AIKIDO-2026-74974

TL;DR

The library parses debug information formats such as ELF and Breakpad from potentially untrusted files. When locating ELF sections it assumes section names begin with a . and slices section data using raw file offsets and sizes without bounds checks, so a malformed section name or an invalid offset/size can trigger a panic or out-of-bounds read. When parsing Breakpad objects it also follows inline records without any nesting-depth limit, so deeply nested inline data can cause excessive processing. The fix skips sections with unexpected names, uses bounds-checked slicing, and enforces a maximum inline nesting depth.