Amiga Hunk HUNK_RELOC32

Hi folks,

The relocation hunk has a number of the number of relocations.

The problem is each relocation has just 1 long word.

I figure that there would be 2.

Let's take this line of assembly:

280: JSR 0x600 (where 0x600 is, in the executable: 0x0... 0x600 is after relocation has been parsed)

1 would specify the offset to the long that needs changing (282, after JSR opcode) and another one to indicate where the long points to (0x600).

So why is there only one 32-bit long? What am I missing?

Jeremy.

[Toni Wilen]

http://eab.abime.net/showthread.php?t=78811 should answer it (and more).

[Leffmann]

Each list of relocations is preceeded by what section (what hunk in your file) it is pointing into, and after having allocated all sections, you only need to add the address of the referenced section to the relocation entry, and the relocation is complete.

[matthey]

Originally Posted by jeremysmith

The relocation hunk has a number of the number of relocations.

The problem is each relocation has just 1 long word.

I figure that there would be 2.

There are 2 longwords of data used in calculating each relocated address.

1) the longword data at the relocated offset (in the code)
2) the longword of the hunk base in memory (not known until the hunk memory is allocated by the scatter loader)

[copse]

I tried to make my wiki page as detailed as possible. There's Python "pseudo-code" taken from my disassembler.

Thanks copse. Actually, I wanted to know about the relocations as I'm writing a 68000 disassembler.

Now that I see you have written one, I can use that, so I don't have to worry about fixing mine!

I've got peasauce working now in Python.

Also thanks to everyone else for their help!

Cheers!

[Hedeon]

Try to make it PC-relative as much as possible is my advice:

JSR Label(pc) instead of JSR Label