[PATCH] D136944: [AMDGPU] Enable `s_sendmsg_rtn` selection with `+gfx11-insts`

[Pierre van Houtryve via Phabricator via llvm-commits]

Pierre-vh marked an inline comment as done.
Pierre-vh added inline comments.


================
Comment at: llvm/lib/Target/AMDGPU/SIInstrInfo.cpp:8002-8004
+    uint16_t NMCOp = AMDGPU::getMCOpcode(Opcode, SIEncodingFamily::GFX11);
+    if (NMCOp != (uint16_t)-1)
+      MCOp = NMCOp;
----------------
arsenm wrote:
> I'm surprised this is necessary. How was this missing before? I thought the encoding didn't change in 11?
Without it, I run into ``Opcode < NumOpcodes && "Invalid opcode!"` without it because that function ends up returning -1 


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D136944/new/

https://reviews.llvm.org/D136944