[ARM][SDAG] Add llvm.lround half promotion.

[davemgreen]

Similar to #161088, add llvm.lround and llvm.llround promotion.

[llvmbot]

@llvm/pr-subscribers-llvm-selectiondag

@llvm/pr-subscribers-backend-arm

Author: David Green (davemgreen)

Changes

Similar to #161088, add llvm.lround and llvm.llround promotion.

---

Full diff: https://github.com/llvm/llvm-project/pull/164235.diff

4 Files Affected:

• (modified) llvm/lib/CodeGen/SelectionDAG/LegalizeFloatTypes.cpp (+2)
• (modified) llvm/lib/Target/ARM/ARMISelLowering.cpp (+1)
• (modified) llvm/test/CodeGen/ARM/llround-conv.ll (+9)
• (modified) llvm/test/CodeGen/ARM/lround-conv.ll (+33-10)

diff --git a/llvm/lib/CodeGen/SelectionDAG/LegalizeFloatTypes.cpp b/llvm/lib/CodeGen/SelectionDAG/LegalizeFloatTypes.cpp
index 437d0f4654096..bf1abfe50327e 100644
--- a/llvm/lib/CodeGen/SelectionDAG/LegalizeFloatTypes.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/LegalizeFloatTypes.cpp
@@ -3765,6 +3765,8 @@ bool DAGTypeLegalizer::SoftPromoteHalfOperand(SDNode *N, unsigned OpNo) {
   case ISD::FP_TO_UINT:
   case ISD::LRINT:
   case ISD::LLRINT:
+  case ISD::LROUND:
+  case ISD::LLROUND:
     Res = SoftPromoteHalfOp_Op0WithStrict(N);
     break;
   case ISD::FP_TO_SINT_SAT:
diff --git a/llvm/lib/Target/ARM/ARMISelLowering.cpp b/llvm/lib/Target/ARM/ARMISelLowering.cpp
index 35e1127000b8a..8aa5f88b050db 100644
--- a/llvm/lib/Target/ARM/ARMISelLowering.cpp
+++ b/llvm/lib/Target/ARM/ARMISelLowering.cpp
@@ -1353,6 +1353,7 @@ ARMTargetLowering::ARMTargetLowering(const TargetMachine &TM_,
     setOperationAction(ISD::FLOG10, MVT::f16, Promote);
     setOperationAction(ISD::FLOG2, MVT::f16, Promote);
     setOperationAction(ISD::LRINT, MVT::f16, Expand);
+    setOperationAction(ISD::LROUND, MVT::f16, Expand);
 
     setOperationAction(ISD::FROUND, MVT::f16, Legal);
     setOperationAction(ISD::FROUNDEVEN, MVT::f16, Legal);
diff --git a/llvm/test/CodeGen/ARM/llround-conv.ll b/llvm/test/CodeGen/ARM/llround-conv.ll
index f734db89af2f7..20fe2728d26d5 100644
--- a/llvm/test/CodeGen/ARM/llround-conv.ll
+++ b/llvm/test/CodeGen/ARM/llround-conv.ll
@@ -1,6 +1,7 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
 ; RUN: llc < %s -mtriple=armv7-none-eabi -float-abi=soft | FileCheck %s --check-prefixes=CHECK,CHECK-SOFT
 ; RUN: llc < %s -mtriple=armv7-none-eabihf -mattr=+vfp2 -float-abi=hard | FileCheck %s --check-prefixes=CHECK,CHECK-NOFP16
+; RUN: llc < %s -mtriple=armv8-none-eabihf -mattr=+fp-armv8 -float-abi=hard | FileCheck %s --check-prefixes=CHECK,CHECK-FPv8
 ; RUN: llc < %s -mtriple=armv8-none-eabihf -mattr=+fp-armv8,+fullfp16 -float-abi=hard | FileCheck %s --check-prefixes=CHECK,CHECK-FP16
 
 define i64 @testmsxh_builtin(half %x) {
@@ -22,6 +23,14 @@ define i64 @testmsxh_builtin(half %x) {
 ; CHECK-NOFP16-NEXT:    bl llroundf
 ; CHECK-NOFP16-NEXT:    pop {r11, pc}
 ;
+; CHECK-FPv8-LABEL: testmsxh_builtin:
+; CHECK-FPv8:       @ %bb.0: @ %entry
+; CHECK-FPv8-NEXT:    .save {r11, lr}
+; CHECK-FPv8-NEXT:    push {r11, lr}
+; CHECK-FPv8-NEXT:    vcvtb.f32.f16 s0, s0
+; CHECK-FPv8-NEXT:    bl llroundf
+; CHECK-FPv8-NEXT:    pop {r11, pc}
+;
 ; CHECK-FP16-LABEL: testmsxh_builtin:
 ; CHECK-FP16:       @ %bb.0: @ %entry
 ; CHECK-FP16-NEXT:    .save {r11, lr}
diff --git a/llvm/test/CodeGen/ARM/lround-conv.ll b/llvm/test/CodeGen/ARM/lround-conv.ll
index 03f7a0d7a44c4..7466bcb982efb 100644
--- a/llvm/test/CodeGen/ARM/lround-conv.ll
+++ b/llvm/test/CodeGen/ARM/lround-conv.ll
@@ -4,11 +4,39 @@
 ; RUN: llc < %s -mtriple=armv8-none-eabihf -mattr=+fp-armv8 -float-abi=hard | FileCheck %s --check-prefixes=CHECK,CHECK-FPv8
 ; RUN: llc < %s -mtriple=armv8-none-eabihf -mattr=+fp-armv8,+fullfp16 -float-abi=hard | FileCheck %s --check-prefixes=CHECK,CHECK-FP16
 
-;define i32 @testmswh_builtin(half %x) {
-;entry:
-;  %0 = tail call i32 @llvm.lround.i32.f16(half %x)
-;  ret i32 %0
-;}
+define i32 @testmswh_builtin(half %x) {
+; CHECK-SOFT-LABEL: testmswh_builtin:
+; CHECK-SOFT:       @ %bb.0: @ %entry
+; CHECK-SOFT-NEXT:    .save {r11, lr}
+; CHECK-SOFT-NEXT:    push {r11, lr}
+; CHECK-SOFT-NEXT:    bl __aeabi_h2f
+; CHECK-SOFT-NEXT:    pop {r11, lr}
+; CHECK-SOFT-NEXT:    b lroundf
+;
+; CHECK-NOFP16-LABEL: testmswh_builtin:
+; CHECK-NOFP16:       @ %bb.0: @ %entry
+; CHECK-NOFP16-NEXT:    .save {r11, lr}
+; CHECK-NOFP16-NEXT:    push {r11, lr}
+; CHECK-NOFP16-NEXT:    vmov r0, s0
+; CHECK-NOFP16-NEXT:    bl __aeabi_h2f
+; CHECK-NOFP16-NEXT:    vmov s0, r0
+; CHECK-NOFP16-NEXT:    pop {r11, lr}
+; CHECK-NOFP16-NEXT:    b lroundf
+;
+; CHECK-FPv8-LABEL: testmswh_builtin:
+; CHECK-FPv8:       @ %bb.0: @ %entry
+; CHECK-FPv8-NEXT:    vcvtb.f32.f16 s0, s0
+; CHECK-FPv8-NEXT:    b lroundf
+;
+; CHECK-FP16-LABEL: testmswh_builtin:
+; CHECK-FP16:       @ %bb.0: @ %entry
+; CHECK-FP16-NEXT:    vcvta.s32.f16 s0, s0
+; CHECK-FP16-NEXT:    vmov r0, s0
+; CHECK-FP16-NEXT:    bx lr
+entry:
+  %0 = tail call i32 @llvm.lround.i32.f16(half %x)
+  ret i32 %0
+}
 
 define i32 @testmsws_builtin(float %x) {
 ; CHECK-LABEL: testmsws_builtin:
@@ -40,8 +68,3 @@ entry:
   ret i32 %0
 }
 
-;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
-; CHECK-FP16: {{.*}}
-; CHECK-FPv8: {{.*}}
-; CHECK-NOFP16: {{.*}}
-; CHECK-SOFT: {{.*}}

[efriedma-quic]

LGTM