Printable View
Hey Monkey could you give me a little rundown on what pi dma transfers are (assuming you have time, since you are a busy person).
When you say this is bad do you mean in relivence to Monkey64s possible development, or simply another obsticle that will take time to get through?
Would be a shame to lose Nintendo roms. Anyways keep up the good work and thanks for your time!
Just a difficult thing to get though is all.Quote:
Originally Posted by drquack02
PI transfer is when the N64 DMA (Direct Memory Access) or Copies a large chunk of data from Rom to Ram/RDRAM. It's a fast way to move sizable chunks of data then to do byte by byte.
Mario64 does a bunch of transfers properly but 3 specific transfers become invalid. I am not sure if it's with in the block of code that made it invalid or if it was invalid before being passed to the block of code.
It's not really complex but I need more time to probe for the bug.
Thanks for the explaination!
Well I guess half the solution is finding the problem. Good luck, and keep us posted!
Wow! This is turning out to be awesome! Any idea on when you think Mario Kart 64 will be working?
Jus a question PSmonkey, i admire how you dont get demmoralised when you get stuck with a bug with ANY of your programming, how do u keep going?
I'm wondering that myself...I've taken several hours in my coding (and I'm coding much simpler things) trying to fix little nitpicky bugs, and after all's said and done, I usually give up for a couple of weeks at a time. How do you manage?
The love of the "game" no pun intended...
everyone has their passions.
I go all ninja/comando style on the bug. Running at it full force and testing all angles till i drop dead of sleep deprivation. :P
Somehow, commando style doesn't fit in that sentence.Quote:
Originally Posted by PSmonkey
Translation:
I sit naked on my chair. Playing Project 64 all night long until girlfriend call me bed ;)
Quote:
Originally Posted by Psycho77
lol. Not close tho. I do reverse debuging on the psp & pj64 at the same time. I find the entry point of a code block. check it on the psp to find what brached to it then keep working backwards.
So far I've traced it back to a tlb instruction. I think I went far back enough to find the different memory addresses for the different loads.
It goes (best i can figure out)
ERET (not sure why, I might have this wrong)
Increase Stack pointer
Store registers to stack
Load branch addr (the start of our function)
Load Addr for PI values <- Different after first 20 transfers
Load RDRAM_ADDR to A2
Load ROM_ADDR to A1
Load WR_LEN to A3
Branch to Addr (Loc for pif transfer)
Increase Stack Pointer
Store Registers to stack
Do Some Work
Load RDRAM_ADDR from Stack
Translate RDRAM ADDR
Store RDRAM_ADDR to PI_RDRAM_ADDR Reg
Load ROM_ADDR
TRanslate ROM_ADDR
Store ROM_ADDR to PI_ROM_ADDR
Load WR_LEN
Add -1 (odd it does this vs calling sub instruction)
Store WR_LEN to PI_WR_LEN
PI_DMA_WRITE triggered by previous instruction
...
...
Branch ...
... = Cleanup but I am not fully sure.
That is what I figured out so far in a nut shell. Atleast now I have an addr for where it gets the values from. So with more tracing. I can find out hopefuly what is writing to that specific address.
:( Hopefuly mario will be going soon. I dont want to have to reverse enginear the entire game in asm just to find the bug. :P