[flang-commits] [flang] [flang] Support -mvscale-min and mvscale-max on all targets (PR #74633)

[via flang-commits]

llvmbot wrote:


<!--LLVM PR SUMMARY COMMENT-->

@llvm/pr-subscribers-flang-driver

Author: Philip Reames (preames)

<details>
<summary>Changes</summary>

We have other targets with scalable vectors (e.g.RISC-V), and there doesn't seem to be any particular reason these options can't be used on those targets.

---
Full diff: https://github.com/llvm/llvm-project/pull/74633.diff


1 Files Affected:

- (modified) flang/lib/Frontend/FrontendActions.cpp (+5-9) 


``````````diff
diff --git a/flang/lib/Frontend/FrontendActions.cpp b/flang/lib/Frontend/FrontendActions.cpp
index 1be95cc27f42c..a9eaae6dd1c49 100644
--- a/flang/lib/Frontend/FrontendActions.cpp
+++ b/flang/lib/Frontend/FrontendActions.cpp
@@ -824,8 +824,8 @@ void CodeGenAction::lowerHLFIRToFIR() {
 // TODO: We should get this from TargetInfo. However, that depends on
 // too much of clang, so for now, replicate the functionality.
 static std::optional<std::pair<unsigned, unsigned>>
-getVScaleRange(CompilerInstance &ci,
-               const Fortran::frontend::LangOptions &langOpts) {
+getVScaleRange(CompilerInstance &ci) {
+  const auto &langOpts = ci.getInvocation().getLangOpts();
   if (langOpts.VScaleMin || langOpts.VScaleMax)
     return std::pair<unsigned, unsigned>(
         langOpts.VScaleMin ? langOpts.VScaleMin : 1, langOpts.VScaleMax);
@@ -860,13 +860,9 @@ void CodeGenAction::generateLLVMIR() {
   const auto targetOpts = ci.getInvocation().getTargetOpts();
   const llvm::Triple triple(targetOpts.triple);
 
-  // Only get the vscale range if AArch64.
-  if (triple.isAArch64()) {
-    auto langOpts = ci.getInvocation().getLangOpts();
-    if (auto vsr = getVScaleRange(ci, langOpts)) {
-      config.VScaleMin = vsr->first;
-      config.VScaleMax = vsr->second;
-    }
+  if (auto vsr = getVScaleRange(ci)) {
+    config.VScaleMin = vsr->first;
+    config.VScaleMax = vsr->second;
   }
 
   // Create the pass pipeline

``````````

</details>


https://github.com/llvm/llvm-project/pull/74633