Add support for demangling C++ frames' symbols

This adds a new optional (but on-by-default) dependency on the `cpp_demangle`
crate. When this feature is enabled, if demangling a frame's symbol as a Rust
symbol fails, then we attempt to demangle it as a C++ symbol.

Author: fitzgen

Cargo.toml

@@ -16,12 +16,15 @@ build = "build.rs"
 [dependencies]
 libc = "0.2"
 cfg-if = "0.1"
-rustc-demangle = "0.1"
+rustc-demangle = "0.1.4"
 
 # Optionally enable the ability to serialize a `Backtrace`
 serde = { version = "0.8", optional = true }
 rustc-serialize = { version = "0.3", optional = true }
 
+# Optionally demangle C++ frames' symbols in backtraces.
+cpp_demangle = { version = "0.2", optional = true }
+
 [target.'cfg(windows)'.dependencies]
 dbghelp-sys = { version = "0.2", optional = true }
 kernel32-sys = { version = "0.2", optional = true }
@@ -32,6 +35,7 @@ backtrace-sys = { path = "backtrace-sys", version = "0.1.3", optional = true }
 
 [build-dependencies]
 serde_codegen = { version = "0.8", optional = true }
+gcc = { version = "0.3.43", optional = true }
 
 # Each feature controls the two phases of finding a backtrace: getting a
 # backtrace and then resolving instruction pointers to symbols. The default
@@ -42,7 +46,7 @@ serde_codegen = { version = "0.8", optional = true }
 # Note that not all features are available on all platforms, so even though a
 # feature is enabled some other feature may be used instead.
 [features]
-default = ["libunwind", "libbacktrace", "coresymbolication", "dladdr", "dbghelp"]
+default = ["libunwind", "libbacktrace", "coresymbolication", "dladdr", "dbghelp", "demangle_cpp"]
 
     #=======================================
     # Methods of acquiring a backtrace
@@ -76,9 +80,12 @@ default = ["libunwind", "libbacktrace", "coresymbolication", "dladdr", "dbghelp"
     #   enough on OSX.
     # - coresymbolication: this feature uses the undocumented core symbolication
     #   framework on OS X to symbolize.
+    # - demangle_cpp: this feature enables demangling C++ symbols from C++
+    #   frames.
     libbacktrace = ["backtrace-sys"]
     dladdr = []
     coresymbolication = []
+    demangle_cpp = ["cpp_demangle", "gcc"]
 
     #=======================================
     # Methods of serialization

build.rs

@@ -25,6 +25,9 @@ fn main() {
 
     expand_serde(out_dir);
     println!("cargo:rerun-if-changed=src/capture.rs");
+
+    compile_cpp();
+    println!("cargo:rerun-if-changed=cpp/trampoline.cpp");
 }
 
 #[cfg(not(feature = "serialize-serde"))]
@@ -49,3 +52,23 @@ fn expand_serde(out_dir: &Path) {
         serde_codegen::expand(&out_dir.join("tmp.rs"), &dst).unwrap();
     }).unwrap().join().unwrap();
 }
+
+// TODO: how to only do this for tests? I tried to do:
+//
+// > #[cfg(all(test, feature = "demangle_cpp"))]
+//
+// but it did not work...
+#[cfg(feature = "demangle_cpp")]
+fn compile_cpp() {
+    extern crate gcc;
+
+    gcc::Config::new()
+        .cpp(true)
+        .debug(true)
+        .opt_level(0)
+        .file("cpp/trampoline.cpp")
+        .compile("libcpptrampoline.a");
+}
+
+#[cfg(not(feature = "demangle_cpp"))]
+fn compile_cpp() {}

cpp/trampoline.cpp

@@ -0,0 +1,14 @@
+#include <stdio.h>
+
+namespace space {
+    template <typename FuncT>
+    void templated_trampoline(FuncT func) {
+        func();
+    }
+}
+
+typedef void (*FuncPtr)();
+
+extern "C" void cpp_trampoline(FuncPtr func) {
+    space::templated_trampoline(func);
+}

src/lib.rs

@@ -85,6 +85,9 @@ extern crate cfg_if;
 
 extern crate rustc_demangle;
 
+#[cfg(feature = "cpp_demangle")]
+extern crate cpp_demangle;
+
 #[allow(dead_code)] // not used everywhere
 #[cfg(unix)]
 #[macro_use]

src/symbolize/mod.rs

@@ -1,8 +1,10 @@
 use std::fmt;
+#[cfg(not(feature = "cpp_demangle"))]
+use std::marker::PhantomData;
 use std::os::raw::c_void;
 use std::path::Path;
 use std::str;
-use rustc_demangle::{demangle, Demangle};
+use rustc_demangle::{try_demangle, Demangle};
 
 /// Resolve an address to a symbol, passing the symbol to the specified
 /// closure.
@@ -108,26 +110,74 @@ impl fmt::Debug for Symbol {
     }
 }
 
+
+cfg_if! {
+    if #[cfg(feature = "cpp_demangle")] {
+        // Maybe a parsed C++ symbol, if parsing the mangled symbol as Rust
+        // failed.
+        struct OptionCppSymbol<'a>(Option<::cpp_demangle::BorrowedSymbol<'a>>);
+
+        impl<'a> OptionCppSymbol<'a> {
+            fn parse(input: &'a [u8]) -> OptionCppSymbol<'a> {
+                OptionCppSymbol(::cpp_demangle::BorrowedSymbol::new(input).ok())
+            }
+
+            fn none() -> OptionCppSymbol<'a> {
+                OptionCppSymbol(None)
+            }
+        }
+    } else {
+        // Make sure to keep this zero-sized, so that the `cpp_demangle` feature
+        // has no cost when disabled.
+        struct OptionCppSymbol<'a>(PhantomData<&'a ()>);
+
+        impl<'a> OptionCppSymbol<'a> {
+            fn parse(_: &'a [u8]) -> OptionCppSymbol<'a> {
+                OptionCppSymbol(PhantomData)
+            }
+
+            fn none() -> OptionCppSymbol<'a> {
+                OptionCppSymbol(PhantomData)
+            }
+        }
+    }
+}
+
 /// A wrapper around a symbol name to provide ergonomic accessors to the
 /// demangled name, the raw bytes, the raw string, etc.
 pub struct SymbolName<'a> {
     bytes: &'a [u8],
     demangled: Option<Demangle<'a>>,
+    cpp_demangled: OptionCppSymbol<'a>,
 }
 
 impl<'a> SymbolName<'a> {
     /// Creates a new symbol name from the raw underlying bytes.
     pub fn new(bytes: &'a [u8]) -> SymbolName<'a> {
-        let demangled = str::from_utf8(bytes).ok().map(demangle);
+        let str_bytes = str::from_utf8(bytes).ok();
+        let demangled = str_bytes.and_then(|s| try_demangle(s).ok());
+
+        let cpp = if demangled.is_none() {
+            OptionCppSymbol::parse(bytes)
+        } else {
+            OptionCppSymbol::none()
+        };
+
         SymbolName {
             bytes: bytes,
             demangled: demangled,
+            cpp_demangled: cpp,
         }
     }
 
-    /// Returns the raw symbol name as `&str` if the symbols is valid utf-8.
+    /// Returns the raw symbol name as a `str` if the symbols is valid utf-8.
     pub fn as_str(&self) -> Option<&'a str> {
-        self.demangled.as_ref().map(|s| s.as_str())
+        self.demangled
+            .as_ref()
+            .map(|s| s.as_str())
+            .or_else(|| {
+                str::from_utf8(self.bytes).ok()
+            })
     }
 
     /// Returns the raw symbol name as a list of bytes
@@ -136,22 +186,54 @@ impl<'a> SymbolName<'a> {
     }
 }
 
-impl<'a> fmt::Display for SymbolName<'a> {
-    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
-        if let Some(ref s) = self.demangled {
-            s.fmt(f)
-        } else {
-            String::from_utf8_lossy(self.bytes).fmt(f)
+cfg_if! {
+    if #[cfg(feature = "cpp_demangle")] {
+        impl<'a> fmt::Display for SymbolName<'a> {
+            fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+                if let Some(ref s) = self.demangled {
+                    s.fmt(f)
+                } else if let Some(ref cpp) = self.cpp_demangled.0 {
+                    cpp.fmt(f)
+                } else {
+                    String::from_utf8_lossy(self.bytes).fmt(f)
+                }
+            }
+        }
+    } else {
+        impl<'a> fmt::Display for SymbolName<'a> {
+            fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+                if let Some(ref s) = self.demangled {
+                    s.fmt(f)
+                } else {
+                    String::from_utf8_lossy(self.bytes).fmt(f)
+                }
+            }
         }
     }
 }
 
-impl<'a> fmt::Debug for SymbolName<'a> {
-    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
-        if let Some(ref s) = self.demangled {
-            s.fmt(f)
-        } else {
-            String::from_utf8_lossy(self.bytes).fmt(f)
+cfg_if! {
+    if #[cfg(feature = "cpp_demangle")] {
+        impl<'a> fmt::Debug for SymbolName<'a> {
+            fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+                if let Some(ref s) = self.demangled {
+                    s.fmt(f)
+                } else if let Some(ref cpp) = self.cpp_demangled.0 {
+                    fmt::Display::fmt(cpp, f)
+                } else {
+                    String::from_utf8_lossy(self.bytes).fmt(f)
+                }
+            }
+        }
+    } else {
+        impl<'a> fmt::Debug for SymbolName<'a> {
+            fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+                if let Some(ref s) = self.demangled {
+                    s.fmt(f)
+                } else {
+                    String::from_utf8_lossy(self.bytes).fmt(f)
+                }
+            }
         }
     }
 }
@@ -185,4 +267,3 @@ cfg_if! {
         use self::noop::Symbol as SymbolImp;
     }
 }
-

tests/smoke.rs

@@ -165,3 +165,71 @@ fn is_serde() {
     is_serialize::<backtrace::Backtrace>();
     is_deserialize::<backtrace::Backtrace>();
 }
+
+#[test]
+#[cfg(all(not(target_os = "windows"), feature = "demangle_cpp"))]
+fn smoke_test_cpp() {
+    use std::sync::atomic::{ATOMIC_BOOL_INIT, AtomicBool, Ordering};
+
+    extern "C" {
+        fn cpp_trampoline(func: extern "C" fn()) -> ();
+    }
+
+    static RAN_ASSERTS: AtomicBool = ATOMIC_BOOL_INIT;
+
+    extern "C" fn assert_cpp_frames() {
+        let mut physical_frames = Vec::new();
+        backtrace::trace(|cx| {
+            physical_frames.push(cx.ip());
+
+            // We only want to capture this closure's frame, assert_cpp_frames,
+            // space::templated_trampoline, and cpp_trampoline. Those are
+            // logical frames, which might be inlined into fewer physical
+            // frames, so we may end up with extra logical frames after
+            // resolving these.
+            physical_frames.len() < 4
+        });
+
+        let names: Vec<_> = physical_frames.into_iter()
+            .flat_map(|ip| {
+                let mut logical_frame_names = vec![];
+
+                backtrace::resolve(ip, |sym| {
+                    let sym_name = sym.name().expect("Should have a symbol name");
+                    let demangled = sym_name.to_string();
+                    logical_frame_names.push(demangled);
+                });
+
+                assert!(!logical_frame_names.is_empty(),
+                        "Should have resolved at least one symbol for the physical frame");
+
+                logical_frame_names
+            })
+            // Skip the backtrace::trace closure and assert_cpp_frames, and then
+            // take the two C++ frame names.
+            .skip(2)
+            .take(2)
+            .collect();
+
+        println!("actual names = {:#?}", names);
+
+        let expected = [
+            "void space::templated_trampoline<void (*)()>(void (*)())",
+            "cpp_trampoline",
+        ];
+        println!("expected names = {:#?}", expected);
+
+        assert_eq!(names.len(), expected.len());
+        for (actual, expected) in names.iter().zip(expected.iter()) {
+            assert_eq!(actual, expected);
+        }
+
+        RAN_ASSERTS.store(true, Ordering::SeqCst);
+    }
+
+    assert!(!RAN_ASSERTS.load(Ordering::SeqCst));
+    unsafe {
+        cpp_trampoline(assert_cpp_frames);
+    }
+    assert!(RAN_ASSERTS.load(Ordering::SeqCst));
+}