NanoTag: Systems Support for Efficient Byte-Granular Overflow Detection on ARM MTE (Artifact)

This is the artifact for the paper NanoTag: Systems Support for Efficient Byte-Granular Overflow Detection on ARM MTE in S&P 2026.

The latest artifact can be found here: https://github.com/ice-rlab/NanoTag.

NanoTag is the first low-overhead technique to detect byte-granular buffer overflows in real hardware. Buffer overflows are the root cause of most software vulnerabilities. ARM's Memory Tagging Extension (MTE) aims to prevent such buffer overflows in hardware by ensuring that a pointer's tag matches the tag of any 16-byte memory the pointer accesses, sanitizing memory accesses. Unfortunately, such a 16-byte tag granularity prevents MTE from sanitizing many memory accesses, exposing a large surface (87% for SPEC CPU 2017) of bypassing MTE checks. To address this limitation, NanoTag enables byte-granular overflow detection via software checks, controlled by a sampling knob to explicitly balance bug-detection capability and performance overhead.