fix: channels parse-failure path writes the negotiation error frame (R4)

Review #002 R4 — a NegotiateRequest parse failure of the open op's
schema-validated input died silently (log + return, channel teardown,
consumer observed NoExitChunk — indistinguishable from a crashed
producer), while the other post-open failure classes (unknown backend,
allocate_failed, ownership denial) wrote the 0x00-prefixed error frame.

- make_tty_open_handler now accepts the channel's BiStream and writes
  a malformed_negotiation frame via the shared
  crate::adapter::send_negotiation_error (now pub(crate)) before
  returning; the consumer's M1 peek surfaces NegotiationRejected
  unchanged
- the frame type and layout are unchanged (ADR-001 wire-stable
  contract); no new frame type, no wire change
- tests: make_tty_open_handler seam test with a hand-built
  schema-bypassing input (cwd: 42) + a real-registry end-to-end test
  via ChannelClient::open_channel (bypasses open_via_channels's local
  fail-fast parse — R5's path — so it exercises the producer handler)
- docs: ADR-009 amended (Parse-failure error frame section);
  tty-adapter.md malformed_negotiation row covers both paths;
  session.rs post-open failure lists updated; review #002 R4 resolved

Note: the review's "unreachable end-to-end" premise was refined —
open_via_channels parses params locally (fail-fast) so a TtySession
consumer never hits the producer-side parse failure, but direct
ChannelClient callers do; the schema is deliberately partial so a
schema-valid value (cwd typed as a number) reaches the handler.

Verification: cargo test 95 lib (default) / 138 (--all-features);
clippy -D warnings native + wasm clean; fmt clean; doc 0 warnings.
This commit is contained in:
2026-09-05 08:37:02 +00:00
parent 5b608117ff
commit 8ee9216a07
7 changed files with 263 additions and 25 deletions
@@ -6,6 +6,11 @@ Accepted (2026-09-05). Resolves review #001 L1. Prerequisites: alkcall
0.4.0 (call-time `input_schema` enforcement) and alkcall 0.4.1
(early-arrival parking for un-adopted channels).
Amended 2026-09-05 (review #002 R4): a `NegotiateRequest` parse failure
of the schema-validated `input` is now a client-visible
`malformed_negotiation` error frame, not a silent teardown — see
§"Parse-failure error frame (R4 amendment, 2026-09-05)".
## Context
Before this ADR, the channels path carried the negotiation twice. The
@@ -73,9 +78,10 @@ so the consumer's ADR-001 §5 disambiguation read applies unchanged.
A `NegotiateRequest` parse failure of the schema-validated `input`
(the schema is deliberately partial — `carriage`/`backend`/`cmd`
required, `tty`/`cwd`/`env`/backend-params free-form so the opaque
ADR-053 params pass through) is a handler-side failure; the handler
logs and returns without writing an error frame (the channel is torn
down by the wrapper's teardown task).
ADR-053 params pass through) was originally a handler-side failure
logged and returned without an error frame. The R4 amendment below
makes it a client-visible error frame like every other post-open
failure.
### Consumer side
@@ -102,6 +108,37 @@ prefix starting `0x00`) from a raw chunk (`stream_type` in `{1, 2, 4}`).
This ADR removes a frame from the channels data stream; it does not
change any frame that remains.
### Parse-failure error frame (R4 amendment, 2026-09-05)
All three post-open failure classes on the channels path are now
client-visible through the same `0x00` error-frame peek
(`from_halves_raw`):
1. **Semantic validation failures** (`carriage != "raw"`, empty `cmd`,
unknown backend, `allocate_failed`, ownership denial) — error
frames from `validate_and_allocate` (unchanged).
2. **`NegotiateRequest` parse failure of schema-valid `input`** — the
open handler accepts the channel's `BiStream` and writes a
`malformed_negotiation` frame
(`{"error":"malformed_negotiation","message":"..."}`) via the shared
`crate::adapter::send_negotiation_error`, then returns. Reachable
despite the registry's schema check because the schema is
deliberately partial: e.g. `cwd` typed as a number passes the schema
(unknown-key/type fields pass through for the opaque ADR-053
params) but fails the typed parse.
3. **Schema-invalid `input`** — rejected at dispatch by the registry
(alkcall 0.4) before any handler runs: a `CallError` on the open
op, no channel allocated, no error frame (unchanged).
This replaces the original behavior (log + return, channel teardown,
consumer observes `NoExitChunk` — indistinguishable from a crashed
producer). The error-frame layout is unchanged (ADR-001's wire-stable
contract); no new frame type, no new stream type. Tests: the
`make_tty_open_handler` seam test (hand-built `input` bypassing the
schema, `channels.rs::tests`) and the real-registry end-to-end test
(`testing.rs`); both assert the consumer-side `0x00` peek observes
`malformed_negotiation`.
## Consequences
**Positive:**
+1 -1
View File
@@ -175,7 +175,7 @@ entering raw mode. The error response shape:
| Error | When | Shape |
|-------|------|------|
| `unknown_backend` | the `backend` string is not in the adapter's backend map | `{"error":"unknown_backend","backend":"..."}` |
| `malformed_negotiation` | the negotiation frame failed to parse as JSON or failed `NegotiateRequest` validation | `{"error":"malformed_negotiation","message":"..."}` |
| `malformed_negotiation` | the negotiation frame failed to parse as JSON or failed `NegotiateRequest` validation — on the direct path the wire frame, on the channels path the open op's `input` (schema-valid values can still fail the typed parse, e.g. `cwd` typed as a number, because the schema is deliberately partial) | `{"error":"malformed_negotiation","message":"..."}` |
| `allocate_failed` | `backend.allocate()` returned a `TtyError` | `{"error":"allocate_failed","message":"..."}` |
After sending the error response, the adapter closes the write half of
+22 -5
View File
@@ -175,6 +175,22 @@ Option 1 is preferred if a test can be written for it (feed a handler
a schema-bypassing value directly — the unit-testable seam is
`make_tty_open_handler` with a hand-built `input`).
**Resolution (2026-09-05)**: option 1 implemented. `make_tty_open_handler`
accepts the channel's `BiStream` and writes a `malformed_negotiation`
error frame via the shared `crate::adapter::send_negotiation_error`
(now `pub(crate)`) before returning — all three post-open failure
classes are client-visible through the unchanged M1 peek. The
review's reachability analysis is refined by the R5 fail-fast
discovery: `TtySession::open_via_channels` parses `params` locally
before opening, so the consumer never sends a value it can't parse
itself — but the producer-side handler is still the reachable seam for
direct `ChannelClient` callers (the schema is deliberately partial; a
schema-valid `cwd: 42` fails the typed parse). Tests: the
`make_tty_open_handler` seam test (channels.rs) and a real-registry
end-to-end test via `ChannelClient::open_channel` (testing.rs). ADR-009
amended (§"Parse-failure error frame"); `tty-adapter.md` error table
updated.
---
### R5. `open_via_channels` fail-fast surfaces as `NegotiationSerialize`
@@ -257,13 +273,14 @@ version-skew note in ADR-009); option 1 at that point.
| R1 | ADR-009 never written | write the ADR | small | none | ✅ resolved (`37ae07a`) |
| R2 | stale docs from the L1 redesign | align with ADR-009 | trivial | none | ✅ resolved (`37ae07a`) |
| R3 | install-time identity snapshot | accepted design (hub-proxy rationale) | none | none | ✅ closed as intended |
| R4 | silent death on parse-failure path | error frame or documented asymmetry | small | low | ⬜ open (deferred) |
| R4 | silent death on parse-failure path | error frame or documented asymmetry | small | low | ✅ resolved (option 1) |
| R5 | `NegotiationSerialize` mislabel on fail-fast | additive variant (with `#[non_exhaustive]` decision) | small | medium (semver) | ⬜ open (deferred) |
R4/R5 are deferred deliberately: both touch the consumer-facing error
surface, both are cheap, and neither is reachable-by-design today.
Batch them with the first post-1.0 API decision rather than churning
the error enum before a consumer exists.
R5 is deferred deliberately: it touches the consumer-facing error
surface, it is cheap, and the producer-side parse-failure arm (R4's
concern) is now client-visible regardless. Batch it with the first
post-1.0 API decision rather than churning the error enum before a
consumer exists.
## Notes
+8 -1
View File
@@ -168,7 +168,14 @@ fn has_scope(identity: &Option<Identity>, scope: &str) -> bool {
/// Send a negotiation error frame and close the write half. Consumes the
/// writer so the underlying transport's shutdown runs after the frame is
/// flushed.
async fn send_negotiation_error<W: AsyncWrite + Unpin>(
///
/// Shared by the direct path's failure arms and the channels open
/// handler (`make_tty_open_handler` writes the same `malformed_negotiation`
/// frame on a `NegotiateRequest` parse failure of the open op's `input` —
/// the pre-negotiated driver has no writer to consume, so the handler
/// builds its own over the channel's write half). `pub(crate)` so the
/// channels handler reuses it; not public API.
pub(crate) async fn send_negotiation_error<W: AsyncWrite + Unpin>(
mut writer: NegotiationWriter<W>,
error: &str,
fields: &[(&str, &str)],
+121 -7
View File
@@ -43,7 +43,7 @@ use tracing::debug;
use crate::adapter::{drive_session_pre_negotiated, TTY_OPEN_SCOPE};
use crate::backend::TtyBackend;
use crate::negotiation::NegotiateRequest;
use crate::negotiation::{NegotiateRequest, NegotiationWriter};
/// The per-ALPN open operation name (`channels/<alpn>/sub` convention,
/// ADR-047). TTY is consumer-opens (the client requests a shell), so
@@ -184,12 +184,27 @@ pub fn tty_open_spec() -> OperationSpec {
/// three-pump session driver as the direct-ALPN path, minus the
/// wire-frame negotiation phase.
///
/// Parse or validation failures (unknown backend, `allocate_failed`,
/// ADR-050 ownership denial) are reported to the client as a
/// negotiation error frame on the channel stream (the same
/// `0x00`-prefixed framing the direct path uses), so a consumer's
/// negotiation-error disambiguation read applies unchanged. The
/// `tty:open` scope gate is enforced by the registry's `AccessControl`
/// All failure classes write a negotiation error frame on the channel
/// stream (the same `0x00`-prefixed framing the direct path uses), so a
/// consumer's negotiation-error disambiguation read applies unchanged:
///
/// - The registry rejects schema-invalid `input` at dispatch (alkcall
/// 0.4) — the open op fails, no channel is ever allocated, and the
/// failure is a `CallError` on the open op, not a channel-stream
/// error frame. But the schema is deliberately partial (the opaque
/// ADR-053 backend params pass through), so a schema-valid value can
/// still fail the full `NegotiateRequest` parse — e.g. `cwd` typed
/// as a number — and that failure is post-open: the handler writes a
/// `malformed_negotiation` error frame (the same
/// [`crate::adapter::send_negotiation_error`] the other failure
/// classes use) and the consumer's peek surfaces
/// `NegotiationRejected`.
/// - Semantic validation failures (unknown backend, `carriage != "raw"`,
/// empty `cmd`, `allocate_failed`, ADR-050 ownership denial) are
/// post-open error frames written by `validate_and_allocate` inside
/// [`crate::adapter::drive_session_pre_negotiated`].
///
/// The `tty:open` scope gate is enforced by the registry's `AccessControl`
/// before this handler runs — the handler does not re-check it.
///
/// The handler's `JoinHandle` is recorded by the channels wrapper for
@@ -214,6 +229,20 @@ fn make_tty_open_handler(
Ok(r) => r,
Err(e) => {
debug!("tty: channels open: invalid negotiate params: {e}");
let stream = match channel_conn.accept_bi().await {
Ok(s) => s,
Err(e) => {
debug!("tty: channels open: accept_bi failed: {e}");
return;
}
};
let (_, mut client_write) = tokio::io::split(stream);
crate::adapter::send_negotiation_error(
NegotiationWriter::new(&mut client_write),
"malformed_negotiation",
&[("message", &e.to_string())],
)
.await;
return;
}
};
@@ -250,6 +279,91 @@ mod tests {
use std::collections::HashMap as StdHashMap;
use tokio::io::duplex;
/// The R4 unit-testable seam: feed `make_tty_open_handler` a
/// hand-built `input` the (partial) schema would accept but the full
/// `NegotiateRequest` parse rejects (`cwd` typed as a number). The
/// handler must write a `0x00`-prefixed `malformed_negotiation` error
/// frame on the channel stream — the same failure class as the
/// post-open semantic failures, not a silent teardown. The read side
/// is the consumer's exact post-open sequence (`from_halves_raw`):
/// peek the first byte, see `0x00`, read the length-prefixed error
/// frame.
#[tokio::test]
async fn open_handler_writes_error_frame_on_schema_bypassing_input() {
use alkcall::core::types::{BiStream, BidiStreamSource, StreamError};
use tokio::io::AsyncReadExt;
struct YieldOnce {
stream: tokio::sync::Mutex<Option<BiStream>>,
}
#[async_trait::async_trait]
impl BidiStreamSource for YieldOnce {
async fn accept_bi(&self) -> Result<BiStream, StreamError> {
self.stream
.lock()
.await
.take()
.ok_or(StreamError::ConnectionClosed)
}
async fn open_bi(&self) -> Result<BiStream, StreamError> {
Err(StreamError::StreamClosed)
}
fn remote_addr(&self) -> Option<std::net::SocketAddr> {
None
}
fn close(&self, _code: u32, _reason: &str) {
// The take-on-drop of the wrapped stream is not needed
// for this test; the connection is never closed.
}
}
let (client_end, server_end) = duplex(64 * 1024);
let (server_read, server_write) = tokio::io::split(server_end);
let bidi = BiStream::from_joined(server_read, server_write);
let channel_conn = Connection::from_source(
YieldOnce {
stream: tokio::sync::Mutex::new(Some(bidi)),
},
TTY_ALPN.as_bytes().to_vec(),
);
let mut backends: HashMap<String, Arc<dyn TtyBackend>> = HashMap::new();
backends.insert("mock".to_string(), Arc::new(MockBackend::with_exit_code(0)));
let handler = make_tty_open_handler(Arc::new(backends), None, None);
let task = handler(
serde_json::json!({
"carriage": "raw",
"backend": "mock",
"cmd": ["true"],
"cwd": 42
}),
channel_conn,
AuthContext::anonymous(b"test"),
);
task.await.expect("handler task");
let mut read = client_end;
let mut first = [0u8; 1];
tokio::time::timeout(
std::time::Duration::from_secs(5),
read.read_exact(&mut first),
)
.await
.expect("no byte from handler")
.expect("read first byte");
assert_eq!(first[0], 0x00, "error frame length prefix starts with 0x00");
let mut len_rest = [0u8; 3];
read.read_exact(&mut len_rest).await.expect("read len rest");
let len = u32::from_be_bytes([first[0], len_rest[0], len_rest[1], len_rest[2]]) as usize;
let mut body = vec![0u8; len];
read.read_exact(&mut body).await.expect("read error body");
let v: serde_json::Value = serde_json::from_slice(&body).expect("parse error frame");
assert_eq!(v["error"], "malformed_negotiation");
assert!(v["message"].as_str().is_some_and(|m| !m.is_empty()));
}
#[test]
fn op_tty_open_is_channels_tty_sub() {
assert_eq!(OP_TTY_OPEN, "channels/tty/sub");
+11 -8
View File
@@ -196,11 +196,13 @@ impl TtySession {
///
/// Failures before the channel opens (ACL denial, unknown op,
/// channel cap, invalid params) surface as
/// [`TtySessionError::ChannelsOpen`]. Post-open failures (unknown
/// backend, allocate failure, ownership denial) arrive as a
/// negotiation error frame on the channel stream — the session
/// surfaces those as [`TtySessionError::NegotiationRejected`] via
/// the same `0x00` disambiguation read the direct path uses.
/// [`TtySessionError::ChannelsOpen`]. Post-open failures (a
/// `NegotiateRequest` parse failure of a schema-valid-but-unparseable
/// params value, unknown backend, allocate failure, ownership
/// denial) arrive as a negotiation error frame on the channel
/// stream — the session surfaces those as
/// [`TtySessionError::NegotiationRejected`] via the same `0x00`
/// disambiguation read the direct path uses.
pub async fn open_via_channels(
client: &ChannelClient,
params: serde_json::Value,
@@ -292,9 +294,10 @@ impl TtySession {
/// Core inner for the channels path (ADR-009): the negotiation
/// already happened in the open op — the stream is already in
/// raw-chunk mode. The peek still applies: the producer sends a
/// `0x00`-prefixed error frame on post-open failures (unknown
/// backend, allocate failure, ownership denial), and a raw chunk
/// (`stream_type` in `{1, 2, 4}`) on success.
/// `0x00`-prefixed error frame on any post-open failure (a
/// `NegotiateRequest` parse failure of the open op's `input`,
/// unknown backend, allocate failure, ownership denial), and a raw
/// chunk (`stream_type` in `{1, 2, 4}`) on success.
async fn from_halves_raw<R, W>(read: R, write: W) -> Result<Self, TtySessionError>
where
R: AsyncRead + Send + Unpin + 'static,
+60
View File
@@ -206,4 +206,64 @@ mod tests {
assert_eq!(first[0], crate::wire::STREAM_STDOUT);
drop(send);
}
/// End-to-end (R4): `input` that passes the registry's (partial)
/// schema but fails the handler's full `NegotiateRequest` parse —
/// `cwd` typed as a number. The open op succeeds (the schema is
/// deliberately partial), the producer-side handler writes a
/// `0x00`-prefixed `malformed_negotiation` error frame on the
/// channel stream, and the consumer-side post-open read sequence
/// (`from_halves_raw`'s peek + error-frame parse) observes it — the
/// same failure class as the post-open semantic failures, not a
/// silent teardown. Uses `ChannelClient::open_channel` directly
/// (the consumer's local fail-fast parse in `open_via_channels`
/// would reject these params before the open op runs — R5's
/// fail-fast path).
#[tokio::test]
async fn schema_valid_but_unparseable_input_gets_malformed_negotiation_frame() {
use tokio::io::AsyncReadExt;
let mut backends: HashMap<String, Arc<dyn TtyBackend>> = HashMap::new();
backends.insert("mock".to_string(), Arc::new(MockBackend::with_exit_code(0)));
let client =
wire_client_and_server(Arc::new(backends), None, Some(tty_identity("alice"))).await;
let (_channel_id, _send, mut recv) = tokio::time::timeout(
std::time::Duration::from_secs(5),
client.open_channel(
OP_TTY_OPEN,
serde_json::json!({
"carriage": "raw",
"backend": "mock",
"cmd": ["true"],
"cwd": 42
}),
crate::channels::TTY_ALPN,
),
)
.await
.expect("open_channel timed out")
.expect("open_channel — the partial schema accepts cwd: 42");
// The consumer-side post-open sequence (`from_halves_raw`):
// peek the first byte; `0x00` = negotiation error frame.
let mut first = [0u8; 1];
tokio::time::timeout(
std::time::Duration::from_secs(5),
recv.read_exact(&mut first),
)
.await
.expect("no response byte from producer")
.expect("read first byte");
assert_eq!(first[0], 0x00, "error frame length prefix starts with 0x00");
let mut len_rest = [0u8; 3];
recv.read_exact(&mut len_rest).await.expect("read len rest");
let len = u32::from_be_bytes([first[0], len_rest[0], len_rest[1], len_rest[2]]) as usize;
let mut body = vec![0u8; len];
recv.read_exact(&mut body).await.expect("read error body");
let v: serde_json::Value = serde_json::from_slice(&body).expect("parse error frame");
assert_eq!(v["error"], "malformed_negotiation");
assert!(v["message"].as_str().is_some_and(|m| !m.is_empty()));
}
}