* Lots of audio matches
Fully matched heap.c, lots of load.c and some stuff in seqplayer.c
Matched heap.c's BSS segement and moved some BSS stuff out of
load.c into data.c, since that where it appears to belong
according to SM64.
Signed-off-by: Taggerung <tyler.taggerung@email.com>
* Add new utility to calculate packed dl end and vertex count
Requires changing a bunch of actual references to packed dl lists
with raw numbers. Which feels extremely bad but it doesn't match
otherwise.
Signed-off-by: Taggerung <tyler.taggerung@email.com>
* Hard code array lenghts in appropriate headers
Use macros to generate the vertex counts and packed dl end points
Heavily dependent on the array lengths being accurate. If they
aren't stuff will break.
Signed-off-by: Taggerung <tyler.taggerung@email.com>
* Match func_800C8920, func_800C8AE4, and..
Introduce struct Unk_800E9F7C and cleanup existing code
* Match func_800C40F0, add the likely other pos vars in the unk struct
* func_800C8CCC
* Update building of packed_dl files
* Expose course_data symbols to linker.
* Remove undefined_syms for course data
* Update ctx_includes
Signed-off-by: Taggerung <tyler.taggerung@email.com>
Also got rid of a couple asm files for functions that were matched
a long, long time ago (thank you ItzSwirlz for finding them)
Signed-off-by: Taggerung <tyler.taggerung@email.com>
* Migrated data and rodata for code_80091750
And it even matches when built!
Minor annoyances:
There are some EUC-JP strings in `func_800A6E94` and `func_800A69C8`
that are currently in byte form. The asm files don't get put
through `icnov` so we can't have them as strings until they're
parent functions are matched.
`func_8009F5E0` has the symbol `D_800E7594` replaced with the
actual number `0x800e7594`. The symbol in question is almost
certainly fake, but I can't figure out what its meant to be.
So I've hardcoded it for the time being to make the migration
possible.
* Fully match all the NEEDS_RODATA functions
Signed-off-by: Taggerung <tyler.taggerung@email.com>
I have historically made prototypes for argument-less functions
with just plain `()` for the argument list. This has caused some
minor pains when using `m2c` and is, pendatically, incorrect.
So I have changed all argument-less function protoypes to instead
have `(void)` as the argument list. This required a couple small
fixes here and there.
I really, really don't like a couple of the matches because of
the handling of the `unk_80` member of the object struct. But
if it matches it matches.
Signed-off-by: Taggerung <tyler.taggerung@email.com>
* All work done in audio
* Migrate audio/port_eu data
* Refereshed m2c
* Got the bss segment matching too
Signed-off-by: Taggerung <tyler.taggerung@email.com>
