Sega Saturn DeBuGger ver. 0.7 alpha. Readme file.
31337 Codin' by TyRaNiD (c) 2000
What this Program is:
--------------------
This program allows you to debug home-brew Sega Saturn code actually on ur saturn
with the help of a Pro Action Replay cart and a PC Commlink. Why u might want to
this is open to debate
Program Requirements:
---------------------
-=Needed for running and debugging code=-
IBM compatible PC, at least a 386 (to run da protected mode code)
A Sega Saturn. Any model should work from any country in theory.
Action replay cartridge. Also called a gameshark or Game Key depending
on where you live.
PC Commlink. A ISA board for your PC which allows connection to your Action Replay.
Without this the program hasnt got a hope of working. Sold mainly now for the psx
action replay but it should work on the saturn as well.
-=Needed to make ur own code=-
Saturn compiler/assembler and libraries.
Documents on the saturn.
I Cannot supply these so dont even ask about it.
BIG NOTE !!!!!!!
----------------
Im am hoping this version should support both datel and ems versions of the par by using
a command line switch or ini file setting. It will definetly work on the ems version
but i dont know if its the same for the datel. Only time will tell.
The program does not support the parallel port commlink developed by Freewing.
Mainly due to lack of documentation and desire. Anyway the PP version
would be hella slow. If of course someone really really wants the support then email
me and ill see what I can do.
This program is farrrrrrrrrr from finished yet. However I feel it has a
very long way to go as im surprised on how solid and easy using the UBC seems.
(lulled into a false sense of security no doubt)
USAGE:
To start debugging first run the program from dos or in a windows dos prompt,
using the following command line parameters.
satdis.exe filename [options]
filename is the name of the program to load. Required.
[options]
-lX X is the load address of the program in hex
-pX X is the port number where X is in 0x3X0 of the commlink port no.
-c Force coff format. Required if input file doesnt have .cof extension.
-d Set Datel PAR support. Without this switch set it would default to ems.
-? Displays help.
Notes on coff support.
I really hope it works with any coff file u want to throw at it but there might
be problems as ive not worked out the format from any official documentation. If
anyone know where I can find the exact specification for the coff format then plz
email me (a direct link please not just its on gccs web site or sommit).
EXAMPLES:
Loading a binary file.
satdis.exe program.bin -l0x6004000
Loading a coff file.
satdis.exe program.cof
NOTES
make sure the saturn is at the PARs main menu (or itll just hang)
(Well it could hang if ur port address is incorrect but doh..)
Oh if it does hang u should be able to CTRL+C it maybe..
It dont matter on the load address if u add a 0x infront or not itll still
work.
The load program can be in bin or coff format. The program will automatically assume
and program with a .cof extension is a coff file. If it does not have this extension
either rename the file or use the -c switch to force load as a coff file. If the
program thinks its loading a coff file when u specify a binary file it will certainly
crash the saturn and probably the debugger as well.
Config file format "satdebug.ini"
Comments r set by adding a # at the start of the line.
[options]
port=X. Allows u to specify the commlink port address. Can be overridden by passing
the -pX parameter.
datel=0|1. Sets datel support. Setting it to 0 means use da ems default and 1 means datel.
Debugger commands
-----------------
F1 : Memory to Screen Dump. Type in the address and press PGUP and PGDOWN to move
the memory dumping address by 256 byte blocks. Press ESC to exit.
Real time memory editor. Pick the range of values in the memory dump to edit
and press F1. Works like a standard hex editor and will modify values in real
time. Press TAB to switch between HEX mode and ASCII mode of editing. Press
ESC to quit and if u have modified any memory it will ask u to confirm the
write. If u press any other key than Y it will reset the memory as before
editing.
F2 : Write a byte to memory.
Type in address and value and it will write to memory
F3 : Write a word to memory.
Type in address and word value is written to memory. U dont have to specify
a word boundry address as it will be able to write it no matter what.
F4 : Write a long word to memory.
Type in an address and long word value. U dont have to specify a long word
boundry as it will write anyway.
F5 : Modify a register. Type the register number (given by the side of the reg dump)
and long value. New value will added and screen updated. It will also modify
the disasm if u have modfied the PC reg.
Oh and u cant modfify the VBR or the SP because this would screw the debugger up
Be very careful with this. U could crash the saturn by puttin in a bad value i.e.
a non long boundry reg when its the address for a long read.
Also u cant set the I in SR flags to any value greater than 0xE (or any value with
the LSB set). This stops u from killing the debugger by disabling the ability to
call the UBC int which is hardware set at priority 15. (why u would want to do
this in the first place is beyond me).
F6 : Dump memory block to a file. Type filename, start address and end
address. (Note. Memory will be read up to end addr -1 ).
F7 : Load file to memory. Type filename and start address. Precautions must be taken
as with any of the memory routines to ensure u dont overwrite important data.
Possibly only truly useful to quickly set a grafix screen to a know state.
F8 : Run program. Execution can be cancelled at any time by pressing a key.
F9 : Step into instruction. Executes 1 op and steps into a jump, branch.
F10 : Step over. If executed on a sub routine jump will skip over it.
Execution can be broken at any time by pressing a key. (note will destroy
the call back breakpoint).
UP,
DOWN,
PGUP,
PGDOWN: Moves the disassembly by a certain amount. Up and down cursors move the disasm by
1 instruction where pgup and down will move by 10 in the respective direction.
HOME : Resets the disassembly to the current instruction position.
Esc : Quit the program and remove the UBC handler. Will allow program to properly
execute at full speed.
ALT+D : Dump from register value. Put in the regsiter dump and it will do a memory dump
at the memory address pointed to by the register. R15 and VBR are included in this
and have the reg nos of 15 and 16 respectivly. Note registers which cannot be used
directly for memory addressing are not included. These being MACL,MACH and SR.
SPACE : Set a breakpoint on current instruction. Press again on instruction to remove.
ALT+L : Display current set breakpoints. When list produced allow setting and deleting
breakpoints without finding them in the disasm.
ALT+I : Inline Asm. Replaces the currently selected instruction.
ALT+U : Undos the previous inline asm. If the asm command has compiled wrong then just
use this command to reset the instruction.
ALT+R : Reset the current program to its original location. Reloads program so it might
take some time to restart.
'?' : Display online help. Not very descriptive just useful if uve forgotten the key u
need.
The memory routines above will only accept certain address ranges at the moment.
To be honest I dont think there are any other main areas which rnt included.
Mem Areas
---------
0x(2)0000000 -> 0x(2)0080000 : Boot ROM (R)
0x(2)0180000 -> 0x(2)0190000 : Backup RAM (R)
0x(2)0200000 -> 0x(2)0300000 : Lower Work RAM (R/W)
0x(2)5C00000 -> 0x(2)5C80000 : VDP1 VRAM (R/W)
0x(2)5E00000 -> 0x(2)5E80000 : VDP2 VRAM (R/W)
0x(2)5F00000 -> 0x(2)5F01000 : VDP2 CRAM (R/W)
0x(2)6000000 -> 0x(2)6100000 : Upper Work RAM (R/W)
Obviously R and W specify the access available for the memory. I have disable write access
to the BIOS and the backup RAM for obvious reason (well I hope they r obvious if not go sit
in a corner until u figure it out).
And also the (2) signifies a cache through address so u can either. However memory dumping
and writing is best done on non cached addresses as this will reflect the programs likely
data.
MORE BIG NOTES!!!!!!! And bugs hehe.
------------------------------------
Here are know problems with the underlying system. Mostly out of my control.
1) Program wont seem to execute delayed ops, interrupt protected instructions
i.e. lds etc, and will not work in areas where the SRs I mask is set to 15.
The debugger should return correctly at the first possible instruction available
will just not be able to debug those instuctions. Due to limitations in the UBC.
(Out of my control)
2) U can only load programs into the higher work ram, (Not really a problem
just a point to note). Will be remedied later on and allow loading to lower
as well.
3) If the size of the program + Start Address > 0x6040000 u will most likely
trash the UBC handler so try and make sure it dont. Ill move it to a better
position when its closer to finishing (basically when I know the rough total
code size of the bootloader). Im looking into setting the cache as 2 way and
storing the boot code into that. As long as ur program doesnt touch the cache
or its regs then in theory it might work well. And then it means that no matter
what u put in main memory it will never overwrite the UBC code. However u cant be
sure that sega demos etc will not adhere to this (or for that matter the sega libs
themselves)
4) I have not extensively tested the step over command so it might not work
inall cases. One case where it will not work is if a subroutine sets the SR
I mask to 15 and doesnt reset it till after the return. As no break would
occur on the instruction required it wouldnt work. However as this is blatent
bad coding practice so dont even go there Ricky.
5) U can only move the disasm within the upper work ram. This will only cause a problem
if the program jumps to a bios routine or a lower work ram address then u will not be able
to move the disasm window. It should still execute however.
6) I think if u modify the instruction u are currently on using the mem commands it will
still execute the original instruction. Only way round this I think would be to do a cache
purge. If u need to get across a branch modify the sr reg instead or sommit.
7) When using the memory dump if u try and move to a address outside the boundries
of the memory block ur in it will give u an invalid memory range error and cancel
the dump. Just irritating more than anything else.
8) Da inline asm needs to be tested properly. It could produce incorrect opcodes for valid
instructions etc or not pick up on errors. Who knows. There r some know perculiarities
in the assembler, such as all displacements immediates and branch addresses have to be
in hex without a 0x at the start. It is also best to avoid any spaces in the instruction
except at the beginning and between the opname and the first arg. The instruction formats
should be the same as in the sh2 manual. If in doubt read up on it.
Things to do now.
-----------------
1) Add some actual functionality to the program. i.e. memory dump, run over
sub routine, modify registers etc. Need to add some meat to the bootloader.
DONE SOME OF IT. Still need to add breakpoints etc.
2) Make the disasm and reg display better and include the missing sh2 regs
(VBR, GBR, MACL, MACH). Didn't add them out of lazyness.
DONE sorta. Still needs alot of work doing to it.
3) Display a text screen on the saturn when the bootloader has started just
to give an indication that everythings working correctly.(Better than no
change at all which occurs at the moment). Probably can reuse PARS predefined
font Possibly wont bother. I might just add a load percentage bar on the pc
side to indicate the program is loading.
4) Add a command to call back from inside a procedure (using the PR reg).
Should in theory be able to drop back to the last sub routine jump.
5) Hopefully dus might be able to port the debugger to linux to get an even
smaller target audience
6) Stuff, stuff and more stuff Of course source level debugging far in
da future (i.e. when I have time to rip off the DJGPP rhide debugger parser
, but at least it looks like it can be done with some hard work).
DISCLAIMER.
----------
I do not provide any warranty on the security of your data on both ur saturn or
pc. While this program is extremely unlikely to damage anything there is always
the tiny chance of it doing so. If used responsibly and correctly nothing will
ever happen and thus I take no responsibilty for any damage cause through the use
of this proggy.
This program is not affilated or approved by either Sega or the manufacturer of
your Action Replay. Plz do not contact them with any problems u have as they will
invariably tell u to go f*#k urself.
Revision History:
----------------
0.7 Alpha : Added inbuilt inline asm capability. This should also make it usable in
my linux port without a copy of gas.
U should get error messages and everything now
Added support for datel and ems versions of the action replay in the same
program.
Unfortunately my commlink has died so i havent been able to test the program
to ensure it work
0.6 Alpha : Real time memory editing facility. HEX/ASCII editor format.
Added a program reset using the ALT + R command.
Finally added cached through memory addresses to the memory routines.
User cant set a non word boundry PC,PR or invalid SR reg value now.
Added a breakpoint editor to disable and set new breakpoints without
directly setting them in the disasm window.
Added easy to read bit by bit display of the SR reg (with bit names above).
Added a load progress bar on the pc side to indicate a program is indeed
loading.
Added a load file to memory command. NOTE F5 and F6 commands reversed.
Added an undo command for the last inline asm command. Useful if uve
have assembled the wrong instruction.
Rnt u lucky boyz and girls. I finally found how to delete filez so no more
files left about cause of gas
Finally sorted out the exact cause of a interface bug which has plagued me
since I wrote satonem in djgpp. Finally realised there must be a bug in the
gets() function and have written my own version to compensate.
Fixed a few things which might have caused problems in last release on
some systems (what systems? No one is using the program)
0.5 Alpha : A super cheat added. Inline assembler function as long as u have
the gcc assembler gas installed and setup in the config file
Config file system make it less hassle to pass the port number etc.
Preliminary coff file support so u dont even have to type in the
programs load/start address
Modified the program parameters slightly.
Made sure any configuration errors where possible occur before the loading
of the boot code. Means u shouldnt have to reset ur saturn if u type in an
invalid load file.
0.41 Alpha: Fixed a bug which caused most of the programs to crash. Its been there
since 0.2a. Whoops..
0.4 Alpha : First public release.
Added breakpoint commands and some pretty colours.
Added online help command. Wahey
0.3 Alpha : Added more commands. Write byte,word,long. Modify registers.
6/2/2000 Reordered the key assigns to make them slightly easier. (well so
dus dont have to press the shift key ).
Added movement of disasm window. However its not much use at the mo.
Slightly adjusted the memory dump to allow u to go back through the address
range as well as forward.
Added a memory dump from register value command.
Fixed a (possible!) bug in the step over command which might occur with
a lds instruction modifying the pr regsiter. Now will only cause a step
over if the pr reg value is equal to the step pc reg.
0.2 Alpha : Added some proper commands.Step in, step over, mem to file, mem
3/2/2000 dump. Prelim key assigns (will change soon as they r crappy).
Made the interface abit neater.
Fixed some so called bugs/niggles which needed addressing first,
i.e. adding all the regs in the Read Reg command (fucked alot up
Added a new comm routine command which speeds up continuous
execution by approx 2000% hehe
0.1 Alpha : Initial test release.
2/2/2000
Program written using DJGPP C compiler (for DOS side) and GCC SH2 assembler
(for Saturn side).
This program is all my own code and so is copyrighted by me It does not use
any commercial libraries or code from either the pc or saturn side so im not
stepping on anybodies toes. The inline asm code is all my own work except for the
opcode table which is from the gcc binutils distribution. As this is under gpl
im not sure whether to release it but it isnt any actual code it just a set of
data which I used to avoid rewriting it. So I probably wont.
Thanx and Greetz:
-----------------
Thanx to dus for program testing and all the other benefits u bestow upon me
Thanx to PI for very helpful PAR documentation/sample code/and for showing that u
can even run programs on a saturn at all.
Thanx to Charles Doty for his saturn dev site.
Thanx to SEGA for producing a machine which seems fairly easy to make a debugger
for
Thanx to Datel UK for producing the PAR cart in the first place.
Thanx to MGhandi (www.kinox.org) for being the only emusite webmaster who
bothered to put up a mention of my debugger.
Greetz to Azuco, of A-Saturn.
Greetz to Alucard and Tunek.
Greetz to da Efnet crew (in no particular order) Gato, Vitor, Ame, Albx, Cowering,
Elmer, Design, Cokeman, The_Morph, IceMan2k, |Chakan|, |Ray|, Kanon, Haljordan, Crokk
and no doubt more i cant remember
Finally greetz to all the other ppl ive met along my way into obscurity
If u feel uve been missed out of the list then I dont care, haha.
Oh and skywalker of hitmen, give me da specs for da DC debugger kthxbye.
Related Links
-------------
SaturnDev
http://saturndev.homepage.com Charles dotys saturn dev site.
Emusaturn
http://saturn.sega-zone.com Me mate alucards saturn emu site.
Titan
http://user.tninet.se/~puc549i/ Tuneks saturn emu site.
Sega Exterme
http://www.litespeedcomputers.com/sx/ Plenty of SCD and Sat izos also home of STG
Unofficial Saturn
Video Game News, Reviews, and Walkthroughs - IGN |Chakan|s saturn site.
Sega Scans
http://scans.sega-zone.com/ Kanons scans site. Nice
Contact info:
-------------
Email : TyRaNiD@020.co.uk. For ya bug reporting, suggestions, gripes and general
lamentations.
EFnet : nick is [TyRaNiD]. U might be able to find me maybe.
Thought for the day..
---------------------
If i thought ud react like that i wouldnt have fucking bothered.
31337 Codin' by TyRaNiD (c) 2000
What this Program is:
--------------------
This program allows you to debug home-brew Sega Saturn code actually on ur saturn
with the help of a Pro Action Replay cart and a PC Commlink. Why u might want to
this is open to debate
Program Requirements:
---------------------
-=Needed for running and debugging code=-
IBM compatible PC, at least a 386 (to run da protected mode code)
A Sega Saturn. Any model should work from any country in theory.
Action replay cartridge. Also called a gameshark or Game Key depending
on where you live.
PC Commlink. A ISA board for your PC which allows connection to your Action Replay.
Without this the program hasnt got a hope of working. Sold mainly now for the psx
action replay but it should work on the saturn as well.
-=Needed to make ur own code=-
Saturn compiler/assembler and libraries.
Documents on the saturn.
I Cannot supply these so dont even ask about it.
BIG NOTE !!!!!!!
----------------
Im am hoping this version should support both datel and ems versions of the par by using
a command line switch or ini file setting. It will definetly work on the ems version
but i dont know if its the same for the datel. Only time will tell.
The program does not support the parallel port commlink developed by Freewing.
Mainly due to lack of documentation and desire. Anyway the PP version
would be hella slow. If of course someone really really wants the support then email
me and ill see what I can do.
This program is farrrrrrrrrr from finished yet. However I feel it has a
very long way to go as im surprised on how solid and easy using the UBC seems.
(lulled into a false sense of security no doubt)
USAGE:
To start debugging first run the program from dos or in a windows dos prompt,
using the following command line parameters.
satdis.exe filename [options]
filename is the name of the program to load. Required.
[options]
-lX X is the load address of the program in hex
-pX X is the port number where X is in 0x3X0 of the commlink port no.
-c Force coff format. Required if input file doesnt have .cof extension.
-d Set Datel PAR support. Without this switch set it would default to ems.
-? Displays help.
Notes on coff support.
I really hope it works with any coff file u want to throw at it but there might
be problems as ive not worked out the format from any official documentation. If
anyone know where I can find the exact specification for the coff format then plz
email me (a direct link please not just its on gccs web site or sommit).
EXAMPLES:
Loading a binary file.
satdis.exe program.bin -l0x6004000
Loading a coff file.
satdis.exe program.cof
NOTES
make sure the saturn is at the PARs main menu (or itll just hang
)(Well it could hang if ur port address is incorrect but doh..)
Oh if it does hang u should be able to CTRL+C it maybe..
It dont matter on the load address if u add a 0x infront or not itll still
work.
The load program can be in bin or coff format. The program will automatically assume
and program with a .cof extension is a coff file. If it does not have this extension
either rename the file or use the -c switch to force load as a coff file. If the
program thinks its loading a coff file when u specify a binary file it will certainly
crash the saturn and probably the debugger as well.
Config file format "satdebug.ini"
Comments r set by adding a # at the start of the line.
[options]
port=X. Allows u to specify the commlink port address. Can be overridden by passing
the -pX parameter.
datel=0|1. Sets datel support. Setting it to 0 means use da ems default and 1 means datel.
Debugger commands
-----------------
F1 : Memory to Screen Dump. Type in the address and press PGUP and PGDOWN to move
the memory dumping address by 256 byte blocks. Press ESC to exit.
Real time memory editor. Pick the range of values in the memory dump to edit
and press F1. Works like a standard hex editor and will modify values in real
time. Press TAB to switch between HEX mode and ASCII mode of editing. Press
ESC to quit and if u have modified any memory it will ask u to confirm the
write. If u press any other key than Y it will reset the memory as before
editing.
F2 : Write a byte to memory.
Type in address and value and it will write to memory
F3 : Write a word to memory.
Type in address and word value is written to memory. U dont have to specify
a word boundry address as it will be able to write it no matter what.
F4 : Write a long word to memory.
Type in an address and long word value. U dont have to specify a long word
boundry as it will write anyway.
F5 : Modify a register. Type the register number (given by the side of the reg dump)
and long value. New value will added and screen updated. It will also modify
the disasm if u have modfied the PC reg.
Oh and u cant modfify the VBR or the SP because this would screw the debugger up
Be very careful with this. U could crash the saturn by puttin in a bad value i.e.
a non long boundry reg when its the address for a long read.
Also u cant set the I in SR flags to any value greater than 0xE (or any value with
the LSB set). This stops u from killing the debugger by disabling the ability to
call the UBC int which is hardware set at priority 15. (why u would want to do
this in the first place is beyond me).
F6 : Dump memory block to a file. Type filename, start address and end
address. (Note. Memory will be read up to end addr -1 ).
F7 : Load file to memory. Type filename and start address. Precautions must be taken
as with any of the memory routines to ensure u dont overwrite important data.
Possibly only truly useful to quickly set a grafix screen to a know state.
F8 : Run program. Execution can be cancelled at any time by pressing a key.
F9 : Step into instruction. Executes 1 op and steps into a jump, branch.
F10 : Step over. If executed on a sub routine jump will skip over it.
Execution can be broken at any time by pressing a key. (note will destroy
the call back breakpoint).
UP,
DOWN,
PGUP,
PGDOWN: Moves the disassembly by a certain amount. Up and down cursors move the disasm by
1 instruction where pgup and down will move by 10 in the respective direction.
HOME : Resets the disassembly to the current instruction position.
Esc : Quit the program and remove the UBC handler. Will allow program to properly
execute at full speed.
ALT+D : Dump from register value. Put in the regsiter dump and it will do a memory dump
at the memory address pointed to by the register. R15 and VBR are included in this
and have the reg nos of 15 and 16 respectivly. Note registers which cannot be used
directly for memory addressing are not included. These being MACL,MACH and SR.
SPACE : Set a breakpoint on current instruction. Press again on instruction to remove.
ALT+L : Display current set breakpoints. When list produced allow setting and deleting
breakpoints without finding them in the disasm.
ALT+I : Inline Asm. Replaces the currently selected instruction.
ALT+U : Undos the previous inline asm. If the asm command has compiled wrong then just
use this command to reset the instruction.
ALT+R : Reset the current program to its original location. Reloads program so it might
take some time to restart.
'?' : Display online help. Not very descriptive just useful if uve forgotten the key u
need.
The memory routines above will only accept certain address ranges at the moment.
To be honest I dont think there are any other main areas which rnt included.
Mem Areas
---------
0x(2)0000000 -> 0x(2)0080000 : Boot ROM (R)
0x(2)0180000 -> 0x(2)0190000 : Backup RAM (R)
0x(2)0200000 -> 0x(2)0300000 : Lower Work RAM (R/W)
0x(2)5C00000 -> 0x(2)5C80000 : VDP1 VRAM (R/W)
0x(2)5E00000 -> 0x(2)5E80000 : VDP2 VRAM (R/W)
0x(2)5F00000 -> 0x(2)5F01000 : VDP2 CRAM (R/W)
0x(2)6000000 -> 0x(2)6100000 : Upper Work RAM (R/W)
Obviously R and W specify the access available for the memory. I have disable write access
to the BIOS and the backup RAM for obvious reason (well I hope they r obvious if not go sit
in a corner until u figure it out).
And also the (2) signifies a cache through address so u can either. However memory dumping
and writing is best done on non cached addresses as this will reflect the programs likely
data.
MORE BIG NOTES!!!!!!! And bugs hehe.
------------------------------------
Here are know problems with the underlying system. Mostly out of my control.
1) Program wont seem to execute delayed ops, interrupt protected instructions
i.e. lds etc, and will not work in areas where the SRs I mask is set to 15.
The debugger should return correctly at the first possible instruction available
will just not be able to debug those instuctions. Due to limitations in the UBC.
(Out of my control)
2) U can only load programs into the higher work ram, (Not really a problem
just a point to note). Will be remedied later on and allow loading to lower
as well.
3) If the size of the program + Start Address > 0x6040000 u will most likely
trash the UBC handler so try and make sure it dont. Ill move it to a better
position when its closer to finishing (basically when I know the rough total
code size of the bootloader). Im looking into setting the cache as 2 way and
storing the boot code into that. As long as ur program doesnt touch the cache
or its regs then in theory it might work well. And then it means that no matter
what u put in main memory it will never overwrite the UBC code. However u cant be
sure that sega demos etc will not adhere to this (or for that matter the sega libs
themselves)
4) I have not extensively tested the step over command so it might not work
inall cases. One case where it will not work is if a subroutine sets the SR
I mask to 15 and doesnt reset it till after the return. As no break would
occur on the instruction required it wouldnt work. However as this is blatent
bad coding practice so dont even go there Ricky.
5) U can only move the disasm within the upper work ram. This will only cause a problem
if the program jumps to a bios routine or a lower work ram address then u will not be able
to move the disasm window. It should still execute however.
6) I think if u modify the instruction u are currently on using the mem commands it will
still execute the original instruction. Only way round this I think would be to do a cache
purge. If u need to get across a branch modify the sr reg instead or sommit.
7) When using the memory dump if u try and move to a address outside the boundries
of the memory block ur in it will give u an invalid memory range error and cancel
the dump. Just irritating more than anything else.
8) Da inline asm needs to be tested properly. It could produce incorrect opcodes for valid
instructions etc or not pick up on errors. Who knows. There r some know perculiarities
in the assembler, such as all displacements immediates and branch addresses have to be
in hex without a 0x at the start. It is also best to avoid any spaces in the instruction
except at the beginning and between the opname and the first arg. The instruction formats
should be the same as in the sh2 manual. If in doubt read up on it.
Things to do now.
-----------------
1) Add some actual functionality to the program. i.e. memory dump, run over
sub routine, modify registers etc. Need to add some meat to the bootloader.
DONE SOME OF IT. Still need to add breakpoints etc.
2) Make the disasm and reg display better and include the missing sh2 regs
(VBR, GBR, MACL, MACH). Didn't add them out of lazyness.
DONE sorta. Still needs alot of work doing to it.
3) Display a text screen on the saturn when the bootloader has started just
to give an indication that everythings working correctly.(Better than no
change at all which occurs at the moment). Probably can reuse PARS predefined
font
Possibly wont bother. I might just add a load percentage bar on the pcside to indicate the program is loading.
4) Add a command to call back from inside a procedure (using the PR reg).
Should in theory be able to drop back to the last sub routine jump.
5) Hopefully dus might be able to port the debugger to linux to get an even
smaller target audience
6) Stuff, stuff and more stuff
Of course source level debugging far inda future (i.e. when I have time to rip off the DJGPP rhide debugger parser
, but at least it looks like it can be done with some hard work). DISCLAIMER.
----------
I do not provide any warranty on the security of your data on both ur saturn or
pc. While this program is extremely unlikely to damage anything there is always
the tiny chance of it doing so. If used responsibly and correctly nothing will
ever happen and thus I take no responsibilty for any damage cause through the use
of this proggy.
This program is not affilated or approved by either Sega or the manufacturer of
your Action Replay. Plz do not contact them with any problems u have as they will
invariably tell u to go f*#k urself.
Revision History:
----------------
0.7 Alpha : Added inbuilt inline asm capability. This should also make it usable in
my linux port without a copy of gas.
U should get error messages and everything now
Added support for datel and ems versions of the action replay in the same
program.
Unfortunately my commlink has died so i havent been able to test the program
to ensure it work
0.6 Alpha : Real time memory editing facility. HEX/ASCII editor format.
Added a program reset using the ALT + R command.
Finally added cached through memory addresses to the memory routines.
User cant set a non word boundry PC,PR or invalid SR reg value now.
Added a breakpoint editor to disable and set new breakpoints without
directly setting them in the disasm window.
Added easy to read bit by bit display of the SR reg (with bit names above).
Added a load progress bar on the pc side to indicate a program is indeed
loading.
Added a load file to memory command. NOTE F5 and F6 commands reversed.
Added an undo command for the last inline asm command. Useful if uve
have assembled the wrong instruction.
Rnt u lucky boyz and girls. I finally found how to delete filez so no more
files left about cause of gas
Finally sorted out the exact cause of a interface bug which has plagued me
since I wrote satonem in djgpp. Finally realised there must be a bug in the
gets() function and have written my own version to compensate.
Fixed a few things which might have caused problems in last release on
some systems (what systems? No one is using the program
)0.5 Alpha : A super cheat added. Inline assembler function as long as u have
the gcc assembler gas installed and setup in the config file
Config file system make it less hassle to pass the port number etc.
Preliminary coff file support so u dont even have to type in the
programs load/start address
Modified the program parameters slightly.
Made sure any configuration errors where possible occur before the loading
of the boot code. Means u shouldnt have to reset ur saturn if u type in an
invalid load file.
0.41 Alpha: Fixed a bug which caused most of the programs to crash. Its been there
since 0.2a. Whoops..
0.4 Alpha : First public release.
Added breakpoint commands and some pretty colours
.Added online help command. Wahey
0.3 Alpha : Added more commands. Write byte,word,long. Modify registers.
6/2/2000 Reordered the key assigns to make them slightly easier. (well so
dus dont have to press the shift key
).Added movement of disasm window. However its not much use at the mo.
Slightly adjusted the memory dump to allow u to go back through the address
range as well as forward.
Added a memory dump from register value command.
Fixed a (possible!) bug in the step over command which might occur with
a lds instruction modifying the pr regsiter. Now will only cause a step
over if the pr reg value is equal to the step pc reg.
0.2 Alpha : Added some proper commands.Step in, step over, mem to file, mem
3/2/2000 dump. Prelim key assigns (will change soon as they r crappy).
Made the interface abit neater.
Fixed some so called bugs/niggles which needed addressing first,
i.e. adding all the regs in the Read Reg command (fucked alot up
Added a new comm routine command which speeds up continuous
execution by approx 2000% hehe
0.1 Alpha : Initial test release.
2/2/2000
Program written using DJGPP C compiler (for DOS side) and GCC SH2 assembler
(for Saturn side).
This program is all my own code and so is copyrighted by me
It does not useany commercial libraries or code from either the pc or saturn side so im not
stepping on anybodies toes. The inline asm code is all my own work except for the
opcode table which is from the gcc binutils distribution. As this is under gpl
im not sure whether to release it but it isnt any actual code it just a set of
data which I used to avoid rewriting it. So I probably wont.
Thanx and Greetz:
-----------------
Thanx to dus for program testing and all the other benefits u bestow upon me
Thanx to PI for very helpful PAR documentation/sample code/and for showing that u
can even run programs on a saturn at all.
Thanx to Charles Doty for his saturn dev site.
Thanx to SEGA for producing a machine which seems fairly easy to make a debugger
for
Thanx to Datel UK for producing the PAR cart in the first place.
Thanx to MGhandi (www.kinox.org) for being the only emusite webmaster who
bothered to put up a mention of my debugger.
Greetz to Azuco, of A-Saturn.
Greetz to Alucard and Tunek.
Greetz to da Efnet crew (in no particular order) Gato, Vitor, Ame, Albx, Cowering,
Elmer, Design, Cokeman, The_Morph, IceMan2k, |Chakan|, |Ray|, Kanon, Haljordan, Crokk
and no doubt more i cant remember
Finally greetz to all the other ppl ive met along my way into obscurity
If u feel uve been missed out of the list then I dont care, haha.
Oh and skywalker of hitmen, give me da specs for da DC debugger kthxbye.
Related Links
-------------
SaturnDev
http://saturndev.homepage.com Charles dotys saturn dev site.
Emusaturn
http://saturn.sega-zone.com Me mate alucards saturn emu site.
Titan
http://user.tninet.se/~puc549i/ Tuneks saturn emu site.
Sega Exterme
http://www.litespeedcomputers.com/sx/ Plenty of SCD and Sat izos also home of STG
Unofficial Saturn
Video Game News, Reviews, and Walkthroughs - IGN |Chakan|s saturn site.
Sega Scans
http://scans.sega-zone.com/ Kanons scans site. Nice
Contact info:
-------------
Email : TyRaNiD@020.co.uk. For ya bug reporting, suggestions, gripes and general
lamentations.
EFnet : nick is [TyRaNiD]. U might be able to find me maybe.
Thought for the day..
---------------------
If i thought ud react like that i wouldnt have fucking bothered.
