das.doc

一个C style Assembler的source code

                 8051 DISASSEMBLER, notes on use.

(0) COMMAND LINE, INPUT FORMAT
   This is a disassembler for the Intel 8051-family of processors.  It is
currently set to read up to and including the 8052.

   The typical conmmand line for this disassembler is:

                     das <asm.hex >asm.s

The input file (read from standard input) is assumed to be in Intel Hex
Format.  The output is placed in standard output (or in a file when
redirection is used).  A third file:

                          entries

is used to specify a list of entry points for das to use.

   Currently, it will only accept input in Intel Hex format located between
the addresses 0000 and 4000 (hex).  It will read entry points from a file:
"entries" and recursively disassemble all the addresses "reachible" from
those listed as entry points.  A sample entries file, input file and output file
are provided with the software.  The format of each address in the file is
(starting at column 1):

HHHH<cr>

with the H's representing the digits of the address in hexadecimal.

   Typically entry points are assigned to 0000, the starting address, and
the address of each interrupt-handler that is used.

(1) OUTPUT FORMAT
   The output of the disassembler is not exactly re-assembleable.  When the
disassembler searches for reachible segments of code, it will fail to resolve
indirect jumps (jmp @A + DPTR, jmp @A + PC), and will fail to recognize
virtual jumps created by manipulating the stack pointer, e.g.

push DPL
push DPH
ret        ;;; An indirect jump to the address pointed to by DPTR.

In addition, if the original program contained data tables, these would
normally be passed by untouched.

   The disassembler will regard anything in the allocated code space that
was not accessible as being data.  If necessary, you will have to visually
inspect the disassembled code for indirect jumps and stack pointer
manipulations to determine what addresses are being indirectly referenced.
Then you could add these addresses to the entry point list in the file
"entries" and disassemble again.  This, in turn, may reveal additional
indirect addresses, so that there is an iterative process implied here to
completely disassemble code.

   For assistance, a list of locations where indirect jumps occur will be
listed in the following format:

Indirect jump at <Address>

   In the sample program provided, there were no indirect jumps.  However,
since the assembly-language software implements a multitasking kernel, it
heavily manipulates the stack pointer.  Consquently, the addresses listed in
the entries file are not just the starting address (0000), and interrupt
handler(s) (0023), but also derived addresses: 0067 (found on the first
iteration). and 0131 (found on the second).

   Also, any addresses referenced that lie outside the range of the Intel Hex
Format file will be listed as external references.  These are listed in the
following form:

REF: <Address>

   Data segments will be clearly marked as such, beginning with the header:

DATA at <Address>

following which will be a hexadecimal listing of the "data" in the format
illustrated below:

34 aa ef 00 00 00 00 00 34 ee ff 00 00 00 00 00 |4       4      |

A list of up to 16 bytes in hexadecimal notation is generated, and then between
vertical bars is the list of corresponding ASCII characters. Control characters
and meta characters (those ranging from 128 to 255) are "blanked" out.  The
character listing allows you to easily recognize embedded string constants
wherever they occur.

Labels are generated for SFR's corresponding to the 8052 processor, and Bit
labels are likewise generated.  These are listed in the tables SFRs, and Bits
contained in the disassembler source, and can be modified to suit whatever
version of the 8051 you are using.

In the output, addresses are labeled in the format illustrated below:

                             C3458

The initial letter indicates the number of times this address is referenced
with the following key:

             B = 1 reference, C = 2 references, D = 3, E = 4, etc.

the next 4 items represent that actual address of the label in hexadecimal.

(2) ERRORS
   The disassembler will duly note where disassembled instructions overlap,
or where it cannot further process input.

No entry points listed.
Bad hexadecimal digit in input.
   The input file "entries" was not found, or contained corrupt data.

Address out of range 0 - 4000h in input.
   The input file contains a portion of a program lying outside the address
   range 0 to 4000 (hex).  The disassembler cannot go beyond 4000h.  This is
   a fatal error, disassembly stops.

Bad format string, PC = HHHH.
   Something's wrong with your compiler or operating system, or the source file
   has been corrupted.  Report this promptly to me, if you should see this. :)

Bad hexadecimal digit in input.
Bad checksum.
Unexpected EOF
   The input file contained a portion not strictly in Intel Hex Format.

Entry into ARG at <Address>.
   An entry point is accessing a point that has already been disassembled as
   the interior of an instruction.

OP into ARG at <Address>.
   An attempt was made to disassemble an instruction starting at an address
   already contained in the interior of another instriction.  The disassembler
   is generating overlapping instructions.

ARG into OP at <Address>.
   An attempt was made to disassemble an argument to an instruction at an
   address already disassembled as the first byte of another instruction.

ARG into ARG at <Address>.
   An attempt was made to disassemble an argument to an instruction at an
   address already disassembled as an argument of another instruction.

   The presence of this error indicates that something is wrong with your
   compiler or operating system, or the source file has been corrupted.
   Report this promptly to me, as well, if you should see this error.

Too many entries, PC = <Address>.
   The number of pending points for the disassembler to process exceeded its
   capacity.

(3) USER SUPPORT and GUARANTEE
   This software has been rigorously verified with respect to the description
given above.  It will also conform to the description of the binary coding
of the 8051 processor supplied by Intel.
   Since this was a hobby project mainly intended for myself and not for a
wider audience, you will be on your own in using this software.  However, I am
interested in hearing any feedback you have.  Send comments to my e-mail
address, [e-mail].  If you want to make upgrades, I encourage
you to try your hand at it.  That's exactly why I'm distributing this package
free.