[flang-commits] [flang] 48eb2bc - [flang][driver] Use `-O{0|1|2|3}` to define LLVM backend pass pipeline

[Andrzej Warzynski via flang-commits]

Author: Andrzej Warzynski
Date: 2022-06-27T15:08:54Z
New Revision: 48eb2bc608e2db4258f9a5432be6458b33e8f551

URL: https://github.com/llvm/llvm-project/commit/48eb2bc608e2db4258f9a5432be6458b33e8f551
DIFF: https://github.com/llvm/llvm-project/commit/48eb2bc608e2db4258f9a5432be6458b33e8f551.diff

LOG: [flang][driver] Use `-O{0|1|2|3}` to define LLVM backend pass pipeline

Support for optimisation flags in LLVM Flang was originally added in
https://reviews.llvm.org/D128043. That patch focused on LLVM
middle-end/optimisation pipelines. With this patch, Flang will
additionally configure LLVM backend pass pipelines accordingly. This
behavior is consistent with Clang.

New hook is added to translate compiler optimisation flags (e.g. `-O3`)
into backend optimisation level: `getCGOptLevel`. Identical hooks are
available in Clang and LLVM. In other words, the meaning of these
optimisation flags remains consistent with other sub-projects that use
LLVM backends.

Differential Revision: https://reviews.llvm.org/D128050

Added: 
    flang/test/Driver/default-backend-pipelines.f90

Modified: 
    flang/lib/Frontend/FrontendActions.cpp

Removed: 
    


################################################################################
diff  --git a/flang/lib/Frontend/FrontendActions.cpp b/flang/lib/Frontend/FrontendActions.cpp
index ae7150900ecf0..c53d9f2347faf 100644
--- a/flang/lib/Frontend/FrontendActions.cpp
+++ b/flang/lib/Frontend/FrontendActions.cpp
@@ -517,6 +517,22 @@ void CodeGenAction::generateLLVMIR() {
   }
 }
 
+static llvm::CodeGenOpt::Level
+getCGOptLevel(const Fortran::frontend::CodeGenOptions &opts) {
+  switch (opts.OptimizationLevel) {
+  default:
+    llvm_unreachable("Invalid optimization level!");
+  case 0:
+    return llvm::CodeGenOpt::None;
+  case 1:
+    return llvm::CodeGenOpt::Less;
+  case 2:
+    return llvm::CodeGenOpt::Default;
+  case 3:
+    return llvm::CodeGenOpt::Aggressive;
+  }
+}
+
 void CodeGenAction::setUpTargetMachine() {
   CompilerInstance &ci = this->getInstance();
 
@@ -535,9 +551,12 @@ void CodeGenAction::setUpTargetMachine() {
   assert(theTarget && "Failed to create Target");
 
   // Create `TargetMachine`
-  tm.reset(theTarget->createTargetMachine(theTriple, /*CPU=*/"",
-                                          /*Features=*/"",
-                                          llvm::TargetOptions(), llvm::None));
+  llvm::CodeGenOpt::Level OptLevel =
+      getCGOptLevel(ci.getInvocation().getCodeGenOpts());
+  tm.reset(theTarget->createTargetMachine(
+      theTriple, /*CPU=*/"",
+      /*Features=*/"", llvm::TargetOptions(), /*Reloc::Model=*/llvm::None,
+      /*CodeModel::Model=*/llvm::None, OptLevel));
   assert(tm && "Failed to create TargetMachine");
 }
 

diff  --git a/flang/test/Driver/default-backend-pipelines.f90 b/flang/test/Driver/default-backend-pipelines.f90
new file mode 100644
index 0000000000000..7d233bb751bc2
--- /dev/null
+++ b/flang/test/Driver/default-backend-pipelines.f90
@@ -0,0 +1,25 @@
+! Verify that`-O{n}` is indeed taken into account when definining the LLVM backend pass pipeline.
+
+! REQUIRES: aarch64-registered-target
+
+!-----------
+! RUN LINES
+!-----------
+! RUN: %flang_fc1 -S -O2 %s -triple aarch64-unknown-linux-gnu -mllvm -debug-pass=Structure -o /dev/null 2>&1 | FileCheck %s --check-prefix=CHECK-O2
+! RUN: %flang_fc1 -S -O3 %s -triple aarch64-unknown-linux-gnu -mllvm -debug-pass=Structure -o /dev/null 2>&1 | FileCheck %s --check-prefix=CHECK-O3
+
+!-----------------------
+! EXPECTED OUTPUT
+!-----------------------
+! CHECK-O2-NOT: SVE intrinsics optimizations
+
+! CHECK-O3: SVE intrinsics optimizations
+
+!-------
+! INPUT
+!-------
+subroutine simple_loop
+  integer :: i
+  do i=1,5
+  end do
+end subroutine