Skip to content

addr2line: detect clang version and dispatch to modern or legacy resolver#4973

Open
TianlongLiang wants to merge 1 commit into
bytecodealliance:mainfrom
TianlongLiang:addr2line_enhance_1
Open

addr2line: detect clang version and dispatch to modern or legacy resolver#4973
TianlongLiang wants to merge 1 commit into
bytecodealliance:mainfrom
TianlongLiang:addr2line_enhance_1

Conversation

@TianlongLiang

Copy link
Copy Markdown
Contributor

Refactor the wasm symbolicator to make its toolchain workarounds explicit. addr2line.py now detects the wasi-sdk's clang major version at startup and routes per-frame address resolution to one of:

  • resolve_address_modern (clang >= 22, wasi-sdk 33+): single llvm-symbolizer call per address. The symbolizer handles address-to-name correctly on this clang.

  • resolve_address_legacy (clang < 22 or unknown): llvm-symbolizer call + llvm-dwarfdump --lookup overlay to fix the outermost function name. Older clang versions emit wasm DWARF that confuses llvm-symbolizer's address-to-name resolver for some addresses (e.g., reports 'recurse' as 'free'). --lookup goes through a different DWARF traversal path that handles them.

The dispatch decision is logged once to stderr so users can see which path was taken without polluting stdout. Stdout is byte- identical between paths for non-buggy inputs.

Also includes accumulated improvements:

  • --mode flag {interp,aot,fast-interp} for different runtime offset conventions (interp post-advance, aot at-instruction-start, fast-interp's transformed in-memory bytecode)
  • Inline frame annotation "(inlined into )" for clarity
  • llvm-symbolizer preferred over llvm-addr2line for column info
  • Fallback for offset=0 (trap at function entry; frame_ip not captured)
  • Last-resort function-index name fallback when DWARF lacks PC ranges

…lver

Refactor the wasm symbolicator to make its toolchain workarounds
explicit. addr2line.py now detects the wasi-sdk's clang major version
at startup and routes per-frame address resolution to one of:

- resolve_address_modern (clang >= 22, wasi-sdk 33+):
    single llvm-symbolizer call per address. The symbolizer handles
    address-to-name correctly on this clang.

- resolve_address_legacy (clang < 22 or unknown):
    llvm-symbolizer call + llvm-dwarfdump --lookup overlay to fix
    the outermost function name. Older clang versions emit wasm DWARF
    that confuses llvm-symbolizer's address-to-name resolver for some
    addresses (e.g., reports 'recurse' as 'free'). --lookup goes
    through a different DWARF traversal path that handles them.

The dispatch decision is logged once to stderr so users can see
which path was taken without polluting stdout. Stdout is byte-
identical between paths for non-buggy inputs.

Also includes accumulated improvements:
- --mode flag {interp,aot,fast-interp} for different runtime offset
  conventions (interp post-advance, aot at-instruction-start,
  fast-interp's transformed in-memory bytecode)
- Inline frame annotation "(inlined into <next>)" for clarity
- llvm-symbolizer preferred over llvm-addr2line for column info
- Fallback for offset=0 (trap at function entry; frame_ip not captured)
- Last-resort function-index name fallback when DWARF lacks PC ranges
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant