Arch-independent demangling and add gnuv2_demangle for old g++ projects

objdiff-core/Cargo.toml

@@ -41,7 +41,8 @@ any-arch = [
     "dep:regex",
     "dep:similar",
     "dep:syn",
-    "dep:encoding_rs"
+    "dep:encoding_rs",
+    "demangler",
 ]
 bindings = [
     "dep:prost",
@@ -88,38 +89,37 @@ std = [
 ]
 mips = [
     "any-arch",
-    "dep:cpp_demangle",
-    "dep:cwdemangle",
     "dep:rabbitizer",
 ]
 ppc = [
     "any-arch",
-    "dep:cwdemangle",
     "dep:cwextab",
     "dep:powerpc",
     "dep:rlwinmdec",
 ]
 x86 = [
     "any-arch",
-    "dep:cpp_demangle",
     "dep:iced-x86",
-    "dep:msvc-demangler",
 ]
 arm = [
     "any-arch",
     "dep:arm-attr",
-    "dep:cpp_demangle",
     "dep:unarm",
 ]
 arm64 = [
     "any-arch",
-    "dep:cpp_demangle",
     "dep:yaxpeax-arch",
     "dep:yaxpeax-arm",
 ]
 superh = [
     "any-arch",
 ]
+demangler = [
+    "dep:cpp_demangle",
+    "dep:cwdemangle",
+    "dep:gnuv2_demangle",
+    "dep:msvc-demangler",
+]
 
 [package.metadata.docs.rs]
 features = ["all"]
@@ -157,6 +157,7 @@ rlwinmdec = { version = "1.1", optional = true }
 
 # mips
 rabbitizer = { version = "2.0.0-alpha.4", default-features = false, features = ["all_extensions"], optional = true }
+gnuv2_demangle ={ version = "0.1.0", optional = true }
 
 # x86
 cpp_demangle = { version = "0.4", default-features = false, features = ["alloc"], optional = true }

objdiff-core/config-schema.json

@@ -25,6 +25,37 @@
         }
       ]
     },
+    {
+      "id": "demangler",
+      "type": "choice",
+      "default": "auto",
+      "name": "Demangler",
+      "description": "Which demangler should be used to demangle each symbol.",
+      "items": [
+        {
+          "value": "auto",
+          "name": "Auto",
+          "description": "Try to automatically guess the mangling format."
+        },
+        {
+          "value": "codewarrior",
+          "name": "CodeWarrior"
+        },
+        {
+          "value": "msvc",
+          "name": "MSVC"
+        },
+        {
+          "value": "itanium",
+          "name": "Itanium"
+        },
+        {
+          "value": "gnu_v2",
+          "name": "GNU g++ (V2)",
+          "description": "Use the old GNU mangling ABI. Used up to g++ 2.9.x"
+        }
+      ]
+    },
     {
       "id": "analyzeDataFlow",
       "type": "boolean",
@@ -259,6 +290,7 @@
       "name": "General",
       "properties": [
         "functionRelocDiffs",
+        "demangler",
         "spaceBetweenArgs",
         "combineDataSections",
         "combineTextSections"

objdiff-core/src/arch/arm.rs

@@ -1,9 +1,4 @@
-use alloc::{
-    collections::BTreeMap,
-    format,
-    string::{String, ToString},
-    vec::Vec,
-};
+use alloc::{collections::BTreeMap, format, string::ToString, vec::Vec};
 
 use anyhow::{Result, bail};
 use arm_attr::{BuildAttrs, enums::CpuArch, tag::Tag};
@@ -409,12 +404,6 @@ impl Arch for ArchArm {
         }
     }
 
-    fn demangle(&self, name: &str) -> Option<String> {
-        cpp_demangle::Symbol::new(name)
-            .ok()
-            .and_then(|s| s.demangle(&cpp_demangle::DemangleOptions::default()).ok())
-    }
-
     fn reloc_name(&self, flags: RelocationFlags) -> Option<&'static str> {
         match flags {
             RelocationFlags::Elf(r_type) => match r_type {

objdiff-core/src/arch/arm64.rs

@@ -1,8 +1,4 @@
-use alloc::{
-    format,
-    string::{String, ToString},
-    vec::Vec,
-};
+use alloc::{format, string::ToString, vec::Vec};
 use core::cmp::Ordering;
 
 use anyhow::Result;
@@ -108,12 +104,6 @@ impl Arch for ArchArm64 {
         Ok(())
     }
 
-    fn demangle(&self, name: &str) -> Option<String> {
-        cpp_demangle::Symbol::new(name)
-            .ok()
-            .and_then(|s| s.demangle(&cpp_demangle::DemangleOptions::default()).ok())
-    }
-
     fn reloc_name(&self, flags: RelocationFlags) -> Option<&'static str> {
         match flags {
             RelocationFlags::Elf(r_type) => match r_type {

objdiff-core/src/arch/mips.rs

@@ -1,6 +1,6 @@
 use alloc::{
     collections::{BTreeMap, BTreeSet},
-    string::{String, ToString},
+    string::ToString,
     vec::Vec,
 };
 
@@ -304,13 +304,6 @@ impl Arch for ArchMips {
         }
     }
 
-    fn demangle(&self, name: &str) -> Option<String> {
-        cpp_demangle::Symbol::new(name)
-            .ok()
-            .and_then(|s| s.demangle(&cpp_demangle::DemangleOptions::default()).ok())
-            .or_else(|| cwdemangle::demangle(name, &cwdemangle::DemangleOptions::default()))
-    }
-
     fn reloc_name(&self, flags: RelocationFlags) -> Option<&'static str> {
         match flags {
             RelocationFlags::Elf(r_type) => match r_type {

objdiff-core/src/arch/mod.rs

@@ -371,8 +371,6 @@ pub trait Arch: Any + Debug + Send + Sync {
         Ok(None)
     }
 
-    fn demangle(&self, _name: &str) -> Option<String> { None }
-
     fn reloc_name(&self, _flags: RelocationFlags) -> Option<&'static str> { None }
 
     fn data_reloc_size(&self, flags: RelocationFlags) -> usize;

objdiff-core/src/arch/ppc/mod.rs

@@ -308,18 +308,6 @@ impl Arch for ArchPpc {
         }
     }
 
-    fn demangle(&self, mut name: &str) -> Option<String> {
-        if name.starts_with('?') {
-            msvc_demangler::demangle(name, msvc_demangler::DemangleFlags::llvm()).ok()
-        } else {
-            name = name.trim_start_matches('.');
-            cpp_demangle::Symbol::new(name)
-                .ok()
-                .and_then(|s| s.demangle(&cpp_demangle::DemangleOptions::default()).ok())
-                .or_else(|| cwdemangle::demangle(name, &cwdemangle::DemangleOptions::default()))
-        }
-    }
-
     fn reloc_name(&self, flags: RelocationFlags) -> Option<&'static str> {
         match flags {
             RelocationFlags::Elf(r_type) => match r_type {

objdiff-core/src/arch/superh/mod.rs

@@ -1,4 +1,4 @@
-use alloc::{collections::BTreeMap, format, string::String, vec, vec::Vec};
+use alloc::{collections::BTreeMap, format, vec, vec::Vec};
 
 use anyhow::Result;
 use object::elf;
@@ -132,12 +132,6 @@ impl Arch for ArchSuperH {
         Ok(())
     }
 
-    fn demangle(&self, name: &str) -> Option<String> {
-        cpp_demangle::Symbol::new(name)
-            .ok()
-            .and_then(|s| s.demangle(&cpp_demangle::DemangleOptions::default()).ok())
-    }
-
     fn reloc_name(&self, flags: RelocationFlags) -> Option<&'static str> {
         match flags {
             RelocationFlags::Elf(r_type) => match r_type {

objdiff-core/src/arch/x86.rs

@@ -1,4 +1,4 @@
-use alloc::{boxed::Box, format, string::String, vec::Vec};
+use alloc::{boxed::Box, format, vec::Vec};
 use core::cmp::Ordering;
 
 use anyhow::{Context, Result, anyhow, bail};
@@ -300,16 +300,6 @@ impl Arch for ArchX86 {
         Ok(Some(RelocationOverride { target: RelocationOverrideTarget::Keep, addend }))
     }
 
-    fn demangle(&self, name: &str) -> Option<String> {
-        if name.starts_with('?') {
-            msvc_demangler::demangle(name, msvc_demangler::DemangleFlags::llvm()).ok()
-        } else {
-            cpp_demangle::Symbol::new(name)
-                .ok()
-                .and_then(|s| s.demangle(&cpp_demangle::DemangleOptions::default()).ok())
-        }
-    }
-
     fn reloc_name(&self, flags: RelocationFlags) -> Option<&'static str> {
         match self.arch {
             Architecture::X86 => match flags {

objdiff-core/src/diff/demangler.rs

@@ -0,0 +1,48 @@
+use alloc::string::String;
+
+use crate::diff::Demangler;
+
+#[cfg(feature = "demangler")]
+impl Demangler {
+    pub fn demangle(&self, name: &str) -> Option<String> {
+        match self {
+            Demangler::Codewarrior => Self::demangle_codewarrior(name),
+            Demangler::Msvc => Self::demangle_msvc(name),
+            Demangler::Itanium => Self::demangle_itanium(name),
+            Demangler::GnuV2 => Self::demangle_gnu_v2(name),
+            Demangler::Auto => {
+                // Try to guess
+                if name.starts_with('?') {
+                    Self::demangle_msvc(name)
+                } else {
+                    Self::demangle_codewarrior(name)
+                        .or_else(|| Self::demangle_gnu_v2(name))
+                        .or_else(|| Self::demangle_itanium(name))
+                }
+            }
+        }
+    }
+
+    fn demangle_codewarrior(name: &str) -> Option<String> {
+        cwdemangle::demangle(name, &cwdemangle::DemangleOptions::default())
+    }
+
+    fn demangle_msvc(name: &str) -> Option<String> {
+        msvc_demangler::demangle(name, msvc_demangler::DemangleFlags::llvm()).ok()
+    }
+
+    fn demangle_itanium(name: &str) -> Option<String> {
+        cpp_demangle::Symbol::new(name)
+            .ok()
+            .and_then(|s| s.demangle(&cpp_demangle::DemangleOptions::default()).ok())
+    }
+
+    fn demangle_gnu_v2(name: &str) -> Option<String> {
+        gnuv2_demangle::demangle(name, &gnuv2_demangle::DemangleConfig::new_no_cfilt_mimics()).ok()
+    }
+}
+
+#[cfg(not(feature = "demangler"))]
+impl Demangler {
+    pub fn demangle(&self, _name: &str) -> Option<String> { None }
+}

objdiff-core/src/diff/mod.rs

@@ -22,6 +22,7 @@ use crate::{
 
 pub mod code;
 pub mod data;
+pub mod demangler;
 pub mod display;
 
 include!(concat!(env!("OUT_DIR"), "/config.gen.rs"));

objdiff-core/src/obj/read.rs

@@ -41,6 +41,7 @@ fn map_symbol(
     symbol: &object::Symbol,
     section_indices: &[usize],
     split_meta: Option<&SplitMeta>,
+    config: &DiffObjConfig,
 ) -> Result<Symbol> {
     let mut name = symbol.name().context("Failed to process symbol name")?.to_string();
     let mut size = symbol.size();
@@ -90,7 +91,7 @@ fn map_symbol(
         _ => SymbolKind::Unknown,
     };
     let address = arch.symbol_address(symbol.address(), kind);
-    let demangled_name = arch.demangle(&name);
+    let demangled_name = config.demangler.demangle(&name);
     // Find the virtual address for the symbol if available
     let virtual_address = split_meta
         .and_then(|m| m.virtual_addresses.as_ref())
@@ -116,6 +117,7 @@ fn map_symbols(
     sections: &[Section],
     section_indices: &[usize],
     split_meta: Option<&SplitMeta>,
+    config: &DiffObjConfig,
 ) -> Result<(Vec<Symbol>, Vec<usize>)> {
     let symbol_count = obj_file.symbols().count();
     let mut symbols = Vec::<Symbol>::with_capacity(symbol_count + obj_file.sections().count());
@@ -124,7 +126,7 @@ fn map_symbols(
         if symbol_indices.len() <= obj_symbol.index().0 {
             symbol_indices.resize(obj_symbol.index().0 + 1, usize::MAX);
         }
-        let symbol = map_symbol(arch, obj_file, &obj_symbol, section_indices, split_meta)?;
+        let symbol = map_symbol(arch, obj_file, &obj_symbol, section_indices, split_meta, config)?;
         symbol_indices[obj_symbol.index().0] = symbols.len();
         symbols.push(symbol);
     }
@@ -997,8 +999,14 @@ pub fn parse(data: &[u8], config: &DiffObjConfig, diff_side: DiffSide) -> Result
     let split_meta = parse_split_meta(&obj_file)?;
     let (mut sections, section_indices) =
         map_sections(arch.as_ref(), &obj_file, split_meta.as_ref())?;
-    let (mut symbols, symbol_indices) =
-        map_symbols(arch.as_ref(), &obj_file, &sections, &section_indices, split_meta.as_ref())?;
+    let (mut symbols, symbol_indices) = map_symbols(
+        arch.as_ref(),
+        &obj_file,
+        &sections,
+        &section_indices,
+        split_meta.as_ref(),
+        config,
+    )?;
     map_relocations(arch.as_ref(), &obj_file, &mut sections, &section_indices, &symbol_indices)?;
     parse_line_info(&obj_file, &mut sections, &section_indices, data)?;
     if config.combine_data_sections || config.combine_text_sections {