fdisasm-1.2.8.zip (info) a zipped hfs disk image and checksum file. The disk image can be mounted with Mini vMac. Includes source code.
fdplusv3-0.4.3.zip (info) a disk image containing all the formatting information needed to disassemble a Macintosh Plus ROM image (version 3, final - checksum 4D1F8172).
ROM Formatting Information - more images to disassemble other Macintosh ROMs.
FDisasm is a formatting disassembler for Motorola 680x0 code. It makes it possible to distribute the means to create an annotated disassembly of copyrighted code (such as Macintosh ROM images), without distributing that code or files derived from it.
To use FDisasm to disassemble the Macintosh Plus ROM, launch Mini vMac and boot from a disk image containing a system folder. Mount the fdplusv3-0.4.3 disk image, and import a copy of the Macintosh Plus ROM onto it. (Such as by using ImportFl.) Mount the fdisasm-1.2.8 disk image, copy the FDisasm application to the fdplus disk image, and then eject the fdisasm image. Rename the ROM image to “bin”, and then launch the FDisasm application. The program displays a progress indicator, and quits when it is finished. (It could take five minutes at 1x speed, so set speed to “all out”. At “all out” speed it takes under 10 seconds to finish on my Mac Mini.) When the program finishes, the output is in a file named “listing”.
This listing file is over 2 megabytes. The program MPW Shell 3.2.3 works in Mini vMac and can work with such large text files. But probably almost nobody else has an old copy of MPW. (The final version of MPW still available for free download requires a Mac II or later.) Instead you can export the listing file to the host computer (such as by using ExportFl), and find some program there that can view it. (For OS X, TextWrangler would work, for example.)
Warning : Not enough people are paying enough attention to FDisasm (especially me) for it to be relied upon. If you notice something odd in the output, double check it, and if it is wrong, please report it to me.
This listing file is derived (in terms of copyright law) from the Macintosh Plus ROM, and so may not be redistributed.
To disassemble the Macintosh 128/512 ROM, follow the above instructions, except use the fdmac128-0.3.0 disk image, and of course use a ROM image from a Macintosh 128 or 512. And similarly for the Macintosh SE, Macintosh II, and PowerBook 100 ROMs.
Here are some details about the formatting information files used by FDisasm:
The only file required by FDisasm is “bin”. But if this is all it finds, it will just create a hexadecimal dump of the file. To disassemble code, it needs to know where the code is. It looks for a file named “bin_map” for this information. It also looks for a file named “bin_names” for names of locations within “bin”, and for a file named “abs_names” for names of absolute memory locations, and for a file named “trap_names” for names of A-line instructions. A file named “format” can give additional formatting information, and comments to include in the listing.
The FindCode tool can help to create the files “bin_map” and “bin_names”.
The FindRes tool will create an initial “format” file for Macintosh resources, and also “bin_map” and a set of entry points to feed to FindCode.
Besides ROM images, FDisasm can also disassemble Macintosh 680x0 applications. But the code and other resources are located in the “resource fork” of such appplications, while FDisasm only operates on the “data fork” of the file “bin”. The GetRsFrk tool gets the resource fork of a file you select, and saves it into the data fork of “bin”. An alternative is asPack, which also saves the data fork and the meta data into “bin”, in AppleSingle format.
FDisasm has some compile time options. The default values are defined in “source/CNFGDLFT.i”, and these defaults can be changed in “source/CONFIGUR.i”. For example, the following variation does not output the address and hex fields, resulting in output that is suitable for passing to an assembler:
fdis0002-1.2.8.zip (info) variation of FDisasm.
The software that has become FDisasm, FindCode, and FindRes was my first serious programming project, started in 1985 in Basic, and later converted to Pascal, and now converted to C when I got back to it recently. I wonder if 23 years to first public release is any kind of record.
See the Compiling page for instructions on compiling FDisasm from the source code.
You can redistribute FDisasm and/or modify it under the terms of version 2 of the GNU General Public License as published by the Free Software Foundation. See the included file COPYING.
FDisasm is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the license for more details.
April 7, 2019 version 1.2.8 will now output “.W” for word sized operations, rather than relying on the default being word. I feel this is better style. ReAsm currently requires this. There are also some new compile time options for improving compatibility with ReAsm, such as forcing labels to be on a separate line, and using tab to indent instructions. These options are used in the fdis0002 variation.
February 17, 2019 version 1.2.7 fixes a bug reported by Ryan that affected disassembling floating point conditional long branches. It also partially reverts for now a change made in 1.2.6. If an instruction (or data) that is about to be output is longer than the space left to be disassembled, it will now not abort outputting the instruction, and just give a warning. This version also starts to provide ways to avoid the situations that lead to this problem. Disassembly of PC relative addressing modes can now use the AC and SC options for adding or subtracting a constant. There is a new ‘,’ option to indicate which argument options like AC and SC apply to. It also now checks that such one time options are actually used by the current instruction, and prints a warning if not.
January 6, 2019 version 1.2.6 adds a “BN” option to modify “B”, “W” or “L” options to output binary instead of hex. Also, if an instruction (or data) that is about to be output is longer than the space left to be disassembled, then the output is aborted, and the remaining space is output as data of unknown format. Also, the AM option (which modifies the D option) now can take an address as well as a name, and checks the address is correct. Similarly the J option can now take a name, making the JM option obsolete. The “A”, “H”, and “Z” options have been changed to compile time options, which can be specified in the new file “CONFIGUR.i” with the default values specified in “CNFGDLFT.i”.
December 30, 2018 version 1.2.5 properly limits the size of the hex field. Previously the hex field would be incorrect for very long lines, such as can be generated with the new “C” option. (For the future, it would be good to limit the length of lines generated by this option.)
December 23, 2018 version 1.2.4 adds an “AL” option in the format file for aligment relative to a base address.
December 16, 2018 version 1.2.3 adds a “C” option in the format file for text fields. Also, for the “D” option (offset fields), there are three new modifiers: “AC” for an added constant, “SC” for a subtracted constant, and “DC” for a constant devided by.
December 9, 2018 version 1.2.2 adds the ability to specify labels in the format file (in addition to the “bin_names” file), so that a single address can have multiple labels. Also in the format file, you can now specify a one time name to use in an address field argument (overriding the “bin_names” file). Similarly, in the format file, can you specify a name to use for the current base address used by the D option. The FindRes tool has been updated to begin to take advantage.
July 13, 2016 version 1.2.1 adds the ability to format jump table offsets as the difference between two symbolic labels (instead of only listing the numeric value). Mac Plus and Mac 128K formatting information has been updated to take advantage.
July 1, 2016 version 1.2.0 adds initial support for FPU instructions, and works well enough to correctly disassemble the Floating-Point Arithmetic Package (PACK 4) of the Macintosh II ROM (verified with the MPW Assembler). The disk image with the Mac II formatting information has also been updated.
June 24, 2016 version 1.1.9 now works well enough that feeding the output to the MPW Assembler gets back the original binary for the Macintosh II ROM. Except that the Floating-Point Arithmetic Package (PACK 4) is not disassembled because FPU instructions are not yet handled. The disk image with the Mac II formatting information has also been updated.
The specific changes include: The 68020 forms of Mul and Div instructions are now supported. Also the ExtB.L instruction. A bug in MoveC is fixed. A bug in Bitfield register width argument is fixed. And also the code is restructured a bit to better match the code in Mini vMac CPU emulation, for easier comparison.
June 15, 2016 version 1.1.8 now works well enough that feeding the output to the MPW Assembler gets back the original binary for the Macintosh PowerBook 100 ROM. Even though the PowerBook 100 has a 68000 CPU, this ROM has some 68020 code. FDisasm can now better handle the full extension word format of addressing mode 6, the absolute long addressing mode, and the Bitfield instructions. The disk image with the PowerBook 100 formatting information has also been updated.
January 1, 2015 version 1.1.7 now works well enough that feeding the output to the MPW Assembler gets back the original binary for the Macintosh Plus ROM, the Macintosh 128/512 ROM, and the Macintosh SE ROM.
This involved figuring out what syntax the MPW Assembler accepts for PC relative address mode, when not using a symbolic label. (It is expecting an address, not the offset.) The MoveC instruction is now better supported. And there is a fix for the MoveP instruction.
The disk images with formatting information have been updated for Macintosh Plus ROM, the Macintosh 128/512 ROM, and the Macintosh SE ROM. The '.' character in labels is not accepted by the MPW Assembler, and so replaced with '_'. Also, in a few cases, two addresses were labeled with the same name. But only one of the two are used by the ROM, and so the other label was removed.
December 30, 2014 version 1.1.6 fixes incorrect size field for BTST instruction as reported by Steve Chamberlin. There are also a number of other changes to better allow the MPW Assembler to assemble the output of FDisasm and get back the original binary.
December 28, 2014 version 1.1.5 changes format of negative hex numbers from “$-A” to “-$A”, so as to work with the Easy 68K assembler, as requested by Steve Chamberlin.
December 26, 2014 version 1.1.4 fixes a number of bugs reported by Steve Chamberlin.
May 17, 2014 fdmac128-0.4.1 includes information reported by Tom Frikker about initials at the end of the Mac 128k ROM. This lead to finding a report by ‘Dog Cow’ about more initials in the Twiggy Mac ROM (for an early prototype). So there may only be two initials in the Mac 128k ROM because the rest got overwritten, like happened in later versions of the Macintosh Plus ROM.
September 12, 2012 version 1.1.3 includes code from Rob Braun for the "pmove, pload, and pflush family of instructions for the 68851 and 68030".
September 7, 2012 version 1.1.2 includes a bug fix from Rob Braun. A 32 bit offset for the Bsr or Bra instruction was not decoded correctly.
October 6, 2011 fdplus-0.4.0 merges in changes from Steve Chamberlin, that "added many more comments of my own, and many new meaningful names for previously-unamed routine entry points, especially in areas related to the boot sequence and Sony disk driver".
September 25, 2011 new versions of all the formatting information disk images, removing the FDisasm application. The version of the application on them was out of date.
June 5, 2011 version 1.1.1 fixes a bug reported by "Ben". The size field was not correct for AddQ and SubQ.
If you find FDisasm useful, please consider helping the Gryphel Project, of which it is a part.