From ceaa8acac65dee0584bc419a17d719d384ae6b41 Mon Sep 17 00:00:00 2001
From: Chris Fallin <chris@cfallin.org>
Date: Fri, 10 Feb 2023 20:55:45 -0800
Subject: [PATCH] Add addr2line usage to attempt to get source-loc information

---
 Cargo.toml       |  1 +
 src/frontend.rs  | 39 ++++++++++++++++++++++++++++++++++++++-
 src/ir.rs        |  2 ++
 src/ir/module.rs |  6 +++++-
 4 files changed, 46 insertions(+), 2 deletions(-)

diff --git a/Cargo.toml b/Cargo.toml
index 756070a..225f23d 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -18,3 +18,4 @@ smallvec = "1.7"
 rayon = "1.5"
 lazy_static = "1.4"
 libc = "0.2"
+addr2line = "0.19"
diff --git a/src/frontend.rs b/src/frontend.rs
index 57b4dbb..7a5bf97 100644
--- a/src/frontend.rs
+++ b/src/frontend.rs
@@ -7,6 +7,7 @@ use crate::errors::FrontendError;
 use crate::ir::*;
 use crate::op_traits::{op_inputs, op_outputs};
 use crate::ops::Operator;
+use addr2line::gimli;
 use anyhow::{bail, Result};
 use fxhash::{FxHashMap, FxHashSet};
 use log::trace;
@@ -19,10 +20,13 @@ pub fn wasm_to_ir(bytes: &[u8]) -> Result<Module<'_>> {
     let mut module = Module::with_orig_bytes(bytes);
     let parser = Parser::new(0);
     let mut next_func = 0;
+    let mut dwarf = gimli::Dwarf::default();
     for payload in parser.parse_all(bytes) {
         let payload = payload?;
-        handle_payload(&mut module, payload, &mut next_func)?;
+        handle_payload(&mut module, payload, &mut next_func, &mut dwarf)?;
     }
+    let debug_map = DebugMap::from_dwarf(dwarf, &mut module.debug);
+    module.debug_map = debug_map;
 
     Ok(module)
 }
@@ -57,6 +61,7 @@ fn handle_payload<'a>(
     module: &mut Module<'a>,
     payload: Payload<'a>,
     next_func: &mut usize,
+    dwarf: &mut gimli::Dwarf<gimli::EndianSlice<'a, gimli::LittleEndian>>,
 ) -> Result<()> {
     trace!("Wasm parser item: {:?}", payload);
     match payload {
@@ -213,6 +218,38 @@ fn handle_payload<'a>(
                 }
             }
         }
+        Payload::CustomSection(reader) if reader.name() == ".debug_info" => {
+            dwarf.debug_info = gimli::DebugInfo::new(reader.data(), gimli::LittleEndian);
+        }
+        Payload::CustomSection(reader) if reader.name() == ".debug_abbrev" => {
+            dwarf.debug_abbrev = gimli::DebugAbbrev::new(reader.data(), gimli::LittleEndian);
+        }
+        Payload::CustomSection(reader) if reader.name() == ".debug_addr" => {
+            dwarf.debug_addr =
+                gimli::DebugAddr::from(gimli::EndianSlice::new(reader.data(), gimli::LittleEndian));
+        }
+        Payload::CustomSection(reader) if reader.name() == ".debug_aranges" => {
+            dwarf.debug_aranges = gimli::DebugAranges::new(reader.data(), gimli::LittleEndian);
+        }
+        Payload::CustomSection(reader) if reader.name() == ".debug_line" => {
+            dwarf.debug_line = gimli::DebugLine::new(reader.data(), gimli::LittleEndian);
+        }
+        Payload::CustomSection(reader) if reader.name() == ".debug_line_str" => {
+            dwarf.debug_line_str = gimli::DebugLineStr::new(reader.data(), gimli::LittleEndian);
+        }
+        Payload::CustomSection(reader) if reader.name() == ".debug_str" => {
+            dwarf.debug_str = gimli::DebugStr::new(reader.data(), gimli::LittleEndian);
+        }
+        Payload::CustomSection(reader) if reader.name() == ".debug_str_offsets" => {
+            dwarf.debug_str_offsets = gimli::DebugStrOffsets::from(gimli::EndianSlice::new(
+                reader.data(),
+                gimli::LittleEndian,
+            ));
+        }
+        Payload::CustomSection(reader) if reader.name() == ".debug_types" => {
+            dwarf.debug_types = gimli::DebugTypes::new(reader.data(), gimli::LittleEndian);
+        }
+        Payload::CustomSection(_) => {}
         Payload::CodeSectionStart { .. } => {}
         Payload::Version { .. } => {}
         Payload::ElementSection(reader) => {
diff --git a/src/ir.rs b/src/ir.rs
index 5fafc10..c7d10ad 100644
--- a/src/ir.rs
+++ b/src/ir.rs
@@ -69,3 +69,5 @@ mod value;
 pub use value::*;
 mod display;
 pub use display::*;
+mod debug;
+pub use debug::*;
diff --git a/src/ir/module.rs b/src/ir/module.rs
index 077248c..ae2e75c 100644
--- a/src/ir/module.rs
+++ b/src/ir/module.rs
@@ -1,6 +1,6 @@
 use super::{Func, FuncDecl, Global, Memory, ModuleDisplay, Signature, Table, Type};
 use crate::entity::{EntityRef, EntityVec};
-use crate::ir::FunctionBody;
+use crate::ir::{Debug, DebugMap, FunctionBody};
 use crate::{backend, frontend};
 use anyhow::Result;
 
@@ -15,6 +15,8 @@ pub struct Module<'a> {
     pub exports: Vec<Export>,
     pub memories: EntityVec<Memory, MemoryData>,
     pub start_func: Option<Func>,
+    pub debug: Debug,
+    pub debug_map: DebugMap,
 }
 
 #[derive(Clone, Debug, PartialEq, Eq, PartialOrd, Ord, Hash)]
@@ -137,6 +139,8 @@ impl<'a> Module<'a> {
             exports: vec![],
             memories: EntityVec::default(),
             start_func: None,
+            debug: Debug::default(),
+            debug_map: DebugMap::default(),
         }
     }
 }