[llvm] [llvm-profgen] Loading binary functions from .symtab when DWARF info is incomplete (PR #163654)

[via llvm-commits]

================
@@ -820,6 +830,65 @@ void ProfiledBinary::populateSymbolAddressList(const ObjectFile *Obj) {
   }
 }
 
+void ProfiledBinary::loadSymbolsFromSymtab(const ObjectFile *Obj) {
+  // Load binary functions from symbol table when Debug info is incomplete.
+  // Strip the internal suffixes which are not reflected in the DWARF info.
+  const SmallVector<StringRef, 10> Suffixes(
+      {// Internal suffixes from CoroSplit pass
+       ".cleanup", ".destroy", ".resume",
+       // Internal suffixes from Bolt
+       ".cold", ".warm",
+       // Compiler/LTO internal
+       ".llvm.", ".part.", ".isra.", ".constprop.", ".lto_priv."});
+  StringRef FileName = Obj->getFileName();
+  for (const SymbolRef &Symbol : Obj->symbols()) {
+    const SymbolRef::Type Type = unwrapOrError(Symbol.getType(), FileName);
+    const uint64_t StartAddr = unwrapOrError(Symbol.getAddress(), FileName);
+    const StringRef Name = unwrapOrError(Symbol.getName(), FileName);
+    uint64_t Size = 0;
+    if (isa<ELFObjectFileBase>(Symbol.getObject())) {
+      ELFSymbolRef ElfSymbol(Symbol);
+      Size = ElfSymbol.getSize();
+    }
+
+    if (Size == 0 || Type != SymbolRef::ST_Function)
+      continue;
+
+    const StringRef SymName =
+        FunctionSamples::getCanonicalFnName(Name, Suffixes);
+
+    auto Range = findFuncRange(StartAddr);
+    if (!Range || Range->StartAddress != StartAddr) {
+      // Function from symbol table not found previously in DWARF, store ranges.
+      auto Ret = BinaryFunctions.emplace(SymName, BinaryFunction());
+      auto &Func = Ret.first->second;
+      if (Ret.second) {
+        Func.FuncName = Ret.first->first;
+        HashBinaryFunctions[MD5Hash(StringRef(SymName))] = &Func;
+      }
+
+      Func.FromSymtab = true;
+      Func.Ranges.emplace_back(StartAddr, StartAddr + Size);
----------------
WenleiHe wrote:

So if a range is found but start address differ, we create a new range, but isn't this creating an overlapping range with existing one (from dwarf)? We should never have two ranges that covers the same address (i.e. ranges must not overlap). 

https://github.com/llvm/llvm-project/pull/163654