 96 2750 JSR SETUP.TRACK 2760 DEC SEEK.COUNT 2770 BEQ .6 NO MORE RE-CALIBRATES 2780 LDA #4 2790 STA SEARCH.COUNT 2800 LDA #0 LOOK FOR TRACK 0 2810 JSR SEEK.TRACK 2820 PLA GET TRACK WE REALLY WANT 2830 .4 JSR SEEK.TRACK 2840 JMP .1 2850 *--------------------------------- 2860 .5 LDY $2E TRACK# IN ADDRESS HEADER 2870 CPY CURRENT.TRACK 2880 BEQ .10 FOUND RIGHT TRACK 2890 LDA CURRENT.TRACK 2900 PHA SAVE TRACK WE REALLY WANT 2910 TYA SET UP TRACK WE ACTUALLY FOUNG 2920 JSR SETUP.TRACK 2930 PLA TRACK WE WANT 2940 DEC SEARCH.COUNT 2950 BNE .4 TRY AGAIN 2960 BEQ .3 TRY TO RE-CALIBRATE AGAIN 2970 *--------------------------------- 2980 * DRIVE ERROR, CANNOT FIND TRACK 2990 *--------------------------------- 3000 .6 PLA REMOVE CURRENT.TRACK 3010 LDA #ERR.BAD.DRIVE 3020 .7 PLP 3030 JMP ERROR.HANDLER 3040 *--------------------------------- 3050 * NULL COMMAND, ON THE WAY OUT.... 3060 *--------------------------------- 3070 .8 BEQ RWTS.EXIT 3080 *--------------------------------- 3090 * FORMAT COMMAND 3100 *--------------------------------- 3110 .9 JMP FORMAT 3120 *--------------------------------- 3130 * READ OR WRITE COMMAND 3140 *--------------------------------- 3150 .10 LDY #3 GET VOLUME# WANTED 3160 LDA (IOB.PNTR),Y 3170 PHA SAVE DESIRED VOLUME# ON STACK 3180 LDA VOLUME 3190 LDY #14 STORE ACTUAL VOLUME NUMBER FOUND 3200 STA (IOB.PNTR),Y 3210 PLA GET DESIRED VOLUME# AGAIN 3220 BEQ .11 IF =0, DON'T CARE 3230 CMP VOLUME SEE IF RIGHT VOLUME 3240 BEQ .11 YES 3250 LDA #ERR.WRONG.VOLUME 3260 BNE .7 UH OH! 3270 *--------------------------------- 3280 .11 LDY #5 GET SECTOR# WANTED 3290 LDA (IOB.PNTR),Y (LOGICAL SECTOR NUMBER) 3300 TAY INDEX INTO PHYSICAL SECTOR VECTOR 3310 LDA PHYSICAL.SECTOR.VECTOR,Y 3320 CMP SECTOR 3330 BNE .21 NOT THE RIGHT SECTOR 3340 PLP GET COMMAND FLAG AGAIN 3350 BCC WRITE 3360 JSR READ.SECTOR 3370 PHP SAVE RESULT; IF BAD, WILL BE COMMAND 3380 BCS .21 BAD READ 3390 PLP THROW AWAY 3400 LDX #0 3410 STX $26 3420 JSR POST.NYBBLE 3430 LDX SLOT 3440 RWTS.EXIT 3450 CLC 3460 .HS 24 "BIT" TO SKIP NEXT INSTRUCTION 3470 *--------------------------------- 3480 ERROR.HANDLER 3490 SEC INDICATE AN ERROR 3500 LDY #13 STORE ERROR CODE 3510 STA (IOB.PNTR),Y 3520 LDA MOTOR.OFF,X 3530 RTS 3540 *--------------------------------- 3550 WRITE JSR WRITE.SECTOR 3560 BCC RWTS.EXIT 3570 LDA #ERR.WRITE.PROTECT 3580 BCS ERROR.HANDLER ...ALWAYS 3590 *--------------------------------- 3600 * SEEK TRACK SUBROUTINE 3610 * (A) = TRACK# TO SEEK 3620 * (DRIVE.NO) IS NEGATIVE IF DRIVE 1 3630 * AND POSITIVE IF DRIVE 2 3640 *--------------------------------- 3650 SEEK.TRACK 3660 PHA SAVE TRACK# 3670 LDY #1 CHECK DEVICE CHARACTERISTICS TABLE 3680 LDA (DCT.PNTR),Y FOR TYPE OF DISK 3690 ROR SET CARRY IF TWO PHASES PER TRACK 3700 PLA GET TRACK# AGAIN 3710 BCC .1 ONE PHASE PER TRACK 3720 ASL TWO PHASES PER TRACK, SO DOUBLE IT 3730 JSR .1 FIND THE TRACK 3740 LSR CURRENT.TRACK DIVIDE IT BACK DOWN 3750 RTS 3760 *--------------------------------- 3770 .1 STA TRACK 3780 JSR GET.SLOT.IN.Y 3790 LDA DRIVE.1.TRACK,Y 3800 BIT DRIVE.NO WHICH DRIVE? 3810 BMI .2 DRIVE 1 3820 LDA DRIVE.2.TRACK,Y 3830 .2 STA CURRENT.TRACK WHERE WE ARE RIGHT NOW 3840 LDA TRACK WHERE WE WANT TO BE 3850 BIT DRIVE.NO WHICH DRIVE? 3860 BMI .3 DRIVE 1 3870 STA DRIVE.2.TRACK,Y DRIVE 2 3880 BPL .4 ...ALWAYS 3890 .3 STA DRIVE.1.TRACK,Y 3900 .4 JMP SEEK.TRACK.ABSOLUTE 3910 *--------------------------------- 3920 * CONVERT SLOT*16 TO SLOT IN Y-REG 3930 *--------------------------------- 3940 GET.SLOT.IN.Y 3950 TXA SLOT*16 FROM X-REG 3960 LSR 3970 LSR 3980 LSR 3990 LSR 4000 TAY SLOT INTO Y 4010 RTS 4020 *--------------------------------- 4030 * SET UP CURRENT TRACK LOCATION 4040 * IN DRIVE.1.TRACK OR DRIVE.2.TRACK VECTORS, 4050 * INDEXED BY SLOT NUMBER. 4060 * 4070 * (A) = TRACK# TO BE SET UP 4080 *--------------------------------- 4090 SETUP.TRACK 4100 PHA SAVE TRACK # WE WANT TO SET UP 4110 LDY #2 GET DRIVE NUMBER FROM IOB 4120 LDA (IOB.PNTR),Y 4130 ROR SET CARRY IF DRIVE 1, CLEAR IF 2 4140 ROR DRIVE.NO MAKE NEGATIVE IF 1, POSITIVE IF 2 4150 JSR GET.SLOT.IN.Y 4160 PLA GET TRACK # 4170 ASL DOUBLE IT 4180 BIT DRIVE.NO WHICH DRIVE? 4190 BMI .1 DRIVE 1 4200 STA DRIVE.2.TRACK,Y 4210 BPL .2 ...ALWAYS 4220 .1 STA DRIVE.1.TRACK,Y 4230 .2 RTS 4240 *--------------------------------- 4250 FORMAT 4260 *--------------------------------- 4270 .BS $BFB8-* 4280 PHYSICAL.SECTOR.VECTOR 4290 .HS 000D0B09070503010E0C0A080604020F