Programmer 7500

home *** CD-ROM | disk | FTP | other *** search

/ Programmer 7500 / MAX_PROGRAMMERS.iso / INFO / CROSSASM / ASM49.ZIP / ASM48.DOC < prev next >

Wrap

Text File | 1988-05-25 | 15.6 KB | 373 lines

ASM48 Version 1.3 Cross Assembler for the Intel 8748 Microcontroller By Vance Navarrette Introduction: The ASM 48 assembler was intended to be a simple, easy to use assembler for the person who only intends to assemble one or two programs for personal use in a hacker type project. This person does not want to spend $200-300 for a professional product, but rather just needs something simple to get him by. The source file is written using any editor that will generate an ascii file (Wordstar, or even (Gag!) Edlin) and is processed using the ASM48 program to produce an assembled object file. Changes for version 1.3: Version 1.3 is now in compiled format using Borland's Turbo Basic. This speeds up the execution by a factor of 8. On a 4.77MHz IBM PC, a 1K object file would have taken approximately 20 minutes to assemble, depending on the number of comments and errors in the source file. This will now take a little over 3 minutes, and on AT class machines the execution time will be measured in seconds, not minutes. The error messages for version 1.3 have been reworked, so that additional kinds of errors are flagged that previously would have gone unflagged (object code generation was still correct, but certain kinds of errors resulted in the source line in which the error occurred to be ignored). To use the program: The program is now much faster than before, but it is impossible to modify to suit your needs unless you have the turbo-basic compiler. Such is the tradeoff for speed. The program can be invoked by typing ASM48 <Enter>. After the program loads, it will immediately execute, and will prompt: "Name of file to assemble?" The user types the name of the source listing file. A sample file name would be "A:Test"<Enter>. Note that the program assumes no extension is present in the file name. If an extension is present, an error will be generated. The program will then prompt: "Do you want to print the assembler listing (Enter=YES)?" The user would type a "N"<Enter> at this point if no hardcopy was wanted, or if there is no printer available for the system. If the assembler is unable to find a printer, or if the printer is not ready, the file will be generated but not printed, and execution will terminate with an error message. It is strongly recommended that a listing be used to de-bug any assembly errors. Assuming the file to be assembled is be found, the assembler then responds : "BEGINNING PASS 1 Press F1 key to abort assembly" The assembler then decodes the listing, looking for opcodes and errors. During this time, a temporary file is created on the disk for use by the assembler to store the partially assembled code listing. For this reason, it is recommended that there be plenty of disk space for large files to assemble. The file will be created on the same disk that the source listing resides. After some time (depending on the amount of code to be assembled), the assembler prints: "BEGINNING PASS 2" and continues to assemble the source. During the second pass, absolute addresses are assigned to labels and branches. Because this assembler produces absolute code, a linker or locator is not required. After assembly is complete, an assembler listing is produced (if requested) and is sent to the list device (printer). In addition to the listing, a label table is dumped to aid in debug, and the number of errors is shown. The object code is placed in a file called <filespec>.OBJ, and the listing, if requested is also placed in a file called <filespec>.LST. The assembler then prints on the CRT screen: "ERRORS THIS ASSEMBLY: XX" and assembly is complete. Syntax of the source file: It is assumed that the user is familiar with assembly language programming, and understands it's elements. For a listing of the Intel mnemonics, see the MCS-48 User's Manual, available from Intel. The structure of a typical source code line is: LABEL OPCODE VARIABLE ;COMMENT <Enter> Note that the above elements may appear in any combination, but only in the order shown. In other words, a label is optional, but if present, must appear before the opcode. The opcode is not needed for a comment to be legal, but must appear BEFORE the comment if present on a line. The assembler uses a FREE FIELD format, that is as long as spaces are used as delimiters, the spacing of the various language elements is irrelevant. A source file might look like this: ; this is a test file for asm48 ;it includes comments and a few sample opcodes label1 MOV P4,A ;COMMENT DEC R1 RR A ;ANOTHER COMMENT INC A LABEL2 XRL A,# label1 ORL P2,#00 ORL BUS,# 0F ADD A,# 55 ORG 000F ; ORG PSEUDO OP DS 05 ; SAVE 5 BYTES DB 00,FF,88 ; DEFINE 3 BYTES ASIC EQU 00 ASID EQU ASIC labelxyz JMP LABEL2 XRL A,#FA ;TEST FOR HEX NOP CALL label1 JMP 0187 ;TEST FOR GOOD JMP DT HELLO, DEVO! ARCANE EQU LABEL2 ; END OF TEST Object Code: The object code is placed in an Intel compatible ascii HEX listing file. The basic format of the hex dump is: :BBAAAA00B1B2B3B4B5B6B7B8B9BABCBDBE...B(N)SS :BBAAAA00B1B2B3B4B5B6B7B8B9BABCBDBE...B(N)SS . . . :0000000000 Where: BB=Number of bytes in the record (max=255) AAAA=Absolute load Address for this record 00=Always two 0s. B1, B2,....B(N)=Object code (Hexadecimal) SS=Two's compliment Check Sum of BB, AAAA, and B1- B(N). The last record (line) will be all zeroes. Note that NO BINARY OBJECT FILE IS PRODUCED. Thus, the output of this assembler cannot be loaded directly and executed. This was deemed un-necessary, since the target system is a controller, and not the host microprocessor. If you need a straight binary file, the INHEX subroutine in the basic listing (If you have it) may be modified to produce one. Points to Remember!: a. The assembler is case sensitive (i.e. "a" does NOT equal "A") b. All mnemonics must be in caps (labels need not be). c. All comments start with ";". d. The vertical bar "|" is a reserved character. Do not use it! The results are unpredictable. e. All numbers are to be in hex. Binary and decimal numbers are not supported. f. No math operators (+, -, etc.) are supported. g. The file to be assembled must be an ascii coded text file. No special control codes are allowed. If you use a word processor like "Word Star" to edit the file, be sure to use the non-document mode or else do a ^QU command in the non document mode to insure that the assembler won't get confused. You will get error messages with nothing (apparently) wrong in the listing file when you try to debug your source code if you fail to observe this rule. h. All Hexadecimal number must have two characters, even if they are both zeros. i. The ^C and the ^Scroll Lock keys are NOT supported. Use F1 key to terminate assembly prematurely. Pseudo Opcodes: In addition to the standard Intel mnemonics, several pseudo opcodes are supported to control the flow of the assembler and aid in producing code with a minimum of fuss. Pseudo opcodes produce no hexadecimal output, and take up no space in the assembled code, although they will show up in your listing and will show an "address" in the listing. They are: DB "define byte" DT "define Text" EQU "equals" ORG "Set Origin" DS "Define space" DB: Used to define a byte or bytes of code, using hexadecimal. Note that hex notation is always assumed, and that multiple bytes are delimited with a single space or comma. Do not use an "H" to designate hex, as it is assumed. It is also permissible to define a constant using the "EQU" opcode, and then use the label previously defined as a variable. DT: Used to define text. This generates the ascii code for each letter in a text string. Delimiters are not necessary. EQU: Sets the label on the left equal to the variable on the right. The variable on the right may be hex or a previously defined label. ORG: This sets the address counter used for assembly to the value of the hex variable or label on the right of the pseudo-op. The ORG opcode forces the assembly of subsequent code to begin at the address given by the variable used for ORG. May be used with hex or a previously defined label. DS: This allocates space in memory immediately following the opcode. The amount of space is defined by the variable used. The Variable may be Hex or previously defined Label. Error Messages: The following error messages are generated during assembly, and are fatal (i.e. object code would have been incorrect, so the object file was deleted). The error messages will appear as com- ments in the listing (.LST file). While the original comment will appear to have been erased and the error message substituted, the original comment is in fact still present in the source file. Correcting the cause of the error and reassembling the source file will cause the original comment to appear in the listing file. ILLEGAL VARIABLE: A non-hexadecimal character was found where only 0-9 or A-F characters were expected, or a label was used as a variable and was not previously defined. This error message will also be generated if an attempt is made to define a variable that is too large (>FFF hexadecimal) or use a variable larger than 0FF hex in a branch instruction that expects a single byte variable (i.e. using a variable larger than 0FF in any branch instruction other than JMP or CALL). UNKNOWN OPCODE: An opcode was used that is not legal, or an opcode is missing, resulting in a variable being interpreted as an opcode. Another possible cause is using lower case letters for an opcode (all opcodes must be in CAPS). Also note that embedded spaces in opcodes are considered significant, and must be present if defined by INTEL mnemonics. MISSING VARIABLE: A variable was not found following an opcode that requires a variable. Also used when a DT pseudo op is used, and there is no text, or when there are no variables following a DB pseudo op. LABEL TOO LONG: A label was encountered that was longer than 10 characters. If a label is too long and subsequently used as a variable, an ILLEGAL VARIABLE error message will result. DUPLICATE DEFINITION: A label that was previously defined was re-defined subsequent to the original definition. The label will appear twice in the label table, but only the first definition will be used in the assembly. PAGE ERROR: The 8748 uses certain jump instructions that are only valid in a paged memory scheme (this is done for the sake of compactness and speed. Most branches in a typical assembly program jump to locations that are close by...).An attempt was made to cross a page boundary (Address in the range of X00-XFF hex) using a branch instruction that only supports branching within the page in which that branch instruction occurs. The assembler has generated code that will result in a jump to an incorrect address. This error is not necessarily even possible to see while writing the source code. The simplest way to correct the problem is to substitute a JMP instruction, but it is possible to simply move the offending subroutine inside the addressing page where the error occurred. A word on debugging the source code: It is possible for more than one error to occur on a given line of source code. ONLY THE MOST RECENT ERROR MESSAGE WILL APPEAR. Thus it is possible to have an "unknown opcode" error message appear, to then correct the opcode error and reassemble, only to have a "Illegal Variable" error message appear. To save time, it is therefore prudent to examine each flagged source line for other mistakes that might have occurred during assembly. The error count given at the close of assembly will not match a physical count of the error messages under these circumstances, and should be taken as an indication of multiple errors on single source lines. It is also possible that correcting one error will generate another. For instance, the following line: TLABEL ORg 5G ;Comment..... Will first result in the "Unknown Opcode" message. After the line is corrected: TLABEL ORG 5G ;Comment..... It will generate an "Illegal Variable" Error, since 5G is not a legal hex number. BASIC error messages: Several errors can arise that are not part of the assembler flow itself, but are generated by BASIC as a result of unexpected conditions during the execution of the assembler program. They are: PRINTER IS OUT OF PAPER OR IS OFF LINE: A listing was requested, but the print device is unavailable. DISK DRIVE IS FULL: The disk drive that the source file was found on has insufficient space to allow the assembler to complete assembly. There are several options to correct this: 1. Copy the source file to a floppy or Hard disk that has more room, and re-assemble the file. 2. Request that no list file be generated. This will greatly reduce the space (and time) required to assemble the file. (It will also make it nearly impossible to debug. The total error count will appear on the screen at the close of assembly.) 3. Create a copy of the source file, and edit it to eliminate all comments, blank lines, etc. Then proceed to assemble the new source file. DISK DRIVE NOT READY: The disk drive named for use by the assembler is unavailable to the system. Usually caused by an open door on the disk drive or a write protected file or disk. SOURCE FILE NOT FOUND: The file specified for assembly is not on the specified disk drive, or is mis-spelled. This is a non fatal error, and can be recovered by retyping the file specification, including the disk drive. Be sure to include the name of the sub-directory that the file may be located in. Example: A:\SRCFILES\TEST Additional error messages may be generated by turbo-basic that are not found here. These will appear as error code numbers. Consult your BASIC manual to correct the error conditions, using the error code number that appears on the screen. Be careful if you do not have access to a Borlan Turbo-Basic manual. While the error codes are in the vast majority of cases identicle, there can be subtle differences in the definition of the error code.