[compiler-rt] [ctxprof] Auto root detection: trie for stack samples (PR #133106)

[Mircea Trofin via llvm-commits]

================
@@ -0,0 +1,90 @@
+//===- RootAutodetector.cpp - detect contextual profiling roots -----------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "RootAutoDetector.h"
+#include "CtxInstrProfiling.h"
+
+#include "sanitizer_common/sanitizer_common.h"
+#include "sanitizer_common/sanitizer_mutex.h"
+#include "sanitizer_common/sanitizer_placement_new.h"
+#include <assert.h>
+#include <dlfcn.h>
+#include <pthread.h>
+
+using namespace __ctx_profile;
+
+uptr PerThreadCallsiteTrie::getFctStartAddr(uptr CallsiteAddress) const {
+  // this requires --linkopt=-Wl,--export-dynamic
+  Dl_info Info;
+  if (dladdr(reinterpret_cast<const void *>(CallsiteAddress), &Info) != 0)
----------------
mtrofin wrote:

In the internal prototype I was caching the result, but I'm actually not 100% sure if I didn't optimize unnecessarily, so decided to introduce that later if needed. Didn't notice any meaningful qps change (on a significant internal binary) during root autodetection - this also happens at the end of the sampling, and we only care about the top portion (up to the first split) of the stacks we collect. 

https://github.com/llvm/llvm-project/pull/133106