[Dwarf-discuss] PROPOSAL: add support for bundled instructions (e.g., Itanium)
Michael Eager
eager@eagercon.com
Wed May 2 19:19:15 GMT 2007
Jim Blandy wrote:
> If Matthew agrees with me that we can handle his very variable
> bundles, then I like the substance of this proposal. There are two
> changes I'd like to see, though:
>
> First, since the imaginary table now has a new slot number column, we
> need a standard opcode for dumping fixed values into it. And perhaps
> DW_LNS_copy should clear the slot number?
>
> Second, instead of the proposed description, I'd like to simply slice
> the available range of special opcodes into a three-dimensional solid,
> instead of a two-dimensional matrix, where the first dimension is the
> line advance, the second is the address advance, and the third is the
> slot advance. By default, the "slot range" would be 1, giving us the
> two-dimensional special opcode matrices we have today, and the "slot
> base" would default to zero, so the opcodes would have no effect on
> the slot number. The slot field would overflow into the address
> advance, before being multiplied by the minimum instruction length.
Let's move discussion of proposals to the Dwarf Committee
mailing list: dwarf-workgroup at lists.dwarfstd.org. That way
all discussions will be in one place.
--
Michael Eager eager at eagercon.com
1960 Park Blvd., Palo Alto, CA 94306 650-325-8077
More information about the Dwarf-discuss
mailing list