[sanitizer_symbolizer] Add initial symbolizer markup support for linux.

[avillega]

This is part of a stack of PRs to add support for symbolizer
markup in linux.

You can check the symbolizer markup specification at:
https://llvm.org/docs/SymbolizerMarkupFormat.html

[llvmbot]

@llvm/pr-subscribers-compiler-rt-sanitizer

Author: Andres Villegas (avillega)

Changes

This is part of a stack of PRs to add support for symbolizer
markup in linux.

You can check the symbolizer markup specification at:
https://llvm.org/docs/SymbolizerMarkupFormat.html

---

Full diff: https://github.com/llvm/llvm-project/pull/73193.diff

5 Files Affected:

• (modified) compiler-rt/lib/sanitizer_common/sanitizer_flags.inc (+4)
• (modified) compiler-rt/lib/sanitizer_common/sanitizer_stacktrace_printer.cpp (+4)
• (modified) compiler-rt/lib/sanitizer_common/sanitizer_symbolizer_markup.cpp (+19)
• (modified) compiler-rt/lib/sanitizer_common/sanitizer_symbolizer_markup.h (+23)
• (modified) compiler-rt/lib/sanitizer_common/sanitizer_symbolizer_posix_libcdep.cpp (+7)

diff --git a/compiler-rt/lib/sanitizer_common/sanitizer_flags.inc b/compiler-rt/lib/sanitizer_common/sanitizer_flags.inc
index 949bdbd148b6b89..7d0c7c4b63c9162 100644
--- a/compiler-rt/lib/sanitizer_common/sanitizer_flags.inc
+++ b/compiler-rt/lib/sanitizer_common/sanitizer_flags.inc
@@ -275,3 +275,7 @@ COMMON_FLAG(bool, test_only_emulate_no_memorymap, false,
 // program.
 COMMON_FLAG(bool, test_only_replace_dlopen_main_program, false,
             "TEST ONLY replace dlopen(<main program>,...) with dlopen(NULL)")
+
+COMMON_FLAG(bool, enable_symbolizer_markup, SANITIZER_FUCHSIA,
+            "Use sanitizer symbolizer markup, available on Linux "
+            "and always set true for fuchsia.")
diff --git a/compiler-rt/lib/sanitizer_common/sanitizer_stacktrace_printer.cpp b/compiler-rt/lib/sanitizer_common/sanitizer_stacktrace_printer.cpp
index 88f186b9c20c105..748d832ccc211d0 100644
--- a/compiler-rt/lib/sanitizer_common/sanitizer_stacktrace_printer.cpp
+++ b/compiler-rt/lib/sanitizer_common/sanitizer_stacktrace_printer.cpp
@@ -16,6 +16,7 @@
 #include "sanitizer_file.h"
 #include "sanitizer_flags.h"
 #include "sanitizer_fuchsia.h"
+#include "sanitizer_symbolizer_markup.h"
 
 namespace __sanitizer {
 
@@ -62,6 +63,9 @@ const char *StackTracePrinter::StripFunctionName(const char *function) {
 #if !SANITIZER_SYMBOLIZER_MARKUP
 
 StackTracePrinter *StackTracePrinter::NewStackTracePrinter() {
+  if (common_flags()->enable_symbolizer_markup)
+    return new (GetGlobalLowLevelAllocator()) MarkupStackTracePrinter();
+
   return new (GetGlobalLowLevelAllocator()) FormattedStackTracePrinter();
 }
 
diff --git a/compiler-rt/lib/sanitizer_common/sanitizer_symbolizer_markup.cpp b/compiler-rt/lib/sanitizer_common/sanitizer_symbolizer_markup.cpp
index c364e1e300225b9..1627908185f2009 100644
--- a/compiler-rt/lib/sanitizer_common/sanitizer_symbolizer_markup.cpp
+++ b/compiler-rt/lib/sanitizer_common/sanitizer_symbolizer_markup.cpp
@@ -45,4 +45,23 @@ void MarkupStackTracePrinter::RenderFrame(InternalScopedString *buffer,
   buffer->AppendF(kFormatFrame, frame_no, address);
 }
 
+bool MarkupSymbolizerTool::SymbolizePC(uptr addr, SymbolizedStack *stack) {
+  char buffer[kFormatFunctionMax];
+  internal_snprintf(buffer, sizeof(buffer), kFormatFunction, addr);
+  stack->info.function = internal_strdup(buffer);
+  return true;
+}
+
+bool MarkupSymbolizerTool::SymbolizeData(uptr addr, DataInfo *info) {
+  info->Clear();
+  info->start = addr;
+  return true;
+}
+
+const char *MarkupSymbolizerTool::Demangle(const char *name) {
+  static char buffer[kFormatDemangleMax];
+  internal_snprintf(buffer, sizeof(buffer), kFormatDemangle, name);
+  return buffer;
+}
+
 }  // namespace __sanitizer
diff --git a/compiler-rt/lib/sanitizer_common/sanitizer_symbolizer_markup.h b/compiler-rt/lib/sanitizer_common/sanitizer_symbolizer_markup.h
index 7cebe520e9bebb4..07630d0b3bdf417 100644
--- a/compiler-rt/lib/sanitizer_common/sanitizer_symbolizer_markup.h
+++ b/compiler-rt/lib/sanitizer_common/sanitizer_symbolizer_markup.h
@@ -16,6 +16,7 @@
 #include "sanitizer_common.h"
 #include "sanitizer_stacktrace_printer.h"
 #include "sanitizer_symbolizer.h"
+#include "sanitizer_symbolizer_internal.h"
 
 namespace __sanitizer {
 
@@ -40,6 +41,28 @@ class MarkupStackTracePrinter : public StackTracePrinter {
   ~MarkupStackTracePrinter() {}
 };
 
+class MarkupSymbolizerTool final : public SymbolizerTool {
+ public:
+  // This is used in some places for suppression checking, which we
+  // don't really support for Fuchsia.  It's also used in UBSan to
+  // identify a PC location to a function name, so we always fill in
+  // the function member with a string containing markup around the PC
+  // value.
+  // TODO(mcgrathr): Under SANITIZER_GO, it's currently used by TSan
+  // to render stack frames, but that should be changed to use
+  // RenderStackFrame.
+  bool SymbolizePC(uptr addr, SymbolizedStack *stack) override;
+
+  // Always claim we succeeded, so that RenderDataInfo will be called.
+  bool SymbolizeData(uptr addr, DataInfo *info) override;
+
+  // May return NULL if demangling failed.
+  // This is used by UBSan for type names, and by ASan for global variable
+  // names. It's expected to return a static buffer that will be reused on each
+  // call.
+  const char *Demangle(const char *name) override;
+};
+
 }  // namespace __sanitizer
 
 #endif  // SANITIZER_SYMBOLIZER_MARKUP_H
diff --git a/compiler-rt/lib/sanitizer_common/sanitizer_symbolizer_posix_libcdep.cpp b/compiler-rt/lib/sanitizer_common/sanitizer_symbolizer_posix_libcdep.cpp
index d92349c04fffabd..28f11352a6b5bbe 100644
--- a/compiler-rt/lib/sanitizer_common/sanitizer_symbolizer_posix_libcdep.cpp
+++ b/compiler-rt/lib/sanitizer_common/sanitizer_symbolizer_posix_libcdep.cpp
@@ -12,6 +12,7 @@
 //===----------------------------------------------------------------------===//
 
 #include "sanitizer_platform.h"
+#include "sanitizer_symbolizer_markup.h"
 #if SANITIZER_POSIX
 #  include <dlfcn.h>  // for dlsym()
 #  include <errno.h>
@@ -475,6 +476,12 @@ static void ChooseSymbolizerTools(IntrusiveList<SymbolizerTool> *list,
     VReport(2, "Symbolizer is disabled.\n");
     return;
   }
+  if (common_flags()->enable_symbolizer_markup) {
+    VReport(2, "Using symbolizer markup");
+    SymbolizerTool *tool = new (*allocator) MarkupSymbolizerTool();
+    CHECK(tool);
+    list->push_back(tool);
+  }
   if (IsAllocatorOutOfMemory()) {
     VReport(2, "Cannot use internal symbolizer: out of memory\n");
   } else if (SymbolizerTool *tool = InternalSymbolizer::get(allocator)) {

[vitalybuka]

LGTM. but looks like it can be squashed with sanitizer_common test

[petrhosek]

Super minor nit, but Fuchsia should be capitalized.

[avillega]

LGTM. but looks like it can be squashed with sanitizer_common test

The tests also test Emitting contextual elements, so I will squash the tests with the RenderContextual elements PR.