[llvm] [AMDGPU][GFX13] MC support for indexed VIMAGE/VSAMPLE resources (PR #223429)

[via llvm-commits]

================
@@ -269,6 +269,16 @@ static DecodeStatus decodeRsrcReg128(MCInst &Inst, unsigned Imm,
   return decodeRsrcRegOp(Inst, Imm, 0, Decoder, OpWidth);
 }
 
+static DecodeStatus decodeRsrcReg256(MCInst &Inst, unsigned Imm,
+                                     uint64_t /* Addr */,
+                                     const MCDisassembler *Decoder) {
+  unsigned OpWidth = 32;
+  // 0-127: Uniform-direct resource in SGPRs (SReg_256).
+  if (Imm < 128)
----------------
sstipano wrote:

You are right, but this just sets the width of the operand. Everything else is already handled later.

https://github.com/llvm/llvm-project/pull/223429