[clang] [llvm] Add support for flag output operand "=@cc" for SystemZ. (PR #125970)

[Ulrich Weigand via llvm-commits]

================
@@ -0,0 +1,786 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
+; Test Flag Output Operands with 14 combinations of CCMASK and optimizations.
+; This test negate of flag_output_operand_ccand, e.g
+; CC != 0 && cc != 1 && cc != 2 for AND for three different functions,
+; including two test cases from heiko.
+
+; RUN: llc < %s -verify-machineinstrs -mtriple=s390x-linux-gnu -O2 | FileCheck %s
+
+; Test CC != 0 && CC != 1.
----------------
uweigand wrote:

Now, these tests are not tautological.  However, you still should make clear that you're not actually testing this code sequence in IL, but rather you're testing the IL that is currently being generated by common optimizers from that source code.

In those cases, it might be better to have a clang-level test that verifies which IL we get, and then have IL-level tests here where necessary.

https://github.com/llvm/llvm-project/pull/125970