Binary Reading Lab

This repository contains public lab artifacts for source-to-machine reading of a small C double-linked-list sample. The tracked files focus on optimization-level differences, stripped versus symbolized disassembly, GCC versus clang output, Hex-Rays observations, and a small RAII cleanup appendix.

Reading Order

1. Start with src/ to see the C and C++ source samples.
2. Read o_level/o_level_survival_table.md, then the O0 through O3 assembly files under o_level/.
3. Read stripped/recognition_signatures.md, then stripped/stripped_vs_symbolized_diff.md and stripped/stripped_O2_pressure_test.md.
4. Read compiler_delta/gcc_vs_clang_delta.md, then compiler_delta/print_list_reload_alias_test.md.
5. Read hexrays/hex_rays_distortion_o2.md with the machine-code excerpts under hexrays/asm/.
6. Read raii_probe/ as appendix evidence for C++ cleanup lowering.
7. Read TOOLCHAIN.md for the recorded compiler and disassembly environment.

Evidence / Inference / Boundary / OPEN

Evidence:

• The source samples are tracked under src/.
• Optimization-level disassembly is tracked under o_level/.
• Stripped and compiler-contrast disassembly is tracked under stripped/ and compiler_delta/.
• Distilled public notes are tracked beside their evidence directories: o_level/o_level_survival_table.md, stripped/recognition_signatures.md, compiler_delta/gcc_vs_clang_delta.md, and hexrays/hex_rays_distortion_o2.md.
• Hex-Rays-facing assembly excerpts are tracked under hexrays/.

Inference:

• Recognition cues from these artifacts are useful for this own-code sample family and as future transfer candidates.

Boundary:

• Compiler contrast is limited to the recorded GCC 13.3 and clang 18.1 artifacts.
• The RAII probe is appendix evidence for cleanup lowering only.
• The RAII probe is not external validation.

OPEN:

• External transferability testing was deferred.
• General stripped-binary capability is not established by this repository.

Reproducibility

TOOLCHAIN.md records the tool versions referenced by these artifacts. Rebuilds should compare generated disassembly against the tracked files instead of assuming byte-identical output across hosts or compiler revisions.