[PATCH] D70653: [AutoFDO] Properly merge context-sensitive profile of inlinee back to outlined function

[Wei Mi via Phabricator via llvm-commits]

wmi accepted this revision.
wmi added a comment.
This revision is now accepted and ready to land.

I rerun perf test and I don't see any performance change. Last run I saw very small improvement on latency in a benchmark. This is fine since the benchmark has some fluctuation by itself.



================
Comment at: llvm/include/llvm/ProfileData/SampleProf.h:394
     const FunctionSamples *R = nullptr;
-    for (const auto &NameFS : iter->second)
-      if (NameFS.second.getTotalSamples() >= MaxTotalSamples) {
-        MaxTotalSamples = NameFS.second.getTotalSamples();
-        R = &NameFS.second;
-      }
+    if (CalleeName.empty()) {
+      for (const auto &NameFS : iter->second)
----------------
Nit: have an early return when CalleeName.empty() is true.


================
Comment at: llvm/test/Transforms/SampleProfile/inline-mergeprof.ll:33-34
+  store i32 %y, i32* %y.addr, align 4
+  %0 = load i32, i32* %x.addr, align 4, !dbg !11
+  %1 = load i32, i32* %y.addr, align 4, !dbg !11
+  %add = add nsw i32 %0, %1, !dbg !11
----------------
run opt -instnamer for the test.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D70653/new/

https://reviews.llvm.org/D70653