GHSA-37F3-6P89-6QR9
Vulnerability from github – Published: 2026-09-02 14:50 – Updated: 2026-09-17 16:34Summary
The regex_replace filter and function are allowlisted in Grav's Twig content sandbox. When Twig processing in page content is enabled security.twig_content.process_enabled: true, authenticated page editors can supply a catastrophically backtracking PCRE pattern, causing unbounded CPU consumption and denying service to the entire web server process.
Details
The Twig sandbox allowlists, defined in system/config/security.yaml, explicitly include regex_replace in both the filter and function permission lists:
Source: system/config/security.yaml
twig_sandbox:
allowed_filters:
# ...
- regex_replace # user-controlled pattern allowed in sandbox
allowed_functions:
# ...
- regex_replace # same
The underlying implementation passes the caller-controlled $pattern directly into PHP's preg_replace() without any pattern complexity validation:
Source: system/src/Grav/Common/Twig/Extension/GravExtension.php:1317-1319
public function regexReplace($subject, $pattern, $replace, $limit = -1)
{
return preg_replace($pattern, $replace, $subject, $limit);
}
When twig_content.process_enabled is true, page body content is sandboxed but can use any allowlisted filter. An editor who embeds a catastrophic backtracking pattern causes the PCRE engine to enter exponential time complexity, consuming 100% CPU until the PHP process is killed or the request times out.
Conditions required:
1. security.twig_content.process_enabled: true (opt-in, false by default on fresh 2.0 installs)
2. security.twig_sandbox.enabled: true (default) - the function is reachable under sandbox
3. Attacker must have page edit access (authenticated contributor / editor role)
PoC
Configuration prerequisite - enable Twig in content:
# user/config/security.yaml
twig_content:
process_enabled: true
Payload — embed in any Grav page body with process: { twig: true } in frontmatter:
---
title: Test
process:
twig: true
---
{{ 'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaab'|regex_replace('/^(a+)+$/', '') }}
Or as a function call in a page where the editor has Twig access:
{{ regex_replace('aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaab', '/^(a+)+$/', '') }}
Result: The PHP-FPM worker (or CLI server process) enters catastrophic PCRE backtracking. On a 2 GHz host, a 32-character string with the above pattern will exhaust one CPU core for seconds to minutes. With a slightly longer string, the time grows exponentially.
Impact
Vulnerability type: Regular Expression Denial of Service - ReDoS
Who is impacted: Server availability. Any Grav installation where:
- An editor-role account exists (or has been compromised), AND
- The operator has enabled twig_content.process_enabled: true
An attacker with page-edit access can render the site unresponsive for all visitors by publishing a page with a catastrophic regex. On single-worker PHP configurations this is a complete outage. On multi-worker setups, multiple concurrent page renders of the malicious page can saturate all workers.
{
"affected": [
{
"package": {
"ecosystem": "Packagist",
"name": "getgrav/grav"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "2.0.4"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-62672"
],
"database_specific": {
"cwe_ids": [
"CWE-1333"
],
"github_reviewed": true,
"github_reviewed_at": "2026-09-02T14:50:57Z",
"nvd_published_at": "2026-08-19T16:18:20Z",
"severity": "MODERATE"
},
"details": "### Summary\n\nThe `regex_replace` filter and function are allowlisted in Grav\u0027s Twig content sandbox. When Twig processing in page content is enabled `security.twig_content.process_enabled: true`, authenticated page editors can supply a catastrophically backtracking PCRE pattern, causing unbounded CPU consumption and denying service to the entire web server process.\n\n---\n\n### Details\n\nThe Twig sandbox allowlists, defined in `system/config/security.yaml`, explicitly include `regex_replace` in both the filter and function permission lists:\n\n**Source: `system/config/security.yaml`**\n\n```yaml\ntwig_sandbox:\n allowed_filters:\n # ...\n - regex_replace # user-controlled pattern allowed in sandbox\n allowed_functions:\n # ...\n - regex_replace # same\n```\n\nThe underlying implementation passes the caller-controlled `$pattern` directly into PHP\u0027s `preg_replace()` without any pattern complexity validation:\n\n**Source: `system/src/Grav/Common/Twig/Extension/GravExtension.php:1317-1319`**\n\n```php\npublic function regexReplace($subject, $pattern, $replace, $limit = -1)\n{\n return preg_replace($pattern, $replace, $subject, $limit);\n}\n```\n\nWhen `twig_content.process_enabled` is `true`, page body content is sandboxed but can use any allowlisted filter. An editor who embeds a catastrophic backtracking pattern causes the PCRE engine to enter exponential time complexity, consuming 100% CPU until the PHP process is killed or the request times out.\n\n**Conditions required**:\n1. `security.twig_content.process_enabled: true` (opt-in, `false` by default on fresh 2.0 installs)\n2. `security.twig_sandbox.enabled: true` (default) - the function is reachable under sandbox\n3. Attacker must have page edit access (authenticated contributor / editor role)\n\n---\n\n### PoC\n\n**Configuration prerequisite** - enable Twig in content:\n\n```yaml\n# user/config/security.yaml\ntwig_content:\n process_enabled: true\n```\n\n**Payload** \u2014 embed in any Grav page body with `process: { twig: true }` in frontmatter:\n\n```\n---\ntitle: Test\nprocess:\n twig: true\n---\n{{ \u0027aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaab\u0027|regex_replace(\u0027/^(a+)+$/\u0027, \u0027\u0027) }}\n```\n\nOr as a function call in a page where the editor has Twig access:\n\n```twig\n{{ regex_replace(\u0027aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaab\u0027, \u0027/^(a+)+$/\u0027, \u0027\u0027) }}\n```\n\n**Result**: The PHP-FPM worker (or CLI server process) enters catastrophic PCRE backtracking. On a 2 GHz host, a 32-character string with the above pattern will exhaust one CPU core for seconds to minutes. With a slightly longer string, the time grows exponentially.\n\n---\n\n### Impact\n\n**Vulnerability type**: Regular Expression Denial of Service - ReDoS\n\n**Who is impacted**: Server availability. Any Grav installation where:\n- An editor-role account exists (or has been compromised), AND\n- The operator has enabled `twig_content.process_enabled: true`\n\nAn attacker with page-edit access can render the site unresponsive for all visitors by publishing a page with a catastrophic regex. On single-worker PHP configurations this is a complete outage. On multi-worker setups, multiple concurrent page renders of the malicious page can saturate all workers.\n\n---",
"id": "GHSA-37f3-6p89-6qr9",
"modified": "2026-09-17T16:34:43Z",
"published": "2026-09-02T14:50:57Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/getgrav/grav/security/advisories/GHSA-37f3-6p89-6qr9"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-62672"
},
{
"type": "WEB",
"url": "https://github.com/getgrav/grav/commit/907116268ad4f48be7472c57f64981918578599a"
},
{
"type": "PACKAGE",
"url": "https://github.com/getgrav/grav"
},
{
"type": "WEB",
"url": "https://github.com/getgrav/grav/releases/tag/2.0.4"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:4.0/AV:N/AC:H/AT:P/PR:L/UI:N/VC:N/VI:N/VA:H/SC:N/SI:N/SA:N",
"type": "CVSS_V4"
}
],
"summary": "Grav: Authenticated ReDoS via regex_replace in Twig Sandbox"
}
Sightings
| Author | Source | Type | Date | Other |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or observed by the user.
- Confirmed: The vulnerability has been validated from an analyst's perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
- Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
- Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
- Not confirmed: The user expressed doubt about the validity of the vulnerability.
- Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.
The approach is described in our paper Mapping CVEs to MITRE ATT&CK Techniques: A Curated Gold-Set Classifier and the Limits of LLM-Assisted Label Expansion.
Browse all ATT&CK techniques and the vulnerabilities related to each.
Related by attack behaviour
Vulnerabilities whose description is nearest to this one in the vector space of the CIRCL/vulnerability-attack-technique-biencoder model. This is a similarity search over the bi-encoder space (plain cosine), not a classification, and it has no measured accuracy.