Skip to main content

harn_vm/vm/
state.rs

1use std::collections::{BTreeMap, HashMap, HashSet};
2use std::path::PathBuf;
3use std::sync::atomic::{AtomicBool, Ordering};
4use std::sync::{Arc, Mutex};
5use std::time::Instant;
6
7use crate::chunk::{Chunk, ChunkRef, Constant};
8use crate::runtime_limits::RuntimeLimits;
9use crate::value::{
10    ModuleFunctionRegistry, VmAsyncBuiltinFn, VmBuiltinFn, VmClosure, VmEnv, VmError, VmMutex,
11    VmTaskHandle, VmValue,
12};
13use crate::BuiltinId;
14
15use super::debug::DebugHook;
16pub(crate) use super::execution_deadline::{ExecutionDeadlinePauseGuard, ExecutionDeadlineState};
17use super::modules::ModuleCache;
18use super::VmBuiltinMetadata;
19
20/// A lazy callable's resolved export set together with the module graph that
21/// was loaded to produce it.
22///
23/// The exported closures — and every function they can transitively reach
24/// through imports — hold only `Weak`s into their home module's function
25/// registry and module state. The child VM that first loaded the graph is
26/// normally their sole strong owner (via its `module_cache`), and it dies once
27/// the hook fire completes. A later child VM that hits this cache never
28/// re-imports the graph, so a transitively imported callee's sibling `pub fn`
29/// would fall through name resolution to host-bridge dispatch (`Undefined
30/// builtin`). Retaining the complete loaded graph here for the cache entry's
31/// lifetime keeps every transitively reachable registry and module state
32/// upgradeable at call time, for every fire — not just the first.
33pub(crate) struct ResolvedLazyCallable {
34    pub(crate) exports: BTreeMap<String, Arc<VmClosure>>,
35    /// Intentionally unread: retained solely to keep the loaded module graph's
36    /// function registries and module states alive for this cache entry (the
37    /// same liveness role [`crate::value::RetainedModuleScope`] plays for a
38    /// single retained closure, generalized across the whole import graph).
39    #[allow(dead_code)]
40    pub(crate) retained_module_graph: ModuleCache,
41}
42
43pub(crate) type LazyCallableResolution = Arc<ResolvedLazyCallable>;
44pub(crate) struct LazyCallableCacheSlot {
45    pub(crate) execution_guard: Option<Arc<harn_modules::package_execution::PackageExecutionGuard>>,
46    pub(crate) resolution: Arc<tokio::sync::OnceCell<LazyCallableResolution>>,
47}
48pub(crate) type LazyCallableModuleCache =
49    Arc<VmMutex<BTreeMap<PathBuf, Vec<LazyCallableCacheSlot>>>>;
50
51/// RAII guard that starts a tracing span on creation and ends it on drop.
52pub(crate) struct ScopeSpan(u64);
53
54impl ScopeSpan {
55    pub(crate) fn new(kind: crate::tracing::SpanKind, name: String) -> Self {
56        Self(crate::tracing::span_start(kind, name))
57    }
58}
59
60impl Drop for ScopeSpan {
61    fn drop(&mut self) {
62        crate::tracing::span_end(self.0);
63    }
64}
65
66#[derive(Clone)]
67pub(crate) struct LocalSlot {
68    pub(crate) value: VmValue,
69    pub(crate) initialized: bool,
70    pub(crate) synced: bool,
71}
72
73impl Drop for LocalSlot {
74    fn drop(&mut self) {
75        // Slot locals hold script values directly (e.g. a `let` bound to a
76        // deeply nested list). When a frame is torn down, the default
77        // recursive drop of such a value would overflow the native stack and
78        // abort the process. For the overwhelmingly common scalar slot this is
79        // a single `matches!` check and then the normal trivial drop; only a
80        // nested container is moved out and torn down iteratively, so hot
81        // frame teardown is unaffected.
82        if crate::value::recursion::is_recursive_container(&self.value) {
83            crate::value::recursion::dismantle(std::mem::replace(&mut self.value, VmValue::Nil));
84        }
85    }
86}
87
88#[derive(Clone)]
89pub(crate) struct InterruptHandler {
90    pub(crate) handle: i64,
91    pub(crate) signals: Vec<String>,
92    pub(crate) once: bool,
93    pub(crate) graceful_timeout_ms: Option<u64>,
94    pub(crate) handler: VmValue,
95}
96
97/// Call frame for function execution.
98pub(crate) struct CallFrame {
99    pub(crate) chunk: ChunkRef,
100    /// VM-local inline-cache set for this frame's chunk. Computed once at
101    /// frame entry so hot opcode dispatch can index cache feedback directly
102    /// instead of hashing the chunk id on every cached opcode.
103    pub(crate) inline_cache_set: usize,
104    pub(crate) ip: usize,
105    pub(crate) stack_base: usize,
106    pub(crate) saved_env: VmEnv,
107    /// Env snapshot captured at call-time, *after* argument binding. Used
108    /// by the debugger's `restartFrame` to rewind this frame to its
109    /// entry state (re-binding args from the original values) without
110    /// re-entering the call site. Cheap to clone because `VmEnv` is
111    /// already cloned into `saved_env` on every call. `None` for
112    /// scratch frames (evaluate, import init) where restart isn't
113    /// meaningful.
114    pub(crate) initial_env: Option<VmEnv>,
115    pub(crate) initial_local_slots: Option<Vec<LocalSlot>>,
116    /// Iterator stack depth to restore when this frame unwinds.
117    pub(crate) saved_iterator_depth: usize,
118    /// Function name for stack traces (empty for top-level pipeline); a
119    /// refcount-bump share of [`CompiledFunction::name`].
120    pub(crate) fn_name: crate::value::HarnStr,
121    /// Number of arguments actually passed by the caller (for default arg support).
122    pub(crate) argc: usize,
123    /// Saved VM_SOURCE_DIR to restore when this frame is popped.
124    /// Set when entering a closure that originated from an imported module.
125    pub(crate) saved_source_dir: Option<std::path::PathBuf>,
126    /// Module-local named functions available to symbolic calls within this frame.
127    pub(crate) module_functions: Option<ModuleFunctionRegistry>,
128    /// Shared module-level env for top-level `let` / `const` bindings of
129    /// this frame's originating module. Looked up after `self.env` and
130    /// before `self.globals` by `GetVar` / `SetVar`, giving each module
131    /// its own live static state that persists across calls. See the
132    /// `module_state` field on `VmClosure` for the full rationale.
133    pub(crate) module_state: Option<crate::value::ModuleState>,
134    /// Slot-indexed locals for compiler-resolved names in this frame.
135    pub(crate) local_slots: Vec<LocalSlot>,
136    /// Env scope index that corresponds to compiler local scope depth 0.
137    pub(crate) local_scope_base: usize,
138    /// Current compiler local scope depth, updated by PushScope/PopScope.
139    pub(crate) local_scope_depth: usize,
140}
141
142pub(crate) struct InlineCacheSite {
143    pub(crate) cache_set: usize,
144    pub(crate) slot_count: usize,
145    pub(crate) slot: Option<usize>,
146}
147
148impl CallFrame {
149    #[inline]
150    pub(crate) fn inline_cache_site_for_previous_op(&self) -> InlineCacheSite {
151        let op_offset = self.ip.saturating_sub(1);
152        InlineCacheSite {
153            cache_set: self.inline_cache_set,
154            slot_count: self.chunk.inline_cache_slot_count(),
155            slot: self.chunk.inline_cache_slot(op_offset),
156        }
157    }
158}
159
160/// Exception handler for try/catch.
161pub(crate) struct ExceptionHandler {
162    pub(crate) catch_ip: usize,
163    pub(crate) stack_depth: usize,
164    pub(crate) frame_depth: usize,
165    pub(crate) env_scope_depth: usize,
166    /// When present, this catch only handles errors whose enum_name matches.
167    pub(crate) error_type: Option<crate::value::HarnStr>,
168}
169
170/// A structured-concurrency nursery (`scope { }`). Tasks spawned while this
171/// scope is innermost record their id here; `TaskScopeExit` joins them.
172pub(crate) struct TaskScope {
173    /// Ids of tasks spawned in this scope that have not been explicitly
174    /// `await`ed away. Joined (normal exit) or cancelled (unwind) on close.
175    pub(crate) task_ids: Vec<String>,
176    /// Frame depth at which the scope was opened, for unwind pruning.
177    pub(crate) frame_depth: usize,
178    /// Env scope depth at open, for unwind pruning.
179    pub(crate) env_scope_depth: usize,
180}
181
182/// Terminal exit requested by any VM in one execution tree. A process exit is
183/// global control flow, so child VMs share this latch with their parent rather
184/// than relying on a task being explicitly awaited.
185pub(crate) struct ProcessExitRequest {
186    code: Mutex<Option<i32>>,
187    requested: AtomicBool,
188}
189
190impl ProcessExitRequest {
191    fn new() -> Self {
192        Self {
193            code: Mutex::new(None),
194            requested: AtomicBool::new(false),
195        }
196    }
197
198    fn request(&self, code: i32) {
199        let mut recorded = self
200            .code
201            .lock()
202            .expect("process exit request lock poisoned");
203        if recorded.is_none() {
204            *recorded = Some(code);
205            self.requested.store(true, Ordering::Release);
206        }
207    }
208
209    fn code(&self) -> Option<i32> {
210        if !self.requested.load(Ordering::Acquire) {
211            return None;
212        }
213        *self
214            .code
215            .lock()
216            .expect("process exit request lock poisoned")
217    }
218}
219
220/// Iterator state for for-in loops.
221pub(crate) enum IterState {
222    Vec {
223        items: Arc<Vec<VmValue>>,
224        idx: usize,
225    },
226    Dict {
227        entries: Arc<crate::value::DictMap>,
228        keys: Vec<crate::value::HarnStr>,
229        idx: usize,
230    },
231    Channel {
232        receiver: std::sync::Arc<tokio::sync::Mutex<tokio::sync::mpsc::Receiver<VmValue>>>,
233        close: std::sync::Arc<crate::value::VmChannelCloseState>,
234    },
235    Generator {
236        gen: Arc<crate::value::VmGenerator>,
237    },
238    Stream {
239        stream: Arc<crate::value::VmStream>,
240    },
241    /// Step through a lazy range without materializing a Vec.
242    /// Inclusive ranges keep `end` as an actual value so `i64::MAX to i64::MAX`
243    /// still yields one item instead of overflowing a one-past-end sentinel.
244    Range {
245        next: i64,
246        end: i64,
247        inclusive: bool,
248        done: bool,
249    },
250    VmIter {
251        handle: crate::vm::iter::VmIterHandle,
252    },
253}
254
255#[derive(Clone)]
256pub(crate) enum VmBuiltinDispatch {
257    Sync(VmBuiltinFn),
258    Async(VmAsyncBuiltinFn),
259}
260
261#[derive(Clone)]
262pub(crate) struct VmBuiltinEntry {
263    pub(crate) name: Arc<str>,
264    pub(crate) dispatch: VmBuiltinDispatch,
265    /// Static effect specs for source-visible capability/privileged builtins.
266    /// Keeping this beside the numeric dispatch entry makes pure direct calls
267    /// pay one `Option` branch instead of consulting the manifest by name.
268    pub(crate) recorded_effects: Option<&'static [harn_builtin_meta::EffectSpec]>,
269}
270
271/// The Harn bytecode virtual machine.
272pub struct Vm {
273    pub(crate) stack: Vec<VmValue>,
274    pub(crate) env: VmEnv,
275    pub(crate) output: String,
276    pub(crate) builtins: Arc<BTreeMap<String, VmBuiltinFn>>,
277    pub(crate) async_builtins: Arc<BTreeMap<String, VmAsyncBuiltinFn>>,
278    /// Host-supplied implementations keyed by the typed capability surface
279    /// they implement. Unlike ordinary builtins these names never enter the
280    /// source namespace: they are reachable only through a `Harness*` value.
281    pub(crate) capability_methods:
282        Arc<BTreeMap<harn_builtin_meta::CapabilityId, BTreeMap<String, VmBuiltinDispatch>>>,
283    pub(crate) builtin_metadata: Arc<BTreeMap<String, VmBuiltinMetadata>>,
284    /// Numeric side index for builtins. Name-keyed maps remain authoritative;
285    /// this index is the hot path for direct builtin bytecode and callback refs.
286    pub(crate) builtins_by_id: Arc<HashMap<BuiltinId, VmBuiltinEntry>>,
287    /// IDs with detected name collisions. Collided names safely fall back to
288    /// the authoritative name-keyed lookup path.
289    pub(crate) builtin_id_collisions: Arc<HashSet<BuiltinId>>,
290    /// Iterator state for for-in loops.
291    pub(crate) iterators: Vec<IterState>,
292    /// Call frame stack.
293    pub(crate) frames: Vec<CallFrame>,
294    /// Exception handler stack.
295    pub(crate) exception_handlers: Vec<ExceptionHandler>,
296    /// Spawned async task handles.
297    pub(crate) spawned_tasks: BTreeMap<String, VmTaskHandle>,
298    /// Shared terminal process-exit latch for this execution tree.
299    pub(crate) process_exit_request: Arc<ProcessExitRequest>,
300    /// Shared process-local synchronization primitives inherited by child VMs.
301    pub(crate) sync_runtime: Arc<crate::synchronization::VmSyncRuntime>,
302    /// Shared process-local cells, maps, and mailboxes inherited by child VMs.
303    pub(crate) shared_state_runtime: Arc<crate::shared_state::VmSharedStateRuntime>,
304    /// Host-agent workers owned by this VM tree. Child interpreters share the
305    /// registry so delayed triggers and task migration address the same worker
306    /// state without exposing it to a sibling VM execution.
307    pub(crate) worker_registry: Arc<crate::stdlib::agents::agents_workers::WorkerRegistry>,
308    /// Managed agent daemons owned by this VM tree.
309    pub(crate) daemon_registry: Arc<crate::stdlib::agents_daemon::DaemonRegistry>,
310    /// Trigger bindings owned by this VM tree, including dynamic auto-resume
311    /// bindings that may be consumed on a different runtime worker.
312    pub(crate) trigger_registry: Arc<crate::triggers::registry::TriggerRegistryRuntime>,
313    /// Agent transcripts and subscribers owned by this VM tree.
314    pub(crate) session_runtime: Arc<crate::agent_sessions::AgentSessionRuntime>,
315    /// Structured spans owned by this VM tree.
316    pub(crate) tracing_runtime: Arc<crate::tracing::TracingRuntime>,
317    /// Host-side agent-loop state owned by this VM tree.
318    pub(crate) agent_host_session_runtime:
319        Arc<crate::llm::agent_session_host::AgentHostSessionRuntime>,
320    /// Connector clients and their on-demand resolver owned by this execution
321    /// tree. Keeping this explicit avoids thread-local loss under task
322    /// migration and prevents sibling VM runs from observing each other.
323    pub(crate) connector_clients: Arc<crate::connectors::VmConnectorClients>,
324    /// Per-isolate inline cache entries. `inline_cache_set_by_chunk` maps a
325    /// compiled chunk identity to an index in this vector at frame entry; the
326    /// dispatch loop uses the frame-local index for per-op reads/writes.
327    pub(crate) inline_cache_sets: Vec<Vec<crate::chunk::InlineCacheEntry>>,
328    pub(crate) inline_cache_set_by_chunk: HashMap<u64, usize>,
329    /// VM-scoped pool registry inherited by child VMs and scoped into Tokio tasks.
330    pub(crate) pool_registry: Arc<crate::stdlib::pool::PoolRegistry>,
331    /// Inline LLM fixtures and observations shared by this VM execution tree.
332    pub(crate) llm_mock_context: crate::llm::mock::LlmMockContext,
333    /// Reader leases opened by package_snapshot_open in this execution tree.
334    /// Child VMs share the registry; the final VM drop releases abandoned
335    /// leases without touching concurrent executions.
336    pub(crate) package_snapshot_registry: Arc<crate::stdlib::PackageSnapshotRegistry>,
337    /// Shared task/channel wait graph for this VM execution tree.
338    pub(crate) wait_for_graph: Arc<crate::wait_for_graph::VmWaitForGraph>,
339    /// Permits acquired by lexical synchronization blocks in this VM.
340    pub(crate) held_sync_guards: Vec<crate::synchronization::VmSyncHeldGuard>,
341    /// Locks held by an ancestor VM that is *suspended on this VM's execution*:
342    /// an inline async-builtin child runs while its parent is parked
343    /// mid-instruction still holding these permits. Re-acquiring more permits
344    /// than the primitive can grant is a provably-unresolvable self-deadlock, so
345    /// HARN-ORC-011 fires across the child boundary. Empty for new concurrent
346    /// tasks (`spawn`/`parallel`/triggers), where the parent keeps running and
347    /// blocking can be legitimately resolvable.
348    pub(crate) inherited_held_keys: Arc<Vec<crate::synchronization::VmSyncHeldKey>>,
349    /// Structured-concurrency nursery stack. Each `scope { }` block pushes a
350    /// `TaskScope`; tasks spawned while it is innermost register their id here.
351    /// On normal exit (`TaskScopeExit`) the scope's tasks are joined and the
352    /// first error propagates; on unwind they are cancelled. Modeled on
353    /// `held_sync_guards` (push on enter, prune/cancel on frame/handler exit).
354    pub(crate) task_scopes: Vec<TaskScope>,
355    /// Counter for generating unique task IDs.
356    pub(crate) task_counter: u64,
357    /// Counter for logical runtime-context task groups.
358    pub(crate) runtime_context_counter: u64,
359    /// Logical runtime task context visible through `runtime_context()`.
360    pub(crate) runtime_context: crate::runtime_context::RuntimeContext,
361    /// Active deadline stack: (deadline_instant, frame_depth).
362    pub(crate) deadlines: Vec<(Instant, usize)>,
363    /// Uncatchable wall-clock deadline imposed by the embedding host.
364    pub(crate) execution_deadline: Arc<ExecutionDeadlineState>,
365    /// Breakpoints, keyed by source-file path so a breakpoint at line N
366    /// in `auto.harn` doesn't also fire when execution hits line N in an
367    /// imported lib. The empty-string key is a wildcard used by callers
368    /// that don't track source paths (legacy `set_breakpoints` API).
369    pub(crate) breakpoints: BTreeMap<String, std::collections::BTreeSet<usize>>,
370    /// Function-name breakpoints. Any closure call whose
371    /// `CompiledFunction.name` matches an entry here raises a stop on
372    /// entry, regardless of the call site's file or line. Lets the IDE
373    /// break on `llm_call` / `host_run_pipeline` / any user pipeline
374    /// function without pinning down a source location first.
375    pub(crate) function_breakpoints: std::collections::BTreeSet<String>,
376    /// Latched on `push_closure_frame` when the callee's name matches
377    /// `function_breakpoints`; consumed by the next step so the stop is
378    /// reported with reason="function breakpoint" and the breakpoint
379    /// name available for the DAP `stopped` event.
380    pub(crate) pending_function_bp: Option<String>,
381    /// Whether the VM is in step mode.
382    pub(crate) step_mode: bool,
383    /// The frame depth at which stepping started (for step-over).
384    pub(crate) step_frame_depth: usize,
385    /// Whether the VM is currently stopped at a debug point.
386    pub(crate) stopped: bool,
387    /// Last source line executed (to detect line changes).
388    pub(crate) last_line: usize,
389    /// Source directory for resolving imports.
390    pub(crate) source_dir: Option<std::path::PathBuf>,
391    /// Installed-package identity retained for the active lazy pipeline.
392    pub(crate) package_execution_guard:
393        Option<Arc<harn_modules::package_execution::PackageExecutionGuard>>,
394    /// Modules currently being imported (cycle prevention).
395    pub(crate) imported_paths: Vec<std::path::PathBuf>,
396    /// Imports that hit an in-progress module (an import cycle) and so could
397    /// not bind inline. Drained by `flush_deferred_cyclic_imports` once the
398    /// involved modules finish loading.
399    pub(crate) deferred_cyclic_imports: Vec<super::modules::DeferredCyclicImport>,
400    /// Loaded module cache keyed by canonical or synthetic module path.
401    pub(crate) module_cache: ModuleCache,
402    /// Immutable hydrated module bytecode shared across fresh VM isolates.
403    /// Runtime closures, registries, state, and init execution are not cached.
404    pub(crate) prepared_module_cache: crate::PreparedModuleCache,
405    /// One graph-validity observation shared by this VM execution tree.
406    pub(crate) prepared_module_validation: crate::prepared_module::PreparedModuleValidation,
407    /// Authority provenance used for the next root module graph. This can only
408    /// move from ordinary user code to trusted host dispatch before loading.
409    pub(crate) module_provenance: crate::module_artifact::ModuleProvenance,
410    /// Optional timing recorder shared by this VM execution tree.
411    pub(crate) module_phase_recorder: Option<super::ModulePhaseRecorder>,
412    /// Lazy manifest modules initialized by any child in this execution tree.
413    /// The complete export set is retained so handlers and predicates from the
414    /// same module share one module state across repeated child invocations.
415    pub(crate) lazy_callable_modules: LazyCallableModuleCache,
416    /// Source text keyed by canonical or synthetic module path for debugger
417    /// retrieval. Entries share the bytes owned by
418    /// [`crate::module_source`] rather than copying them.
419    pub(crate) source_cache: Arc<BTreeMap<std::path::PathBuf, Arc<str>>>,
420    /// This spawn's import-graph link table, when the entry-chunk cache proved
421    /// one current. Lets module loading name a module's artifact from a recorded
422    /// digest instead of reading the file to recompute the same digest. `None`
423    /// means every module takes the ordinary read-and-hash path.
424    pub(crate) graph_link_table: Option<Arc<crate::context_manifest::GraphLinkTable>>,
425    /// Verified, closed-program module templates for this execution tree.
426    /// Unlike the generic graph link table, absence is never interpreted as a
427    /// source-cache miss by the linked-program path.
428    pub(crate) linked_program_repository:
429        Option<Arc<crate::linked_program::LinkedProgramRepository>>,
430    /// Source file path for error reporting.
431    pub(crate) source_file: Option<String>,
432    /// Source text for error reporting.
433    pub(crate) source_text: Option<String>,
434    /// Line-coverage accumulator. `Some` only while a coverage session is
435    /// active (see [`crate::coverage`]); folded into the global report on drop.
436    pub(crate) coverage: Option<crate::coverage::Coverage>,
437    /// Optional bridge for delegating unknown builtins in bridge mode.
438    pub(crate) bridge: Option<Arc<crate::bridge::HostBridge>>,
439    /// Builtins denied by sandbox mode (`--deny` / `--allow` flags).
440    pub(crate) denied_builtins: Arc<HashSet<String>>,
441    /// Cancellation token for cooperative graceful shutdown (set by parent).
442    pub(crate) cancel_token: Option<std::sync::Arc<std::sync::atomic::AtomicBool>>,
443    pub(crate) interrupt_signal_token: Option<std::sync::Arc<std::sync::Mutex<Option<String>>>>,
444    /// Remaining instruction-boundary checks before a requested host
445    /// cancellation is forcefully raised. This gives `is_cancelled()` loops a
446    /// deterministic chance to return cleanly without letting non-cooperative
447    /// CPU-bound code run forever.
448    pub(crate) cancel_grace_instructions_remaining: Option<usize>,
449    /// User-visible interrupt handlers registered through `std/signal`.
450    pub(crate) interrupt_handlers: Vec<InterruptHandler>,
451    pub(crate) next_interrupt_handle: i64,
452    pub(crate) pending_interrupt_signal: Option<String>,
453    pub(crate) interrupted: bool,
454    pub(crate) dispatching_interrupt: bool,
455    pub(crate) interrupt_handler_deadline: Option<Instant>,
456    /// Captured stack trace from the most recent error (fn_name, line, col).
457    pub(crate) error_stack_trace: Vec<(String, usize, usize, Option<String>)>,
458    /// Yield channel sender for generator execution. When set, `Op::Yield`
459    /// sends values through this channel instead of being a no-op.
460    pub(crate) yield_sender: Option<tokio::sync::mpsc::Sender<Result<VmValue, VmError>>>,
461    /// Project root directory (detected via harn.toml).
462    /// Used as base directory for metadata, store, and checkpoint operations.
463    pub(crate) project_root: Option<std::path::PathBuf>,
464    /// Global constants (e.g. `pi`, `e`). Checked as a fallback in `GetVar`
465    /// after the environment, so user-defined variables can shadow them.
466    pub(crate) globals: Arc<crate::value::DictMap>,
467    /// Root authority injected by the embedder for the compiler-generated
468    /// `main(harness: Harness)` call. This is deliberately not an environment
469    /// binding or global: user code can receive it only through an explicit
470    /// parameter.
471    pub(crate) root_harness: Option<VmValue>,
472    /// Runtime effect receipt state (shared recorder + per-VM call memo).
473    pub(crate) runtime_effects: crate::orchestration::RuntimeEffectState,
474    /// Optional debugger hook invoked when execution advances to a new source line.
475    pub(crate) debug_hook: Option<parking_lot::Mutex<Box<DebugHook>>>,
476    /// Effective runtime ceilings for this VM execution.
477    pub(crate) runtime_limits: RuntimeLimits,
478}
479
480/// Reusable VM baseline for hosts that need many clean executions with the
481/// same stable builtin/source setup.
482///
483/// The baseline intentionally does not snapshot execution state. Each
484/// instantiation gets fresh stacks, frames, tasks, cancellation fields, sync
485/// primitives, shared cells/maps/mailboxes, and debug state. Builtin tables are
486/// shared through `Arc` until a per-execution rebind needs copy-on-write.
487#[derive(Clone)]
488pub struct VmBaseline {
489    builtins: Arc<BTreeMap<String, VmBuiltinFn>>,
490    async_builtins: Arc<BTreeMap<String, VmAsyncBuiltinFn>>,
491    capability_methods:
492        Arc<BTreeMap<harn_builtin_meta::CapabilityId, BTreeMap<String, VmBuiltinDispatch>>>,
493    builtin_metadata: Arc<BTreeMap<String, VmBuiltinMetadata>>,
494    builtins_by_id: Arc<HashMap<BuiltinId, VmBuiltinEntry>>,
495    builtin_id_collisions: Arc<HashSet<BuiltinId>>,
496    source_dir: Option<std::path::PathBuf>,
497    source_file: Option<String>,
498    source_text: Option<String>,
499    project_root: Option<std::path::PathBuf>,
500    globals: Arc<crate::value::DictMap>,
501    root_harness: Option<VmValue>,
502    denied_builtins: Arc<HashSet<String>>,
503    prepared_module_cache: crate::PreparedModuleCache,
504    module_provenance: crate::module_artifact::ModuleProvenance,
505    /// Carried so a VM rebuilt from this baseline keeps resolving modules
506    /// through the same link table the original spawn was handed.
507    graph_link_table: Option<Arc<crate::context_manifest::GraphLinkTable>>,
508    linked_program_repository: Option<Arc<crate::linked_program::LinkedProgramRepository>>,
509    runtime_limits: RuntimeLimits,
510}
511
512impl VmBaseline {
513    pub fn from_vm(vm: &Vm) -> Self {
514        Self {
515            builtins: Arc::clone(&vm.builtins),
516            async_builtins: Arc::clone(&vm.async_builtins),
517            capability_methods: Arc::clone(&vm.capability_methods),
518            builtin_metadata: Arc::clone(&vm.builtin_metadata),
519            builtins_by_id: Arc::clone(&vm.builtins_by_id),
520            builtin_id_collisions: Arc::clone(&vm.builtin_id_collisions),
521            source_dir: vm.source_dir.clone(),
522            source_file: vm.source_file.clone(),
523            source_text: vm.source_text.clone(),
524            project_root: vm.project_root.clone(),
525            globals: Arc::clone(&vm.globals),
526            root_harness: vm.root_harness.clone(),
527            denied_builtins: Arc::clone(&vm.denied_builtins),
528            prepared_module_cache: vm.prepared_module_cache.clone(),
529            module_provenance: vm.module_provenance,
530            graph_link_table: vm.graph_link_table.clone(),
531            linked_program_repository: vm.linked_program_repository.clone(),
532            runtime_limits: vm.runtime_limits,
533        }
534    }
535
536    pub fn instantiate(&self) -> Vm {
537        crate::initialize_runtime_assets();
538        let mut source_cache = BTreeMap::new();
539        if let (Some(file), Some(text)) = (&self.source_file, &self.source_text) {
540            source_cache.insert(std::path::PathBuf::from(file), Arc::from(text.as_str()));
541        }
542        if let Some(dir) = &self.source_dir {
543            crate::stdlib::set_thread_source_dir(dir);
544        }
545
546        let mut vm = Vm {
547            stack: Vec::with_capacity(256),
548            env: VmEnv::new(),
549            output: String::new(),
550            builtins: Arc::clone(&self.builtins),
551            async_builtins: Arc::clone(&self.async_builtins),
552            capability_methods: Arc::clone(&self.capability_methods),
553            builtin_metadata: Arc::clone(&self.builtin_metadata),
554            builtins_by_id: Arc::clone(&self.builtins_by_id),
555            builtin_id_collisions: Arc::clone(&self.builtin_id_collisions),
556            iterators: Vec::new(),
557            frames: Vec::new(),
558            exception_handlers: Vec::new(),
559            spawned_tasks: BTreeMap::new(),
560            process_exit_request: Arc::new(ProcessExitRequest::new()),
561            sync_runtime: Arc::new(crate::synchronization::VmSyncRuntime::new()),
562            shared_state_runtime: Arc::new(crate::shared_state::VmSharedStateRuntime::new()),
563            worker_registry: crate::stdlib::agents::agents_workers::active_worker_registry(),
564            daemon_registry: crate::stdlib::agents_daemon::active_daemon_registry(),
565            trigger_registry: crate::triggers::registry::active_trigger_registry(),
566            session_runtime: crate::agent_sessions::active_session_runtime(),
567            tracing_runtime: crate::tracing::active_tracing_runtime(),
568            agent_host_session_runtime:
569                crate::llm::agent_session_host::active_agent_host_session_runtime(),
570            connector_clients: Arc::new(crate::connectors::VmConnectorClients::default()),
571            inline_cache_sets: Vec::new(),
572            inline_cache_set_by_chunk: HashMap::new(),
573            pool_registry: crate::stdlib::pool::new_pool_registry(),
574            llm_mock_context: crate::llm::mock::LlmMockContext::for_new_vm(),
575            package_snapshot_registry: Arc::new(Default::default()),
576            wait_for_graph: Arc::new(crate::wait_for_graph::VmWaitForGraph::new()),
577            held_sync_guards: Vec::new(),
578            inherited_held_keys: Arc::new(Vec::new()),
579            task_scopes: Vec::new(),
580            task_counter: 0,
581            runtime_context_counter: 0,
582            runtime_context: crate::runtime_context::RuntimeContext::root(),
583            deadlines: Vec::new(),
584            execution_deadline: super::execution::new_execution_deadline_state(None),
585            breakpoints: BTreeMap::new(),
586            function_breakpoints: std::collections::BTreeSet::new(),
587            pending_function_bp: None,
588            step_mode: false,
589            step_frame_depth: 0,
590            stopped: false,
591            last_line: 0,
592            source_dir: self.source_dir.clone(),
593            package_execution_guard: None,
594            imported_paths: Vec::new(),
595            deferred_cyclic_imports: Vec::new(),
596            module_cache: Arc::new(BTreeMap::new()),
597            prepared_module_cache: self.prepared_module_cache.clone(),
598            prepared_module_validation: crate::prepared_module::PreparedModuleValidation::default(),
599            module_provenance: self.module_provenance,
600            module_phase_recorder: None,
601            lazy_callable_modules: Arc::new(crate::value::VmMutex::new(BTreeMap::new())),
602            source_cache: Arc::new(source_cache),
603            graph_link_table: self.graph_link_table.clone(),
604            linked_program_repository: self.linked_program_repository.clone(),
605            source_file: self.source_file.clone(),
606            source_text: self.source_text.clone(),
607            coverage: crate::coverage::for_primary(self.source_file.as_deref()),
608            bridge: None,
609            denied_builtins: Arc::clone(&self.denied_builtins),
610            cancel_token: None,
611            interrupt_signal_token: None,
612            cancel_grace_instructions_remaining: None,
613            interrupt_handlers: Vec::new(),
614            next_interrupt_handle: 1,
615            pending_interrupt_signal: None,
616            interrupted: false,
617            dispatching_interrupt: false,
618            interrupt_handler_deadline: None,
619            error_stack_trace: Vec::new(),
620            yield_sender: None,
621            project_root: self.project_root.clone(),
622            globals: Arc::clone(&self.globals),
623            root_harness: self.root_harness.clone(),
624            runtime_effects: crate::orchestration::RuntimeEffectState::fresh(),
625            debug_hook: None,
626            runtime_limits: self.runtime_limits,
627        };
628
629        crate::stdlib::rebind_execution_state_builtins(&mut vm);
630        vm
631    }
632}
633
634impl Vm {
635    pub(crate) fn ensure_execution_available(&self) -> Result<(), VmError> {
636        if self.execution_deadline.is_abandoned() {
637            return Err(VmError::AbandonedExecution);
638        }
639        Ok(())
640    }
641
642    pub(crate) fn fresh_local_slots(chunk: &Chunk) -> Vec<LocalSlot> {
643        chunk
644            .local_slots
645            .iter()
646            .map(|_| LocalSlot {
647                value: VmValue::Nil,
648                initialized: false,
649                synced: false,
650            })
651            .collect()
652    }
653
654    pub(crate) fn bind_param_slots(
655        slots: &mut [LocalSlot],
656        func: &crate::chunk::CompiledFunction,
657        args: &[VmValue],
658        synced: bool,
659    ) {
660        Self::bind_param_slots_args(slots, func, &super::CallArgs::Slice(args), synced);
661    }
662
663    pub(crate) fn bind_param_slots_args(
664        slots: &mut [LocalSlot],
665        func: &crate::chunk::CompiledFunction,
666        args: &super::CallArgs<'_>,
667        synced: bool,
668    ) {
669        let param_count = func.params.len();
670        for (i, _param) in func.params.iter().enumerate() {
671            if i >= slots.len() {
672                break;
673            }
674            if func.has_rest_param && i == param_count - 1 {
675                let rest_args = args.to_vec_from(i);
676                slots[i].value = VmValue::List(std::sync::Arc::new(rest_args));
677                slots[i].initialized = true;
678                slots[i].synced = synced;
679            } else if let Some(arg) = args.get(i) {
680                slots[i].value = arg.clone();
681                slots[i].initialized = true;
682                slots[i].synced = synced;
683            }
684        }
685    }
686
687    pub(crate) fn visible_variables(&self) -> crate::value::DictMap {
688        let mut vars = self.env.all_variables();
689        let Some(frame) = self.frames.last() else {
690            return vars;
691        };
692        for (slot, info) in frame.local_slots.iter().zip(frame.chunk.local_slots.iter()) {
693            if slot.initialized && info.scope_depth <= frame.local_scope_depth {
694                vars.insert(crate::value::intern_key(&info.name), slot.value.clone());
695            }
696        }
697        vars
698    }
699
700    pub(crate) fn sync_current_frame_locals_to_env(&mut self) {
701        let frames = &mut self.frames;
702        let env = &mut self.env;
703        let Some(frame) = frames.last_mut() else {
704            return;
705        };
706        let local_scope_base = frame.local_scope_base;
707        let local_scope_depth = frame.local_scope_depth;
708        for (slot, info) in frame
709            .local_slots
710            .iter_mut()
711            .zip(frame.chunk.local_slots.iter())
712        {
713            if slot.initialized && !slot.synced && info.scope_depth <= local_scope_depth {
714                slot.synced = true;
715                let scope_idx = local_scope_base + info.scope_depth;
716                while env.scopes.len() <= scope_idx {
717                    env.push_scope();
718                }
719                // Slot-backed locals are never boxed captures (those resolve
720                // to `None` in the compiler and are defined straight into the
721                // env as `Cell`s), so syncing a slot always yields a `Value`.
722                Arc::make_mut(&mut env.scopes[scope_idx].vars).insert(
723                    info.name.clone(),
724                    crate::value::Binding::Value {
725                        value: slot.value.clone(),
726                        mutable: info.mutable,
727                    },
728                );
729            }
730        }
731    }
732
733    pub(crate) fn closure_call_env_for_current_frame(
734        &self,
735        closure: &crate::value::VmClosure,
736    ) -> VmEnv {
737        if closure.module_state().is_some() {
738            return closure.env.cloned_for_call();
739        }
740        let call_env = Self::closure_call_env(&self.env, closure);
741        // Same compile-time short-circuit as the env walk in
742        // `closure_call_env`: when the callee body never resolves an
743        // outer name through the env, injecting closure-typed *slot*
744        // locals from the caller's frame is wasted work too.
745        if !closure.func.chunk.references_outer_names {
746            return call_env;
747        }
748        let mut call_env = call_env;
749        let Some(frame) = self.frames.last() else {
750            return call_env;
751        };
752        for (slot, info) in frame
753            .local_slots
754            .iter()
755            .zip(frame.chunk.local_slots.iter())
756            .filter(|(slot, info)| slot.initialized && info.scope_depth <= frame.local_scope_depth)
757        {
758            if matches!(slot.value, VmValue::Closure(_)) && !call_env.contains(&info.name) {
759                let _ = call_env.define(&info.name, slot.value.clone(), info.mutable);
760            }
761        }
762        call_env
763    }
764
765    pub(crate) fn active_local_slot_value(&self, name: &str) -> Option<VmValue> {
766        let frame = self.frames.last()?;
767        let idx = self.active_local_slot_index(name)?;
768        frame.local_slots.get(idx).map(|slot| slot.value.clone())
769    }
770
771    /// Returns the slot index of an initialized active local with the given
772    /// name, walking from innermost to outermost scope. Used by legacy by-name
773    /// hot paths that still want to mutate the slot value in place without
774    /// paying a defensive `VmValue::clone` first.
775    pub(crate) fn active_local_slot_index(&self, name: &str) -> Option<usize> {
776        let frame = self.frames.last()?;
777        for (idx, info) in frame.chunk.local_slots.iter().enumerate().rev() {
778            if info.name == name && info.scope_depth <= frame.local_scope_depth {
779                if let Some(slot) = frame.local_slots.get(idx) {
780                    if slot.initialized {
781                        return Some(idx);
782                    }
783                }
784            }
785        }
786        None
787    }
788
789    pub(crate) fn assign_active_local_slot(
790        &mut self,
791        name: &str,
792        value: VmValue,
793        debug: bool,
794    ) -> Result<bool, VmError> {
795        let Some(frame) = self.frames.last_mut() else {
796            return Ok(false);
797        };
798        for (idx, info) in frame.chunk.local_slots.iter().enumerate().rev() {
799            if info.name == name && info.scope_depth <= frame.local_scope_depth {
800                if !debug && !info.mutable {
801                    return Err(VmError::ImmutableAssignment(name.to_string()));
802                }
803                if let Some(slot) = frame.local_slots.get_mut(idx) {
804                    crate::value::recursion::dismantle(std::mem::replace(&mut slot.value, value));
805                    slot.initialized = true;
806                    slot.synced = false;
807                    return Ok(true);
808                }
809            }
810        }
811        Ok(false)
812    }
813
814    pub fn new() -> Self {
815        crate::initialize_runtime_assets();
816        Self {
817            stack: Vec::with_capacity(256),
818            env: VmEnv::new(),
819            output: String::new(),
820            builtins: Arc::new(BTreeMap::new()),
821            async_builtins: Arc::new(BTreeMap::new()),
822            capability_methods: Arc::new(BTreeMap::new()),
823            builtin_metadata: Arc::new(BTreeMap::new()),
824            builtins_by_id: Arc::new(HashMap::new()),
825            builtin_id_collisions: Arc::new(HashSet::new()),
826            iterators: Vec::new(),
827            frames: Vec::new(),
828            exception_handlers: Vec::new(),
829            spawned_tasks: BTreeMap::new(),
830            process_exit_request: Arc::new(ProcessExitRequest::new()),
831            sync_runtime: Arc::new(crate::synchronization::VmSyncRuntime::new()),
832            shared_state_runtime: Arc::new(crate::shared_state::VmSharedStateRuntime::new()),
833            worker_registry: crate::stdlib::agents::agents_workers::active_worker_registry(),
834            daemon_registry: crate::stdlib::agents_daemon::active_daemon_registry(),
835            trigger_registry: crate::triggers::registry::active_trigger_registry(),
836            session_runtime: crate::agent_sessions::active_session_runtime(),
837            tracing_runtime: crate::tracing::active_tracing_runtime(),
838            agent_host_session_runtime:
839                crate::llm::agent_session_host::active_agent_host_session_runtime(),
840            connector_clients: Arc::new(crate::connectors::VmConnectorClients::default()),
841            inline_cache_sets: Vec::new(),
842            inline_cache_set_by_chunk: HashMap::new(),
843            pool_registry: crate::stdlib::pool::new_pool_registry(),
844            llm_mock_context: crate::llm::mock::LlmMockContext::for_new_vm(),
845            package_snapshot_registry: Arc::new(Default::default()),
846            wait_for_graph: Arc::new(crate::wait_for_graph::VmWaitForGraph::new()),
847            held_sync_guards: Vec::new(),
848            inherited_held_keys: Arc::new(Vec::new()),
849            task_scopes: Vec::new(),
850            task_counter: 0,
851            runtime_context_counter: 0,
852            runtime_context: crate::runtime_context::RuntimeContext::root(),
853            deadlines: Vec::new(),
854            execution_deadline: super::execution::new_execution_deadline_state(None),
855            breakpoints: BTreeMap::new(),
856            function_breakpoints: std::collections::BTreeSet::new(),
857            pending_function_bp: None,
858            step_mode: false,
859            step_frame_depth: 0,
860            stopped: false,
861            last_line: 0,
862            source_dir: None,
863            package_execution_guard: None,
864            imported_paths: Vec::new(),
865            deferred_cyclic_imports: Vec::new(),
866            module_cache: Arc::new(BTreeMap::new()),
867            prepared_module_cache: crate::PreparedModuleCache::default(),
868            prepared_module_validation: crate::prepared_module::PreparedModuleValidation::default(),
869            module_provenance: crate::module_artifact::ModuleProvenance::User,
870            module_phase_recorder: None,
871            lazy_callable_modules: Arc::new(crate::value::VmMutex::new(BTreeMap::new())),
872            source_cache: Arc::new(BTreeMap::new()),
873            graph_link_table: None,
874            linked_program_repository: None,
875            source_file: None,
876            source_text: None,
877            coverage: crate::coverage::for_primary(None),
878            bridge: None,
879            denied_builtins: Arc::new(HashSet::new()),
880            cancel_token: None,
881            interrupt_signal_token: None,
882            cancel_grace_instructions_remaining: None,
883            interrupt_handlers: Vec::new(),
884            next_interrupt_handle: 1,
885            pending_interrupt_signal: None,
886            interrupted: false,
887            dispatching_interrupt: false,
888            interrupt_handler_deadline: None,
889            error_stack_trace: Vec::new(),
890            yield_sender: None,
891            project_root: None,
892            globals: Arc::new(crate::value::DictMap::new()),
893            root_harness: None,
894            runtime_effects: crate::orchestration::RuntimeEffectState::fresh(),
895            debug_hook: None,
896            runtime_limits: RuntimeLimits::default(),
897        }
898    }
899
900    pub fn baseline(&self) -> VmBaseline {
901        VmBaseline::from_vm(self)
902    }
903
904    /// Typed, runtime-resolved effects attempted by this VM execution tree.
905    pub fn executed_effects(&self) -> Vec<crate::orchestration::EffectRecord> {
906        self.runtime_effects.snapshot()
907    }
908
909    /// Clear the execution-local effect recorder and per-VM memo.
910    pub fn clear_executed_effects(&mut self) {
911        self.runtime_effects.clear();
912    }
913
914    pub(crate) fn record_capability_effects(
915        &mut self,
916        capability: harn_builtin_meta::CapabilityId,
917        method: &str,
918        args: &[VmValue],
919    ) {
920        self.runtime_effects
921            .record_capability(capability, method, args);
922    }
923
924    pub(crate) fn record_builtin_contract_effects(&mut self, name: &str, args: &[VmValue]) {
925        let Some(entry) = crate::stdlib::recorded_effect_builtin_manifest_entry(name) else {
926            return;
927        };
928        self.record_builtin_effect_specs(entry.contract.effects, args);
929    }
930
931    pub(crate) fn record_builtin_effect_specs(
932        &mut self,
933        specs: &'static [harn_builtin_meta::EffectSpec],
934        args: &[VmValue],
935    ) {
936        self.runtime_effects.record_specs(specs, args);
937    }
938
939    /// Replace the scoped immutable module-template cache used by this VM.
940    /// Fresh runtime module state is still instantiated for every VM.
941    pub fn set_prepared_module_cache(&mut self, cache: crate::PreparedModuleCache) {
942        self.prepared_module_cache = cache;
943        self.prepared_module_validation =
944            crate::prepared_module::PreparedModuleValidation::default();
945    }
946
947    /// Hand this VM the link table for the import graph it is about to run.
948    ///
949    /// Only an entry-chunk cache hit can supply one, because only that path holds
950    /// a manifest proving the recorded digests still describe the files on disk.
951    /// A table never changes what a module resolves to: one the table does not
952    /// name, or whose artifact is no longer on disk, is read and compiled as
953    /// usual.
954    pub fn set_graph_link_table(
955        &mut self,
956        link_table: Option<Arc<crate::context_manifest::GraphLinkTable>>,
957    ) {
958        self.graph_link_table = link_table;
959    }
960
961    /// Install the verified module-template repository carried by one linked
962    /// program. Templates remain scoped to this VM tree and never enter the
963    /// ordinary caller-independent cache.
964    pub fn set_linked_program_runtime(
965        &mut self,
966        runtime: &crate::linked_program::LinkedProgramRuntime,
967    ) {
968        self.linked_program_repository = Some(Arc::clone(&runtime.repository));
969        self.graph_link_table = None;
970    }
971
972    /// Return the effective runtime limit profile for this VM.
973    pub fn runtime_limits(&self) -> RuntimeLimits {
974        self.runtime_limits
975    }
976
977    /// Return a host/debug report describing the VM's effective runtime limits.
978    pub fn runtime_limit_report(&self) -> crate::RuntimeLimitsReport {
979        self.runtime_limits.report()
980    }
981
982    /// Returns true if any debugging affordance is active — DAP hook,
983    /// line breakpoints, or function breakpoints. Call-site code uses
984    /// this to decide whether to capture per-frame restart snapshots
985    /// (`initial_env`, `initial_local_slots`); without a debugger those
986    /// snapshots are dead weight, so skipping them removes two
987    /// allocations from every function call hot path.
988    ///
989    /// All three signals are stable across a function call's lifetime
990    /// (they're set before pipeline execution starts), so the gate is
991    /// consistent between frame creation and any later `restart_frame`
992    /// invocation. The three `is_empty` checks compile to a handful of
993    /// branch-predicted memory probes — cheaper than a single
994    /// `BTreeMap` clone, which is what we're avoiding.
995    #[inline]
996    pub(crate) fn debugger_attached(&self) -> bool {
997        self.debug_hook.is_some()
998            || !self.breakpoints.is_empty()
999            || !self.function_breakpoints.is_empty()
1000    }
1001
1002    /// Set the bridge for delegating unknown builtins in bridge mode.
1003    pub fn set_bridge(&mut self, bridge: Arc<crate::bridge::HostBridge>) {
1004        self.bridge = Some(bridge);
1005    }
1006
1007    /// Set builtins that are denied in sandbox mode.
1008    /// When called, the given builtin names will produce a permission error.
1009    pub fn set_denied_builtins(&mut self, mut denied: HashSet<String>) {
1010        // A builtin may have a private registry name plus one or more public
1011        // aliases. Normalize the whole family once so no dispatch path can
1012        // escape policy merely by using another name for the same handler.
1013        let denied_canonical_names = denied
1014            .iter()
1015            .filter_map(|name| crate::stdlib::builtin_manifest_entry(name))
1016            .map(|entry| entry.canonical_name)
1017            .collect::<HashSet<_>>();
1018        if !denied_canonical_names.is_empty() {
1019            denied.extend(
1020                crate::stdlib::all_builtin_manifest()
1021                    .iter()
1022                    .filter(|entry| denied_canonical_names.contains(entry.canonical_name))
1023                    .map(|entry| entry.name.to_string()),
1024            );
1025        }
1026        self.denied_builtins = Arc::new(denied);
1027    }
1028
1029    /// Set source info for error reporting (file path and source text).
1030    pub fn set_source_info(&mut self, file: &str, text: &str) {
1031        self.source_file = Some(file.to_string());
1032        self.source_text = Some(text.to_string());
1033        if let Some(cov) = self.coverage.as_mut() {
1034            cov.set_primary_file(file);
1035        }
1036        Arc::make_mut(&mut self.source_cache)
1037            .insert(std::path::PathBuf::from(file), Arc::from(text));
1038    }
1039
1040    /// Initialize execution (push the initial frame).
1041    pub fn start(&mut self, chunk: &Chunk) -> Result<(), VmError> {
1042        self.ensure_execution_available()?;
1043        // The top-level pipeline frame captures env at start so
1044        // restartFrame on the outermost frame rewinds to the
1045        // pre-pipeline state — basically "restart session" in
1046        // debugger terms. Skipped when no debugger is attached:
1047        // the snapshot is dead weight in that case and dominates
1048        // call-overhead bench numbers (~5-10%).
1049        let debugger = self.debugger_attached();
1050        let initial_env = if debugger {
1051            Some(self.env.clone())
1052        } else {
1053            None
1054        };
1055        let initial_local_slots = if debugger {
1056            Some(Self::fresh_local_slots(chunk))
1057        } else {
1058            None
1059        };
1060        let chunk = Arc::new(chunk.clone());
1061        let local_slots = Self::fresh_local_slots(&chunk);
1062        let inline_cache_set = self.inline_cache_set_index_for_chunk(&chunk);
1063        self.frames.push(CallFrame {
1064            chunk,
1065            inline_cache_set,
1066            ip: 0,
1067            stack_base: self.stack.len(),
1068            saved_env: self.env.clone(),
1069            initial_env,
1070            initial_local_slots,
1071            saved_iterator_depth: self.iterators.len(),
1072            fn_name: crate::value::HarnStr::new(),
1073            argc: 0,
1074            saved_source_dir: None,
1075            module_functions: None,
1076            module_state: None,
1077            local_slots,
1078            local_scope_base: self.env.scope_depth().saturating_sub(1),
1079            local_scope_depth: 0,
1080        });
1081        Ok(())
1082    }
1083
1084    /// Create a child VM that shares builtins and env but has fresh execution state.
1085    /// Used for parallel/spawn to fork the VM for concurrent tasks.
1086    pub(crate) fn child_vm(&self) -> Vm {
1087        Vm {
1088            stack: Vec::with_capacity(64),
1089            env: self.env.clone(),
1090            output: String::new(),
1091            builtins: Arc::clone(&self.builtins),
1092            async_builtins: Arc::clone(&self.async_builtins),
1093            capability_methods: Arc::clone(&self.capability_methods),
1094            builtin_metadata: Arc::clone(&self.builtin_metadata),
1095            builtins_by_id: Arc::clone(&self.builtins_by_id),
1096            builtin_id_collisions: Arc::clone(&self.builtin_id_collisions),
1097            iterators: Vec::new(),
1098            frames: Vec::new(),
1099            exception_handlers: Vec::new(),
1100            spawned_tasks: BTreeMap::new(),
1101            process_exit_request: Arc::clone(&self.process_exit_request),
1102            sync_runtime: self.sync_runtime.clone(),
1103            shared_state_runtime: self.shared_state_runtime.clone(),
1104            worker_registry: self.worker_registry.clone(),
1105            daemon_registry: self.daemon_registry.clone(),
1106            trigger_registry: self.trigger_registry.clone(),
1107            session_runtime: self.session_runtime.clone(),
1108            tracing_runtime: self.tracing_runtime.clone(),
1109            agent_host_session_runtime: self.agent_host_session_runtime.clone(),
1110            connector_clients: self.connector_clients.clone(),
1111            inline_cache_sets: Vec::new(),
1112            inline_cache_set_by_chunk: HashMap::new(),
1113            pool_registry: self.pool_registry.clone(),
1114            llm_mock_context: self.llm_mock_context.clone(),
1115            package_snapshot_registry: self.package_snapshot_registry.clone(),
1116            wait_for_graph: self.wait_for_graph.clone(),
1117            held_sync_guards: Vec::new(),
1118            inherited_held_keys: Arc::new(Vec::new()),
1119            task_scopes: Vec::new(),
1120            task_counter: 0,
1121            runtime_context_counter: self.runtime_context_counter,
1122            runtime_context: self.runtime_context.clone(),
1123            deadlines: self.deadlines.clone(),
1124            execution_deadline: self.execution_deadline.fork(),
1125            breakpoints: BTreeMap::new(),
1126            function_breakpoints: std::collections::BTreeSet::new(),
1127            pending_function_bp: None,
1128            step_mode: false,
1129            step_frame_depth: 0,
1130            stopped: false,
1131            last_line: 0,
1132            source_dir: self.source_dir.clone(),
1133            package_execution_guard: self.package_execution_guard.clone(),
1134            imported_paths: Vec::new(),
1135            deferred_cyclic_imports: Vec::new(),
1136            module_cache: Arc::clone(&self.module_cache),
1137            prepared_module_cache: self.prepared_module_cache.clone(),
1138            prepared_module_validation: self.prepared_module_validation.clone(),
1139            module_provenance: self.module_provenance,
1140            module_phase_recorder: self.module_phase_recorder.clone(),
1141            lazy_callable_modules: Arc::clone(&self.lazy_callable_modules),
1142            source_cache: Arc::clone(&self.source_cache),
1143            graph_link_table: self.graph_link_table.clone(),
1144            linked_program_repository: self.linked_program_repository.clone(),
1145            source_file: self.source_file.clone(),
1146            source_text: self.source_text.clone(),
1147            coverage: crate::coverage::for_primary(self.source_file.as_deref()),
1148            bridge: self.bridge.clone(),
1149            denied_builtins: Arc::clone(&self.denied_builtins),
1150            cancel_token: self.cancel_token.clone(),
1151            interrupt_signal_token: self.interrupt_signal_token.clone(),
1152            cancel_grace_instructions_remaining: None,
1153            interrupt_handlers: Vec::new(),
1154            next_interrupt_handle: 1,
1155            pending_interrupt_signal: None,
1156            interrupted: self.interrupted,
1157            dispatching_interrupt: false,
1158            interrupt_handler_deadline: None,
1159            error_stack_trace: Vec::new(),
1160            yield_sender: None,
1161            project_root: self.project_root.clone(),
1162            globals: Arc::clone(&self.globals),
1163            root_harness: self.root_harness.clone(),
1164            runtime_effects: crate::orchestration::RuntimeEffectState::with_shared_recorder(
1165                Arc::clone(&self.runtime_effects.recorder),
1166            ),
1167            debug_hook: None,
1168            runtime_limits: self.runtime_limits,
1169        }
1170    }
1171
1172    /// Create a child VM for external adapters that need to invoke Harn
1173    /// closures while sharing the parent's builtins, globals, and module state.
1174    pub(crate) fn child_vm_for_host(&self) -> Vm {
1175        self.child_vm()
1176    }
1177
1178    /// Return the cooperative cancellation/deadline sources that a blocking
1179    /// host adapter must install on its worker thread. The values are cloned
1180    /// once at the async-boundary; the token remains shared with the owning VM
1181    /// so `parallel` aborts and host cancellation reach an in-flight wait.
1182    pub(crate) fn interrupt_sources(
1183        &self,
1184    ) -> (
1185        Option<std::sync::Arc<std::sync::atomic::AtomicBool>>,
1186        Option<std::time::Instant>,
1187    ) {
1188        let scope_deadline = self.deadlines.last().map(|(deadline, _)| *deadline);
1189        let deadline = match (scope_deadline, self.interrupt_handler_deadline) {
1190            (Some(scope), Some(interrupt)) => Some(scope.min(interrupt)),
1191            (scope, interrupt) => scope.or(interrupt),
1192        };
1193        (self.cancel_token.clone(), deadline)
1194    }
1195
1196    pub(crate) fn request_process_exit(&self, code: i32) {
1197        self.process_exit_request.request(code);
1198    }
1199
1200    pub(crate) fn requested_process_exit(&self) -> Option<i32> {
1201        self.process_exit_request.code()
1202    }
1203
1204    /// Request cancellation for every outstanding child task owned by this VM
1205    /// and then abort the join handles. This prevents un-awaited spawned tasks
1206    /// from outliving their parent execution scope.
1207    pub(crate) fn cancel_spawned_tasks(&mut self) {
1208        for (_, task) in std::mem::take(&mut self.spawned_tasks) {
1209            task.cancel_token
1210                .store(true, std::sync::atomic::Ordering::SeqCst);
1211            task.handle.abort();
1212        }
1213    }
1214
1215    /// Set the source directory for import resolution and introspection.
1216    /// Also auto-detects the project root if not already set.
1217    pub fn set_source_dir(&mut self, dir: &std::path::Path) {
1218        let dir = crate::stdlib::process::normalize_context_path(dir);
1219        self.source_dir = Some(dir.clone());
1220        crate::stdlib::set_thread_source_dir(&dir);
1221        // Auto-detect project root if not explicitly set.
1222        if self.project_root.is_none() {
1223            self.project_root = crate::stdlib::process::find_project_root(&dir);
1224        }
1225    }
1226
1227    /// Explicitly set the project root directory.
1228    /// Used by ACP/CLI to override auto-detection.
1229    pub fn set_project_root(&mut self, root: &std::path::Path) {
1230        self.project_root = Some(root.to_path_buf());
1231    }
1232
1233    /// Get only the explicit or auto-detected project root, without falling
1234    /// back to `source_dir`.
1235    pub(crate) fn explicit_project_root(&self) -> Option<&std::path::Path> {
1236        self.project_root.as_deref()
1237    }
1238
1239    /// Get the project root directory, falling back to source_dir.
1240    pub fn project_root(&self) -> Option<&std::path::Path> {
1241        self.project_root.as_deref().or(self.source_dir.as_deref())
1242    }
1243
1244    /// Set a global constant (e.g. `pi`, `e`).
1245    /// Stored separately from the environment so user-defined variables can shadow them.
1246    pub fn set_global(&mut self, name: &str, value: VmValue) {
1247        Arc::make_mut(&mut self.globals).insert(crate::value::intern_key(name), value);
1248    }
1249
1250    /// Read a previously-installed ordinary global constant.
1251    pub fn global(&self, name: &str) -> Option<&VmValue> {
1252        self.globals.get(name)
1253    }
1254
1255    /// Install the script's root authority for the compiler-generated
1256    /// `main(harness: Harness)` call. It is kept out of all source-visible
1257    /// namespaces.
1258    pub fn set_harness(&mut self, harness: crate::harness::Harness) {
1259        self.root_harness = Some(harness.into_vm_value());
1260    }
1261
1262    /// Install the connector projection owned by this VM execution tree.
1263    pub fn set_connector_clients(&mut self, clients: crate::connectors::VmConnectorClients) {
1264        self.connector_clients = Arc::new(clients);
1265    }
1266
1267    pub(crate) fn harness(&self) -> Option<&crate::harness::VmHarness> {
1268        match self.root_harness.as_ref() {
1269            Some(VmValue::Harness(handle)) => Some(handle),
1270            _ => None,
1271        }
1272    }
1273
1274    /// Clone the embedder-owned root authority for an explicit exported
1275    /// function parameter. This does not install a source-visible binding.
1276    pub fn root_harness_value(&self) -> Option<VmValue> {
1277        self.root_harness.clone()
1278    }
1279
1280    /// Get the captured output.
1281    pub fn output(&self) -> &str {
1282        &self.output
1283    }
1284
1285    /// Drain and return the captured output, leaving the buffer empty.
1286    /// Used by the async-builtin dispatch path to forward closure output
1287    /// from a child VM back to its parent.
1288    pub fn take_output(&mut self) -> String {
1289        std::mem::take(&mut self.output)
1290    }
1291
1292    /// Append text to this VM's captured output. Used to forward output
1293    /// from child VMs (e.g. closures invoked via `call_closure_pub`)
1294    /// back into the parent stream.
1295    pub fn append_output(&mut self, text: &str) {
1296        self.output.push_str(text);
1297    }
1298
1299    pub(crate) fn pop(&mut self) -> Result<VmValue, VmError> {
1300        self.stack.pop().ok_or(VmError::StackUnderflow)
1301    }
1302
1303    pub(crate) fn peek(&self) -> Result<&VmValue, VmError> {
1304        self.stack.last().ok_or(VmError::StackUnderflow)
1305    }
1306
1307    pub(crate) fn const_str(c: &Constant) -> Result<&str, VmError> {
1308        match c {
1309            Constant::String(s) => Ok(s.as_str()),
1310            _ => Err(VmError::TypeError("expected string constant".into())),
1311        }
1312    }
1313
1314    pub(crate) fn release_sync_guards_for_current_scope(&mut self) {
1315        let depth = self.env.scope_depth();
1316        self.held_sync_guards
1317            .retain(|guard| guard.env_scope_depth < depth);
1318        // A `scope { }` torn down without a normal `TaskScopeExit` (break /
1319        // continue out of it) leaves a dangling nursery — cancel its tasks.
1320        self.cancel_task_scopes_where(|s| s.env_scope_depth >= depth);
1321    }
1322
1323    pub(crate) fn release_sync_guards_after_unwind(
1324        &mut self,
1325        frame_depth: usize,
1326        env_scope_depth: usize,
1327    ) {
1328        self.held_sync_guards.retain(|guard| {
1329            guard.frame_depth <= frame_depth && guard.env_scope_depth <= env_scope_depth
1330        });
1331        // Cancel nurseries opened above the catch handler (a `throw` unwound
1332        // past their `TaskScopeExit`).
1333        self.cancel_task_scopes_where(|s| {
1334            !(s.frame_depth <= frame_depth && s.env_scope_depth <= env_scope_depth)
1335        });
1336    }
1337
1338    pub(crate) fn release_sync_guards_for_frame(&mut self, frame_depth: usize) {
1339        self.held_sync_guards
1340            .retain(|guard| guard.frame_depth != frame_depth);
1341        // Cancel any nursery whose `scope {}` block belonged to the frame being
1342        // torn down (e.g. a `return` jumped past its `TaskScopeExit`).
1343        self.cancel_task_scopes_where(|s| s.frame_depth == frame_depth);
1344    }
1345
1346    pub(crate) fn adopt_sync_permit_for_current_scope(
1347        &mut self,
1348        permit: crate::value::VmSyncPermitHandle,
1349    ) {
1350        if permit.is_released()
1351            || self
1352                .held_sync_guards
1353                .iter()
1354                .any(|guard| guard._permit.same_lease(&permit))
1355        {
1356            return;
1357        }
1358        self.held_sync_guards
1359            .push(crate::synchronization::VmSyncHeldGuard {
1360                _permit: permit,
1361                frame_depth: self.frames.len(),
1362                env_scope_depth: self.env.scope_depth(),
1363            });
1364    }
1365
1366    /// Deregister a task id from every open nursery (it was explicitly
1367    /// `await`ed, so it must not be double-joined or cancelled at scope exit).
1368    pub(crate) fn deregister_task_from_scopes(&mut self, id: &str) {
1369        for scope in &mut self.task_scopes {
1370            scope.task_ids.retain(|t| t != id);
1371        }
1372    }
1373
1374    /// Cancel and remove every task scope matching `doomed`, aborting its bound
1375    /// tasks (used when a `scope {}` is torn down without a normal join).
1376    fn cancel_task_scopes_where<F: Fn(&TaskScope) -> bool>(&mut self, doomed: F) {
1377        let mut i = 0;
1378        while i < self.task_scopes.len() {
1379            if doomed(&self.task_scopes[i]) {
1380                let scope = self.task_scopes.remove(i);
1381                for id in &scope.task_ids {
1382                    if let Some(task) = self.spawned_tasks.remove(id) {
1383                        task.cancel_token
1384                            .store(true, std::sync::atomic::Ordering::SeqCst);
1385                        task.handle.abort();
1386                    }
1387                }
1388            } else {
1389                i += 1;
1390            }
1391        }
1392    }
1393
1394    /// Total live permits this VM already holds for `kind:key`. The held-set is
1395    /// tiny (bounded by lexical nesting and explicit sync acquisitions), so this
1396    /// scan is cheap and only runs on the rare blocking-acquire path.
1397    pub(crate) fn held_permits_for(&self, kind: &str, key: &str) -> u32 {
1398        let own: u32 = self
1399            .held_sync_guards
1400            .iter()
1401            .filter(|guard| {
1402                !guard._permit.is_released()
1403                    && guard._permit.kind() == kind
1404                    && guard._permit.key() == key
1405            })
1406            .map(|guard| guard._permit.permits())
1407            .sum();
1408        let inherited: u32 = self
1409            .inherited_held_keys
1410            .iter()
1411            .filter(|held| held.kind == kind && held.key == key)
1412            .map(|held| held.permits)
1413            .sum();
1414        own + inherited
1415    }
1416
1417    /// Every live sync permit held by this VM *and* its suspended ancestors: the
1418    /// transitive held-set seen by an inline child.
1419    pub(crate) fn combined_held_keys(&self) -> Vec<crate::synchronization::VmSyncHeldKey> {
1420        let mut keys: Vec<crate::synchronization::VmSyncHeldKey> = self
1421            .held_sync_guards
1422            .iter()
1423            .filter_map(|guard| crate::synchronization::VmSyncHeldKey::from_permit(&guard._permit))
1424            .collect();
1425        keys.extend(self.inherited_held_keys.iter().cloned());
1426        keys
1427    }
1428
1429    /// Clone a child VM for an **inline, same-task** execution (an async builtin
1430    /// awaited while this VM is parked, or a user closure that builtin runs and
1431    /// awaits). The child inherits this VM's transitive held-lock keys so a
1432    /// re-acquire of a parent-held lock is caught as a self-deadlock
1433    /// (HARN-ORC-011). Use plain `child_vm()` for new concurrent tasks.
1434    pub(crate) fn child_vm_inline(&self) -> Vm {
1435        let mut child = self.child_vm();
1436        child.inherited_held_keys = Arc::new(self.combined_held_keys());
1437        child.execution_deadline = Arc::clone(&self.execution_deadline);
1438        child
1439    }
1440}
1441
1442impl Drop for Vm {
1443    fn drop(&mut self) {
1444        if let Some(coverage) = self.coverage.take() {
1445            crate::coverage::merge_into_global(coverage);
1446        }
1447        self.cancel_spawned_tasks();
1448    }
1449}
1450
1451impl Default for Vm {
1452    fn default() -> Self {
1453        Self::new()
1454    }
1455}
1456
1457#[cfg(test)]
1458#[path = "state_tests.rs"]
1459mod tests;