GHSA-GX4C-2HQX-CW2R
Vulnerability from github – Published: 2026-08-05 20:43 – Updated: 2026-08-05 20:43Vulnerability Details
File: backend/s3/s3.go
Lines: 1359-1380 (functions s3CheckRedirect / s3RedirectCrossesHost)
Root Cause
Commit e7b1eb774 (released in v1.74.3) added a CheckRedirect policy for
the S3 HTTP client whose purpose is to strip the X-Amz-Security-Token
header (the AWS STS session token) whenever a redirect chain "crosses a
host", so the token isn't forwarded to an unintended origin.
s3RedirectCrossesHost decides this purely by comparing url.URL.Host
(hostname[:port]); it never looks at url.URL.Scheme. A redirect that keeps
the exact same host:port but changes the scheme from https:// to http://
therefore compares as "same host" and X-Amz-Security-Token is not
stripped — it is sent again, this time over plaintext HTTP.
func s3RedirectCrossesHost(req *http.Request, via []*http.Request) bool {
if len(via) == 0 {
return false
}
host := via[0].URL.Host
for _, redirect := range via[1:] {
if redirect.URL.Host != host {
return true
}
}
return host != req.URL.Host
}
Attack Scenario
- The user configures an
s3remote (or--s3-endpointpointing at a self-hosted/third-party S3-compatible service) using temporary credentials that include an STSsession_token(common for assumed-role / CI / Kubernetes IRSA setups). - The configured endpoint responds to a request with a 3xx redirect to the
same host:port but with
http://instead ofhttps://(TLS-front misconfiguration, maintenance redirect, or a malicious/compromised storage provider trying to harvest the token). - rclone's S3 HTTP client follows the redirect and re-sends the request,
including
X-Amz-Security-Token, over the now-unencrypted connection to that same host. - Any passive observer on that now-plaintext network path can read the STS session token from the request headers.
Impact
Disclosure of the AWS STS session token (X-Amz-Security-Token) in
cleartext for the remainder of its validity window. This is the exact class
of leak that e7b1eb774 was written to close — it just doesn't cover the
scheme-downgrade axis of "crossing a host".
Vulnerable Code
func s3RedirectCrossesHost(req *http.Request, via []*http.Request) bool {
if len(via) == 0 {
return false
}
host := via[0].URL.Host
for _, redirect := range via[1:] {
if redirect.URL.Host != host {
return true
}
}
return host != req.URL.Host
}
Recommended Fix
Also compare URL.Scheme, so a scheme downgrade on the same host is treated
the same as a host change:
func s3RedirectCrossesHost(req *http.Request, via []*http.Request) bool {
if len(via) == 0 {
return false
}
scheme, host := via[0].URL.Scheme, via[0].URL.Host
for _, redirect := range via[1:] {
if redirect.URL.Host != host || redirect.URL.Scheme != scheme {
return true
}
}
return host != req.URL.Host || scheme != req.URL.Scheme
}
Verification
Added a unit test (backend/s3/redirect_scheme_test.go) that calls the real,
unmodified s3RedirectCrossesHost / s3CheckRedirect with an
https://bucket.example.com -> http://bucket.example.com redirect chain.
On unpatched code (commit 16091ce365, current master / v1.74.3):
- s3RedirectCrossesHost returns false
- s3CheckRedirect leaves X-Amz-Security-Token: SECRET-SESSION-TOKEN
intact on the outgoing (plaintext) request.
=== RUN TestSchemeDowngradeNotDetectedAsCrossHost
redirect_scheme_test.go:23: initial=https://bucket.example.com final=http://bucket.example.com s3RedirectCrossesHost=false
--- PASS: TestSchemeDowngradeNotDetectedAsCrossHost (0.00s)
After applying the one-line fix above (also adding scheme comparison), the
token is correctly stripped and all existing redirect tests
(TestClientRemovesSecurityTokenOnCrossHostRedirect,
TestClientDoesNotRestoreSecurityTokenAfterCrossHostRedirect,
TestClientKeepsSecurityTokenOnSameHostRedirect,
TestClientStopsAfterTenRedirects) continue to pass.
A minimal fix commit is ready and can be pushed to a private fork once this report is acknowledged.
{
"affected": [
{
"database_specific": {
"last_known_affected_version_range": "\u003c= 1.74.3"
},
"package": {
"ecosystem": "Go",
"name": "github.com/rclone/rclone"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "1.74.4"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [],
"database_specific": {
"cwe_ids": [
"CWE-319",
"CWE-522"
],
"github_reviewed": true,
"github_reviewed_at": "2026-08-05T20:43:11Z",
"nvd_published_at": null,
"severity": "LOW"
},
"details": "## Vulnerability Details\n\n**File**: `backend/s3/s3.go`\n**Lines**: 1359-1380 (functions `s3CheckRedirect` / `s3RedirectCrossesHost`)\n\n### Root Cause\nCommit `e7b1eb774` (released in v1.74.3) added a `CheckRedirect` policy for\nthe S3 HTTP client whose purpose is to strip the `X-Amz-Security-Token`\nheader (the AWS STS session token) whenever a redirect chain \"crosses a\nhost\", so the token isn\u0027t forwarded to an unintended origin.\n\n`s3RedirectCrossesHost` decides this purely by comparing `url.URL.Host`\n(hostname[:port]); it never looks at `url.URL.Scheme`. A redirect that keeps\nthe exact same host:port but changes the scheme from `https://` to `http://`\ntherefore compares as \"same host\" and `X-Amz-Security-Token` is *not*\nstripped \u2014 it is sent again, this time over plaintext HTTP.\n\n```go\nfunc s3RedirectCrossesHost(req *http.Request, via []*http.Request) bool {\n\tif len(via) == 0 {\n\t\treturn false\n\t}\n\thost := via[0].URL.Host\n\tfor _, redirect := range via[1:] {\n\t\tif redirect.URL.Host != host {\n\t\t\treturn true\n\t\t}\n\t}\n\treturn host != req.URL.Host\n}\n```\n\n### Attack Scenario\n1. The user configures an `s3` remote (or `--s3-endpoint` pointing at a\n self-hosted/third-party S3-compatible service) using temporary\n credentials that include an STS `session_token` (common for assumed-role\n / CI / Kubernetes IRSA setups).\n2. The configured endpoint responds to a request with a 3xx redirect to the\n *same* host:port but with `http://` instead of `https://` (TLS-front\n misconfiguration, maintenance redirect, or a malicious/compromised\n storage provider trying to harvest the token).\n3. rclone\u0027s S3 HTTP client follows the redirect and re-sends the request,\n including `X-Amz-Security-Token`, over the now-unencrypted connection to\n that same host.\n4. Any passive observer on that now-plaintext network path can read the STS\n session token from the request headers.\n\n### Impact\nDisclosure of the AWS STS session token (`X-Amz-Security-Token`) in\ncleartext for the remainder of its validity window. This is the exact class\nof leak that `e7b1eb774` was written to close \u2014 it just doesn\u0027t cover the\nscheme-downgrade axis of \"crossing a host\".\n\n### Vulnerable Code\n```go\nfunc s3RedirectCrossesHost(req *http.Request, via []*http.Request) bool {\n\tif len(via) == 0 {\n\t\treturn false\n\t}\n\thost := via[0].URL.Host\n\tfor _, redirect := range via[1:] {\n\t\tif redirect.URL.Host != host {\n\t\t\treturn true\n\t\t}\n\t}\n\treturn host != req.URL.Host\n}\n```\n\n### Recommended Fix\nAlso compare `URL.Scheme`, so a scheme downgrade on the same host is treated\nthe same as a host change:\n\n```go\nfunc s3RedirectCrossesHost(req *http.Request, via []*http.Request) bool {\n\tif len(via) == 0 {\n\t\treturn false\n\t}\n\tscheme, host := via[0].URL.Scheme, via[0].URL.Host\n\tfor _, redirect := range via[1:] {\n\t\tif redirect.URL.Host != host || redirect.URL.Scheme != scheme {\n\t\t\treturn true\n\t\t}\n\t}\n\treturn host != req.URL.Host || scheme != req.URL.Scheme\n}\n```\n\n### Verification\nAdded a unit test (`backend/s3/redirect_scheme_test.go`) that calls the real,\nunmodified `s3RedirectCrossesHost` / `s3CheckRedirect` with an\n`https://bucket.example.com` -\u003e `http://bucket.example.com` redirect chain.\n\nOn unpatched code (commit 16091ce365, current master / v1.74.3):\n- `s3RedirectCrossesHost` returns `false`\n- `s3CheckRedirect` leaves `X-Amz-Security-Token: SECRET-SESSION-TOKEN`\n intact on the outgoing (plaintext) request.\n\n```\n=== RUN TestSchemeDowngradeNotDetectedAsCrossHost\n redirect_scheme_test.go:23: initial=https://bucket.example.com final=http://bucket.example.com s3RedirectCrossesHost=false\n--- PASS: TestSchemeDowngradeNotDetectedAsCrossHost (0.00s)\n```\n\nAfter applying the one-line fix above (also adding scheme comparison), the\ntoken is correctly stripped and all existing redirect tests\n(`TestClientRemovesSecurityTokenOnCrossHostRedirect`,\n`TestClientDoesNotRestoreSecurityTokenAfterCrossHostRedirect`,\n`TestClientKeepsSecurityTokenOnSameHostRedirect`,\n`TestClientStopsAfterTenRedirects`) continue to pass.\n\nA minimal fix commit is ready and can be pushed to a private fork once this\nreport is acknowledged.",
"id": "GHSA-gx4c-2hqx-cw2r",
"modified": "2026-08-05T20:43:11Z",
"published": "2026-08-05T20:43:11Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/rclone/rclone/security/advisories/GHSA-gx4c-2hqx-cw2r"
},
{
"type": "WEB",
"url": "https://github.com/rclone/rclone/commit/1a28451ea6fc8ac1806b0e9923dcb5b3f543f7fa"
},
{
"type": "PACKAGE",
"url": "https://github.com/rclone/rclone"
},
{
"type": "WEB",
"url": "https://github.com/rclone/rclone/releases/tag/v1.74.4"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:A/AC:H/PR:N/UI:N/S:U/C:L/I:N/A:N",
"type": "CVSS_V3"
}
],
"summary": "rclone: S3 backend does not strip X-Amz-Security-Token on a same-host HTTPS-\u003eHTTP redirect"
}
Drag nodes to arrange them. Double-click a related vulnerability to open it, or a source to visit its link. The Physics button on the left tunes the layout and the simulation; the legend at the bottom left hides or shows nodes by kind and by sighting type.
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.
Vulnerability timeline
Sightings and recorded dates, grouped by day · UTC
How durations are calculated
Following the VTEM framework (Black Hat USA 2015), time to patch starts when the vendor or a coordinator was informed, total time to patch at discovery, and disclosure to patch at public disclosure; vendor response time runs from the report to its acknowledgement. Time to exploit and time unpatched start at the earliest discovery or report, falling back to public disclosure. Unpatched time ends at the first recorded patch, or the calculation time when no patch is recorded. Exploitation to patch shows the interval between those two events; a published proof of concept is a public exploit, not exploitation.
Each result shows its source dates. A negative interval means the end event came first; when date-only inputs leave the order open, the result says so. Date-only inputs are approximate. The earliest evidence wins: KEV report dates and sightings are labelled as such. Records describing another vulnerability are listed but never calculated from, and a PoC alone does not establish exploitation. Missing patch evidence does not prove that no patch exists. Filters below change the event list only.
Loading timeline…
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.