Allow specifying replacement bytes in dtk extab clean (#103)

* Allow specifying replacement bytes in dtk extab clean

* Simplify extab padding replacement

* Reword log message

* clippy has bad taste

* Don't specify revision number for cwextab

---------

Co-authored-by: Amber Brault <[email protected]>

Author: cadmic

Cargo.lock

@@ -838,7 +838,7 @@ fn load_dol_module(
     };
     if config.clean_extab.unwrap_or(false) {
         log::debug!("Cleaning extab for {}", config.name());
-        clean_extab(&mut obj)?;
+        clean_extab(&mut obj, std::iter::empty())?;
     }
     Ok((obj, object_path))
 }

src/cmd/dol.rs

@@ -30,15 +30,18 @@ enum SubCommand {
 }
 
 #[derive(FromArgs, PartialEq, Eq, Debug)]
-/// Rewrites extab data in a DOL or ELF file, zeroing out any uninitialized padding bytes.
+/// Rewrites extab data in a DOL or ELF file, replacing any uninitialized padding bytes.
 #[argp(subcommand, name = "clean")]
 pub struct CleanArgs {
     #[argp(positional, from_str_fn(native_path))]
-    /// path to input file
+    /// Path to input file
     input: Utf8NativePathBuf,
     #[argp(positional, from_str_fn(native_path))]
-    /// path to output file
+    /// Path to output file
     output: Utf8NativePathBuf,
+    #[argp(option, short = 'p')]
+    /// Data to replace padding bytes with, encoded as a hexadecimal string. If not specified, padding bytes will be zeroed instead.
+    padding: Option<String>,
 }
 
 pub fn run(args: Args) -> Result<()> {
@@ -56,7 +59,13 @@ fn clean_extab(args: CleanArgs) -> Result<()> {
         let name = args.input.file_stem().unwrap_or_default();
         process_dol(file.map()?, name)?
     };
-    let num_cleaned = util::extab::clean_extab(&mut obj)?;
+    let padding: Vec<u8> = match args.padding {
+        None => Vec::new(),
+        Some(padding_str) => {
+            hex::decode(padding_str).context("Failed to decode padding bytes from hex")?
+        }
+    };
+    let num_cleaned = util::extab::clean_extab(&mut obj, padding.iter().copied())?;
     tracing::debug!("Cleaned {num_cleaned} extab symbols");
     let mut out = buf_writer(&args.output)?;
     if is_elf {

src/cmd/extab.rs

@@ -3,7 +3,7 @@ use itertools::Itertools;
 
 use crate::obj::ObjInfo;
 
-pub fn clean_extab(obj: &mut ObjInfo) -> Result<usize> {
+pub fn clean_extab(obj: &mut ObjInfo, mut padding: impl Iterator<Item = u8>) -> Result<usize> {
     let (extab_section_index, extab_section) = obj
         .sections
         .iter_mut()
@@ -27,19 +27,25 @@ pub fn clean_extab(obj: &mut ObjInfo) -> Result<usize> {
         })?;
         let mut updated = false;
         for action in &decoded.exception_actions {
-            let section_offset =
-                (symbol.address - extab_section.address) as usize + action.action_offset as usize;
-            let clean_data = action.get_exaction_bytes(true);
-            let orig_data =
-                &mut extab_section.data[section_offset..section_offset + clean_data.len()];
-            if orig_data != clean_data {
-                updated = true;
+            // Check if the current action has padding
+            if let Some(padding_offset) = action.get_struct_padding_offset() {
+                let index = padding_offset as usize;
+                let section_offset = (symbol.address - extab_section.address) as usize
+                    + action.action_offset as usize;
+                let mut clean_data: Vec<u8> = action.get_exaction_bytes(false);
+                // Write the two padding bytes
+                clean_data[index] = padding.next().unwrap_or(0);
+                clean_data[index + 1] = padding.next().unwrap_or(0);
+
+                let orig_data =
+                    &mut extab_section.data[section_offset..section_offset + clean_data.len()];
                 orig_data.copy_from_slice(&clean_data);
+                updated = true;
             }
         }
         if updated {
             tracing::debug!(
-                "Removed uninitialized bytes in {} (extab {:#010X}..{:#010X})",
+                "Replaced uninitialized bytes in {} (extab {:#010X}..{:#010X})",
                 symbol.name,
                 symbol.address,
                 symbol.address + symbol.size