predis/predis security advisory
CVE-2026-84372: Predis: Redis command injection and denial of service via CRLF smuggling in pipelined commands on aggregate connections
CVE-2026-84372 is a critical-severity denial of service vulnerability in the PHP package predis/predis (CVSS 9.8), published 1 September 2026. It is fixed in 3.3.0; earlier versions are affected. It can be exploited without an account.
critical 9.8 published 2026-09-01 · updated 2026-09-09
- Affected versions
- >= 3.0.0-RC1, < 3.3.0
- Fixed in
- 3.3.0
- Newest release
- 3.7.1
- Severity
- Critical · CVSS 9.8
- Type
- Denial of service
CWE-93 - Exploited in the wild
- Not on the CISA KEV list
checked 2026-11-04
Is your predis/predis version affected?
Who can exploit CVE-2026-84372
- Access: No account. Anyone who can send a request to an app that uses the affected code.
CVSS vector (scored by GitHub): CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H
How to fix CVE-2026-84372 in predis/predis
- Find the installed version:
composer show predis/predis. If you never required it yourself,composer why predis/predisshows which package pulls it in. - Update it:
composer update predis/predis --with-all-dependencies. You need 3.3.0 or newer. - If Composer will not go that high, your composer.json constraint is holding it back. Widen it (for example
"predis/predis": "^3.3.0"if you can take that version), or upgrade the package that requires it. - Run
composer audit. It reads the same advisory data and should no longer list CVE-2026-84372. - Commit composer.lock and deploy it. On the server, run
composer install --no-devfrom the lock file, notcomposer update.
If you cannot upgrade yet
- Rate-limit the affected endpoint and cap request body, header and upload sizes at the web server.
- If your app does not use the affected feature, note that in your audit, but still plan the upgrade.
The CVE-2026-84372 advisory
Predis is a flexible and feature-complete Redis and Valkey client for PHP. From version 3.0.0-RC1 until version 3.3.0, pipeline handling on aggregate cluster and replication connections reparses an already serialized RESP buffer in AbstractAggregateConnection::write() by splitting it with explode("\r\n") instead of honoring RESP length prefixes. Attacker-controlled keys or values containing CRLF sequences can therefore be interpreted by Command::deserializeCommand() as additional commands. On cluster connections, ClusterStrategy::getFakeKey() can route injected keyless commands using the literal fake key value "key", permitting operations such as shard-wide cache deletion, targeted data modification, data reads, or node disruption. On replication connections, malformed reparsing can throw an uncaught exception and repeatedly terminate affected requests. Only pipeline() reaches this vulnerable path; transaction() and MULTI are not affected. This issue is fixed in version 3.3.0.
References
- GitHub advisory GHSA-w6f5-v2h6-g786 github.com
- Fix commit 053cb4b github.com
- Pull request #1586 github.com
- Release v3.3.0 github.com
- github.com github.com
Related
Need help patching or upgrading? Laskyr does PHP and Laravel security fixes and upgrades. Tell us what you run and we reply with a plan and a price.
Ask about a fixWhere this data comes from
- Affected and fixed versions: the Packagist security advisories API (GitHub Advisory Database and FriendsOfPHP), the same data
composer audituses. Synced 2026-11-03. - Description, score, weakness and references: NIST National Vulnerability Database (NVD), checked 2026-11-04. This product uses data from the NVD API but is not endorsed or certified by the NVD.
- Exploited-in-the-wild status: the CISA Known Exploited Vulnerabilities catalog, checked 2026-11-04.
- Compiled by Daniel Lasek at Laskyr from the sources above. The summary, “who can exploit it”, “how to fix it” and hardening sections are our plain-English reading of the record and are generated from it. The quoted description is the record's own text. Spot an error? Tell us.