[flang-commits] [flang] [flang][SystemZ] Fix test failures on s390x for target alignment, endianness and vectorizer remarks (PR #230565)

[Ulrich Weigand via flang-commits]

================
@@ -6,8 +6,8 @@
 ! added to this directory and sub-directories.
 !===----------------------------------------------------------------------===!
 
-!RUN: %flang_fc1 -emit-llvm -fopenmp -mmlir --enable-delayed-privatization-staging=false -fopenmp-version=51 -fopenmp-targets=amdgcn-amd-amdhsa %s -o - | FileCheck %s --check-prefixes=CHECK,CHECK-NO-FPRIV
-!RUN: %flang_fc1 -emit-llvm -fopenmp -mmlir --enable-delayed-privatization-staging=true -fopenmp-version=51 -fopenmp-targets=amdgcn-amd-amdhsa %s -o - | FileCheck %s --check-prefixes=CHECK,CHECK-FPRIV
+!RUN: %flang_fc1 -emit-llvm -fopenmp -mmlir --enable-delayed-privatization-staging=false -fopenmp-version=51 -fopenmp-targets=amdgcn-amd-amdhsa %s -o - | FileCheck %s -DALLOCA_ALIGN=%if target=s390x{{.*}} %{4%} %else %{8%} --check-prefixes=CHECK,CHECK-NO-FPRIV
----------------
uweigand wrote:

Can we call this `AGG_ALIGN` here as well?  The core difference is preferred alignment of aggregates after all, it's nothing to do with alloca as such.

https://github.com/llvm/llvm-project/pull/230565