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.lock

@@ -722,7 +722,7 @@ checksum = "fe6d2e5af09e8c8ad56c969f2157a3d4238cebc7c55f0a517728c38f7b200f81"
 dependencies = [
  "serde",
  "termcolor",
- "unicode-width 0.1.14",
+ "unicode-width 0.2.0",
 ]
 
 [[package]]
@@ -3437,7 +3437,7 @@ dependencies = [
 
 [[package]]
 name = "objdiff-cli"
-version = "3.0.0-beta.12"
+version = "3.0.0-beta.13"
 dependencies = [
  "anyhow",
  "argp",
@@ -3460,7 +3460,7 @@ dependencies = [
 
 [[package]]
 name = "objdiff-core"
-version = "3.0.0-beta.12"
+version = "3.0.0-beta.13"
 dependencies = [
  "anyhow",
  "arm-attr",
@@ -3505,6 +3505,7 @@ dependencies = [
  "syn",
  "tempfile",
  "time",
+ "typed-arena",
  "typed-path",
  "unarm",
  "winapi",
@@ -3514,7 +3515,7 @@ dependencies = [
 
 [[package]]
 name = "objdiff-gui"
-version = "3.0.0-beta.12"
+version = "3.0.0-beta.13"
 dependencies = [
  "anyhow",
  "cfg-if",
@@ -3550,7 +3551,7 @@ dependencies = [
 
 [[package]]
 name = "objdiff-wasm"
-version = "3.0.0-beta.12"
+version = "3.0.0-beta.13"
 dependencies = [
  "log",
  "objdiff-core",
@@ -4014,7 +4015,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "ac6c3320f9abac597dcbc668774ef006702672474aad53c6d596b62e487b40b1"
 dependencies = [
  "heck",
- "itertools 0.13.0",
+ "itertools 0.14.0",
  "log",
  "multimap",
  "once_cell",
@@ -4034,7 +4035,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "9120690fafc389a67ba3803df527d0ec9cbbc9cc45e4cc20b332996dfb672425"
 dependencies = [
  "anyhow",
- "itertools 0.13.0",
+ "itertools 0.14.0",
  "proc-macro2",
  "quote",
  "syn",
@@ -5620,6 +5621,12 @@ dependencies = [
  "rustc-hash 2.1.1",
 ]
 
+[[package]]
+name = "typed-arena"
+version = "2.0.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "6af6ae20167a9ece4bcb41af5b80f8a1f1df981f6391189ce00fd257af04126a"
+
 [[package]]
 name = "typed-path"
 version = "0.11.0"
@@ -6257,7 +6264,7 @@ version = "0.1.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "cf221c93e13a30d793f7645a0e7762c55d169dbb0a49671918a2319d289b10bb"
 dependencies = [
- "windows-sys 0.48.0",
+ "windows-sys 0.59.0",
 ]
 
 [[package]]

Cargo.toml

@@ -14,7 +14,7 @@ strip = "debuginfo"
 codegen-units = 1
 
 [workspace.package]
-version = "3.0.0-beta.12"
+version = "3.0.0-beta.13"
 authors = ["Luke Street <luke@street.dev>"]
 edition = "2024"
 license = "MIT OR Apache-2.0"

objdiff-cli/src/cmd/report.rs

@@ -248,13 +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,

objdiff-cli/src/views/function_diff.rs

@@ -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 =
@@ -561,10 +560,12 @@ 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,

objdiff-core/Cargo.toml

@@ -62,7 +62,10 @@ config = [
     "dep:semver",
     "dep:typed-path",
 ]
-dwarf = ["dep:gimli"]
+dwarf = [
+    "dep:gimli",
+    "dep:typed-arena",
+]
 serde = [
     "dep:pbjson",
     "dep:pbjson-build",
@@ -78,6 +81,7 @@ std = [
     "prost?/std",
     "serde?/std",
     "similar?/std",
+    "typed-arena?/std",
     "typed-path?/std",
     "dep:filetime",
     "dep:memmap2",
@@ -143,6 +147,7 @@ serde_json = { version = "1.0", default-features = false, features = ["alloc"],
 
 # dwarf
 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 }

objdiff-core/src/arch/mips.rs

@@ -15,7 +15,7 @@ use crate::{
     diff::{DiffObjConfig, MipsAbi, MipsInstrCategory, display::InstructionPart},
     obj::{
         InstructionArg, InstructionArgValue, InstructionRef, Relocation, RelocationFlags,
-        ResolvedInstructionRef, ResolvedRelocation, SymbolFlag, SymbolFlagSet,
+        ResolvedInstructionRef, ResolvedRelocation, Section, Symbol, SymbolFlag, SymbolFlagSet,
     },
 };
 
@@ -140,6 +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,
@@ -151,7 +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,
@@ -234,17 +242,16 @@ impl Arch for ArchMips {
             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 {
-                        if let Some(addend) = self
+                    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,
-                            }));
-                        }
+                    {
+                        return Ok(Some(RelocationOverride {
+                            target: RelocationOverrideTarget::Keep,
+                            addend,
+                        }));
                     }
 
                     let data = section.data()?;
@@ -331,6 +338,36 @@ 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(

objdiff-core/src/arch/mod.rs

@@ -215,10 +215,10 @@ impl dyn Arch {
 
         // Remove any branch destinations that are outside the function range
         for ins in result.iter_mut() {
-            if let Some(branch_dest) = ins.branch_dest {
-                if branch_dest < function_start || branch_dest >= function_end {
-                    ins.branch_dest = None;
-                }
+            if let Some(branch_dest) = ins.branch_dest
+                && (branch_dest < function_start || branch_dest >= function_end)
+            {
+                ins.branch_dest = None;
             }
         }
 
@@ -406,6 +406,15 @@ 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>> {

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

@@ -514,14 +514,14 @@ pub fn ppc_data_flow_analysis(
 }
 
 fn get_string_data(obj: &Object, symbol_index: usize, offset: Simm) -> Option<&str> {
-    if let Some(sym) = obj.symbols.get(symbol_index) {
-        if sym.name.starts_with("@stringBase") && offset.0 != 0 {
-            if let Some(data) = obj.symbol_data(symbol_index) {
-                let bytes = &data[offset.0 as usize..];
-                if let Ok(Ok(str)) = CStr::from_bytes_until_nul(bytes).map(|x| x.to_str()) {
-                    return Some(str);
-                }
-            }
+    if let Some(sym) = obj.symbols.get(symbol_index)
+        && sym.name.starts_with("@stringBase")
+        && offset.0 != 0
+        && let Some(data) = obj.symbol_data(symbol_index)
+    {
+        let bytes = &data[offset.0 as usize..];
+        if let Ok(Ok(str)) = CStr::from_bytes_until_nul(bytes).map(|x| x.to_str()) {
+            return Some(str);
         }
     }
     None
@@ -577,19 +577,17 @@ fn generate_flow_analysis_result(
         let registers = register_state_at.get(index as usize).unwrap_or(&default_register_state);
         if let (powerpc::Opcode::Addi, Argument::GPR(rel), Argument::Simm(offset)) =
             (ins.op, args[1], args[2])
+            && let RegisterContent::Symbol(sym_index) = registers[rel]
+            && let Some(str) = get_string_data(obj, sym_index, offset)
         {
-            if let RegisterContent::Symbol(sym_index) = registers[rel] {
-                if let Some(str) = get_string_data(obj, sym_index, offset) {
-                    // Show the string constant in the analysis result
-                    let formatted = format!("\"{str}\"");
-                    analysis_result.set_argument_value_at_address(
-                        ins_address,
-                        2,
-                        FlowAnalysisValue::Text(clamp_text_length(formatted, 20)),
-                    );
-                    // Don't continue, we want to show the stringbase value as well
-                }
-            }
+            // Show the string constant in the analysis result
+            let formatted = format!("\"{str}\"");
+            analysis_result.set_argument_value_at_address(
+                ins_address,
+                2,
+                FlowAnalysisValue::Text(clamp_text_length(formatted, 20)),
+            );
+            // Don't continue, we want to show the stringbase value as well
         }
 
         let is_store = is_store_instruction(ins.op);

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

@@ -20,7 +20,7 @@ use crate::{
     },
     obj::{
         FlowAnalysisResult, InstructionRef, Object, Relocation, RelocationFlags,
-        ResolvedInstructionRef, ResolvedRelocation, Symbol, SymbolFlag, SymbolFlagSet,
+        ResolvedInstructionRef, ResolvedRelocation, Section, Symbol, SymbolFlag, SymbolFlagSet,
     },
 };
 
@@ -66,7 +66,9 @@ impl ArchPpc {
                 if file.is_64() {
                     powerpc::Extension::Ppc64 | powerpc::Extension::AltiVec
                 } else {
-                    powerpc::Extension::AltiVec.into()
+                    // Gekko/Broadway objects often use the EF_PPC_EMB flag,
+                    // but ProDG in particular does not emit it.
+                    powerpc::Extensions::gekko_broadway()
                 }
             }
         };
@@ -455,6 +457,22 @@ impl Arch for ArchPpc {
         }
         out
     }
+
+    fn infer_function_size(
+        &self,
+        symbol: &Symbol,
+        section: &Section,
+        mut next_address: u64,
+    ) -> Result<u64> {
+        // Trim any trailing 4-byte zeroes from the end (padding)
+        while next_address >= symbol.address + 4
+            && let Some(data) = section.data_range(next_address - 4, 4)
+            && data == [0u8; 4]
+        {
+            next_address -= 4;
+        }
+        Ok(next_address.saturating_sub(symbol.address))
+    }
 }
 
 impl ArchPpc {
@@ -850,44 +868,43 @@ fn generate_fake_pool_relocations_for_function(
                     break;
                 }
             }
-            if let Some(branch_dest) = branch_dest {
-                if branch_dest >= func_address as u32
-                    && (branch_dest - func_address as u32) < code.len() as u32
-                {
-                    let dest_offset_into_func = branch_dest - func_address as u32;
-                    let dest_code_slice = &code[dest_offset_into_func as usize..];
-                    match ins.op {
-                        Opcode::Bc => {
-                            // Conditional branch.
-                            // Add the branch destination to the queue to do later.
+            if let Some(branch_dest) = branch_dest
+                && branch_dest >= func_address as u32
+                && (branch_dest - func_address as u32) < code.len() as u32
+            {
+                let dest_offset_into_func = branch_dest - func_address as u32;
+                let dest_code_slice = &code[dest_offset_into_func as usize..];
+                match ins.op {
+                    Opcode::Bc => {
+                        // Conditional branch.
+                        // Add the branch destination to the queue to do later.
+                        ins_iters_with_gpr_state.push((
+                            InsIter::new(dest_code_slice, branch_dest, extensions),
+                            gpr_pool_relocs.clone(),
+                        ));
+                        // Then continue on with the current iterator.
+                    }
+                    Opcode::B => {
+                        if simplified.mnemonic != "bl" {
+                            // Unconditional branch.
+                            // Add the branch destination to the queue.
                             ins_iters_with_gpr_state.push((
                                 InsIter::new(dest_code_slice, branch_dest, extensions),
                                 gpr_pool_relocs.clone(),
                             ));
-                            // Then continue on with the current iterator.
+                            // Break out of the current iterator so we can do the newly added one.
+                            break;
                         }
-                        Opcode::B => {
-                            if simplified.mnemonic != "bl" {
-                                // Unconditional branch.
-                                // Add the branch destination to the queue.
-                                ins_iters_with_gpr_state.push((
-                                    InsIter::new(dest_code_slice, branch_dest, extensions),
-                                    gpr_pool_relocs.clone(),
-                                ));
-                                // Break out of the current iterator so we can do the newly added one.
-                                break;
-                            }
-                        }
-                        _ => unreachable!(),
                     }
+                    _ => unreachable!(),
                 }
             }
-            if let Opcode::Bcctr = ins.op {
-                if simplified.mnemonic == "bctr" {
-                    // Unconditional branch to count register.
-                    // Likely a jump table.
-                    gpr_state_at_bctr.insert(cur_addr, gpr_pool_relocs.clone());
-                }
+            if let Opcode::Bcctr = ins.op
+                && simplified.mnemonic == "bctr"
+            {
+                // Unconditional branch to count register.
+                // Likely a jump table.
+                gpr_state_at_bctr.insert(cur_addr, gpr_pool_relocs.clone());
             }
 
             // Then handle keeping track of which GPR contains which pool relocation.

objdiff-core/src/arch/x86.rs

@@ -11,7 +11,7 @@ use object::{Endian as _, Object as _, ObjectSection as _, elf, pe};
 use crate::{
     arch::{Arch, RelocationOverride, RelocationOverrideTarget},
     diff::{DiffObjConfig, X86Formatter, display::InstructionPart},
-    obj::{InstructionRef, Relocation, RelocationFlags, ResolvedInstructionRef},
+    obj::{InstructionRef, Relocation, RelocationFlags, ResolvedInstructionRef, Section, Symbol},
 };
 
 #[derive(Debug)]
@@ -303,6 +303,52 @@ impl Arch for ArchX86 {
     fn data_reloc_size(&self, flags: RelocationFlags) -> usize {
         self.reloc_size(flags).unwrap_or(1)
     }
+
+    fn infer_function_size(
+        &self,
+        symbol: &Symbol,
+        section: &Section,
+        next_address: u64,
+    ) -> Result<u64> {
+        let Ok(size) = (next_address - symbol.address).try_into() else {
+            return Ok(next_address.saturating_sub(symbol.address));
+        };
+        let Some(code) = section.data_range(symbol.address, size) else {
+            return Ok(0);
+        };
+        // Decode instructions to find the last non-NOP instruction
+        let mut decoder = self.decoder(code, symbol.address);
+        let mut instruction = Instruction::default();
+        let mut new_address = 0;
+        let mut reloc_iter = section.relocations.iter().peekable();
+        'outer: while decoder.can_decode() {
+            let address = decoder.ip();
+            while let Some(reloc) = reloc_iter.peek() {
+                match reloc.address.cmp(&address) {
+                    Ordering::Less => {
+                        reloc_iter.next();
+                    }
+                    Ordering::Equal => {
+                        // If the instruction starts at a relocation, it's inline data
+                        let reloc_size = self.reloc_size(reloc.flags).with_context(|| {
+                            format!("Unsupported inline x86 relocation {:?}", reloc.flags)
+                        })?;
+                        if decoder.set_position(decoder.position() + reloc_size).is_ok() {
+                            new_address = address + reloc_size as u64;
+                            decoder.set_ip(new_address);
+                            continue 'outer;
+                        }
+                    }
+                    Ordering::Greater => break,
+                }
+            }
+            decoder.decode_out(&mut instruction);
+            if instruction.mnemonic() != iced_x86::Mnemonic::Nop {
+                new_address = instruction.next_ip();
+            }
+        }
+        Ok(new_address.saturating_sub(symbol.address))
+    }
 }
 
 struct InstructionFormatterOutput<'a> {

objdiff-core/src/diff/display.rs

@@ -385,13 +385,13 @@ pub fn symbol_context(obj: &Object, symbol_index: usize) -> Vec<ContextItem> {
     if let Some(name) = &symbol.demangled_name {
         out.push(ContextItem::Copy { value: name.clone(), label: None });
     }
-    if symbol.section.is_some() {
-        if let Some(address) = symbol.virtual_address {
-            out.push(ContextItem::Copy {
-                value: format!("{address:x}"),
-                label: Some("virtual address".to_string()),
-            });
-        }
+    if symbol.section.is_some()
+        && let Some(address) = symbol.virtual_address
+    {
+        out.push(ContextItem::Copy {
+            value: format!("{address:x}"),
+            label: Some("virtual address".to_string()),
+        });
     }
     out.append(&mut obj.arch.symbol_context(obj, symbol_index));
     out

objdiff-core/src/diff/mod.rs

@@ -467,15 +467,15 @@ fn apply_symbol_mappings(
 ) -> Result<()> {
     // If we're selecting a symbol to use as a comparison, mark it as used
     // This ensures that we don't match it to another symbol at any point
-    if let Some(left_name) = &mapping_config.selecting_left {
-        if let Some(left_symbol) = left.symbol_by_name(left_name) {
-            left_used.insert(left_symbol);
-        }
+    if let Some(left_name) = &mapping_config.selecting_left
+        && let Some(left_symbol) = left.symbol_by_name(left_name)
+    {
+        left_used.insert(left_symbol);
     }
-    if let Some(right_name) = &mapping_config.selecting_right {
-        if let Some(right_symbol) = right.symbol_by_name(right_name) {
-            right_used.insert(right_symbol);
-        }
+    if let Some(right_name) = &mapping_config.selecting_right
+        && let Some(right_symbol) = right.symbol_by_name(right_name)
+    {
+        right_used.insert(right_symbol);
     }
 
     // Apply manual symbol mappings
@@ -639,17 +639,16 @@ fn find_symbol(
     // If they are at the same address in the same section
     if in_symbol.name.starts_with('@')
         && matches!(section_kind, SectionKind::Data | SectionKind::Bss)
-    {
-        if let Some((symbol_idx, _)) = unmatched_symbols(obj, used).find(|(_, symbol)| {
+        && let Some((symbol_idx, _)) = unmatched_symbols(obj, used).find(|(_, symbol)| {
             let Some(section_index) = symbol.section else {
                 return false;
             };
             symbol.name.starts_with('@')
                 && symbol.address == in_symbol.address
                 && obj.sections[section_index].name == section_name
-        }) {
-            return Some(symbol_idx);
-        }
+        })
+    {
+        return Some(symbol_idx);
     }
     // Match Metrowerks symbol$1234 against symbol$2345
     if let Some((prefix, suffix)) = in_symbol.name.split_once('$') {

objdiff-core/src/obj/dwarf2.rs

@@ -0,0 +1,109 @@
+use alloc::{borrow::Cow, vec::Vec};
+
+use anyhow::{Context, Result};
+use object::{Object as _, ObjectSection as _};
+use typed_arena::Arena;
+
+use crate::obj::{Section, SectionKind};
+
+/// Parse line information from DWARF 2+ sections.
+pub(crate) fn parse_line_info_dwarf2(
+    obj_file: &object::File,
+    sections: &mut [Section],
+) -> Result<()> {
+    let arena_data = Arena::new();
+    let arena_relocations = Arena::new();
+    let endian = match obj_file.endianness() {
+        object::Endianness::Little => gimli::RunTimeEndian::Little,
+        object::Endianness::Big => gimli::RunTimeEndian::Big,
+    };
+    let dwarf = gimli::Dwarf::load(|id: gimli::SectionId| -> Result<_> {
+        load_file_section(id, obj_file, endian, &arena_data, &arena_relocations)
+    })
+    .context("loading DWARF sections")?;
+
+    let mut iter = dwarf.units();
+    if let Some(header) = iter.next().map_err(|e| gimli_error(e, "iterating over DWARF units"))? {
+        let unit = dwarf.unit(header).map_err(|e| gimli_error(e, "loading DWARF unit"))?;
+        if let Some(program) = unit.line_program.clone() {
+            let mut text_sections = sections.iter_mut().filter(|s| s.kind == SectionKind::Code);
+            let mut lines = text_sections.next().map(|section| &mut section.line_info);
+
+            let mut rows = program.rows();
+            while let Some((_header, row)) =
+                rows.next_row().map_err(|e| gimli_error(e, "loading program row"))?
+            {
+                if let (Some(line), Some(lines)) = (row.line(), &mut lines) {
+                    lines.insert(row.address(), line.get() as u32);
+                }
+                if row.end_sequence() {
+                    // The next row is the start of a new sequence, which means we must
+                    // advance to the next .text section.
+                    lines = text_sections.next().map(|section| &mut section.line_info);
+                }
+            }
+        }
+    }
+    if iter.next().map_err(|e| gimli_error(e, "checking for next unit"))?.is_some() {
+        log::warn!("Multiple units found in DWARF data, only processing the first");
+    }
+
+    Ok(())
+}
+
+#[derive(Debug, Default)]
+struct RelocationMap(object::read::RelocationMap);
+
+impl RelocationMap {
+    fn add(&mut self, file: &object::File, section: &object::Section) {
+        for (offset, relocation) in section.relocations() {
+            if let Err(e) = self.0.add(file, offset, relocation) {
+                log::error!(
+                    "Relocation error for section {} at offset 0x{:08x}: {}",
+                    section.name().unwrap(),
+                    offset,
+                    e
+                );
+            }
+        }
+    }
+}
+
+impl gimli::read::Relocate for &'_ RelocationMap {
+    fn relocate_address(&self, offset: usize, value: u64) -> gimli::Result<u64> {
+        Ok(self.0.relocate(offset as u64, value))
+    }
+
+    fn relocate_offset(&self, offset: usize, value: usize) -> gimli::Result<usize> {
+        <usize as gimli::ReaderOffset>::from_u64(self.0.relocate(offset as u64, value as u64))
+    }
+}
+
+type Relocate<'a, R> = gimli::RelocateReader<R, &'a RelocationMap>;
+
+fn load_file_section<'input, 'arena, Endian: gimli::Endianity>(
+    id: gimli::SectionId,
+    file: &object::File<'input>,
+    endian: Endian,
+    arena_data: &'arena Arena<Cow<'input, [u8]>>,
+    arena_relocations: &'arena Arena<RelocationMap>,
+) -> Result<Relocate<'arena, gimli::EndianSlice<'arena, Endian>>> {
+    let mut relocations = RelocationMap::default();
+    let data = match file.section_by_name(id.name()) {
+        Some(ref section) => {
+            relocations.add(file, section);
+            section.uncompressed_data()?
+        }
+        // Use a non-zero capacity so that `ReaderOffsetId`s are unique.
+        None => Cow::Owned(Vec::with_capacity(1)),
+    };
+    let data_ref = arena_data.alloc(data);
+    let section = gimli::EndianSlice::new(data_ref, endian);
+    let relocations = arena_relocations.alloc(relocations);
+    Ok(Relocate::new(section, relocations))
+}
+
+#[inline]
+fn gimli_error(e: gimli::Error, context: &str) -> anyhow::Error {
+    anyhow::anyhow!("gimli error {context}: {e:?}")
+}

objdiff-core/src/obj/mod.rs

@@ -1,3 +1,5 @@
+#[cfg(feature = "dwarf")]
+mod dwarf2;
 pub mod read;
 pub mod split_meta;
 

objdiff-core/src/obj/read.rs

@@ -122,7 +122,7 @@ fn map_symbols(
     }
 
     // Infer symbol sizes for 0-size symbols
-    infer_symbol_sizes(&mut symbols, sections);
+    infer_symbol_sizes(arch, &mut symbols, sections)?;
 
     Ok((symbols, symbol_indices))
 }
@@ -136,7 +136,7 @@ fn is_local_label(symbol: &Symbol) -> bool {
         && LABEL_PREFIXES.iter().any(|p| symbol.name.starts_with(p))
 }
 
-fn infer_symbol_sizes(symbols: &mut [Symbol], sections: &[Section]) {
+fn infer_symbol_sizes(arch: &dyn Arch, symbols: &mut [Symbol], sections: &[Section]) -> Result<()> {
     // Create a sorted list of symbol indices by section
     let mut symbols_with_section = Vec::<usize>::with_capacity(symbols.len());
     for (i, symbol) in symbols.iter().enumerate() {
@@ -206,18 +206,13 @@ fn infer_symbol_sizes(symbols: &mut [Symbol], sections: &[Section]) {
             iter_idx += 1;
         };
         let section = &sections[section_idx];
-        let mut next_address =
+        let next_address =
             next_symbol.map(|s| s.address).unwrap_or_else(|| section.address + section.size);
-        if section.kind == SectionKind::Code {
-            // For functions, trim any trailing 4-byte zeroes from the end (padding, nops)
-            while next_address > symbol.address + 4
-                && let Some(data) = section.data_range(next_address - 4, 4)
-                && data == [0u8; 4]
-            {
-                next_address -= 4;
-            }
-        }
-        let new_size = next_address.saturating_sub(symbol.address);
+        let new_size = if section.kind == SectionKind::Code {
+            arch.infer_function_size(symbol, section, next_address)?
+        } else {
+            next_address.saturating_sub(symbol.address)
+        };
         if new_size > 0 {
             let symbol = &mut symbols[symbol_idx];
             symbol.size = new_size;
@@ -234,6 +229,7 @@ fn infer_symbol_sizes(symbols: &mut [Symbol], sections: &[Section]) {
             }
         }
     }
+    Ok(())
 }
 
 fn map_sections(
@@ -554,12 +550,11 @@ fn perform_data_flow_analysis(obj: &mut Object, config: &DiffObjConfig) -> Resul
             }
 
             // Optional full data flow analysis
-            if config.analyze_data_flow {
-                if let Some(flow_result) =
+            if config.analyze_data_flow
+                && let Some(flow_result) =
                     obj.arch.data_flow_analysis(obj, symbol, code, &section.relocations)
-                {
-                    generated_flow_results.push((symbol.clone(), flow_result));
-                }
+            {
+                generated_flow_results.push((symbol.clone(), flow_result));
             }
         }
     }
@@ -582,6 +577,28 @@ fn parse_line_info(
     obj_data: &[u8],
 ) -> Result<()> {
     // DWARF 1.1
+    if let Err(e) = parse_line_info_dwarf1(obj_file, sections) {
+        log::warn!("Failed to parse DWARF 1.1 line info: {e}");
+    }
+
+    // DWARF 2+
+    #[cfg(feature = "dwarf")]
+    if let Err(e) = super::dwarf2::parse_line_info_dwarf2(obj_file, sections) {
+        log::warn!("Failed to parse DWARF 2+ line info: {e}");
+    }
+
+    // COFF
+    if let object::File::Coff(coff) = obj_file
+        && let Err(e) = parse_line_info_coff(coff, sections, section_indices, obj_data)
+    {
+        log::warn!("Failed to parse COFF line info: {e}");
+    }
+
+    Ok(())
+}
+
+/// Parse .line section from DWARF 1.1 format.
+fn parse_line_info_dwarf1(obj_file: &object::File, sections: &mut [Section]) -> Result<()> {
     if let Some(section) = obj_file.section_by_name(".line") {
         let data = section.uncompressed_data()?;
         let mut reader: &[u8] = data.as_ref();
@@ -609,55 +626,6 @@ fn parse_line_info(
             }
         }
     }
-
-    // DWARF 2+
-    #[cfg(feature = "dwarf")]
-    {
-        fn gimli_error(e: gimli::Error) -> anyhow::Error { anyhow::anyhow!("DWARF error: {e:?}") }
-        let dwarf_cow = gimli::DwarfSections::load(|id| {
-            Ok::<_, gimli::Error>(
-                obj_file
-                    .section_by_name(id.name())
-                    .and_then(|section| section.uncompressed_data().ok())
-                    .unwrap_or(alloc::borrow::Cow::Borrowed(&[][..])),
-            )
-        })
-        .map_err(gimli_error)?;
-        let endian = match obj_file.endianness() {
-            object::Endianness::Little => gimli::RunTimeEndian::Little,
-            object::Endianness::Big => gimli::RunTimeEndian::Big,
-        };
-        let dwarf = dwarf_cow.borrow(|section| gimli::EndianSlice::new(section, endian));
-        let mut iter = dwarf.units();
-        if let Some(header) = iter.next().map_err(gimli_error)? {
-            let unit = dwarf.unit(header).map_err(gimli_error)?;
-            if let Some(program) = unit.line_program.clone() {
-                let mut text_sections = sections.iter_mut().filter(|s| s.kind == SectionKind::Code);
-                let mut lines = text_sections.next().map(|section| &mut section.line_info);
-
-                let mut rows = program.rows();
-                while let Some((_header, row)) = rows.next_row().map_err(gimli_error)? {
-                    if let (Some(line), Some(lines)) = (row.line(), &mut lines) {
-                        lines.insert(row.address(), line.get() as u32);
-                    }
-                    if row.end_sequence() {
-                        // The next row is the start of a new sequence, which means we must
-                        // advance to the next .text section.
-                        lines = text_sections.next().map(|section| &mut section.line_info);
-                    }
-                }
-            }
-        }
-        if iter.next().map_err(gimli_error)?.is_some() {
-            log::warn!("Multiple units found in DWARF data, only processing the first");
-        }
-    }
-
-    // COFF
-    if let object::File::Coff(coff) = obj_file {
-        parse_line_info_coff(coff, sections, section_indices, obj_data)?;
-    }
-
     Ok(())
 }
 

objdiff-core/src/util.rs

@@ -48,7 +48,7 @@ pub fn align_data_to_4<W: std::io::Write + ?Sized>(
     len: usize,
 ) -> std::io::Result<()> {
     const ALIGN_BYTES: &[u8] = &[0; 4];
-    if len % 4 != 0 {
+    if !len.is_multiple_of(4) {
         writer.write_all(&ALIGN_BYTES[..4 - len % 4])?;
     }
     Ok(())

objdiff-core/tests/snapshots/arch_x86__read_x86_jumptable-2.snap

@@ -507,116 +507,4 @@ expression: diff.instruction_rows
         ),
         arg_diff: [],
     },
-    InstructionDiffRow {
-        ins_ref: Some(
-            InstructionRef {
-                address: 88,
-                size: 1,
-                opcode: 465,
-                branch_dest: None,
-            },
-        ),
-        kind: None,
-        branch_from: None,
-        branch_to: None,
-        arg_diff: [],
-    },
-    InstructionDiffRow {
-        ins_ref: Some(
-            InstructionRef {
-                address: 89,
-                size: 1,
-                opcode: 465,
-                branch_dest: None,
-            },
-        ),
-        kind: None,
-        branch_from: None,
-        branch_to: None,
-        arg_diff: [],
-    },
-    InstructionDiffRow {
-        ins_ref: Some(
-            InstructionRef {
-                address: 90,
-                size: 1,
-                opcode: 465,
-                branch_dest: None,
-            },
-        ),
-        kind: None,
-        branch_from: None,
-        branch_to: None,
-        arg_diff: [],
-    },
-    InstructionDiffRow {
-        ins_ref: Some(
-            InstructionRef {
-                address: 91,
-                size: 1,
-                opcode: 465,
-                branch_dest: None,
-            },
-        ),
-        kind: None,
-        branch_from: None,
-        branch_to: None,
-        arg_diff: [],
-    },
-    InstructionDiffRow {
-        ins_ref: Some(
-            InstructionRef {
-                address: 92,
-                size: 1,
-                opcode: 465,
-                branch_dest: None,
-            },
-        ),
-        kind: None,
-        branch_from: None,
-        branch_to: None,
-        arg_diff: [],
-    },
-    InstructionDiffRow {
-        ins_ref: Some(
-            InstructionRef {
-                address: 93,
-                size: 1,
-                opcode: 465,
-                branch_dest: None,
-            },
-        ),
-        kind: None,
-        branch_from: None,
-        branch_to: None,
-        arg_diff: [],
-    },
-    InstructionDiffRow {
-        ins_ref: Some(
-            InstructionRef {
-                address: 94,
-                size: 1,
-                opcode: 465,
-                branch_dest: None,
-            },
-        ),
-        kind: None,
-        branch_from: None,
-        branch_to: None,
-        arg_diff: [],
-    },
-    InstructionDiffRow {
-        ins_ref: Some(
-            InstructionRef {
-                address: 95,
-                size: 1,
-                opcode: 465,
-                branch_dest: None,
-            },
-        ),
-        kind: None,
-        branch_from: None,
-        branch_to: None,
-        arg_diff: [],
-    },
 ]

objdiff-core/tests/snapshots/arch_x86__read_x86_jumptable-3.snap

@@ -29,11 +29,3 @@ expression: output
 [(Address(76), Normal, 5), (Spacing(4), Normal, 0), (Opcode(".dword", 65534), Normal, 10), (BranchDest(41), Normal, 0), (Basic(" ~>"), Rotating(6), 0), (Eol, Normal, 0)]
 [(Address(80), Normal, 5), (Spacing(4), Normal, 0), (Opcode(".dword", 65534), Normal, 10), (BranchDest(47), Normal, 0), (Basic(" ~>"), Rotating(7), 0), (Eol, Normal, 0)]
 [(Address(84), Normal, 5), (Spacing(4), Normal, 0), (Opcode(".dword", 65534), Normal, 10), (BranchDest(53), Normal, 0), (Basic(" ~>"), Rotating(8), 0), (Eol, Normal, 0)]
-[(Address(88), Normal, 5), (Spacing(4), Normal, 0), (Opcode("nop", 465), Normal, 10), (Eol, Normal, 0)]
-[(Address(89), Normal, 5), (Spacing(4), Normal, 0), (Opcode("nop", 465), Normal, 10), (Eol, Normal, 0)]
-[(Address(90), Normal, 5), (Spacing(4), Normal, 0), (Opcode("nop", 465), Normal, 10), (Eol, Normal, 0)]
-[(Address(91), Normal, 5), (Spacing(4), Normal, 0), (Opcode("nop", 465), Normal, 10), (Eol, Normal, 0)]
-[(Address(92), Normal, 5), (Spacing(4), Normal, 0), (Opcode("nop", 465), Normal, 10), (Eol, Normal, 0)]
-[(Address(93), Normal, 5), (Spacing(4), Normal, 0), (Opcode("nop", 465), Normal, 10), (Eol, Normal, 0)]
-[(Address(94), Normal, 5), (Spacing(4), Normal, 0), (Opcode("nop", 465), Normal, 10), (Eol, Normal, 0)]
-[(Address(95), Normal, 5), (Spacing(4), Normal, 0), (Opcode("nop", 465), Normal, 10), (Eol, Normal, 0)]

objdiff-core/tests/snapshots/arch_x86__read_x86_jumptable.snap

@@ -63,7 +63,7 @@ Object {
                 "int __cdecl test(int)",
             ),
             address: 0,
-            size: 96,
+            size: 88,
             kind: Function,
             section: Some(
                 1,