AMDGPU: Cleanup extract_subvector actions (NFC)

[arsenm]

The base AMDGPUISelLowering was setting custom action on 16-bit
vector types, but also set in SIISelLowering.

[arsenm]

• AMDGPU: Cleanup extract_subvector actions (NFC) #101454 👈
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

Join @arsenm and the rest of your teammates on Graphite

[llvmbot]

@llvm/pr-subscribers-backend-amdgpu

Author: Matt Arsenault (arsenm)

Changes

The base AMDGPUISelLowering was setting custom action on 16-bit
vector types, but also set in SIISelLowering.

---

Full diff: https://github.com/llvm/llvm-project/pull/101454.diff

2 Files Affected:

• (modified) llvm/lib/Target/AMDGPU/AMDGPUISelLowering.cpp (+6-10)
• (modified) llvm/lib/Target/AMDGPU/SIISelLowering.cpp (+1-1)

diff --git a/llvm/lib/Target/AMDGPU/AMDGPUISelLowering.cpp b/llvm/lib/Target/AMDGPU/AMDGPUISelLowering.cpp
index 2ad91de566323..862b5c7e3e3d7 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUISelLowering.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPUISelLowering.cpp
@@ -444,19 +444,15 @@ AMDGPUTargetLowering::AMDGPUTargetLowering(const TargetMachine &TM,
                       MVT::v11i32, MVT::v11f32, MVT::v12i32, MVT::v12f32},
                      Custom);
 
-  // FIXME: Why is v8f16/v8bf16 missing?
   setOperationAction(
       ISD::EXTRACT_SUBVECTOR,
-      {MVT::v2f16,  MVT::v2bf16, MVT::v2i16,  MVT::v4f16,  MVT::v4bf16,
-       MVT::v4i16,  MVT::v2f32,  MVT::v2i32,  MVT::v3f32,  MVT::v3i32,
-       MVT::v4f32,  MVT::v4i32,  MVT::v5f32,  MVT::v5i32,  MVT::v6f32,
-       MVT::v6i32,  MVT::v7f32,  MVT::v7i32,  MVT::v8f32,  MVT::v8i32,
-       MVT::v9f32,  MVT::v9i32,  MVT::v10i32, MVT::v10f32, MVT::v11i32,
-       MVT::v11f32, MVT::v12i32, MVT::v12f32, MVT::v16f16, MVT::v16bf16,
-       MVT::v16i16, MVT::v16f32, MVT::v16i32, MVT::v32f32, MVT::v32i32,
+      {MVT::v2f32,  MVT::v2i32,  MVT::v3f32,  MVT::v3i32,  MVT::v4f32,
+       MVT::v4i32,  MVT::v5f32,  MVT::v5i32,  MVT::v6f32,  MVT::v6i32,
+       MVT::v7f32,  MVT::v7i32,  MVT::v8f32,  MVT::v8i32,  MVT::v9f32,
+       MVT::v9i32,  MVT::v10i32, MVT::v10f32, MVT::v11i32, MVT::v11f32,
+       MVT::v12i32, MVT::v12f32, MVT::v16i32, MVT::v32f32, MVT::v32i32,
        MVT::v2f64,  MVT::v2i64,  MVT::v3f64,  MVT::v3i64,  MVT::v4f64,
-       MVT::v4i64,  MVT::v8f64,  MVT::v8i64,  MVT::v16f64, MVT::v16i64,
-       MVT::v32i16, MVT::v32f16, MVT::v32bf16},
+       MVT::v4i64,  MVT::v8f64,  MVT::v8i64,  MVT::v16f64, MVT::v16i64},
       Custom);
 
   setOperationAction(ISD::FP16_TO_FP, MVT::f64, Expand);
diff --git a/llvm/lib/Target/AMDGPU/SIISelLowering.cpp b/llvm/lib/Target/AMDGPU/SIISelLowering.cpp
index 8a811f7a7c02d..d5d7a2522a09b 100644
--- a/llvm/lib/Target/AMDGPU/SIISelLowering.cpp
+++ b/llvm/lib/Target/AMDGPU/SIISelLowering.cpp
@@ -629,10 +629,10 @@ SITargetLowering::SITargetLowering(const TargetMachine &TM,
         case ISD::EXTRACT_VECTOR_ELT:
         case ISD::INSERT_VECTOR_ELT:
         case ISD::INSERT_SUBVECTOR:
-        case ISD::EXTRACT_SUBVECTOR:
         case ISD::SCALAR_TO_VECTOR:
         case ISD::IS_FPCLASS:
           break;
+        case ISD::EXTRACT_SUBVECTOR:
         case ISD::CONCAT_VECTORS:
           setOperationAction(Op, VT, Custom);
           break;

[Pierre-vh]

Looks like a NFC, if it indeed is, please mark it as such