[llvm] [X86] Sink NOT to be fold into ANDN (PR #172329)

[Piotr Fusik via llvm-commits]

================
@@ -0,0 +1,3168 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=i686-- -mattr=-bmi | FileCheck %s --check-prefixes=X86-NOBMI,X86
+; RUN: llc < %s -mtriple=i686-- -mattr=-bmi,+sse | FileCheck %s --check-prefixes=X86-NOBMI,X86-SSE
+; RUN: llc < %s -mtriple=i686-- -mattr=-bmi,+sse2 | FileCheck %s --check-prefixes=X86-NOBMI,X86-SSE2
+; RUN: llc < %s -mtriple=i686-- -mattr=+bmi | FileCheck %s --check-prefixes=X86-BMI
+; RUN: llc < %s -mtriple=x86_64-- -mattr=-bmi | FileCheck %s --check-prefixes=X64,X64-NOAVX2,X64-NOBMI
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+bmi | FileCheck %s --check-prefixes=X64,X64-BMI,X64-NOAVX2
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+bmi,+avx2 | FileCheck %s --check-prefixes=X64,X64-BMI,X64-AVX2
----------------
pfusik wrote:

AVX2 has `vandnps`.

https://github.com/llvm/llvm-project/pull/172329