feat(daemon): alarms by name + RequestControlState + WatchHealth (Phase A complete)

A2 — alarms: optional 'name:' on alarm config (a LOCAL key — SEMI only
defines numeric ALID + freetext ALTX; field appended last so existing
{id, text, category} brace-inits compile unchanged), parsed by the loader,
checked by the validator, shipped in equipment.yaml. SetAlarm/ClearAlarm
RPCs resolve config name OR stringified ALID via a constructor snapshot.

A3 — control state + health: RequestControlState fires operator events on
the io thread (read_sync) and reports what the E30 table actually did —
ACCEPT iff the equipment landed in the requested state, CANNOT_DO_NOW naming
the actual state otherwise (the shipped table has no operator path to
EquipmentOffline; the test pins that honesty). ATTEMPT_ONLINE is rejected as
transient. WatchHealth streams an immediate snapshot then pushes on link/
control-state changes via service observers (add_link_observer +
add_control_state_observer — the HandlerSlot work paying off), spool depth
sampled at the 500ms poll; ends on cancel or engine stop.

Tests: daemon suite 61 -> 101 assertions (alarm lifecycle by name/id/unknown,
WatchHealth initial + change push, all four RequestControlState semantics);
loader test for the alarm name (present + absent fallback); core 467/3055.
Interop now 15 checks incl. gRPC SetAlarm -> host receives S5F1 ALCD=0x84
ALID=1, and RequestControlState(HOST_OFFLINE) -> GetControlState confirms.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
2026-06-10 19:47:31 +02:00
parent 83593bb508
commit 1da56f973f
9 changed files with 326 additions and 15 deletions
+157 -2
View File
@@ -12,8 +12,13 @@
#include <grpcpp/grpcpp.h>
#include <atomic>
#include <chrono>
#include <condition_variable>
#include <cstdint>
#include <map>
#include <mutex>
#include <optional>
#include <string>
#include <utility>
#include <vector>
@@ -155,8 +160,9 @@ inline pb::ControlState::State to_proto_state(gem::ControlState s) {
class EquipmentService final : public pb::Equipment::Service {
public:
// Snapshots the (immutable) name->id/format dictionaries. Construct before
// run_async() so the model is read while the io thread isn't running yet.
// Snapshots the (immutable) name->id/format dictionaries and registers the
// health observers. Construct before run_async() so the model is read (and
// observers land) while the io thread isn't running yet.
explicit EquipmentService(gem::EquipmentRuntime& rt) : rt_(rt) {
for (const auto& sv : rt.model().svids.all())
vars_.insert({sv.name, {sv.id, sv.value.format()}});
@@ -164,6 +170,21 @@ class EquipmentService final : public pb::Equipment::Service {
vars_.insert({dv.name, {dv.id, dv.value.format()}});
for (const auto& ev : rt.model().events.all_events())
events_.insert({ev.name, ev.id});
for (const auto& al : rt.model().alarms.all()) {
if (!al.name.empty()) alarms_.insert({al.name, al.id});
alarms_.insert({std::to_string(al.id), al.id}); // always addressable by id
}
// Health signals: bump a version + wake WatchHealth streams whenever the
// HSMS link or the control state changes (observers fire on the io thread;
// add_ observers survive register_default_handlers' primary set_).
rt.add_link_observer([this](bool selected) {
link_selected_.store(selected, std::memory_order_relaxed);
bump_health();
});
rt.add_control_state_observer(
[this](gem::ControlState, gem::ControlState, gem::ControlEvent) {
bump_health();
});
}
grpc::Status SetVariables(grpc::ServerContext*, const pb::VariableUpdate* req,
@@ -214,6 +235,16 @@ class EquipmentService final : public pb::Equipment::Service {
return grpc::Status::OK;
}
grpc::Status SetAlarm(grpc::ServerContext*, const pb::Alarm* req,
pb::Ack* resp) override {
return alarm_action(req->name(), /*set=*/true, resp);
}
grpc::Status ClearAlarm(grpc::ServerContext*, const pb::Alarm* req,
pb::Ack* resp) override {
return alarm_action(req->name(), /*set=*/false, resp);
}
grpc::Status GetVariables(grpc::ServerContext*, const pb::VariableQuery* req,
pb::VariableSnapshot* resp) override {
// Resolve names against the snapshot maps (empty query = everything).
@@ -258,7 +289,123 @@ class EquipmentService final : public pb::Equipment::Service {
return grpc::Status::OK;
}
// Operator-panel control-state transitions (e.g. "offline for maintenance").
// Fires the operator events on the io thread and reports what the E30 table
// actually did: ACCEPT iff the equipment landed in the requested state,
// CANNOT_DO_NOW (naming the actual state) otherwise. Note the shipped table
// has no operator path to EQUIPMENT_OFFLINE — operator_offline lands
// HOST_OFFLINE — so honesty matters here.
grpc::Status RequestControlState(grpc::ServerContext*,
const pb::ControlStateRequest* req,
pb::Ack* resp) override {
using PS = pb::ControlState;
const auto desired = req->desired();
if (desired == PS::ATTEMPT_ONLINE) {
resp->set_code(pb::Ack::PARAMETER_INVALID);
resp->set_message("ATTEMPT_ONLINE is transient; request a settled state");
return grpc::Status::OK;
}
auto final_state = rt_.read_sync([this, desired]() {
auto& sm = rt_.control();
switch (desired) {
case PS::ONLINE_LOCAL:
if (sm.state() == gem::ControlState::OnlineRemote) sm.operator_local();
else if (!sm.online()) sm.operator_online(); // table chains to OnlineLocal
break;
case PS::ONLINE_REMOTE:
if (sm.state() == gem::ControlState::OnlineLocal) sm.operator_remote();
else if (!sm.online()) {
sm.operator_online();
sm.operator_remote();
}
break;
case PS::HOST_OFFLINE:
case PS::EQUIPMENT_OFFLINE:
sm.operator_offline();
break;
default:
break;
}
return sm.state();
});
if (!final_state) {
resp->set_code(pb::Ack::CANNOT_DO_NOW);
resp->set_message("engine io thread did not answer (not running?)");
return grpc::Status::OK;
}
if (to_proto_state(*final_state) == desired) {
resp->set_code(pb::Ack::ACCEPT);
} else {
resp->set_code(pb::Ack::CANNOT_DO_NOW);
resp->set_message(std::string("equipment is in ") +
gem::control_state_name(*final_state));
}
return grpc::Status::OK;
}
// Streams a Health snapshot immediately, then again whenever the link or
// control state changes (and on spool-depth changes, sampled at the poll
// interval). Ends when the client cancels or the engine stops.
grpc::Status WatchHealth(grpc::ServerContext* ctx, const pb::Empty*,
grpc::ServerWriter<pb::Health>* writer) override {
bool first = true;
pb::Health last;
while (!ctx->IsCancelled()) {
const uint64_t seen = health_version_.load(std::memory_order_relaxed);
auto snap = make_health();
if (!snap) break; // engine stopped — end the stream
if (first || snap->link() != last.link() ||
snap->control_state() != last.control_state() ||
snap->spool_depth() != last.spool_depth()) {
if (!writer->Write(*snap)) break;
last = *snap;
first = false;
}
std::unique_lock<std::mutex> lk(health_mu_);
health_cv_.wait_for(lk, std::chrono::milliseconds(500), [&] {
return health_version_.load(std::memory_order_relaxed) != seen;
});
}
return grpc::Status::OK;
}
private:
void bump_health() {
health_version_.fetch_add(1, std::memory_order_relaxed);
health_cv_.notify_all();
}
// Build a Health snapshot. Control state + link come from atomics; spool
// depth is mutable engine state, read via read_sync. nullopt = engine down.
std::optional<pb::Health> make_health() {
auto depth = rt_.read_sync(
[this]() { return static_cast<uint32_t>(rt_.model().spool.size()); });
if (!depth) return std::nullopt;
pb::Health h;
// CONNECTED (TCP up, not yet SELECTED) is reserved: the runtime's link
// observer fires on SELECTED/closed only. TODO(daemon): surface the
// intermediate state if a tool ever needs it.
h.set_link(link_selected_.load(std::memory_order_relaxed)
? pb::Health::SELECTED
: pb::Health::DISCONNECTED);
h.set_spool_depth(*depth);
h.set_control_state(to_proto_state(rt_.control_state()));
return h;
}
grpc::Status alarm_action(const std::string& name, bool set, pb::Ack* resp) {
auto it = alarms_.find(name);
if (it == alarms_.end()) {
resp->set_code(pb::Ack::PARAMETER_INVALID);
resp->set_message("no alarm named '" + name + "'");
return grpc::Status::OK;
}
if (set) rt_.set_alarm(it->second);
else rt_.clear_alarm(it->second);
resp->set_code(pb::Ack::ACCEPT);
return grpc::Status::OK;
}
struct VarRef {
uint32_t vid;
s2::Format format; // declared wire format from equipment.yaml
@@ -267,6 +414,14 @@ class EquipmentService final : public pb::Equipment::Service {
gem::EquipmentRuntime& rt_;
std::map<std::string, VarRef> vars_; // SVIDs + DVIDs (SVIDs win on clash)
std::map<std::string, uint32_t> events_; // CEID by name
std::map<std::string, uint32_t> alarms_; // ALID by name AND stringified id
// WatchHealth plumbing: observers (io thread) bump the version and wake the
// per-stream wait loops (gRPC threads).
std::atomic<bool> link_selected_{false};
std::atomic<uint64_t> health_version_{0};
std::mutex health_mu_;
std::condition_variable health_cv_;
};
} // namespace secsgem::daemon
+5 -3
View File
@@ -56,10 +56,12 @@ ceids:
- {id: 400, name: ControlJobExecuting} # E94 CJ entered Executing
- {id: 401, name: ControlJobCompleted} # E94 CJ entered Completed
# Reported on S5F5 / S5F1. `category` is the lower-7 of ALCD.
# Reported on S5F5 / S5F1. `category` is the lower-7 of ALCD. `name` is an
# optional local key for the daemon's name-based API (gRPC SetAlarm/ClearAlarm);
# it never goes on the wire — SEMI only defines numeric ALID + freetext text.
alarms:
- {id: 1, text: "Chiller Temp High", category: 4}
- {id: 2, text: "Door Open", category: 1}
- {id: 1, name: chiller_temp_high, text: "Chiller Temp High", category: 4}
- {id: 2, name: door_open, text: "Door Open", category: 1}
# Reported on S7F19. Body served by S7F5/F6.
recipes:
+18 -10
View File
@@ -29,7 +29,7 @@ host and stays conformant while the tool software restarts/upgrades/crashes.
| Daemon interop vs **secsgem-py** reference host | ✅ | `interop/daemon_interop.py` (via `gemd` compose service): gRPC `SetVariables(ChamberPressure=2.5)` + `FireEvent` → host receives `S6F11 CEID 300` carrying `<F4 2.5>` — value *and declared format* flow gRPC→engine→HSMS→host |
| Daemon interop vs **secs4j** (Java) | ⬜ | mirror the secsgem-py harness against `interop/secs4j` |
| `Subscribe` host→tool command stream | ⬜ | design settled (HCACK-4, see below); not implemented |
| Remaining universal RPCs (alarms, `RequestControlState`, `WatchHealth`) | | see plan (Phase A) |
| Universal RPC surface complete (vars/events/alarms/control-state/health) | | Phase A done; daemon tests 101 assertions, interop 15 checks |
| Python client package (the "beautiful API") | ⬜ | thin wrapper over generated stubs |
## Known issues (found in the 2026-06-10 audit; honest list)
@@ -39,10 +39,9 @@ host and stays conformant while the tool software restarts/upgrades/crashes.
observer (`HandlerSlot` primary+observers pattern), so the mirror survives
`register_default_handlers` claiming the primary slot. `control_state()` is
now safe from any thread.
- **Alarms have no name key.** `equipment.yaml` alarms carry only numeric
`id` + freetext `text` (matches SEMI: ALID/ALTX; there is no standard short
name). The name-based `SetAlarm`/`ClearAlarm` RPCs need an optional local
`name:` field in the alarm config (fallback: stringified id).
- ~~**Alarms have no name key.**~~ Optional `name:` added to the alarm
config (loader + validator + shipped equipment.yaml); daemon RPCs accept
the name or the stringified ALID.
-**`pvd_tool` predates the behaviour hook.** It still hard-codes
`if (rcmd=="START") recipe->start(...)` in a router handler. Migrate it to
`commands.set_handler` so the flagship example showcases the intended seam.
@@ -132,11 +131,20 @@ debts tax every later phase, and the most valuable tests aren't automated.
through the API, read back through the API — plus empty-query-returns-all,
INVALID_ARGUMENT on unknown names, and a live round-trip check in
`daemon_interop.py`.
2. Alarm `name:` config field + `SetAlarm`/`ClearAlarm` RPCs + tests.
3.`RequestControlState` (operator online/offline) + control-state atomic
mirror (fixes the known race) + `WatchHealth` stream (link state from the
selected/closed handlers, spool depth, control state).
4. ⬜ Extend `test_daemon_service.cpp` + `daemon_interop.py` for all of the above.
2. Alarm `name:` config field (optional local key; `name` appended LAST on
the Alarm struct so existing brace-inits compile unchanged) + `SetAlarm`/
`ClearAlarm` RPCs (addressable by config name AND stringified ALID).
Validated end-to-end: gRPC `SetAlarm(chiller_temp_high)` -> secsgem-py host
receives `S5F1 ALCD=0x84 ALID=1`.
3.`RequestControlState` — fires operator events on the io thread and
reports what the E30 table actually did (ACCEPT iff landed in the requested
state; the shipped table has NO operator path to EquipmentOffline and the
test pins that honesty). ✅ `WatchHealth` — initial snapshot + push on
link/control-state change (+ spool depth sampled at 500ms); unit-tested
incl. the change push; link state still SELECTED/DISCONNECTED only
(CONNECTED reserved, TODO in code). Interop covers RequestControlState;
WatchHealth external check rides with Phase B.
4. ✅ Done per-item above (daemon suite at 101 assertions; interop at 15 checks).
### Phase B — the command stream (the big one)
5. ⬜ Implement `Subscribe`/`CompleteCommand` per the design above, including
+5
View File
@@ -39,6 +39,11 @@ struct Alarm {
// Lower 7 bits of ALCD: severity bitmap (see AlarmSeverity). Bit 7
// marks set vs cleared and is applied at emit time.
uint8_t severity_category;
// Optional local key for name-based APIs (the gRPC daemon, future Python
// client). NOT on the wire — SEMI defines only numeric ALID + freetext
// ALTX. Last field so existing {id, text, category} brace-inits compile
// unchanged; empty = unnamed (address it by id).
std::string name;
bool has(AlarmSeverity bit) const { return has_severity(severity_category, bit); }
bool is_safety() const {
+38
View File
@@ -99,6 +99,8 @@ def run(grpc_addr: str, hsms_host: str, hsms_port: int, session_id: int,
ceid300 = threading.Event()
last_s6f11 = {}
s5f1_seen = threading.Event()
last_s5f1 = {}
def on_s6f11(_handler, message):
decoded = client.settings.streams_functions.decode(message)
@@ -110,7 +112,17 @@ def run(grpc_addr: str, hsms_host: str, hsms_port: int, session_id: int,
ceid300.set()
client.send_response(F.SecsS06F12(0), message.header.system)
def on_s5f1(_handler, message):
decoded = client.settings.streams_functions.decode(message)
body = decoded.get()
LOG.info("[alm] S5F1 body=%r", body)
if isinstance(body, dict):
last_s5f1.update(body)
s5f1_seen.set()
client.send_response(F.SecsS05F02(0), message.header.system)
client.register_stream_function(6, 11, on_s6f11)
client.register_stream_function(5, 1, on_s5f1)
client.enable()
try:
@@ -172,6 +184,32 @@ def run(grpc_addr: str, hsms_host: str, hsms_port: int, session_id: int,
for v in vals)
check("S6F11 report carries ChamberPressure=2.5 (value flowed end-to-end)",
near, f"scalars={vals}")
# --- alarms: host enables ALID 1 (S5F3), gRPC raises it BY NAME,
# host receives the unsolicited S5F1 with set-bit + ALID 1 ---
client.send_and_waitfor_response(
F.SecsS05F03({"ALED": 0x80, "ALID": 1}))
ack = stub.SetAlarm(pb.Alarm(name="chiller_temp_high"))
check("gRPC SetAlarm(chiller_temp_high) -> ACCEPT",
ack.code == pb.Ack.ACCEPT, pb.Ack.Code.Name(ack.code))
got_alarm = s5f1_seen.wait(timeout=10)
check("host received S5F1 (gRPC SetAlarm bridged to HSMS)", got_alarm)
if got_alarm:
check("S5F1 carries ALID 1 with the set bit",
last_s5f1.get("ALID") == 1
and (int(last_s5f1.get("ALCD") or 0) & 0x80) != 0,
f"body={last_s5f1}")
stub.ClearAlarm(pb.Alarm(name="chiller_temp_high"))
# --- operator-panel control state: take the tool offline via gRPC ---
ack = stub.RequestControlState(pb.ControlStateRequest(
desired=pb.ControlState.HOST_OFFLINE))
check("gRPC RequestControlState(HOST_OFFLINE) -> ACCEPT",
ack.code == pb.Ack.ACCEPT, pb.Ack.Code.Name(ack.code))
cs = stub.GetControlState(pb.Empty())
check("control state is HOST_OFFLINE after operator request",
cs.state == pb.ControlState.HOST_OFFLINE,
pb.ControlState.State.Name(cs.state))
finally:
client.disable()
+1
View File
@@ -206,6 +206,7 @@ EquipmentDescriptor load_equipment(const std::string& path, gem::EquipmentDataMo
static_cast<uint32_t>(a["id"].as<int>()),
a["text"].as<std::string>(),
static_cast<uint8_t>(a["category"].as<int>()),
a["name"] ? a["name"].as<std::string>() : "", // optional local key
});
}
}
+4
View File
@@ -218,6 +218,10 @@ void validate_equipment_block(YAML::Node& root, Sink& sink) {
// is the set/clear flag, set at emit time — must not be in YAML.
as_int_in_range<uint8_t>(als[i]["category"], sink,
path + ".category", 0, 127);
// Optional local key for name-based APIs (not on the wire). If
// present it must be non-empty.
if (als[i]["name"])
as_nonempty_string(als[i]["name"], sink, path + ".name");
if (id && !alarm_ids.insert(*id).second) {
sink.error(path + ".id",
"duplicate ALID " + std::to_string(*id), als[i]);
+76
View File
@@ -125,6 +125,30 @@ TEST_CASE("Equipment gRPC service over an in-process channel") {
check_all(rt.model().dvids.all());
}
SUBCASE("SetAlarm / ClearAlarm by config name, by stringified id, unknown rejected") {
auto call = [&](auto method, const std::string& name) {
grpc::ClientContext ctx;
pb::Alarm req;
pb::Ack resp;
req.set_name(name);
REQUIRE((stub.get()->*method)(&ctx, req, &resp).ok());
return resp.code();
};
// By config name.
CHECK(call(&pb::Equipment::Stub::SetAlarm, "chiller_temp_high") == pb::Ack::ACCEPT);
rt.poll();
CHECK(rt.model().alarms.active(1));
CHECK(call(&pb::Equipment::Stub::ClearAlarm, "chiller_temp_high") == pb::Ack::ACCEPT);
rt.poll();
CHECK_FALSE(rt.model().alarms.active(1));
// By stringified ALID — always works, even for unnamed alarms.
CHECK(call(&pb::Equipment::Stub::SetAlarm, "2") == pb::Ack::ACCEPT);
rt.poll();
CHECK(rt.model().alarms.active(2));
// Unknown name.
CHECK(call(&pb::Equipment::Stub::SetAlarm, "no_such_alarm") == pb::Ack::PARAMETER_INVALID);
}
SUBCASE("FireEvent accepts a known event and rejects an unknown one") {
{
grpc::ClientContext ctx;
@@ -205,6 +229,58 @@ TEST_CASE("GetVariables round-trip under run_async (production threading mode)")
CHECK(st.error_message().find("definitely_not_a_var") != std::string::npos);
}
SUBCASE("RequestControlState walks the E30 table; WatchHealth pushes the change") {
// Open the health stream first: the initial snapshot arrives immediately.
grpc::ClientContext health_ctx;
pb::Empty empty;
auto reader = stub->WatchHealth(&health_ctx, empty);
pb::Health h;
REQUIRE(reader->Read(&h));
CHECK(h.link() == pb::Health::DISCONNECTED); // no HSMS host in this test
CHECK(h.control_state() == pb::ControlState::HOST_OFFLINE);
CHECK(h.spool_depth() == 0);
auto request = [&](pb::ControlState::State desired) {
grpc::ClientContext ctx;
pb::ControlStateRequest req;
pb::Ack resp;
req.set_desired(desired);
REQUIRE(stub->RequestControlState(&ctx, req, &resp).ok());
return resp;
};
// HostOffline --operator_online--> (AttemptOnline) --> OnlineLocal.
CHECK(request(pb::ControlState::ONLINE_LOCAL).code() == pb::Ack::ACCEPT);
CHECK(rt.control_state() == gem::ControlState::OnlineLocal);
// The stream pushes the change (well inside its 500ms poll interval).
REQUIRE(reader->Read(&h));
CHECK(h.control_state() == pb::ControlState::ONLINE_LOCAL);
// OnlineLocal --operator_remote--> OnlineRemote.
CHECK(request(pb::ControlState::ONLINE_REMOTE).code() == pb::Ack::ACCEPT);
CHECK(rt.control_state() == gem::ControlState::OnlineRemote);
// Transient state is not requestable.
CHECK(request(pb::ControlState::ATTEMPT_ONLINE).code() == pb::Ack::PARAMETER_INVALID);
// The shipped table has NO operator path to EquipmentOffline:
// operator_offline lands HostOffline — the API must say so honestly.
auto resp = request(pb::ControlState::EQUIPMENT_OFFLINE);
CHECK(resp.code() == pb::Ack::CANNOT_DO_NOW);
CHECK(resp.message().find("HostOffline") != std::string::npos);
CHECK(rt.control_state() == gem::ControlState::HostOffline);
// Requesting HOST_OFFLINE (the state operators actually get) succeeds —
// idempotently, since we're already there.
CHECK(request(pb::ControlState::HOST_OFFLINE).code() == pb::Ack::ACCEPT);
health_ctx.TryCancel();
pb::Health drain;
while (reader->Read(&drain)) {}
(void)reader->Finish(); // CANCELLED — expected
}
server->Shutdown();
rt.stop();
}
+22
View File
@@ -1,5 +1,6 @@
#include <doctest/doctest.h>
#include <filesystem>
#include <fstream>
#include <string>
@@ -144,3 +145,24 @@ TEST_CASE("loader: control_job_state.yaml -> non-empty table") {
REQUIRE(row->to.has_value());
CHECK(*row->to == gem::ControlJobState::Completed);
}
TEST_CASE("alarm optional name: parsed when present, empty when absent") {
gem::EquipmentDataModel m;
config::load_equipment(SECSGEM_DATA_DIR "/equipment.yaml", m);
// data/equipment.yaml names both alarms.
CHECK(m.alarms.get(1)->name == "chiller_temp_high");
CHECK(m.alarms.get(2)->name == "door_open");
// A config without `name:` yields an empty name (back-compat).
const auto path = std::filesystem::temp_directory_path() / "alarm_noname.yaml";
{
std::ofstream f(path);
f << "device: {id: 0, model_name: M, software_rev: R}\n"
"alarms:\n"
" - {id: 9, text: \"Unnamed\", category: 1}\n";
}
gem::EquipmentDataModel m2;
config::load_equipment(path.string(), m2);
CHECK(m2.alarms.get(9)->name.empty());
std::filesystem::remove(path);
}