ARM: Fix "Combine text sections" confusing code and data (#195)

* ARM: Fix parsing of mapping symbols when "Combine text sections" is enabled

* Add test

objdiff-core/src/arch/arm.rs

@@ -15,7 +15,7 @@ use crate::{
     diff::{ArmArchVersion, ArmR9Usage, DiffObjConfig, display::InstructionPart},
     obj::{
         InstructionRef, Relocation, RelocationFlags, ResolvedInstructionRef, ResolvedRelocation,
-        ScannedInstruction, SymbolFlag, SymbolFlagSet, SymbolKind,
+        ScannedInstruction, Section, SectionKind, Symbol, SymbolFlag, SymbolFlagSet, SymbolKind,
     },
 };
 
@@ -32,7 +32,8 @@ impl ArchArm {
         let endianness = file.endianness();
         match file {
             object::File::Elf32(_) => {
-                let disasm_modes = Self::elf_get_mapping_symbols(file);
+                // The disasm_modes mapping is populated later in the post_init step so that we have access to merged sections.
+                let disasm_modes = BTreeMap::new();
                 let detected_version = Self::elf_detect_arm_version(file)?;
                 Ok(Self { disasm_modes, detected_version, endianness })
             }
@@ -73,18 +74,22 @@ impl ArchArm {
         Ok(None)
     }
 
-    fn elf_get_mapping_symbols(file: &object::File) -> BTreeMap<usize, Vec<DisasmMode>> {
-        file.sections()
-            .filter(|s| s.kind() == object::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))
+    fn get_mapping_symbols(
+        sections: &[Section],
+        symbols: &[Symbol],
+    ) -> BTreeMap<usize, Vec<DisasmMode>> {
+        sections
+            .iter()
+            .enumerate()
+            .filter(|(_, section)| section.kind == SectionKind::Code)
+            .map(|(index, _)| {
+                let mut mapping_symbols: Vec<_> = symbols
+                    .iter()
+                    .filter(|s| s.section.map(|i| i == index).unwrap_or(false))
+                    .filter_map(DisasmMode::from_symbol)
                     .collect();
                 mapping_symbols.sort_unstable_by_key(|x| x.address);
-                (s.index().0 - 1, mapping_symbols)
+                (index, mapping_symbols)
             })
             .collect()
     }
@@ -178,6 +183,10 @@ impl ArchArm {
 }
 
 impl Arch for ArchArm {
+    fn post_init(&mut self, sections: &[Section], symbols: &[Symbol]) {
+        self.disasm_modes = Self::get_mapping_symbols(sections, symbols);
+    }
+
     fn scan_instructions(
         &self,
         address: u64,
@@ -441,7 +450,7 @@ impl Arch for ArchArm {
 
     fn extra_symbol_flags(&self, symbol: &object::Symbol) -> SymbolFlagSet {
         let mut flags = SymbolFlagSet::default();
-        if DisasmMode::from_symbol(symbol).is_some() {
+        if DisasmMode::from_object_symbol(symbol).is_some() {
             flags |= SymbolFlag::Hidden;
         }
         flags
@@ -455,12 +464,17 @@ struct DisasmMode {
 }
 
 impl DisasmMode {
-    fn from_symbol<'a>(sym: &object::Symbol<'a, '_, &'a [u8]>) -> Option<Self> {
+    fn from_object_symbol<'a>(sym: &object::Symbol<'a, '_, &'a [u8]>) -> Option<Self> {
         sym.name()
             .ok()
             .and_then(unarm::ParseMode::from_mapping_symbol)
             .map(|mapping| DisasmMode { address: sym.address() as u32, mapping })
     }
+
+    fn from_symbol(sym: &Symbol) -> Option<Self> {
+        unarm::ParseMode::from_mapping_symbol(&sym.name)
+            .map(|mapping| DisasmMode { address: sym.address as u32, mapping })
+    }
 }
 
 fn push_args(

objdiff-core/src/arch/mod.rs

@@ -12,7 +12,7 @@ use crate::{
     },
     obj::{
         InstructionArg, Object, ParsedInstruction, Relocation, RelocationFlags,
-        ResolvedInstructionRef, ScannedInstruction, Symbol, SymbolFlagSet, SymbolKind,
+        ResolvedInstructionRef, ScannedInstruction, Section, Symbol, SymbolFlagSet, SymbolKind,
     },
     util::ReallySigned,
 };
@@ -183,6 +183,9 @@ impl DataType {
 }
 
 pub trait Arch: Send + Sync + Debug {
+    // Finishes arch-specific initialization that must be done after sections have been combined.
+    fn post_init(&mut self, _sections: &[Section], _symbols: &[Symbol]) {}
+
     /// Generate a list of instructions references (offset, size, opcode) from the given code.
     ///
     /// The opcode IDs are used to generate the initial diff. Implementations should do as little

objdiff-core/src/obj/read.rs

@@ -843,7 +843,7 @@ pub fn read(obj_path: &std::path::Path, config: &DiffObjConfig) -> Result<Object
 
 pub fn parse(data: &[u8], config: &DiffObjConfig) -> Result<Object> {
     let obj_file = object::File::parse(data)?;
-    let arch = new_arch(&obj_file)?;
+    let mut arch = new_arch(&obj_file)?;
     let split_meta = parse_split_meta(&obj_file)?;
     let (mut sections, section_indices) =
         map_sections(arch.as_ref(), &obj_file, split_meta.as_ref())?;
@@ -857,6 +857,7 @@ pub fn parse(data: &[u8], config: &DiffObjConfig) -> Result<Object> {
     if config.combine_data_sections || config.combine_text_sections {
         combine_sections(&mut sections, &mut symbols, config)?;
     }
+    arch.post_init(&sections, &symbols);
     Ok(Object {
         arch,
         endianness: obj_file.endianness(),

objdiff-core/tests/arch_arm.rs

@@ -5,7 +5,7 @@ mod common;
 #[test]
 #[cfg(feature = "arm")]
 fn read_arm() {
-    let diff_config = diff::DiffObjConfig { mips_register_prefix: true, ..Default::default() };
+    let diff_config = diff::DiffObjConfig { ..Default::default() };
     let obj = obj::read::parse(include_object!("data/arm/LinkStateItem.o"), &diff_config).unwrap();
     insta::assert_debug_snapshot!(obj);
     let symbol_idx =
@@ -19,7 +19,7 @@ fn read_arm() {
 #[test]
 #[cfg(feature = "arm")]
 fn read_thumb() {
-    let diff_config = diff::DiffObjConfig { mips_register_prefix: true, ..Default::default() };
+    let diff_config = diff::DiffObjConfig { ..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
@@ -32,3 +32,15 @@ fn read_thumb() {
     let output = common::display_diff(&obj, &diff, symbol_idx, &diff_config);
     insta::assert_snapshot!(output);
 }
+
+#[test]
+#[cfg(feature = "arm")]
+fn combine_text_sections() {
+    let diff_config = diff::DiffObjConfig { combine_text_sections: true, ..Default::default() };
+    let obj = obj::read::parse(include_object!("data/arm/enemy300.o"), &diff_config).unwrap();
+    let symbol_idx = obj.symbols.iter().position(|s| s.name == "Enemy300Draw").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);
+}

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

@@ -0,0 +1,8 @@
+---
+source: objdiff-core/tests/arch_arm.rs
+assertion_line: 45
+expression: output
+---
+[(Line(90), Dim, 5), (Address(0), Normal, 5), (Spacing(4), Normal, 0), (Opcode("ldr", 32799), Normal, 10), (Argument(Opaque("r12")), Normal, 0), (Basic(", "), Normal, 0), (Basic("["), Normal, 0), (Argument(Opaque("pc")), Normal, 0), (Basic(", "), Normal, 0), (Basic("#"), Normal, 0), (Argument(Signed(0)), Normal, 0), (Basic("]"), Normal, 0), (Eol, Normal, 0)]
+[(Line(90), Dim, 5), (Address(4), Normal, 5), (Spacing(4), Normal, 0), (Opcode("bx", 32779), Normal, 10), (Argument(Opaque("r12")), Normal, 0), (Eol, Normal, 0)]
+[(Line(90), Dim, 5), (Address(8), Normal, 5), (Spacing(4), Normal, 0), (Opcode(".word", 65535), Normal, 10), (Symbol(Symbol { name: "esEnemyDraw", demangled_name: None, address: 0, size: 0, kind: Unknown, section: None, flags: FlagSet(Global), align: None, virtual_address: None }), Bright, 0), (Eol, Normal, 0)]

objdiff-core/tests/snapshots/arch_arm__combine_text_sections.snap

@@ -0,0 +1,46 @@
+---
+source: objdiff-core/tests/arch_arm.rs
+assertion_line: 43
+expression: diff.instruction_rows
+---
+[
+    InstructionDiffRow {
+        ins_ref: Some(
+            InstructionRef {
+                address: 76,
+                size: 4,
+                opcode: 32799,
+            },
+        ),
+        kind: None,
+        branch_from: None,
+        branch_to: None,
+        arg_diff: [],
+    },
+    InstructionDiffRow {
+        ins_ref: Some(
+            InstructionRef {
+                address: 80,
+                size: 4,
+                opcode: 32779,
+            },
+        ),
+        kind: None,
+        branch_from: None,
+        branch_to: None,
+        arg_diff: [],
+    },
+    InstructionDiffRow {
+        ins_ref: Some(
+            InstructionRef {
+                address: 84,
+                size: 4,
+                opcode: 65535,
+            },
+        ),
+        kind: None,
+        branch_from: None,
+        branch_to: None,
+        arg_diff: [],
+    },
+]