[clang] [llvm] [AMDGPU][True16][MC][CodeGen] true16 for v_alignbyte_b32 (PR #119750)

[Brox Chen via llvm-commits]

================
@@ -3802,6 +3802,26 @@ def : FPMinCanonMaxPat<V_MINIMUMMAXIMUM_F16_fake16_e64, f16, DivergentBinFrag<fm
 def : FPMinCanonMaxPat<V_MAXIMUMMINIMUM_F16_fake16_e64, f16, DivergentBinFrag<fminimum>, fmaximum_oneuse>;
 }
 
+let True16Predicate = UseFakeTrue16Insts in
+def : GCNPat <
+(i32 (int_amdgcn_alignbyte (i32 (VOP3OpSelMods i32:$src0, i32:$src0_modifiers)),
----------------
broxigarchen wrote:

I tried this but tablegen complained about thte type of the src2 operand

https://github.com/llvm/llvm-project/pull/119750