objdiff-core API adjustments

encounter · encounter · commit 3f82c1a50fd4 · 2024-06-03T18:52:32.000-06:00
- Allows using process_code without
  constructing an ObjInfo
- Allows creating an arch without
  having to provide an object

Used in decomp-toolkit
diff --git a/objdiff-core/src/arch/mips.rs b/objdiff-core/src/arch/mips.rs
@@ -1,4 +1,4 @@
-use std::{borrow::Cow, sync::Mutex};
+use std::{borrow::Cow, collections::BTreeMap, sync::Mutex};
 
 use anyhow::{anyhow, bail, Result};
 use object::{elf, Endian, Endianness, File, FileFlags, Object, Relocation, RelocationFlags};
@@ -7,7 +7,7 @@ use rabbitizer::{config, Abi, InstrCategory, Instruction, OperandType};
 use crate::{
     arch::{ObjArch, ProcessCodeResult},
     diff::{DiffObjConfig, MipsAbi, MipsInstrCategory},
-    obj::{ObjInfo, ObjIns, ObjInsArg, ObjInsArgValue, ObjReloc, ObjSection, SymbolRef},
+    obj::{ObjIns, ObjInsArg, ObjInsArgValue, ObjReloc, ObjSection},
 };
 
 static RABBITIZER_MUTEX: Mutex<()> = Mutex::new(());
@@ -57,15 +57,12 @@ impl ObjArchMips {
 impl ObjArch for ObjArchMips {
     fn process_code(
         &self,
-        obj: &ObjInfo,
-        symbol_ref: SymbolRef,
+        address: u64,
+        code: &[u8],
+        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,
@@ -82,14 +79,14 @@ impl ObjArch for ObjArchMips {
             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, instr_category);
 
@@ -155,7 +152,7 @@ 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,
diff --git a/objdiff-core/src/arch/mod.rs b/objdiff-core/src/arch/mod.rs
@@ -1,25 +1,27 @@
-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 = "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],
+        relocations: &[ObjReloc],
+        line_info: &BTreeMap<u64, u64>,
         config: &DiffObjConfig,
     ) -> Result<ProcessCodeResult>;
 
diff --git a/objdiff-core/src/arch/ppc.rs b/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 +31,17 @@ impl ObjArchPpc {
 impl ObjArch for ObjArchPpc {
     fn process_code(
         &self,
-        obj: &ObjInfo,
-        symbol_ref: SymbolRef,
+        address: u64,
+        code: &[u8],
+        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 +130,7 @@ 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,
diff --git a/objdiff-core/src/arch/x86.rs b/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 +28,14 @@ impl ObjArchX86 {
 impl ObjArch for ObjArchX86 {
     fn process_code(
         &self,
-        obj: &ObjInfo,
-        symbol_ref: SymbolRef,
+        address: u64,
+        code: &[u8],
+        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 +67,10 @@ 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,
diff --git a/objdiff-core/src/diff/code.rs b/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 +16,41 @@ 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.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);
diff --git a/objdiff-core/src/diff/mod.rs b/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,
@@ -13,8 +13,8 @@ use crate::{
     obj::{ObjInfo, ObjIns, ObjSection, ObjSectionKind, ObjSymbol, SymbolRef},
 };
 
-mod code;
-mod data;
+pub mod code;
+pub mod data;
 pub mod display;
 
 #[derive(
@@ -321,9 +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,
@@ -333,9 +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,
@@ -369,8 +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) =
@@ -382,8 +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) =
