[llvm] bdd1eba - [ARM] Add additional vecreduce float legalization test (NFC)

[Nikita Popov via llvm-commits]

Author: Nikita Popov
Date: 2020-09-12T22:40:39+02:00
New Revision: bdd1eba37b64e64c2d93d3e79223b5933d631447

URL: https://github.com/llvm/llvm-project/commit/bdd1eba37b64e64c2d93d3e79223b5933d631447
DIFF: https://github.com/llvm/llvm-project/commit/bdd1eba37b64e64c2d93d3e79223b5933d631447.diff

LOG: [ARM] Add additional vecreduce float legalization test (NFC)

Added: 
    

Modified: 
    llvm/test/CodeGen/ARM/vecreduce-fadd-legalization-soft-float.ll

Removed: 
    


################################################################################
diff  --git a/llvm/test/CodeGen/ARM/vecreduce-fadd-legalization-soft-float.ll b/llvm/test/CodeGen/ARM/vecreduce-fadd-legalization-soft-float.ll
index f3eeb11a17fd..164cfe1d8848 100644
--- a/llvm/test/CodeGen/ARM/vecreduce-fadd-legalization-soft-float.ll
+++ b/llvm/test/CodeGen/ARM/vecreduce-fadd-legalization-soft-float.ll
@@ -1,10 +1,49 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc < %s -mtriple=arm-none-eabi -mattr=-neon | FileCheck %s --check-prefix=CHECK
 
+declare half @llvm.experimental.vector.reduce.v2.fadd.f16.v4f16(half, <4 x half>)
 declare float @llvm.experimental.vector.reduce.v2.fadd.f32.v4f32(float, <4 x float>)
 declare double @llvm.experimental.vector.reduce.v2.fadd.f64.v2f64(double, <2 x double>)
 declare fp128 @llvm.experimental.vector.reduce.v2.fadd.f128.v2f128(fp128, <2 x fp128>)
 
+define half @test_v4f16(<4 x half> %a) nounwind {
+; CHECK-LABEL: test_v4f16:
+; CHECK:       @ %bb.0:
+; CHECK-NEXT:    .save {r4, r5, r6, r7, r8, lr}
+; CHECK-NEXT:    push {r4, r5, r6, r7, r8, lr}
+; CHECK-NEXT:    mov r7, #255
+; CHECK-NEXT:    mov r6, r0
+; CHECK-NEXT:    orr r7, r7, #65280
+; CHECK-NEXT:    mov r4, r3
+; CHECK-NEXT:    and r0, r1, r7
+; CHECK-NEXT:    mov r5, r2
+; CHECK-NEXT:    bl __aeabi_h2f
+; CHECK-NEXT:    mov r8, r0
+; CHECK-NEXT:    and r0, r4, r7
+; CHECK-NEXT:    bl __aeabi_h2f
+; CHECK-NEXT:    mov r4, r0
+; CHECK-NEXT:    and r0, r6, r7
+; CHECK-NEXT:    bl __aeabi_h2f
+; CHECK-NEXT:    mov r6, r0
+; CHECK-NEXT:    and r0, r5, r7
+; CHECK-NEXT:    bl __aeabi_h2f
+; CHECK-NEXT:    mov r1, r0
+; CHECK-NEXT:    mov r0, r6
+; CHECK-NEXT:    bl __aeabi_fadd
+; CHECK-NEXT:    mov r5, r0
+; CHECK-NEXT:    mov r0, r8
+; CHECK-NEXT:    mov r1, r4
+; CHECK-NEXT:    bl __aeabi_fadd
+; CHECK-NEXT:    mov r1, r0
+; CHECK-NEXT:    mov r0, r5
+; CHECK-NEXT:    bl __aeabi_fadd
+; CHECK-NEXT:    bl __aeabi_f2h
+; CHECK-NEXT:    pop {r4, r5, r6, r7, r8, lr}
+; CHECK-NEXT:    mov pc, lr
+  %b = call fast half @llvm.experimental.vector.reduce.v2.fadd.f16.v4f16(half 0.0, <4 x half> %a)
+  ret half %b
+}
+
 define float @test_v4f32(<4 x float> %a) nounwind {
 ; CHECK-LABEL: test_v4f32:
 ; CHECK:       @ %bb.0: