[sanitizer] Skip stack symbolization when not required for print format

[tejohnson]

Summary

Adds a check to avoid symbolization when printing stack traces if the
stack_trace_format flag does not need it. While there is a symbolize
flag that can be turned off to skip some of the symbolization,
SymbolizePC() still unconditionally looks up the module name and offset.
Avoid invoking SymbolizePC() at all if not needed.

This is an efficiency improvement when dumping all stack traces as part
of the memory profiler in D87120, for large stripped apps where we want
to symbolize as a post pass.

[vitalybuka]

compiler-rt/lib/sanitizer_common/sanitizer_stacktrace_printer.cpp

109–111

maybe
and pass info as nullptr for frames without symbolization call
to crash if RenderNeedsSymbolization was wrong.

[tejohnson]

Implement suggestion

[leonardchan]

Hi! It seems that this change is leading to the undefined symbol errors we're seeing on our builders (https://luci-milo.appspot.com/p/fuchsia/builders/ci/clang-linux-x64/b8867057367989385504):

[716/753] Linking CXX shared library /b/s/w/ir/x/w/staging/llvm_build/lib/clang/12.0.0/lib/x86_64-unknown-fuchsia/libclang_rt.asan.so
FAILED: /b/s/w/ir/x/w/staging/llvm_build/lib/clang/12.0.0/lib/x86_64-unknown-fuchsia/libclang_rt.asan.so 
...
ld.lld: error: undefined symbol: __sanitizer::RenderNeedsSymbolization(char const*)
>>> referenced by sanitizer_stacktrace_libcdep.cpp:29 (compiler-rt/lib/sanitizer_common/sanitizer_stacktrace_libcdep.cpp:29)
>>>               compiler-rt/lib/sanitizer_common/CMakeFiles/RTSanitizerCommonSymbolizer.x86_64.dir/sanitizer_stacktrace_libcdep.cpp.obj:(__sanitizer::StackTrace::Print() const)
>>> referenced by sanitizer_stacktrace_libcdep.cpp:118 (compiler-rt/lib/sanitizer_common/sanitizer_stacktrace_libcdep.cpp:118)
>>>               compiler-rt/lib/sanitizer_common/CMakeFiles/RTSanitizerCommonSymbolizer.x86_64.dir/sanitizer_stacktrace_libcdep.cpp.obj:(__sanitizer_symbolize_pc)

ld.lld: error: undefined symbol: __sanitizer::RenderFrame(__sanitizer::InternalScopedString*, char const*, int, unsigned long, __sanitizer::AddressInfo const*, bool, char const*, char const*)
>>> referenced by sanitizer_stacktrace_libcdep.cpp:43 (compiler-rt/lib/sanitizer_common/sanitizer_stacktrace_libcdep.cpp:43)
>>>               compiler-rt/lib/sanitizer_common/CMakeFiles/RTSanitizerCommonSymbolizer.x86_64.dir/sanitizer_stacktrace_libcdep.cpp.obj:(__sanitizer::StackTrace::Print() const)
>>> referenced by sanitizer_stacktrace_libcdep.cpp:135 (compiler-rt/lib/sanitizer_common/sanitizer_stacktrace_libcdep.cpp:135)
>>>               compiler-rt/lib/sanitizer_common/CMakeFiles/RTSanitizerCommonSymbolizer.x86_64.dir/sanitizer_stacktrace_libcdep.cpp.obj:(__sanitizer_symbolize_pc)
>>> referenced by sanitizer_symbolizer_report.cpp:36 (compiler-rt/lib/sanitizer_common/sanitizer_symbolizer_report.cpp:36)
>>>               compiler-rt/lib/sanitizer_common/CMakeFiles/RTSanitizerCommonSymbolizer.x86_64.dir/sanitizer_symbolizer_report.cpp.obj:(__sanitizer::ReportErrorSummary(char const*, __sanitizer::AddressInfo const&, char const*))

Could you take a look? Thanks.

[tejohnson]

In D88361#2319693, @leonardchan wrote:

Hi! It seems that this change is leading to the undefined symbol errors we're seeing on our builders (https://luci-milo.appspot.com/p/fuchsia/builders/ci/clang-linux-x64/b8867057367989385504):

[716/753] Linking CXX shared library /b/s/w/ir/x/w/staging/llvm_build/lib/clang/12.0.0/lib/x86_64-unknown-fuchsia/libclang_rt.asan.so
FAILED: /b/s/w/ir/x/w/staging/llvm_build/lib/clang/12.0.0/lib/x86_64-unknown-fuchsia/libclang_rt.asan.so 
...
ld.lld: error: undefined symbol: __sanitizer::RenderNeedsSymbolization(char const*)
>>> referenced by sanitizer_stacktrace_libcdep.cpp:29 (compiler-rt/lib/sanitizer_common/sanitizer_stacktrace_libcdep.cpp:29)
>>>               compiler-rt/lib/sanitizer_common/CMakeFiles/RTSanitizerCommonSymbolizer.x86_64.dir/sanitizer_stacktrace_libcdep.cpp.obj:(__sanitizer::StackTrace::Print() const)
>>> referenced by sanitizer_stacktrace_libcdep.cpp:118 (compiler-rt/lib/sanitizer_common/sanitizer_stacktrace_libcdep.cpp:118)
>>>               compiler-rt/lib/sanitizer_common/CMakeFiles/RTSanitizerCommonSymbolizer.x86_64.dir/sanitizer_stacktrace_libcdep.cpp.obj:(__sanitizer_symbolize_pc)

ld.lld: error: undefined symbol: __sanitizer::RenderFrame(__sanitizer::InternalScopedString*, char const*, int, unsigned long, __sanitizer::AddressInfo const*, bool, char const*, char const*)
>>> referenced by sanitizer_stacktrace_libcdep.cpp:43 (compiler-rt/lib/sanitizer_common/sanitizer_stacktrace_libcdep.cpp:43)
>>>               compiler-rt/lib/sanitizer_common/CMakeFiles/RTSanitizerCommonSymbolizer.x86_64.dir/sanitizer_stacktrace_libcdep.cpp.obj:(__sanitizer::StackTrace::Print() const)
>>> referenced by sanitizer_stacktrace_libcdep.cpp:135 (compiler-rt/lib/sanitizer_common/sanitizer_stacktrace_libcdep.cpp:135)
>>>               compiler-rt/lib/sanitizer_common/CMakeFiles/RTSanitizerCommonSymbolizer.x86_64.dir/sanitizer_stacktrace_libcdep.cpp.obj:(__sanitizer_symbolize_pc)
>>> referenced by sanitizer_symbolizer_report.cpp:36 (compiler-rt/lib/sanitizer_common/sanitizer_symbolizer_report.cpp:36)
>>>               compiler-rt/lib/sanitizer_common/CMakeFiles/RTSanitizerCommonSymbolizer.x86_64.dir/sanitizer_symbolizer_report.cpp.obj:(__sanitizer::ReportErrorSummary(char const*, __sanitizer::AddressInfo const&, char const*))

Could you take a look? Thanks.

I see the issue. Fuschia uses a different implementation of RenderFrame, I'll need to update that one too and add a dummy version of the new RenderNeedsSymbolization as well (looks like this version in sanitizer_symbolizer_markup.cpp is quite simple and doesn't attempt to symbolize at all, so RenderNeedsSymbolization can simply return false). Fix coming up.

[tejohnson]

In D88361#2319756, @tejohnson wrote:

In D88361#2319693, @leonardchan wrote:

Hi! It seems that this change is leading to the undefined symbol errors we're seeing on our builders (https://luci-milo.appspot.com/p/fuchsia/builders/ci/clang-linux-x64/b8867057367989385504):

[716/753] Linking CXX shared library /b/s/w/ir/x/w/staging/llvm_build/lib/clang/12.0.0/lib/x86_64-unknown-fuchsia/libclang_rt.asan.so
FAILED: /b/s/w/ir/x/w/staging/llvm_build/lib/clang/12.0.0/lib/x86_64-unknown-fuchsia/libclang_rt.asan.so 
...
ld.lld: error: undefined symbol: __sanitizer::RenderNeedsSymbolization(char const*)
>>> referenced by sanitizer_stacktrace_libcdep.cpp:29 (compiler-rt/lib/sanitizer_common/sanitizer_stacktrace_libcdep.cpp:29)
>>>               compiler-rt/lib/sanitizer_common/CMakeFiles/RTSanitizerCommonSymbolizer.x86_64.dir/sanitizer_stacktrace_libcdep.cpp.obj:(__sanitizer::StackTrace::Print() const)
>>> referenced by sanitizer_stacktrace_libcdep.cpp:118 (compiler-rt/lib/sanitizer_common/sanitizer_stacktrace_libcdep.cpp:118)
>>>               compiler-rt/lib/sanitizer_common/CMakeFiles/RTSanitizerCommonSymbolizer.x86_64.dir/sanitizer_stacktrace_libcdep.cpp.obj:(__sanitizer_symbolize_pc)

ld.lld: error: undefined symbol: __sanitizer::RenderFrame(__sanitizer::InternalScopedString*, char const*, int, unsigned long, __sanitizer::AddressInfo const*, bool, char const*, char const*)
>>> referenced by sanitizer_stacktrace_libcdep.cpp:43 (compiler-rt/lib/sanitizer_common/sanitizer_stacktrace_libcdep.cpp:43)
>>>               compiler-rt/lib/sanitizer_common/CMakeFiles/RTSanitizerCommonSymbolizer.x86_64.dir/sanitizer_stacktrace_libcdep.cpp.obj:(__sanitizer::StackTrace::Print() const)
>>> referenced by sanitizer_stacktrace_libcdep.cpp:135 (compiler-rt/lib/sanitizer_common/sanitizer_stacktrace_libcdep.cpp:135)
>>>               compiler-rt/lib/sanitizer_common/CMakeFiles/RTSanitizerCommonSymbolizer.x86_64.dir/sanitizer_stacktrace_libcdep.cpp.obj:(__sanitizer_symbolize_pc)
>>> referenced by sanitizer_symbolizer_report.cpp:36 (compiler-rt/lib/sanitizer_common/sanitizer_symbolizer_report.cpp:36)
>>>               compiler-rt/lib/sanitizer_common/CMakeFiles/RTSanitizerCommonSymbolizer.x86_64.dir/sanitizer_symbolizer_report.cpp.obj:(__sanitizer::ReportErrorSummary(char const*, __sanitizer::AddressInfo const&, char const*))

Could you take a look? Thanks.

I see the issue. Fuschia uses a different implementation of RenderFrame, I'll need to update that one too and add a dummy version of the new RenderNeedsSymbolization as well (looks like this version in sanitizer_symbolizer_markup.cpp is quite simple and doesn't attempt to symbolize at all, so RenderNeedsSymbolization can simply return false). Fix coming up.

Should be fixed by f775cb8994ccb367d68cdea5ce453d747e47d891.

[leonardchan]

Should be fixed by f775cb8994ccb367d68cdea5ce453d747e47d891.

Awesome! Thanks for the quick fix.

compiler-rt/lib/sanitizer_common/sanitizer_stacktrace_libcdep.cpp

 void StackTrace::Print() const {
   if (trace == nullptr || size == 0) {
     Printf("    <empty stack>\n\n");
     return;
   }
   InternalScopedString frame_desc(GetPageSizeCached() * 2);
   InternalScopedString dedup_token(GetPageSizeCached());
   int dedup_frames = common_flags()->dedup_token_length;
+  bool symbolize = RenderNeedsSymbolization(common_flags()->stack_trace_format);
   uptr frame_num = 0;
   for (uptr i = 0; i < size && trace[i]; i++) {
     // PCs in stack traces are actually the return addresses, that is,
     // addresses of the next instructions after the call.
     uptr pc = GetPreviousInstructionPc(trace[i]);
-    SymbolizedStack *frames = Symbolizer::GetOrInit()->SymbolizePC(pc);
+    SymbolizedStack *frames;
+    if (symbolize)
+      frames = Symbolizer::GetOrInit()->SymbolizePC(pc);
+    else
+      frames = SymbolizedStack::New(pc);
     CHECK(frames);
     for (SymbolizedStack *cur = frames; cur; cur = cur->next) {
       frame_desc.clear();
       RenderFrame(&frame_desc, common_flags()->stack_trace_format, frame_num++,
                   cur->info, common_flags()->symbolize_vs_style,
                   common_flags()->strip_path_prefix);
       Printf("%s\n", frame_desc.data());
       if (dedup_frames-- > 0) {

 using namespace __sanitizer;
 
 extern "C" {
 SANITIZER_INTERFACE_ATTRIBUTE
 void __sanitizer_symbolize_pc(uptr pc, const char *fmt, char *out_buf,
                               uptr out_buf_size) {
   if (!out_buf_size) return;
   pc = StackTrace::GetPreviousInstructionPc(pc);
-  SymbolizedStack *frame = Symbolizer::GetOrInit()->SymbolizePC(pc);
+  SymbolizedStack *frame;
+  if (RenderNeedsSymbolization(fmt))
+    frame = Symbolizer::GetOrInit()->SymbolizePC(pc);
+  else
+    frame = SymbolizedStack::New(pc);
   if (!frame) {
     internal_strncpy(out_buf, "<can't symbolize>", out_buf_size);
     out_buf[out_buf_size - 1] = 0;
     return;
   }
   InternalScopedString frame_desc(GetPageSizeCached());
   uptr frame_num = 0;
   // Reserve one byte for the final 0.

compiler-rt/lib/sanitizer_common/sanitizer_stacktrace_printer.h

 //   %L - prints location information: file/line/column, if it is known, or
 //        module+offset if it is known, or (<unknown module>) string.
 //   %M - prints module basename and offset, if it is known, or PC.
 void RenderFrame(InternalScopedString *buffer, const char *format, int frame_no,
                  const AddressInfo &info, bool vs_style,
                  const char *strip_path_prefix = "",
                  const char *strip_func_prefix = "");
 
+bool RenderNeedsSymbolization(const char *format);
+
 void RenderSourceLocation(InternalScopedString *buffer, const char *file,
                           int line, int column, bool vs_style,
                           const char *strip_path_prefix);
 
 void RenderModuleLocation(InternalScopedString *buffer, const char *module,
                           uptr offset, ModuleArch arch,
                           const char *strip_path_prefix);
 

compiler-rt/lib/sanitizer_common/sanitizer_stacktrace_printer.cpp

if (!internal_strcmp(function, "__libc_thr_sigsetmask"))
     return "pthread_sigmask";
 #endif
 
   return function;
 }
 
 static const char kDefaultFormat[] = "    #%n %p %F %L";
 
 void RenderFrame(InternalScopedString *buffer, const char *format, int frame_no,
                  const AddressInfo &info, bool vs_style,
                  const char *strip_path_prefix, const char *strip_func_prefix) {
   if (0 == internal_strcmp(format, "DEFAULT"))
     format = kDefaultFormat;
   for (const char *p = format; *p != '\0'; p++) {
     if (*p != '%') {
       buffer->append("%c", *p);
       continue;
     }
     p++;
for (const char *p = format; *p != '\0'; p++) {
     default:
       Report("Unsupported specifier in stack frame format: %c (0x%zx)!\n", *p,
              *p);
       Die();
     }
   }
 }
 
+bool RenderNeedsSymbolization(const char *format) {
+  if (0 == internal_strcmp(format, "DEFAULT"))
+    format = kDefaultFormat;
+  for (const char *p = format; *p != '\0'; p++) {
+    if (*p != '%')
+      continue;
+    p++;
+    switch (*p) {
+      case '%':
+        break;
+      case 'n':
+        // frame_no
+        break;
+      case 'p':
+        // address
+        break;
+      default:
+        return true;
+    }
+  }
+  return false;
+}
+
 void RenderData(InternalScopedString *buffer, const char *format,
                 const DataInfo *DI, const char *strip_path_prefix) {
   for (const char *p = format; *p != '\0'; p++) {
     if (*p != '%') {
       buffer->append("%c", *p);
       continue;
     }
     p++;

compiler-rt/test/sanitizer_common/TestCases/print-stack-trace.cpp

 // RUN: %clangxx -O0 %s -o %t && %env_tool_opts=stack_trace_format=DEFAULT %run %t 2>&1 | FileCheck %s
 // RUN: %clangxx -O3 %s -o %t && %env_tool_opts=stack_trace_format=DEFAULT %run %t 2>&1 | FileCheck %s
 // RUN: %env_tool_opts=stack_trace_format=frame%n_lineno%l %run %t 2>&1 | FileCheck %s --check-prefix=CUSTOM
 // RUN: %env_tool_opts=symbolize_inline_frames=false:stack_trace_format=DEFAULT %run %t 2>&1 | FileCheck %s --check-prefix=NOINLINE
+// RUN: %env_tool_opts=stack_trace_format='"frame:%n address:%%p"' %run %t 2>&1 | FileCheck %s --check-prefix=NOSYMBOLIZE
 
 // UNSUPPORTED: darwin
 
 // TODO(yln): temporary failing due to refactoring
 // UNSUPPORTED: ubsan
 
 #include <sanitizer/common_interface_defs.h>
 

 // CHECK: {{    #1 0x.* in FooBarBaz(\(\))? .*}}print-stack-trace.cpp:[[@LINE-8]]
 // CHECK: {{    #2 0x.* in main.*}}print-stack-trace.cpp:[[@LINE-5]]
 
 // CUSTOM: frame1_lineno[[@LINE-11]]
 // CUSTOM: frame2_lineno[[@LINE-8]]
 
 // NOINLINE: #0 0x{{.*}} in __sanitizer_print_stack_trace
 // NOINLINE: #1 0x{{.*}} in main{{.*}}print-stack-trace.cpp:[[@LINE-15]]
+
+// NOSYMBOLIZE: frame:0 address:{{0x.*}}
+// NOSYMBOLIZE: frame:1 address:{{0x.*}}
+// NOSYMBOLIZE: frame:2 address:{{0x.*}}