feat(wasm): support for wasm (#166)

Author: mitsuhiko

symbolic-cabi/include/symbolic.h

@@ -3,7 +3,7 @@
 #ifndef SYMBOLIC_H_INCLUDED
 #define SYMBOLIC_H_INCLUDED
 
-/* Generated with cbindgen:0.14.2 */
+/* Generated with cbindgen:0.14.3 */
 
 /* Warning, this file is autogenerated. Do not modify this manually. */
 
@@ -33,6 +33,7 @@ enum SymbolicErrorCode {
   SYMBOLIC_ERROR_CODE_OBJECT_ERROR_BAD_PDB_OBJECT = 2105,
   SYMBOLIC_ERROR_CODE_OBJECT_ERROR_BAD_PE_OBJECT = 2106,
   SYMBOLIC_ERROR_CODE_OBJECT_ERROR_BAD_SOURCE_BUNDLE = 2107,
+  SYMBOLIC_ERROR_CODE_OBJECT_ERROR_BAD_WASM_OBJECT = 2108,
   SYMBOLIC_ERROR_CODE_DWARF_ERROR_UNKNOWN = 2200,
   SYMBOLIC_ERROR_CODE_DWARF_ERROR_INVALID_UNIT_REF = 2201,
   SYMBOLIC_ERROR_CODE_DWARF_ERROR_INVALID_FILE_REF = 2202,

symbolic-cabi/src/core.rs

@@ -159,6 +159,7 @@ pub enum SymbolicErrorCode {
     ObjectErrorBadPdbObject = 2105,
     ObjectErrorBadPeObject = 2106,
     ObjectErrorBadSourceBundle = 2107,
+    ObjectErrorBadWasmObject = 2108,
     DwarfErrorUnknown = 2200,
     DwarfErrorInvalidUnitRef = 2201,
     DwarfErrorInvalidFileRef = 2202,
@@ -262,6 +263,7 @@ impl SymbolicErrorCode {
                     ObjectError::MachO(_) => SymbolicErrorCode::ObjectErrorBadMachOObject,
                     ObjectError::Pdb(_) => SymbolicErrorCode::ObjectErrorBadPdbObject,
                     ObjectError::Pe(_) => SymbolicErrorCode::ObjectErrorBadPeObject,
+                    ObjectError::Wasm(_) => SymbolicErrorCode::ObjectErrorBadWasmObject,
                     ObjectError::Dwarf(ref e) => match e {
                         DwarfError::InvalidUnitRef(_) => {
                             SymbolicErrorCode::DwarfErrorInvalidUnitRef

symbolic-common/src/types.rs

@@ -90,6 +90,8 @@ pub enum CpuFamily {
     Mips64 = 8,
     /// ILP32 ABI on 64-bit ARM.
     Arm64_32 = 9,
+    /// Virtual WASM architecture.
+    Wasm = 10,
 }
 
 impl CpuFamily {
@@ -111,6 +113,7 @@ impl CpuFamily {
     pub fn pointer_size(self) -> Option<usize> {
         match self {
             CpuFamily::Unknown => None,
+            CpuFamily::Wasm => Some(4),
             CpuFamily::Amd64
             | CpuFamily::Arm64
             | CpuFamily::Ppc64
@@ -139,6 +142,7 @@ impl CpuFamily {
     /// ```
     pub fn instruction_alignment(self) -> Option<u64> {
         match self {
+            CpuFamily::Wasm => Some(4),
             CpuFamily::Arm32 => Some(2),
             CpuFamily::Arm64 | CpuFamily::Arm64_32 => Some(4),
             CpuFamily::Ppc32 | CpuFamily::Mips32 | CpuFamily::Mips64 => Some(4),
@@ -174,6 +178,7 @@ impl CpuFamily {
             CpuFamily::Arm32 | CpuFamily::Arm64 | CpuFamily::Arm64_32 => Some("pc"),
             CpuFamily::Ppc32 | CpuFamily::Ppc64 => Some("srr0"),
             CpuFamily::Mips32 | CpuFamily::Mips64 => Some("pc"),
+            CpuFamily::Wasm => None,
             CpuFamily::Unknown => None,
         }
     }
@@ -280,6 +285,7 @@ pub enum Arch {
     Arm64_32 = 901,
     Arm64_32V8 = 902,
     Arm64_32Unknown = 999,
+    Wasm = 1001,
 }
 
 impl Arch {
@@ -325,6 +331,7 @@ impl Arch {
             901 => Arch::Arm64_32,
             902 => Arch::Arm64_32V8,
             999 => Arch::Arm64_32Unknown,
+            1001 => Arch::Wasm,
             _ => Arch::Unknown,
         }
     }
@@ -361,6 +368,7 @@ impl Arch {
             Arch::Mips => CpuFamily::Mips32,
             Arch::Mips64 => CpuFamily::Mips64,
             Arch::Arm64_32 | Arch::Arm64_32V8 | Arch::Arm64_32Unknown => CpuFamily::Arm64_32,
+            Arch::Wasm => CpuFamily::Wasm,
         }
     }
 
@@ -384,6 +392,7 @@ impl Arch {
     pub fn name(self) -> &'static str {
         match self {
             Arch::Unknown => "unknown",
+            Arch::Wasm => "wasm",
             Arch::X86 => "x86",
             Arch::X86Unknown => "x86_unknown",
             Arch::Amd64 => "x86_64",
@@ -491,6 +500,9 @@ impl str::FromStr for Arch {
             "x86-64" => Arch::Amd64,
             "arm-64" => Arch::Arm64,
 
+            // wasm extensions
+            "wasm" => Arch::Wasm,
+
             _ => return Err(UnknownArchError),
         })
     }

symbolic-debuginfo/Cargo.toml

@@ -41,7 +41,9 @@ smallvec = "1.2.0"
 symbolic-common = { version = "7.5.0", path = "../symbolic-common" }
 thiserror = "1.0.20"
 zip = "0.5.2"
+walrus = "0.18.0"
+wasmparser = "0.68.0"
 
 [dev-dependencies]
-insta = "1.1.0"
+insta = "1.3.0"
 symbolic-testutils = { path = "../symbolic-testutils" }

symbolic-debuginfo/src/base.rs

@@ -156,6 +156,8 @@ pub enum FileFormat {
     Pe,
     /// Source code bundle ZIP.
     SourceBundle,
+    /// WASM container.
+    Wasm,
 }
 
 impl FileFormat {
@@ -169,6 +171,7 @@ impl FileFormat {
             FileFormat::Pdb => "pdb",
             FileFormat::Pe => "pe",
             FileFormat::SourceBundle => "sourcebundle",
+            FileFormat::Wasm => "wasm",
         }
     }
 }
@@ -190,6 +193,7 @@ impl FromStr for FileFormat {
             "pdb" => FileFormat::Pdb,
             "pe" => FileFormat::Pe,
             "sourcebundle" => FileFormat::SourceBundle,
+            "wasm" => FileFormat::Wasm,
             _ => return Err(UnknownFileFormatError),
         })
     }

symbolic-debuginfo/src/dwarf.rs

@@ -43,6 +43,14 @@ type IncompleteLineNumberProgram<'a> = gimli::read::IncompleteLineProgram<Slice<
 type LineNumberProgramHeader<'a> = gimli::read::LineProgramHeader<Slice<'a>>;
 type LineProgramFileEntry<'a> = gimli::read::FileEntry<Slice<'a>>;
 
+/// This applies the offset to the address.
+///
+/// This function does not panic but would wrap around if too large or small
+/// numbers are passed.
+fn offset(addr: u64, offset: i64) -> u64 {
+    (addr as i64).wrapping_sub(offset as i64) as u64
+}
+
 /// An error handling [`DWARF`](trait.Dwarf.html) debugging information.
 #[non_exhaustive]
 #[derive(Debug, Error)]
@@ -536,7 +544,7 @@ impl<'d, 'a> DwarfUnit<'d, 'a> {
             if let Some((first, rows)) = rows.split_first() {
                 let mut last_file = first.file_index;
                 let mut last_info = LineInfo {
-                    address: range.begin - self.inner.info.load_address,
+                    address: offset(range.begin, self.inner.info.address_offset),
                     size: first.size.map(|s| s + first.address - range.begin),
                     file: self.resolve_file(first.file_index).unwrap_or_default(),
                     line: first.line.unwrap_or(0),
@@ -560,7 +568,7 @@ impl<'d, 'a> DwarfUnit<'d, 'a> {
 
                     last_file = row.file_index;
                     last_info = LineInfo {
-                        address: row.address - self.inner.info.load_address,
+                        address: offset(row.address, self.inner.info.address_offset),
                         size: row.size,
                         file: self.resolve_file(row.file_index).unwrap_or_default(),
                         line,
@@ -569,7 +577,7 @@ impl<'d, 'a> DwarfUnit<'d, 'a> {
 
                 // Fix the size of the last line
                 if let Some(size) = last_info.size.as_mut() {
-                    *size = range.end - self.inner.info.load_address - last_info.address;
+                    *size = offset(range.end, self.inner.info.address_offset) - last_info.address;
                 }
 
                 lines.push(last_info);
@@ -659,7 +667,7 @@ impl<'d, 'a> DwarfUnit<'d, 'a> {
                 continue;
             }
 
-            let function_address = range_buf[0].begin - self.inner.info.load_address;
+            let function_address = offset(range_buf[0].begin, self.inner.info.address_offset);
             let function_size = range_buf[range_buf.len() - 1].end - range_buf[0].begin;
             let function_end = function_address + function_size;
 
@@ -715,8 +723,8 @@ impl<'d, 'a> DwarfUnit<'d, 'a> {
 
                     let mut index = 0;
                     for range in range_buf.iter() {
-                        let range_begin = range.begin - self.inner.info.load_address;
-                        let range_end = range.end - self.inner.info.load_address;
+                        let range_begin = offset(range.begin, self.inner.info.address_offset);
+                        let range_end = offset(range.end, self.inner.info.address_offset);
 
                         // Check if there is a line record covering the start of this range,
                         // otherwise insert a new record pointing to the correct call location.
@@ -952,7 +960,7 @@ struct DwarfInfo<'data> {
     headers: Vec<CompilationUnitHeader<'data>>,
     units: Vec<LazyCell<Option<Unit<'data>>>>,
     symbol_map: SymbolMap<'data>,
-    load_address: u64,
+    address_offset: i64,
     kind: ObjectKind,
 }
 
@@ -969,7 +977,7 @@ impl<'d> DwarfInfo<'d> {
     pub fn parse(
         sections: &'d DwarfSections<'d>,
         symbol_map: SymbolMap<'d>,
-        load_address: u64,
+        address_offset: i64,
         kind: ObjectKind,
     ) -> Result<Self, DwarfError> {
         let inner = gimli::read::Dwarf {
@@ -998,7 +1006,7 @@ impl<'d> DwarfInfo<'d> {
             headers,
             units,
             symbol_map,
-            load_address,
+            address_offset,
             kind,
         })
     }
@@ -1074,7 +1082,7 @@ impl fmt::Debug for DwarfInfo<'_> {
         f.debug_struct("DwarfInfo")
             .field("headers", &self.headers)
             .field("symbol_map", &self.symbol_map)
-            .field("load_address", &self.load_address)
+            .field("address_offset", &self.address_offset)
             .finish()
     }
 }
@@ -1122,15 +1130,15 @@ impl<'data> DwarfDebugSession<'data> {
     pub fn parse<D>(
         dwarf: &D,
         symbol_map: SymbolMap<'data>,
-        load_address: u64,
+        address_offset: i64,
         kind: ObjectKind,
     ) -> Result<Self, DwarfError>
     where
         D: Dwarf<'data>,
     {
         let sections = DwarfSections::from_dwarf(dwarf)?;
         let cell = SelfCell::try_new(Box::new(sections), |sections| {
-            DwarfInfo::parse(unsafe { &*sections }, symbol_map, load_address, kind)
+            DwarfInfo::parse(unsafe { &*sections }, symbol_map, address_offset, kind)
         })?;
 
         Ok(DwarfDebugSession { cell })

symbolic-debuginfo/src/elf.rs

@@ -241,7 +241,7 @@ impl<'data> ElfObject<'data> {
     /// [`has_debug_info`](struct.ElfObject.html#method.has_debug_info).
     pub fn debug_session(&self) -> Result<DwarfDebugSession<'data>, DwarfError> {
         let symbols = self.symbol_map();
-        DwarfDebugSession::parse(self, symbols, self.load_address(), self.kind())
+        DwarfDebugSession::parse(self, symbols, self.load_address() as i64, self.kind())
     }
 
     /// Determines whether this object contains stack unwinding information.

symbolic-debuginfo/src/lib.rs

@@ -46,6 +46,7 @@ pub mod macho;
 pub mod pdb;
 pub mod pe;
 pub mod sourcebundle;
+pub mod wasm;
 
 pub use crate::base::*;
 pub use crate::object::*;

symbolic-debuginfo/src/macho.rs

@@ -217,7 +217,7 @@ impl<'d> MachObject<'d> {
     /// [`has_debug_info`](struct.MachObject.html#method.has_debug_info).
     pub fn debug_session(&self) -> Result<DwarfDebugSession<'d>, DwarfError> {
         let symbols = self.symbol_map();
-        DwarfDebugSession::parse(self, symbols, self.load_address(), self.kind())
+        DwarfDebugSession::parse(self, symbols, self.load_address() as i64, self.kind())
     }
 
     /// Determines whether this object contains stack unwinding information.