New from Wiz Research: Critical NGINX flaw. Understand the risk and impact.
Vulnerability DatabaseGHSA-9wh3-m3m5-jffx

GHSA-9wh3-m3m5-jffx: 
PHP vulnerability analysis and mitigation

In transaction-doctrine from 1.3.0, when a WriteUnit attempt failed with a transient conflict (deadlock, lock wait timeout, serialization failure) and was replayed, every open entity manager was cleared between attempts unless the unit was scoped with forManagers(), the host application's default manager included. Every entity the host had loaded before run() was detached, even when the unit itself wrote only through DBAL. Changes the host made to those entities afterwards were no longer tracked, so its next flush() silently wrote nothing for them, while the request still reported success. Bundles and applications that use WriteUnit together with the default entity manager are affected.

Update (2026-11-20): 1.8.0 and 1.8.1 are only a partial fix

1.8.0 no longer clears an open manager, and restores it when the failure leaves it open (for example a failure at commit time, in a later statement, or a lock wait timeout after the flush). But when the failure happens inside the attempt's own flush() — the usual case for a deadlock on MySQL and MariaDB — Doctrine closes the entity manager and transaction-doctrine resets it before the replay. Entities the host loaded before run() are then detached, and writes to them are still silently lost in 1.8.0 and 1.8.1. The same happens in 1.8.1 when code inside the attempt clears the manager or detaches every host entity it held. A complete fix (re-attaching the host's entities to the reset manager) is planned for 1.8.2; this advisory will be updated when it is released.

Patches

Partially fixed in 1.8.0 (see the update above). An open manager that managed entities before the first attempt is no longer cleared: it is put back to what the database holds (host entities deleted by the attempt are managed again, the attempt's unflushed work is dropped, entities the attempt brought in are unlinked and detached, touched host entities are reloaded without cascading). When a host entity cannot be brought back, the unit gives up with a ConcurrencyConflict (HTTP 409, Retry-After, CONFLICT_001).

Workarounds

Load again, inside the unit's closure, every entity you change in the unit, instead of using entities loaded before run(); this works on every version. Scoping write units with forManagers() to the managers they actually write through also avoids clearing unrelated managers.


Source: NVD

Related PHP vulnerabilities:

CVE ID

Severity

Score

Technologies

Component name

CISA KEV exploit

Has fix

Published date

GHSA-hg3m-q4fj-ggmqCRITICAL9.3
  • PHP logoPHP
  • wallabag/wallabag
NoNoNov 20, 2026
CVE-2026-63012MEDIUM6.5
  • PHP logoPHP
  • baserproject/baser-core
NoYesNov 19, 2026
GHSA-crpp-mf59-rhpqMEDIUMN/A
  • PHP logoPHP
  • api-platform/core
NoYesNov 20, 2026
GHSA-9wh3-m3m5-jffxMEDIUMN/A
  • PHP logoPHP
  • nowo-tech/transaction-doctrine
NoYesNov 20, 2026
GHSA-65w2-9r4h-848gMEDIUMN/A
  • PHP logoPHP
  • nowo-tech/transaction-doctrine
NoYesNov 20, 2026

Free Vulnerability Assessment

Benchmark your Cloud Security Posture

Evaluate your cloud security practices across 9 security domains to benchmark your risk level and identify gaps in your defenses.

Request assessment

Additional Wiz resources

Get a personalized demo

Ready to see Wiz in action?

"Best User Experience I have ever seen, provides full visibility to cloud workloads."
David EstlickCISO
"Wiz provides a single pane of glass to see what is going on in our cloud environments."
Adam FletcherChief Security Officer
"We know that if Wiz identifies something as critical, it actually is."
Greg PoniatowskiHead of Threat and Vulnerability Management