[Prev][Next][Index][Thread]
A92: A STRANGE AND VERY LUCKY DISCOVER...
Hello !
First, I want to tell you I have a 1.11 ROM.
Here are the facts : Yesterday, I was playing with the tiger emulator
and the excellent db92 debbuger.
Here is what I made :
Well, I run db92 from my program PCT98, there isn't a test program.
So the program point to 0x0, and print a lot of 'ori' instructions.
I tell me 'well, ben, you could go in the ROM, for fun'.
So I press [G] (Go function), and enter the address 420000. Why
420000 could you wonder ? Why not I would answer you ?
So I go to this offset, and press [S] (Set PC function)
There I trace it instruction by instruction. After less than around 10-20
instruction, I see with astonishment that I am returned in my program
PCTOOLS !!!!! I have run db92 from pctools, and I am now come back
in PCTools with db92, and I trace it !!
Funny, isn't it ?
So after I think that if I ran db92 from the home, It would come
back in the ROM, and I could trace it ! And it seems to work !!!!
So I decide to know how and Why it made that, and he is what I found :
In the 420000 offset, there is (with my comments) :
move.l $#FFE0(a6),a7 ;it put for the next rts the address
;#$01F714. that is at the end of the RAM,
;it must be a Fargo function
move.l $#FFDC(a6),a7 ;it is the file address from I run db92 I saw
jsr 04331964 ;a very short function
move.w #$0001,d0
bra #$004200E4
...
movem.l #3C38(a6),d2/d5??? ;there may be more registers, but db92
;had no more space
unlnk a6
rts ;rts, might go to the fargo function
And after you are in the program which called db92 !!!
(When you want to exit db92, it first make a stack corrupted, and then
a illegal instruction, but it is normal...)
So We could explore the ROM, isn't it ? But only 1.11 ROM will
be able to test that I think.
(we can't use a test program, since db92 check if the program is
terminated. But we could patch db92 to never says 'program
terminated' and so go back in the ROM, without crashing the calc.
Well, it was funny...
Benoit SCHERRER
Follow-Ups: