[llvm] [MemProf] Extend MemProfUse pass to make use of data access profiles to partition data (PR #151238)

[Mingming Liu via llvm-commits]

================
@@ -752,3 +773,58 @@ PreservedAnalyses MemProfUsePass::run(Module &M, ModuleAnalysisManager &AM) {
 
   return PreservedAnalyses::none();
 }
+
+bool MemProfUsePass::annotateGlobalVariables(
+    Module &M, const memprof::DataAccessProfData *DataAccessProf) {
+  if (!AnnotationStaticDataPrefix || M.globals().empty() || !DataAccessProf)
+    return false;
+
+  bool Changed = false;
+  for (GlobalVariable &GVar : M.globals()) {
+    assert(!GVar.getSectionPrefix().has_value() &&
+           "GVar shouldn't have section prefix yet");
+    if (GVar.isDeclarationForLinker())
+      continue;
+
+    StringRef Name = GVar.getName();
+    // Skip string literals whose mangled names doesn't stay stable across
+    // binary releases.
+    // TODO: Track string content hash in the profiles and compute it inside the
+    // compiler to categeorize the hotness string literals.
+    if (Name.starts_with(".str"))
+      continue;
+
+    // DataAccessProfRecord's look-up methods will canonicalize the variable
+    // name before looking up methods, so optimizer doesn't need to do it.
+    std::optional<DataAccessProfRecord> Record =
+        DataAccessProf->getProfileRecord(Name);
+    // Annotate a global variable as hot if it has non-zero sampled count, and
+    // annotate it as cold if it's seen in the profiled binary
+    // file but doesn't have any access sample.
+    if (Record && Record->AccessCount > 0) {
+      GVar.setSectionPrefix("hot");
+      Changed = true;
+    } else if (DataAccessProf->isKnownColdSymbol(Name)) {
+      GVar.setSectionPrefix("unlikely");
+      Changed = true;
+    }
+  }
+
+  // Optimization remark emitter requires a llvm::Function, but it's not well
+  // defined to associate a global variable with a function. So we just print
+  // out the static data section prefix in errs().
+  if (PrintStaticDataPrefix) {
----------------
mingmingl-llvm wrote:

done.

errs() was originally preferred over LLVM_DEBUG since it's easier to enable errs() with an option in a release build compiler (i.e., one that doesn't enable assertion), but both should work reasonably well when it comes to debugging.

https://github.com/llvm/llvm-project/pull/151238