---
id: "CVE-2026-120634"
kind: "vulnerability"
aliases: ["CVE-2026-120634"]
title: "Perl CGI::Carp leaks one user's warnings into another user's page"
severity: "unknown"
cvss: null
epss: null
exploited: false
status: "open"
category: "package"
products: []
ecosystems: []
tags: ["cgi-carp", "fastcgi", "mod_perl", "package", "perl", "psgi"]
not_affected_if: ["plain-cgi", "warningstobrowser-off"]
published_at: "2026-11-04T00:17:24+00:00"
first_seen_at: "2026-11-04T00:46:16+00:00"
updated_at: "2026-11-04T00:48:23+00:00"
sources: [{"source": "nvd", "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-120634", "first_seen_at": "2026-11-04T00:46:16+00:00"}]
---

# Perl CGI::Carp leaks one user's warnings into another user's page

CGI::Carp before 5.00 keeps warnings in a package-level buffer that is never cleared between requests. Under persistent runtimes such as mod_perl, FastCGI or PSGI, warnings containing one user's request data can be written into another user's HTML response.

_Summary, check and fix written by a model from the sources; labelled as such._

## How to check

- Check the installed CGI::Carp version with `perl -MCGI::Carp -e 'print $CGI::Carp::VERSION'`
- Look for CGI::Carp use with warningsToBrowser enabled under mod_perl, FastCGI or PSGI

## Fix

- Upgrade CGI::Carp to `5.00` or later

## Timeline

- 2026-11-04T00:46:16+00:00 nvd https://nvd.nist.gov/vuln/detail/CVE-2026-120634

## Documents

- [fix] Patch: github.com https://github.com/leejo/CGI.pm/commit/604a83fec105f75fe06f1a75026c4c34fd5a1f61.patch

## From the sources

CGI::Carp versions before 5.00 for Perl leak warnings across requests in persistent environments via the warningsToBrowser buffer.

Warnings are held in a package-level buffer until warningsToBrowser writes them into the page as HTML comments, and the buffer is never cleared between requests. Under mod_perl, FastCGI or PSGI, warnings from one request are written into a later one.

Warnings that contain one user's request data can appear in the HTML sent to another user.

---

_Free to read at a human pace. An API with keys and an MCP server are in preparation; early access: hello@infrastack.ai_
