[llvm] [ConstantantFolding] Add support for `sinh` and `cosh` intrinsics constant folding (PR #132671)

[Matt Arsenault via llvm-commits]

================
@@ -0,0 +1,36 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
+; RUN: opt -S -passes=early-cse < %s | FileCheck %s
+
+define double @test_sinh() {
+; CHECK-LABEL: define double @test_sinh() {
+; CHECK-NEXT:    ret double 0.000000e+00
+;
+  %result = call double @llvm.sinh.f64(double 0.0)
+  ret double %result
+}
+
+define <2 x double> @test_sinh_v2() {
+; CHECK-LABEL: define <2 x double> @test_sinh_v2() {
+; CHECK-NEXT:    ret <2 x double> zeroinitializer
+;
+  %result = call <2 x double> @llvm.sinh.v2f64(<2 x double> zeroinitializer)
+  ret <2 x double> %result
+}
+
+define double @test_cosh() {
+; CHECK-LABEL: define double @test_cosh() {
+; CHECK-NEXT:    ret double 1.000000e+00
+;
+  %result = call double @llvm.cosh.f64(double 0.0)
+  ret double %result
+}
+
+define <2 x double> @test_cosh_v2() {
+; CHECK-LABEL: define <2 x double> @test_cosh_v2() {
+; CHECK-NEXT:    ret <2 x double> splat (double 1.000000e+00)
+;
+  %result = call <2 x double> @llvm.cosh.v2f64(<2 x double> zeroinitializer)
+  ret <2 x double> %result
+}
+
----------------
arsenm wrote:

Ignore it, should show folding to nan or ignored 

https://github.com/llvm/llvm-project/pull/132671