AMDGPU: Use tablegen pattern for sendmsg intrinsics

Since this now emits a direct copy to m0, SIFixSGPRCopies has to
handle a physical register.

llvm-svn: 367593
diff --git a/llvm/lib/Target/AMDGPU/SOPInstructions.td b/llvm/lib/Target/AMDGPU/SOPInstructions.td
index 1a57509..58b0c4b 100644
--- a/llvm/lib/Target/AMDGPU/SOPInstructions.td
+++ b/llvm/lib/Target/AMDGPU/SOPInstructions.td
@@ -1110,12 +1110,11 @@
 let Uses = [EXEC, M0] in {
 // FIXME: Should this be mayLoad+mayStore?
 def S_SENDMSG : SOPP <0x00000010, (ins SendMsgImm:$simm16), "s_sendmsg $simm16",
-  [(AMDGPUsendmsg (i32 imm:$simm16))]
->;
+  [(int_amdgcn_s_sendmsg (i32 imm:$simm16), M0)]>;
 
 def S_SENDMSGHALT : SOPP <0x00000011, (ins SendMsgImm:$simm16), "s_sendmsghalt $simm16",
-  [(AMDGPUsendmsghalt (i32 imm:$simm16))]
->;
+  [(int_amdgcn_s_sendmsghalt (i32 imm:$simm16), M0)]>;
+
 } // End Uses = [EXEC, M0]
 
 def S_TRAP : SOPP <0x00000012, (ins i16imm:$simm16), "s_trap $simm16"> {