[llvm] Improve TLI for Darwin libsystem_m functions

... to ensure we can vectorize these under -veclib=Darwin_libsystem_m

Author: jroelofs

llvm/include/llvm/Analysis/TargetLibraryInfo.h

@@ -409,12 +409,18 @@ class TargetLibraryInfo {
     default: break;
       // clang-format off
     case LibFunc_acos:         case LibFunc_acosf:      case LibFunc_acosl:
+    case LibFunc_acosh:        case LibFunc_acoshf:     case LibFunc_acoshl:
     case LibFunc_asin:         case LibFunc_asinf:      case LibFunc_asinl:
+    case LibFunc_asinh:        case LibFunc_asinhf:     case LibFunc_asinhl:
+    case LibFunc_atan2:        case LibFunc_atan2f:     case LibFunc_atan2l:
     case LibFunc_atan:         case LibFunc_atanf:      case LibFunc_atanl:
+    case LibFunc_atanh:        case LibFunc_atanhf:     case LibFunc_atanhl:
+    case LibFunc_cbrt:         case LibFunc_cbrtf:      case LibFunc_cbrtl:
     case LibFunc_ceil:         case LibFunc_ceilf:      case LibFunc_ceill:
     case LibFunc_copysign:     case LibFunc_copysignf:  case LibFunc_copysignl:
     case LibFunc_cos:          case LibFunc_cosf:       case LibFunc_cosl:
     case LibFunc_cosh:         case LibFunc_coshf:      case LibFunc_coshl:
+    case LibFunc_erf:          case LibFunc_erff:       case LibFunc_erfl:
     case LibFunc_exp2:         case LibFunc_exp2f:      case LibFunc_exp2l:
     case LibFunc_fabs:         case LibFunc_fabsf:      case LibFunc_fabsl:
     case LibFunc_floor:        case LibFunc_floorf:     case LibFunc_floorl:
@@ -425,6 +431,7 @@ class TargetLibraryInfo {
     case LibFunc_memcmp:       case LibFunc_bcmp:       case LibFunc_strcmp:
     case LibFunc_memcpy:       case LibFunc_memset:     case LibFunc_memmove:
     case LibFunc_nearbyint:    case LibFunc_nearbyintf: case LibFunc_nearbyintl:
+    case LibFunc_pow:          case LibFunc_powf:       case LibFunc_powl:
     case LibFunc_rint:         case LibFunc_rintf:      case LibFunc_rintl:
     case LibFunc_round:        case LibFunc_roundf:     case LibFunc_roundl:
     case LibFunc_sin:          case LibFunc_sinf:       case LibFunc_sinl:

llvm/test/Transforms/LoopVectorize/AArch64/vectorize-atan2-darwin.ll

@@ -0,0 +1,43 @@
+; RUN: opt -vector-library=Darwin_libsystem_m -passes=inject-tli-mappings,loop-vectorize -S < %s | FileCheck %s
+
+target datalayout = "e-m:o-i64:64-i128:128-n32:64-S128"
+target triple = "arm64-apple-ios"
+
+declare float @atan2f(float, float)
+
+define void @foo(ptr noalias nocapture %ptrA,
+                 ptr noalias nocapture readonly %ptrB,
+                 ptr noalias nocapture readonly %ptrC,
+                 i64 %size) {
+; CHECK-LABEL: @foo(
+; CHECK: call <4 x float> @_simd_atan2_f4(<4 x float>
+;
+entry:
+  br label %for.cond
+
+for.cond:                                         ; preds = %for.body, %entry
+  %indvars.iv = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %entry ]
+  %exitcond = icmp eq i64 %indvars.iv, %size
+  br i1 %exitcond, label %for.cond.cleanup, label %for.body
+
+for.body:                                         ; preds = %for.cond
+  %arrayidx = getelementptr inbounds float, ptr %ptrB, i64 %indvars.iv
+  %src1 = load float, ptr %arrayidx, align 4
+
+  %arrayidx2 = getelementptr inbounds float, ptr %ptrC, i64 %indvars.iv
+  %src2 = load float, ptr %arrayidx, align 4
+
+  %arrayidx3 = getelementptr inbounds float, ptr %ptrA, i64 %indvars.iv
+
+  %phase = call float @atan2f(float %src1, float %src2)
+
+  store float %phase, ptr %arrayidx3, align 4
+  %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
+  br label %for.cond, !llvm.loop !0
+
+for.cond.cleanup:                                 ; preds = %for.cond
+  ret void
+}
+
+!0 = distinct !{!0, !1}
+!1 = !{!"llvm.loop.vectorize.enable", i1 true}