bson@rice-chex.ai.mit.edu (Jan Brittenson) (09/15/90)
I have just (sort of) finished a Saturn Disassembler. I would like a couple of people out there who have previously disassembled parts of the HP-48SX ROMs to verify that the output is correct by disassembling previously disassembled areas and comparing the results. If you are interested in doing this, drop me a line. All others, please hang on until most bugs are out, and a couple of tools have been added to update the symbol and comment databases using edited listings as input, as well as a GNU Emacs mode. The preparatory work required is to dump areas of the ROMs into your UNIX system (DOS users will be on their own here), and convert it into Core format. SAD consults three databases: Core, Symbols, and Comments. Only Core is necessary for the verification. Included is a sample SAD 1.01 listing ([...] identify comments inserted by SAD, the rest are picked from the Comments database): 05b79 05b79 ; Allocate string on heap. 05b79 ; C is to contain the string length, in bytes. 05b79 ; Sets r0 to object header address, and d0 to data address. 05b79 ; Sets mode to hex. 05b79 05b79 stralloc: 05b79 04 sethex 05b7b c6 add.a c, c ; Bytes to nibbles 05b7d 84a clrb #a, st ; No GC performed 05b80 24 move.1 4, p 05b82 809 add.a p+1, c 05b85 108 move.w c, r0 05b88 20 move.1 0, p 05b8a 8f60860 call.a c=free_mem ; [<#6806>] 05b91 da move.a c, a 05b93 118 move.w r0, c 05b96 2f move.1 15, p 05b98 809 add.a p+1, c 05b9b 20 move.1 0, p 05b9d 8bed0 brle.a c, a, stralloc+#34 ; [<#5bad>] 05ba2 8fd3361 call.a #1633d ; GC? 05ba9 6edf jump.3 stralloc+#f ; [<#5b88>] 05bad 118 move.w r0, c 05bb0 24 move.1 4, p 05bb2 809 add.a p+1, c 05bb5 20 move.1 0, p 05bb7 8f8da60 call.a #6ad8 ; blkalloc? 05bbe 132 swap.a a, d0 05bc1 130 move.a a, d0 05bc4 120 swap.w a, r0 05bc7 34c2a20 move.p5 type_String, c ; [<#2a2c>] 05bce 144 move.a c, @d0 05bd1 164 add.a 5, d0 05bd4 140 move.a a, @d0 05bd7 164 add.a 5, d0 05bda 1f97507 move.5 TOS, d1 ; [<#70579>] 05be1 147 move.a @d1, c 05be4 135 move.a c, d1 05be7 03 retclrc
alonzo@microsoft.UUCP (Alonzo GARIEPY) (09/16/90)
In article <10712@life.ai.mit.edu> bson@rice-chex.ai.mit.edu (Jan Brittenson) writes: > I have just (sort of) finished a Saturn Disassembler. I would like > a couple of people out there who have previously disassembled parts of > the HP-48SX ROMs to verify that the output is correct by disassembling > previously disassembled areas and comparing the results. > > 05b79 stralloc: > 05b79 04 sethex > [...] Congratulations, this is excellent work. I verified the stralloc routine and everything checks out okay. More tests are definately in order to catch some of the trickier problems. I do have some suggestions to make based on how my assembler/disassembler and Andreas's assembler work. These aren't necessarily the best ways to do things so we should discuss them in order to come to some standard. 1. All hex numbers should be preceded by the # symbol, including the machine code and addresses in listing files. 2. Standard symbols begin with a letter or underscore and contain only numbers, letters and underscores (we can talk about adding such things as the $ and .). 3. Nonstandard symbols are enclosed in quotes and can contain any characters. Example: "Garbage Collector!" When used as labels, the colon goes outside the quotes. 4. All symbols used as arguments should be defined in the disassembly (controllable by command line option). The simplest way to define an address symbol is #xxxxx: symbol: Andreas also supports assignment for address and nonaddress symbols symbol=#xxxxx This has the disadvantage of a potential parsing problem if 'symbol' conflicts with a keyword. One modification is that a standard symbol cannot be a keyword (i.e., if you want the symbol "move" it must always be enclosed in quotes). I imagine we can come up with a scheme where numeric symbols don't conflict with keywords. 5. I have seen many assemblers get into trouble with poor syntax for macro capabilities, so I am inclined to think this through and make it part of the standard. My preference would be to use an existing, portable, macro processor. Cpp is one possibility, but not very attractive. The ideal would be something that is powerful enough to implement the assembler and disassembler itself. Any suggestions out there? 6. Whether arguments are expressed in hex and commented with symbol names, or expressed symbolically and commented with hex equivalents, should be controllable by a command line switch. 7. Comments should not be used for machine readable information. Thus the syntax [<#xxxxx>] should be replaced with the simpler #xxxxx. 8. The data pseudo op is used to put non instructions into the code. Your disassembler should use the data op for anything that is not a valid instruction. Here are a few examples of its use from the CHIP48 source by Andreas Gustafsson... data.b 'A' data.a #2dcc ; machine code object begin: data.a end-begin ; length of object * * * At one point I was planning a better disassembler (like yours!) and I have some interesting ideas about RPL and synchronization (always a problem with disassembly) that you might like to hear. Drop me a line. Alonzo Gariepy alonzo@microsoft