makepad/libs/asset/store/tests/http/security.rs
Admin e37c263b9c ai backbone: the hub era — makepad_ai deleted, every backend is a hub pipe; machine residency elections, job leases, ETA placement; the store only stores; creator pipelines run in the app (aicore)
Squashed from work:
- asset-ai: FastH3 4-step fast video backend; clip keyframes on the wire
- asset-ui: loop video chains — text→image→video that ends where it began
- h3: safetensors -> pruned-Q4_K GGUF quantizer for the 24GB DiT tiers
- h3_quant_gguf verify: row-error gates calibrated to the measured Q4_K floor
- asset-ai realtime: the feedback loop — the source anchors, the drifted frame inits
- asset-ai realtime: a feedback loop survives a resize and travels by default
- asset-ai realtime: the feedback loop frees itself from the feed handshake and pauses for its listener
- asset-ai realtime: the outbound encode leaves the loop's critical path
- asset-ai ocr: the ocr domain — Chandra 2 at page resolution, and the tower goes planner-owned
- llm slots: a lane can hold an image span — embedding prefill and a rope cursor of its own
- vision tower on CUDA: the encode leg gets its two missing kernels
- llm/ocr: one M-RoPE grid encoder for both image paths, and a livelock made an error
- vision tower on CUDA: the f16 GEMM keeps the precision it was throwing away
- live: a feed that moves box takes its trip with it — one seed image
- vision tower on CUDA: the tiled attention becomes bit-exact, and tensor cores go
- llm prefill on CUDA: the MMA attention kernel gets the tile a 4-to-1 model needs
- asset-ai ocr: the CUDA encode lane joins the integration — vision-parity sits beside run's three arms, and the kernels
- Merge branch 'ocr-perf-integration' into work
- asset-ai: the live anchor can follow the trip, and text leaves the 5090
- asset-ai: the camera moves the world, and the world starts still
- asset-import: the EA strategy classics, in the one 2D contract
- rtsmap: one seeded generator for tiled strategy maps
- asset-ui: one card for the strategy classics, with a pack dropdown
- asset-ai: music3 reference-audio path, ocr/h3 backends, registry
- asset: mp4 sample index for range-streaming, chat tools, import profiles
- cnc: tiberium is twelve growth frames, not twelve empty variants
- platform: native file and save dialogs, in-house on all three desktops
- chat: the scan holds out for a lane home
- chat: a full home queues you — take the free lane
- chat: the preload has a percentage, and the boundless cap stops showing
- llm cuda: the 32x2 attention tile — even GQA ratios stay on MMA
- sa3 gets a bake path: the sfx model's tables precomputed by a diffusion-side bin
- sqlite_query: anti-join regression test
- td import: HARV's second frame block is its harvesting cycle, not a turret
- asset-ui: sprite enhancement runs on the 32B dev DiT — distillation, not the prompt, was the ceiling
- ai-hub: makepad-asset-ai becomes makepad-ai-hub at libs/ai/hub, the chat pane becomes makepad-chat-ui, the service bin
- asset-ui: test health fixtures grow the realtime field they were born without
- ai-hub: one home at ~/.makepad — weights/ run/ cache/ logs/, the service cache migrates from ai_content by a single re
- ai-hub: subprocess workers die with the node — process groups everywhere, PDEATHSIG on linux, one KILL_ON_JOB_CLOSE Jo
- ai-hub: the hub object — AiHub::in_process, pipes vocabulary, and the local LLM engine generalized out of mpfiles (aic
- strict-json: the dependency-free JSON module gets its own crate; asset-client re-exports it so nothing downstream move
- ai-hub: the machine layer — node entries, the 0600 machine token, and the residency election that IS the lock (aicore
- ai-hub: MPHUB1 — the fabric beacon only dedicated nodes can send (aicore §4)
- ai-hub: job leases — work lives only while it is renewed (aicore §8)
- asset-creator: the pipeline library is born — specs, the deps gate, and the derived-state law (aicore §9)
- ai-hub: RAM residency facts — the CPU-side twin of residency.rs (aicore §3)
- ai-hub: ETA placement primitives — relative GPU throughput, the four-term estimate, and an observable breakdown (aicor
- ai-hub: leases go live on the wire — origin fields on submit, /job/<id>/keepalive, /bye, and the reaper that cancels w
- ai-hub: the chat providers move in — fleet qwen, openai, grok, claude/codex/grok CLIs, the responses driver, and the w
- asset-creator: the engine — one pipeline run against the hub, deps-gated, spliced, cancellable, resumable-by-construct
- ai-hub: the machine node mode — --machine binds loopback, registers in ~/.makepad/run, and exits on its own once idle
- asset-creator: makepad-creator-run — the detached client for runs that must outlive a window (aicore §9)
- ai-hub: a native Claude Messages-API provider — API-key or Claude Code OAuth, bounded SSE streaming, injected tools (a
- route + converse: off makepad_ai — the Agent seam moves to converse, route's cloud dispatcher rides the hub's Claude p
- asset-creator: the preset tables move in — fifteen chain-policy constants shared by every creator app (aicore §9 / P6)
- makepad_ai is deleted — every backend is a hub pipe, the agent seam lives with its consumers (aicore §14, decided 2026
- ai-hub: loads hold the machine residency election — set_model_state claims on Loaded and publishes the service port (a
- ai-hub: chats run the machine election — route to a serving holder, wait on a loading one, claim and publish when open
- ai-hub: pick_for_domain_eta — ETA-ranked placement over the shared hard-filter core (aicore §6 / P4)
- asset-creator: the engine picks a provider per stage at dispatch time — a chain's later stages see fresh fleet state (
- ai-hub: the fabric secret gates the service HTTP surface — bearer on everything but /health and the ticketed peer path
- vj: DREAM runs execute in the app — pipelines.rs becomes the run it used to watch (aicore §9 / F1)
- asset-creator: the runner — generate one thing and put it in the catalog, one implementation for every surface (aicore
- chat-ui: the session runs in the app — no broker anywhere on the chat path (aicore P8 / F5)
- asset-store: assets.query is a first-class query endpoint — the bounded SQL surface outlives the broker (aicore P8 / F
- asset-creator: CreatorTools — the chat tool pack for a store that only stores (aicore §9 / P8)
- asset-store: the shrink — the store stores (aicore P7)
- importer + asset-server host: the coordination era ends (aicore P7)
- store config purge + asset-ui goes fleet-direct; the derive protocol gets its route proof (aicore P7)
- client + chat dispatcher: the dead wire comes out (aicore P7/P8)
- ai-hub: 0.3.0 — the health version says which era a node runs
- ai-hub: the default fleet is 'gen' — apps hear the LAN without env plumbing
- ai-hub: the preload note percents the prefill, not the job bar
- ai-hub: conversations keep their KV — the wire mirror, the lane identity, the in-turn dynamic context (aicore §7)
- ai-hub: an open-think model is thinking from its first token
- libs: the zero-warning sweep — stitch casts say what they mean, xatlas keeps upstream's surface quietly
- zero-warning sweep, round two — the first full-workspace pass
- zero-warning sweep, round three — the model lanes and the deep examples
- zero-warning sweep, round four — the last stragglers
- zero-warning sweep, round five — vj and chat-ui
- zero-warning sweep, round six — three cascades

Co-authored-by: Claude <info@makepad.nl>
2026-09-01 16:46:31 +02:00

379 lines
15 KiB
Rust

//! Security-focused tests: credential uniformity, capability scoping and
//! delegation limits, worker identity binding, the namespace claim gate,
//! hostile HTTP input, upload refusals, and discovery-plane hostility.
mod common;
use common::*;
use makepad_asset_store::discovery::{caps, Beacon, DiscoveryListener, PROTOCOL_VERSION};
use makepad_asset_store::json::Value;
use makepad_asset_data::BlobId;
use std::net::UdpSocket;
// ---------------------------------------------------------------------------
// credentials
// ---------------------------------------------------------------------------
#[test]
fn refused_credentials_are_uniform() {
let ts = start_server("uniform401");
let token = ts.admin_token();
let mut admin = ts.control(Some(&token));
// A real token that will be expired, and one that will be revoked.
let victim = {
let r = admin.post_json("/v1/auth/principals", &jobj(vec![("name", jstr("victim"))]));
r.str_field("principal")
};
let short = admin
.post_json(
"/v1/auth/tokens",
&jobj(vec![("principal", jstr(victim.clone())), ("ttl_ms", Value::Int(50))]),
)
.str_field("token");
let revoked = admin
.post_json("/v1/auth/tokens", &jobj(vec![("principal", jstr(victim))]))
.str_field("token");
let r = admin.post_json("/v1/auth/tokens/revoke", &jobj(vec![("token", jstr(revoked.clone()))]));
assert_eq!(r.status, 204);
std::thread::sleep(std::time::Duration::from_millis(80)); // expire `short`
let unknown = format!("mpat_{}", "ab".repeat(32));
let cases: Vec<Option<String>> = vec![
None, // no header
Some("not-a-token".into()), // malformed shape
Some("mpat_zz".into()), // malformed hex
Some(unknown), // valid shape, unknown
Some(short), // expired
Some(revoked), // revoked
];
let mut bodies = Vec::new();
for case in cases {
let mut c = ts.control(case.as_deref());
let r = c.get("/v1/auth/whoami");
assert_eq!(r.status, 401);
assert_eq!(r.header("WWW-Authenticate"), Some("Bearer"));
bodies.push(r.body);
}
// Every refusal is byte-identical: no oracle distinguishes unknown,
// malformed, expired and revoked credentials.
for b in &bodies[1..] {
assert_eq!(b, &bodies[0]);
}
}
// ---------------------------------------------------------------------------
// capability scoping
// ---------------------------------------------------------------------------
#[test]
fn capability_scoping_and_revocation() {
let ts = start_server("scoping");
let token = ts.admin_token();
let mut admin = ts.control(Some(&token));
let scoped = principal_with(&mut admin, &[("asset_register", "demo")]);
let mut c = ts.control(Some(&scoped));
// In-scope namespace works; any other namespace is denied with the
// capability named.
let r = c.post_json("/v1/assets", &jobj(vec![("namespace", jstr("demo"))]));
assert_eq!(r.status, 201);
let asset_id = r.str_field("asset_id");
let r = c.post_json("/v1/assets", &jobj(vec![("namespace", jstr("other"))]));
assert_eq!(r.status, 403);
assert_eq!(r.json().get("capability").unwrap().as_str(), Some("asset_register"));
// A capability the principal never had is denied even in-scope.
let fake_rev = format!("arev_{}", "0".repeat(64));
let r = c.post_json(
&format!("/v1/assets/{asset_id}/revisions/{fake_rev}/publish"),
&jobj(vec![]),
);
assert_eq!(r.status, 403);
assert_eq!(r.json().get("capability").unwrap().as_str(), Some("asset_publish"));
// Revocation takes effect on the next request.
let whoami = c.get("/v1/auth/whoami").str_field("principal");
let r = admin.post_json(
"/v1/auth/grants/revoke",
&jobj(vec![
("principal", jstr(whoami)),
("capability", jstr("asset_register")),
("scope", jstr("demo")),
]),
);
assert_eq!(r.status, 204);
let r = c.post_json("/v1/assets", &jobj(vec![("namespace", jstr("demo"))]));
assert_eq!(r.status, 403);
}
#[test]
fn auth_admin_delegation_is_namespace_bounded() {
let ts = start_server("delegation");
let token = ts.admin_token();
let mut admin = ts.control(Some(&token));
let ns_admin = principal_with(&mut admin, &[("auth_admin", "demo")]);
let subject = principal_with(&mut admin, &[]);
let mut c = ts.control(Some(&ns_admin));
let subject_principal = {
let mut s = ts.control(Some(&subject));
s.get("/v1/auth/whoami").str_field("principal")
};
// A namespace-scoped auth admin may grant within its namespace...
let r = c.post_json(
"/v1/auth/grants",
&jobj(vec![
("principal", jstr(subject_principal.clone())),
("capability", jstr("blob_write")),
("scope", jstr("demo")),
]),
);
assert_eq!(r.status, 204);
// ...but not in another namespace, not with wildcard scope, and it may
// not mint principals or tokens (root-only).
let r = c.post_json(
"/v1/auth/grants",
&jobj(vec![
("principal", jstr(subject_principal.clone())),
("capability", jstr("blob_write")),
("scope", jstr("other")),
]),
);
assert_eq!(r.status, 403);
let r = c.post_json(
"/v1/auth/grants",
&jobj(vec![
("principal", jstr(subject_principal.clone())),
("capability", jstr("blob_write")),
("scope", jstr("*")),
]),
);
assert_eq!(r.status, 403);
assert_eq!(
c.post_json("/v1/auth/principals", &jobj(vec![("name", jstr("sneaky"))])).status,
403
);
let r = c.post_json(
"/v1/auth/tokens",
&jobj(vec![("principal", jstr(subject_principal))]),
);
assert_eq!(r.status, 403);
// The bootstrap admin principal cannot be disabled.
let root_principal = admin.get("/v1/auth/whoami").str_field("principal");
let r = admin.post_json(&format!("/v1/auth/principals/{root_principal}/disable"), &jobj(vec![]));
assert_eq!(r.status, 409);
}
// ---------------------------------------------------------------------------
// jobs: claim gate + identity binding + visibility
// ---------------------------------------------------------------------------
#[test]
fn hostile_http_input_is_refused() {
let ts = start_server("hostile");
let addr = ts.server.control_addr();
// Request smuggling and framing hostility.
let cases: Vec<(&[u8], u16)> = vec![
(b"POST /v1/catalog HTTP/1.1\r\nHost: x\r\nContent-Length: 3\r\nTransfer-Encoding: chunked\r\n\r\n", 400),
(b"POST /v1/catalog HTTP/1.1\r\nHost: x\r\nTransfer-Encoding: gzip, chunked\r\n\r\n", 501),
(b"GET /v1/health HTTP/1.0\r\nHost: x\r\n\r\n", 505),
(b"BREW /v1/health HTTP/1.1\r\nHost: x\r\n\r\n", 501),
(b"GET /v1/health HTTP/1.1\r\n\r\n", 400),
(b"GET /v1/health HTTP/1.1\r\nHost: a\r\nHost: b\r\n\r\n", 400),
(b"GET /v1/%2e%2e/secrets HTTP/1.1\r\nHost: x\r\n\r\n", 400),
(b"GET /v1/../etc/passwd HTTP/1.1\r\nHost: x\r\n\r\n", 400),
(b"GET /v1//health HTTP/1.1\r\nHost: x\r\n\r\n", 400),
(b"GET /v1/health?a=1&a=2 HTTP/1.1\r\nHost: x\r\n\r\n", 400),
(b"GET /v1/health HTTP/1.1\r\nHost: x\r\nBad: a\x01b\r\n\r\n", 400),
(b"GET /v1/health HTTP/1.1\r\nHost: x\r\nContent-Length: 5\r\n\r\n", 400),
];
for (raw, want) in cases {
let resp = raw_roundtrip(addr, raw);
assert_eq!(status_of(&resp), want, "{}", String::from_utf8_lossy(raw));
let text = String::from_utf8_lossy(&resp);
assert!(text.contains("Connection: close"), "hostile input closes");
assert!(text.contains("X-Content-Type-Options: nosniff"));
}
// Oversized head: refused with 431.
let mut big = b"GET /v1/health HTTP/1.1\r\nHost: x\r\n".to_vec();
big.extend_from_slice(format!("X-Filler: {}\r\n", "a".repeat(40 * 1024)).as_bytes());
big.extend_from_slice(b"\r\n");
assert_eq!(status_of(&raw_roundtrip(addr, &big)), 431);
// Chunked body over the JSON budget: 413, connection closed.
let token = ts.admin_token();
let mut chunked = format!(
"POST /v1/catalog HTTP/1.1\r\nHost: x\r\nAuthorization: Bearer {token}\r\nTransfer-Encoding: chunked\r\n\r\n"
)
.into_bytes();
let chunk = vec![b'a'; 64 * 1024];
for _ in 0..6 {
chunked.extend_from_slice(format!("{:x}\r\n", chunk.len()).as_bytes());
chunked.extend_from_slice(&chunk);
chunked.extend_from_slice(b"\r\n");
}
chunked.extend_from_slice(b"0\r\n\r\n");
assert_eq!(status_of(&raw_roundtrip(addr, &chunked)), 413);
// Chunk-extension and non-hex size lines are malformed.
let bad_chunk = format!(
"POST /v1/catalog HTTP/1.1\r\nHost: x\r\nAuthorization: Bearer {token}\r\nTransfer-Encoding: chunked\r\n\r\n5;ext=1\r\nhello\r\n0\r\n\r\n"
);
assert_eq!(status_of(&raw_roundtrip(addr, bad_chunk.as_bytes())), 400);
// Pipelined requests on one connection are answered in order.
let two = b"GET /v1/health HTTP/1.1\r\nHost: x\r\n\r\nGET /v1/health HTTP/1.1\r\nHost: x\r\nConnection: close\r\n\r\n";
let resp = raw_roundtrip(addr, two);
let text = String::from_utf8_lossy(&resp);
assert_eq!(text.matches("HTTP/1.1 200").count(), 2);
}
// ---------------------------------------------------------------------------
// upload refusals
// ---------------------------------------------------------------------------
#[test]
fn blob_upload_refusals_leave_no_state() {
// Shrunken blob budget makes over-budget cheap to hit.
let ts = start_server_with("uploadrefuse", |cfg| {
cfg.budgets.max_blob_bytes = 4096;
});
let token = ts.admin_token();
let mut admin = ts.control(Some(&token));
let writer = principal_with(&mut admin, &[("blob_write", "demo")]);
let mut data = ts.data(Some(&writer));
// Over budget: refused, and the digest stays unknown afterwards.
let big = vec![7u8; 8192];
let r = data.post_bytes("/v1/blobs?ns=demo", &big);
assert_eq!(r.status, 413);
let id = BlobId::hash_of(&big);
assert_eq!(data.get(&format!("/v1/blobs/{id}")).status, 404);
// Declared-digest mismatch: refused, nothing recorded.
let payload = b"honest bytes".to_vec();
let wrong = "1".repeat(64);
let r = data.request(
"POST",
&format!("/v1/blobs?ns=demo&sha256={wrong}"),
&[],
Some(&payload),
);
assert_eq!(r.status, 422);
let id = BlobId::hash_of(&payload);
assert_eq!(data.get(&format!("/v1/blobs/{id}")).status, 404);
// Missing namespace and missing capability refuse before any byte is
// admitted.
assert_eq!(data.post_bytes("/v1/blobs", &payload).status, 400);
let outsider = principal_with(&mut admin, &[]);
let mut noperm = ts.data(Some(&outsider));
let r = noperm.post_bytes("/v1/blobs?ns=demo", &payload);
assert_eq!(r.status, 403);
// Unauthenticated data plane: refused uniformly.
let mut anon = ts.data(None);
assert_eq!(anon.post_bytes("/v1/blobs?ns=demo", &payload).status, 401);
assert_eq!(anon.get(&format!("/v1/blobs/{id}")).status, 401);
}
// ---------------------------------------------------------------------------
// connection capacity
// ---------------------------------------------------------------------------
#[test]
fn over_capacity_connections_get_an_explicit_503() {
let ts = start_server_with("capacity", |cfg| {
cfg.control_max_conns = 2;
});
// Two idle keep-alive connections occupy the whole plane.
let mut c1 = ts.control(None);
let mut c2 = ts.control(None);
assert_eq!(c1.get("/v1/health").status, 200);
assert_eq!(c2.get("/v1/health").status, 200);
// The third connection is refused immediately and explicitly — no silent
// backlog starvation behind idle keep-alives.
let resp = raw_roundtrip(
ts.server.control_addr(),
b"GET /v1/health HTTP/1.1\r\nHost: x\r\n\r\n",
);
assert_eq!(status_of(&resp), 503);
// The occupied connections still work, and freeing one readmits others.
assert_eq!(c1.get("/v1/health").status, 200);
drop(c2);
let deadline = std::time::Instant::now() + std::time::Duration::from_secs(5);
loop {
let resp = raw_roundtrip(
ts.server.control_addr(),
b"GET /v1/health HTTP/1.1\r\nHost: x\r\nConnection: close\r\n\r\n",
);
if status_of(&resp) == 200 {
break;
}
assert!(std::time::Instant::now() < deadline, "slot never freed");
std::thread::sleep(std::time::Duration::from_millis(20));
}
}
// ---------------------------------------------------------------------------
// discovery hostility
// ---------------------------------------------------------------------------
#[test]
fn discovery_listener_survives_hostile_floods() {
let mut listener = DiscoveryListener::start(0, 60_000).expect("listener");
let port = listener.port();
let sock = UdpSocket::bind("127.0.0.1:0").unwrap();
let dest = ("127.0.0.1", port);
// Garbage of every shape: ignored.
sock.send_to(b"", dest).unwrap();
sock.send_to(b"short", dest).unwrap();
sock.send_to(&[0u8; 36], dest).unwrap();
sock.send_to(&[0xffu8; 512], dest).unwrap();
let mut zero_port = Beacon {
protocol_version: PROTOCOL_VERSION,
server_id: [1; 16],
control_port: 0,
data_port: 1,
auth_required: true,
tls: false,
capability_bits: caps::ALL_V1,
};
sock.send_to(&zero_port.encode(), dest).unwrap();
zero_port.control_port = 1;
zero_port.protocol_version = 0;
sock.send_to(&zero_port.encode(), dest).unwrap();
// A flood of distinct identities cannot grow the cache past its bound.
for i in 0..300u32 {
let mut id = [0u8; 16];
id[..4].copy_from_slice(&i.to_be_bytes());
let b = Beacon {
protocol_version: PROTOCOL_VERSION,
server_id: id,
control_port: 9701,
data_port: 9702,
auth_required: true,
tls: false,
capability_bits: caps::ALL_V1,
};
sock.send_to(&b.encode(), dest).unwrap();
}
// Give the receive thread time to drain the socket.
let deadline = std::time::Instant::now() + std::time::Duration::from_secs(5);
loop {
let n = listener.snapshot(makepad_asset_store::util::now_ms()).len();
assert!(n <= DiscoveryListener::MAX_ENTRIES, "cache bounded");
if n == DiscoveryListener::MAX_ENTRIES || std::time::Instant::now() >= deadline {
assert!(n > 0, "well-formed beacons were received");
break;
}
std::thread::sleep(std::time::Duration::from_millis(20));
}
listener.stop();
}