[flang][driver] Add support for -S and implement -c/-emit-obj

This patch adds support for:
  * `-S` in Flang's compiler and frontend drivers,
and implements:
  * `-emit-obj` in Flang's frontend driver and `-c` in Flang's compiler
    driver (this is consistent with Clang).
(these options were already available before, but only as placeholders).
The semantics of these options in Clang and Flang are identical.

The `EmitObjAction` frontend action is renamed as `BackendAction`. This
new name more accurately reflects the fact that this action will
primarily run the code-gen/backend pipeline in LLVM. It also makes more
sense as an action implementing both `-emit-obj` and `-S` (originally,
it was just `-emit-obj`).

`tripleName` from FirContext.cpp is deleted and, when a target triple is
required, `mlir::LLVM::LLVMDialect::getTargetTripleAttrName()` is used
instead. In practice, this means that `fir.triple` is replaced with
`llvm.target_triple`. The former was effectively ignored. The latter is
used when lowering from the LLVM dialect in MLIR to LLVM IR (i.e. it's
embedded in the generated LLVM IR module). The driver can then re-use
it when configuring the backend. With this change, the LLVM IR files
generated by e.g. `tco` will from now on contain the correct target
triple.

The code-gen.f90 test is replaced with code-gen-x86.f90 and
code-gen-aarch64.f90. With 2 seperate files we can verify that
`--target` is correctly taken into account. LIT configuration is updated
to enable e.g.:
```
! REQUIRES: aarch64-registered-target
```

Differential Revision: https://reviews.llvm.org/D120568

23 files changed, 285 insertions, 39 deletions

diff --git a/clang/include/clang/Driver/Options.td b/clang/include/clang/Driver/Options.td
index 3d49e6b..fa54d4a 100644
--- a/clang/include/clang/Driver/Options.td
+++ b/clang/include/clang/Driver/Options.td
@@ -750,7 +750,7 @@ def Rpass_analysis_EQ : Joined<["-"], "Rpass-analysis=">, Group<R_value_Group>,
            "name matches the given POSIX regular expression">;
 def R_Joined : Joined<["-"], "R">, Group<R_Group>, Flags<[CC1Option, CoreOption]>,
   MetaVarName<"<remark>">, HelpText<"Enable the specified remark">;
-def S : Flag<["-"], "S">, Flags<[NoXarchOption,CC1Option]>, Group<Action_Group>,
+def S : Flag<["-"], "S">, Flags<[NoXarchOption,CC1Option,FlangOption,FC1Option]>, Group<Action_Group>,
   HelpText<"Only run preprocess and compilation steps">;
 def Tbss : JoinedOrSeparate<["-"], "Tbss">, Group<T_Group>,
   MetaVarName<"<addr>">, HelpText<"Set starting address of BSS to <addr>">;
diff --git a/flang/include/flang/Frontend/FrontendActions.h b/flang/include/flang/Frontend/FrontendActions.h
index 0cb6d63..34cabea 100644
--- a/flang/include/flang/Frontend/FrontendActions.h
+++ b/flang/include/flang/Frontend/FrontendActions.h
@@ -192,8 +192,19 @@ class EmitLLVMAction : public CodeGenAction {
   void ExecuteAction() override;
 };
 
-class EmitObjAction : public CodeGenAction {
+class BackendAction : public CodeGenAction {
+public:
+  enum class BackendActionTy {
+    Backend_EmitAssembly, ///< Emit native assembly files
+    Backend_EmitObj ///< Emit native object files
+  };
+
+  BackendAction(BackendActionTy act) : action{act} {};
+
+private:
   void ExecuteAction() override;
+
+  BackendActionTy action;
 };
 
 } // namespace Fortran::frontend
diff --git a/flang/include/flang/Frontend/FrontendOptions.h b/flang/include/flang/Frontend/FrontendOptions.h
index c396822..b299c87 100644
--- a/flang/include/flang/Frontend/FrontendOptions.h
+++ b/flang/include/flang/Frontend/FrontendOptions.h
@@ -40,6 +40,9 @@ enum ActionKind {
   /// Emit a .o file.
   EmitObj,
 
+  /// Emit a .s file.
+  EmitAssembly,
+
   /// Parse, unparse the parse-tree and output a Fortran source file
   DebugUnparse,
 
diff --git a/flang/lib/Frontend/CMakeLists.txt b/flang/lib/Frontend/CMakeLists.txt
index 4f54ab7..a68e0e9 100644
--- a/flang/lib/Frontend/CMakeLists.txt
+++ b/flang/lib/Frontend/CMakeLists.txt
@@ -27,6 +27,8 @@ add_flang_library(flangFrontend
   FortranLower
   clangBasic
   clangDriver
+  LLVMAnalysis
+  LLVMTarget
   FIRDialect
   FIRSupport
   FIRBuilder
diff --git a/flang/lib/Frontend/CompilerInvocation.cpp b/flang/lib/Frontend/CompilerInvocation.cpp
index a1d7309..d1b0d0e 100644
--- a/flang/lib/Frontend/CompilerInvocation.cpp
+++ b/flang/lib/Frontend/CompilerInvocation.cpp
@@ -153,6 +153,9 @@ static bool ParseFrontendArgs(FrontendOptions &opts, llvm::opt::ArgList &args,
     case clang::driver::options::OPT_emit_obj:
       opts.programAction = EmitObj;
       break;
+    case clang::driver::options::OPT_S:
+      opts.programAction = EmitAssembly;
+      break;
     case clang::driver::options::OPT_fdebug_unparse:
       opts.programAction = DebugUnparse;
       break;
diff --git a/flang/lib/Frontend/FrontendActions.cpp b/flang/lib/Frontend/FrontendActions.cpp
index c72460f..e30f6af 100644
--- a/flang/lib/Frontend/FrontendActions.cpp
+++ b/flang/lib/Frontend/FrontendActions.cpp
@@ -31,7 +31,13 @@
 #include "mlir/Pass/PassManager.h"
 #include "mlir/Target/LLVMIR/ModuleTranslation.h"
 #include "llvm/ADT/StringRef.h"
+#include "llvm/Analysis/TargetLibraryInfo.h"
+#include "llvm/Analysis/TargetTransformInfo.h"
+#include "llvm/IR/LegacyPassManager.h"
+#include "llvm/MC/TargetRegistry.h"
+#include "llvm/Passes/PassBuilder.h"
 #include "llvm/Support/ErrorHandling.h"
+#include "llvm/Target/TargetMachine.h"
 #include <clang/Basic/Diagnostic.h>
 #include <memory>
 
@@ -417,7 +423,6 @@ void CodeGenAction::GenerateLLVMIR() {
 
   pm.addPass(std::make_unique<Fortran::lower::VerifierPass>());
   pm.enableVerifier(/*verifyPasses=*/true);
-  mlir::PassPipelineCLParser passPipeline("", "Compiler passes to run");
 
   // Create the pass pipeline
   fir::createMLIRToLLVMPassPipeline(pm);
@@ -490,11 +495,90 @@ void EmitMLIRAction::ExecuteAction() {
   mlirModule->print(*os);
 }
 
-void EmitObjAction::ExecuteAction() {
+void BackendAction::ExecuteAction() {
   CompilerInstance &ci = this->instance();
-  unsigned DiagID = ci.diagnostics().getCustomDiagID(
-      clang::DiagnosticsEngine::Error, "code-generation is not available yet");
-  ci.diagnostics().Report(DiagID);
+  // Generate an LLVM module if it's not already present (it will already be
+  // present if the input file is an LLVM IR/BC file).
+  if (!llvmModule)
+    GenerateLLVMIR();
+
+  // Create `Target`
+  std::string error;
+  const std::string &theTriple = llvmModule->getTargetTriple();
+  const llvm::Target *theTarget =
+      llvm::TargetRegistry::lookupTarget(theTriple, error);
+  // TODO: Make this a diagnostic once `flang-new` can consume LLVM IR files
+  // (in which users could use unsupported triples)
+  assert(theTarget && "Failed to create Target");
+
+  // Create `TargetMachine`
+  std::unique_ptr<llvm::TargetMachine> TM;
+  TM.reset(theTarget->createTargetMachine(theTriple, /*CPU=*/"",
+      /*Features=*/"", llvm::TargetOptions(), llvm::None));
+  assert(TM && "Failed to create TargetMachine");
+  llvmModule->setDataLayout(TM->createDataLayout());
+
+  // If the output stream is a file, generate it and define the corresponding
+  // output stream. If a pre-defined output stream is available, we will use
+  // that instead.
+  //
+  // NOTE: `os` is a smart pointer that will be destroyed at the end of this
+  // method. However, it won't be written to until `CodeGenPasses` is
+  // destroyed. By defining `os` before `CodeGenPasses`, we make sure that the
+  // output stream won't be destroyed before it is written to. This only
+  // applies when an output file is used (i.e. there is no pre-defined output
+  // stream).
+  // TODO: Revisit once the new PM is ready (i.e. when `CodeGenPasses` is
+  // updated to use it).
+  std::unique_ptr<llvm::raw_pwrite_stream> os;
+  if (ci.IsOutputStreamNull()) {
+    // Get the output buffer/file
+    switch (action) {
+    case BackendActionTy::Backend_EmitAssembly:
+      os = ci.CreateDefaultOutputFile(
+          /*Binary=*/false, /*InFile=*/GetCurrentFileOrBufferName(), "s");
+      break;
+    case BackendActionTy::Backend_EmitObj:
+      os = ci.CreateDefaultOutputFile(
+          /*Binary=*/true, /*InFile=*/GetCurrentFileOrBufferName(), "o");
+      break;
+    }
+    if (!os) {
+      unsigned diagID = ci.diagnostics().getCustomDiagID(
+          clang::DiagnosticsEngine::Error, "failed to create the output file");
+      ci.diagnostics().Report(diagID);
+      return;
+    }
+  }
+
+  // Create an LLVM code-gen pass pipeline. Currently only the legacy pass
+  // manager is supported.
+  // TODO: Switch to the new PM once it's available in the backend.
+  llvm::legacy::PassManager CodeGenPasses;
+  CodeGenPasses.add(
+      createTargetTransformInfoWrapperPass(TM->getTargetIRAnalysis()));
+  llvm::Triple triple(theTriple);
+
+  std::unique_ptr<llvm::TargetLibraryInfoImpl> TLII =
+      std::make_unique<llvm::TargetLibraryInfoImpl>(triple);
+  assert(TLII && "Failed to create TargetLibraryInfo");
+  CodeGenPasses.add(new llvm::TargetLibraryInfoWrapperPass(*TLII));
+
+  llvm::CodeGenFileType cgft = (action == BackendActionTy::Backend_EmitAssembly)
+      ? llvm::CodeGenFileType::CGFT_AssemblyFile
+      : llvm::CodeGenFileType::CGFT_ObjectFile;
+  if (TM->addPassesToEmitFile(CodeGenPasses,
+          ci.IsOutputStreamNull() ? *os : ci.GetOutputStream(), nullptr,
+          cgft)) {
+    unsigned diagID =
+        ci.diagnostics().getCustomDiagID(clang::DiagnosticsEngine::Error,
+            "emission of this file type is not supported");
+    ci.diagnostics().Report(diagID);
+    return;
+  }
+
+  // Run the code-gen passes
+  CodeGenPasses.run(*llvmModule);
 }
 
 void InitOnlyAction::ExecuteAction() {
diff --git a/flang/lib/FrontendTool/ExecuteCompilerInvocation.cpp b/flang/lib/FrontendTool/ExecuteCompilerInvocation.cpp
index d97833b..4e8c225 100644
--- a/flang/lib/FrontendTool/ExecuteCompilerInvocation.cpp
+++ b/flang/lib/FrontendTool/ExecuteCompilerInvocation.cpp
@@ -38,7 +38,11 @@ static std::unique_ptr<FrontendAction> CreateFrontendBaseAction(
   case EmitLLVM:
     return std::make_unique<EmitLLVMAction>();
   case EmitObj:
-    return std::make_unique<EmitObjAction>();
+    return std::make_unique<BackendAction>(
+        BackendAction::BackendActionTy::Backend_EmitObj);
+  case EmitAssembly:
+    return std::make_unique<BackendAction>(
+        BackendAction::BackendActionTy::Backend_EmitAssembly);
   case DebugUnparse:
     return std::make_unique<DebugUnparseAction>();
   case DebugUnparseNoSema:
diff --git a/flang/lib/Optimizer/Support/FIRContext.cpp b/flang/lib/Optimizer/Support/FIRContext.cpp
index 890364a..80dfa1c 100644
--- a/flang/lib/Optimizer/Support/FIRContext.cpp
+++ b/flang/lib/Optimizer/Support/FIRContext.cpp
@@ -12,19 +12,20 @@
 
 #include "flang/Optimizer/Support/FIRContext.h"
 #include "flang/Optimizer/Support/KindMapping.h"
+#include "mlir/Dialect/LLVMIR/LLVMDialect.h"
 #include "mlir/IR/BuiltinAttributes.h"
 #include "mlir/IR/BuiltinOps.h"
 #include "llvm/Support/Host.h"
 
-static constexpr const char *tripleName = "fir.triple";
-
 void fir::setTargetTriple(mlir::ModuleOp mod, llvm::StringRef triple) {
   auto target = fir::determineTargetTriple(triple);
-  mod->setAttr(tripleName, mlir::StringAttr::get(mod.getContext(), target));
+  mod->setAttr(mlir::LLVM::LLVMDialect::getTargetTripleAttrName(),
+               mlir::StringAttr::get(mod.getContext(), target));
 }
 
 llvm::Triple fir::getTargetTriple(mlir::ModuleOp mod) {
-  if (auto target = mod->getAttrOfType<mlir::StringAttr>(tripleName))
+  if (auto target = mod->getAttrOfType<mlir::StringAttr>(
+          mlir::LLVM::LLVMDialect::getTargetTripleAttrName()))
     return llvm::Triple(target.getValue());
   return llvm::Triple(llvm::sys::getDefaultTargetTriple());
 }
diff --git a/flang/test/CMakeLists.txt b/flang/test/CMakeLists.txt
index e2725d2..282426e 100644
--- a/flang/test/CMakeLists.txt
+++ b/flang/test/CMakeLists.txt
@@ -46,7 +46,7 @@ set(FLANG_TEST_PARAMS
   flang_site_config=${CMAKE_CURRENT_BINARY_DIR}/lit.site.cfg.py)
 
 set(FLANG_TEST_DEPENDS
-  flang-new llvm-config FileCheck count not module_files fir-opt tco bbc
+  flang-new llvm-config FileCheck count not module_files fir-opt tco bbc llvm-objdump
 )
 
 if (FLANG_INCLUDE_TESTS)
diff --git a/flang/test/Driver/code-gen-aarch64.f90 b/flang/test/Driver/code-gen-aarch64.f90
new file mode 100644
index 0000000..aabfabb
--- /dev/null
+++ b/flang/test/Driver/code-gen-aarch64.f90
@@ -0,0 +1,31 @@
+! Test -emit-obj (X86)
+
+! REQUIRES: aarch64-registered-target, x86-registered-target
+
+!-------------
+! RUN COMMANDS
+!-------------
+! RUN: rm -f %t.o
+! RUN: %flang_fc1 -emit-obj -triple aarch64-unknown-linux-gnu %s -o %t.o
+! RUN: llvm-objdump --triple aarch64-unknown-linux-gnu --disassemble-all %t.o | FileCheck %s --check-prefix=CORRECT_TRIPLE
+! RUN: rm -f %t.o
+! RUN: %flang -c --target=aarch64-unknown-linux-gnu %s -o %t.o
+! RUN: llvm-objdump --triple aarch64-unknown-linux-gnu --disassemble-all %t.o | FileCheck %s --check-prefix=CORRECT_TRIPLE
+
+! RUN: %flang -c --target=aarch64-unknown-linux-gnu %s -o %t.o
+! RUN: llvm-objdump --triple x86_64-unknown-linux-gnu --disassemble-all %t.o | FileCheck %s --check-prefix=INCORRECT_TRIPLE
+
+!----------------
+! EXPECTED OUTPUT
+!----------------
+! CORRECT_TRIPLE-LABEL: <_QQmain>:
+! CORRECT_TRIPLE-NEXT:  	ret
+
+! When incorrect triple is used to disassemble, there won't be a ret instruction at all.
+! INCORRECT_TRIPLE-LABEL: <_QQmain>:
+! INCORRECT_TRIPLE-NOT:  	ret
+
+!------
+! INPUT
+!------
+end program
diff --git a/flang/test/Driver/code-gen-x86.f90 b/flang/test/Driver/code-gen-x86.f90
new file mode 100644
index 0000000..e630677
--- /dev/null
+++ b/flang/test/Driver/code-gen-x86.f90
@@ -0,0 +1,32 @@
+! Test -emit-obj (X86)
+
+! REQUIRES: x86-registered-target, aarch64-registered-target
+! UNSUPPORTED: darwin, macos
+
+!-------------
+! RUN COMMANDS
+!-------------
+! RUN: rm -f %t.o
+! RUN: %flang_fc1 -triple x86_64-unknown-linux-gnu -emit-obj %s -o %t.o
+! RUN: llvm-objdump --triple x86_64-unknown-linux-gnu --disassemble-all %t.o | FileCheck %s --check-prefix=CORRECT_TRIPLE
+! RUN: rm -f %t.o
+! RUN: %flang --target=x86_64-unknown-linux-gnu -c %s -o %t.o
+! RUN: llvm-objdump --triple x86_64-unknown-linux-gnu --disassemble-all %t.o | FileCheck %s --check-prefix=CORRECT_TRIPLE
+
+! RUN: %flang -c --target=x86_64-unknown-linux-gnu %s -o %t.o
+! RUN: llvm-objdump --triple aarch64-unknown-linux-gnu --disassemble-all %t.o | FileCheck %s --check-prefix=INCORRECT_TRIPLE
+
+!----------------
+! EXPECTED OUTPUT
+!----------------
+! CORRECT_TRIPLE-LABEL: <_QQmain>:
+! CORRECT_TRIPLE-NEXT:  	retq
+
+! When incorrect triple is used to disassemble, there won't be a ret(q) instruction at all.
+! INCORRECT_TRIPLE-LABEL: <_QQmain>:
+! INCORRECT_TRIPLE-NOT:  	ret
+
+!------
+! INPUT
+!------
+end program
diff --git a/flang/test/Driver/code-gen.f90 b/flang/test/Driver/code-gen.f90
deleted file mode 100644
index 53ee40d..0000000
--- a/flang/test/Driver/code-gen.f90
+++ /dev/null
@@ -1,19 +0,0 @@
-! Although code-generation is not yet available, we do have frontend actions
-! that correspond to `-c` and `-emit-obj`. For now these actions are just a
-! placeholder and running them leads to a driver error. This test makes sure
-! that these actions are indeed run (rather than `-c` or `-emit-obj` being
-! rejected earlier).
-! TODO: Replace this file with a proper test once code-generation is available.
-
-!-----------
-! RUN LINES
-!-----------
-! RUN: not %flang %s 2>&1 | FileCheck %s --check-prefix=ERROR
-! RUN: not %flang -c %s 2>&1 | FileCheck %s --check-prefix=ERROR
-! RUN: not %flang -emit-obj %s 2>&1 | FileCheck %s --check-prefix=ERROR
-! RUN: not %flang -fc1 -emit-obj %s 2>&1 | FileCheck %s --check-prefix=ERROR
-
-!-----------------------
-! EXPECTED OUTPUT
-!-----------------------
-! ERROR: code-generation is not available yet
diff --git a/flang/test/Driver/driver-help-hidden.f90 b/flang/test/Driver/driver-help-hidden.f90
index 5aed6f2..150900a 100644
--- a/flang/test/Driver/driver-help-hidden.f90
+++ b/flang/test/Driver/driver-help-hidden.f90
@@ -55,6 +55,7 @@
 ! CHECK-NEXT: -print-target-triple    Print the normalized target triple
 ! CHECK-NEXT: -P                     Disable linemarker output in -E mode
 ! CHECK-NEXT: -std=<value>           Language standard to compile for
+! CHECK-NEXT: -S                     Only run preprocess and compilation steps
 ! CHECK-NEXT: --target=<value>        Generate code for the given target
 ! CHECK-NEXT: -U <macro>             Undefine macro <macro>
 ! CHECK-NEXT: --version Print version information
diff --git a/flang/test/Driver/driver-help.f90 b/flang/test/Driver/driver-help.f90
index 560c675..fd19220 100644
--- a/flang/test/Driver/driver-help.f90
+++ b/flang/test/Driver/driver-help.f90
@@ -55,6 +55,7 @@
 ! HELP-NEXT: -print-target-triple    Print the normalized target triple
 ! HELP-NEXT: -P                     Disable linemarker output in -E mode
 ! HELP-NEXT: -std=<value>           Language standard to compile for
+! HELP-NEXT: -S                     Only run preprocess and compilation steps
 ! HELP-NEXT: --target=<value>       Generate code for the given target
 ! HELP-NEXT: -U <macro>             Undefine macro <macro>
 ! HELP-NEXT: --version              Print version information
@@ -128,6 +129,7 @@
 ! HELP-FC1-NEXT: -plugin <name>         Use the named plugin action instead of the default action (use "help" to list available options)
 ! HELP-FC1-NEXT: -P                     Disable linemarker output in -E mode
 ! HELP-FC1-NEXT: -std=<value>           Language standard to compile for
+! HELP-FC1-NEXT: -S                     Only run preprocess and compilation steps
 ! HELP-FC1-NEXT: -test-io               Run the InputOuputTest action. Use for development and testing only.
 ! HELP-FC1-NEXT: -triple <value>        Specify target triple (e.g. i686-apple-darwin9)
 ! HELP-FC1-NEXT: -U <macro>             Undefine macro <macro>
diff --git a/flang/test/Driver/emit-asm-aarch64.f90 b/flang/test/Driver/emit-asm-aarch64.f90
new file mode 100644
index 0000000..4c84985
--- /dev/null
+++ b/flang/test/Driver/emit-asm-aarch64.f90
@@ -0,0 +1,21 @@
+! Test -S (AArch64)
+
+! REQUIRES: aarch64-registered-target
+
+!-------------
+! RUN COMMANDS
+!-------------
+! RUN: %flang_fc1 -S -triple aarch64-unknown-linux-gnu %s -o - | FileCheck %s
+! RUN: %flang -S -target aarch64-unknown-linux-gnu %s -o - | FileCheck %s
+
+!----------------
+! EXPECTED OUTPUT
+!----------------
+! CHECK-LABEL: _QQmain:
+! CHECK-NEXT: .Lfunc_begin0:
+! CHECK: ret
+
+!------
+! INPUT
+!------
+end program
diff --git a/flang/test/Driver/emit-asm-x86.f90 b/flang/test/Driver/emit-asm-x86.f90
new file mode 100644
index 0000000..c79406c
--- /dev/null
+++ b/flang/test/Driver/emit-asm-x86.f90
@@ -0,0 +1,21 @@
+! Test -S (X86)
+
+! REQUIRES: x86-registered-target
+
+!-------------
+! RUN COMMANDS
+!-------------
+! RUN: %flang_fc1 -S -triple x86_64-unknown-linux-gnu %s -o - | FileCheck %s
+! RUN: %flang -S -target x86_64-unknown-linux-gnu %s -o - | FileCheck %s
+
+!----------------
+! EXPECTED OUTPUT
+!----------------
+! CHECK-LABEL: _QQmain:
+! CHECK-NEXT: .Lfunc_begin0:
+! CHECK: ret
+
+!------
+! INPUT
+!------
+end program
diff --git a/flang/test/Driver/syntax-only.f90 b/flang/test/Driver/syntax-only.f90
index a97abee..df831e0 100644
--- a/flang/test/Driver/syntax-only.f90
+++ b/flang/test/Driver/syntax-only.f90
@@ -13,14 +13,15 @@
 ! RUN: %flang -fsyntax-only %s 2>&1 | FileCheck %s --allow-empty
 ! RUN: %flang_fc1 %s 2>&1 | FileCheck %s --allow-empty
 
-! RUN: not %flang  %s 2>&1 | FileCheck %s --check-prefix=NO_FSYNTAX_ONLY
-! RUN: not %flang_fc1 -emit-obj %s 2>&1 | FileCheck %s --check-prefix=NO_FSYNTAX_ONLY
+! RUN: rm -rf %t/non-existent-dir/
+! RUN: not %flang -c %s -o %t/non-existent-dir/syntax-only.o 2>&1 | FileCheck %s --check-prefix=NO_FSYNTAX_ONLY
+! RUN: not %flang_fc1 -emit-obj %s -o %t/non-existent-dir/syntax-only.o 2>&1 | FileCheck %s --check-prefix=NO_FSYNTAX_ONLY
 
 !-----------------
 ! EXPECTED OUTPUT
 !-----------------
 ! CHECK-NOT: error
-! NO_FSYNTAX_ONLY: error: code-generation is not available yet
+! NO_FSYNTAX_ONLY: error: failed to create the output file
 
 !-------
 ! INPUT
diff --git a/flang/test/Fir/target-rewrite-triple.fir b/flang/test/Fir/target-rewrite-triple.fir
index ccaf4fa..b9e61b3 100644
--- a/flang/test/Fir/target-rewrite-triple.fir
+++ b/flang/test/Fir/target-rewrite-triple.fir
@@ -1,10 +1,10 @@
 // RUN: fir-opt --target-rewrite %s | FileCheck %s --check-prefix=UNCHANGED
 // RUN: fir-opt --target-rewrite="target=x86_64-unknown-linux-gnu" %s | FileCheck %s --check-prefix=CHANGED
 
-// UNCHANGED: fir.triple = "aarch64-unknown-linux-gnu"
-// CHANGED: fir.triple = "x86_64-unknown-linux-gnu"
-// CHANGED-NOT: fir.triple = "aarch64-unknown-linux-gnu"
-module attributes {fir.triple = "aarch64-unknown-linux-gnu"}  {
+// UNCHANGED: llvm.target_triple = "aarch64-unknown-linux-gnu"
+// CHANGED: llvm.target_triple = "x86_64-unknown-linux-gnu"
+// CHANGED-NOT: llvm.target_triple = "aarch64-unknown-linux-gnu"
+module attributes {llvm.target_triple = "aarch64-unknown-linux-gnu"}  {
   func @dummyfunc() -> () {
     return
   }
diff --git a/flang/test/lit.cfg.py b/flang/test/lit.cfg.py
index 3942198..9578ab9 100644
--- a/flang/test/lit.cfg.py
+++ b/flang/test/lit.cfg.py
@@ -39,6 +39,11 @@ llvm_config.use_default_substitutions()
 llvm_config.feature_config(
     [('--assertion-mode', {'ON': 'asserts'})])
 
+# Targets
+config.targets = frozenset(config.targets_to_build.split())
+for arch in config.targets_to_build.split():
+    config.available_features.add(arch.lower() + '-registered-target')
+
 # excludes: A list of directories to exclude from the testsuite. The 'Inputs'
 # subdirectories contain auxiliary inputs for various tests in their parent
 # directories.
diff --git a/flang/test/lit.site.cfg.py.in b/flang/test/lit.site.cfg.py.in
index 378602d..5619dd6 100644
--- a/flang/test/lit.site.cfg.py.in
+++ b/flang/test/lit.site.cfg.py.in
@@ -18,6 +18,7 @@ config.python_executable = "@PYTHON_EXECUTABLE@"
 config.flang_standalone_build = @FLANG_STANDALONE_BUILD@
 config.has_plugins = @LLVM_ENABLE_PLUGINS@
 config.cc = "@CMAKE_C_COMPILER@"
+config.targets_to_build = "@TARGETS_TO_BUILD@"
 
 # Support substitution of the tools_dir with user parameters. This is
 # used when we can't determine the tool dir at configuration time.
diff --git a/flang/tools/flang-driver/fc1_main.cpp b/flang/tools/flang-driver/fc1_main.cpp
index ecdceb9..ff9d3c7 100644
--- a/flang/tools/flang-driver/fc1_main.cpp
+++ b/flang/tools/flang-driver/fc1_main.cpp
@@ -20,6 +20,7 @@
 #include "llvm/Option/Arg.h"
 #include "llvm/Option/ArgList.h"
 #include "llvm/Option/OptTable.h"
+#include "llvm/Support/TargetSelect.h"
 
 #include <cstdio>
 
@@ -48,6 +49,11 @@ int fc1_main(llvm::ArrayRef<const char *> argv, const char *argv0) {
   bool success =
       CompilerInvocation::CreateFromArgs(flang->invocation(), argv, diags);
 
+  // Initialize targets first, so that --version shows registered targets.
+  llvm::InitializeAllTargets();
+  llvm::InitializeAllTargetMCs();
+  llvm::InitializeAllAsmPrinters();
+
   diagsBuffer->FlushDiagnostics(flang->diagnostics());
 
   if (!success)
diff --git a/flang/unittests/Frontend/CMakeLists.txt b/flang/unittests/Frontend/CMakeLists.txt
index b78adf3f..739412c 100644
--- a/flang/unittests/Frontend/CMakeLists.txt
+++ b/flang/unittests/Frontend/CMakeLists.txt
@@ -1,3 +1,7 @@
+set(LLVM_LINK_COMPONENTS
+  ${LLVM_TARGETS_TO_BUILD}
+)
+
 add_flang_unittest(FlangFrontendTests
   CompilerInstanceTest.cpp
   FrontendActionTest.cpp
diff --git a/flang/unittests/Frontend/FrontendActionTest.cpp b/flang/unittests/Frontend/FrontendActionTest.cpp
index 81a57be..b88f1ab 100644
--- a/flang/unittests/Frontend/FrontendActionTest.cpp
+++ b/flang/unittests/Frontend/FrontendActionTest.cpp
@@ -11,6 +11,7 @@
 #include "flang/Frontend/FrontendOptions.h"
 #include "flang/FrontendTool/Utils.h"
 #include "llvm/Support/FileSystem.h"
+#include "llvm/Support/TargetSelect.h"
 #include "llvm/Support/raw_ostream.h"
 
 #include "gtest/gtest.h"
@@ -188,4 +189,35 @@ TEST_F(FrontendActionTest, EmitLLVM) {
   EXPECT_TRUE(llvm::StringRef(outputFileBuffer.data())
                   .contains("define void @_QQmain()"));
 }
+
+TEST_F(FrontendActionTest, EmitAsm) {
+  // Populate the input file with the pre-defined input and flush it.
+  *(inputFileOs_) << "end program";
+  inputFileOs_.reset();
+
+  // Set-up the action kind.
+  compInst_.invocation().frontendOpts().programAction = EmitAssembly;
+  compInst_.invocation().preprocessorOpts().noReformat = true;
+
+  // Initialise LLVM backend
+  llvm::InitializeAllTargets();
+  llvm::InitializeAllTargetMCs();
+  llvm::InitializeAllAsmPrinters();
+
+  // Set-up the output stream. We are using output buffer wrapped as an output
+  // stream, as opposed to an actual file (or a file descriptor).
+  llvm::SmallVector<char, 256> outputFileBuffer;
+  std::unique_ptr<llvm::raw_pwrite_stream> outputFileStream(
+      new llvm::raw_svector_ostream(outputFileBuffer));
+  compInst_.set_outputStream(std::move(outputFileStream));
+
+  // Execute the action.
+  bool success = ExecuteCompilerInvocation(&compInst_);
+
+  // Validate the expected output.
+  EXPECT_TRUE(success);
+  EXPECT_TRUE(!outputFileBuffer.empty());
+
+  EXPECT_TRUE(llvm::StringRef(outputFileBuffer.data()).contains("_QQmain"));
+}
 } // namespace