Implement diffing relocations within data sections (#154)

* Data view: Show data bytes with differing relocations as a diff

* Data view: Show differing relocations on hover

* Symbol list view: Adjust symbol/section match %s when relocations differ

* Improve data reloc diffing logic

* Don't make reloc diffs cause bytes to show as red or green

* Properly detect byte size of each relocation

* Data view: Add context menu for copying relocation target symbols

* Also show already-matching relocations on hover/right click

* Change font color for nonmatching relocs on hover

.github/workflows/build.yaml

@@ -65,7 +65,7 @@ jobs:
     continue-on-error: ${{ matrix.checks == 'advisories' }}
     steps:
       - uses: actions/checkout@v4
-      - uses: EmbarkStudios/cargo-deny-action@v1
+      - uses: EmbarkStudios/cargo-deny-action@v2
         with:
           command: check ${{ matrix.checks }}
 

Cargo.toml

@@ -13,9 +13,9 @@ strip = "debuginfo"
 codegen-units = 1
 
 [workspace.package]
-version = "2.6.0"
+version = "2.7.0"
 authors = ["Luke Street <luke@street.dev>"]
 edition = "2021"
 license = "MIT OR Apache-2.0"
 repository = "https://github.com/encounter/objdiff"
-rust-version = "1.74"
+rust-version = "1.81"

deny.toml

@@ -73,7 +73,6 @@ ignore = [
     #{ id = "RUSTSEC-0000-0000", reason = "you can specify a reason the advisory is ignored" },
     #"a-crate-that-is-yanked@0.1.1", # you can also ignore yanked crate versions if you wish
     #{ crate = "a-crate-that-is-yanked@0.1.1", reason = "you can specify why you are ignoring the yanked crate" },
-    { id = "RUSTSEC-2024-0384", reason = "Unmaintained indirect dependency" },
 ]
 # If this is true, then cargo deny will use the git executable to fetch advisory database.
 # If this is false, then it uses a built-in git library.
@@ -240,7 +239,7 @@ allow-git = []
 
 [sources.allow-org]
 # github.com organizations to allow git sources for
-github = ["notify-rs"]
+github = []
 # gitlab.com organizations to allow git sources for
 gitlab = []
 # bitbucket.org organizations to allow git sources for

objdiff-cli/src/cmd/report.rs

@@ -169,7 +169,10 @@ fn report_object(
         }
         _ => {}
     }
-    let diff_config = diff::DiffObjConfig { relax_reloc_diffs: true, ..Default::default() };
+    let diff_config = diff::DiffObjConfig {
+        function_reloc_diffs: diff::FunctionRelocDiffs::None,
+        ..Default::default()
+    };
     let mapping_config = diff::MappingConfig::default();
     let target = object
         .target_path

objdiff-cli/src/views/function_diff.rs

@@ -3,7 +3,7 @@ use crossterm::event::{Event, KeyCode, KeyEventKind, KeyModifiers, MouseButton,
 use objdiff_core::{
     diff::{
         display::{display_diff, DiffText, HighlightKind},
-        ObjDiff, ObjInsDiffKind, ObjSymbolDiff,
+        FunctionRelocDiffs, ObjDiff, ObjInsDiffKind, ObjSymbolDiff,
     },
     obj::{ObjInfo, ObjSectionKind, ObjSymbol, SymbolRef},
 };
@@ -368,10 +368,15 @@ impl UiView for FunctionDiffUi {
                         self.scroll_x = self.scroll_x.saturating_sub(1);
                         result.redraw = true;
                     }
-                    // Toggle relax relocation diffs
+                    // Cycle through function relocation diff mode
                     KeyCode::Char('x') => {
-                        state.diff_obj_config.relax_reloc_diffs =
-                            !state.diff_obj_config.relax_reloc_diffs;
+                        state.diff_obj_config.function_reloc_diffs =
+                            match state.diff_obj_config.function_reloc_diffs {
+                                FunctionRelocDiffs::None => FunctionRelocDiffs::NameAddress,
+                                FunctionRelocDiffs::NameAddress => FunctionRelocDiffs::DataValue,
+                                FunctionRelocDiffs::DataValue => FunctionRelocDiffs::All,
+                                FunctionRelocDiffs::All => FunctionRelocDiffs::None,
+                            };
                         result.redraw = true;
                         return EventControlFlow::Reload;
                     }

objdiff-core/Cargo.toml

@@ -149,7 +149,7 @@ gimli = { version = "0.31", default-features = false, features = ["read-all"], o
 
 # ppc
 cwdemangle = { version = "1.0", optional = true }
-cwextab = { version = "1.0.2", optional = true }
+cwextab = { version = "1.0", optional = true }
 ppc750cl = { version = "0.3", optional = true }
 
 # mips
@@ -169,10 +169,10 @@ yaxpeax-arch = { version = "0.3", default-features = false, features = ["std"],
 yaxpeax-arm = { version = "0.3", default-features = false, features = ["std"], optional = true }
 
 # build
-notify = { git = "https://github.com/notify-rs/notify", rev = "128bf6230c03d39dbb7f301ff7b20e594e34c3a2", version = "6.1.1", optional = true }
-notify-debouncer-full = { git = "https://github.com/notify-rs/notify", rev = "128bf6230c03d39dbb7f301ff7b20e594e34c3a2", version = "0.4.0", optional = true }
+notify = { version = "8.0.0", optional = true }
+notify-debouncer-full = { version = "0.5.0", optional = true }
 shell-escape = { version = "0.1", optional = true }
-tempfile = { version = "3.14", optional = true }
+tempfile = { version = "3.15", optional = true }
 time = { version = "0.3", optional = true }
 
 [target.'cfg(windows)'.dependencies]

objdiff-core/config-schema.json

@@ -1,11 +1,29 @@
 {
   "properties": [
     {
-      "id": "relaxRelocDiffs",
-      "type": "boolean",
-      "default": false,
-      "name": "Relax relocation diffs",
-      "description": "Ignores differences in relocation targets. (Address, name, etc)"
+      "id": "functionRelocDiffs",
+      "type": "choice",
+      "default": "name_address",
+      "name": "Function relocation diffs",
+      "description": "How relocation targets will be diffed in the function view.",
+      "items": [
+        {
+          "value": "none",
+          "name": "None"
+        },
+        {
+          "value": "name_address",
+          "name": "Name or address"
+        },
+        {
+          "value": "data_value",
+          "name": "Data value"
+        },
+        {
+          "value": "all",
+          "name": "Name or address, data value"
+        }
+      ]
     },
     {
       "id": "spaceBetweenArgs",
@@ -193,7 +211,7 @@
       "id": "general",
       "name": "General",
       "properties": [
-        "relaxRelocDiffs",
+        "functionRelocDiffs",
         "spaceBetweenArgs",
         "combineDataSections"
       ]

objdiff-core/src/arch/arm.rs

@@ -276,6 +276,19 @@ impl ObjArch for ObjArchArm {
     fn display_reloc(&self, flags: RelocationFlags) -> Cow<'static, str> {
         Cow::Owned(format!("<{flags:?}>"))
     }
+
+    fn get_reloc_byte_size(&self, flags: RelocationFlags) -> usize {
+        match flags {
+            RelocationFlags::Elf { r_type } => match r_type {
+                elf::R_ARM_ABS32 => 4,
+                elf::R_ARM_REL32 => 4,
+                elf::R_ARM_ABS16 => 2,
+                elf::R_ARM_ABS8 => 1,
+                _ => 1,
+            },
+            _ => 1,
+        }
+    }
 }
 
 #[derive(Clone, Copy, Debug)]

objdiff-core/src/arch/arm64.rs

@@ -173,6 +173,21 @@ impl ObjArch for ObjArchArm64 {
             _ => Cow::Owned(format!("<{flags:?}>")),
         }
     }
+
+    fn get_reloc_byte_size(&self, flags: RelocationFlags) -> usize {
+        match flags {
+            RelocationFlags::Elf { r_type } => match r_type {
+                elf::R_AARCH64_ABS64 => 8,
+                elf::R_AARCH64_ABS32 => 4,
+                elf::R_AARCH64_ABS16 => 2,
+                elf::R_AARCH64_PREL64 => 8,
+                elf::R_AARCH64_PREL32 => 4,
+                elf::R_AARCH64_PREL16 => 2,
+                _ => 1,
+            },
+            _ => 1,
+        }
+    }
 }
 
 struct DisplayCtx<'a> {

objdiff-core/src/arch/mips.rs

@@ -271,6 +271,17 @@ impl ObjArch for ObjArchMips {
             _ => Cow::Owned(format!("<{flags:?}>")),
         }
     }
+
+    fn get_reloc_byte_size(&self, flags: RelocationFlags) -> usize {
+        match flags {
+            RelocationFlags::Elf { r_type } => match r_type {
+                elf::R_MIPS_16 => 2,
+                elf::R_MIPS_32 => 4,
+                _ => 1,
+            },
+            _ => 1,
+        }
+    }
 }
 
 fn push_reloc(args: &mut Vec<ObjInsArg>, reloc: &ObjReloc) -> Result<()> {

objdiff-core/src/arch/mod.rs

@@ -148,6 +148,8 @@ pub trait ObjArch: Send + Sync {
 
     fn display_reloc(&self, flags: RelocationFlags) -> Cow<'static, str>;
 
+    fn get_reloc_byte_size(&self, flags: RelocationFlags) -> usize;
+
     fn symbol_address(&self, symbol: &Symbol) -> u64 { symbol.address() }
 
     fn guess_data_type(&self, _instruction: &ObjIns) -> Option<DataType> { None }
@@ -156,6 +158,17 @@ pub trait ObjArch: Send + Sync {
         Some(format!("Bytes: {:#x?}", bytes))
     }
 
+    fn display_ins_data(&self, ins: &ObjIns) -> Option<String> {
+        let reloc = ins.reloc.as_ref()?;
+        if reloc.addend >= 0 && reloc.target.bytes.len() > reloc.addend as usize {
+            self.guess_data_type(ins).and_then(|ty| {
+                self.display_data_type(ty, &reloc.target.bytes[reloc.addend as usize..])
+            })
+        } else {
+            None
+        }
+    }
+
     // Downcast methods
     #[cfg(feature = "ppc")]
     fn ppc(&self) -> Option<&ppc::ObjArchPpc> { None }

objdiff-core/src/arch/ppc.rs

@@ -1,6 +1,6 @@
 use std::{
     borrow::Cow,
-    collections::{BTreeMap, HashMap},
+    collections::{BTreeMap, HashMap, HashSet},
 };
 
 use anyhow::{bail, ensure, Result};
@@ -10,7 +10,7 @@ use object::{
     elf, File, Object, ObjectSection, ObjectSymbol, Relocation, RelocationFlags, RelocationTarget,
     Symbol, SymbolKind,
 };
-use ppc750cl::{Argument, InsIter, Opcode, ParsedIns, GPR};
+use ppc750cl::{Argument, Arguments, Ins, InsIter, Opcode, ParsedIns, GPR};
 
 use crate::{
     arch::{DataType, ObjArch, ProcessCodeResult},
@@ -143,6 +143,15 @@ impl ObjArch for ObjArchPpc {
                 }
             }
 
+            if reloc.is_none() {
+                if let Some(fake_pool_reloc) = fake_pool_reloc_for_addr.get(&cur_addr) {
+                    // If this instruction has a fake pool relocation, show it as a fake argument
+                    // at the end of the line.
+                    args.push(ObjInsArg::PlainText(" ".into()));
+                    push_reloc(&mut args, fake_pool_reloc)?;
+                }
+            }
+
             ops.push(ins.op as u16);
             let line = line_info.range(..=cur_addr as u64).last().map(|(_, &b)| b);
             insts.push(ObjIns {
@@ -193,24 +202,23 @@ impl ObjArch for ObjArchPpc {
         }
     }
 
+    fn get_reloc_byte_size(&self, flags: RelocationFlags) -> usize {
+        match flags {
+            RelocationFlags::Elf { r_type } => match r_type {
+                elf::R_PPC_ADDR32 => 4,
+                elf::R_PPC_UADDR32 => 4,
+                _ => 1,
+            },
+            _ => 1,
+        }
+    }
+
     fn guess_data_type(&self, instruction: &ObjIns) -> Option<super::DataType> {
         if instruction.reloc.as_ref().is_some_and(|r| r.target.name.starts_with("@stringBase")) {
             return Some(DataType::String);
         }
 
-        let op = Opcode::from(instruction.op as u8);
-        if let Some(ty) = guess_data_type_from_load_store_inst_op(op) {
-            Some(ty)
-        } else if op == Opcode::Addi {
-            // Assume that any addi instruction that references a local symbol is loading a string.
-            // This hack is not ideal and results in tons of false positives where it will show
-            // garbage strings (e.g. misinterpreting arrays, float literals, etc).
-            // But not all strings are in the @stringBase pool, so the condition above that checks
-            // the target symbol name would miss some.
-            Some(DataType::String)
-        } else {
-            None
-        }
+        guess_data_type_from_load_store_inst_op(Opcode::from(instruction.op as u8))
     }
 
     fn display_data_type(&self, ty: DataType, bytes: &[u8]) -> Option<String> {
@@ -248,6 +256,12 @@ fn push_reloc(args: &mut Vec<ObjInsArg>, reloc: &ObjReloc) -> Result<()> {
             elf::R_PPC_ADDR32 | elf::R_PPC_UADDR32 | elf::R_PPC_REL24 | elf::R_PPC_REL14 => {
                 args.push(ObjInsArg::Reloc);
             }
+            elf::R_PPC_NONE => {
+                // Fake pool relocation.
+                args.push(ObjInsArg::PlainText("<".into()));
+                args.push(ObjInsArg::Reloc);
+                args.push(ObjInsArg::PlainText(">".into()));
+            }
             _ => bail!("Unsupported ELF PPC relocation type {r_type}"),
         },
         flags => bail!("Unsupported PPC relocation kind: {flags:?}"),
@@ -442,16 +456,43 @@ fn get_offset_and_addr_gpr_for_possible_pool_reference(
                 Argument::Simm(simm),
             ) => Some((simm.0, addr_src_gpr, Some(addr_dst_gpr))),
             (
+                // `mr` or `mr.`
                 Opcode::Or,
                 Argument::GPR(addr_dst_gpr),
                 Argument::GPR(addr_src_gpr),
                 Argument::None,
-            ) => Some((0, addr_src_gpr, Some(addr_dst_gpr))), // `mr` or `mr.`
+            ) => Some((0, addr_src_gpr, Some(addr_dst_gpr))),
+            (
+                Opcode::Add,
+                Argument::GPR(addr_dst_gpr),
+                Argument::GPR(addr_src_gpr),
+                Argument::GPR(_offset_gpr),
+            ) => Some((0, addr_src_gpr, Some(addr_dst_gpr))),
             _ => None,
         }
     }
 }
 
+// Remove the relocation we're keeping track of in a particular register when an instruction reuses
+// that register to hold some other value, unrelated to pool relocation addresses.
+fn clear_overwritten_gprs(ins: Ins, gpr_pool_relocs: &mut HashMap<u8, ObjReloc>) {
+    let mut def_args = Arguments::default();
+    ins.parse_defs(&mut def_args);
+    for arg in def_args {
+        if let Argument::GPR(gpr) = arg {
+            if ins.op == Opcode::Lmw {
+                // `lmw` overwrites all registers from rd to r31.
+                // ppc750cl only returns rd itself, so we manually clear the rest of them.
+                for reg in gpr.0..31 {
+                    gpr_pool_relocs.remove(&reg);
+                }
+                break;
+            }
+            gpr_pool_relocs.remove(&gpr.0);
+        }
+    }
+}
+
 // We create a fake relocation for an instruction, vaguely simulating what the actual relocation
 // might have looked like if it wasn't pooled. This is so minimal changes are needed to display
 // pooled accesses vs non-pooled accesses. We set the relocation type to R_PPC_NONE to indicate that
@@ -461,35 +502,51 @@ fn get_offset_and_addr_gpr_for_possible_pool_reference(
 fn make_fake_pool_reloc(offset: i16, cur_addr: u32, pool_reloc: &ObjReloc) -> Option<ObjReloc> {
     let offset_from_pool = pool_reloc.addend + offset as i64;
     let target_address = pool_reloc.target.address.checked_add_signed(offset_from_pool)?;
-    let orig_section_index = pool_reloc.target.orig_section_index?;
-    // We also need to create a fake target symbol to go inside our fake relocation.
-    // This is because we don't have access to list of all symbols in this section, so we can't find
-    // the real symbol yet. Instead we make a placeholder that has the correct `orig_section_index`
-    // and `address` fields, and then later on when this information is displayed to the user, we
-    // can find the real symbol by searching through the object's section's symbols for one that
-    // contains this address.
-    let fake_target_symbol = ObjSymbol {
-        name: "".to_string(),
-        demangled_name: None,
-        address: target_address,
-        section_address: 0,
-        size: 0,
-        size_known: false,
-        kind: Default::default(),
-        flags: Default::default(),
-        orig_section_index: Some(orig_section_index),
-        virtual_address: None,
-        original_index: None,
-        bytes: vec![],
+    let target;
+    let addend;
+    if pool_reloc.target.orig_section_index.is_some() {
+        // If the target symbol is within this current object, then we also need to create a fake
+        // target symbol to go inside our fake relocation. This is because we don't have access to
+        // list of all symbols in this section, so we can't find the real symbol within the pool
+        // based on its address yet. Instead we make a placeholder that has the correct
+        // `orig_section_index` and `address` fields, and then later on when this information is
+        // displayed to the user, we can find the real symbol by searching through the object's
+        // section's symbols for one that contains this address.
+        target = ObjSymbol {
+            name: "".to_string(),
+            demangled_name: None,
+            address: target_address,
+            section_address: 0,
+            size: 0,
+            size_known: false,
+            kind: Default::default(),
+            flags: Default::default(),
+            orig_section_index: pool_reloc.target.orig_section_index,
+            virtual_address: None,
+            original_index: None,
+            bytes: vec![],
+        };
+        // The addend is also fake because we don't know yet if the `target_address` here is the exact
+        // start of the symbol or if it's in the middle of it.
+        addend = 0;
+    } else {
+        // But if the target symbol is in a different object (extern), then we simply copy the pool
+        // relocation's target. This is because it won't be possible to locate the actual symbol
+        // later on based only off of an offset without knowing the object or section it's in. And
+        // doing that for external symbols would also be unnecessary, because when the compiler
+        // generates an instruction that accesses an external "pool" plus some offset, that won't be
+        // a normal pool that contains other symbols within it that we want to display. It will be
+        // something like a vtable for a class with multiple inheritance (for example, dCcD_Cyl in
+        // The Wind Waker). So just showing that vtable symbol plus an addend to represent the
+        // offset into it works fine in this case, no fake symbol to hold an address is necessary.
+        target = pool_reloc.target.clone();
+        addend = pool_reloc.addend;
     };
-    // The addend is also fake because we don't know yet if the `target_address` here is the exact
-    // start of the symbol or if it's in the middle of it.
-    let fake_addend = 0;
     Some(ObjReloc {
         flags: RelocationFlags::Elf { r_type: elf::R_PPC_NONE },
         address: cur_addr as u64,
-        target: fake_target_symbol,
-        addend: fake_addend,
+        target,
+        addend,
     })
 }
 
@@ -497,78 +554,184 @@ fn make_fake_pool_reloc(offset: i16, cur_addr: u32, pool_reloc: &ObjReloc) -> Op
 // of pooled data relocations in them, finding which instructions load data from those addresses,
 // and constructing a mapping of the address of that instruction to a "fake pool relocation" that
 // simulates what that instruction's relocation would look like if data hadn't been pooled.
-// Limitations: This method currently only goes through the instructions in a function in linear
-// order, from start to finish. It does *not* follow any branches. This means that it could have
-// false positives or false negatives in determining which relocation is currently loaded in which
-// register at any given point in the function, as control flow is not respected.
-// There are currently no known examples of this method producing inaccurate results in reality, but
-// if examples are found, it may be possible to update this method to also follow all branches so
-// that it produces more accurate results.
+// This method tries to follow the function's proper control flow. It keeps track of a queue of
+// states it hasn't traversed yet, where each state holds an instruction address and a HashMap of
+// which registers hold which pool relocations at that point.
+// When a conditional or unconditional branch is encountered, the destination of the branch is added
+// to the queue. Conditional branches will traverse both the path where the branch is taken and the
+// one where it's not. Unconditional branches only follow the branch, ignoring any code immediately
+// after the branch instruction.
+// Limitations: This method cannot read jump tables. This is because the jump tables are located in
+// the .data section, but ObjArch.process_code only has access to the .text section. In order to
+// work around this limitation and avoid completely missing most code inside switch statements that
+// use jump tables, we instead guess that any parts of a function we missed were switch cases, and
+// traverse them as if the last `bctr` before that address had branched there. This should be fairly
+// accurate in practice - in testing the only instructions it seems to miss are double branches that
+// the compiler generates in error which can never be reached during normal execution anyway.
 fn generate_fake_pool_reloc_for_addr_mapping(
-    address: u64,
+    func_address: u64,
     code: &[u8],
     relocations: &[ObjReloc],
 ) -> HashMap<u32, ObjReloc> {
-    let mut active_pool_relocs = HashMap::new();
+    let mut visited_ins_addrs = HashSet::new();
     let mut pool_reloc_for_addr = HashMap::new();
-    for (cur_addr, ins) in InsIter::new(code, address as u32) {
-        let simplified = ins.simplified();
-        let reloc = relocations.iter().find(|r| (r.address as u32 & !3) == cur_addr);
+    let mut ins_iters_with_gpr_state =
+        vec![(InsIter::new(code, func_address as u32), HashMap::new())];
+    let mut gpr_state_at_bctr = BTreeMap::new();
+    while let Some((ins_iter, mut gpr_pool_relocs)) = ins_iters_with_gpr_state.pop() {
+        for (cur_addr, ins) in ins_iter {
+            if visited_ins_addrs.contains(&cur_addr) {
+                // Avoid getting stuck in an infinite loop when following looping branches.
+                break;
+            }
+            visited_ins_addrs.insert(cur_addr);
 
-        if let Some(reloc) = reloc {
-            // This instruction has a real relocation, so it may be a pool load we want to keep
-            // track of.
-            let args = &simplified.args;
-            match (ins.op, args[0], args[1], args[2]) {
-                (
-                    Opcode::Addi,
-                    Argument::GPR(addr_dst_gpr),
-                    Argument::GPR(_addr_src_gpr),
-                    Argument::Simm(_simm),
-                ) => {
-                    active_pool_relocs.insert(addr_dst_gpr.0, reloc.clone()); // `lis` + `addi`
+            let simplified = ins.simplified();
+
+            // First handle traversing the function's control flow.
+            let mut branch_dest = None;
+            for arg in simplified.args_iter() {
+                if let Argument::BranchDest(dest) = arg {
+                    let dest = cur_addr.wrapping_add_signed(dest.0);
+                    branch_dest = Some(dest);
+                    break;
                 }
-                (
-                    Opcode::Ori,
-                    Argument::GPR(addr_dst_gpr),
-                    Argument::GPR(_addr_src_gpr),
-                    Argument::Uimm(_uimm),
-                ) => {
-                    active_pool_relocs.insert(addr_dst_gpr.0, reloc.clone()); // `lis` + `ori`
-                }
-                (Opcode::B, _, _, _) => {
-                    if simplified.mnemonic == "bl" {
-                        // When encountering a function call, clear any active pool relocations from
-                        // the volatile registers (r0, r3-r12), but not the nonvolatile registers.
-                        active_pool_relocs.remove(&0);
-                        for gpr in 3..12 {
-                            active_pool_relocs.remove(&gpr);
+            }
+            if let Some(branch_dest) = branch_dest {
+                if branch_dest >= func_address as u32
+                    && (branch_dest - func_address as u32) < code.len() as u32
+                {
+                    let dest_offset_into_func = branch_dest - func_address as u32;
+                    let dest_code_slice = &code[dest_offset_into_func as usize..];
+                    match ins.op {
+                        Opcode::Bc => {
+                            // Conditional branch.
+                            // Add the branch destination to the queue to do later.
+                            ins_iters_with_gpr_state.push((
+                                InsIter::new(dest_code_slice, branch_dest),
+                                gpr_pool_relocs.clone(),
+                            ));
+                            // Then continue on with the current iterator.
                         }
+                        Opcode::B => {
+                            if simplified.mnemonic != "bl" {
+                                // Unconditional branch.
+                                // Add the branch destination to the queue.
+                                ins_iters_with_gpr_state.push((
+                                    InsIter::new(dest_code_slice, branch_dest),
+                                    gpr_pool_relocs.clone(),
+                                ));
+                                // Break out of the current iterator so we can do the newly added one.
+                                break;
+                            }
+                        }
+                        _ => unreachable!(),
                     }
                 }
-                _ => {}
             }
-        } else if let Some((offset, addr_src_gpr, addr_dst_gpr)) =
-            get_offset_and_addr_gpr_for_possible_pool_reference(ins.op, &simplified)
-        {
-            // This instruction doesn't have a real relocation, so it may be a reference to one of
-            // the already-loaded pools.
-            if let Some(pool_reloc) = active_pool_relocs.get(&addr_src_gpr.0) {
-                if let Some(fake_pool_reloc) = make_fake_pool_reloc(offset, cur_addr, pool_reloc) {
-                    pool_reloc_for_addr.insert(cur_addr, fake_pool_reloc);
+            if let Opcode::Bcctr = ins.op {
+                if simplified.mnemonic == "bctr" {
+                    // Unconditional branch to count register.
+                    // Likely a jump table.
+                    gpr_state_at_bctr.insert(cur_addr, gpr_pool_relocs.clone());
                 }
-                if let Some(addr_dst_gpr) = addr_dst_gpr {
-                    // If the address of the pool relocation got copied into another register, we
-                    // need to keep track of it in that register too as future instructions may
-                    // reference the symbol indirectly via this new register, instead of the
-                    // register the symbol's address was originally loaded into.
-                    // For example, the start of the function might `lis` + `addi` the start of the
-                    // ...data pool into r25, and then later the start of a loop will `addi` r25
-                    // with the offset within the .data section of an array variable into r21.
-                    // Then the body of the loop will `lwzx` one of the array elements from r21.
-                    let mut new_reloc = pool_reloc.clone();
-                    new_reloc.addend += offset as i64;
-                    active_pool_relocs.insert(addr_dst_gpr.0, new_reloc);
+            }
+
+            // Then handle keeping track of which GPR contains which pool relocation.
+            let reloc = relocations.iter().find(|r| (r.address as u32 & !3) == cur_addr);
+            if let Some(reloc) = reloc {
+                // This instruction has a real relocation, so it may be a pool load we want to keep
+                // track of.
+                let args = &simplified.args;
+                match (ins.op, args[0], args[1], args[2]) {
+                    (
+                        // `lis` + `addi`
+                        Opcode::Addi,
+                        Argument::GPR(addr_dst_gpr),
+                        Argument::GPR(_addr_src_gpr),
+                        Argument::Simm(_simm),
+                    ) => {
+                        gpr_pool_relocs.insert(addr_dst_gpr.0, reloc.clone());
+                    }
+                    (
+                        // `lis` + `ori`
+                        Opcode::Ori,
+                        Argument::GPR(addr_dst_gpr),
+                        Argument::GPR(_addr_src_gpr),
+                        Argument::Uimm(_uimm),
+                    ) => {
+                        gpr_pool_relocs.insert(addr_dst_gpr.0, reloc.clone());
+                    }
+                    (Opcode::B, _, _, _) => {
+                        if simplified.mnemonic == "bl" {
+                            // When encountering a function call, clear any active pool relocations from
+                            // the volatile registers (r0, r3-r12), but not the nonvolatile registers.
+                            gpr_pool_relocs.remove(&0);
+                            for gpr in 3..12 {
+                                gpr_pool_relocs.remove(&gpr);
+                            }
+                        }
+                    }
+                    _ => {
+                        clear_overwritten_gprs(ins, &mut gpr_pool_relocs);
+                    }
+                }
+            } else if let Some((offset, addr_src_gpr, addr_dst_gpr)) =
+                get_offset_and_addr_gpr_for_possible_pool_reference(ins.op, &simplified)
+            {
+                // This instruction doesn't have a real relocation, so it may be a reference to one of
+                // the already-loaded pools.
+                if let Some(pool_reloc) = gpr_pool_relocs.get(&addr_src_gpr.0) {
+                    if let Some(fake_pool_reloc) =
+                        make_fake_pool_reloc(offset, cur_addr, pool_reloc)
+                    {
+                        pool_reloc_for_addr.insert(cur_addr, fake_pool_reloc);
+                    }
+                    if let Some(addr_dst_gpr) = addr_dst_gpr {
+                        // If the address of the pool relocation got copied into another register, we
+                        // need to keep track of it in that register too as future instructions may
+                        // reference the symbol indirectly via this new register, instead of the
+                        // register the symbol's address was originally loaded into.
+                        // For example, the start of the function might `lis` + `addi` the start of the
+                        // ...data pool into r25, and then later the start of a loop will `addi` r25
+                        // with the offset within the .data section of an array variable into r21.
+                        // Then the body of the loop will `lwzx` one of the array elements from r21.
+                        let mut new_reloc = pool_reloc.clone();
+                        new_reloc.addend += offset as i64;
+                        gpr_pool_relocs.insert(addr_dst_gpr.0, new_reloc);
+                    } else {
+                        clear_overwritten_gprs(ins, &mut gpr_pool_relocs);
+                    }
+                } else {
+                    clear_overwritten_gprs(ins, &mut gpr_pool_relocs);
+                }
+            } else {
+                clear_overwritten_gprs(ins, &mut gpr_pool_relocs);
+            }
+        }
+
+        // Finally, if we're about to finish the outer loop and don't have any more control flow to
+        // follow, we check if there are any instruction addresses in this function that we missed.
+        // If so, and if there were any `bctr` instructions before those points in this function,
+        // then we try to traverse those missing spots as switch cases.
+        if ins_iters_with_gpr_state.is_empty() {
+            let unseen_addrs = (func_address as u32..func_address as u32 + code.len() as u32)
+                .step_by(4)
+                .filter(|addr| !visited_ins_addrs.contains(addr));
+            for unseen_addr in unseen_addrs {
+                let prev_bctr_gpr_state = gpr_state_at_bctr
+                    .iter()
+                    .filter(|(&addr, _)| addr < unseen_addr)
+                    .min_by_key(|(&addr, _)| addr)
+                    .map(|(_, gpr_state)| gpr_state);
+                if let Some(gpr_pool_relocs) = prev_bctr_gpr_state {
+                    let dest_offset_into_func = unseen_addr - func_address as u32;
+                    let dest_code_slice = &code[dest_offset_into_func as usize..];
+                    ins_iters_with_gpr_state.push((
+                        InsIter::new(dest_code_slice, unseen_addr),
+                        gpr_pool_relocs.clone(),
+                    ));
+                    break;
                 }
             }
         }

objdiff-core/src/arch/x86.rs

@@ -162,6 +162,19 @@ impl ObjArch for ObjArchX86 {
             _ => Cow::Owned(format!("<{flags:?}>")),
         }
     }
+
+    fn get_reloc_byte_size(&self, flags: RelocationFlags) -> usize {
+        match flags {
+            RelocationFlags::Coff { typ } => match typ {
+                pe::IMAGE_REL_I386_DIR16 => 2,
+                pe::IMAGE_REL_I386_REL16 => 2,
+                pe::IMAGE_REL_I386_DIR32 => 4,
+                pe::IMAGE_REL_I386_REL32 => 4,
+                _ => 1,
+            },
+            _ => 1,
+        }
+    }
 }
 
 fn replace_arg(

objdiff-core/src/diff/code.rs

@@ -3,13 +3,17 @@ use std::{cmp::max, collections::BTreeMap};
 use anyhow::{anyhow, Result};
 use similar::{capture_diff_slices_deadline, Algorithm};
 
+use super::FunctionRelocDiffs;
 use crate::{
     arch::ProcessCodeResult,
     diff::{
         DiffObjConfig, ObjInsArgDiff, ObjInsBranchFrom, ObjInsBranchTo, ObjInsDiff, ObjInsDiffKind,
         ObjSymbolDiff,
     },
-    obj::{ObjInfo, ObjInsArg, ObjReloc, ObjSection, ObjSymbol, ObjSymbolFlags, SymbolRef},
+    obj::{
+        ObjInfo, ObjIns, ObjInsArg, ObjReloc, ObjSection, ObjSymbol, ObjSymbolFlags, ObjSymbolKind,
+        SymbolRef,
+    },
 };
 
 pub fn process_code_symbol(
@@ -192,7 +196,7 @@ fn address_eq(left: &ObjReloc, right: &ObjReloc) -> bool {
     left.target.address as i64 + left.addend == right.target.address as i64 + right.addend
 }
 
-fn section_name_eq(
+pub fn section_name_eq(
     left_obj: &ObjInfo,
     right_obj: &ObjInfo,
     left_orig_section_index: usize,
@@ -215,16 +219,19 @@ fn reloc_eq(
     config: &DiffObjConfig,
     left_obj: &ObjInfo,
     right_obj: &ObjInfo,
-    left_reloc: Option<&ObjReloc>,
-    right_reloc: Option<&ObjReloc>,
+    left_ins: Option<&ObjIns>,
+    right_ins: Option<&ObjIns>,
 ) -> bool {
-    let (Some(left), Some(right)) = (left_reloc, right_reloc) else {
+    let (Some(left_ins), Some(right_ins)) = (left_ins, right_ins) else {
+        return false;
+    };
+    let (Some(left), Some(right)) = (&left_ins.reloc, &right_ins.reloc) else {
         return false;
     };
     if left.flags != right.flags {
         return false;
     }
-    if config.relax_reloc_diffs {
+    if config.function_reloc_diffs == FunctionRelocDiffs::None {
         return true;
     }
 
@@ -233,7 +240,13 @@ fn reloc_eq(
         (Some(sl), Some(sr)) => {
             // Match if section and name or address match
             section_name_eq(left_obj, right_obj, *sl, *sr)
-                && (symbol_name_matches || address_eq(left, right))
+                && (config.function_reloc_diffs == FunctionRelocDiffs::DataValue
+                    || symbol_name_matches
+                    || address_eq(left, right))
+                && (config.function_reloc_diffs == FunctionRelocDiffs::NameAddress
+                    || left.target.kind != ObjSymbolKind::Object
+                    || left_obj.arch.display_ins_data(left_ins)
+                        == left_obj.arch.display_ins_data(right_ins))
         }
         (Some(_), None) => false,
         (None, Some(_)) => {
@@ -262,7 +275,7 @@ fn arg_eq(
             ObjInsArg::Arg(r) => l.loose_eq(r),
             // If relocations are relaxed, match if left is a constant and right is a reloc
             // Useful for instances where the target object is created without relocations
-            ObjInsArg::Reloc => config.relax_reloc_diffs,
+            ObjInsArg::Reloc => config.function_reloc_diffs == FunctionRelocDiffs::None,
             _ => false,
         },
         ObjInsArg::Reloc => {
@@ -271,8 +284,8 @@ fn arg_eq(
                     config,
                     left_obj,
                     right_obj,
-                    left_diff.ins.as_ref().and_then(|i| i.reloc.as_ref()),
-                    right_diff.ins.as_ref().and_then(|i| i.reloc.as_ref()),
+                    left_diff.ins.as_ref(),
+                    right_diff.ins.as_ref(),
                 )
         }
         ObjInsArg::BranchDest(_) => match right {
@@ -283,7 +296,7 @@ fn arg_eq(
             }
             // If relocations are relaxed, match if left is a constant and right is a reloc
             // Useful for instances where the target object is created without relocations
-            ObjInsArg::Reloc => config.relax_reloc_diffs,
+            ObjInsArg::Reloc => config.function_reloc_diffs == FunctionRelocDiffs::None,
             _ => false,
         },
     }

objdiff-core/src/diff/data.rs

@@ -1,11 +1,15 @@
-use std::cmp::{max, min, Ordering};
+use std::{
+    cmp::{max, min, Ordering},
+    ops::Range,
+};
 
 use anyhow::{anyhow, Result};
 use similar::{capture_diff_slices_deadline, get_diff_ratio, Algorithm};
 
+use super::code::section_name_eq;
 use crate::{
     diff::{ObjDataDiff, ObjDataDiffKind, ObjSectionDiff, ObjSymbolDiff},
-    obj::{ObjInfo, ObjSection, SymbolRef},
+    obj::{ObjInfo, ObjReloc, ObjSection, ObjSymbolFlags, SymbolRef},
 };
 
 pub fn diff_bss_symbol(
@@ -37,8 +41,110 @@ pub fn no_diff_symbol(_obj: &ObjInfo, symbol_ref: SymbolRef) -> ObjSymbolDiff {
     ObjSymbolDiff { symbol_ref, target_symbol: None, instructions: vec![], match_percent: None }
 }
 
+fn address_eq(left: &ObjReloc, right: &ObjReloc) -> bool {
+    if right.target.size == 0 && left.target.size != 0 {
+        // The base relocation is against a pool but the target relocation isn't.
+        // This can happen in rare cases where the compiler will generate a pool+addend relocation
+        // in the base, but the one detected in the target is direct with no addend.
+        // Just check that the final address is the same so these count as a match.
+        left.target.address as i64 + left.addend == right.target.address as i64 + right.addend
+    } else {
+        // But otherwise, if the compiler isn't using a pool, we're more strict and check that the
+        // target symbol address and relocation addend both match exactly.
+        left.target.address == right.target.address && left.addend == right.addend
+    }
+}
+
+fn reloc_eq(left_obj: &ObjInfo, right_obj: &ObjInfo, left: &ObjReloc, right: &ObjReloc) -> bool {
+    if left.flags != right.flags {
+        return false;
+    }
+
+    let symbol_name_matches = left.target.name == right.target.name;
+    match (&left.target.orig_section_index, &right.target.orig_section_index) {
+        (Some(sl), Some(sr)) => {
+            // Match if section and name+addend or address match
+            section_name_eq(left_obj, right_obj, *sl, *sr)
+                && ((symbol_name_matches && left.addend == right.addend) || address_eq(left, right))
+        }
+        (Some(_), None) => false,
+        (None, Some(_)) => {
+            // Match if possibly stripped weak symbol
+            (symbol_name_matches && left.addend == right.addend)
+                && right.target.flags.0.contains(ObjSymbolFlags::Weak)
+        }
+        (None, None) => symbol_name_matches,
+    }
+}
+
+/// Compares relocations contained with a certain data range.
+/// The ObjDataDiffKind for each diff will either be `None`` (if the relocation matches),
+/// or `Replace` (if a relocation was changed, added, or removed).
+/// `Insert` and `Delete` are not used when a relocation is added or removed to avoid confusing diffs
+/// where it looks like the bytes themselves were changed but actually only the relocations changed.
+fn diff_data_relocs_for_range(
+    left_obj: &ObjInfo,
+    right_obj: &ObjInfo,
+    left: &ObjSection,
+    right: &ObjSection,
+    left_range: Range<usize>,
+    right_range: Range<usize>,
+) -> Vec<(ObjDataDiffKind, Option<ObjReloc>, Option<ObjReloc>)> {
+    let mut diffs = Vec::new();
+    for left_reloc in left.relocations.iter() {
+        if !left_range.contains(&(left_reloc.address as usize)) {
+            continue;
+        }
+        let left_offset = left_reloc.address as usize - left_range.start;
+        let Some(right_reloc) = right.relocations.iter().find(|r| {
+            if !right_range.contains(&(r.address as usize)) {
+                return false;
+            }
+            let right_offset = r.address as usize - right_range.start;
+            right_offset == left_offset
+        }) else {
+            diffs.push((ObjDataDiffKind::Replace, Some(left_reloc.clone()), None));
+            continue;
+        };
+        if reloc_eq(left_obj, right_obj, left_reloc, right_reloc) {
+            diffs.push((
+                ObjDataDiffKind::None,
+                Some(left_reloc.clone()),
+                Some(right_reloc.clone()),
+            ));
+        } else {
+            diffs.push((
+                ObjDataDiffKind::Replace,
+                Some(left_reloc.clone()),
+                Some(right_reloc.clone()),
+            ));
+        }
+    }
+    for right_reloc in right.relocations.iter() {
+        if !right_range.contains(&(right_reloc.address as usize)) {
+            continue;
+        }
+        let right_offset = right_reloc.address as usize - right_range.start;
+        let Some(_) = left.relocations.iter().find(|r| {
+            if !left_range.contains(&(r.address as usize)) {
+                return false;
+            }
+            let left_offset = r.address as usize - left_range.start;
+            left_offset == right_offset
+        }) else {
+            diffs.push((ObjDataDiffKind::Replace, None, Some(right_reloc.clone())));
+            continue;
+        };
+        // No need to check the cases for relocations being deleted or matching again.
+        // They were already handled in the loop over the left relocs.
+    }
+    diffs
+}
+
 /// Compare the data sections of two object files.
 pub fn diff_data_section(
+    left_obj: &ObjInfo,
+    right_obj: &ObjInfo,
     left: &ObjSection,
     right: &ObjSection,
     left_section_diff: &ObjSectionDiff,
@@ -70,6 +176,94 @@ pub fn diff_data_section(
                 ObjDataDiffKind::Replace
             }
         };
+        if kind == ObjDataDiffKind::None {
+            let mut found_any_relocs = false;
+            let mut left_curr_addr = left_range.start;
+            let mut right_curr_addr = right_range.start;
+            for (diff_kind, left_reloc, right_reloc) in diff_data_relocs_for_range(
+                left_obj,
+                right_obj,
+                left,
+                right,
+                left_range.clone(),
+                right_range.clone(),
+            ) {
+                found_any_relocs = true;
+
+                if let Some(left_reloc) = left_reloc {
+                    let left_reloc_addr = left_reloc.address as usize;
+                    if left_reloc_addr > left_curr_addr {
+                        let len = left_reloc_addr - left_curr_addr;
+                        let left_data = &left.data[left_curr_addr..left_reloc_addr];
+                        left_diff.push(ObjDataDiff {
+                            data: left_data[..min(len, left_data.len())].to_vec(),
+                            kind: ObjDataDiffKind::None,
+                            len,
+                            ..Default::default()
+                        });
+                    }
+                    let reloc_diff_len = left_obj.arch.get_reloc_byte_size(left_reloc.flags);
+                    let left_data = &left.data[left_reloc_addr..left_reloc_addr + reloc_diff_len];
+                    left_diff.push(ObjDataDiff {
+                        data: left_data[..min(reloc_diff_len, left_data.len())].to_vec(),
+                        kind: diff_kind,
+                        len: reloc_diff_len,
+                        reloc: Some(left_reloc.clone()),
+                        ..Default::default()
+                    });
+                    left_curr_addr = left_reloc_addr + reloc_diff_len;
+                }
+
+                if let Some(right_reloc) = right_reloc {
+                    let right_reloc_addr = right_reloc.address as usize;
+                    if right_reloc_addr > right_curr_addr {
+                        let len = right_reloc_addr - right_curr_addr;
+                        let right_data = &right.data[right_curr_addr..right_reloc_addr];
+                        right_diff.push(ObjDataDiff {
+                            data: right_data[..min(len, right_data.len())].to_vec(),
+                            kind: ObjDataDiffKind::None,
+                            len,
+                            ..Default::default()
+                        });
+                    }
+                    let reloc_diff_len = right_obj.arch.get_reloc_byte_size(right_reloc.flags);
+                    let right_data =
+                        &right.data[right_reloc_addr..right_reloc_addr + reloc_diff_len];
+                    right_diff.push(ObjDataDiff {
+                        data: right_data[..min(reloc_diff_len, right_data.len())].to_vec(),
+                        kind: diff_kind,
+                        len: reloc_diff_len,
+                        reloc: Some(right_reloc.clone()),
+                        ..Default::default()
+                    });
+                    right_curr_addr = right_reloc_addr + reloc_diff_len;
+                }
+            }
+
+            if found_any_relocs {
+                if left_curr_addr < left_range.end - 1 {
+                    let len = left_range.end - left_curr_addr;
+                    let left_data = &left.data[left_curr_addr..left_range.end];
+                    left_diff.push(ObjDataDiff {
+                        data: left_data[..min(len, left_data.len())].to_vec(),
+                        kind: ObjDataDiffKind::None,
+                        len,
+                        ..Default::default()
+                    });
+                }
+                if right_curr_addr < right_range.end - 1 {
+                    let len = right_range.end - right_curr_addr;
+                    let right_data = &right.data[right_curr_addr..right_range.end];
+                    right_diff.push(ObjDataDiff {
+                        data: right_data[..min(len, right_data.len())].to_vec(),
+                        kind: ObjDataDiffKind::None,
+                        len,
+                        ..Default::default()
+                    });
+                }
+                continue;
+            }
+        }
         let left_data = &left.data[left_range];
         let right_data = &right.data[right_range];
         left_diff.push(ObjDataDiff {
@@ -123,14 +317,19 @@ pub fn diff_data_section(
 
     let (mut left_section_diff, mut right_section_diff) =
         diff_generic_section(left, right, left_section_diff, right_section_diff)?;
+    let all_left_relocs_match =
+        left_diff.iter().all(|d| d.kind == ObjDataDiffKind::None || d.reloc.is_none());
     left_section_diff.data_diff = left_diff;
     right_section_diff.data_diff = right_diff;
-    // Use the highest match percent between two options:
-    // - Left symbols matching right symbols by name
-    // - Diff of the data itself
-    if left_section_diff.match_percent.unwrap_or(-1.0) < match_percent {
-        left_section_diff.match_percent = Some(match_percent);
-        right_section_diff.match_percent = Some(match_percent);
+    if all_left_relocs_match {
+        // Use the highest match percent between two options:
+        // - Left symbols matching right symbols by name
+        // - Diff of the data itself
+        // We only do this when all relocations on the left side match.
+        if left_section_diff.match_percent.unwrap_or(-1.0) < match_percent {
+            left_section_diff.match_percent = Some(match_percent);
+            right_section_diff.match_percent = Some(match_percent);
+        }
     }
     Ok((left_section_diff, right_section_diff))
 }
@@ -147,13 +346,54 @@ pub fn diff_data_symbol(
     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 left_range = left_symbol.section_address as usize
+        ..(left_symbol.section_address + left_symbol.size) as usize;
+    let right_range = right_symbol.section_address as usize
+        ..(right_symbol.section_address + right_symbol.size) as usize;
+    let left_data = &left_section.data[left_range.clone()];
+    let right_data = &right_section.data[right_range.clone()];
+
+    let reloc_diffs = diff_data_relocs_for_range(
+        left_obj,
+        right_obj,
+        left_section,
+        right_section,
+        left_range,
+        right_range,
+    );
 
     let ops = capture_diff_slices_deadline(Algorithm::Patience, left_data, right_data, None);
-    let match_percent = get_diff_ratio(&ops, left_data.len(), right_data.len()) * 100.0;
+    let bytes_match_ratio = get_diff_ratio(&ops, left_data.len(), right_data.len());
+
+    let mut match_ratio = bytes_match_ratio;
+    if !reloc_diffs.is_empty() {
+        let mut total_reloc_bytes = 0;
+        let mut matching_reloc_bytes = 0;
+        for (diff_kind, left_reloc, right_reloc) in reloc_diffs {
+            let reloc_diff_len = match (left_reloc, right_reloc) {
+                (None, None) => unreachable!(),
+                (None, Some(right_reloc)) => right_obj.arch.get_reloc_byte_size(right_reloc.flags),
+                (Some(left_reloc), _) => left_obj.arch.get_reloc_byte_size(left_reloc.flags),
+            };
+            total_reloc_bytes += reloc_diff_len;
+            if diff_kind == ObjDataDiffKind::None {
+                matching_reloc_bytes += reloc_diff_len;
+            }
+        }
+        if total_reloc_bytes > 0 {
+            let relocs_match_ratio = matching_reloc_bytes as f32 / total_reloc_bytes as f32;
+            // Adjust the overall match ratio to include relocation differences.
+            // We calculate it so that bytes that contain a relocation are counted twice: once for the
+            // byte's raw value, and once for its relocation.
+            // e.g. An 8 byte symbol that has 8 matching raw bytes and a single 4 byte relocation that
+            // doesn't match would show as 66% (weighted average of 100% and 0%).
+            match_ratio = ((bytes_match_ratio * (left_data.len() as f32))
+                + (relocs_match_ratio * total_reloc_bytes as f32))
+                / (left_data.len() + total_reloc_bytes) as f32;
+        }
+    }
+
+    let match_percent = match_ratio * 100.0;
 
     Ok((
         ObjSymbolDiff {

objdiff-core/src/diff/mod.rs

@@ -11,7 +11,9 @@ use crate::{
             diff_generic_section, no_diff_symbol,
         },
     },
-    obj::{ObjInfo, ObjIns, ObjSection, ObjSectionKind, ObjSymbol, SymbolRef, SECTION_COMMON},
+    obj::{
+        ObjInfo, ObjIns, ObjReloc, ObjSection, ObjSectionKind, ObjSymbol, SymbolRef, SECTION_COMMON,
+    },
 };
 
 pub mod code;
@@ -85,6 +87,7 @@ pub struct ObjDataDiff {
     pub kind: ObjDataDiffKind,
     pub len: usize,
     pub symbol: String,
+    pub reloc: Option<ObjReloc>,
 }
 
 #[derive(Debug, Copy, Clone, Eq, PartialEq, Default)]
@@ -153,6 +156,7 @@ impl ObjDiff {
                     kind: ObjDataDiffKind::None,
                     len: section.data.len(),
                     symbol: section.name.clone(),
+                    ..Default::default()
                 }],
                 match_percent: None,
             });
@@ -345,6 +349,8 @@ pub fn diff_objs(
                     let left_section_diff = left_out.section_diff(left_section_idx);
                     let right_section_diff = right_out.section_diff(right_section_idx);
                     let (left_diff, right_diff) = diff_data_section(
+                        left_obj,
+                        right_obj,
                         left_section,
                         right_section,
                         left_section_diff,

objdiff-gui/Cargo.toml

@@ -93,4 +93,4 @@ tracing-wasm = "0.2"
 anyhow = "1.0"
 
 [target.'cfg(windows)'.build-dependencies]
-tauri-winres = "0.1"
+tauri-winres = "0.2"

objdiff-gui/src/app_config.rs

@@ -3,8 +3,11 @@ use std::path::PathBuf;
 use eframe::Storage;
 use globset::Glob;
 use objdiff_core::{
-    config::ScratchConfig,
-    diff::{ArmArchVersion, ArmR9Usage, DiffObjConfig, MipsAbi, MipsInstrCategory, X86Formatter},
+    config::{ScratchConfig, SymbolMappings},
+    diff::{
+        ArmArchVersion, ArmR9Usage, DiffObjConfig, FunctionRelocDiffs, MipsAbi, MipsInstrCategory,
+        X86Formatter,
+    },
 };
 
 use crate::app::{AppConfig, ObjectConfig, CONFIG_KEY};
@@ -15,7 +18,7 @@ pub struct AppConfigVersion {
 }
 
 impl Default for AppConfigVersion {
-    fn default() -> Self { Self { version: 2 } }
+    fn default() -> Self { Self { version: 3 } }
 }
 
 /// Deserialize the AppConfig from storage, handling upgrades from older versions.
@@ -23,7 +26,8 @@ pub fn deserialize_config(storage: &dyn Storage) -> Option<AppConfig> {
     let str = storage.get_string(CONFIG_KEY)?;
     match ron::from_str::<AppConfigVersion>(&str) {
         Ok(version) => match version.version {
-            2 => from_str::<AppConfig>(&str),
+            3 => from_str::<AppConfig>(&str),
+            2 => from_str::<AppConfigV2>(&str).map(|c| c.into_config()),
             1 => from_str::<AppConfigV1>(&str).map(|c| c.into_config()),
             _ => {
                 log::warn!("Unknown config version: {}", version.version);
@@ -49,6 +53,119 @@ where T: serde::de::DeserializeOwned {
     }
 }
 
+#[derive(serde::Deserialize, serde::Serialize)]
+pub struct ScratchConfigV2 {
+    #[serde(default)]
+    pub platform: Option<String>,
+    #[serde(default)]
+    pub compiler: Option<String>,
+    #[serde(default)]
+    pub c_flags: Option<String>,
+    #[serde(default)]
+    pub ctx_path: Option<PathBuf>,
+    #[serde(default)]
+    pub build_ctx: Option<bool>,
+    #[serde(default)]
+    pub preset_id: Option<u32>,
+}
+
+impl ScratchConfigV2 {
+    fn into_config(self) -> ScratchConfig {
+        ScratchConfig {
+            platform: self.platform,
+            compiler: self.compiler,
+            c_flags: self.c_flags,
+            ctx_path: self.ctx_path,
+            build_ctx: self.build_ctx,
+            preset_id: self.preset_id,
+        }
+    }
+}
+
+#[derive(serde::Deserialize, serde::Serialize)]
+pub struct ObjectConfigV2 {
+    pub name: String,
+    pub target_path: Option<PathBuf>,
+    pub base_path: Option<PathBuf>,
+    pub reverse_fn_order: Option<bool>,
+    pub complete: Option<bool>,
+    pub scratch: Option<ScratchConfigV2>,
+    pub source_path: Option<String>,
+    #[serde(default)]
+    pub symbol_mappings: SymbolMappings,
+}
+
+impl ObjectConfigV2 {
+    fn into_config(self) -> ObjectConfig {
+        ObjectConfig {
+            name: self.name,
+            target_path: self.target_path,
+            base_path: self.base_path,
+            reverse_fn_order: self.reverse_fn_order,
+            complete: self.complete,
+            scratch: self.scratch.map(|scratch| scratch.into_config()),
+            source_path: self.source_path,
+            symbol_mappings: self.symbol_mappings,
+        }
+    }
+}
+
+#[derive(serde::Deserialize, serde::Serialize)]
+pub struct AppConfigV2 {
+    pub version: u32,
+    #[serde(default)]
+    pub custom_make: Option<String>,
+    #[serde(default)]
+    pub custom_args: Option<Vec<String>>,
+    #[serde(default)]
+    pub selected_wsl_distro: Option<String>,
+    #[serde(default)]
+    pub project_dir: Option<PathBuf>,
+    #[serde(default)]
+    pub target_obj_dir: Option<PathBuf>,
+    #[serde(default)]
+    pub base_obj_dir: Option<PathBuf>,
+    #[serde(default)]
+    pub selected_obj: Option<ObjectConfigV2>,
+    #[serde(default = "bool_true")]
+    pub build_base: bool,
+    #[serde(default)]
+    pub build_target: bool,
+    #[serde(default = "bool_true")]
+    pub rebuild_on_changes: bool,
+    #[serde(default)]
+    pub auto_update_check: bool,
+    #[serde(default)]
+    pub watch_patterns: Vec<Glob>,
+    #[serde(default)]
+    pub recent_projects: Vec<PathBuf>,
+    #[serde(default)]
+    pub diff_obj_config: DiffObjConfigV1,
+}
+
+impl AppConfigV2 {
+    fn into_config(self) -> AppConfig {
+        log::info!("Upgrading configuration from v2");
+        AppConfig {
+            custom_make: self.custom_make,
+            custom_args: self.custom_args,
+            selected_wsl_distro: self.selected_wsl_distro,
+            project_dir: self.project_dir,
+            target_obj_dir: self.target_obj_dir,
+            base_obj_dir: self.base_obj_dir,
+            selected_obj: self.selected_obj.map(|obj| obj.into_config()),
+            build_base: self.build_base,
+            build_target: self.build_target,
+            rebuild_on_changes: self.rebuild_on_changes,
+            auto_update_check: self.auto_update_check,
+            watch_patterns: self.watch_patterns,
+            recent_projects: self.recent_projects,
+            diff_obj_config: self.diff_obj_config.into_config(),
+            ..Default::default()
+        }
+    }
+}
+
 #[derive(serde::Deserialize, serde::Serialize)]
 pub struct ScratchConfigV1 {
     #[serde(default)]
@@ -147,7 +264,11 @@ impl Default for DiffObjConfigV1 {
 impl DiffObjConfigV1 {
     fn into_config(self) -> DiffObjConfig {
         DiffObjConfig {
-            relax_reloc_diffs: self.relax_reloc_diffs,
+            function_reloc_diffs: if self.relax_reloc_diffs {
+                FunctionRelocDiffs::None
+            } else {
+                FunctionRelocDiffs::default()
+            },
             space_between_args: self.space_between_args,
             combine_data_sections: self.combine_data_sections,
             x86_formatter: self.x86_formatter,

objdiff-gui/src/views/data_diff.rs

@@ -1,4 +1,8 @@
-use std::{cmp::min, default::Default, mem::take};
+use std::{
+    cmp::{min, Ordering},
+    default::Default,
+    mem::take,
+};
 
 use egui::{text::LayoutJob, Id, Label, RichText, Sense, Widget};
 use objdiff_core::{
@@ -23,7 +27,88 @@ fn find_section(obj: &ObjInfo, section_name: &str) -> Option<usize> {
     obj.sections.iter().position(|section| section.name == section_name)
 }
 
-fn data_row_ui(ui: &mut egui::Ui, address: usize, diffs: &[ObjDataDiff], appearance: &Appearance) {
+fn data_row_hover_ui(
+    ui: &mut egui::Ui,
+    obj: &ObjInfo,
+    diffs: &[ObjDataDiff],
+    appearance: &Appearance,
+) {
+    ui.scope(|ui| {
+        ui.style_mut().override_text_style = Some(egui::TextStyle::Monospace);
+        ui.style_mut().wrap_mode = Some(egui::TextWrapMode::Extend);
+
+        for diff in diffs {
+            let Some(reloc) = &diff.reloc else {
+                continue;
+            };
+
+            // Use a slightly different font color for nonmatching relocations so they stand out.
+            let color = match diff.kind {
+                ObjDataDiffKind::None => appearance.highlight_color,
+                _ => appearance.replace_color,
+            };
+
+            // TODO: Most of this code is copy-pasted from ins_hover_ui.
+            // Try to separate this out into a shared function.
+            ui.label(format!("Relocation type: {}", obj.arch.display_reloc(reloc.flags)));
+            let addend_str = match reloc.addend.cmp(&0i64) {
+                Ordering::Greater => format!("+{:x}", reloc.addend),
+                Ordering::Less => format!("-{:x}", -reloc.addend),
+                _ => "".to_string(),
+            };
+            ui.colored_label(color, format!("Name: {}{}", reloc.target.name, addend_str));
+            if let Some(orig_section_index) = reloc.target.orig_section_index {
+                if let Some(section) =
+                    obj.sections.iter().find(|s| s.orig_index == orig_section_index)
+                {
+                    ui.colored_label(color, format!("Section: {}", section.name));
+                }
+                ui.colored_label(
+                    color,
+                    format!("Address: {:x}{}", reloc.target.address, addend_str),
+                );
+                ui.colored_label(color, format!("Size: {:x}", reloc.target.size));
+                if reloc.addend >= 0 && reloc.target.bytes.len() > reloc.addend as usize {}
+            } else {
+                ui.colored_label(color, "Extern".to_string());
+            }
+        }
+    });
+}
+
+fn data_row_context_menu(ui: &mut egui::Ui, diffs: &[ObjDataDiff]) {
+    ui.scope(|ui| {
+        ui.style_mut().override_text_style = Some(egui::TextStyle::Monospace);
+        ui.style_mut().wrap_mode = Some(egui::TextWrapMode::Extend);
+
+        for diff in diffs {
+            let Some(reloc) = &diff.reloc else {
+                continue;
+            };
+
+            // TODO: This code is copy-pasted from ins_context_menu.
+            // Try to separate this out into a shared function.
+            if let Some(name) = &reloc.target.demangled_name {
+                if ui.button(format!("Copy \"{name}\"")).clicked() {
+                    ui.output_mut(|output| output.copied_text.clone_from(name));
+                    ui.close_menu();
+                }
+            }
+            if ui.button(format!("Copy \"{}\"", reloc.target.name)).clicked() {
+                ui.output_mut(|output| output.copied_text.clone_from(&reloc.target.name));
+                ui.close_menu();
+            }
+        }
+    });
+}
+
+fn data_row_ui(
+    ui: &mut egui::Ui,
+    obj: Option<&ObjInfo>,
+    address: usize,
+    diffs: &[ObjDataDiff],
+    appearance: &Appearance,
+) {
     if diffs.iter().any(|d| d.kind != ObjDataDiffKind::None) {
         ui.painter().rect_filled(ui.available_rect_before_wrap(), 0.0, ui.visuals().faint_bg_color);
     }
@@ -94,9 +179,13 @@ fn data_row_ui(ui: &mut egui::Ui, address: usize, diffs: &[ObjDataDiff], appeara
             write_text(text.as_str(), base_color, &mut job, appearance.code_font.clone());
         }
     }
-    Label::new(job).sense(Sense::click()).ui(ui);
-    //     .on_hover_ui_at_pointer(|ui| ins_hover_ui(ui, ins))
-    //     .context_menu(|ui| ins_context_menu(ui, ins));
+
+    let response = Label::new(job).sense(Sense::click()).ui(ui);
+    if let Some(obj) = obj {
+        response
+            .on_hover_ui_at_pointer(|ui| data_row_hover_ui(ui, obj, diffs, appearance))
+            .context_menu(|ui| data_row_context_menu(ui, diffs));
+    }
 }
 
 fn split_diffs(diffs: &[ObjDataDiff]) -> Vec<Vec<ObjDataDiff>> {
@@ -117,8 +206,8 @@ fn split_diffs(diffs: &[ObjDataDiff]) -> Vec<Vec<ObjDataDiff>> {
                 },
                 kind: diff.kind,
                 len,
-                // TODO
-                symbol: String::new(),
+                symbol: String::new(), // TODO
+                reloc: diff.reloc.clone(),
             });
             remaining_in_row -= len;
             cur_len += len;
@@ -161,6 +250,8 @@ fn data_table_ui(
     right_ctx: Option<SectionDiffContext<'_>>,
     config: &Appearance,
 ) -> Option<()> {
+    let left_obj = left_ctx.map(|ctx| ctx.obj);
+    let right_obj = right_ctx.map(|ctx| ctx.obj);
     let left_section = left_ctx
         .and_then(|ctx| ctx.section_index.map(|i| (&ctx.obj.sections[i], &ctx.diff.sections[i])));
     let right_section = right_ctx
@@ -187,11 +278,11 @@ fn data_table_ui(
         row.col(|ui| {
             if column == 0 {
                 if let Some(left_diffs) = &left_diffs {
-                    data_row_ui(ui, address, &left_diffs[i], config);
+                    data_row_ui(ui, left_obj, address, &left_diffs[i], config);
                 }
             } else if column == 1 {
                 if let Some(right_diffs) = &right_diffs {
-                    data_row_ui(ui, address, &right_diffs[i], config);
+                    data_row_ui(ui, right_obj, address, &right_diffs[i], config);
                 }
             }
         });

objdiff-gui/src/views/function_diff.rs

@@ -149,12 +149,8 @@ fn ins_hover_ui(
                     appearance.highlight_color,
                     format!("Size: {:x}", reloc.target.size),
                 );
-                if reloc.addend >= 0 && reloc.target.bytes.len() > reloc.addend as usize {
-                    if let Some(s) = obj.arch.guess_data_type(ins).and_then(|ty| {
-                        obj.arch.display_data_type(ty, &reloc.target.bytes[reloc.addend as usize..])
-                    }) {
-                        ui.colored_label(appearance.highlight_color, s);
-                    }
+                if let Some(s) = obj.arch.display_ins_data(ins) {
+                    ui.colored_label(appearance.highlight_color, s);
                 }
             } else {
                 ui.colored_label(appearance.highlight_color, "Extern".to_string());