Use fm's new matching features for x86 register matching.

[ltratt]

Instead of using concrete register names, one can refer to a class of registers e.g. r.8 is all 8-bit registers. To match a register name but ignore its value uses r.8._: to match a register name use r.8.x. Suffixes: must be unique (i.e. r.8.x and r.8.y must refer to different registers); and they refer to the "underlying" register (e.g. r.8.x and r.64.x might match al and RAX which is considered the same register, but will fail to match against al and RBX).

To make this concrete these asserts all succeed:

let fmm = fmatcher("r.8.x r.8.y r.64.x");
assert!(fmm.matches("al bl rax").is_ok());
assert!(fmm.matches("al al rax").is_err());
assert!(fmm.matches("al bl rbx").is_err());
assert!(fmm.matches("al bl eax").is_err());

This commit is unfinished in the sense that not all the tests are ported: this is an RFC in its current state. Needs softdevteam/fm#50.

[vext01]

The user-observable functionality looks great, but I don't feel like I've understood name_matching_processor() sufficiently to do more than a superficial review (I've often struggled with understanding name matchers in general, but I usually manage to muddle through).

If you are OK with that, then we can proceed.

[ptersilie]

e.g. r.8.x and r.64.x might match al and RAX which is considered the same register, but will fail to match against al and RBX

Does this mean that r.8.x and r.64.x will only match if the register name is the same, even if the sizes are different. So the x part will bind to the A part of AL, RAX, etc?

[vext01]

My understanding is that r.8.x would never match a r.64.x. So al and rax are treated as distinct. I think that's the way we'd want it anyway.

[ptersilie]

My understanding is that r.8.x would never match a r.64.x. So al and rax are treated as distinct. I think that's the way we'd want it anyway.

Yes, they are distinct, but the x binds all other sizes to the RAX register. So r.16.x needs to be ax, and r.32.x needs to be eax. They can't suddenly be ebx.

[ltratt]

Does this mean that r.8.x and r.64.x will only match if the register name is the same, even if the sizes are different. So the x part will bind to the A part of AL, RAX, etc?

Yes, though it's not done by text but by the underlying register: so if r.8.x is al then r.64.x can only match rax (and not rbx etc).

[ltratt]

Ported all remaining tests (including floating point register support) in 8d4a343. Hopefully ready for squashing!

[vext01]

If @ptersilie is OK with this, I think I am.

[ltratt]

@ptersilie OK to squash?

[ptersilie]

Please squash.

[ltratt]

Squashed.