Fix read error on objects with no .text section (#67)

* Fix read error on objects with no .text section

* Fix read error on DWARF 1.1 objects

* Revert DWARF 1 changes

---------

Co-authored-by: Luke Street <luke@street.dev>

Cargo.lock

@@ -3081,7 +3081,7 @@ dependencies = [
 
 [[package]]
 name = "objdiff-cli"
-version = "2.0.0-alpha.1"
+version = "2.0.0-alpha.3"
 dependencies = [
  "anyhow",
  "argp",
@@ -3100,7 +3100,7 @@ dependencies = [
 
 [[package]]
 name = "objdiff-core"
-version = "2.0.0-alpha.1"
+version = "2.0.0-alpha.3"
 dependencies = [
  "anyhow",
  "byteorder",
@@ -3123,11 +3123,13 @@ dependencies = [
  "serde_json",
  "serde_yaml",
  "similar",
+ "strum",
+ "unarm",
 ]
 
 [[package]]
 name = "objdiff-gui"
-version = "2.0.0-alpha.1"
+version = "2.0.0-alpha.3"
 dependencies = [
  "anyhow",
  "bytes",
@@ -3157,6 +3159,7 @@ dependencies = [
  "serde",
  "serde_json",
  "shell-escape",
+ "strum",
  "tempfile",
  "time",
  "tracing-subscriber",
@@ -3543,9 +3546,9 @@ dependencies = [
 
 [[package]]
 name = "rabbitizer"
-version = "1.10.0"
+version = "1.11.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f570a07717dcb6edf85e6652d92671c49e8728014c6ec1577a81b4d949725e27"
+checksum = "305e6fb948a8a884ba996ac4efb1d7b0ee44d0bbfcd86b9c0f0fc0aa0aa0fc21"
 dependencies = [
  "cc",
  "glob",
@@ -4693,6 +4696,12 @@ dependencies = [
  "winapi",
 ]
 
+[[package]]
+name = "unarm"
+version = "1.0.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1c13fc9a9c95348bf7565e5c30688fc288239962958cac0ccdc7cd009141d850"
+
 [[package]]
 name = "unicase"
 version = "2.7.0"

Cargo.toml

@@ -8,5 +8,7 @@ resolver = "2"
 
 [profile.release-lto]
 inherits = "release"
-lto = "thin"
+# Temporarily disabled to fix notify crash
+# See https://github.com/encounter/objdiff/issues/66
+#lto = "thin"
 strip = "debuginfo"

README.md

@@ -14,9 +14,10 @@ Features:
 - Click to highlight all instances of values and registers.
 
 Supports:
-- PowerPC 750CL (GameCube & Wii)
-- MIPS (Nintendo 64 & PS2)
+- PowerPC 750CL (GameCube, Wii)
+- MIPS (N64, PS1, PS2, PSP)
 - x86 (COFF only at the moment)
+- ARMv5 (DS)
 
 See [Usage](#usage) for more information.
 

objdiff-cli/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "objdiff-cli"
-version = "2.0.0-alpha.1"
+version = "2.0.0-alpha.3"
 edition = "2021"
 rust-version = "1.70"
 authors = ["Luke Street <luke@street.dev>"]

objdiff-cli/src/cmd/diff.rs

@@ -821,8 +821,10 @@ impl FunctionDiffUi {
             .transpose()?;
         let config = diff::DiffObjConfig {
             relax_reloc_diffs: self.relax_reloc_diffs,
-            space_between_args: true,          // TODO
-            x86_formatter: Default::default(), // TODO
+            space_between_args: true,                // TODO
+            x86_formatter: Default::default(),       // TODO
+            mips_abi: Default::default(),            // TODO
+            mips_instr_category: Default::default(), // TODO
         };
         let result = diff::diff_objs(&config, target.as_ref(), base.as_ref(), prev.as_ref())?;
 

objdiff-core/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "objdiff-core"
-version = "2.0.0-alpha.1"
+version = "2.0.0-alpha.3"
 edition = "2021"
 rust-version = "1.70"
 authors = ["Luke Street <luke@street.dev>"]
@@ -12,13 +12,14 @@ A local diffing tool for decompilation projects.
 """
 
 [features]
-all = ["config", "dwarf", "mips", "ppc", "x86"]
+all = ["config", "dwarf", "mips", "ppc", "x86", "arm"]
 any-arch = [] # Implicit, used to check if any arch is enabled
 config = ["globset", "semver", "serde_json", "serde_yaml"]
 dwarf = ["gimli"]
 mips = ["any-arch", "rabbitizer"]
 ppc = ["any-arch", "cwdemangle", "ppc750cl"]
 x86 = ["any-arch", "cpp_demangle", "iced-x86", "msvc-demangler"]
+arm = ["any-arch", "cpp_demangle", "unarm"]
 
 [dependencies]
 anyhow = "1.0.82"
@@ -31,6 +32,7 @@ num-traits = "0.2.18"
 object = { version = "0.35.0", features = ["read_core", "std", "elf", "pe"], default-features = false }
 serde = { version = "1", features = ["derive"] }
 similar = { version = "2.5.0", default-features = false }
+strum = { version = "0.26.2", features = ["derive"] }
 
 # config
 globset = { version = "0.4.14", features = ["serde1"], optional = true }
@@ -46,9 +48,12 @@ cwdemangle = { version = "1.0.0", optional = true }
 ppc750cl = { git = "https://github.com/encounter/ppc750cl", rev = "6cbd7d888c7082c2c860f66cbb9848d633f753ed", optional = true }
 
 # mips
-rabbitizer = { version = "1.10.0", optional = true }
+rabbitizer = { version = "1.11.0", optional = true }
 
 # x86
 cpp_demangle = { version = "0.4.3", optional = true }
 iced-x86 = { version = "1.21.0", default-features = false, features = ["std", "decoder", "intel", "gas", "masm", "nasm", "exhaustive_enums"], optional = true }
 msvc-demangler = { version = "0.10.0", optional = true }
+
+# arm
+unarm = { version = "1.0.0", optional = true }

objdiff-core/src/arch/arm.rs

@@ -0,0 +1,296 @@
+use std::{
+    borrow::Cow,
+    collections::{BTreeMap, HashMap},
+};
+
+use anyhow::{bail, Result};
+use object::{
+    elf, File, Object, ObjectSection, ObjectSymbol, Relocation, RelocationFlags, SectionIndex,
+    SectionKind, Symbol,
+};
+use unarm::{
+    args::{Argument, OffsetImm, OffsetReg, Register},
+    parse::{ArmVersion, ParseMode, Parser},
+    ParsedIns,
+};
+
+use crate::{
+    arch::{ObjArch, ProcessCodeResult},
+    diff::DiffObjConfig,
+    obj::{ObjIns, ObjInsArg, ObjInsArgValue, ObjReloc, ObjSection},
+};
+
+pub struct ObjArchArm {
+    /// Maps section index, to list of disasm modes (arm, thumb or data) sorted by address
+    disasm_modes: HashMap<SectionIndex, Vec<DisasmMode>>,
+}
+
+impl ObjArchArm {
+    pub fn new(file: &File) -> Result<Self> {
+        match file {
+            File::Elf32(_) => {
+                let disasm_modes: HashMap<_, _> = file
+                    .sections()
+                    .filter(|s| s.kind() == SectionKind::Text)
+                    .map(|s| {
+                        let index = s.index();
+                        let mut mapping_symbols: Vec<_> = file
+                            .symbols()
+                            .filter(|s| s.section_index().map(|i| i == index).unwrap_or(false))
+                            .filter_map(|s| DisasmMode::from_symbol(&s))
+                            .collect();
+                        mapping_symbols.sort_unstable_by_key(|x| x.address);
+                        (s.index(), mapping_symbols)
+                    })
+                    .collect();
+                Ok(Self { disasm_modes })
+            }
+            _ => bail!("Unsupported file format {:?}", file.format()),
+        }
+    }
+}
+
+impl ObjArch for ObjArchArm {
+    fn process_code(
+        &self,
+        address: u64,
+        code: &[u8],
+        section_index: usize,
+        relocations: &[ObjReloc],
+        line_info: &BTreeMap<u64, u64>,
+        config: &DiffObjConfig,
+    ) -> Result<ProcessCodeResult> {
+        let start_addr = address as u32;
+        let end_addr = start_addr + code.len() as u32;
+
+        // Mapping symbols decide what kind of data comes after it. $a for ARM code, $t for Thumb code and $d for data.
+        let fallback_mappings = [DisasmMode { address: start_addr, mapping: ParseMode::Arm }];
+        let mapping_symbols = self
+            .disasm_modes
+            .get(&SectionIndex(section_index))
+            .map(|x| x.as_slice())
+            .unwrap_or(&fallback_mappings);
+        let first_mapping_idx =
+            match mapping_symbols.binary_search_by_key(&start_addr, |x| x.address) {
+                Ok(idx) => idx,
+                Err(idx) => idx - 1,
+            };
+        let first_mapping = mapping_symbols[first_mapping_idx].mapping;
+
+        let mut mappings_iter =
+            mapping_symbols.iter().skip(first_mapping_idx + 1).take_while(|x| x.address < end_addr);
+        let mut next_mapping = mappings_iter.next();
+
+        let ins_count = code.len() / first_mapping.instruction_size();
+        let mut ops = Vec::<u16>::with_capacity(ins_count);
+        let mut insts = Vec::<ObjIns>::with_capacity(ins_count);
+        let mut parser = Parser::new(ArmVersion::V5Te, first_mapping, start_addr, code);
+
+        while let Some((address, op, ins)) = parser.next() {
+            if let Some(next) = next_mapping {
+                let next_address = parser.address;
+                if next_address >= next.address {
+                    // Change mapping
+                    parser.mode = next.mapping;
+                    next_mapping = mappings_iter.next();
+                }
+            }
+
+            let line = line_info.range(..=address as u64).last().map(|(_, &b)| b);
+
+            let reloc = relocations.iter().find(|r| (r.address as u32 & !1) == address).cloned();
+
+            let mut reloc_arg = None;
+            if let Some(reloc) = &reloc {
+                match reloc.flags {
+                    RelocationFlags::Elf { r_type: elf::R_ARM_THM_XPC22 }
+                    | RelocationFlags::Elf { r_type: elf::R_ARM_PC24 } => {
+                        reloc_arg =
+                            ins.args.iter().rposition(|a| matches!(a, Argument::BranchDest(_)));
+                    }
+                    _ => (),
+                }
+            };
+
+            let (args, branch_dest) = if reloc.is_some() && parser.mode == ParseMode::Data {
+                (vec![ObjInsArg::Reloc], None)
+            } else {
+                push_args(&ins, config, reloc_arg, address)?
+            };
+
+            ops.push(op.id());
+            insts.push(ObjIns {
+                address: address as u64,
+                size: (parser.address - address) as u8,
+                op: op.id(),
+                mnemonic: ins.mnemonic.to_string(),
+                args,
+                reloc,
+                branch_dest,
+                line,
+                formatted: ins.to_string(),
+                orig: None,
+            });
+        }
+
+        Ok(ProcessCodeResult { ops, insts })
+    }
+
+    fn implcit_addend(
+        &self,
+        _section: &ObjSection,
+        address: u64,
+        reloc: &Relocation,
+    ) -> anyhow::Result<i64> {
+        bail!("Unsupported ARM implicit relocation {:#x}{:?}", address, reloc.flags())
+    }
+
+    fn demangle(&self, name: &str) -> Option<String> {
+        cpp_demangle::Symbol::new(name)
+            .ok()
+            .and_then(|s| s.demangle(&cpp_demangle::DemangleOptions::default()).ok())
+    }
+
+    fn display_reloc(&self, flags: RelocationFlags) -> Cow<'static, str> {
+        Cow::Owned(format!("<{flags:?}>"))
+    }
+}
+
+#[derive(Clone, Copy, Debug)]
+struct DisasmMode {
+    address: u32,
+    mapping: ParseMode,
+}
+
+impl DisasmMode {
+    fn from_symbol<'a>(sym: &Symbol<'a, '_, &'a [u8]>) -> Option<Self> {
+        if let Ok(name) = sym.name() {
+            ParseMode::from_mapping_symbol(name)
+                .map(|mapping| DisasmMode { address: sym.address() as u32, mapping })
+        } else {
+            None
+        }
+    }
+}
+
+fn push_args(
+    parsed_ins: &ParsedIns,
+    config: &DiffObjConfig,
+    reloc_arg: Option<usize>,
+    cur_addr: u32,
+) -> Result<(Vec<ObjInsArg>, Option<u64>)> {
+    let mut args = vec![];
+    let mut branch_dest = None;
+    let mut writeback = false;
+    let mut deref = false;
+    for (i, arg) in parsed_ins.args_iter().enumerate() {
+        // Emit punctuation before separator
+        if deref {
+            match arg {
+                Argument::OffsetImm(OffsetImm { post_indexed: true, value: _ })
+                | Argument::OffsetReg(OffsetReg { add: _, post_indexed: true, reg: _ })
+                | Argument::CoOption(_) => {
+                    deref = false;
+                    args.push(ObjInsArg::PlainText("]".into()));
+                    if writeback {
+                        writeback = false;
+                        args.push(ObjInsArg::Arg(ObjInsArgValue::Opaque("!".into())));
+                    }
+                }
+                _ => {}
+            }
+        }
+
+        if i > 0 {
+            args.push(ObjInsArg::PlainText(config.separator().into()));
+        }
+
+        if reloc_arg == Some(i) {
+            args.push(ObjInsArg::Reloc);
+        } else {
+            match arg {
+                Argument::Reg(reg) => {
+                    if reg.deref {
+                        deref = true;
+                        args.push(ObjInsArg::PlainText("[".into()));
+                    }
+                    args.push(ObjInsArg::Arg(ObjInsArgValue::Opaque(reg.reg.to_string().into())));
+                    if reg.writeback {
+                        if reg.deref {
+                            writeback = true;
+                        } else {
+                            args.push(ObjInsArg::Arg(ObjInsArgValue::Opaque("!".into())));
+                        }
+                    }
+                }
+                Argument::RegList(reg_list) => {
+                    args.push(ObjInsArg::PlainText("{".into()));
+                    let mut first = true;
+                    for i in 0..16 {
+                        if (reg_list.regs & (1 << i)) != 0 {
+                            if !first {
+                                args.push(ObjInsArg::PlainText(config.separator().into()));
+                            }
+                            args.push(ObjInsArg::Arg(ObjInsArgValue::Opaque(
+                                Register::parse(i).to_string().into(),
+                            )));
+                            first = false;
+                        }
+                    }
+                    args.push(ObjInsArg::PlainText("}".into()));
+                    if reg_list.user_mode {
+                        args.push(ObjInsArg::Arg(ObjInsArgValue::Opaque("^".to_string().into())));
+                    }
+                }
+                Argument::UImm(value) | Argument::CoOpcode(value) => {
+                    args.push(ObjInsArg::PlainText("#".into()));
+                    args.push(ObjInsArg::Arg(ObjInsArgValue::Unsigned(*value as u64)));
+                }
+                Argument::SImm(value)
+                | Argument::OffsetImm(OffsetImm { post_indexed: _, value }) => {
+                    args.push(ObjInsArg::PlainText("#".into()));
+                    args.push(ObjInsArg::Arg(ObjInsArgValue::Signed(*value as i64)));
+                }
+                Argument::BranchDest(value) => {
+                    let dest = cur_addr.wrapping_add_signed(*value) as u64;
+                    args.push(ObjInsArg::BranchDest(dest));
+                    branch_dest = Some(dest);
+                }
+                Argument::CoOption(value) => {
+                    args.push(ObjInsArg::PlainText("{".into()));
+                    args.push(ObjInsArg::Arg(ObjInsArgValue::Unsigned(*value as u64)));
+                    args.push(ObjInsArg::PlainText("}".into()));
+                }
+                Argument::CoprocNum(value) => {
+                    args.push(ObjInsArg::Arg(ObjInsArgValue::Opaque(format!("p{}", value).into())));
+                }
+                Argument::ShiftImm(shift) => {
+                    args.push(ObjInsArg::Arg(ObjInsArgValue::Opaque(shift.op.to_string().into())));
+                    args.push(ObjInsArg::PlainText(" #".into()));
+                    args.push(ObjInsArg::Arg(ObjInsArgValue::Unsigned(shift.imm as u64)));
+                }
+                Argument::ShiftReg(shift) => {
+                    args.push(ObjInsArg::Arg(ObjInsArgValue::Opaque(shift.op.to_string().into())));
+                    args.push(ObjInsArg::PlainText(" ".into()));
+                    args.push(ObjInsArg::Arg(ObjInsArgValue::Opaque(shift.reg.to_string().into())));
+                }
+                Argument::OffsetReg(offset) => {
+                    if !offset.add {
+                        args.push(ObjInsArg::PlainText("-".into()));
+                    }
+                    args.push(ObjInsArg::Arg(ObjInsArgValue::Opaque(
+                        offset.reg.to_string().into(),
+                    )));
+                }
+                _ => args.push(ObjInsArg::Arg(ObjInsArgValue::Opaque(arg.to_string().into()))),
+            }
+        }
+    }
+    if deref {
+        args.push(ObjInsArg::PlainText("]".into()));
+        if writeback {
+            args.push(ObjInsArg::Arg(ObjInsArgValue::Opaque("!".into())));
+        }
+    }
+    Ok((args, branch_dest))
+}

objdiff-core/src/arch/mips.rs

@@ -1,54 +1,95 @@
-use std::borrow::Cow;
+use std::{borrow::Cow, collections::BTreeMap, sync::Mutex};
 
 use anyhow::{anyhow, bail, Result};
-use object::{elf, Endian, Endianness, File, Object, Relocation, RelocationFlags};
+use object::{elf, Endian, Endianness, File, FileFlags, Object, Relocation, RelocationFlags};
 use rabbitizer::{config, Abi, InstrCategory, Instruction, OperandType};
 
 use crate::{
     arch::{ObjArch, ProcessCodeResult},
-    diff::DiffObjConfig,
-    obj::{ObjInfo, ObjIns, ObjInsArg, ObjInsArgValue, ObjReloc, ObjSection, SymbolRef},
+    diff::{DiffObjConfig, MipsAbi, MipsInstrCategory},
+    obj::{ObjIns, ObjInsArg, ObjInsArgValue, ObjReloc, ObjSection},
 };
 
-fn configure_rabbitizer() {
+static RABBITIZER_MUTEX: Mutex<()> = Mutex::new(());
+
+fn configure_rabbitizer(abi: Abi) {
     unsafe {
-        config::RabbitizerConfig_Cfg.reg_names.fpr_abi_names = Abi::O32;
+        config::RabbitizerConfig_Cfg.reg_names.fpr_abi_names = abi;
     }
 }
 
 pub struct ObjArchMips {
     pub endianness: Endianness,
+    pub abi: Abi,
+    pub instr_category: InstrCategory,
 }
 
+const EF_MIPS_ABI: u32 = 0x0000F000;
+const EF_MIPS_MACH: u32 = 0x00FF0000;
+
+const E_MIPS_MACH_ALLEGREX: u32 = 0x00840000;
+const E_MIPS_MACH_5900: u32 = 0x00920000;
+
 impl ObjArchMips {
     pub fn new(object: &File) -> Result<Self> {
-        configure_rabbitizer();
-        Ok(Self { endianness: object.endianness() })
+        let mut abi = Abi::NUMERIC;
+        let mut instr_category = InstrCategory::CPU;
+        match object.flags() {
+            FileFlags::None => {}
+            FileFlags::Elf { e_flags, .. } => {
+                abi = match e_flags & EF_MIPS_ABI {
+                    elf::EF_MIPS_ABI_O32 => Abi::O32,
+                    elf::EF_MIPS_ABI_EABI32 | elf::EF_MIPS_ABI_EABI64 => Abi::N32,
+                    _ => Abi::NUMERIC,
+                };
+                instr_category = match e_flags & EF_MIPS_MACH {
+                    E_MIPS_MACH_ALLEGREX => InstrCategory::R4000ALLEGREX,
+                    E_MIPS_MACH_5900 => InstrCategory::R5900,
+                    _ => InstrCategory::CPU,
+                };
+            }
+            _ => bail!("Unsupported MIPS file flags"),
+        }
+        Ok(Self { endianness: object.endianness(), abi, instr_category })
     }
 }
 
 impl ObjArch for ObjArchMips {
     fn process_code(
         &self,
-        obj: &ObjInfo,
-        symbol_ref: SymbolRef,
+        address: u64,
+        code: &[u8],
+        _section_index: usize,
+        relocations: &[ObjReloc],
+        line_info: &BTreeMap<u64, u64>,
         config: &DiffObjConfig,
     ) -> Result<ProcessCodeResult> {
-        let (section, symbol) = obj.section_symbol(symbol_ref);
-        let section = section.ok_or_else(|| anyhow!("Code symbol section not found"))?;
-        let code = &section.data
-            [symbol.section_address as usize..(symbol.section_address + symbol.size) as usize];
+        let _guard = RABBITIZER_MUTEX.lock().map_err(|e| anyhow!("Failed to lock mutex: {e}"))?;
+        configure_rabbitizer(match config.mips_abi {
+            MipsAbi::Auto => self.abi,
+            MipsAbi::O32 => Abi::O32,
+            MipsAbi::N32 => Abi::N32,
+            MipsAbi::N64 => Abi::N64,
+        });
+        let instr_category = match config.mips_instr_category {
+            MipsInstrCategory::Auto => self.instr_category,
+            MipsInstrCategory::Cpu => InstrCategory::CPU,
+            MipsInstrCategory::Rsp => InstrCategory::RSP,
+            MipsInstrCategory::R3000Gte => InstrCategory::R3000GTE,
+            MipsInstrCategory::R4000Allegrex => InstrCategory::R4000ALLEGREX,
+            MipsInstrCategory::R5900 => InstrCategory::R5900,
+        };
 
-        let start_address = symbol.address;
-        let end_address = symbol.address + symbol.size;
+        let start_address = address;
+        let end_address = address + code.len() as u64;
         let ins_count = code.len() / 4;
         let mut ops = Vec::<u16>::with_capacity(ins_count);
         let mut insts = Vec::<ObjIns>::with_capacity(ins_count);
         let mut cur_addr = start_address as u32;
         for chunk in code.chunks_exact(4) {
-            let reloc = section.relocations.iter().find(|r| (r.address as u32 & !3) == cur_addr);
+            let reloc = relocations.iter().find(|r| (r.address as u32 & !3) == cur_addr);
             let code = self.endianness.read_u32_bytes(chunk.try_into()?);
-            let instruction = Instruction::new(code, cur_addr, InstrCategory::CPU);
+            let instruction = Instruction::new(code, cur_addr, instr_category);
 
             let formatted = instruction.disassemble(None, 0);
             let op = instruction.unique_id as u16;
@@ -112,7 +153,7 @@ impl ObjArch for ObjArchMips {
                     }
                 }
             }
-            let line = section.line_info.range(..=cur_addr as u64).last().map(|(_, &b)| b);
+            let line = line_info.range(..=cur_addr as u64).last().map(|(_, &b)| b);
             insts.push(ObjIns {
                 address: cur_addr as u64,
                 size: 4,

objdiff-core/src/arch/mod.rs

@@ -1,25 +1,30 @@
-use std::borrow::Cow;
+use std::{borrow::Cow, collections::BTreeMap};
 
 use anyhow::{bail, Result};
 use object::{Architecture, Object, Relocation, RelocationFlags};
 
 use crate::{
     diff::DiffObjConfig,
-    obj::{ObjInfo, ObjIns, ObjSection, SymbolRef},
+    obj::{ObjIns, ObjReloc, ObjSection},
 };
 
+#[cfg(feature = "arm")]
+mod arm;
 #[cfg(feature = "mips")]
-mod mips;
+pub mod mips;
 #[cfg(feature = "ppc")]
-mod ppc;
+pub mod ppc;
 #[cfg(feature = "x86")]
-mod x86;
+pub mod x86;
 
 pub trait ObjArch: Send + Sync {
     fn process_code(
         &self,
-        obj: &ObjInfo,
-        symbol_ref: SymbolRef,
+        address: u64,
+        code: &[u8],
+        section_index: usize,
+        relocations: &[ObjReloc],
+        line_info: &BTreeMap<u64, u64>,
         config: &DiffObjConfig,
     ) -> Result<ProcessCodeResult>;
 
@@ -44,6 +49,8 @@ pub fn new_arch(object: &object::File) -> Result<Box<dyn ObjArch>> {
         Architecture::Mips => Box::new(mips::ObjArchMips::new(object)?),
         #[cfg(feature = "x86")]
         Architecture::I386 | Architecture::X86_64 => Box::new(x86::ObjArchX86::new(object)?),
+        #[cfg(feature = "arm")]
+        Architecture::Arm => Box::new(arm::ObjArchArm::new(object)?),
         arch => bail!("Unsupported architecture: {arch:?}"),
     })
 }

objdiff-core/src/arch/ppc.rs

@@ -1,13 +1,13 @@
-use std::borrow::Cow;
+use std::{borrow::Cow, collections::BTreeMap};
 
-use anyhow::{anyhow, bail, Result};
+use anyhow::{bail, Result};
 use object::{elf, File, Relocation, RelocationFlags};
 use ppc750cl::{Argument, InsIter, GPR};
 
 use crate::{
     arch::{ObjArch, ProcessCodeResult},
     diff::DiffObjConfig,
-    obj::{ObjInfo, ObjIns, ObjInsArg, ObjInsArgValue, ObjReloc, ObjSection, SymbolRef},
+    obj::{ObjIns, ObjInsArg, ObjInsArgValue, ObjReloc, ObjSection},
 };
 
 // Relative relocation, can be Simm, Offset or BranchDest
@@ -31,20 +31,18 @@ impl ObjArchPpc {
 impl ObjArch for ObjArchPpc {
     fn process_code(
         &self,
-        obj: &ObjInfo,
-        symbol_ref: SymbolRef,
+        address: u64,
+        code: &[u8],
+        _section_index: usize,
+        relocations: &[ObjReloc],
+        line_info: &BTreeMap<u64, u64>,
         config: &DiffObjConfig,
     ) -> Result<ProcessCodeResult> {
-        let (section, symbol) = obj.section_symbol(symbol_ref);
-        let section = section.ok_or_else(|| anyhow!("Code symbol section not found"))?;
-        let code = &section.data
-            [symbol.section_address as usize..(symbol.section_address + symbol.size) as usize];
-
         let ins_count = code.len() / 4;
         let mut ops = Vec::<u16>::with_capacity(ins_count);
         let mut insts = Vec::<ObjIns>::with_capacity(ins_count);
-        for (cur_addr, mut ins) in InsIter::new(code, symbol.address as u32) {
-            let reloc = section.relocations.iter().find(|r| (r.address as u32 & !3) == cur_addr);
+        for (cur_addr, mut ins) in InsIter::new(code, address as u32) {
+            let reloc = relocations.iter().find(|r| (r.address as u32 & !3) == cur_addr);
             if let Some(reloc) = reloc {
                 // Zero out relocations
                 ins.code = match reloc.flags {
@@ -133,7 +131,7 @@ impl ObjArch for ObjArchPpc {
             }
 
             ops.push(ins.op as u16);
-            let line = section.line_info.range(..=cur_addr as u64).last().map(|(_, &b)| b);
+            let line = line_info.range(..=cur_addr as u64).last().map(|(_, &b)| b);
             insts.push(ObjIns {
                 address: cur_addr as u64,
                 size: 4,

objdiff-core/src/arch/x86.rs

@@ -1,4 +1,4 @@
-use std::borrow::Cow;
+use std::{borrow::Cow, collections::BTreeMap};
 
 use anyhow::{anyhow, bail, ensure, Result};
 use iced_x86::{
@@ -11,7 +11,7 @@ use object::{pe, Endian, Endianness, File, Object, Relocation, RelocationFlags};
 use crate::{
     arch::{ObjArch, ProcessCodeResult},
     diff::{DiffObjConfig, X86Formatter},
-    obj::{ObjInfo, ObjIns, ObjInsArg, ObjInsArgValue, ObjSection, SymbolRef},
+    obj::{ObjIns, ObjInsArg, ObjInsArgValue, ObjReloc, ObjSection},
 };
 
 pub struct ObjArchX86 {
@@ -28,17 +28,15 @@ impl ObjArchX86 {
 impl ObjArch for ObjArchX86 {
     fn process_code(
         &self,
-        obj: &ObjInfo,
-        symbol_ref: SymbolRef,
+        address: u64,
+        code: &[u8],
+        _section_index: usize,
+        relocations: &[ObjReloc],
+        line_info: &BTreeMap<u64, u64>,
         config: &DiffObjConfig,
     ) -> Result<ProcessCodeResult> {
-        let (section, symbol) = obj.section_symbol(symbol_ref);
-        let section = section.ok_or_else(|| anyhow!("Code symbol section not found"))?;
-        let code = &section.data
-            [symbol.section_address as usize..(symbol.section_address + symbol.size) as usize];
-
         let mut result = ProcessCodeResult { ops: Vec::new(), insts: Vec::new() };
-        let mut decoder = Decoder::with_ip(self.bits, code, symbol.address, DecoderOptions::NONE);
+        let mut decoder = Decoder::with_ip(self.bits, code, address, DecoderOptions::NONE);
         let mut formatter: Box<dyn Formatter> = match config.x86_formatter {
             X86Formatter::Intel => Box::new(IntelFormatter::new()),
             X86Formatter::Gas => Box::new(GasFormatter::new()),
@@ -70,11 +68,10 @@ impl ObjArch for ObjArchX86 {
 
             let address = instruction.ip();
             let op = instruction.mnemonic() as u16;
-            let reloc = section
-                .relocations
+            let reloc = relocations
                 .iter()
                 .find(|r| r.address >= address && r.address < address + instruction.len() as u64);
-            let line = section.line_info.range(..=address).last().map(|(_, &b)| b);
+            let line = line_info.range(..=address).last().map(|(_, &b)| b);
             output.ins = ObjIns {
                 address,
                 size: instruction.len() as u8,

objdiff-core/src/diff/code.rs

@@ -4,7 +4,7 @@ use std::{
     time::{Duration, Instant},
 };
 
-use anyhow::Result;
+use anyhow::{anyhow, Result};
 use similar::{capture_diff_slices_deadline, Algorithm};
 
 use crate::{
@@ -16,34 +16,48 @@ use crate::{
     obj::{ObjInfo, ObjInsArg, ObjReloc, ObjSymbol, ObjSymbolFlags, SymbolRef},
 };
 
-pub fn no_diff_code(
+pub fn process_code_symbol(
     obj: &ObjInfo,
     symbol_ref: SymbolRef,
     config: &DiffObjConfig,
-) -> Result<ObjSymbolDiff> {
-    let out = obj.arch.process_code(obj, symbol_ref, config)?;
+) -> Result<ProcessCodeResult> {
+    let (section, symbol) = obj.section_symbol(symbol_ref);
+    let section = section.ok_or_else(|| anyhow!("Code symbol section not found"))?;
+    let code = &section.data
+        [symbol.section_address as usize..(symbol.section_address + symbol.size) as usize];
+    obj.arch.process_code(
+        symbol.address,
+        code,
+        section.orig_index,
+        &section.relocations,
+        &section.line_info,
+        config,
+    )
+}
 
+pub fn no_diff_code(out: &ProcessCodeResult, symbol_ref: SymbolRef) -> Result<ObjSymbolDiff> {
     let mut diff = Vec::<ObjInsDiff>::new();
-    for i in out.insts {
-        diff.push(ObjInsDiff { ins: Some(i), kind: ObjInsDiffKind::None, ..Default::default() });
+    for i in &out.insts {
+        diff.push(ObjInsDiff {
+            ins: Some(i.clone()),
+            kind: ObjInsDiffKind::None,
+            ..Default::default()
+        });
     }
     resolve_branches(&mut diff);
     Ok(ObjSymbolDiff { symbol_ref, diff_symbol: None, instructions: diff, match_percent: None })
 }
 
 pub fn diff_code(
-    left_obj: &ObjInfo,
-    right_obj: &ObjInfo,
+    left_out: &ProcessCodeResult,
+    right_out: &ProcessCodeResult,
     left_symbol_ref: SymbolRef,
     right_symbol_ref: SymbolRef,
     config: &DiffObjConfig,
 ) -> Result<(ObjSymbolDiff, ObjSymbolDiff)> {
-    let left_out = left_obj.arch.process_code(left_obj, left_symbol_ref, config)?;
-    let right_out = right_obj.arch.process_code(right_obj, right_symbol_ref, config)?;
-
     let mut left_diff = Vec::<ObjInsDiff>::new();
     let mut right_diff = Vec::<ObjInsDiff>::new();
-    diff_instructions(&mut left_diff, &mut right_diff, &left_out, &right_out)?;
+    diff_instructions(&mut left_diff, &mut right_diff, left_out, right_out)?;
 
     resolve_branches(&mut left_diff);
     resolve_branches(&mut right_diff);

objdiff-core/src/diff/mod.rs

@@ -4,7 +4,7 @@ use anyhow::Result;
 
 use crate::{
     diff::{
-        code::{diff_code, no_diff_code},
+        code::{diff_code, no_diff_code, process_code_symbol},
         data::{
             diff_bss_section, diff_bss_symbol, diff_data_section, diff_data_symbol,
             diff_text_section, no_diff_symbol,
@@ -13,29 +13,112 @@ use crate::{
     obj::{ObjInfo, ObjIns, ObjSection, ObjSectionKind, ObjSymbol, SymbolRef},
 };
 
-mod code;
-mod data;
+pub mod code;
+pub mod data;
 pub mod display;
 
-#[derive(Debug, Copy, Clone, Default, Eq, PartialEq, serde::Deserialize, serde::Serialize)]
+#[derive(
+    Debug,
+    Copy,
+    Clone,
+    Default,
+    Eq,
+    PartialEq,
+    serde::Deserialize,
+    serde::Serialize,
+    strum::VariantArray,
+    strum::EnumMessage,
+)]
 pub enum X86Formatter {
     #[default]
+    #[strum(message = "Intel (default)")]
     Intel,
+    #[strum(message = "AT&T")]
     Gas,
+    #[strum(message = "NASM")]
     Nasm,
+    #[strum(message = "MASM")]
     Masm,
 }
 
+#[derive(
+    Debug,
+    Copy,
+    Clone,
+    Default,
+    Eq,
+    PartialEq,
+    serde::Deserialize,
+    serde::Serialize,
+    strum::VariantArray,
+    strum::EnumMessage,
+)]
+pub enum MipsAbi {
+    #[default]
+    #[strum(message = "Auto (default)")]
+    Auto,
+    #[strum(message = "O32")]
+    O32,
+    #[strum(message = "N32")]
+    N32,
+    #[strum(message = "N64")]
+    N64,
+}
+
+#[derive(
+    Debug,
+    Copy,
+    Clone,
+    Default,
+    Eq,
+    PartialEq,
+    serde::Deserialize,
+    serde::Serialize,
+    strum::VariantArray,
+    strum::EnumMessage,
+)]
+pub enum MipsInstrCategory {
+    #[default]
+    #[strum(message = "Auto (default)")]
+    Auto,
+    #[strum(message = "CPU")]
+    Cpu,
+    #[strum(message = "RSP (N64)")]
+    Rsp,
+    #[strum(message = "R3000 GTE (PS1)")]
+    R3000Gte,
+    #[strum(message = "R4000 ALLEGREX (PSP)")]
+    R4000Allegrex,
+    #[strum(message = "R5900 EE (PS2)")]
+    R5900,
+}
+
 #[inline]
 const fn default_true() -> bool { true }
 
-#[derive(Debug, Clone, Default, Eq, PartialEq, serde::Deserialize, serde::Serialize)]
+#[derive(Debug, Clone, Eq, PartialEq, serde::Deserialize, serde::Serialize)]
 #[serde(default)]
 pub struct DiffObjConfig {
     pub relax_reloc_diffs: bool,
     #[serde(default = "default_true")]
     pub space_between_args: bool,
+    // x86
     pub x86_formatter: X86Formatter,
+    // MIPS
+    pub mips_abi: MipsAbi,
+    pub mips_instr_category: MipsInstrCategory,
+}
+
+impl Default for DiffObjConfig {
+    fn default() -> Self {
+        Self {
+            relax_reloc_diffs: false,
+            space_between_args: true,
+            x86_formatter: Default::default(),
+            mips_abi: Default::default(),
+            mips_instr_category: Default::default(),
+        }
+    }
 }
 
 impl DiffObjConfig {
@@ -238,9 +321,11 @@ pub fn diff_objs(
                 let (right_obj, right_out) = right.as_mut().unwrap();
                 match section_kind {
                     ObjSectionKind::Code => {
+                        let left_code = process_code_symbol(left_obj, left_symbol_ref, config)?;
+                        let right_code = process_code_symbol(right_obj, right_symbol_ref, config)?;
                         let (left_diff, right_diff) = diff_code(
-                            left_obj,
-                            right_obj,
+                            &left_code,
+                            &right_code,
                             left_symbol_ref,
                             right_symbol_ref,
                             config,
@@ -250,9 +335,10 @@ pub fn diff_objs(
 
                         if let Some(prev_symbol_ref) = prev_symbol_ref {
                             let (prev_obj, prev_out) = prev.as_mut().unwrap();
+                            let prev_code = process_code_symbol(prev_obj, prev_symbol_ref, config)?;
                             let (_, prev_diff) = diff_code(
-                                right_obj,
-                                prev_obj,
+                                &right_code,
+                                &prev_code,
                                 right_symbol_ref,
                                 prev_symbol_ref,
                                 config,
@@ -286,8 +372,9 @@ pub fn diff_objs(
                 let (left_obj, left_out) = left.as_mut().unwrap();
                 match section_kind {
                     ObjSectionKind::Code => {
+                        let code = process_code_symbol(left_obj, left_symbol_ref, config)?;
                         *left_out.symbol_diff_mut(left_symbol_ref) =
-                            no_diff_code(left_obj, left_symbol_ref, config)?;
+                            no_diff_code(&code, left_symbol_ref)?;
                     }
                     ObjSectionKind::Data | ObjSectionKind::Bss => {
                         *left_out.symbol_diff_mut(left_symbol_ref) =
@@ -299,8 +386,9 @@ pub fn diff_objs(
                 let (right_obj, right_out) = right.as_mut().unwrap();
                 match section_kind {
                     ObjSectionKind::Code => {
+                        let code = process_code_symbol(right_obj, right_symbol_ref, config)?;
                         *right_out.symbol_diff_mut(right_symbol_ref) =
-                            no_diff_code(right_obj, right_symbol_ref, config)?;
+                            no_diff_code(&code, right_symbol_ref)?;
                     }
                     ObjSectionKind::Data | ObjSectionKind::Bss => {
                         *right_out.symbol_diff_mut(right_symbol_ref) =

objdiff-core/src/obj/read.rs

@@ -328,15 +328,10 @@ fn line_info(obj_file: &File<'_>, sections: &mut [ObjSection]) -> Result<()> {
             if let Some(program) = unit.line_program.clone() {
                 let mut text_sections =
                     obj_file.sections().filter(|s| s.kind() == SectionKind::Text);
-                let section_index = text_sections
-                    .next()
-                    .ok_or_else(|| anyhow!("Next text section not found for line info"))?
-                    .index()
-                    .0;
-                let mut lines = sections
-                    .iter_mut()
-                    .find(|s| s.orig_index == section_index)
-                    .map(|s| &mut s.line_info);
+                let section_index = text_sections.next().map(|s| s.index().0);
+                let mut lines = section_index.map(|index| {
+                    &mut sections.iter_mut().find(|s| s.orig_index == index).unwrap().line_info
+                });
 
                 let mut rows = program.rows();
                 while let Some((_header, row)) = rows.next_row()? {

objdiff-gui/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "objdiff-gui"
-version = "2.0.0-alpha.1"
+version = "2.0.0-alpha.3"
 edition = "2021"
 rust-version = "1.70"
 authors = ["Luke Street <luke@street.dev>"]
@@ -46,6 +46,7 @@ ron = "0.8.1"
 serde = { version = "1", features = ["derive"] }
 serde_json = "1.0.116"
 shell-escape = "0.1.5"
+strum = { version = "0.26.2", features = ["derive"] }
 tempfile = "3.10.1"
 time = { version = "0.3.36", features = ["formatting", "local-offset"] }
 

objdiff-gui/src/app.rs

@@ -30,7 +30,7 @@ use crate::{
     views::{
         appearance::{appearance_window, Appearance},
         config::{
-            config_ui, diff_config_window, project_window, ConfigViewState, CONFIG_DISABLED_TEXT,
+            arch_config_window, config_ui, project_window, ConfigViewState, CONFIG_DISABLED_TEXT,
         },
         data_diff::data_diff_ui,
         debug::debug_window,
@@ -52,7 +52,7 @@ pub struct ViewState {
     pub show_appearance_config: bool,
     pub show_demangle: bool,
     pub show_project_config: bool,
-    pub show_diff_config: bool,
+    pub show_arch_config: bool,
     pub show_debug: bool,
 }
 
@@ -414,7 +414,7 @@ impl eframe::App for App {
             show_appearance_config,
             show_demangle,
             show_project_config,
-            show_diff_config,
+            show_arch_config,
             show_debug,
         } = view_state;
 
@@ -468,8 +468,8 @@ impl eframe::App for App {
                     }
                 });
                 ui.menu_button("Diff Options", |ui| {
-                    if ui.button("More…").clicked() {
-                        *show_diff_config = !*show_diff_config;
+                    if ui.button("Arch Settings…").clicked() {
+                        *show_arch_config = !*show_arch_config;
                         ui.close_menu();
                     }
                     let mut config = config.write().unwrap();
@@ -541,7 +541,7 @@ impl eframe::App for App {
         project_window(ctx, config, show_project_config, config_state, appearance);
         appearance_window(ctx, show_appearance_config, appearance);
         demangle_window(ctx, show_demangle, demangle_state, appearance);
-        diff_config_window(ctx, config, show_diff_config, appearance);
+        arch_config_window(ctx, config, show_arch_config, appearance);
         debug_window(ctx, show_debug, frame_history, appearance);
 
         self.post_update(ctx);

objdiff-gui/src/jobs/objdiff.rs

@@ -143,7 +143,7 @@ fn run_make_cmd(config: &BuildConfig, cwd: &Path, arg: &Path) -> Result<BuildSta
         cmdline.push(' ');
         cmdline.push_str(shell_escape::escape(arg.to_string_lossy()).as_ref());
     }
-    let output = command.output().context("Failed to execute build")?;
+    let output = command.output().map_err(|e| anyhow!("Failed to execute build: {e}"))?;
     let stdout = from_utf8(&output.stdout).context("Failed to process stdout")?;
     let stderr = from_utf8(&output.stderr).context("Failed to process stderr")?;
     Ok(BuildStatus {

objdiff-gui/src/views/appearance.rs

@@ -119,6 +119,8 @@ impl Appearance {
                 self.delete_color = Color32::from_rgb(200, 40, 41);
             }
         }
+        style.spacing.scroll = egui::style::ScrollStyle::solid();
+        style.spacing.scroll.bar_width = 10.0;
         ctx.set_style(style);
     }
 

objdiff-gui/src/views/config.rs

@@ -16,9 +16,10 @@ use egui::{
 use globset::Glob;
 use objdiff_core::{
     config::{ProjectObject, DEFAULT_WATCH_PATTERNS},
-    diff::X86Formatter,
+    diff::{MipsAbi, MipsInstrCategory, X86Formatter},
 };
 use self_update::cargo_crate_version;
+use strum::{EnumMessage, VariantArray};
 
 use crate::{
     app::{AppConfig, AppConfigRef, ObjectConfig},
@@ -842,29 +843,28 @@ fn split_obj_config_ui(
     });
 }
 
-pub fn diff_config_window(
+pub fn arch_config_window(
     ctx: &egui::Context,
     config: &AppConfigRef,
     show: &mut bool,
     appearance: &Appearance,
 ) {
     let mut config_guard = config.write().unwrap();
-    egui::Window::new("Diff Config").open(show).show(ctx, |ui| {
-        diff_config_ui(ui, &mut config_guard, appearance);
+    egui::Window::new("Arch Settings").open(show).show(ctx, |ui| {
+        arch_config_ui(ui, &mut config_guard, appearance);
     });
 }
 
-fn diff_config_ui(ui: &mut egui::Ui, config: &mut AppConfig, _appearance: &Appearance) {
-    egui::ComboBox::new("x86_formatter", "X86 Format")
-        .selected_text(format!("{:?}", config.diff_obj_config.x86_formatter))
+fn arch_config_ui(ui: &mut egui::Ui, config: &mut AppConfig, _appearance: &Appearance) {
+    ui.heading("x86");
+    egui::ComboBox::new("x86_formatter", "Format")
+        .selected_text(config.diff_obj_config.x86_formatter.get_message().unwrap())
         .show_ui(ui, |ui| {
-            for &formatter in
-                &[X86Formatter::Intel, X86Formatter::Gas, X86Formatter::Nasm, X86Formatter::Masm]
-            {
+            for &formatter in X86Formatter::VARIANTS {
                 if ui
                     .selectable_label(
                         config.diff_obj_config.x86_formatter == formatter,
-                        format!("{:?}", formatter),
+                        formatter.get_message().unwrap(),
                     )
                     .clicked()
                 {
@@ -873,4 +873,38 @@ fn diff_config_ui(ui: &mut egui::Ui, config: &mut AppConfig, _appearance: &Appea
                 }
             }
         });
+    ui.separator();
+    ui.heading("MIPS");
+    egui::ComboBox::new("mips_abi", "ABI")
+        .selected_text(config.diff_obj_config.mips_abi.get_message().unwrap())
+        .show_ui(ui, |ui| {
+            for &abi in MipsAbi::VARIANTS {
+                if ui
+                    .selectable_label(
+                        config.diff_obj_config.mips_abi == abi,
+                        abi.get_message().unwrap(),
+                    )
+                    .clicked()
+                {
+                    config.diff_obj_config.mips_abi = abi;
+                    config.queue_reload = true;
+                }
+            }
+        });
+    egui::ComboBox::new("mips_instr_category", "Instruction Category")
+        .selected_text(config.diff_obj_config.mips_instr_category.get_message().unwrap())
+        .show_ui(ui, |ui| {
+            for &category in MipsInstrCategory::VARIANTS {
+                if ui
+                    .selectable_label(
+                        config.diff_obj_config.mips_instr_category == category,
+                        category.get_message().unwrap(),
+                    )
+                    .clicked()
+                {
+                    config.diff_obj_config.mips_instr_category = category;
+                    config.queue_reload = true;
+                }
+            }
+        });
 }