Debugging: preserve original Wasm bytecode inside of compiled ELF artifact.

crates/environ/src/compile/module_artifacts.rs

@@ -245,6 +245,32 @@ impl<'a> ObjectBuilder<'a> {
         dwarf.push((T::id() as u8, offset..offset + data.len() as u64));
     }
 
+    /// Appends the original Wasm bytecode for one or more core modules as a
+    /// pair of new ELF sections.
+    ///
+    /// `modules` is an iterator of raw Wasm binary slices, one per core
+    /// module, in `StaticModuleIndex` order.
+    pub fn append_wasm_bytecode<'b>(&mut self, modules: impl IntoIterator<Item = &'b [u8]>) {
+        let bytecode_id = self.obj.add_section(
+            self.obj.segment_name(StandardSegment::Data).to_vec(),
+            obj::ELF_WASMTIME_WASM_BYTECODE.as_bytes().to_vec(),
+            SectionKind::ReadOnlyData,
+        );
+        let ends_id = self.obj.add_section(
+            self.obj.segment_name(StandardSegment::Data).to_vec(),
+            obj::ELF_WASMTIME_WASM_BYTECODE_ENDS.as_bytes().to_vec(),
+            SectionKind::ReadOnlyData,
+        );
+        let mut end: u32 = 0;
+        for wasm in modules {
+            self.obj.append_section_data(bytecode_id, wasm, 1);
+            end = end
+                .checked_add(u32::try_from(wasm.len()).expect("module bytecode exceeds 4 GiB"))
+                .expect("total bytecode exceeds 4 GiB");
+            self.obj.append_section_data(ends_id, &end.to_le_bytes(), 4);
+        }
+    }
+
     /// Creates the `ELF_WASMTIME_INFO` section from the given serializable data
     /// structure.
     pub fn serialize_info<T>(&mut self, info: &T)

crates/environ/src/obj.rs

@@ -177,6 +177,24 @@ pub const ELF_NAME_DATA: &'static str = ".name.wasm";
 /// metadata.
 pub const ELF_WASMTIME_DWARF: &str = ".wasmtime.dwarf";
 
+/// This is the name of the section in the final ELF image which contains the
+/// original Wasm bytecode for the module, preserved verbatim to support
+/// debugger access to the source bytecode.
+///
+/// This section is only emitted when the `guest-debug` tunable is enabled at
+/// compile time. Its contents are the concatenated raw bytes of all core
+/// module Wasm binaries in the artifact.
+pub const ELF_WASMTIME_WASM_BYTECODE: &str = ".wasmtime.wasm_bytecode";
+
+/// This is the name of the companion section to [`ELF_WASMTIME_WASM_BYTECODE`]
+/// that stores the end-offset table used to locate individual module bytecodes
+/// within the concatenated data.
+///
+/// The section contains one little-endian `u32` per core module in
+/// the artifact giving the *end* of that module's bytecode in the
+/// concatenated bytecode section above.
+pub const ELF_WASMTIME_WASM_BYTECODE_ENDS: &str = ".wasmtime.wasm_bytecode_ends";
+
 /// Workaround to implement `core::error::Error` until
 /// gimli-rs/object#747 is settled.
 pub struct ObjectCrateErrorWrapper(pub object::Error);

crates/wasmtime/src/compile.rs

@@ -119,6 +119,10 @@ pub(crate) fn build_module_artifacts<T: FinishedObject>(
         dwarf_package,
     )?;
 
+    if tunables.debug_guest {
+        object.append_wasm_bytecode(std::iter::once(wasm));
+    }
+
     let (info, index) = compilation_artifacts.unwrap_as_module_info();
     let types = types.finish();
     object.serialize_info(&(&info, &index, &types));
@@ -181,6 +185,16 @@ pub(crate) fn build_component_artifacts<T: FinishedObject>(
         t.module.needs_gc_heap |= needs_gc_heap
     }
 
+    // Collect bytecode slices here before moving `module_translations` below.
+    let module_wasms = if tunables.debug_guest {
+        module_translations
+            .values()
+            .map(|t| t.wasm)
+            .collect::<Vec<_>>()
+    } else {
+        vec![]
+    };
+
     let mut object = compiler.object(ObjectKind::Component)?;
     engine.append_compiler_info(&mut object)?;
     engine.append_bti(&mut object);
@@ -192,6 +206,11 @@ pub(crate) fn build_component_artifacts<T: FinishedObject>(
         module_translations,
         None, // TODO: Support dwarf packages for components.
     )?;
+
+    if tunables.debug_guest {
+        object.append_wasm_bytecode(module_wasms);
+    }
+
     let (types, ty) = types.finish(&component.component);
 
     let info = CompiledComponentInfo {

crates/wasmtime/src/runtime/code.rs

@@ -10,6 +10,7 @@ use alloc::sync::Arc;
 use core::ops::{Add, Range, Sub};
 use wasmtime_environ::DefinedFuncIndex;
 use wasmtime_environ::ModuleTypes;
+use wasmtime_environ::StaticModuleIndex;
 #[cfg(feature = "component-model")]
 use wasmtime_environ::component::ComponentTypes;
 
@@ -227,6 +228,13 @@ impl EngineCode {
         self.original_code.wasm_dwarf()
     }
 
+    /// Returns the original Wasm bytecode section if preserved in the
+    /// compiled artifact.
+    #[inline]
+    pub fn wasm_bytecode_for_module(&self, module: StaticModuleIndex) -> Option<&[u8]> {
+        self.original_code.wasm_bytecode_for_module(module)
+    }
+
     /// Returns the raw image as bytes (in our internal image format).
     pub fn image(&self) -> &[u8] {
         &self.original_code.mmap()[..]

crates/wasmtime/src/runtime/code_memory.rs

@@ -5,13 +5,14 @@ use crate::prelude::*;
 use crate::runtime::vm::MmapVec;
 use alloc::sync::Arc;
 use core::ops::Range;
-use object::SectionIndex;
 use object::read::elf::SectionTable;
+use object::{LittleEndian, SectionIndex, U32Bytes};
 use object::{
     elf::{FileHeader64, SectionHeader64},
     endian::Endianness,
     read::elf::{FileHeader as _, SectionHeader as _},
 };
+use wasmtime_environ::StaticModuleIndex;
 use wasmtime_environ::{Trap, lookup_trap_code, obj};
 use wasmtime_unwinder::ExceptionTable;
 
@@ -45,6 +46,8 @@ pub struct CodeMemory {
     func_name_data: Range<usize>,
     info_data: Range<usize>,
     wasm_dwarf: Range<usize>,
+    wasm_bytecode: Range<usize>,
+    wasm_bytecode_ends: Range<usize>,
 }
 
 impl Drop for CodeMemory {
@@ -153,6 +156,8 @@ impl CodeMemory {
         let mut func_name_data = 0..0;
         let mut info_data = 0..0;
         let mut wasm_dwarf = 0..0;
+        let mut wasm_bytecode = 0..0;
+        let mut wasm_bytecode_ends = 0..0;
         for section_header in sections.iter() {
             let data = section_header
                 .data(endian, mmap_data)
@@ -212,6 +217,8 @@ impl CodeMemory {
                 obj::ELF_NAME_DATA => func_name_data = range,
                 obj::ELF_WASMTIME_INFO => info_data = range,
                 obj::ELF_WASMTIME_DWARF => wasm_dwarf = range,
+                obj::ELF_WASMTIME_WASM_BYTECODE => wasm_bytecode = range,
+                obj::ELF_WASMTIME_WASM_BYTECODE_ENDS => wasm_bytecode_ends = range,
 
                 #[cfg(feature = "debug-builtins")]
                 ".debug_info" => has_native_debug_info = true,
@@ -269,6 +276,8 @@ impl CodeMemory {
             wasm_dwarf,
             info_data,
             wasm_data,
+            wasm_bytecode,
+            wasm_bytecode_ends,
         })
     }
 
@@ -332,6 +341,17 @@ impl CodeMemory {
         &self.mmap[self.frame_tables_data.clone()]
     }
 
+    /// Returns the concatenated Wasm bytecode section, or an empty slice if
+    /// the artifact was not compiled with `guest-debug` enabled.
+    pub fn wasm_bytecode(&self) -> &[u8] {
+        &self.mmap[self.wasm_bytecode.clone()]
+    }
+
+    /// Returns the Wasm bytecode section end-offset array.
+    pub fn wasm_bytecode_ends(&self) -> &[u8] {
+        &self.mmap[self.wasm_bytecode_ends.clone()]
+    }
+
     /// Returns the contents of the `ELF_WASMTIME_INFO` section, or an empty
     /// slice if it wasn't found.
     #[inline]
@@ -346,6 +366,36 @@ impl CodeMemory {
         &self.mmap[self.trap_data.clone()]
     }
 
+    /// Returns the Wasm bytecode section end-offset for a given core
+    /// module, or `None` if no bytecode is present.
+    ///
+    /// # Panics
+    ///
+    /// Panics if index is out-of-range.
+    fn wasm_bytecode_end_for_module(&self, index: StaticModuleIndex) -> Option<usize> {
+        if self.wasm_bytecode_ends().is_empty() {
+            return None;
+        }
+        let ends = self.wasm_bytecode_ends();
+        let count = ends.len() / core::mem::size_of::<u32>();
+        let (ends, _) = object::slice_from_bytes::<U32Bytes<LittleEndian>>(ends, count)
+            .expect("Invalid alignment of `ends` section");
+        let index = usize::try_from(index.as_u32()).unwrap();
+        Some(usize::try_from(ends[index].get(LittleEndian)).unwrap())
+    }
+
+    /// Returns the Wasm bytecode for the a core module in this
+    /// artifact, or `None` if bytecode was not preserved.
+    pub(crate) fn wasm_bytecode_for_module(&self, index: StaticModuleIndex) -> Option<&[u8]> {
+        let start = if index.as_u32() == 0 {
+            0
+        } else {
+            self.wasm_bytecode_end_for_module(StaticModuleIndex::from_u32(index.as_u32() - 1))?
+        };
+        let end = self.wasm_bytecode_end_for_module(index)?;
+        Some(&self.wasm_bytecode()[start..end])
+    }
+
     /// Publishes the internal ELF image to be ready for execution.
     ///
     /// This method can only be when the image is not published (its

crates/wasmtime/src/runtime/instantiate.rs

@@ -261,6 +261,12 @@ impl CompiledModule {
     pub fn has_address_map(&self) -> bool {
         !self.engine_code.address_map_data().is_empty()
     }
+
+    /// Returns the original Wasm bytecode for this module, if it is available.
+    pub fn bytecode(&self) -> Option<&[u8]> {
+        self.engine_code
+            .wasm_bytecode_for_module(self.module.module_index)
+    }
 }
 
 #[cfg(feature = "addr2line")]

crates/wasmtime/src/runtime/module.rs

@@ -688,6 +688,18 @@ impl Module {
         Some(&module.strings[name])
     }
 
+    /// Returns the original Wasm bytecode for this module, if it is
+    /// available.
+    ///
+    /// Bytecode is only retained when the [`Engine`] was configured with
+    /// `guest-debug` support enabled (see [`Config::guest_debug`]). Returns
+    /// `None` when the module was compiled without that option.
+    ///
+    /// [`Config::guest_debug`]: crate::Config::guest_debug
+    pub fn debug_bytecode(&self) -> Option<&[u8]> {
+        self.compiled_module().bytecode()
+    }
+
     /// Returns the list of imports that this [`Module`] has and must be
     /// satisfied.
     ///

tests/all/debug.rs

@@ -1116,3 +1116,87 @@ async fn invalidated_frame_handles_in_dropped_future() -> wasmtime::Result<()> {
 
     Ok(())
 }
+
+#[test]
+#[cfg_attr(miri, ignore)]
+fn module_bytecode() -> wasmtime::Result<()> {
+    let wasm = wat::parse_str(
+        r#"
+        (module
+            (func (export "add") (param i32 i32) (result i32)
+                local.get 0
+                local.get 1
+                i32.add
+            )
+        )
+        "#,
+    )
+    .unwrap();
+
+    let mut config = Config::default();
+    config.guest_debug(true);
+    let engine = Engine::new(&config)?;
+    let module = Module::new(&engine, &wasm)?;
+
+    assert_eq!(module.debug_bytecode(), Some(&wasm[..]));
+
+    Ok(())
+}
+
+#[test]
+#[cfg_attr(miri, ignore)]
+fn module_bytecode_absent_without_debug() -> wasmtime::Result<()> {
+    let wasm = wat::parse_str("(module)").unwrap();
+
+    let mut config = Config::default();
+    config.guest_debug(false);
+    let engine = Engine::new(&config)?;
+    let module = Module::new(&engine, &wasm)?;
+
+    assert_eq!(module.debug_bytecode(), None);
+
+    Ok(())
+}
+
+#[test]
+#[cfg_attr(miri, ignore)]
+fn component_bytecode() -> wasmtime::Result<()> {
+    use wasmtime::component::{Component, Linker};
+
+    // Build the bytecode for each core module by compiling them
+    // standalone.
+    let m1_body = r#"(func (export "f1") (result i32) i32.const 42)"#;
+    let m2_body = r#"(func (export "f2") (result i32) i32.const 99)"#;
+    let m1_wasm = wat::parse_str(&format!("(module $m1 {m1_body})")).unwrap();
+    let m2_wasm = wat::parse_str(&format!("(module $m2 {m2_body})")).unwrap();
+
+    // Build a component that embeds both core modules inline.
+    let component_wasm = wat::parse_str(&format!(
+        r#"(component
+               (core module $m1 {m1_body})
+               (core instance $i1 (instantiate (module $m1)))
+               (core module $m2 {m2_body})
+               (core instance $i2 (instantiate (module $m2))))
+        "#,
+    ))
+    .unwrap();
+
+    let mut config = Config::default();
+    config.guest_debug(true);
+    let engine = Engine::new(&config)?;
+
+    let component = Component::new(&engine, &component_wasm)?;
+    let linker: Linker<()> = Linker::new(&engine);
+    let mut store = Store::new(&engine, ());
+    linker.instantiate(&mut store, &component)?;
+
+    let modules = store.debug_all_modules();
+    assert_eq!(modules.len(), 2);
+
+    // Modules should be registered in offset order. The API doesn't
+    // guarantee this, but this suffices for a test.
+    assert_eq!(modules[0].debug_bytecode().unwrap(), &m1_wasm[..]);
+    assert_eq!(modules[1].debug_bytecode().unwrap(), &m2_wasm[..]);
+
+    Ok(())
+}