CVE-2026-42482: Hashcat Buffer Overflow Vulnerability

CVE-2026-42482 Overview

CVE-2026-42482 is a stack-based buffer overflow in mangle_to_hex_lower() and mangle_to_hex_upper() within src/rp_cpu.c in hashcat v7.1.2. The flaw allows an attacker to trigger a denial of service or potentially execute arbitrary code. Exploitation requires a crafted rule file or use of the -j or -k rule options with password candidates of 128 or more characters. The vulnerability stems from a bounds check that fails to account for the 2x expansion that occurs when password bytes are converted to hexadecimal representation. The issue is tracked under [CWE-787] and [CWE-121].

Critical Impact
A crafted rule file processed by hashcat 7.1.2 can corrupt the stack and lead to arbitrary code execution in the context of the user running the tool.

Affected Products

hashcat 7.1.2
Deployments using -j or -k rule options with long password candidates
Distributions and CI pipelines bundling hashcat 7.1.2

Discovery Timeline

2026-05-01 - CVE-2026-42482 published to NVD
2026-05-01 - Last updated in NVD database

Technical Details for CVE-2026-42482

Vulnerability Analysis

The vulnerability resides in hashcat's rule-based candidate mutation engine. The mangle_to_hex_lower() and mangle_to_hex_upper() functions in src/rp_cpu.c convert each byte of a password candidate into two hexadecimal characters. The bounds check enforced before writing to the destination buffer compares against the input length rather than the doubled output length. When a candidate of 128 or more bytes is processed, the resulting hex-encoded output exceeds the fixed-size stack buffer. Stack memory adjacent to the buffer is overwritten, including saved registers and return addresses on architectures without stack protections.

Root Cause

The bounds check fails to account for the 2x size expansion inherent to hexadecimal encoding. Each input byte produces two output bytes, so a buffer sized for N bytes of input requires 2N bytes of output capacity. The original validation logic treats input and output capacity as equivalent, producing a classic off-by-multiple write past the end of a stack-allocated buffer.

Attack Vector

Attackers exploit the flaw by supplying a malicious rule file or by passing crafted rules through the -j (left rule) or -k (right rule) command-line options. The triggering condition is a password candidate length of 128 characters or more combined with the hex-mangling rule. Shared rule files distributed through public repositories, package archives, or red-team toolkits represent a realistic delivery channel. While CVSS labels the attack vector as network, practical exploitation requires the victim to run hashcat against attacker-controlled inputs.

No public proof-of-concept code is verified. A reference describing the issue is available at the GitHub Gist PoC Example.

Detection Methods for CVE-2026-42482

Indicators of Compromise

Hashcat processes terminating with SIGSEGV or stack-smashing diagnostics during rule application
Execution of hashcat 7.1.2 with -j or -k flags referencing untrusted rule files
Rule files containing hex-mangling directives paired with wordlists containing entries 128 characters or longer
Unexpected child processes spawned by hashcat outside its normal execution profile

Detection Strategies

Inventory all hosts running hashcat version 7.1.2 using software asset management or endpoint telemetry
Alert on command-line invocations of hashcat that include -j or -k with externally sourced rule files
Capture process crash dumps from systems running hashcat to identify exploitation attempts

Monitoring Recommendations

Forward hashcat process execution events and crash telemetry to a centralized logging platform for correlation
Monitor file integrity on rule directories such as rules/ within hashcat installations
Track network egress from research and password-audit workstations for anomalous post-exploitation traffic

How to Mitigate CVE-2026-42482

Immediate Actions Required

Stop using hashcat 7.1.2 with untrusted rule files until a fixed release is installed
Restrict execution of hashcat to vetted users on isolated, non-privileged workstations
Validate password wordlists and reject entries 128 characters or longer when hex-mangling rules are in use
Review existing rule files for u and l hex-conversion directives originating from untrusted sources

Patch Information

No vendor patch is referenced in the NVD entry at the time of publication. Track the hashcat project and the vendor advisory reference for an official fixed release. Once published, upgrade beyond version 7.1.2 and rebuild any container images or packaging that bundle the affected binary.

Workarounds

Run hashcat under a non-privileged user account inside a sandbox or container with no outbound network access
Compile hashcat with stack protector flags such as -fstack-protector-strong and -D_FORTIFY_SOURCE=2
Pre-filter candidate wordlists to remove entries with length greater than or equal to 128 characters before applying hex rules
Disable use of -j and -k rule options when processing rule files from external sources

bash

# Configuration example: filter long candidates and run hashcat in a constrained shell
awk 'length($0) < 128' wordlist.txt > wordlist.filtered.txt
firejail --net=none --private-tmp \
  hashcat -m 0 -a 0 wordlist.filtered.txt \
  --rules-file ./rules/trusted.rule