Version v3.0.0-beta.4

arm: Fix thumb & data decoding, add tests
Add pre-commit hooks
2026-07-10 12:18:36 -07:00 · 2025-03-18 21:36:37 -06:00 · 2025-03-18 21:26:28 -06:00 · 2025-03-18 21:26:28 -06:00 · 2025-03-14 09:49:31 -06:00 · 2025-03-14 09:44:15 -06:00
38 changed files with 6520 additions and 169 deletions
@@ -0,0 +1,2 @@
+[*.md]
+trim_trailing_whitespace = false
@@ -4,8 +4,8 @@ on:
  pull_request:
  push:
    paths-ignore:
-      - '*.md'
-      - 'LICENSE*'
+      - "*.md"
+      - "LICENSE*"
  workflow_dispatch:

 env:
@@ -32,9 +32,9 @@ jobs:
      - name: Cache Rust workspace
        uses: Swatinem/rust-cache@v2
      - name: Cargo check
-        run: cargo check --features all
+        run: cargo check --all-targets --all-features
      - name: Cargo clippy
-        run: cargo clippy --features all
+        run: cargo clippy --all-targets --all-features

  fmt:
    name: Format
@@ -72,7 +72,7 @@ jobs:
    name: Test
    strategy:
      matrix:
-        platform: [ ubuntu-latest, windows-latest, macos-latest ]
+        platform: [ubuntu-latest, windows-latest, macos-latest]
      fail-fast: false
    runs-on: ${{ matrix.platform }}
    steps:
@@ -159,7 +159,7 @@ jobs:
          key: ${{ matrix.target }}
      - name: Cargo build
        run: >
-          cargo ${{ matrix.build }} --profile ${{ env.BUILD_PROFILE }} --target ${{ matrix.target }} 
+          cargo ${{ matrix.build }} --profile ${{ env.BUILD_PROFILE }} --target ${{ matrix.target }}
          --bin ${{ env.CARGO_BIN_NAME }} --features ${{ matrix.features }}
      - name: Upload artifacts
        uses: actions/upload-artifact@v4
@@ -226,7 +226,7 @@ jobs:
          key: ${{ matrix.target }}
      - name: Cargo build
        run: >
-          cargo ${{ matrix.build }} --profile ${{ env.BUILD_PROFILE }} --target ${{ matrix.target }} 
+          cargo ${{ matrix.build }} --profile ${{ env.BUILD_PROFILE }} --target ${{ matrix.target }}
          --bin ${{ env.CARGO_BIN_NAME }} --features ${{ matrix.features }}
      - name: Upload artifacts
        uses: actions/upload-artifact@v4
@@ -258,7 +258,7 @@ jobs:
    name: Release
    if: startsWith(github.ref, 'refs/tags/')
    runs-on: ubuntu-latest
-    needs: [ build-cli, build-gui ]
+    needs: [build-cli, build-gui]
    permissions:
      contents: write
    steps:
@@ -0,0 +1,36 @@
+# See https://pre-commit.com for more information
+# See https://pre-commit.com/hooks.html for more hooks
+repos:
+  - repo: https://github.com/pre-commit/pre-commit-hooks
+    rev: v5.0.0
+    hooks:
+      - id: trailing-whitespace
+        args: [--markdown-linebreak-ext=md]
+      - id: end-of-file-fixer
+      - id: fix-byte-order-marker
+      - id: check-yaml
+      - id: check-added-large-files
+  - repo: local
+    hooks:
+      - id: cargo-fmt
+        name: cargo fmt
+        description: Run cargo fmt on all project files.
+        language: system
+        entry: cargo
+        args: ["+nightly", "fmt", "--all"]
+        pass_filenames: false
+      - id: cargo clippy
+        name: cargo clippy
+        description: Run cargo clippy on all project files.
+        language: system
+        entry: cargo
+        args: ["+nightly", "clippy", "--all-targets", "--all-features"]
+        pass_filenames: false
+      - id: cargo-deny
+        name: cargo deny
+        description: Run cargo deny on all project files.
+        language: system
+        entry: cargo
+        args: ["deny", "check"]
+        pass_filenames: false
+        always_run: true
@@ -450,6 +450,18 @@ version = "1.1.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "1505bd5d3d116872e7271a6d4e16d81d0c8570876c8de68093a09ac269d8aac0"

+[[package]]
+name = "auditable-serde"
+version = "0.8.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5c7bf8143dfc3c0258df908843e169b5cc5fcf76c7718bd66135ef4a9cd558c5"
+dependencies = [
+ "semver",
+ "serde",
+ "serde_json",
+ "topological-sort",
+]
+
 [[package]]
 name = "autocfg"
 version = "1.4.0"
@@ -2652,7 +2664,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "fc2f4eb4bc735547cfed7c0a4922cbd04a4655978c09b54f1f7b228750664c34"
 dependencies = [
 "cfg-if",
- "windows-targets 0.48.5",
+ "windows-targets 0.52.6",
 ]

 [[package]]
@@ -3268,7 +3280,7 @@ dependencies = [

 [[package]]
 name = "objdiff-cli"
-version = "3.0.0-beta.2"
+version = "3.0.0-beta.4"
 dependencies = [
 "anyhow",
 "argp",
@@ -3291,7 +3303,7 @@ dependencies = [

 [[package]]
 name = "objdiff-core"
-version = "3.0.0-beta.2"
+version = "3.0.0-beta.4"
 dependencies = [
 "anyhow",
 "arm-attr",
@@ -3344,7 +3356,7 @@ dependencies = [

 [[package]]
 name = "objdiff-gui"
-version = "3.0.0-beta.2"
+version = "3.0.0-beta.4"
 dependencies = [
 "anyhow",
 "cfg-if",
@@ -3380,7 +3392,7 @@ dependencies = [

 [[package]]
 name = "objdiff-wasm"
-version = "3.0.0-beta.2"
+version = "3.0.0-beta.4"
 dependencies = [
 "log",
 "objdiff-core",
@@ -4474,6 +4486,9 @@ name = "semver"
 version = "1.0.25"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "f79dfe2d285b0488816f30e700a7438c5a73d816b5b7d3ac72fbc48b0d185e03"
+dependencies = [
+ "serde",
+]

 [[package]]
 name = "serde"
@@ -5165,6 +5180,12 @@ dependencies = [
 "winnow 0.7.3",
 ]

+[[package]]
+name = "topological-sort"
+version = "0.2.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ea68304e134ecd095ac6c3574494fc62b909f416c4fca77e440530221e549d3d"
+
 [[package]]
 name = "tower"
 version = "0.5.2"
@@ -5537,9 +5558,9 @@ dependencies = [

 [[package]]
 name = "wasm-encoder"
-version = "0.225.0"
+version = "0.227.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "6f7eac0445cac73bcf09e6a97f83248d64356dccf9f2b100199769b6b42464e5"
+checksum = "80bb72f02e7fbf07183443b27b0f3d4144abf8c114189f2e088ed95b696a7822"
 dependencies = [
 "leb128fmt",
 "wasmparser",
@@ -5547,11 +5568,13 @@ dependencies = [

 [[package]]
 name = "wasm-metadata"
-version = "0.225.0"
+version = "0.227.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f1d20d0bf2c73c32a5114cf35a5c10ccf9f9aa37a3a2c0114b3e11cbf6faac12"
+checksum = "ce1ef0faabbbba6674e97a56bee857ccddf942785a336c8b47b42373c922a91d"
 dependencies = [
 "anyhow",
+ "auditable-serde",
+ "flate2",
 "indexmap",
 "serde",
 "serde_derive",
@@ -5577,9 +5600,9 @@ dependencies = [

 [[package]]
 name = "wasmparser"
-version = "0.225.0"
+version = "0.227.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "36e5456165f81e64cb9908a0fe9b9d852c2c74582aa3fe2be3c2da57f937d3ae"
+checksum = "0f51cad774fb3c9461ab9bccc9c62dfb7388397b5deda31bf40e8108ccd678b2"
 dependencies = [
 "bitflags 2.9.0",
 "hashbrown",
@@ -6300,19 +6323,19 @@ dependencies = [

 [[package]]
 name = "wit-bindgen"
-version = "0.39.0"
+version = "0.40.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e4dd9a372b25d6f35456b0a730d2adabeb0c4878066ba8f8089800349be6ecb5"
+checksum = "8e7091ed6c9abfa4e0a2ef3b39d0539da992d841fcf32c255f64fb98764ffee5"
 dependencies = [
- "wit-bindgen-rt 0.39.0",
+ "wit-bindgen-rt 0.40.0",
 "wit-bindgen-rust-macro",
 ]

 [[package]]
 name = "wit-bindgen-core"
-version = "0.39.0"
+version = "0.40.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f108fa9b77a346372858b30c11ea903680e7e2b9d820b1a5883e9d530bf51c7e"
+checksum = "398c650cec1278cfb72e214ba26ef3440ab726e66401bd39c04f465ee3979e6b"
 dependencies = [
 "anyhow",
 "heck",
@@ -6330,18 +6353,18 @@ dependencies = [

 [[package]]
 name = "wit-bindgen-rt"
-version = "0.39.0"
+version = "0.40.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "6f42320e61fe2cfd34354ecb597f86f413484a798ba44a8ca1165c58d42da6c1"
+checksum = "68faed92ae696b93ea9a7b67ba6c37bf09d72c6d9a70fa824a743c3020212f11"
 dependencies = [
 "bitflags 2.9.0",
 ]

 [[package]]
 name = "wit-bindgen-rust"
-version = "0.39.0"
+version = "0.40.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e5ba5b852e976d35dbf6cb745746bf1bd4fc26782bab1e0c615fc71a7d8aac05"
+checksum = "83903c8dcd8084a8a67ae08190122cf0e25dc37bdc239070a00f47e22d3f0aae"
 dependencies = [
 "anyhow",
 "heck",
@@ -6355,9 +6378,9 @@ dependencies = [

 [[package]]
 name = "wit-bindgen-rust-macro"
-version = "0.39.0"
+version = "0.40.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "401529c9af9304a20ed99fa01799e467b7d37727126f0c9a958895471268ad7a"
+checksum = "a7bf7f20495bcc7dc9f24c5fbcac9e919ca5ebdb7a1b1841d74447d3c8dd0c60"
 dependencies = [
 "anyhow",
 "prettyplease",
@@ -6370,9 +6393,9 @@ dependencies = [

 [[package]]
 name = "wit-component"
-version = "0.225.0"
+version = "0.227.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "2505c917564c1d74774563bbcd3e4f8c216a6508050862fd5f449ee56e3c5125"
+checksum = "635c3adc595422cbf2341a17fb73a319669cc8d33deed3a48368a841df86b676"
 dependencies = [
 "anyhow",
 "bitflags 2.9.0",
@@ -6414,9 +6437,9 @@ dependencies = [

 [[package]]
 name = "wit-parser"
-version = "0.225.0"
+version = "0.227.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ebefaa234e47224f10ce60480c5bfdece7497d0f3b87a12b41ff39e5c8377a78"
+checksum = "ddf445ed5157046e4baf56f9138c124a0824d4d1657e7204d71886ad8ce2fc11"
 dependencies = [
 "anyhow",
 "id-arena",
@@ -14,7 +14,7 @@ strip = "debuginfo"
 codegen-units = 1

 [workspace.package]
-version = "3.0.0-beta.2"
+version = "3.0.0-beta.4"
 authors = ["Luke Street <luke@street.dev>"]
 edition = "2024"
 license = "MIT OR Apache-2.0"
@@ -143,15 +143,15 @@ If not specified, objdiff will use the default patterns listed above.
 `units` _(optional)_: If specified, objdiff will display a list of objects in the sidebar for easy navigation.

 > `name` _(optional)_: The name of the object in the UI. If not specified, the object's `path` will be used.
-> 
+>
 > `target_path`: Path to the "target" or "expected" object from the project root.  
 > This object is the **intended result** of the match.
-> 
+>
 > `base_path`: Path to the "base" or "actual" object from the project root.  
 > This object is built from the **current source code**.
-> 
+>
 > `metadata.auto_generated` _(optional)_: Hides the object from the object list, but still includes it in reports.
-> 
+>
 > `metadata.complete` _(optional)_: Marks the object as "complete" (or "linked") in the object list.  
 > This is useful for marking objects that are fully decompiled. A value of `false` will mark the object as "incomplete".

@@ -173,6 +173,21 @@ $ cargo install --locked --git https://github.com/encounter/objdiff.git objdiff-

 The binaries will be installed to `~/.cargo/bin` as `objdiff` and `objdiff-cli`.

+## Installing `pre-commit`
+
+When contributing, it's recommended to install `pre-commit` to automatically run the linter and formatter before a commit.
+
+[`uv`](https://github.com/astral-sh/uv#installation) is recommended to manage Python version and tools.
+
+Rust nightly is required for `cargo +nightly fmt` and `cargo +nightly clippy`.
+
+```shell
+$ cargo install --locked cargo-deny
+$ rustup toolchain install nightly
+$ uv tool install pre-commit
+$ pre-commit install
+```
+
 ## License

 Licensed under either of
@@ -232,6 +232,7 @@ fn report_object(
            if symbol.section != Some(section_idx)
                || symbol.size == 0
                || symbol.flags.contains(SymbolFlag::Hidden)
+                || symbol.flags.contains(SymbolFlag::Ignored)
            {
                continue;
            }
@@ -81,6 +81,8 @@ std = [
 ]
 mips = [
    "any-arch",
+    "dep:cpp_demangle",
+    "dep:cwdemangle",
    "dep:rabbitizer",
 ]
 ppc = [
@@ -275,4 +275,4 @@
      ]
    }
  ]
-}
+}
@@ -84,18 +84,11 @@ impl ArchArm {
                    .filter_map(|s| DisasmMode::from_symbol(&s))
                    .collect();
                mapping_symbols.sort_unstable_by_key(|x| x.address);
-                (s.index().0, mapping_symbols)
+                (s.index().0 - 1, mapping_symbols)
            })
            .collect()
    }

-    fn endian(&self) -> unarm::Endian {
-        match self.endianness {
-            object::Endianness::Little => unarm::Endian::Little,
-            object::Endianness::Big => unarm::Endian::Big,
-        }
-    }
-
    fn parse_flags(&self, diff_config: &DiffObjConfig) -> unarm::ParseFlags {
        unarm::ParseFlags {
            ual: diff_config.arm_unified_syntax,
@@ -130,6 +123,31 @@ impl ArchArm {
        code: &[u8],
        diff_config: &DiffObjConfig,
    ) -> Result<(unarm::Ins, unarm::ParsedIns)> {
+        if ins_ref.opcode == thumb::Opcode::BlH as u16 && ins_ref.size == 4 {
+            // Special case: combined thumb BL instruction
+            let parse_flags = self.parse_flags(diff_config);
+            let first_ins = thumb::Ins {
+                code: match self.endianness {
+                    object::Endianness::Little => u16::from_le_bytes([code[0], code[1]]),
+                    object::Endianness::Big => u16::from_be_bytes([code[0], code[1]]),
+                } as u32,
+                op: thumb::Opcode::BlH,
+            };
+            let second_ins = thumb::Ins::new(
+                match self.endianness {
+                    object::Endianness::Little => u16::from_le_bytes([code[2], code[3]]),
+                    object::Endianness::Big => u16::from_be_bytes([code[2], code[3]]),
+                } as u32,
+                &parse_flags,
+            );
+            let first_parsed = first_ins.parse(&parse_flags);
+            let second_parsed = second_ins.parse(&parse_flags);
+            return Ok((
+                unarm::Ins::Thumb(first_ins),
+                first_parsed.combine_thumb_bl(&second_parsed),
+            ));
+        }
+
        let code = match (self.endianness, ins_ref.size) {
            (object::Endianness::Little, 2) => u16::from_le_bytes([code[0], code[1]]) as u32,
            (object::Endianness::Little, 4) => {
@@ -153,11 +171,7 @@ impl ArchArm {
        } else {
            let ins = thumb::Ins { code, op: thumb::Opcode::from(ins_ref.opcode as u8) };
            let parsed = ins.parse(&self.parse_flags(diff_config));
-            if ins.is_half_bl() {
-                todo!("Combine thumb BL instructions");
-            } else {
-                (unarm::Ins::Thumb(ins), parsed)
-            }
+            (unarm::Ins::Thumb(ins), parsed)
        };
        Ok((ins, parsed_ins))
    }
@@ -185,60 +199,127 @@ impl Arch for ArchArm {
        let first_mapping_idx = mapping_symbols
            .binary_search_by_key(&start_addr, |x| x.address)
            .unwrap_or_else(|idx| idx - 1);
-        let first_mapping = mapping_symbols[first_mapping_idx].mapping;
+        let mut mode = 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 mappings_iter = mapping_symbols
+            .iter()
+            .copied()
+            .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(start_addr);
+        let ins_count = code.len() / mode.instruction_size(start_addr);
        let mut ops = Vec::<ScannedInstruction>::with_capacity(ins_count);

-        let endian = self.endian();
        let parse_flags = self.parse_flags(diff_config);
-        let mut parser = unarm::Parser::new(first_mapping, start_addr, endian, parse_flags, code);

-        while let Some((address, ins, _parsed_ins)) = parser.next() {
-            let size = parser.mode.instruction_size(address);
-            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 mut address = start_addr;
+        while address < end_addr {
+            while let Some(next) = next_mapping.take_if(|x| address >= x.address) {
+                // Change mapping
+                mode = next.mapping;
+                next_mapping = mappings_iter.next();
            }
-            let (opcode, branch_dest) = match ins {
-                unarm::Ins::Arm(x) => {
-                    let opcode = x.op as u16 | (1 << 15);
-                    let branch_dest = match x.op {
+
+            let mut ins_size = mode.instruction_size(address);
+            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,
+                    },
+                    branch_dest: None,
+                });
+                break;
+            }
+            let code = match (self.endianness, ins_size) {
+                (object::Endianness::Little, 2) => u16::from_le_bytes([data[0], data[1]]) as u32,
+                (object::Endianness::Little, 4) => {
+                    u32::from_le_bytes([data[0], data[1], data[2], data[3]])
+                }
+                (object::Endianness::Big, 2) => u16::from_be_bytes([data[0], data[1]]) as u32,
+                (object::Endianness::Big, 4) => {
+                    u32::from_be_bytes([data[0], data[1], data[2], data[3]])
+                }
+                _ => {
+                    // Invalid instruction size
+                    ops.push(ScannedInstruction {
+                        ins_ref: InstructionRef {
+                            address: address as u64,
+                            size: ins_size as u8,
+                            opcode: u16::MAX,
+                        },
+                        branch_dest: None,
+                    });
+                    address += ins_size as u32;
+                    continue;
+                }
+            };
+
+            let (opcode, branch_dest) = match mode {
+                unarm::ParseMode::Arm => {
+                    let ins = arm::Ins::new(code, &parse_flags);
+                    let opcode = ins.op as u16 | (1 << 15);
+                    let branch_dest = match ins.op {
                        arm::Opcode::B | arm::Opcode::Bl => {
-                            address.checked_add_signed(x.field_branch_offset())
+                            address.checked_add_signed(ins.field_branch_offset())
                        }
-                        arm::Opcode::BlxI => address.checked_add_signed(x.field_blx_offset()),
+                        arm::Opcode::BlxI => address.checked_add_signed(ins.field_blx_offset()),
                        _ => None,
                    };
                    (opcode, branch_dest)
                }
-                unarm::Ins::Thumb(x) => {
-                    let opcode = x.op as u16;
-                    let branch_dest = match x.op {
+                unarm::ParseMode::Thumb => {
+                    let ins = thumb::Ins::new(code, &parse_flags);
+                    let opcode = ins.op as u16;
+                    let branch_dest = match ins.op {
                        thumb::Opcode::B | thumb::Opcode::Bl => {
-                            address.checked_add_signed(x.field_branch_offset_8())
+                            address.checked_add_signed(ins.field_branch_offset_8())
+                        }
+                        thumb::Opcode::BlH if data.len() >= 4 => {
+                            // Combine BL instructions
+                            let second_ins = thumb::Ins::new(
+                                match self.endianness {
+                                    object::Endianness::Little => {
+                                        u16::from_le_bytes([data[2], data[3]]) as u32
+                                    }
+                                    object::Endianness::Big => {
+                                        u16::from_be_bytes([data[2], data[3]]) as u32
+                                    }
+                                },
+                                &parse_flags,
+                            );
+                            if let Some(low) = match second_ins.op {
+                                thumb::Opcode::Bl => Some(second_ins.field_low_branch_offset_11()),
+                                thumb::Opcode::BlxI => Some(second_ins.field_low_blx_offset_11()),
+                                _ => None,
+                            } {
+                                ins_size = 4;
+                                address.checked_add_signed(
+                                    (ins.field_high_branch_offset_11() + (low as i32)) << 9 >> 9,
+                                )
+                            } else {
+                                None
+                            }
                        }
                        thumb::Opcode::BLong => {
-                            address.checked_add_signed(x.field_branch_offset_11())
+                            address.checked_add_signed(ins.field_branch_offset_11())
                        }
                        _ => None,
                    };
                    (opcode, branch_dest)
                }
-                unarm::Ins::Data => (u16::MAX, None),
+                unarm::ParseMode::Data => (u16::MAX, None),
            };
+
            ops.push(ScannedInstruction {
-                ins_ref: InstructionRef { address: address as u64, size: size as u8, opcode },
+                ins_ref: 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;
        }

        Ok(ops)
@@ -391,7 +472,7 @@ fn push_args(
    let reloc_arg = find_reloc_arg(parsed_ins, relocation);
    let mut writeback = false;
    let mut deref = false;
-    for (i, arg) in parsed_ins.args_iter().enumerate() {
+    for (i, &arg) in parsed_ins.args_iter().enumerate() {
        // Emit punctuation before separator
        if deref {
            match arg {
@@ -463,19 +544,19 @@ fn push_args(
                | args::Argument::CoOpcode(value)
                | args::Argument::SatImm(value) => {
                    arg_cb(InstructionPart::basic("#"))?;
-                    arg_cb(InstructionPart::unsigned(*value))?;
+                    arg_cb(InstructionPart::unsigned(value))?;
                }
                args::Argument::SImm(value)
                | args::Argument::OffsetImm(args::OffsetImm { post_indexed: _, value }) => {
                    arg_cb(InstructionPart::basic("#"))?;
-                    arg_cb(InstructionPart::signed(*value))?;
+                    arg_cb(InstructionPart::signed(value))?;
                }
                args::Argument::BranchDest(value) => {
-                    arg_cb(InstructionPart::branch_dest(cur_addr.wrapping_add_signed(*value)))?;
+                    arg_cb(InstructionPart::branch_dest(cur_addr.wrapping_add_signed(value)))?;
                }
                args::Argument::CoOption(value) => {
                    arg_cb(InstructionPart::basic("{"))?;
-                    arg_cb(InstructionPart::unsigned(*value))?;
+                    arg_cb(InstructionPart::unsigned(value))?;
                    arg_cb(InstructionPart::basic("}"))?;
                }
                args::Argument::CoprocNum(value) => {
@@ -1,4 +1,8 @@
-use alloc::{collections::BTreeMap, string::ToString, vec::Vec};
+use alloc::{
+    collections::{BTreeMap, BTreeSet},
+    string::{String, ToString},
+    vec::Vec,
+};
 use core::ops::Range;

 use anyhow::{Result, bail};
@@ -15,7 +19,7 @@ use crate::{
    diff::{DiffObjConfig, MipsAbi, MipsInstrCategory, display::InstructionPart},
    obj::{
        InstructionArg, InstructionArgValue, InstructionRef, Relocation, RelocationFlags,
-        ResolvedInstructionRef, ResolvedRelocation, ScannedInstruction,
+        ResolvedInstructionRef, ResolvedRelocation, ScannedInstruction, SymbolFlag, SymbolFlagSet,
    },
 };

@@ -26,6 +30,7 @@ pub struct ArchMips {
    pub isa_extension: Option<IsaExtension>,
    pub ri_gp_value: i32,
    pub paired_relocations: Vec<BTreeMap<u64, i64>>,
+    pub ignored_symbols: BTreeSet<usize>,
 }

 const EF_MIPS_ABI: u32 = 0x0000F000;
@@ -118,7 +123,25 @@ impl ArchMips {
            paired_relocations[section_index] = addends;
        }

-        Ok(Self { endianness, abi, isa_extension, ri_gp_value, paired_relocations })
+        let mut ignored_symbols = BTreeSet::new();
+        for obj_symbol in object.symbols() {
+            let Ok(name) = obj_symbol.name() else { continue };
+            if let Some(prefix) = name.strip_suffix(".NON_MATCHING") {
+                ignored_symbols.insert(obj_symbol.index().0);
+                if let Some(target_symbol) = object.symbol_by_name(prefix) {
+                    ignored_symbols.insert(target_symbol.index().0);
+                }
+            }
+        }
+
+        Ok(Self {
+            endianness,
+            abi,
+            isa_extension,
+            ri_gp_value,
+            paired_relocations,
+            ignored_symbols,
+        })
    }

    fn instruction_flags(&self, diff_config: &DiffObjConfig) -> rabbitizer::InstructionFlags {
@@ -263,6 +286,13 @@ impl Arch for ArchMips {
        })
    }

+    fn demangle(&self, name: &str) -> Option<String> {
+        cpp_demangle::Symbol::new(name)
+            .ok()
+            .and_then(|s| s.demangle(&cpp_demangle::DemangleOptions::default()).ok())
+            .or_else(|| cwdemangle::demangle(name, &cwdemangle::DemangleOptions::default()))
+    }
+
    fn reloc_name(&self, flags: RelocationFlags) -> Option<&'static str> {
        match flags {
            RelocationFlags::Elf(r_type) => match r_type {
@@ -294,6 +324,14 @@ impl Arch for ArchMips {
            _ => 1,
        }
    }
+
+    fn extra_symbol_flags(&self, symbol: &object::Symbol) -> SymbolFlagSet {
+        let mut flags = SymbolFlagSet::default();
+        if self.ignored_symbols.contains(&symbol.index().0) {
+            flags |= SymbolFlag::Ignored;
+        }
+        flags
+    }
 }

 fn push_args(
@@ -137,7 +137,7 @@ impl Arch for ArchX86 {
            {
                instruction.set_memory_displacement64(PLACEHOLDER);
                // Formatter always writes the displacement as Int32
-                reloc_replace = Some((OpKind::Memory, NumberKind::Int32, PLACEHOLDER));
+                reloc_replace = Some((OpKind::Memory, 4, PLACEHOLDER));
            } else if reloc_offset == offsets.immediate_offset() as u64
                && reloc_size == offsets.immediate_size()
            {
@@ -155,18 +155,12 @@ impl Arch for ArchX86 {
                        _ => OpKind::default(),
                    }
                };
-                let number_kind = match reloc_size {
-                    2 => NumberKind::UInt16,
-                    4 => NumberKind::UInt32,
-                    8 => NumberKind::UInt64,
-                    _ => NumberKind::default(),
-                };
                if is_branch {
                    instruction.set_near_branch64(PLACEHOLDER);
                } else {
                    instruction.set_immediate32(PLACEHOLDER as u32);
                }
-                reloc_replace = Some((op_kind, number_kind, PLACEHOLDER));
+                reloc_replace = Some((op_kind, reloc_size, PLACEHOLDER));
            }
        }

@@ -251,7 +245,7 @@ impl Arch for ArchX86 {

 struct InstructionFormatterOutput<'a> {
    cb: &'a mut dyn FnMut(InstructionPart<'_>) -> Result<()>,
-    reloc_replace: Option<(OpKind, NumberKind, u64)>,
+    reloc_replace: Option<(OpKind, usize, u64)>,
    error: Option<anyhow::Error>,
    skip_next: bool,
 }
@@ -326,11 +320,18 @@ impl FormatterOutput for InstructionFormatterOutput<'_> {
            return;
        }

-        if let (Some(operand), Some((target_op_kind, target_number_kind, target_value))) =
+        if let (Some(operand), Some((target_op_kind, reloc_size, target_value))) =
            (instruction_operand, self.reloc_replace)
        {
+            #[allow(clippy::match_like_matches_macro)]
            if instruction.op_kind(operand) == target_op_kind
-                && number_kind == target_number_kind
+                && match (number_kind, reloc_size) {
+                    (NumberKind::Int8 | NumberKind::UInt8, 1)
+                    | (NumberKind::Int16 | NumberKind::UInt16, 2)
+                    | (NumberKind::Int32 | NumberKind::UInt32, 4)
+                    | (NumberKind::Int64 | NumberKind::UInt64, 8) => true,
+                    _ => false,
+                }
                && value == target_value
            {
                if let Err(e) = (self.cb)(InstructionPart::reloc()) {
@@ -571,4 +572,43 @@ mod test {
        .unwrap();
        assert_eq!(parts, &[InstructionPart::opcode("call", opcode), InstructionPart::reloc()]);
    }
+
+    #[test]
+    fn test_process_instruction_with_reloc_4() {
+        let arch = ArchX86 { arch: Architecture::X86, endianness: object::Endianness::Little };
+        let code = [0x8b, 0x15, 0xa4, 0x21, 0x7e, 0x00];
+        let opcode = iced_x86::Mnemonic::Mov as u16;
+        let mut parts = Vec::new();
+        arch.display_instruction(
+            ResolvedInstructionRef {
+                ins_ref: InstructionRef { address: 0x1234, size: 6, opcode },
+                code: &code,
+                relocation: Some(ResolvedRelocation {
+                    relocation: &Relocation {
+                        flags: RelocationFlags::Coff(pe::IMAGE_REL_I386_DIR32),
+                        address: 0x1234 + 2,
+                        target_symbol: 0,
+                        addend: 0,
+                    },
+                    symbol: &Default::default(),
+                }),
+                ..Default::default()
+            },
+            &DiffObjConfig::default(),
+            &mut |part| {
+                parts.push(part.into_static());
+                Ok(())
+            },
+        )
+        .unwrap();
+        assert_eq!(parts, &[
+            InstructionPart::opcode("mov", opcode),
+            InstructionPart::opaque("edx"),
+            InstructionPart::basic(","),
+            InstructionPart::basic(" "),
+            InstructionPart::basic("["),
+            InstructionPart::reloc(),
+            InstructionPart::basic("]"),
+        ]);
+    }
 }
@@ -139,27 +139,13 @@ pub fn diff_data_section(
 ) -> Result<(SectionDiff, SectionDiff)> {
    let left_section = &left_obj.sections[left_section_idx];
    let right_section = &right_obj.sections[right_section_idx];
-    let left_max = left_obj
-        .symbols
-        .iter()
-        .filter_map(|s| {
-            if s.section != Some(left_section_idx) || s.kind == SymbolKind::Section {
-                return None;
-            }
-            s.address.checked_sub(left_section.address).map(|a| a + s.size)
-        })
+    let left_max = symbols_matching_section(&left_obj.symbols, left_section_idx)
+        .filter_map(|(_, s)| s.address.checked_sub(left_section.address).map(|a| a + s.size))
        .max()
        .unwrap_or(0)
        .min(left_section.size);
-    let right_max = right_obj
-        .symbols
-        .iter()
-        .filter_map(|s| {
-            if s.section != Some(right_section_idx) || s.kind == SymbolKind::Section {
-                return None;
-            }
-            s.address.checked_sub(right_section.address).map(|a| a + s.size)
-        })
+    let right_max = symbols_matching_section(&right_obj.symbols, right_section_idx)
+        .filter_map(|(_, s)| s.address.checked_sub(right_section.address).map(|a| a + s.size))
        .max()
        .unwrap_or(0)
        .min(right_section.size);
@@ -412,21 +398,13 @@ pub fn diff_generic_section(
    left_section_idx: usize,
    _right_section_idx: usize,
 ) -> Result<(SectionDiff, SectionDiff)> {
-    let match_percent = if left_obj
-        .symbols
-        .iter()
-        .enumerate()
-        .filter(|(_, s)| s.section == Some(left_section_idx) && s.kind != SymbolKind::Section)
+    let match_percent = if symbols_matching_section(&left_obj.symbols, left_section_idx)
        .map(|(i, _)| &left_diff.symbols[i])
        .all(|d| d.match_percent == Some(100.0))
    {
        100.0 // Avoid fp precision issues
    } else {
-        let (matched, total) = left_obj
-            .symbols
-            .iter()
-            .enumerate()
-            .filter(|(_, s)| s.section == Some(left_section_idx) && s.kind != SymbolKind::Section)
+        let (matched, total) = symbols_matching_section(&left_obj.symbols, left_section_idx)
            .map(|(i, s)| (s, &left_diff.symbols[i]))
            .fold((0.0, 0.0), |(matched, total), (s, d)| {
                (matched + d.match_percent.unwrap_or(0.0) * s.size as f32, total + s.size as f32)
@@ -449,19 +427,11 @@ pub fn diff_bss_section(
    right_section_idx: usize,
 ) -> Result<(SectionDiff, SectionDiff)> {
    let left_section = &left_obj.sections[left_section_idx];
-    let left_sizes = left_obj
-        .symbols
-        .iter()
-        .enumerate()
-        .filter(|(_, s)| s.section == Some(left_section_idx) && s.kind != SymbolKind::Section)
+    let left_sizes = symbols_matching_section(&left_obj.symbols, left_section_idx)
        .filter_map(|(_, s)| s.address.checked_sub(left_section.address).map(|a| (a, s.size)))
        .collect::<Vec<_>>();
    let right_section = &right_obj.sections[right_section_idx];
-    let right_sizes = right_obj
-        .symbols
-        .iter()
-        .enumerate()
-        .filter(|(_, s)| s.section == Some(right_section_idx) && s.kind != SymbolKind::Section)
+    let right_sizes = symbols_matching_section(&right_obj.symbols, right_section_idx)
        .filter_map(|(_, s)| s.address.checked_sub(right_section.address).map(|a| (a, s.size)))
        .collect::<Vec<_>>();
    let ops = capture_diff_slices(Algorithm::Patience, &left_sizes, &right_sizes);
@@ -487,3 +457,15 @@ pub fn diff_bss_section(
        SectionDiff { match_percent: Some(match_percent), data_diff: vec![], reloc_diff: vec![] },
    ))
 }
+
+fn symbols_matching_section(
+    symbols: &[Symbol],
+    section_idx: usize,
+) -> impl Iterator<Item = (usize, &Symbol)> + '_ {
+    symbols.iter().enumerate().filter(move |(_, s)| {
+        s.section == Some(section_idx)
+            && s.kind != SymbolKind::Section
+            && s.size > 0
+            && !s.flags.contains(SymbolFlag::Ignored)
+    })
+}
@@ -563,7 +563,7 @@ pub fn instruction_hover(
    out
 }

-#[derive(Copy, Clone)]
+#[derive(Debug, Copy, Clone)]
 pub enum SymbolFilter<'a> {
    None,
    Search(&'a Regex),
@@ -580,7 +580,11 @@ fn symbol_matches_filter(
    if symbol.section.is_none() && !symbol.flags.contains(SymbolFlag::Common) {
        return false;
    }
-    if !show_hidden_symbols && (symbol.size == 0 || symbol.flags.contains(SymbolFlag::Hidden)) {
+    if !show_hidden_symbols
+        && (symbol.size == 0
+            || symbol.flags.contains(SymbolFlag::Hidden)
+            || symbol.flags.contains(SymbolFlag::Ignored))
+    {
        return false;
    }
    match filter {
@@ -701,7 +705,7 @@ pub fn display_sections(
            });
        }
    }
-    sections.sort_by(|a, b| a.id.cmp(&b.id));
+    sections.sort_by(|a, b| a.name.cmp(&b.name));
    sections
 }

@@ -371,7 +371,10 @@ fn generate_mapping_symbols(
    };
    let base_section_kind = symbol_section_kind(base_obj, &base_obj.symbols[base_symbol_ref]);
    for (target_symbol_index, target_symbol) in target_obj.symbols.iter().enumerate() {
-        if symbol_section_kind(target_obj, target_symbol) != base_section_kind {
+        if target_symbol.size == 0
+            || target_symbol.flags.contains(SymbolFlag::Ignored)
+            || symbol_section_kind(target_obj, target_symbol) != base_section_kind
+        {
            continue;
        }
        match base_section_kind {
@@ -526,6 +529,9 @@ fn matching_symbols(
            )?;
        }
        for (symbol_idx, symbol) in left.symbols.iter().enumerate() {
+            if symbol.size == 0 || symbol.flags.contains(SymbolFlag::Ignored) {
+                continue;
+            }
            let section_kind = symbol_section_kind(left, symbol);
            if section_kind == SectionKind::Unknown {
                continue;
@@ -547,6 +553,9 @@ fn matching_symbols(
    }
    if let Some(right) = right {
        for (symbol_idx, symbol) in right.symbols.iter().enumerate() {
+            if symbol.size == 0 || symbol.flags.contains(SymbolFlag::Ignored) {
+                continue;
+            }
            let section_kind = symbol_section_kind(right, symbol);
            if section_kind == SectionKind::Unknown {
                continue;
@@ -572,9 +581,10 @@ fn unmatched_symbols<'obj, 'used>(
 where
    'obj: 'used,
 {
-    obj.symbols.iter().enumerate().filter(move |&(symbol_idx, _)| {
-        // Skip symbols that have already been matched
-        !used.is_some_and(|u| u.contains(&symbol_idx))
+    obj.symbols.iter().enumerate().filter(move |&(symbol_idx, symbol)| {
+        !symbol.flags.contains(SymbolFlag::Ignored)
+            // Skip symbols that have already been matched
+            && !used.is_some_and(|u| u.contains(&symbol_idx))
    })
 }

@@ -42,6 +42,8 @@ flags! {
        HasExtra,
        /// Symbol size was missing and was inferred
        SizeInferred,
+        /// Symbol should be ignored by any diffing
+        Ignored,
    }
 }

@@ -15,3 +15,20 @@ fn read_arm() {
    let output = common::display_diff(&obj, &diff, symbol_idx, &diff_config);
    insta::assert_snapshot!(output);
 }
+
+#[test]
+#[cfg(feature = "arm")]
+fn read_thumb() {
+    let diff_config = diff::DiffObjConfig { mips_register_prefix: true, ..Default::default() };
+    let obj = obj::read::parse(include_object!("data/arm/thumb.o"), &diff_config).unwrap();
+    insta::assert_debug_snapshot!(obj);
+    let symbol_idx = obj
+        .symbols
+        .iter()
+        .position(|s| s.name == "THUMB_BRANCH_ServerDisplay_UncategorizedMove")
+        .unwrap();
+    let diff = diff::code::no_diff_code(&obj, symbol_idx, &diff_config).unwrap();
+    insta::assert_debug_snapshot!(diff.instruction_rows);
+    let output = common::display_diff(&obj, &diff, symbol_idx, &diff_config);
+    insta::assert_snapshot!(output);
+}
@@ -37,3 +37,11 @@ fn cross_endian_diff() {
    assert_eq!(left_diff.instruction_rows[2].kind, diff::InstructionDiffKind::None);
    assert_eq!(right_diff.instruction_rows[2].kind, diff::InstructionDiffKind::None);
 }
+
+#[test]
+#[cfg(feature = "mips")]
+fn filter_non_matching() {
+    let diff_config = diff::DiffObjConfig::default();
+    let obj = obj::read::parse(include_object!("data/mips/vw_main.c.o"), &diff_config).unwrap();
+    insta::assert_debug_snapshot!(obj.symbols);
+}
@@ -61,7 +61,7 @@ fn diff_ppc() {
    let base_diff = diff.right.as_ref().unwrap();
    let sections_display = display::display_sections(
        &target_obj,
-        &target_diff,
+        target_diff,
        display::SymbolFilter::None,
        false,
        false,
@@ -1,4 +1,4 @@
-use objdiff_core::{diff, obj};
+use objdiff_core::{diff, diff::display::SymbolFilter, obj};

 mod common;

@@ -40,3 +40,18 @@ fn read_x86_64() {
    let output = common::display_diff(&obj, &diff, symbol_idx, &diff_config);
    insta::assert_snapshot!(output);
 }
+
+#[test]
+#[cfg(feature = "x86")]
+fn display_section_ordering() {
+    let diff_config = diff::DiffObjConfig::default();
+    let obj = obj::read::parse(include_object!("data/x86/basenode.obj"), &diff_config).unwrap();
+    let obj_diff =
+        diff::diff_objs(Some(&obj), None, None, &diff_config, &diff::MappingConfig::default())
+            .unwrap()
+            .left
+            .unwrap();
+    let section_display =
+        diff::display::display_sections(&obj, &obj_diff, SymbolFilter::None, false, false, false);
+    insta::assert_debug_snapshot!(section_display);
+}
--- a/Show More
+++ b/Show More
Author	SHA1	Message	Date
Luke Street	6768df9d80	Version v3.0.0-beta.4	2025-03-18 21:36:37 -06:00
Luke Street	eaba2391e0	arm: Fix thumb & data decoding, add tests	2025-03-18 21:26:28 -06:00
Luke Street	bd8f5ede23	Add pre-commit hooks	2025-03-18 21:26:28 -06:00
Luke Street	acf46c6b54	Fix no_std build	2025-03-14 09:49:31 -06:00
Luke Street	9358d8ec60	mips: Add C++ symbol demangling (CW & modern GCC)	2025-03-14 09:44:15 -06:00
Luke Street	809e2ffddc	cargo fmt	2025-03-11 21:43:32 -06:00
Luke Street	87fa29e8b0	objdiff-wasm: Fix symbol filtering regex crate needed the `unicode-case` feature	2025-03-11 21:42:14 -06:00
Luke Street	42d4c38079	mips: Ignore `.NON_MATCHING` functions from `INLINE_ASM` macros	2025-03-11 21:39:17 -06:00
Luke Street	fa26200ed7	Silence warning	2025-03-10 22:22:05 -06:00
Luke Street	a0e7f9bc37	Fix x86 mov relocations with uint32	2025-03-10 22:09:58 -06:00
Luke Street	5898d7aebf	Fix section ordering with many same-named sections Before, this was comparing, for example, `.text-2` with `.text-10` with standard string comparison, yielding `.text-10` before `.text-2`. Instead, this simply uses a stable sort by name, which preserves the relative ordering of sections.	2025-03-10 21:51:54 -06:00