Add map config for generating symbols/splits

Useful for extracting information from
map files that aren't fully well-formed,
such as ones from existing decompilation
projects.

src/cmd/map.rs

@@ -1,12 +1,15 @@
-use std::path::PathBuf;
+use std::{fs::DirBuilder, path::PathBuf};
 
 use anyhow::{bail, ensure, Result};
 use argp::FromArgs;
 use cwdemangle::{demangle, DemangleOptions};
+use tracing::error;
 
 use crate::util::{
+    config::{write_splits_file, write_symbols_file},
     file::map_file,
-    map::{process_map, SymbolEntry, SymbolRef},
+    map::{create_obj, process_map, SymbolEntry, SymbolRef},
+    split::update_splits,
 };
 
 #[derive(FromArgs, PartialEq, Debug)]
@@ -22,6 +25,7 @@ pub struct Args {
 enum SubCommand {
     Entries(EntriesArgs),
     Symbol(SymbolArgs),
+    Config(ConfigArgs),
 }
 
 #[derive(FromArgs, PartialEq, Eq, Debug)]
@@ -48,10 +52,23 @@ pub struct SymbolArgs {
     symbol: String,
 }
 
+#[derive(FromArgs, PartialEq, Eq, Debug)]
+/// Generates project configuration files from a map. (symbols.txt, splits.txt)
+#[argp(subcommand, name = "config")]
+pub struct ConfigArgs {
+    #[argp(positional)]
+    /// path to input map
+    map_file: PathBuf,
+    #[argp(positional)]
+    /// output directory for symbols.txt and splits.txt
+    out_dir: PathBuf,
+}
+
 pub fn run(args: Args) -> Result<()> {
     match args.command {
         SubCommand::Entries(c_args) => entries(c_args),
         SubCommand::Symbol(c_args) => symbol(c_args),
+        SubCommand::Config(c_args) => config(c_args),
     }
 }
 
@@ -160,3 +177,18 @@ fn symbol(args: SymbolArgs) -> Result<()> {
     println!("\n");
     Ok(())
 }
+
+fn config(args: ConfigArgs) -> Result<()> {
+    let file = map_file(&args.map_file)?;
+    log::info!("Processing map...");
+    let entries = process_map(&mut file.as_reader(), None, None)?;
+    let mut obj = create_obj(&entries)?;
+    if let Err(e) = update_splits(&mut obj, None, false) {
+        error!("Failed to update splits: {}", e)
+    }
+    DirBuilder::new().recursive(true).create(&args.out_dir)?;
+    write_symbols_file(args.out_dir.join("symbols.txt"), &obj, None)?;
+    write_splits_file(args.out_dir.join("splits.txt"), &obj, false, None)?;
+    log::info!("Done!");
+    Ok(())
+}

src/obj/mod.rs

@@ -13,7 +13,7 @@ use std::{
 use anyhow::{anyhow, bail, ensure, Result};
 use objdiff_core::obj::split_meta::SplitMeta;
 pub use relocations::{ObjReloc, ObjRelocKind, ObjRelocations};
-pub use sections::{ObjSection, ObjSectionKind, ObjSections};
+pub use sections::{ObjSection, ObjSectionKind, ObjSections, section_kind_for_section};
 pub use splits::{ObjSplit, ObjSplits};
 pub use symbols::{
     best_match_for_reloc, ObjDataKind, ObjSymbol, ObjSymbolFlagSet, ObjSymbolFlags, ObjSymbolKind,

src/obj/sections.rs

@@ -218,7 +218,7 @@ impl ObjSection {
     }
 }
 
-fn section_kind_for_section(section_name: &str) -> Result<ObjSectionKind> {
+pub fn section_kind_for_section(section_name: &str) -> Result<ObjSectionKind> {
     Ok(match section_name {
         ".init" | ".text" | ".dbgtext" | ".vmtext" => ObjSectionKind::Code,
         ".ctors" | ".dtors" | ".rodata" | ".sdata2" | "extab" | "extabindex" | ".BINARY" => {

src/util/map.rs

@@ -18,8 +18,9 @@ use regex::{Captures, Regex};
 
 use crate::{
     obj::{
-        ObjInfo, ObjKind, ObjSplit, ObjSymbol, ObjSymbolFlagSet, ObjSymbolFlags, ObjSymbolKind,
+        section_kind_for_section, ObjArchitecture, ObjInfo, ObjKind, ObjSection, ObjSectionKind,
-        ObjUnit,
+        ObjSections, ObjSplit, ObjSymbol, ObjSymbolFlagSet, ObjSymbolFlags, ObjSymbolKind,
+        ObjSymbols, ObjUnit,
     },
     util::{file::map_file, nested::NestedVec},
 };
@@ -814,6 +815,108 @@ pub fn apply_map(result: &MapInfo, obj: &mut ObjInfo) -> Result<()> {
     Ok(())
 }
 
+pub fn create_obj(result: &MapInfo) -> Result<ObjInfo> {
+    let sections = result
+        .sections
+        .iter()
+        .map(|s| {
+            let name = s.name.clone();
+            let address = s.address as u64;
+            let size = s.size as u64;
+            let file_offset = s.file_offset as u64;
+            let kind = section_kind_for_section(&name).unwrap_or(ObjSectionKind::ReadOnlyData);
+            ObjSection {
+                name,
+                kind,
+                address,
+                size,
+                data: vec![],
+                align: 0,
+                elf_index: 0,
+                relocations: Default::default(),
+                virtual_address: None,
+                file_offset,
+                section_known: true,
+                splits: Default::default(),
+            }
+        })
+        .collect();
+    let mut obj = ObjInfo {
+        kind: ObjKind::Executable,
+        architecture: ObjArchitecture::PowerPc,
+        name: "".to_string(),
+        symbols: ObjSymbols::new(ObjKind::Executable, vec![]),
+        sections: ObjSections::new(ObjKind::Executable, sections),
+        entry: None, // TODO result.entry_point
+        mw_comment: None,
+        split_meta: None,
+        sda2_base: None,
+        sda_base: None,
+        stack_address: None,
+        stack_end: None,
+        db_stack_addr: None,
+        arena_lo: None,
+        arena_hi: None,
+        link_order: vec![],
+        blocked_relocation_sources: Default::default(),
+        blocked_relocation_targets: Default::default(),
+        known_functions: Default::default(),
+        module_id: 0,
+        unresolved_relocations: vec![],
+    };
+
+    // Add section symbols
+    for (section_name, symbol_map) in &result.section_symbols {
+        let (section_index, _) = obj
+            .sections
+            .by_name(section_name)?
+            .ok_or_else(|| anyhow!("Failed to locate section {section_name} from map"))?;
+        for symbol_entry in symbol_map.values().flatten() {
+            add_symbol(&mut obj, symbol_entry, Some(section_index))?;
+        }
+    }
+
+    // Add splits
+    for (section_name, unit_order) in &result.section_units {
+        let (_, section) = obj
+            .sections
+            .iter_mut()
+            .find(|(_, s)| s.name == *section_name)
+            .ok_or_else(|| anyhow!("Failed to locate section '{}'", section_name))?;
+        let mut iter = unit_order.iter().peekable();
+        while let Some((addr, unit)) = iter.next() {
+            let next = iter
+                .peek()
+                .map(|(addr, _)| *addr)
+                .unwrap_or_else(|| (section.address + section.size) as u32);
+            let common = section_name == ".bss"
+                && matches!(result.common_bss_start, Some(start) if *addr >= start);
+            let unit = unit.replace(' ', "/");
+
+            // Disable mw_comment_version for assembly units
+            if unit.ends_with(".s") && !obj.link_order.iter().any(|u| u.name == unit) {
+                obj.link_order.push(ObjUnit {
+                    name: unit.clone(),
+                    autogenerated: false,
+                    comment_version: Some(0),
+                    order: None,
+                });
+            }
+
+            section.splits.push(*addr, ObjSplit {
+                unit,
+                end: next,
+                align: None,
+                common,
+                autogenerated: false,
+                skip: false,
+                rename: None,
+            });
+        }
+    }
+    Ok(obj)
+}
+
 fn add_symbol(obj: &mut ObjInfo, symbol_entry: &SymbolEntry, section: Option<usize>) -> Result<()> {
     let demangled_name = demangle(&symbol_entry.name, &DemangleOptions::default());
     let mut flags: FlagSet<ObjSymbolFlags> = match symbol_entry.visibility {

src/util/split.rs

@@ -758,26 +758,32 @@ fn trim_linker_generated_symbols(obj: &mut ObjInfo) -> Result<()> {
 pub fn update_splits(obj: &mut ObjInfo, common_start: Option<u32>, fill_gaps: bool) -> Result<()> {
     // Create splits for extab and extabindex entries
     if let Some((section_index, section)) = obj.sections.by_name("extabindex")? {
-        let start = SectionAddress::new(section_index, section.address as u32);
+        if !section.data.is_empty() {
-        split_extabindex(obj, start)?;
+            let start = SectionAddress::new(section_index, section.address as u32);
+            split_extabindex(obj, start)?;
+        }
     }
 
     // Create splits for .ctors entries
     if let Some((section_index, section)) = obj.sections.by_name(".ctors")? {
-        let start = SectionAddress::new(section_index, section.address as u32);
+        if !section.data.is_empty() {
-        let end = start + (section.size as u32 - 4);
+            let start = SectionAddress::new(section_index, section.address as u32);
-        split_ctors_dtors(obj, start, end)?;
+            let end = start + (section.size as u32 - 4);
+            split_ctors_dtors(obj, start, end)?;
+        }
     }
 
     // Create splits for .dtors entries
     if let Some((section_index, section)) = obj.sections.by_name(".dtors")? {
-        let mut start = SectionAddress::new(section_index, section.address as u32);
+        if !section.data.is_empty() {
-        let end = start + (section.size as u32 - 4);
+            let mut start = SectionAddress::new(section_index, section.address as u32);
-        if obj.kind == ObjKind::Executable {
+            let end = start + (section.size as u32 - 4);
-            // Skip __destroy_global_chain_reference
+            if obj.kind == ObjKind::Executable {
-            start += 4;
+                // Skip __destroy_global_chain_reference
+                start += 4;
+            }
+            split_ctors_dtors(obj, start, end)?;
         }
-        split_ctors_dtors(obj, start, end)?;
     }
 
     // Remove linker generated symbols from splits
@@ -1352,6 +1358,9 @@ pub fn end_for_section(obj: &ObjInfo, section_index: usize) -> Result<SectionAdd
         .get(section_index)
         .ok_or_else(|| anyhow!("Invalid section index: {}", section_index))?;
     let mut section_end = (section.address + section.size) as u32;
+    if section.data.is_empty() {
+        return Ok(SectionAddress::new(section_index, section_end));
+    }
     // .ctors and .dtors end with a linker-generated null pointer,
     // adjust section size appropriately
     if matches!(section.name.as_str(), ".ctors" | ".dtors")