[llvm] [InstCombine] Optimize `sinh` and `cosh` divisions (PR #81433)

[Felix Kellenbenz via llvm-commits]

================
@@ -0,0 +1,114 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt -S -passes=instcombine < %s | FileCheck %s
+
+define double @fdiv_cosh_sinh(double %a) {
+; CHECK-LABEL: @fdiv_cosh_sinh(
+; CHECK-NEXT:    [[TMP1:%.*]] = call double @cosh(double [[A:%.*]])
+; CHECK-NEXT:    [[TMP2:%.*]] = call double @sinh(double [[A]])
+; CHECK-NEXT:    [[DIV:%.*]] = fdiv double [[TMP1]], [[TMP2]]
+; CHECK-NEXT:    ret double [[DIV]]
+;
+  %1 = call double @cosh(double %a)
+  %2 = call double @sinh(double %a)
+  %div = fdiv double %1, %2
+  ret double %div
+}
+
+define double @fdiv_strict_cosh_strict_sinh_reassoc(double %a) {
+; CHECK-LABEL: @fdiv_strict_cosh_strict_sinh_reassoc(
+; CHECK-NEXT:    [[TMP1:%.*]] = call double @cosh(double [[A:%.*]])
+; CHECK-NEXT:    [[TMP2:%.*]] = call reassoc double @sinh(double [[A]])
+; CHECK-NEXT:    [[DIV:%.*]] = fdiv double [[TMP1]], [[TMP2]]
+; CHECK-NEXT:    ret double [[DIV]]
+;
+  %1 = call double @cosh(double %a)
+  %2 = call reassoc double @sinh(double %a)
+  %div = fdiv double %1, %2
+  ret double %div
+}
+
+define double @fdiv_reassoc_cosh_strict_sinh_strict(double %a, ptr dereferenceable(2) %dummy) {
+; CHECK-LABEL: @fdiv_reassoc_cosh_strict_sinh_strict(
+; CHECK-NEXT:    [[TANH:%.*]] = call reassoc double @tanh(double [[A]])
+; CHECK-NEXT:    [[DIV:%.*]] = fdiv reassoc double 1.000000e+00, [[TANH]]
+; CHECK-NEXT:    ret double [[DIV]]
+;
+  %1 = call double @cosh(double %a)
+  %2 = call double @sinh(double %a)
+  %div = fdiv reassoc double %1, %2
----------------
felixkellenbenz wrote:

Since I thought the intended behavior for the sinh(x)/cosh(x) and cosh(x)/sinh(x) should be similar to the behavior of the non hyperbolic functions I thought it made sense to check for the same flags that are checked for the non hyperbolic functions. After reading your comment and learning about the inaccuracy the new optimization introduces, it makes more sense for me if we would check for `afn` instead.
I will look into this and make changes so that we check for `afn` instead of `reassoc`.

https://github.com/llvm/llvm-project/pull/81433