Use deprecated egui::menu as temp workaround

egui 0.32 refactored menus, and now having a
ComboBox within the menu does not work properly.
Temporarily use the deprecated menu instead.

.cargo/config.toml

@@ -1,5 +1,4 @@
-[target.x86_64-pc-windows-msvc]
-linker = "rust-lld"
-
-[target.aarch64-pc-windows-msvc]
-linker = "rust-lld"
+# statically link the C runtime so the executable does not depend on
+# that shared/dynamic library.
+[target.'cfg(all(target_env = "msvc", target_os = "windows"))']
+rustflags = ["-C", "target-feature=+crt-static"]

Cargo.toml

@@ -14,9 +14,9 @@ strip = "debuginfo"
 codegen-units = 1
 
 [workspace.package]
-version = "3.0.0-beta.6"
+version = "3.0.0-beta.13"
 authors = ["Luke Street <luke@street.dev>"]
 edition = "2024"
 license = "MIT OR Apache-2.0"
 repository = "https://github.com/encounter/objdiff"
-rust-version = "1.85"
+rust-version = "1.88"

README.md

@@ -16,11 +16,12 @@ Features:
 
 Supports:
 
-- PowerPC 750CL (GameCube, Wii)
-- MIPS (N64, PS1, PS2, PSP)
-- x86 (COFF only at the moment)
 - ARM (GBA, DS, 3DS)
-- ARM64 (Switch, experimental)
+- ARM64 (Switch)
+- MIPS (N64, PS1, PS2, PSP)
+- PowerPC (GameCube, Wii, PS3, Xbox 360)
+- SuperH (Saturn, Dreamcast)
+- x86, x86_64 (PC)
 
 See [Usage](#usage) for more information.
 

config.schema.json

@@ -175,6 +175,10 @@
           "type": "boolean",
           "description": "If true, objdiff will run the build command with the context file as an argument to generate it.",
           "default": false
+        },
+        "preset_id": {
+          "type": "number",
+          "description": "The decomp.me preset ID to use for the scratch.\nCompiler and flags in the config will take precedence over the preset, but the preset is useful for organizational purposes."
         }
       },
       "required": [

deny.toml

@@ -103,6 +103,7 @@ allow = [
     "0BSD",
     "OFL-1.1",
     "Ubuntu-font-1.0",
+    "CDLA-Permissive-2.0",
 ]
 # The confidence threshold for detecting a license from license text.
 # The higher the value, the more closely the license text must be to the

objdiff-cli/Cargo.toml

@@ -15,11 +15,11 @@ publish = false
 [dependencies]
 anyhow = "1.0"
 argp = "0.4"
-crossterm = "0.28"
+crossterm = "0.29"
 enable-ansi-support = "0.2"
 memmap2 = "0.9"
 objdiff-core = { path = "../objdiff-core", features = ["all"] }
-prost = "0.13"
+prost = "0.14"
 ratatui = "0.29"
 rayon = "1.10"
 serde = { version = "1.0", features = ["derive"] }
@@ -28,7 +28,7 @@ supports-color = "3.0"
 time = { version = "0.3", features = ["formatting", "local-offset"] }
 tracing = "0.1"
 tracing-subscriber = { version = "0.3", features = ["env-filter"] }
-typed-path = "0.10"
+typed-path = "0.11"
 
 [target.'cfg(target_env = "musl")'.dependencies]
 mimalloc = "0.1"

objdiff-cli/src/cmd/diff.rs

@@ -203,14 +203,10 @@ fn run_oneshot(
     let output_format = OutputFormat::from_option(args.format.as_deref())?;
     let (diff_config, mapping_config) = build_config_from_args(args)?;
     let target = target_path
-        .map(|p| {
-            obj::read::read(p.as_ref(), &diff_config).with_context(|| format!("Loading {}", p))
-        })
+        .map(|p| obj::read::read(p.as_ref(), &diff_config).with_context(|| format!("Loading {p}")))
         .transpose()?;
     let base = base_path
-        .map(|p| {
-            obj::read::read(p.as_ref(), &diff_config).with_context(|| format!("Loading {}", p))
-        })
+        .map(|p| obj::read::read(p.as_ref(), &diff_config).with_context(|| format!("Loading {p}")))
         .transpose()?;
     let result =
         diff::diff_objs(target.as_ref(), base.as_ref(), None, &diff_config, &mapping_config)?;
@@ -399,7 +395,7 @@ fn run_interactive(
         stdout(),
         EnterAlternateScreen,
         EnableMouseCapture,
-        SetTitle(format!("{} - objdiff", symbol_name)),
+        SetTitle(format!("{symbol_name} - objdiff")),
     )?;
     let backend = CrosstermBackend::new(stdout());
     let mut terminal = Terminal::new(backend)?;

objdiff-cli/src/cmd/report.rs

@@ -85,6 +85,9 @@ pub fn run(args: Args) -> Result<()> {
 fn generate(args: GenerateArgs) -> Result<()> {
     let mut diff_config = diff::DiffObjConfig {
         function_reloc_diffs: diff::FunctionRelocDiffs::None,
+        combine_data_sections: true,
+        combine_text_sections: true,
+        ppc_calculate_pool_relocations: false,
         ..Default::default()
     };
     apply_config_args(&mut diff_config, &args.config)?;
@@ -174,16 +177,14 @@ fn report_object(
         .target_path
         .as_ref()
         .map(|p| {
-            obj::read::read(p.as_ref(), diff_config)
-                .with_context(|| format!("Failed to open {}", p))
+            obj::read::read(p.as_ref(), diff_config).with_context(|| format!("Failed to open {p}"))
         })
         .transpose()?;
     let base = object
         .base_path
         .as_ref()
         .map(|p| {
-            obj::read::read(p.as_ref(), diff_config)
-                .with_context(|| format!("Failed to open {}", p))
+            obj::read::read(p.as_ref(), diff_config).with_context(|| format!("Failed to open {p}"))
         })
         .transpose()?;
     let result =
@@ -225,6 +226,7 @@ fn report_object(
                 demangled_name: None,
                 virtual_address: section.virtual_address,
             }),
+            address: None,
         });
 
         match section.kind {
@@ -246,13 +248,12 @@ fn report_object(
             {
                 continue;
             }
-            if let Some(existing_functions) = &mut existing_functions {
-                if (symbol.flags.contains(SymbolFlag::Global)
+            if let Some(existing_functions) = &mut existing_functions
+                && (symbol.flags.contains(SymbolFlag::Global)
                     || symbol.flags.contains(SymbolFlag::Weak))
-                    && !existing_functions.insert(symbol.name.clone())
-                {
-                    continue;
-                }
+                && !existing_functions.insert(symbol.name.clone())
+            {
+                continue;
             }
             let match_percent = symbol_diff.match_percent.unwrap_or_else(|| {
                 // Support cases where we don't have a target object,
@@ -272,6 +273,7 @@ fn report_object(
                     demangled_name: symbol.demangled_name.clone(),
                     virtual_address: symbol.virtual_address,
                 }),
+                address: symbol.address.checked_sub(section.address),
             });
             if match_percent == 100.0 {
                 measures.matched_functions += 1;
@@ -279,6 +281,16 @@ fn report_object(
             measures.total_functions += 1;
         }
     }
+    sections.sort_by(|a, b| a.name.cmp(&b.name));
+    let reverse_fn_order = object.metadata.reverse_fn_order.unwrap_or(false);
+    functions.sort_by(|a, b| {
+        if reverse_fn_order {
+            b.address.unwrap_or(0).cmp(&a.address.unwrap_or(0))
+        } else {
+            a.address.unwrap_or(u64::MAX).cmp(&b.address.unwrap_or(u64::MAX))
+        }
+        .then_with(|| a.size.cmp(&b.size))
+    });
     if metadata.complete.unwrap_or(false) {
         measures.complete_code = measures.total_code;
         measures.complete_data = measures.total_data;
@@ -418,8 +430,8 @@ fn read_report(path: &Utf8PlatformPath) -> Result<Report> {
         std::io::stdin().read_to_end(&mut data)?;
         return Report::parse(&data).with_context(|| "Failed to load report from stdin");
     }
-    let file = File::open(path).with_context(|| format!("Failed to open {}", path))?;
+    let file = File::open(path).with_context(|| format!("Failed to open {path}"))?;
     let mmap =
-        unsafe { memmap2::Mmap::map(&file) }.with_context(|| format!("Failed to map {}", path))?;
-    Report::parse(mmap.as_ref()).with_context(|| format!("Failed to load report {}", path))
+        unsafe { memmap2::Mmap::map(&file) }.with_context(|| format!("Failed to map {path}"))?;
+    Report::parse(mmap.as_ref()).with_context(|| format!("Failed to load report {path}"))
 }

objdiff-cli/src/views/function_diff.rs

@@ -87,7 +87,7 @@ impl UiView for FunctionDiffUi {
             .and_then(|(_, _, d)| d.match_percent)
         {
             line_r.spans.push(Span::styled(
-                format!("{:.2}% ", percent),
+                format!("{percent:.2}% "),
                 Style::new().fg(match_percent_color(percent)),
             ));
         }
@@ -97,7 +97,7 @@ impl UiView for FunctionDiffUi {
             .and_then(|t| t.format(&state.time_format).ok())
             .unwrap_or_else(|| "N/A".to_string());
         line_r.spans.push(Span::styled(
-            format!("Last reload: {}", reload_time),
+            format!("Last reload: {reload_time}"),
             Style::new().fg(Color::White),
         ));
         line_r.spans.push(Span::styled(
@@ -170,32 +170,31 @@ impl UiView for FunctionDiffUi {
 
         let mut prev_text = None;
         let mut prev_margin_text = None;
-        if self.three_way {
-            if let Some((obj, symbol_idx, symbol_diff)) =
+        if self.three_way
+            && let Some((obj, symbol_idx, symbol_diff)) =
                 get_symbol(state.prev_obj.as_ref(), self.prev_sym)
-            {
-                let mut text = Text::default();
-                let rect = content_chunks[4].inner(Margin::new(0, 1));
-                self.print_sym(
-                    &mut text,
-                    obj,
-                    symbol_idx,
-                    symbol_diff,
-                    &state.diff_obj_config,
-                    rect,
-                    &self.right_highlight,
-                    result,
-                    true,
-                );
-                max_width = max_width.max(text.width());
-                prev_text = Some(text);
+        {
+            let mut text = Text::default();
+            let rect = content_chunks[4].inner(Margin::new(0, 1));
+            self.print_sym(
+                &mut text,
+                obj,
+                symbol_idx,
+                symbol_diff,
+                &state.diff_obj_config,
+                rect,
+                &self.right_highlight,
+                result,
+                true,
+            );
+            max_width = max_width.max(text.width());
+            prev_text = Some(text);
 
-                // Render margin
-                let mut text = Text::default();
-                let rect = content_chunks[3].inner(Margin::new(1, 1));
-                self.print_margin(&mut text, symbol_diff, rect);
-                prev_margin_text = Some(text);
-            }
+            // Render margin
+            let mut text = Text::default();
+            let rect = content_chunks[3].inner(Margin::new(1, 1));
+            self.print_margin(&mut text, symbol_diff, rect);
+            prev_margin_text = Some(text);
         }
 
         let max_scroll_x =
@@ -450,11 +449,11 @@ impl UiView for FunctionDiffUi {
 
     fn reload(&mut self, state: &AppState) -> Result<()> {
         let left_sym =
-            state.left_obj.as_ref().and_then(|(o, _)| find_function(o, &self.symbol_name));
+            state.left_obj.as_ref().and_then(|(o, _)| o.symbol_by_name(&self.symbol_name));
         let right_sym =
-            state.right_obj.as_ref().and_then(|(o, _)| find_function(o, &self.symbol_name));
+            state.right_obj.as_ref().and_then(|(o, _)| o.symbol_by_name(&self.symbol_name));
         let prev_sym =
-            state.prev_obj.as_ref().and_then(|(o, _)| find_function(o, &self.symbol_name));
+            state.prev_obj.as_ref().and_then(|(o, _)| o.symbol_by_name(&self.symbol_name));
         self.num_rows = match (
             get_symbol(state.left_obj.as_ref(), left_sym),
             get_symbol(state.right_obj.as_ref(), right_sym),
@@ -538,7 +537,7 @@ impl FunctionDiffUi {
                 let label_text = match segment.text {
                     DiffText::Basic(text) => text.to_string(),
                     DiffText::Line(num) => format!("{num} "),
-                    DiffText::Address(addr) => format!("{:x}:", addr),
+                    DiffText::Address(addr) => format!("{addr:x}:"),
                     DiffText::Opcode(mnemonic, _op) => format!("{mnemonic} "),
                     DiffText::Argument(arg) => arg.to_string(),
                     DiffText::BranchDest(addr) => format!("{addr:x}"),
@@ -546,7 +545,7 @@ impl FunctionDiffUi {
                         sym.demangled_name.as_ref().unwrap_or(&sym.name).clone()
                     }
                     DiffText::Addend(addend) => match addend.cmp(&0i64) {
-                        Ordering::Greater => format!("+{:#x}", addend),
+                        Ordering::Greater => format!("+{addend:#x}"),
                         Ordering::Less => format!("-{:#x}", -addend),
                         _ => String::new(),
                     },
@@ -561,15 +560,18 @@ impl FunctionDiffUi {
                 let len = label_text.len();
                 let highlighted =
                     highlight_kind != HighlightKind::None && *highlight == highlight_kind;
-                if let Some((cx, cy)) = result.click_xy {
-                    if cx >= sx && cx < sx + len as u16 && cy == sy {
-                        new_highlight = Some(highlight_kind);
-                    }
+                if let Some((cx, cy)) = result.click_xy
+                    && cx >= sx
+                    && cx < sx + len as u16
+                    && cy == sy
+                {
+                    new_highlight = Some(highlight_kind);
                 }
                 let mut style = Style::new().fg(match segment.color {
                     DiffTextColor::Normal => Color::Gray,
                     DiffTextColor::Dim => Color::DarkGray,
                     DiffTextColor::Bright => Color::White,
+                    DiffTextColor::DataFlow => Color::LightCyan,
                     DiffTextColor::Replace => Color::Cyan,
                     DiffTextColor::Delete => Color::Red,
                     DiffTextColor::Insert => Color::Green,
@@ -650,12 +652,3 @@ fn get_symbol(
     let sym = sym?;
     Some((obj, sym, &diff.symbols[sym]))
 }
-
-fn find_function(obj: &Object, name: &str) -> Option<usize> {
-    for (symbol_idx, symbol) in obj.symbols.iter().enumerate() {
-        if symbol.name == name {
-            return Some(symbol_idx);
-        }
-    }
-    None
-}

objdiff-core/Cargo.toml

@@ -41,6 +41,7 @@ any-arch = [
     "dep:regex",
     "dep:similar",
     "dep:syn",
+    "dep:encoding_rs"
 ]
 bindings = [
     "dep:prost",
@@ -61,7 +62,10 @@ config = [
     "dep:semver",
     "dep:typed-path",
 ]
-dwarf = ["dep:gimli"]
+dwarf = [
+    "dep:gimli",
+    "dep:typed-arena",
+]
 serde = [
     "dep:pbjson",
     "dep:pbjson-build",
@@ -77,6 +81,7 @@ std = [
     "prost?/std",
     "serde?/std",
     "similar?/std",
+    "typed-arena?/std",
     "typed-path?/std",
     "dep:filetime",
     "dep:memmap2",
@@ -91,7 +96,7 @@ ppc = [
     "any-arch",
     "dep:cwdemangle",
     "dep:cwextab",
-    "dep:ppc750cl",
+    "dep:powerpc",
     "dep:rlwinmdec",
 ]
 x86 = [
@@ -122,18 +127,18 @@ features = ["all"]
 [dependencies]
 anyhow = { version = "1.0", default-features = false }
 filetime = { version = "0.2", optional = true }
-flagset = { version = "0.4", default-features = false, optional = true, git = "https://github.com/enarx/flagset.git", rev = "a1fe9369b3741e43fec45da1998e83b9d78966a2" }
+flagset = { version = "0.4", default-features = false, optional = true }
 itertools = { version = "0.14", default-features = false, features = ["use_alloc"] }
 log = { version = "0.4", default-features = false, optional = true }
 memmap2 = { version = "0.9", optional = true }
 num-traits = { version = "0.2", default-features = false, optional = true }
-object = { git = "https://github.com/gimli-rs/object", rev = "a74579249e21ab8fcd3a86be588de336f18297cb", default-features = false, features = ["read_core", "elf", "pe"] }
-pbjson = { version = "0.7", default-features = false, optional = true }
-prost = { version = "0.13", default-features = false, features = ["prost-derive"], optional = true }
+object = { git = "https://github.com/gimli-rs/object", rev = "16ff70aa6fbd97d6bb7b92375929f4d72414c32b", default-features = false, features = ["read_core", "elf", "coff"] }
+pbjson = { version = "0.8", default-features = false, optional = true }
+prost = { version = "0.14", default-features = false, features = ["derive"], optional = true }
 regex = { version = "1.11", default-features = false, features = [], optional = true }
 serde = { version = "1.0", default-features = false, features = ["derive"], optional = true }
 similar = { version = "2.7", default-features = false, features = ["hashbrown"], optional = true, git = "https://github.com/encounter/similar.git", branch = "no_std" }
-typed-path = { version = "0.10", default-features = false, optional = true }
+typed-path = { version = "0.11", default-features = false, optional = true }
 
 # config
 globset = { version = "0.4", default-features = false, optional = true }
@@ -141,16 +146,17 @@ semver = { version = "1.0", default-features = false, optional = true }
 serde_json = { version = "1.0", default-features = false, features = ["alloc"], optional = true }
 
 # dwarf
-gimli = { version = "0.31", default-features = false, features = ["read"], optional = true }
+gimli = { version = "0.32", default-features = false, features = ["read"], optional = true }
+typed-arena = { version = "2.0", default-features = false, optional = true }
 
 # ppc
 cwdemangle = { version = "1.0", optional = true }
-cwextab = { version = "1.0", optional = true }
-ppc750cl = { version = "0.3", optional = true }
+cwextab = { version = "1.1", optional = true }
+powerpc = { version = "0.4", optional = true }
 rlwinmdec = { version = "1.1", optional = true }
 
 # mips
-rabbitizer = { version = "2.0.0-alpha.1", default-features = false, features = ["all_extensions"], optional = true }
+rabbitizer = { version = "2.0.0-alpha.4", default-features = false, features = ["all_extensions"], optional = true }
 
 # x86
 cpp_demangle = { version = "0.4", default-features = false, features = ["alloc"], optional = true }
@@ -158,7 +164,7 @@ iced-x86 = { version = "1.21", default-features = false, features = ["decoder",
 msvc-demangler = { version = "0.11", optional = true }
 
 # arm
-unarm = { version = "1.7", optional = true }
+unarm = { version = "1.9", optional = true }
 arm-attr = { version = "0.2", optional = true }
 
 # arm64
@@ -166,15 +172,15 @@ yaxpeax-arch = { version = "0.3", default-features = false, optional = true }
 yaxpeax-arm = { version = "0.3", default-features = false, optional = true }
 
 # build
-notify = { version = "8.0.0", optional = true }
+notify = { version = "8.1.0", optional = true }
 notify-debouncer-full = { version = "0.5.0", optional = true }
 shell-escape = { version = "0.1", optional = true }
-tempfile = { version = "3.19", optional = true }
+tempfile = { version = "3.20", optional = true }
 time = { version = "0.3", optional = true }
-encoding_rs = "0.8.35"
+encoding_rs = { version = "0.8.35", optional = true }
 
 [target.'cfg(windows)'.dependencies]
-winapi = { version = "0.3", optional = true }
+winapi = { version = "0.3", optional = true, features = ["winbase"] }
 
 # For Linux static binaries, use rustls
 [target.'cfg(target_os = "linux")'.dependencies]
@@ -188,10 +194,10 @@ self_update = { version = "0.42", optional = true }
 
 [build-dependencies]
 heck = { version = "0.5", optional = true }
-pbjson-build = { version = "0.7", optional = true }
+pbjson-build = { version = "0.8", optional = true }
 prettyplease = { version = "0.2", optional = true }
 proc-macro2 = { version = "1.0", optional = true }
-prost-build = { version = "0.13", optional = true }
+prost-build = { version = "0.14", optional = true }
 quote = { version = "1.0", optional = true }
 serde = { version = "1.0", features = ["derive"] }
 serde_json = { version = "1.0" }
@@ -200,4 +206,4 @@ syn = { version = "2.0", optional = true }
 [dev-dependencies]
 # Enable all features for tests
 objdiff-core = { path = ".", features = ["all"] }
-insta = "1.42"
+insta = "1.43"

objdiff-core/README.md

@@ -5,11 +5,12 @@ objdiff-core contains the core functionality of [objdiff](https://github.com/enc
 ## Crate feature flags
 
 - **`all`**: Enables all main features.
+- **`bindings`**: Enables serialization and deserialization of objdiff data structures.
 - **`config`**: Enables objdiff configuration file support.
 - **`dwarf`**: Enables extraction of line number information from DWARF debug sections.
-- **`mips`**: Enables the MIPS backend powered by [rabbitizer](https://github.com/Decompollaborate/rabbitizer). (Note: C library with Rust bindings)
-- **`ppc`**: Enables the PowerPC backend powered by [ppc750cl](https://github.com/encounter/ppc750cl).
-- **`x86`**: Enables the x86 backend powered by [iced-x86](https://crates.io/crates/iced-x86).
-- **`arm`**: Enables the ARM backend powered by [unarm](https://github.com/AetiasHax/unarm).
 - **`arm64`**: Enables the ARM64 backend powered by [yaxpeax-arm](https://github.com/iximeow/yaxpeax-arm).
-- **`bindings`**: Enables serialization and deserialization of objdiff data structures.
+- **`arm`**: Enables the ARM backend powered by [unarm](https://github.com/AetiasHax/unarm).
+- **`mips`**: Enables the MIPS backend powered by [rabbitizer](https://github.com/Decompollaborate/rabbitizer).
+- **`ppc`**: Enables the PowerPC backend powered by [powerpc](https://github.com/encounter/powerpc-rs).
+- **`superh`**: Enables the SuperH backend powered by an included disassembler.
+- **`x86`**: Enables the x86 backend powered by [iced-x86](https://crates.io/crates/iced-x86).

objdiff-core/config-schema.json

@@ -25,6 +25,20 @@
         }
       ]
     },
+    {
+      "id": "analyzeDataFlow",
+      "type": "boolean",
+      "default": false,
+      "name": "(Experimental) Perform data flow analysis",
+      "description": "Use data flow analysis to display known information about register contents where possible"
+    },
+    {
+      "id": "showDataFlow",
+      "type": "boolean",
+      "default": true,
+      "name": "Show data flow",
+      "description": "Show data flow analysis results in place of register name where present"
+    },
     {
       "id": "spaceBetweenArgs",
       "type": "boolean",
@@ -264,7 +278,8 @@
       "id": "ppc",
       "name": "PowerPC",
       "properties": [
-        "ppc.calculatePoolRelocations"
+        "ppc.calculatePoolRelocations",
+        "analyzeDataFlow"
       ]
     },
     {

objdiff-core/config_gen.rs

@@ -60,10 +60,10 @@ pub struct ConfigGroup {
 }
 
 fn build_doc(name: &str, description: Option<&str>) -> TokenStream {
-    let mut doc = format!(" {}", name);
+    let mut doc = format!(" {name}");
     let mut out = quote! { #[doc = #doc] };
     if let Some(description) = description {
-        doc = format!(" {}", description);
+        doc = format!(" {description}");
         out.extend(quote! { #[doc = ""] });
         out.extend(quote! { #[doc = #doc] });
     }
@@ -443,9 +443,9 @@ pub fn generate_diff_config() {
         }
         impl core::fmt::Display for ConfigPropertyValue {
             fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
-                match self {
-                    ConfigPropertyValue::Boolean(value) => write!(f, "{}", value),
-                    ConfigPropertyValue::Choice(value) => write!(f, "{}", value),
+                match *self {
+                    ConfigPropertyValue::Boolean(value) => write!(f, "{value}"),
+                    ConfigPropertyValue::Choice(value) => f.write_str(value),
                 }
             }
         }

objdiff-core/protos/report.proto

@@ -99,6 +99,8 @@ message ReportItem {
   float fuzzy_match_percent = 3;
   // Extra metadata for this item
   optional ReportItemMetadata metadata = 4;
+  // Address of the item (section-relative offset)
+  optional uint64 address = 5;
 }
 
 // Extra metadata for an item

objdiff-core/src/arch/arm.rs

@@ -11,11 +11,11 @@ use object::{Endian as _, Object as _, ObjectSection as _, ObjectSymbol as _, el
 use unarm::{args, arm, thumb};
 
 use crate::{
-    arch::Arch,
+    arch::{Arch, RelocationOverride, RelocationOverrideTarget},
     diff::{ArmArchVersion, ArmR9Usage, DiffObjConfig, display::InstructionPart},
     obj::{
         InstructionRef, Relocation, RelocationFlags, ResolvedInstructionRef, ResolvedRelocation,
-        ScannedInstruction, Section, SectionKind, Symbol, SymbolFlag, SymbolFlagSet, SymbolKind,
+        Section, SectionKind, Symbol, SymbolFlag, SymbolFlagSet, SymbolKind,
     },
 };
 
@@ -187,14 +187,14 @@ impl Arch for ArchArm {
         self.disasm_modes = Self::get_mapping_symbols(sections, symbols);
     }
 
-    fn scan_instructions(
+    fn scan_instructions_internal(
         &self,
         address: u64,
         code: &[u8],
         section_index: usize,
         _relocations: &[Relocation],
         diff_config: &DiffObjConfig,
-    ) -> Result<Vec<ScannedInstruction>> {
+    ) -> Result<Vec<InstructionRef>> {
         let start_addr = address as u32;
         let end_addr = start_addr + code.len() as u32;
 
@@ -219,13 +219,13 @@ impl Arch for ArchArm {
         let mut next_mapping = mappings_iter.next();
 
         let ins_count = code.len() / mode.instruction_size(start_addr);
-        let mut ops = Vec::<ScannedInstruction>::with_capacity(ins_count);
+        let mut ops = Vec::<InstructionRef>::with_capacity(ins_count);
 
         let parse_flags = self.parse_flags(diff_config);
 
         let mut address = start_addr;
         while address < end_addr {
-            while let Some(next) = next_mapping.take_if(|x| address >= x.address) {
+            while let Some(next) = next_mapping.filter(|x| address >= x.address) {
                 // Change mapping
                 mode = next.mapping;
                 next_mapping = mappings_iter.next();
@@ -235,12 +235,10 @@ impl Arch for ArchArm {
             let data = &code[(address - start_addr) as usize..];
             if data.len() < ins_size {
                 // Push the remainder as data
-                ops.push(ScannedInstruction {
-                    ins_ref: InstructionRef {
-                        address: address as u64,
-                        size: data.len() as u8,
-                        opcode: u16::MAX,
-                    },
+                ops.push(InstructionRef {
+                    address: address as u64,
+                    size: data.len() as u8,
+                    opcode: u16::MAX,
                     branch_dest: None,
                 });
                 break;
@@ -256,12 +254,10 @@ impl Arch for ArchArm {
                 }
                 _ => {
                     // Invalid instruction size
-                    ops.push(ScannedInstruction {
-                        ins_ref: InstructionRef {
-                            address: address as u64,
-                            size: ins_size as u8,
-                            opcode: u16::MAX,
-                        },
+                    ops.push(InstructionRef {
+                        address: address as u64,
+                        size: ins_size as u8,
+                        opcode: u16::MAX,
                         branch_dest: None,
                     });
                     address += ins_size as u32;
@@ -325,8 +321,10 @@ impl Arch for ArchArm {
                 unarm::ParseMode::Data => (u16::MAX, None),
             };
 
-            ops.push(ScannedInstruction {
-                ins_ref: InstructionRef { address: address as u64, size: ins_size as u8, opcode },
+            ops.push(InstructionRef {
+                address: address as u64,
+                size: ins_size as u8,
+                opcode,
                 branch_dest: branch_dest.map(|x| x as u64),
             });
             address += ins_size as u32;
@@ -358,47 +356,57 @@ impl Arch for ArchArm {
         Ok(())
     }
 
-    fn implcit_addend(
+    fn relocation_override(
         &self,
         _file: &object::File<'_>,
         section: &object::Section,
         address: u64,
-        _relocation: &object::Relocation,
-        flags: RelocationFlags,
-    ) -> Result<i64> {
-        let section_data = section.data()?;
-        let address = address as usize;
-        Ok(match flags {
-            // ARM calls
-            RelocationFlags::Elf(elf::R_ARM_PC24)
-            | RelocationFlags::Elf(elf::R_ARM_XPC25)
-            | RelocationFlags::Elf(elf::R_ARM_CALL) => {
-                let data = section_data[address..address + 4].try_into()?;
-                let addend = self.endianness.read_i32_bytes(data);
-                let imm24 = addend & 0xffffff;
-                (imm24 << 2) << 8 >> 8
+        relocation: &object::Relocation,
+    ) -> Result<Option<RelocationOverride>> {
+        match relocation.flags() {
+            // Handle ELF implicit relocations
+            object::RelocationFlags::Elf { r_type } => {
+                if relocation.has_implicit_addend() {
+                    let section_data = section.data()?;
+                    let address = address as usize;
+                    let addend = match r_type {
+                        // ARM calls
+                        elf::R_ARM_PC24 | elf::R_ARM_XPC25 | elf::R_ARM_CALL => {
+                            let data = section_data[address..address + 4].try_into()?;
+                            let addend = self.endianness.read_i32_bytes(data);
+                            let imm24 = addend & 0xffffff;
+                            (imm24 << 2) << 8 >> 8
+                        }
+
+                        // Thumb calls
+                        elf::R_ARM_THM_PC22 | elf::R_ARM_THM_XPC22 => {
+                            let data = section_data[address..address + 2].try_into()?;
+                            let high = self.endianness.read_i16_bytes(data) as i32;
+                            let data = section_data[address + 2..address + 4].try_into()?;
+                            let low = self.endianness.read_i16_bytes(data) as i32;
+
+                            let imm22 = ((high & 0x7ff) << 11) | (low & 0x7ff);
+                            (imm22 << 1) << 9 >> 9
+                        }
+
+                        // Data
+                        elf::R_ARM_ABS32 => {
+                            let data = section_data[address..address + 4].try_into()?;
+                            self.endianness.read_i32_bytes(data)
+                        }
+
+                        flags => bail!("Unsupported ARM implicit relocation {flags:?}"),
+                    };
+                    Ok(Some(RelocationOverride {
+                        target: RelocationOverrideTarget::Keep,
+                        addend: addend as i64,
+                    }))
+                } else {
+                    Ok(None)
+                }
             }
-
-            // Thumb calls
-            RelocationFlags::Elf(elf::R_ARM_THM_PC22)
-            | RelocationFlags::Elf(elf::R_ARM_THM_XPC22) => {
-                let data = section_data[address..address + 2].try_into()?;
-                let high = self.endianness.read_i16_bytes(data) as i32;
-                let data = section_data[address + 2..address + 4].try_into()?;
-                let low = self.endianness.read_i16_bytes(data) as i32;
-
-                let imm22 = ((high & 0x7ff) << 11) | (low & 0x7ff);
-                (imm22 << 1) << 9 >> 9
-            }
-
-            // Data
-            RelocationFlags::Elf(elf::R_ARM_ABS32) => {
-                let data = section_data[address..address + 4].try_into()?;
-                self.endianness.read_i32_bytes(data)
-            }
-
-            flags => bail!("Unsupported ARM implicit relocation {flags:?}"),
-        } as i64)
+            _ => Ok(None),
+        }
     }
 
     fn demangle(&self, name: &str) -> Option<String> {
@@ -577,7 +585,7 @@ fn push_args(
                     arg_cb(InstructionPart::basic("}"))?;
                 }
                 args::Argument::CoprocNum(value) => {
-                    arg_cb(InstructionPart::opaque(format!("p{}", value)))?;
+                    arg_cb(InstructionPart::opaque(format!("p{value}")))?;
                 }
                 args::Argument::ShiftImm(shift) => {
                     arg_cb(InstructionPart::opaque(shift.op.to_string()))?;

objdiff-core/src/arch/mips.rs

@@ -3,23 +3,19 @@ use alloc::{
     string::{String, ToString},
     vec::Vec,
 };
-use core::ops::Range;
 
 use anyhow::{Result, bail};
 use object::{Endian as _, Object as _, ObjectSection as _, ObjectSymbol as _, elf};
 use rabbitizer::{
-    IsaExtension, IsaVersion, Vram,
-    abi::Abi,
-    operands::{IU16, ValuedOperand},
-    registers_meta::Register,
+    IsaExtension, IsaVersion, Vram, abi::Abi, operands::ValuedOperand, registers_meta::Register,
 };
 
 use crate::{
-    arch::Arch,
+    arch::{Arch, RelocationOverride, RelocationOverrideTarget},
     diff::{DiffObjConfig, MipsAbi, MipsInstrCategory, display::InstructionPart},
     obj::{
         InstructionArg, InstructionArgValue, InstructionRef, Relocation, RelocationFlags,
-        ResolvedInstructionRef, ResolvedRelocation, ScannedInstruction, SymbolFlag, SymbolFlagSet,
+        ResolvedInstructionRef, ResolvedRelocation, Section, Symbol, SymbolFlag, SymbolFlagSet,
     },
 };
 
@@ -144,6 +140,14 @@ impl ArchMips {
         })
     }
 
+    fn default_instruction_flags(&self) -> rabbitizer::InstructionFlags {
+        match self.isa_extension {
+            Some(extension) => rabbitizer::InstructionFlags::new_extension(extension),
+            None => rabbitizer::InstructionFlags::new(IsaVersion::MIPS_III),
+        }
+        .with_abi(self.abi)
+    }
+
     fn instruction_flags(&self, diff_config: &DiffObjConfig) -> rabbitizer::InstructionFlags {
         let isa_extension = match diff_config.mips_instr_category {
             MipsInstrCategory::Auto => self.isa_extension,
@@ -155,7 +159,7 @@ impl ArchMips {
         };
         match isa_extension {
             Some(extension) => rabbitizer::InstructionFlags::new_extension(extension),
-            None => rabbitizer::InstructionFlags::new_isa(IsaVersion::MIPS_III, None),
+            None => rabbitizer::InstructionFlags::new(IsaVersion::MIPS_III),
         }
         .with_abi(match diff_config.mips_abi {
             MipsAbi::Auto => self.abi,
@@ -189,16 +193,16 @@ impl ArchMips {
 }
 
 impl Arch for ArchMips {
-    fn scan_instructions(
+    fn scan_instructions_internal(
         &self,
         address: u64,
         code: &[u8],
         _section_index: usize,
         _relocations: &[Relocation],
         diff_config: &DiffObjConfig,
-    ) -> Result<Vec<ScannedInstruction>> {
+    ) -> Result<Vec<InstructionRef>> {
         let instruction_flags = self.instruction_flags(diff_config);
-        let mut ops = Vec::<ScannedInstruction>::with_capacity(code.len() / 4);
+        let mut ops = Vec::<InstructionRef>::with_capacity(code.len() / 4);
         let mut cur_addr = address as u32;
         for chunk in code.chunks_exact(4) {
             let code = self.endianness.read_u32_bytes(chunk.try_into()?);
@@ -206,10 +210,7 @@ impl Arch for ArchMips {
                 rabbitizer::Instruction::new(code, Vram::new(cur_addr), instruction_flags);
             let opcode = instruction.opcode() as u16;
             let branch_dest = instruction.get_branch_vram_generic().map(|v| v.inner() as u64);
-            ops.push(ScannedInstruction {
-                ins_ref: InstructionRef { address: cur_addr as u64, size: 4, opcode },
-                branch_dest,
-            });
+            ops.push(InstructionRef { address: cur_addr as u64, size: 4, opcode, branch_dest });
             cur_addr += 4;
         }
         Ok(ops)
@@ -225,66 +226,70 @@ impl Arch for ArchMips {
         let display_flags = self.instruction_display_flags(diff_config);
         let opcode = instruction.opcode();
         cb(InstructionPart::opcode(opcode.name(), opcode as u16))?;
-        let start_address = resolved.symbol.address;
-        let function_range = start_address..start_address + resolved.symbol.size;
-        push_args(
-            &instruction,
-            resolved.relocation,
-            function_range,
-            resolved.section_index,
-            &display_flags,
-            cb,
-        )?;
+        push_args(&instruction, resolved.relocation, &display_flags, cb)?;
         Ok(())
     }
 
-    fn implcit_addend(
+    fn relocation_override(
         &self,
         file: &object::File<'_>,
         section: &object::Section,
         address: u64,
-        reloc: &object::Relocation,
-        flags: RelocationFlags,
-    ) -> Result<i64> {
-        // Check for paired R_MIPS_HI16 and R_MIPS_LO16 relocations.
-        if let RelocationFlags::Elf(elf::R_MIPS_HI16 | elf::R_MIPS_LO16) = flags {
-            if let Some(addend) = self
-                .paired_relocations
-                .get(section.index().0)
-                .and_then(|m| m.get(&address).copied())
-            {
-                return Ok(addend);
-            }
-        }
+        relocation: &object::Relocation,
+    ) -> Result<Option<RelocationOverride>> {
+        match relocation.flags() {
+            // Handle ELF implicit relocations
+            object::RelocationFlags::Elf { r_type } => {
+                if relocation.has_implicit_addend() {
+                    // Check for paired R_MIPS_HI16 and R_MIPS_LO16 relocations.
+                    if let elf::R_MIPS_HI16 | elf::R_MIPS_LO16 = r_type
+                        && let Some(addend) = self
+                            .paired_relocations
+                            .get(section.index().0)
+                            .and_then(|m| m.get(&address).copied())
+                    {
+                        return Ok(Some(RelocationOverride {
+                            target: RelocationOverrideTarget::Keep,
+                            addend,
+                        }));
+                    }
 
-        let data = section.data()?;
-        let code = data[address as usize..address as usize + 4].try_into()?;
-        let addend = self.endianness.read_u32_bytes(code);
-        Ok(match flags {
-            RelocationFlags::Elf(elf::R_MIPS_32) => addend as i64,
-            RelocationFlags::Elf(elf::R_MIPS_26) => ((addend & 0x03FFFFFF) << 2) as i64,
-            RelocationFlags::Elf(elf::R_MIPS_HI16) => ((addend & 0x0000FFFF) << 16) as i32 as i64,
-            RelocationFlags::Elf(elf::R_MIPS_LO16 | elf::R_MIPS_GOT16 | elf::R_MIPS_CALL16) => {
-                (addend & 0x0000FFFF) as i16 as i64
-            }
-            RelocationFlags::Elf(elf::R_MIPS_GPREL16 | elf::R_MIPS_LITERAL) => {
-                let object::RelocationTarget::Symbol(idx) = reloc.target() else {
-                    bail!("Unsupported R_MIPS_GPREL16 relocation against a non-symbol");
-                };
-                let sym = file.symbol_by_index(idx)?;
+                    let data = section.data()?;
+                    let code = self
+                        .endianness
+                        .read_u32_bytes(data[address as usize..address as usize + 4].try_into()?);
+                    let addend = match r_type {
+                        elf::R_MIPS_32 => code as i64,
+                        elf::R_MIPS_26 => ((code & 0x03FFFFFF) << 2) as i64,
+                        elf::R_MIPS_HI16 => ((code & 0x0000FFFF) << 16) as i32 as i64,
+                        elf::R_MIPS_LO16 | elf::R_MIPS_GOT16 | elf::R_MIPS_CALL16 => {
+                            (code & 0x0000FFFF) as i16 as i64
+                        }
+                        elf::R_MIPS_GPREL16 | elf::R_MIPS_LITERAL => {
+                            let object::RelocationTarget::Symbol(idx) = relocation.target() else {
+                                bail!("Unsupported R_MIPS_GPREL16 relocation against a non-symbol");
+                            };
+                            let sym = file.symbol_by_index(idx)?;
 
-                // if the symbol we are relocating against is in a local section we need to add
-                // the ri_gp_value from .reginfo to the addend.
-                if sym.section().index().is_some() {
-                    ((addend & 0x0000FFFF) as i16 as i64) + self.ri_gp_value as i64
+                            // if the symbol we are relocating against is in a local section we need to add
+                            // the ri_gp_value from .reginfo to the addend.
+                            if sym.section().index().is_some() {
+                                ((code & 0x0000FFFF) as i16 as i64) + self.ri_gp_value as i64
+                            } else {
+                                (code & 0x0000FFFF) as i16 as i64
+                            }
+                        }
+                        elf::R_MIPS_PC16 => 0, // PC-relative relocation
+                        R_MIPS15_S3 => ((code & 0x001FFFC0) >> 3) as i64,
+                        flags => bail!("Unsupported MIPS implicit relocation {flags:?}"),
+                    };
+                    Ok(Some(RelocationOverride { target: RelocationOverrideTarget::Keep, addend }))
                 } else {
-                    (addend & 0x0000FFFF) as i16 as i64
+                    Ok(None)
                 }
             }
-            RelocationFlags::Elf(elf::R_MIPS_PC16) => 0, // PC-relative relocation
-            RelocationFlags::Elf(R_MIPS15_S3) => ((addend & 0x001FFFC0) >> 3) as i64,
-            flags => bail!("Unsupported MIPS implicit relocation {flags:?}"),
-        })
+            _ => Ok(None),
+        }
     }
 
     fn demangle(&self, name: &str) -> Option<String> {
@@ -333,13 +338,41 @@ impl Arch for ArchMips {
         }
         flags
     }
+
+    fn infer_function_size(
+        &self,
+        symbol: &Symbol,
+        section: &Section,
+        next_address: u64,
+    ) -> Result<u64> {
+        // Trim any trailing 4-byte zeroes from the end (nops)
+        let mut new_address = next_address;
+        while new_address >= symbol.address + 4
+            && let Some(data) = section.data_range(new_address - 4, 4)
+            && data == [0u8; 4]
+        {
+            new_address -= 4;
+        }
+        // Check if the last instruction has a delay slot, if so, include the delay slot nop
+        if new_address + 4 <= next_address
+            && new_address >= symbol.address + 4
+            && let Some(data) = section.data_range(new_address - 4, 4)
+            && let instruction = rabbitizer::Instruction::new(
+                self.endianness.read_u32_bytes(data.try_into().unwrap()),
+                Vram::new((new_address - 4) as u32),
+                self.default_instruction_flags(),
+            )
+            && instruction.opcode().has_delay_slot()
+        {
+            new_address += 4;
+        }
+        Ok(new_address.saturating_sub(symbol.address))
+    }
 }
 
 fn push_args(
     instruction: &rabbitizer::Instruction,
     relocation: Option<ResolvedRelocation>,
-    function_range: Range<u64>,
-    section_index: usize,
     display_flags: &rabbitizer::InstructionDisplayFlags,
     mut arg_cb: impl FnMut(InstructionPart) -> Result<()>,
 ) -> Result<()> {
@@ -350,35 +383,23 @@ fn push_args(
         }
 
         match op {
-            ValuedOperand::core_immediate(imm) => {
+            ValuedOperand::core_imm_i16(imm) => {
                 if let Some(resolved) = relocation {
                     push_reloc(resolved.relocation, &mut arg_cb)?;
                 } else {
-                    arg_cb(match imm {
-                        IU16::Integer(s) => InstructionPart::signed(s),
-                        IU16::Unsigned(u) => InstructionPart::unsigned(u),
-                    })?;
+                    arg_cb(InstructionPart::signed(imm))?;
+                }
+            }
+            ValuedOperand::core_imm_u16(imm) => {
+                if let Some(resolved) = relocation {
+                    push_reloc(resolved.relocation, &mut arg_cb)?;
+                } else {
+                    arg_cb(InstructionPart::unsigned(imm))?;
                 }
             }
             ValuedOperand::core_label(..) | ValuedOperand::core_branch_target_label(..) => {
                 if let Some(resolved) = relocation {
-                    // If the relocation target is within the current function, we can
-                    // convert it into a relative branch target. Note that we check
-                    // target_address > start_address instead of >= so that recursive
-                    // tail calls are not considered branch targets.
-                    let target_address =
-                        resolved.symbol.address.checked_add_signed(resolved.relocation.addend);
-                    if resolved.symbol.section == Some(section_index)
-                        && target_address.is_some_and(|addr| {
-                            addr > function_range.start && addr < function_range.end
-                        })
-                    {
-                        // TODO move this logic up a level
-                        let target_address = target_address.unwrap();
-                        arg_cb(InstructionPart::branch_dest(target_address))?;
-                    } else {
-                        push_reloc(resolved.relocation, &mut arg_cb)?;
-                    }
+                    push_reloc(resolved.relocation, &mut arg_cb)?;
                 } else if let Some(branch_dest) = instruction
                     .get_branch_offset_generic()
                     .map(|o| (instruction.vram() + o).inner() as u64)
@@ -390,14 +411,13 @@ fn push_args(
                     ))?;
                 }
             }
-            ValuedOperand::core_immediate_base(imm, base) => {
+            ValuedOperand::core_imm_rs(imm, base) => {
                 if let Some(resolved) = relocation {
                     push_reloc(resolved.relocation, &mut arg_cb)?;
                 } else {
-                    arg_cb(InstructionPart::Arg(InstructionArg::Value(match imm {
-                        IU16::Integer(s) => InstructionArgValue::Signed(s as i64),
-                        IU16::Unsigned(u) => InstructionArgValue::Unsigned(u as u64),
-                    })))?;
+                    arg_cb(InstructionPart::Arg(InstructionArg::Value(
+                        InstructionArgValue::Signed(imm as i64),
+                    )))?;
                 }
                 arg_cb(InstructionPart::basic("("))?;
                 arg_cb(InstructionPart::opaque(base.either_name(

objdiff-core/src/arch/mod.rs

@@ -1,5 +1,14 @@
-use alloc::{borrow::Cow, boxed::Box, format, string::String, vec::Vec};
-use core::{ffi::CStr, fmt, fmt::Debug};
+use alloc::{
+    borrow::Cow,
+    boxed::Box,
+    format,
+    string::{String, ToString},
+    vec::Vec,
+};
+use core::{
+    ffi::CStr,
+    fmt::{self, Debug},
+};
 
 use anyhow::{Result, bail};
 use encoding_rs::SHIFT_JIS;
@@ -11,8 +20,9 @@ use crate::{
         display::{ContextItem, HoverItem, InstructionPart},
     },
     obj::{
-        InstructionArg, Object, ParsedInstruction, Relocation, RelocationFlags,
-        ResolvedInstructionRef, ScannedInstruction, Section, Symbol, SymbolFlagSet, SymbolKind,
+        FlowAnalysisResult, InstructionArg, InstructionRef, Object, ParsedInstruction, Relocation,
+        RelocationFlags, ResolvedInstructionRef, ResolvedSymbol, Section, Symbol, SymbolFlagSet,
+        SymbolKind,
     },
     util::ReallySigned,
 };
@@ -31,6 +41,7 @@ pub mod superh;
 pub mod x86;
 
 /// Represents the type of data associated with an instruction
+#[derive(PartialEq)]
 pub enum DataType {
     Int8,
     Int16,
@@ -44,16 +55,16 @@ pub enum DataType {
 
 impl fmt::Display for DataType {
     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
-        match self {
-            DataType::Int8 => write!(f, "Int8"),
-            DataType::Int16 => write!(f, "Int16"),
-            DataType::Int32 => write!(f, "Int32"),
-            DataType::Int64 => write!(f, "Int64"),
-            DataType::Float => write!(f, "Float"),
-            DataType::Double => write!(f, "Double"),
-            DataType::Bytes => write!(f, "Bytes"),
-            DataType::String => write!(f, "String"),
-        }
+        f.write_str(match self {
+            DataType::Int8 => "Int8",
+            DataType::Int16 => "Int16",
+            DataType::Int32 => "Int32",
+            DataType::Int64 => "Int64",
+            DataType::Float => "Float",
+            DataType::Double => "Double",
+            DataType::Bytes => "Bytes",
+            DataType::String => "String",
+        })
     }
 }
 
@@ -61,8 +72,8 @@ impl DataType {
     pub fn display_labels(&self, endian: object::Endianness, bytes: &[u8]) -> Vec<String> {
         let mut strs = Vec::new();
         for (literal, label_override) in self.display_literals(endian, bytes) {
-            let label = label_override.unwrap_or_else(|| format!("{}", self));
-            strs.push(format!("{}: {}", label, literal))
+            let label = label_override.unwrap_or_else(|| self.to_string());
+            strs.push(format!("{label}: {literal}"))
         }
         strs
     }
@@ -95,7 +106,7 @@ impl DataType {
         match self {
             DataType::Int8 => {
                 let i = i8::from_ne_bytes(bytes.try_into().unwrap());
-                strs.push((format!("{:#x}", i), None));
+                strs.push((format!("{i:#x}"), None));
 
                 if i < 0 {
                     strs.push((format!("{:#x}", ReallySigned(i)), None));
@@ -103,7 +114,7 @@ impl DataType {
             }
             DataType::Int16 => {
                 let i = endian.read_i16_bytes(bytes.try_into().unwrap());
-                strs.push((format!("{:#x}", i), None));
+                strs.push((format!("{i:#x}"), None));
 
                 if i < 0 {
                     strs.push((format!("{:#x}", ReallySigned(i)), None));
@@ -111,7 +122,7 @@ impl DataType {
             }
             DataType::Int32 => {
                 let i = endian.read_i32_bytes(bytes.try_into().unwrap());
-                strs.push((format!("{:#x}", i), None));
+                strs.push((format!("{i:#x}"), None));
 
                 if i < 0 {
                     strs.push((format!("{:#x}", ReallySigned(i)), None));
@@ -119,7 +130,7 @@ impl DataType {
             }
             DataType::Int64 => {
                 let i = endian.read_i64_bytes(bytes.try_into().unwrap());
-                strs.push((format!("{:#x}", i), None));
+                strs.push((format!("{i:#x}"), None));
 
                 if i < 0 {
                     strs.push((format!("{:#x}", ReallySigned(i)), None));
@@ -146,16 +157,16 @@ impl DataType {
                 ));
             }
             DataType::Bytes => {
-                strs.push((format!("{:#?}", bytes), None));
+                strs.push((format!("{bytes:#?}"), None));
             }
             DataType::String => {
                 if let Ok(cstr) = CStr::from_bytes_until_nul(bytes) {
-                    strs.push((format!("{:?}", cstr), None));
+                    strs.push((format!("{cstr:?}"), None));
                 }
                 if let Some(nul_idx) = bytes.iter().position(|&c| c == b'\0') {
                     let (cow, _, had_errors) = SHIFT_JIS.decode(&bytes[..nul_idx]);
                     if !had_errors {
-                        let str = format!("{:?}", cow);
+                        let str = format!("{cow:?}");
                         // Only add the Shift JIS string if it's different from the ASCII string.
                         if !strs.iter().any(|x| x.0 == str) {
                             strs.push((str, Some("Shift JIS".into())));
@@ -182,46 +193,108 @@ impl DataType {
     }
 }
 
-pub trait Arch: Send + Sync + Debug {
-    // Finishes arch-specific initialization that must be done after sections have been combined.
-    fn post_init(&mut self, _sections: &[Section], _symbols: &[Symbol]) {}
-
+impl dyn Arch {
     /// Generate a list of instructions references (offset, size, opcode) from the given code.
     ///
-    /// The opcode IDs are used to generate the initial diff. Implementations should do as little
-    /// parsing as possible here: just enough to identify the base instruction opcode, size, and
-    /// possible branch destination (for visual representation). As needed, instructions are parsed
-    /// via `process_instruction` to compare their arguments.
-    fn scan_instructions(
+    /// See [`scan_instructions_internal`] for more details.
+    pub fn scan_instructions(
         &self,
-        address: u64,
-        code: &[u8],
-        section_index: usize,
-        relocations: &[Relocation],
+        resolved: ResolvedSymbol,
         diff_config: &DiffObjConfig,
-    ) -> Result<Vec<ScannedInstruction>>;
+    ) -> Result<Vec<InstructionRef>> {
+        let mut result = self.scan_instructions_internal(
+            resolved.symbol.address,
+            resolved.data,
+            resolved.section_index,
+            &resolved.section.relocations,
+            diff_config,
+        )?;
+
+        let function_start = resolved.symbol.address;
+        let function_end = function_start + resolved.symbol.size;
+
+        // Remove any branch destinations that are outside the function range
+        for ins in result.iter_mut() {
+            if let Some(branch_dest) = ins.branch_dest
+                && (branch_dest < function_start || branch_dest >= function_end)
+            {
+                ins.branch_dest = None;
+            }
+        }
+
+        // Resolve relocation targets within the same function to branch destinations
+        let mut ins_iter = result.iter_mut().peekable();
+        'outer: for reloc in resolved
+            .section
+            .relocations
+            .iter()
+            .skip_while(|r| r.address < function_start)
+            .take_while(|r| r.address < function_end)
+        {
+            let ins = loop {
+                let Some(ins) = ins_iter.peek_mut() else {
+                    break 'outer;
+                };
+                if reloc.address < ins.address {
+                    continue 'outer;
+                }
+                let ins = ins_iter.next().unwrap();
+                if reloc.address >= ins.address && reloc.address < ins.address + ins.size as u64 {
+                    break ins;
+                }
+            };
+            // Clear existing branch destination for instructions with relocations
+            ins.branch_dest = None;
+            let Some(target) = resolved.obj.symbols.get(reloc.target_symbol) else {
+                continue;
+            };
+            if target.section != Some(resolved.section_index) {
+                continue;
+            }
+            let Some(target_address) = target.address.checked_add_signed(reloc.addend) else {
+                continue;
+            };
+            // If the target address is within the function range, set it as a branch destination
+            if target_address >= function_start && target_address < function_end {
+                ins.branch_dest = Some(target_address);
+            }
+        }
+
+        Ok(result)
+    }
 
     /// Parse an instruction to gather its mnemonic and arguments for more detailed comparison.
     ///
     /// This is called only when we need to compare the arguments of an instruction.
-    fn process_instruction(
+    pub fn process_instruction(
         &self,
         resolved: ResolvedInstructionRef,
         diff_config: &DiffObjConfig,
     ) -> Result<ParsedInstruction> {
         let mut mnemonic = None;
         let mut args = Vec::with_capacity(8);
+        let mut relocation_emitted = false;
         self.display_instruction(resolved, diff_config, &mut |part| {
             match part {
                 InstructionPart::Opcode(m, _) => mnemonic = Some(Cow::Owned(m.into_owned())),
-                InstructionPart::Arg(arg) => args.push(arg.into_static()),
+                InstructionPart::Arg(arg) => {
+                    if arg == InstructionArg::Reloc {
+                        relocation_emitted = true;
+                        // If the relocation was resolved to a branch destination, emit that instead.
+                        if let Some(dest) = resolved.ins_ref.branch_dest {
+                            args.push(InstructionArg::BranchDest(dest));
+                            return Ok(());
+                        }
+                    }
+                    args.push(arg.into_static());
+                }
                 _ => {}
             }
             Ok(())
         })?;
         // If the instruction has a relocation, but we didn't format it in the display, add it to
         // the end of the arguments list.
-        if resolved.relocation.is_some() && !args.contains(&InstructionArg::Reloc) {
+        if resolved.relocation.is_some() && !relocation_emitted {
             args.push(InstructionArg::Reloc);
         }
         Ok(ParsedInstruction {
@@ -230,6 +303,26 @@ pub trait Arch: Send + Sync + Debug {
             args,
         })
     }
+}
+
+pub trait Arch: Send + Sync + Debug {
+    /// Finishes arch-specific initialization that must be done after sections have been combined.
+    fn post_init(&mut self, _sections: &[Section], _symbols: &[Symbol]) {}
+
+    /// Generate a list of instructions references (offset, size, opcode) from the given code.
+    ///
+    /// The opcode IDs are used to generate the initial diff. Implementations should do as little
+    /// parsing as possible here: just enough to identify the base instruction opcode, size, and
+    /// possible branch destination (for visual representation). As needed, instructions are parsed
+    /// via `process_instruction` to compare their arguments.
+    fn scan_instructions_internal(
+        &self,
+        address: u64,
+        code: &[u8],
+        section_index: usize,
+        relocations: &[Relocation],
+        diff_config: &DiffObjConfig,
+    ) -> Result<Vec<InstructionRef>>;
 
     /// Format an instruction for display.
     ///
@@ -253,14 +346,26 @@ pub trait Arch: Send + Sync + Debug {
         Vec::new()
     }
 
-    fn implcit_addend(
+    // Perform detailed data flow analysis
+    fn data_flow_analysis(
         &self,
-        file: &object::File<'_>,
-        section: &object::Section,
-        address: u64,
-        relocation: &object::Relocation,
-        flags: RelocationFlags,
-    ) -> Result<i64>;
+        _obj: &Object,
+        _symbol: &Symbol,
+        _code: &[u8],
+        _relocations: &[Relocation],
+    ) -> Option<Box<dyn FlowAnalysisResult>> {
+        None
+    }
+
+    fn relocation_override(
+        &self,
+        _file: &object::File<'_>,
+        _section: &object::Section,
+        _address: u64,
+        _relocation: &object::Relocation,
+    ) -> Result<Option<RelocationOverride>> {
+        Ok(None)
+    }
 
     fn demangle(&self, _name: &str) -> Option<String> { None }
 
@@ -301,13 +406,24 @@ pub trait Arch: Send + Sync + Debug {
     ) -> Vec<ContextItem> {
         Vec::new()
     }
+
+    fn infer_function_size(
+        &self,
+        symbol: &Symbol,
+        _section: &Section,
+        next_address: u64,
+    ) -> Result<u64> {
+        Ok(next_address.saturating_sub(symbol.address))
+    }
 }
 
 pub fn new_arch(object: &object::File) -> Result<Box<dyn Arch>> {
     use object::Object as _;
     Ok(match object.architecture() {
         #[cfg(feature = "ppc")]
-        object::Architecture::PowerPc => Box::new(ppc::ArchPpc::new(object)?),
+        object::Architecture::PowerPc | object::Architecture::PowerPc64 => {
+            Box::new(ppc::ArchPpc::new(object)?)
+        }
         #[cfg(feature = "mips")]
         object::Architecture::Mips => Box::new(mips::ArchMips::new(object)?),
         #[cfg(feature = "x86")]
@@ -332,14 +448,14 @@ impl ArchDummy {
 }
 
 impl Arch for ArchDummy {
-    fn scan_instructions(
+    fn scan_instructions_internal(
         &self,
         _address: u64,
         _code: &[u8],
         _section_index: usize,
         _relocations: &[Relocation],
         _diff_config: &DiffObjConfig,
-    ) -> Result<Vec<ScannedInstruction>> {
+    ) -> Result<Vec<InstructionRef>> {
         Ok(Vec::new())
     }
 
@@ -352,16 +468,19 @@ impl Arch for ArchDummy {
         Ok(())
     }
 
-    fn implcit_addend(
-        &self,
-        _file: &object::File<'_>,
-        _section: &object::Section,
-        _address: u64,
-        _relocation: &object::Relocation,
-        _flags: RelocationFlags,
-    ) -> Result<i64> {
-        Ok(0)
-    }
-
     fn data_reloc_size(&self, _flags: RelocationFlags) -> usize { 0 }
 }
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub enum RelocationOverrideTarget {
+    Keep,
+    Skip,
+    Symbol(object::SymbolIndex),
+    Section(object::SectionIndex),
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub struct RelocationOverride {
+    pub target: RelocationOverrideTarget,
+    pub addend: i64,
+}