[PATCH] D110891: [inliner] Mandatory inlining decisions produce remarks

Arthur Eubanks via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Fri Oct 1 11:35:50 PDT 2021


aeubanks added inline comments.


================
Comment at: llvm/lib/Analysis/InlineAdvisor.cpp:72-89
+  void recordUnsuccessfulInliningImpl(const InlineResult &Result) override {
+    if (IsInliningRecommended)
+      ORE.emit([&]() {
+        return OptimizationRemarkMissed(DEBUG_TYPE, "NotInlined", DLoc, Block)
+               << "'" << NV("Callee", Callee) << "' is not AlwaysInline into '"
+               << NV("Caller", Caller)
+               << "': " << NV("Reason", Result.getFailureReason());
----------------
mtrofin wrote:
> aeubanks wrote:
> > can we add a test for these?
> I think that would be tricky, because they should not actually happen - the way we determine whether a site is alwaysinlinable checks (but not thoroughly) for legality. Let me see if I can find a regression test. It may be we can synthesize such a case in IR only, though, so not much of a help for the frontend tests?
yeah some IR tests is what I was thinking


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D110891/new/

https://reviews.llvm.org/D110891



More information about the llvm-commits mailing list