Skip to content
KitploitKITPLOIT
ToolsExploitsBlog
Log in
Submit
ToolsExploitsBlog
Submit

Hacking, PenTest, and Cybersecurity Tools for Your Security Arsenal!

Kitploit is a directory of hacking, cybersecurity, and pentesting tools. Discover the latest project updates to find vulnerabilities, analyze systems, automate testing, and strengthen your security.

FeedsContactPrivacy© 2026 Kitploit

Tool Directory

Categories

View all categories
Loading categories
php-tzcache-rce — CVE-2026-129496 - PHP - Critical: 9.2 - Remote - RCE - timezone cache use-after-free. No password. A startup warning logged to a file lets the next request run a command as the web worker. Lab pins 8.4.27 vs 8.4.28. | Kitploit
Tools/GitHubGitHub/abraxas/php-tzcache-rce
Memory ForensicsVulnerability AnalysisExploitationWeb Application ExploitationPenetration TestingLabs & Practice
GitHubabraxas/php-tzcache-rce

php-tzcache-rce

CVE-2026-129496 - PHP - Critical: 9.2 - Remote - RCE - timezone cache use-after-free. No password. A startup warning logged to a file lets the next request run a command as the web worker. Lab pins 8.4.27 vs 8.4.28.

View Repository
2218h 2m agoNot yet reviewed
Share

Abraxas Labs - php-tzcache-rce

php-tzcache-rce

Someone on the internet can talk to a PHP site with no password. If that server wrote a startup warning into a file log, the first request a fresh worker handles can run a command as the account that serves the site. This lab does that on PHP 8.4.27. The same request on 8.4.28 does not. The proof reads the worker's memory map from inside the lab. That is the "known addresses" condition the PHP Group already named. It is not a one-packet attack against a server whose addresses you have never seen.

Class: RCE Reach: Remote

PHP 8.4.27 (php:8.4.27-apache) vs patched 8.4.28 - GHSA-gr4f-wgx5-jpfv

A file error log during module startup allocates the date extension's timezone cache on a heap that startup then frees. The pointer stays. The date extension clears it only after the next request has already started parsing input. A POST bigger than memory_limit dies in that window, and cleanup calls a destructor planted in the freed cache. I pinned Apache prefork to one worker, set date.timezone=Not/AZone, and logged to /var/log/php-error.log. 8.4.27 writes CVE-2026-129496-RCE and the connection resets. 8.4.28 returns HTTP 500 and leaves the file absent.

IDCVE-2026-129496
AdvisoryGHSA-gr4f-wgx5-jpfv
ClassRCE
ReachRemote (unauthenticated HTTP, fresh worker)
CWECWE-416 / CWE-459
CVSSCritical: 9.2 CVSS:4.0/AV:N/AC:H/AT:P/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N (PHP Group)
ProductPHP
AffectedPHP 8.2 before 8.2.36, 8.3 before 8.3.37, 8.4 before 8.4.28, 8.5 before 8.5.13 (lab pin 8.4.27)
Fixed8.2.36 / 8.3.37 / 8.4.28 / 8.5.13 (19 Nov 2026)
Authnone
LicenseGNU Affero GPL v3.0
Lab127.0.0.1 only

Thread-safe builds do not take this path. FPM's own error log does not either. The diagnostic has to go through PHP's file error_log.

What an attacker can do

MethodSinkPrivilege
Unauthenticated POST /index.php with a Basic token and a body over memory_limit, on a fresh prefork childstale date_globals.tzcache freed at module startup, destructor called from request cleanupIntegrity (lab-proven): worker runs printf and writes CVE-2026-129496-RCE as www-data on 8.4.27. Availability (lab-proven): that connection resets. Addresses come from the lab process maps.
Same request on 8.4.28log timestamps no longer use the request timezone cache outside a live requestClosed (lab-proven): HTTP 500, witness file absent, worker stays up.
max_input_vars warning on the first requestsame stale cache, early in input parsingAvailability (advisory): crash without the destructor chain. This lab used the memory-limit cleanup path.
Thread-safe build, or error_log set to syslogstartup log does not allocate the cache this wayClosed. Not this bug.
Blind remote host, no memory mapASLRNot shown. The PHP Group said a remote ASLR bypass is plausible and that a complete independent bypass was not demonstrated. This lab does not add one.
Login, reverse shell, custom .son/aClosed. The witness command is printf into /tmp/w.

Lab-proven vs inferred: command execution and the patched miss are lab-proven on official php:8.4.27-apache vs php:8.4.28-apache, aarch64, with addresses read from that container. A crash from max_input_vars alone, and code execution without a local map, are not what this run showed.

Requires: non-thread-safe Apache apache2handler, file error_log, a startup diagnostic (date.timezone=Not/AZone), memory_limit=2M, LimitRequestFieldSize large enough for the Basic token, one prefork worker, and SYS_PTRACE so the lab can read /proc/<pid>/mem. Loopback only.

Lab

python3 php-tzcache-rce-Abraxas-Labs.py

or cd lab && ./run.sh. Loopback 127.0.0.1:18224 (8.4.27) vs 127.0.0.1:18225 (8.4.28). Compose project cve-2026-129496. Witness CVE-2026-129496-RCE. Last line:

SUCCESS CVE-2026-129496 rce=yes

Authorized lab only. It talks to loopback.

The fix

Update to PHP 8.4.28, or 8.2.36 / 8.3.37 / 8.5.13. The fix formats those log timestamps in UTC outside a fully initialized request. Until then, correct the setting that prints at startup, send PHP errors to syslog instead of a file, and restart the master and every worker.

References

  • GHSA-gr4f-wgx5-jpfv
  • CVE-2026-129496
  • php/php-src tag php-8.4.27 vs php-8.4.28
  • CWE-416
  • CWE-459

Abraxas Labs [email protected]

Download Tool

Most Popular

View all →

Discover the most used tools by our community.

  • 1

    OsintHunter

    letox4/osinthunter

    🎯 The ultimate OSINT toolkit: investigation graph, confidence scoring, alias families, map, timeline and reports. 49 modules, 400+ sites, no API keys required.

    977
  • 2

    ghostlock-app

    yukonga/ghostlock-app

    GhostLock One-Tap Execution App (CVE-2026-43499)

    564
  • 3

    looking-glass

    anonghosty/looking-glass

    CSIRT platform that ingests Shadowserver reports into MongoDB and replays them as interactive maps, analysis, findings, advisories and executive dashboards.

    352
  • 4

    redforge

    cybersecurity91/redforge

    Browser-side red team toolkit with 102 offline tools for recon, payload generation, credential attacks, crypto, encoding and engagement reporting.

    344
  • 5

    ARTEX

    baarai/artex

    LLM multi-agent autonomous penetration testing platform with planner/worker agents, asset and exploration graphs, MITM traffic recording, and vulnerability retesting.

    322
  • 6

    CyberStrikeAI

    ed1s0nz/cyberstrikeai

    The system of action for AI-native cybersecurity—where intent becomes governed execution, evidence becomes operational memory, and every operation improves the next.

    308

Explore all tools

Browse our collection of tools

View all tools →