TableGen machine code emitter backend

In my travels through the TableGen backends, I just learned that the machine code emitter can build instructions no longer than 8 bytes. Is there any interest in extending this to, say, 16 bytes?

James Malloy added support for this recently. I’m using it in a downstream back-end today, as are others.
(Similarly, Daniel Sanders extended the FixedLenDecoderEmitter to support instruction words > 64 bits).

commit 60aadd19cbffc3793476a14d2e3529214119e2f5
Author: James Molloy

[CodeEmitter] Support instruction widths > 64 bits

Some VLIW instruction sets are Very Long Indeed. Using uint64_t constricts the Inst encoding to 64 bits (naturally).

This change switches CodeEmitter to a mode that uses APInts when Inst’s bitwidth is > 64 bits (NFC for existing targets).

When Inst.BitWidth > 64 the prototype changes to:

void TargetMCCodeEmitter::getBinaryCodeForInstr(const MCInst &MI,
SmallVectorImpl &Fixups,
APInt &Inst,
APInt &Scratch,
const MCSubtargetInfo &STI);

The Inst parameter returns the encoded instruction, the Scratch parameter is used internally for manipulating operands and is exposed so that the underlying storage can be reused between calls to getBinaryCodeForInstr. The goal is to elide any APInt constructions that we can.

Similarly the operand encoding prototype changes to:

getMachineOpValue(const MCInst &MI, const MCOperand &MO, APInt &op, SmallVectorImpl &Fixups, const MCSubtargetInfo &STI);

That is, the operand is passed by reference as APInt rather than returned as uint64_t.

To reiterate, this APInt mode is enabled only when Inst.BitWidth > 64, so this change is NFC for existing targets.

llvm-svn: 371928

Excellent! I didn't look hard enough to notice that APInt's were being used.