Diff data symbols & improve symbol match logic

encounter · encounter · commit 22a24f37f549 · 2024-05-20T23:53:37.000-06:00
diff --git a/objdiff-core/src/arch/mips.rs b/objdiff-core/src/arch/mips.rs
@@ -1,6 +1,6 @@
 use std::borrow::Cow;
 
-use anyhow::{bail, Result};
+use anyhow::{anyhow, bail, Result};
 use object::{elf, Endian, Endianness, File, Object, Relocation, RelocationFlags};
 use rabbitizer::{config, Abi, InstrCategory, Instruction, OperandType};
 
@@ -35,6 +35,7 @@ impl ObjArch for ObjArchMips {
         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];
 
diff --git a/objdiff-core/src/arch/ppc.rs b/objdiff-core/src/arch/ppc.rs
@@ -1,6 +1,6 @@
 use std::borrow::Cow;
 
-use anyhow::{bail, Result};
+use anyhow::{anyhow, bail, Result};
 use object::{elf, File, Relocation, RelocationFlags};
 use ppc750cl::{Argument, InsIter, GPR};
 
@@ -36,6 +36,7 @@ impl ObjArch for ObjArchPpc {
         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];
 
diff --git a/objdiff-core/src/arch/x86.rs b/objdiff-core/src/arch/x86.rs
@@ -33,6 +33,7 @@ impl ObjArch for ObjArchX86 {
         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];
 
diff --git a/objdiff-core/src/diff/code.rs b/objdiff-core/src/diff/code.rs
@@ -28,7 +28,7 @@ pub fn no_diff_code(
         diff.push(ObjInsDiff { ins: Some(i), kind: ObjInsDiffKind::None, ..Default::default() });
     }
     resolve_branches(&mut diff);
-    Ok(ObjSymbolDiff { diff_symbol: None, instructions: diff, match_percent: None })
+    Ok(ObjSymbolDiff { symbol_ref, diff_symbol: None, instructions: diff, match_percent: None })
 }
 
 pub fn diff_code(
@@ -66,11 +66,13 @@ pub fn diff_code(
 
     Ok((
         ObjSymbolDiff {
+            symbol_ref: left_symbol_ref,
             diff_symbol: Some(right_symbol_ref),
             instructions: left_diff,
             match_percent: Some(percent),
         },
         ObjSymbolDiff {
+            symbol_ref: right_symbol_ref,
             diff_symbol: Some(left_symbol_ref),
             instructions: right_diff,
             match_percent: Some(percent),
diff --git a/objdiff-core/src/diff/data.rs b/objdiff-core/src/diff/data.rs
@@ -3,8 +3,8 @@ use std::{
     time::{Duration, Instant},
 };
 
-use anyhow::Result;
-use similar::{capture_diff_slices_deadline, Algorithm};
+use anyhow::{anyhow, Result};
+use similar::{capture_diff_slices_deadline, get_diff_ratio, Algorithm};
 
 use crate::{
     diff::{ObjDataDiff, ObjDataDiffKind, ObjSectionDiff, ObjSymbolDiff},
@@ -22,20 +22,22 @@ pub fn diff_bss_symbol(
     let percent = if left_symbol.size == right_symbol.size { 100.0 } else { 50.0 };
     Ok((
         ObjSymbolDiff {
+            symbol_ref: left_symbol_ref,
             diff_symbol: Some(right_symbol_ref),
             instructions: vec![],
             match_percent: Some(percent),
         },
         ObjSymbolDiff {
+            symbol_ref: right_symbol_ref,
             diff_symbol: Some(left_symbol_ref),
             instructions: vec![],
             match_percent: Some(percent),
         },
     ))
 }
 
-pub fn no_diff_bss_symbol(_obj: &ObjInfo, _symbol_ref: SymbolRef) -> ObjSymbolDiff {
-    ObjSymbolDiff { diff_symbol: None, instructions: vec![], match_percent: Some(0.0) }
+pub fn no_diff_symbol(_obj: &ObjInfo, symbol_ref: SymbolRef) -> ObjSymbolDiff {
+    ObjSymbolDiff { symbol_ref, diff_symbol: None, instructions: vec![], match_percent: None }
 }
 
 pub fn diff_data(
@@ -45,6 +47,7 @@ pub fn diff_data(
     let deadline = Instant::now() + Duration::from_secs(5);
     let ops =
         capture_diff_slices_deadline(Algorithm::Patience, &left.data, &right.data, Some(deadline));
+    let match_percent = get_diff_ratio(&ops, left.data.len(), right.data.len()) * 100.0;
 
     let mut left_diff = Vec::<ObjDataDiff>::new();
     let mut right_diff = Vec::<ObjDataDiff>::new();
@@ -118,14 +121,50 @@ pub fn diff_data(
         ObjSectionDiff {
             symbols: vec![],
             data_diff: left_diff,
-            // TODO
-            match_percent: None,
+            match_percent: Some(match_percent),
         },
         ObjSectionDiff {
             symbols: vec![],
             data_diff: right_diff,
-            // TODO
-            match_percent: None,
+            match_percent: Some(match_percent),
+        },
+    ))
+}
+
+pub fn diff_data_symbol(
+    left_obj: &ObjInfo,
+    right_obj: &ObjInfo,
+    left_symbol_ref: SymbolRef,
+    right_symbol_ref: SymbolRef,
+) -> Result<(ObjSymbolDiff, ObjSymbolDiff)> {
+    let (left_section, left_symbol) = left_obj.section_symbol(left_symbol_ref);
+    let (right_section, right_symbol) = right_obj.section_symbol(right_symbol_ref);
+
+    let left_section = left_section.ok_or_else(|| anyhow!("Data symbol section not found"))?;
+    let right_section = right_section.ok_or_else(|| anyhow!("Data symbol section not found"))?;
+
+    let left_data = &left_section.data[left_symbol.section_address as usize
+        ..(left_symbol.section_address + left_symbol.size) as usize];
+    let right_data = &right_section.data[right_symbol.section_address as usize
+        ..(right_symbol.section_address + right_symbol.size) as usize];
+
+    let deadline = Instant::now() + Duration::from_secs(5);
+    let ops =
+        capture_diff_slices_deadline(Algorithm::Patience, left_data, right_data, Some(deadline));
+    let match_percent = get_diff_ratio(&ops, left_data.len(), right_data.len()) * 100.0;
+
+    Ok((
+        ObjSymbolDiff {
+            symbol_ref: left_symbol_ref,
+            diff_symbol: Some(right_symbol_ref),
+            instructions: vec![],
+            match_percent: Some(match_percent),
+        },
+        ObjSymbolDiff {
+            symbol_ref: right_symbol_ref,
+            diff_symbol: Some(left_symbol_ref),
+            instructions: vec![],
+            match_percent: Some(match_percent),
         },
     ))
 }
diff --git a/objdiff-core/src/diff/mod.rs b/objdiff-core/src/diff/mod.rs
@@ -1,19 +1,19 @@
-mod code;
-mod data;
-pub mod display;
-
 use std::collections::HashSet;
 
 use anyhow::Result;
 
 use crate::{
     diff::{
         code::{diff_code, no_diff_code},
-        data::{diff_bss_symbol, diff_data, no_diff_bss_symbol},
+        data::{diff_bss_symbol, diff_data, diff_data_symbol, no_diff_symbol},
     },
-    obj::{ObjInfo, ObjIns, ObjSectionKind, SymbolRef},
+    obj::{ObjInfo, ObjIns, ObjSection, ObjSectionKind, ObjSymbol, SymbolRef},
 };
 
+mod code;
+mod data;
+pub mod display;
+
 #[derive(Debug, Copy, Clone, Default, Eq, PartialEq, serde::Deserialize, serde::Serialize)]
 pub enum X86Formatter {
     #[default]
@@ -62,6 +62,7 @@ impl ObjSectionDiff {
 
 #[derive(Debug, Clone, Default)]
 pub struct ObjSymbolDiff {
+    pub symbol_ref: SymbolRef,
     pub diff_symbol: Option<SymbolRef>,
     pub instructions: Vec<ObjInsDiff>,
     pub match_percent: Option<f32>,
@@ -142,10 +143,11 @@ impl ObjDiff {
             sections: Vec::with_capacity(obj.sections.len()),
             common: Vec::with_capacity(obj.common.len()),
         };
-        for section in &obj.sections {
+        for (section_idx, section) in obj.sections.iter().enumerate() {
             let mut symbols = Vec::with_capacity(section.symbols.len());
-            for _ in &section.symbols {
+            for (symbol_idx, _) in section.symbols.iter().enumerate() {
                 symbols.push(ObjSymbolDiff {
+                    symbol_ref: SymbolRef { section_idx, symbol_idx },
                     diff_symbol: None,
                     instructions: vec![],
                     match_percent: None,
@@ -162,8 +164,9 @@ impl ObjDiff {
                 match_percent: None,
             });
         }
-        for _ in &obj.common {
+        for (symbol_idx, _) in obj.common.iter().enumerate() {
             result.common.push(ObjSymbolDiff {
+                symbol_ref: SymbolRef { section_idx: obj.sections.len(), symbol_idx },
                 diff_symbol: None,
                 instructions: vec![],
                 match_percent: None,
@@ -184,12 +187,20 @@ impl ObjDiff {
 
     #[inline]
     pub fn symbol_diff(&self, symbol_ref: SymbolRef) -> &ObjSymbolDiff {
-        &self.section_diff(symbol_ref.section_idx).symbols[symbol_ref.symbol_idx]
+        if symbol_ref.section_idx == self.sections.len() {
+            &self.common[symbol_ref.symbol_idx]
+        } else {
+            &self.section_diff(symbol_ref.section_idx).symbols[symbol_ref.symbol_idx]
+        }
     }
 
     #[inline]
     pub fn symbol_diff_mut(&mut self, symbol_ref: SymbolRef) -> &mut ObjSymbolDiff {
-        &mut self.section_diff_mut(symbol_ref.section_idx).symbols[symbol_ref.symbol_idx]
+        if symbol_ref.section_idx == self.sections.len() {
+            &mut self.common[symbol_ref.symbol_idx]
+        } else {
+            &mut self.section_diff_mut(symbol_ref.section_idx).symbols[symbol_ref.symbol_idx]
+        }
     }
 }
 
@@ -247,7 +258,14 @@ pub fn diff_objs(
                         }
                     }
                     ObjSectionKind::Data => {
-                        // TODO diff data symbol
+                        let (left_diff, right_diff) = diff_data_symbol(
+                            left_obj,
+                            right_obj,
+                            left_symbol_ref,
+                            right_symbol_ref,
+                        )?;
+                        *left_out.symbol_diff_mut(left_symbol_ref) = left_diff;
+                        *right_out.symbol_diff_mut(right_symbol_ref) = right_diff;
                     }
                     ObjSectionKind::Bss => {
                         let (left_diff, right_diff) = diff_bss_symbol(
@@ -268,10 +286,9 @@ pub fn diff_objs(
                         *left_out.symbol_diff_mut(left_symbol_ref) =
                             no_diff_code(left_obj, left_symbol_ref, config)?;
                     }
-                    ObjSectionKind::Data => {}
-                    ObjSectionKind::Bss => {
+                    ObjSectionKind::Data | ObjSectionKind::Bss => {
                         *left_out.symbol_diff_mut(left_symbol_ref) =
-                            no_diff_bss_symbol(left_obj, left_symbol_ref);
+                            no_diff_symbol(left_obj, left_symbol_ref);
                     }
                 }
             }
@@ -282,10 +299,9 @@ pub fn diff_objs(
                         *right_out.symbol_diff_mut(right_symbol_ref) =
                             no_diff_code(right_obj, right_symbol_ref, config)?;
                     }
-                    ObjSectionKind::Data => {}
-                    ObjSectionKind::Bss => {
+                    ObjSectionKind::Data | ObjSectionKind::Bss => {
                         *right_out.symbol_diff_mut(right_symbol_ref) =
-                            no_diff_bss_symbol(right_obj, right_symbol_ref);
+                            no_diff_symbol(right_obj, right_symbol_ref);
                     }
                 }
             }
@@ -357,8 +373,8 @@ fn matching_symbols(
             for (symbol_idx, symbol) in section.symbols.iter().enumerate() {
                 let symbol_match = SymbolMatch {
                     left: Some(SymbolRef { section_idx, symbol_idx }),
-                    right: find_symbol(right, &symbol.name, section.kind),
-                    prev: find_symbol(prev, &symbol.name, section.kind),
+                    right: find_symbol(right, symbol, section),
+                    prev: find_symbol(prev, symbol, section),
                     section_kind: section.kind,
                 };
                 matches.push(symbol_match);
@@ -367,6 +383,18 @@ fn matching_symbols(
                 }
             }
         }
+        for (symbol_idx, symbol) in left.common.iter().enumerate() {
+            let symbol_match = SymbolMatch {
+                left: Some(SymbolRef { section_idx: left.sections.len(), symbol_idx }),
+                right: find_common_symbol(right, symbol),
+                prev: find_common_symbol(prev, symbol),
+                section_kind: ObjSectionKind::Bss,
+            };
+            matches.push(symbol_match);
+            if let Some(right) = symbol_match.right {
+                right_used.insert(right);
+            }
+        }
     }
     if let Some(right) = right {
         for (section_idx, section) in right.sections.iter().enumerate() {
@@ -378,29 +406,68 @@ fn matching_symbols(
                 matches.push(SymbolMatch {
                     left: None,
                     right: Some(symbol_ref),
-                    prev: find_symbol(prev, &symbol.name, section.kind),
+                    prev: find_symbol(prev, symbol, section),
                     section_kind: section.kind,
                 });
             }
         }
+        for (symbol_idx, symbol) in right.common.iter().enumerate() {
+            let symbol_ref = SymbolRef { section_idx: right.sections.len(), symbol_idx };
+            if right_used.contains(&symbol_ref) {
+                continue;
+            }
+            matches.push(SymbolMatch {
+                left: None,
+                right: Some(symbol_ref),
+                prev: find_common_symbol(prev, symbol),
+                section_kind: ObjSectionKind::Bss,
+            });
+        }
     }
     Ok(matches)
 }
 
 fn find_symbol(
     obj: Option<&ObjInfo>,
-    name: &str,
-    section_kind: ObjSectionKind,
+    in_symbol: &ObjSymbol,
+    in_section: &ObjSection,
 ) -> Option<SymbolRef> {
-    for (section_idx, section) in obj?.sections.iter().enumerate() {
-        if section.kind != section_kind {
+    let obj = obj?;
+    // Try to find an exact name match
+    for (section_idx, section) in obj.sections.iter().enumerate() {
+        if section.kind != in_section.kind {
             continue;
         }
-        let symbol_idx = match section.symbols.iter().position(|symbol| symbol.name == name) {
-            Some(symbol_idx) => symbol_idx,
-            None => continue,
-        };
-        return Some(SymbolRef { section_idx, symbol_idx });
+        if let Some(symbol_idx) =
+            section.symbols.iter().position(|symbol| symbol.name == in_symbol.name)
+        {
+            return Some(SymbolRef { section_idx, symbol_idx });
+        }
+    }
+    // Match compiler-generated symbols against each other (e.g. @251 -> @60)
+    // If they are at the same address in the same section
+    if in_symbol.name.starts_with('@')
+        && matches!(in_section.kind, ObjSectionKind::Data | ObjSectionKind::Bss)
+    {
+        if let Some((section_idx, section)) =
+            obj.sections.iter().enumerate().find(|(_, s)| s.name == in_section.name)
+        {
+            if let Some(symbol_idx) = section.symbols.iter().position(|symbol| {
+                symbol.address == in_symbol.address && symbol.name.starts_with('@')
+            }) {
+                return Some(SymbolRef { section_idx, symbol_idx });
+            }
+        }
+    }
+    None
+}
+
+fn find_common_symbol(obj: Option<&ObjInfo>, in_symbol: &ObjSymbol) -> Option<SymbolRef> {
+    let obj = obj?;
+    for (symbol_idx, symbol) in obj.common.iter().enumerate() {
+        if symbol.name == in_symbol.name {
+            return Some(SymbolRef { section_idx: obj.sections.len(), symbol_idx });
+        }
     }
     None
 }
diff --git a/objdiff-core/src/obj/mod.rs b/objdiff-core/src/obj/mod.rs
diff --git a/objdiff-gui/src/views/function_diff.rs b/objdiff-gui/src/views/function_diff.rs
diff --git a/objdiff-gui/src/views/symbol_diff.rs b/objdiff-gui/src/views/symbol_diff.rs
