Add component support to addr2line (#1312)

Previously this didn't take into account components so components with
multiple modules internally wouldn't work with `addr2line`.

Author: alexcrichton

src/bin/wasm-tools/addr2line.rs

@@ -1,10 +1,11 @@
 use addr2line::{Context, LookupResult};
-use anyhow::{bail, Context as _, Result};
+use anyhow::{anyhow, bail, Context as _, Result};
 use gimli::EndianSlice;
 use std::collections::HashMap;
 use std::io::Write;
+use std::ops::Range;
 use std::u64;
-use wasmparser::{Parser, Payload};
+use wasmparser::{Encoding, Parser, Payload};
 
 /// Translate a WebAssembly address to a filename and line number using DWARF
 /// debugging information.
@@ -40,6 +41,12 @@ pub struct Opts {
     code_section_relative: bool,
 }
 
+struct Module<'a> {
+    range: Range<u64>,
+    code_start: Option<u64>,
+    custom_sections: HashMap<&'a str, &'a [u8]>,
+}
+
 impl Opts {
     pub fn general_opts(&self) -> &wasm_tools::GeneralOpts {
         self.io.general_opts()
@@ -48,54 +55,60 @@ impl Opts {
     pub fn run(&self) -> Result<()> {
         let wasm = self.io.parse_input_wasm()?;
 
-        let (code_start, custom_sections) = self
+        let modules = self
             .parse_custom_sections(&wasm)
             .context("failed to parse input and read custom sections")?;
-
-        let dwarf = gimli::Dwarf::load(|id| -> Result<_> {
-            let data = custom_sections.get(id.name()).copied().unwrap_or(&[]);
-            Ok(EndianSlice::new(data, gimli::LittleEndian))
-        })?;
-        let cx = Context::from_dwarf(dwarf)
-            .context("failed to create addr2line dwarf mapping context")?;
-
         let mut output = self.io.output_writer()?;
 
         for addr in self.addresses.iter() {
-            self.addr2line(&addr, code_start, &cx, &mut output)
+            self.addr2line(&addr, &modules, &mut output)
                 .with_context(|| format!("failed to find frames for `{addr}`"))?;
         }
 
         Ok(())
     }
 
-    fn parse_custom_sections<'a>(
-        &self,
-        wasm: &'a [u8],
-    ) -> Result<(Option<u64>, HashMap<&'a str, &'a [u8]>)> {
-        let mut ret = HashMap::new();
-        let mut code_start = None;
+    fn parse_custom_sections<'a>(&self, wasm: &'a [u8]) -> Result<Vec<Module<'a>>> {
+        let mut ret = Vec::new();
+        let mut cur_module = None;
         for payload in Parser::new(0).parse_all(wasm) {
             match payload? {
+                Payload::Version {
+                    encoding: Encoding::Module,
+                    range,
+                    ..
+                } => {
+                    assert!(cur_module.is_none());
+                    cur_module = Some(Module {
+                        range: range.start as u64..0,
+                        code_start: None,
+                        custom_sections: HashMap::new(),
+                    });
+                }
+
                 Payload::CustomSection(s) => {
-                    ret.insert(s.name(), s.data());
+                    if let Some(cur) = &mut cur_module {
+                        cur.custom_sections.insert(s.name(), s.data());
+                    }
                 }
                 Payload::CodeSectionStart { range, .. } => {
-                    code_start = Some(range.start as u64);
+                    assert!(cur_module.is_some());
+                    cur_module.as_mut().unwrap().code_start = Some(range.start as u64);
+                }
+
+                Payload::End(offset) => {
+                    if let Some(mut module) = cur_module.take() {
+                        module.range.end = offset as u64;
+                        ret.push(module);
+                    }
                 }
                 _ => {}
             }
         }
-        Ok((code_start, ret))
+        Ok(ret)
     }
 
-    fn addr2line(
-        &self,
-        addr: &str,
-        code_start: Option<u64>,
-        cx: &Context<EndianSlice<gimli::LittleEndian>>,
-        out: &mut dyn Write,
-    ) -> Result<()> {
+    fn addr2line(&self, addr: &str, modules: &[Module<'_>], out: &mut dyn Write) -> Result<()> {
         // Support either `0x` or `@` prefixes for hex addresses since 0x is
         // standard and @ is used by wasmprinter (and web browsers I think?)
         let addr = if let Some(hex) = addr.strip_prefix("0x").or_else(|| addr.strip_prefix("@")) {
@@ -104,12 +117,28 @@ impl Opts {
             addr.parse()?
         };
 
+        let module = modules
+            .iter()
+            .find(|module| module.range.start <= addr && addr <= module.range.end)
+            .ok_or_else(|| anyhow!("no module found which contains this address"))?;
+
+        let dwarf = gimli::Dwarf::load(|id| -> Result<_> {
+            let data = module
+                .custom_sections
+                .get(id.name())
+                .copied()
+                .unwrap_or(&[]);
+            Ok(EndianSlice::new(data, gimli::LittleEndian))
+        })?;
+        let cx = Context::from_dwarf(dwarf)
+            .context("failed to create addr2line dwarf mapping context")?;
+
         // Addresses in DWARF are relative to the start of the text section, so
         // factor that in here.
         let text_relative_addr = if self.code_section_relative {
             addr
         } else {
-            match code_start {
+            match module.code_start {
                 Some(start) => addr
                     .checked_sub(start)
                     .context("address is before the beginning of the text section")?,