The full debugger source has been commented, a total of just over 2500 lines of code and comments. Of course there are other things that could be done, such as making the code position independent (as it probably was originally), so that it could be linked at arbitrary addresses. However, let's be honest, it is unlikely anyone is actually going to use the results, so it is probably a wasted effort. What I will do is to write all my findings up as a document. I hope that, with the blessing of the Blue Sky Rangers, this will include the full, commented source code.
In the mean time, here are the contents of the debugger's workspace RAM as a taster:
; ------------------------------------------------------------------------------
; Debugger workspace RAM
L_D279:
DECLE 0 ; $d279 - Keystroke debounce count
DECLE 0 ; $d27a - Last right controller input
DECLE 0 ; $d27b - Last left controller input
DECLE 0 ; $d27c - Last valid controller cmd
DECLE 0 ; $d27d - SDBD flag
BIDECLE 0 ; $d27e - cached R0 register
BIDECLE 0 ; $d280 - cached R1 register
BIDECLE 0 ; $d282 - cached R2 register
BIDECLE 0 ; $d284 - cached R3 register
BIDECLE 0 ; $d286 - cached R4 register
BIDECLE 0 ; $d288 - cached R5 register
BIDECLE 0 ; $d28a - cached R6 register (SP)
BIDECLE 0 ; $d28c - cached R7 register (PC)
BIDECLE 0 ; $d28e - cached Sign flag
BIDECLE 0 ; $d290 - cached Zerp flag
BIDECLE 0 ; $d292 - cached Overflow flag
BIDECLE 0 ; $d294 - cached Carry flag
BIDECLE 0 ; $d296 - breakpoint A active
BIDECLE 0 ; $d298 - breakpoint B active
BIDECLE 0 ; $d29a - breakpoint C active
BIDECLE 0 ; $d29c - breakpoint D active
BIDECLE 0 ; $d29e - breakpoint E active
BIDECLE 0 ; $d2a0 - breakpoint F active
BIDECLE 0 ; $d2a2 - breakpoint G active
BIDECLE 0 ; $d2a4 - breakpoint H active
BIDECLE $5000 ; $d2a6 - memory inspector 1 address
BIDECLE $6000 ; $d2a8 - memory inspector 2 address
DECLE 0 ; $d2aa - focussed field / breakpoint
DECLE 0 ; $d2ab - subroutine register
DECLE 0 ; $d2ac - breakpoints active
;
L_D2AD: ;
DECLE 0, 0, 0, 0, 0, 0, 0 ; $d2ad - instruction cache
DECLE 0 ; $d2b4 - instruction running flag
BIDECLE 0 ; $d2b5 - subroutine R6 (SP)
DECLE 0 ; $d2b7 - debugger mode
; - 0 = single step
; - 1 = trace
; - 2 = trace to user instruction
; - 3 = trace to subroutine exit
;
REPEAT 30 ; screen BACKTAB cache
BIDECLE 0, 0, 0, 0 ;
ENDR ;
Speaking of findings, the only other snippet I discovered going through the last of the source is that instructions prefixed with SDBD are treated as a single compound opcode, rather than two distinct instructions. The existence of the compound SDBD instruction is then highlighted in the disassembly by doubling the # or @ symbol at the start of the operand. Again, a nice usability tweak.
In the absence of any advice from Rick or the other BSRs, it does seem as though the two memory inspectors behave the same. There is no great mystery there after all, apparently.
So, looking back at the questions we set at the beginning of this RetroChallenge, I think we now have the answers...
- Is Rick's debugger in WCB? - Yes
- Is it complete? - Yes, it seems to be
- Can it be isolated, reverse-engineered and tinkered with? - Hell yes!
Well that is perhaps one for the next post...
Comments
Post a Comment