Skip to main content

harn_vm/vm/
state.rs

1use std::collections::{BTreeMap, HashMap, HashSet};
2use std::path::PathBuf;
3use std::sync::atomic::{AtomicBool, AtomicU64, 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;
16use super::modules::ModuleCache;
17use super::VmBuiltinMetadata;
18
19/// A lazy callable's resolved export set together with the module graph that
20/// was loaded to produce it.
21///
22/// The exported closures — and every function they can transitively reach
23/// through imports — hold only `Weak`s into their home module's function
24/// registry and module state. The child VM that first loaded the graph is
25/// normally their sole strong owner (via its `module_cache`), and it dies once
26/// the hook fire completes. A later child VM that hits this cache never
27/// re-imports the graph, so a transitively imported callee's sibling `pub fn`
28/// would fall through name resolution to host-bridge dispatch (`Undefined
29/// builtin`). Retaining the complete loaded graph here for the cache entry's
30/// lifetime keeps every transitively reachable registry and module state
31/// upgradeable at call time, for every fire — not just the first.
32pub(crate) struct ResolvedLazyCallable {
33    pub(crate) exports: BTreeMap<String, Arc<VmClosure>>,
34    /// Intentionally unread: retained solely to keep the loaded module graph's
35    /// function registries and module states alive for this cache entry (the
36    /// same liveness role [`crate::value::RetainedModuleScope`] plays for a
37    /// single retained closure, generalized across the whole import graph).
38    #[allow(dead_code)]
39    pub(crate) retained_module_graph: ModuleCache,
40}
41
42pub(crate) type LazyCallableResolution = Arc<ResolvedLazyCallable>;
43pub(crate) struct LazyCallableCacheSlot {
44    pub(crate) execution_guard: Option<Arc<harn_modules::package_execution::PackageExecutionGuard>>,
45    pub(crate) resolution: Arc<tokio::sync::OnceCell<LazyCallableResolution>>,
46}
47pub(crate) type LazyCallableModuleCache =
48    Arc<VmMutex<BTreeMap<PathBuf, Vec<LazyCallableCacheSlot>>>>;
49
50/// RAII guard that starts a tracing span on creation and ends it on drop.
51pub(crate) struct ScopeSpan(u64);
52
53impl ScopeSpan {
54    pub(crate) fn new(kind: crate::tracing::SpanKind, name: String) -> Self {
55        Self(crate::tracing::span_start(kind, name))
56    }
57}
58
59/// Cancellation-safe host deadline state. The guard owns the shared state so
60/// dropping an in-flight `execute_with_timeout` future restores the previous
61/// deadline and poisons the abandoned VM even though the future still holds
62/// `&mut Vm`.
63pub(crate) struct ExecutionDeadlineState {
64    origin: Instant,
65    /// Nanoseconds from `origin`, plus one so zero remains the inactive value.
66    deadline_offset: AtomicU64,
67    /// Set when the host drops a polled execution future. Arbitrary async
68    /// cancellation cannot unwind interpreter state, so subsequent execution
69    /// entries fail loudly instead of resuming a partial frame.
70    abandoned: AtomicBool,
71}
72
73impl ExecutionDeadlineState {
74    pub(crate) fn new(origin: Instant, deadline: Option<Instant>) -> Arc<Self> {
75        Arc::new(Self {
76            origin,
77            deadline_offset: AtomicU64::new(Self::encode(origin, deadline)),
78            abandoned: AtomicBool::new(false),
79        })
80    }
81
82    #[inline]
83    pub(crate) fn is_active(&self) -> bool {
84        self.deadline_offset.load(Ordering::Acquire) != 0
85    }
86
87    #[inline]
88    pub(crate) fn is_abandoned(&self) -> bool {
89        self.abandoned.load(Ordering::Acquire)
90    }
91
92    pub(crate) fn fork(&self) -> Arc<Self> {
93        let state = Self::new(self.origin, self.current());
94        state
95            .abandoned
96            .store(self.is_abandoned(), Ordering::Release);
97        state
98    }
99
100    pub(crate) fn current(&self) -> Option<Instant> {
101        let encoded = self.deadline_offset.load(Ordering::Acquire);
102        (encoded != 0)
103            .then(|| self.origin + std::time::Duration::from_nanos(encoded.saturating_sub(1)))
104    }
105
106    pub(crate) fn install(self: &Arc<Self>, deadline: Instant) -> ExecutionDeadlineGuard {
107        let previous = self.deadline_offset.load(Ordering::Acquire);
108        let requested = Self::encode(self.origin, Some(deadline));
109        let active = if previous == 0 {
110            requested
111        } else {
112            previous.min(requested)
113        };
114        self.deadline_offset.store(active, Ordering::Release);
115        ExecutionDeadlineGuard {
116            state: Arc::clone(self),
117            previous,
118            completed: false,
119        }
120    }
121
122    fn encode(origin: Instant, deadline: Option<Instant>) -> u64 {
123        deadline.map_or(0, |deadline| {
124            let nanos = deadline.saturating_duration_since(origin).as_nanos();
125            u64::try_from(nanos)
126                .unwrap_or(u64::MAX - 1)
127                .saturating_add(1)
128        })
129    }
130}
131
132pub(crate) struct ExecutionDeadlineGuard {
133    state: Arc<ExecutionDeadlineState>,
134    previous: u64,
135    completed: bool,
136}
137
138impl ExecutionDeadlineGuard {
139    /// Mark an awaited execution as terminal before restoring its prior host
140    /// deadline. Dropping without this acknowledgement poisons the VM.
141    pub(crate) fn complete(mut self) {
142        self.completed = true;
143    }
144}
145
146impl Drop for ExecutionDeadlineGuard {
147    fn drop(&mut self) {
148        self.state
149            .deadline_offset
150            .store(self.previous, Ordering::Release);
151        if !self.completed {
152            self.state.abandoned.store(true, Ordering::Release);
153        }
154    }
155}
156
157impl Drop for ScopeSpan {
158    fn drop(&mut self) {
159        crate::tracing::span_end(self.0);
160    }
161}
162
163#[derive(Clone)]
164pub(crate) struct LocalSlot {
165    pub(crate) value: VmValue,
166    pub(crate) initialized: bool,
167    pub(crate) synced: bool,
168}
169
170impl Drop for LocalSlot {
171    fn drop(&mut self) {
172        // Slot locals hold script values directly (e.g. a `let` bound to a
173        // deeply nested list). When a frame is torn down, the default
174        // recursive drop of such a value would overflow the native stack and
175        // abort the process. For the overwhelmingly common scalar slot this is
176        // a single `matches!` check and then the normal trivial drop; only a
177        // nested container is moved out and torn down iteratively, so hot
178        // frame teardown is unaffected.
179        if crate::value::recursion::is_recursive_container(&self.value) {
180            crate::value::recursion::dismantle(std::mem::replace(&mut self.value, VmValue::Nil));
181        }
182    }
183}
184
185#[derive(Clone)]
186pub(crate) struct InterruptHandler {
187    pub(crate) handle: i64,
188    pub(crate) signals: Vec<String>,
189    pub(crate) once: bool,
190    pub(crate) graceful_timeout_ms: Option<u64>,
191    pub(crate) handler: VmValue,
192}
193
194/// Call frame for function execution.
195pub(crate) struct CallFrame {
196    pub(crate) chunk: ChunkRef,
197    /// VM-local inline-cache set for this frame's chunk. Computed once at
198    /// frame entry so hot opcode dispatch can index cache feedback directly
199    /// instead of hashing the chunk id on every cached opcode.
200    pub(crate) inline_cache_set: usize,
201    pub(crate) ip: usize,
202    pub(crate) stack_base: usize,
203    pub(crate) saved_env: VmEnv,
204    /// Env snapshot captured at call-time, *after* argument binding. Used
205    /// by the debugger's `restartFrame` to rewind this frame to its
206    /// entry state (re-binding args from the original values) without
207    /// re-entering the call site. Cheap to clone because `VmEnv` is
208    /// already cloned into `saved_env` on every call. `None` for
209    /// scratch frames (evaluate, import init) where restart isn't
210    /// meaningful.
211    pub(crate) initial_env: Option<VmEnv>,
212    pub(crate) initial_local_slots: Option<Vec<LocalSlot>>,
213    /// Iterator stack depth to restore when this frame unwinds.
214    pub(crate) saved_iterator_depth: usize,
215    /// Function name for stack traces (empty for top-level pipeline).
216    pub(crate) fn_name: String,
217    /// Number of arguments actually passed by the caller (for default arg support).
218    pub(crate) argc: usize,
219    /// Saved VM_SOURCE_DIR to restore when this frame is popped.
220    /// Set when entering a closure that originated from an imported module.
221    pub(crate) saved_source_dir: Option<std::path::PathBuf>,
222    /// Module-local named functions available to symbolic calls within this frame.
223    pub(crate) module_functions: Option<ModuleFunctionRegistry>,
224    /// Shared module-level env for top-level `let` / `const` bindings of
225    /// this frame's originating module. Looked up after `self.env` and
226    /// before `self.globals` by `GetVar` / `SetVar`, giving each module
227    /// its own live static state that persists across calls. See the
228    /// `module_state` field on `VmClosure` for the full rationale.
229    pub(crate) module_state: Option<crate::value::ModuleState>,
230    /// Slot-indexed locals for compiler-resolved names in this frame.
231    pub(crate) local_slots: Vec<LocalSlot>,
232    /// Env scope index that corresponds to compiler local scope depth 0.
233    pub(crate) local_scope_base: usize,
234    /// Current compiler local scope depth, updated by PushScope/PopScope.
235    pub(crate) local_scope_depth: usize,
236}
237
238pub(crate) struct InlineCacheSite {
239    pub(crate) cache_set: usize,
240    pub(crate) slot_count: usize,
241    pub(crate) slot: Option<usize>,
242}
243
244impl CallFrame {
245    #[inline]
246    pub(crate) fn inline_cache_site_for_previous_op(&self) -> InlineCacheSite {
247        let op_offset = self.ip.saturating_sub(1);
248        InlineCacheSite {
249            cache_set: self.inline_cache_set,
250            slot_count: self.chunk.inline_cache_slot_count(),
251            slot: self.chunk.inline_cache_slot(op_offset),
252        }
253    }
254}
255
256/// Exception handler for try/catch.
257pub(crate) struct ExceptionHandler {
258    pub(crate) catch_ip: usize,
259    pub(crate) stack_depth: usize,
260    pub(crate) frame_depth: usize,
261    pub(crate) env_scope_depth: usize,
262    /// When present, this catch only handles errors whose enum_name matches.
263    pub(crate) error_type: Option<crate::value::HarnStr>,
264}
265
266/// A structured-concurrency nursery (`scope { }`). Tasks spawned while this
267/// scope is innermost record their id here; `TaskScopeExit` joins them.
268pub(crate) struct TaskScope {
269    /// Ids of tasks spawned in this scope that have not been explicitly
270    /// `await`ed away. Joined (normal exit) or cancelled (unwind) on close.
271    pub(crate) task_ids: Vec<String>,
272    /// Frame depth at which the scope was opened, for unwind pruning.
273    pub(crate) frame_depth: usize,
274    /// Env scope depth at open, for unwind pruning.
275    pub(crate) env_scope_depth: usize,
276}
277
278/// Terminal exit requested by any VM in one execution tree. A process exit is
279/// global control flow, so child VMs share this latch with their parent rather
280/// than relying on a task being explicitly awaited.
281pub(crate) struct ProcessExitRequest {
282    code: Mutex<Option<i32>>,
283    requested: AtomicBool,
284}
285
286impl ProcessExitRequest {
287    fn new() -> Self {
288        Self {
289            code: Mutex::new(None),
290            requested: AtomicBool::new(false),
291        }
292    }
293
294    fn request(&self, code: i32) {
295        let mut recorded = self
296            .code
297            .lock()
298            .expect("process exit request lock poisoned");
299        if recorded.is_none() {
300            *recorded = Some(code);
301            self.requested.store(true, Ordering::Release);
302        }
303    }
304
305    fn code(&self) -> Option<i32> {
306        if !self.requested.load(Ordering::Acquire) {
307            return None;
308        }
309        *self
310            .code
311            .lock()
312            .expect("process exit request lock poisoned")
313    }
314}
315
316/// Iterator state for for-in loops.
317pub(crate) enum IterState {
318    Vec {
319        items: Arc<Vec<VmValue>>,
320        idx: usize,
321    },
322    Dict {
323        entries: Arc<crate::value::DictMap>,
324        keys: Vec<String>,
325        idx: usize,
326    },
327    Channel {
328        receiver: std::sync::Arc<tokio::sync::Mutex<tokio::sync::mpsc::Receiver<VmValue>>>,
329        close: std::sync::Arc<crate::value::VmChannelCloseState>,
330    },
331    Generator {
332        gen: Arc<crate::value::VmGenerator>,
333    },
334    Stream {
335        stream: Arc<crate::value::VmStream>,
336    },
337    /// Step through a lazy range without materializing a Vec.
338    /// Inclusive ranges keep `end` as an actual value so `i64::MAX to i64::MAX`
339    /// still yields one item instead of overflowing a one-past-end sentinel.
340    Range {
341        next: i64,
342        end: i64,
343        inclusive: bool,
344        done: bool,
345    },
346    VmIter {
347        handle: crate::vm::iter::VmIterHandle,
348    },
349}
350
351#[derive(Clone)]
352pub(crate) enum VmBuiltinDispatch {
353    Sync(VmBuiltinFn),
354    Async(VmAsyncBuiltinFn),
355}
356
357#[derive(Clone)]
358pub(crate) struct VmBuiltinEntry {
359    pub(crate) name: Arc<str>,
360    pub(crate) dispatch: VmBuiltinDispatch,
361    /// Static effect specs for source-visible capability/privileged builtins.
362    /// Keeping this beside the numeric dispatch entry makes pure direct calls
363    /// pay one `Option` branch instead of consulting the manifest by name.
364    pub(crate) recorded_effects: Option<&'static [harn_builtin_meta::EffectSpec]>,
365}
366
367/// The Harn bytecode virtual machine.
368pub struct Vm {
369    pub(crate) stack: Vec<VmValue>,
370    pub(crate) env: VmEnv,
371    pub(crate) output: String,
372    pub(crate) builtins: Arc<BTreeMap<String, VmBuiltinFn>>,
373    pub(crate) async_builtins: Arc<BTreeMap<String, VmAsyncBuiltinFn>>,
374    /// Host-supplied implementations keyed by the typed capability surface
375    /// they implement. Unlike ordinary builtins these names never enter the
376    /// source namespace: they are reachable only through a `Harness*` value.
377    pub(crate) capability_methods:
378        Arc<BTreeMap<harn_builtin_meta::CapabilityId, BTreeMap<String, VmBuiltinDispatch>>>,
379    pub(crate) builtin_metadata: Arc<BTreeMap<String, VmBuiltinMetadata>>,
380    /// Numeric side index for builtins. Name-keyed maps remain authoritative;
381    /// this index is the hot path for direct builtin bytecode and callback refs.
382    pub(crate) builtins_by_id: Arc<HashMap<BuiltinId, VmBuiltinEntry>>,
383    /// IDs with detected name collisions. Collided names safely fall back to
384    /// the authoritative name-keyed lookup path.
385    pub(crate) builtin_id_collisions: Arc<HashSet<BuiltinId>>,
386    /// Iterator state for for-in loops.
387    pub(crate) iterators: Vec<IterState>,
388    /// Call frame stack.
389    pub(crate) frames: Vec<CallFrame>,
390    /// Exception handler stack.
391    pub(crate) exception_handlers: Vec<ExceptionHandler>,
392    /// Spawned async task handles.
393    pub(crate) spawned_tasks: BTreeMap<String, VmTaskHandle>,
394    /// Shared terminal process-exit latch for this execution tree.
395    pub(crate) process_exit_request: Arc<ProcessExitRequest>,
396    /// Shared process-local synchronization primitives inherited by child VMs.
397    pub(crate) sync_runtime: Arc<crate::synchronization::VmSyncRuntime>,
398    /// Shared process-local cells, maps, and mailboxes inherited by child VMs.
399    pub(crate) shared_state_runtime: Arc<crate::shared_state::VmSharedStateRuntime>,
400    /// Per-isolate inline cache entries. `inline_cache_set_by_chunk` maps a
401    /// compiled chunk identity to an index in this vector at frame entry; the
402    /// dispatch loop uses the frame-local index for per-op reads/writes.
403    pub(crate) inline_cache_sets: Vec<Vec<crate::chunk::InlineCacheEntry>>,
404    pub(crate) inline_cache_set_by_chunk: HashMap<u64, usize>,
405    /// VM-scoped pool registry inherited by child VMs and scoped into Tokio tasks.
406    pub(crate) pool_registry: Arc<crate::stdlib::pool::PoolRegistry>,
407    /// Inline LLM fixtures and observations shared by this VM execution tree.
408    pub(crate) llm_mock_context: crate::llm::mock::LlmMockContext,
409    /// Reader leases opened by package_snapshot_open in this execution tree.
410    /// Child VMs share the registry; the final VM drop releases abandoned
411    /// leases without touching concurrent executions.
412    pub(crate) package_snapshot_registry: Arc<crate::stdlib::PackageSnapshotRegistry>,
413    /// Shared task/channel wait graph for this VM execution tree.
414    pub(crate) wait_for_graph: Arc<crate::wait_for_graph::VmWaitForGraph>,
415    /// Permits acquired by lexical synchronization blocks in this VM.
416    pub(crate) held_sync_guards: Vec<crate::synchronization::VmSyncHeldGuard>,
417    /// Locks held by an ancestor VM that is *suspended on this VM's execution*:
418    /// an inline async-builtin child runs while its parent is parked
419    /// mid-instruction still holding these permits. Re-acquiring more permits
420    /// than the primitive can grant is a provably-unresolvable self-deadlock, so
421    /// HARN-ORC-011 fires across the child boundary. Empty for new concurrent
422    /// tasks (`spawn`/`parallel`/triggers), where the parent keeps running and
423    /// blocking can be legitimately resolvable.
424    pub(crate) inherited_held_keys: Arc<Vec<crate::synchronization::VmSyncHeldKey>>,
425    /// Structured-concurrency nursery stack. Each `scope { }` block pushes a
426    /// `TaskScope`; tasks spawned while it is innermost register their id here.
427    /// On normal exit (`TaskScopeExit`) the scope's tasks are joined and the
428    /// first error propagates; on unwind they are cancelled. Modeled on
429    /// `held_sync_guards` (push on enter, prune/cancel on frame/handler exit).
430    pub(crate) task_scopes: Vec<TaskScope>,
431    /// Counter for generating unique task IDs.
432    pub(crate) task_counter: u64,
433    /// Counter for logical runtime-context task groups.
434    pub(crate) runtime_context_counter: u64,
435    /// Logical runtime task context visible through `runtime_context()`.
436    pub(crate) runtime_context: crate::runtime_context::RuntimeContext,
437    /// Active deadline stack: (deadline_instant, frame_depth).
438    pub(crate) deadlines: Vec<(Instant, usize)>,
439    /// Uncatchable wall-clock deadline imposed by the embedding host.
440    pub(crate) execution_deadline: Arc<ExecutionDeadlineState>,
441    /// Breakpoints, keyed by source-file path so a breakpoint at line N
442    /// in `auto.harn` doesn't also fire when execution hits line N in an
443    /// imported lib. The empty-string key is a wildcard used by callers
444    /// that don't track source paths (legacy `set_breakpoints` API).
445    pub(crate) breakpoints: BTreeMap<String, std::collections::BTreeSet<usize>>,
446    /// Function-name breakpoints. Any closure call whose
447    /// `CompiledFunction.name` matches an entry here raises a stop on
448    /// entry, regardless of the call site's file or line. Lets the IDE
449    /// break on `llm_call` / `host_run_pipeline` / any user pipeline
450    /// function without pinning down a source location first.
451    pub(crate) function_breakpoints: std::collections::BTreeSet<String>,
452    /// Latched on `push_closure_frame` when the callee's name matches
453    /// `function_breakpoints`; consumed by the next step so the stop is
454    /// reported with reason="function breakpoint" and the breakpoint
455    /// name available for the DAP `stopped` event.
456    pub(crate) pending_function_bp: Option<String>,
457    /// Whether the VM is in step mode.
458    pub(crate) step_mode: bool,
459    /// The frame depth at which stepping started (for step-over).
460    pub(crate) step_frame_depth: usize,
461    /// Whether the VM is currently stopped at a debug point.
462    pub(crate) stopped: bool,
463    /// Last source line executed (to detect line changes).
464    pub(crate) last_line: usize,
465    /// Source directory for resolving imports.
466    pub(crate) source_dir: Option<std::path::PathBuf>,
467    /// Installed-package identity retained for the active lazy pipeline.
468    pub(crate) package_execution_guard:
469        Option<Arc<harn_modules::package_execution::PackageExecutionGuard>>,
470    /// Modules currently being imported (cycle prevention).
471    pub(crate) imported_paths: Vec<std::path::PathBuf>,
472    /// Imports that hit an in-progress module (an import cycle) and so could
473    /// not bind inline. Drained by `flush_deferred_cyclic_imports` once the
474    /// involved modules finish loading.
475    pub(crate) deferred_cyclic_imports: Vec<super::modules::DeferredCyclicImport>,
476    /// Loaded module cache keyed by canonical or synthetic module path.
477    pub(crate) module_cache: ModuleCache,
478    /// Immutable hydrated module bytecode shared across fresh VM isolates.
479    /// Runtime closures, registries, state, and init execution are not cached.
480    pub(crate) prepared_module_cache: crate::PreparedModuleCache,
481    /// Authority provenance used for the next root module graph. This can only
482    /// move from ordinary user code to trusted host dispatch before loading.
483    pub(crate) module_provenance: crate::module_artifact::ModuleProvenance,
484    /// Optional timing recorder shared by this VM execution tree.
485    pub(crate) module_phase_recorder: Option<super::ModulePhaseRecorder>,
486    /// Lazy manifest modules initialized by any child in this execution tree.
487    /// The complete export set is retained so handlers and predicates from the
488    /// same module share one module state across repeated child invocations.
489    pub(crate) lazy_callable_modules: LazyCallableModuleCache,
490    /// Source text keyed by canonical or synthetic module path for debugger
491    /// retrieval. Entries share the bytes owned by
492    /// [`crate::module_source`] rather than copying them.
493    pub(crate) source_cache: Arc<BTreeMap<std::path::PathBuf, Arc<str>>>,
494    /// This spawn's import-graph link table, when the entry-chunk cache proved
495    /// one current. Lets module loading name a module's artifact from a recorded
496    /// digest instead of reading the file to recompute the same digest. `None`
497    /// means every module takes the ordinary read-and-hash path.
498    pub(crate) graph_link_table: Option<Arc<crate::context_manifest::GraphLinkTable>>,
499    /// Source file path for error reporting.
500    pub(crate) source_file: Option<String>,
501    /// Source text for error reporting.
502    pub(crate) source_text: Option<String>,
503    /// Line-coverage accumulator. `Some` only while a coverage session is
504    /// active (see [`crate::coverage`]); folded into the global report on drop.
505    pub(crate) coverage: Option<crate::coverage::Coverage>,
506    /// Optional bridge for delegating unknown builtins in bridge mode.
507    pub(crate) bridge: Option<Arc<crate::bridge::HostBridge>>,
508    /// Builtins denied by sandbox mode (`--deny` / `--allow` flags).
509    pub(crate) denied_builtins: Arc<HashSet<String>>,
510    /// Cancellation token for cooperative graceful shutdown (set by parent).
511    pub(crate) cancel_token: Option<std::sync::Arc<std::sync::atomic::AtomicBool>>,
512    pub(crate) interrupt_signal_token: Option<std::sync::Arc<std::sync::Mutex<Option<String>>>>,
513    /// Remaining instruction-boundary checks before a requested host
514    /// cancellation is forcefully raised. This gives `is_cancelled()` loops a
515    /// deterministic chance to return cleanly without letting non-cooperative
516    /// CPU-bound code run forever.
517    pub(crate) cancel_grace_instructions_remaining: Option<usize>,
518    /// User-visible interrupt handlers registered through `std/signal`.
519    pub(crate) interrupt_handlers: Vec<InterruptHandler>,
520    pub(crate) next_interrupt_handle: i64,
521    pub(crate) pending_interrupt_signal: Option<String>,
522    pub(crate) interrupted: bool,
523    pub(crate) dispatching_interrupt: bool,
524    pub(crate) interrupt_handler_deadline: Option<Instant>,
525    /// Captured stack trace from the most recent error (fn_name, line, col).
526    pub(crate) error_stack_trace: Vec<(String, usize, usize, Option<String>)>,
527    /// Yield channel sender for generator execution. When set, `Op::Yield`
528    /// sends values through this channel instead of being a no-op.
529    pub(crate) yield_sender: Option<tokio::sync::mpsc::Sender<Result<VmValue, VmError>>>,
530    /// Project root directory (detected via harn.toml).
531    /// Used as base directory for metadata, store, and checkpoint operations.
532    pub(crate) project_root: Option<std::path::PathBuf>,
533    /// Global constants (e.g. `pi`, `e`). Checked as a fallback in `GetVar`
534    /// after the environment, so user-defined variables can shadow them.
535    pub(crate) globals: Arc<crate::value::DictMap>,
536    /// Root authority injected by the embedder for the compiler-generated
537    /// `main(harness: Harness)` call. This is deliberately not an environment
538    /// binding or global: user code can receive it only through an explicit
539    /// parameter.
540    pub(crate) root_harness: Option<VmValue>,
541    /// Runtime effect receipt state (shared recorder + per-VM call memo).
542    pub(crate) runtime_effects: crate::orchestration::RuntimeEffectState,
543    /// Optional debugger hook invoked when execution advances to a new source line.
544    pub(crate) debug_hook: Option<parking_lot::Mutex<Box<DebugHook>>>,
545    /// Effective runtime ceilings for this VM execution.
546    pub(crate) runtime_limits: RuntimeLimits,
547}
548
549/// Reusable VM baseline for hosts that need many clean executions with the
550/// same stable builtin/source setup.
551///
552/// The baseline intentionally does not snapshot execution state. Each
553/// instantiation gets fresh stacks, frames, tasks, cancellation fields, sync
554/// primitives, shared cells/maps/mailboxes, and debug state. Builtin tables are
555/// shared through `Arc` until a per-execution rebind needs copy-on-write.
556#[derive(Clone)]
557pub struct VmBaseline {
558    builtins: Arc<BTreeMap<String, VmBuiltinFn>>,
559    async_builtins: Arc<BTreeMap<String, VmAsyncBuiltinFn>>,
560    capability_methods:
561        Arc<BTreeMap<harn_builtin_meta::CapabilityId, BTreeMap<String, VmBuiltinDispatch>>>,
562    builtin_metadata: Arc<BTreeMap<String, VmBuiltinMetadata>>,
563    builtins_by_id: Arc<HashMap<BuiltinId, VmBuiltinEntry>>,
564    builtin_id_collisions: Arc<HashSet<BuiltinId>>,
565    source_dir: Option<std::path::PathBuf>,
566    source_file: Option<String>,
567    source_text: Option<String>,
568    project_root: Option<std::path::PathBuf>,
569    globals: Arc<crate::value::DictMap>,
570    root_harness: Option<VmValue>,
571    denied_builtins: Arc<HashSet<String>>,
572    prepared_module_cache: crate::PreparedModuleCache,
573    module_provenance: crate::module_artifact::ModuleProvenance,
574    /// Carried so a VM rebuilt from this baseline keeps resolving modules
575    /// through the same link table the original spawn was handed.
576    graph_link_table: Option<Arc<crate::context_manifest::GraphLinkTable>>,
577    runtime_limits: RuntimeLimits,
578}
579
580impl VmBaseline {
581    pub fn from_vm(vm: &Vm) -> Self {
582        Self {
583            builtins: Arc::clone(&vm.builtins),
584            async_builtins: Arc::clone(&vm.async_builtins),
585            capability_methods: Arc::clone(&vm.capability_methods),
586            builtin_metadata: Arc::clone(&vm.builtin_metadata),
587            builtins_by_id: Arc::clone(&vm.builtins_by_id),
588            builtin_id_collisions: Arc::clone(&vm.builtin_id_collisions),
589            source_dir: vm.source_dir.clone(),
590            source_file: vm.source_file.clone(),
591            source_text: vm.source_text.clone(),
592            project_root: vm.project_root.clone(),
593            globals: Arc::clone(&vm.globals),
594            root_harness: vm.root_harness.clone(),
595            denied_builtins: Arc::clone(&vm.denied_builtins),
596            prepared_module_cache: vm.prepared_module_cache.clone(),
597            module_provenance: vm.module_provenance,
598            graph_link_table: vm.graph_link_table.clone(),
599            runtime_limits: vm.runtime_limits,
600        }
601    }
602
603    pub fn instantiate(&self) -> Vm {
604        crate::initialize_runtime_assets();
605        let mut source_cache = BTreeMap::new();
606        if let (Some(file), Some(text)) = (&self.source_file, &self.source_text) {
607            source_cache.insert(std::path::PathBuf::from(file), Arc::from(text.as_str()));
608        }
609        if let Some(dir) = &self.source_dir {
610            crate::stdlib::set_thread_source_dir(dir);
611        }
612
613        let mut vm = Vm {
614            stack: Vec::with_capacity(256),
615            env: VmEnv::new(),
616            output: String::new(),
617            builtins: Arc::clone(&self.builtins),
618            async_builtins: Arc::clone(&self.async_builtins),
619            capability_methods: Arc::clone(&self.capability_methods),
620            builtin_metadata: Arc::clone(&self.builtin_metadata),
621            builtins_by_id: Arc::clone(&self.builtins_by_id),
622            builtin_id_collisions: Arc::clone(&self.builtin_id_collisions),
623            iterators: Vec::new(),
624            frames: Vec::new(),
625            exception_handlers: Vec::new(),
626            spawned_tasks: BTreeMap::new(),
627            process_exit_request: Arc::new(ProcessExitRequest::new()),
628            sync_runtime: Arc::new(crate::synchronization::VmSyncRuntime::new()),
629            shared_state_runtime: Arc::new(crate::shared_state::VmSharedStateRuntime::new()),
630            inline_cache_sets: Vec::new(),
631            inline_cache_set_by_chunk: HashMap::new(),
632            pool_registry: crate::stdlib::pool::new_pool_registry(),
633            llm_mock_context: crate::llm::mock::LlmMockContext::for_new_vm(),
634            package_snapshot_registry: Arc::new(Default::default()),
635            wait_for_graph: Arc::new(crate::wait_for_graph::VmWaitForGraph::new()),
636            held_sync_guards: Vec::new(),
637            inherited_held_keys: Arc::new(Vec::new()),
638            task_scopes: Vec::new(),
639            task_counter: 0,
640            runtime_context_counter: 0,
641            runtime_context: crate::runtime_context::RuntimeContext::root(),
642            deadlines: Vec::new(),
643            execution_deadline: super::execution::new_execution_deadline_state(None),
644            breakpoints: BTreeMap::new(),
645            function_breakpoints: std::collections::BTreeSet::new(),
646            pending_function_bp: None,
647            step_mode: false,
648            step_frame_depth: 0,
649            stopped: false,
650            last_line: 0,
651            source_dir: self.source_dir.clone(),
652            package_execution_guard: None,
653            imported_paths: Vec::new(),
654            deferred_cyclic_imports: Vec::new(),
655            module_cache: Arc::new(BTreeMap::new()),
656            prepared_module_cache: self.prepared_module_cache.clone(),
657            module_provenance: self.module_provenance,
658            module_phase_recorder: None,
659            lazy_callable_modules: Arc::new(crate::value::VmMutex::new(BTreeMap::new())),
660            source_cache: Arc::new(source_cache),
661            graph_link_table: self.graph_link_table.clone(),
662            source_file: self.source_file.clone(),
663            source_text: self.source_text.clone(),
664            coverage: crate::coverage::for_primary(self.source_file.as_deref()),
665            bridge: None,
666            denied_builtins: Arc::clone(&self.denied_builtins),
667            cancel_token: None,
668            interrupt_signal_token: None,
669            cancel_grace_instructions_remaining: None,
670            interrupt_handlers: Vec::new(),
671            next_interrupt_handle: 1,
672            pending_interrupt_signal: None,
673            interrupted: false,
674            dispatching_interrupt: false,
675            interrupt_handler_deadline: None,
676            error_stack_trace: Vec::new(),
677            yield_sender: None,
678            project_root: self.project_root.clone(),
679            globals: Arc::clone(&self.globals),
680            root_harness: self.root_harness.clone(),
681            runtime_effects: crate::orchestration::RuntimeEffectState::fresh(),
682            debug_hook: None,
683            runtime_limits: self.runtime_limits,
684        };
685
686        crate::stdlib::rebind_execution_state_builtins(&mut vm);
687        vm
688    }
689}
690
691impl Vm {
692    pub(crate) fn ensure_execution_available(&self) -> Result<(), VmError> {
693        if self.execution_deadline.is_abandoned() {
694            return Err(VmError::AbandonedExecution);
695        }
696        Ok(())
697    }
698
699    pub(crate) fn fresh_local_slots(chunk: &Chunk) -> Vec<LocalSlot> {
700        chunk
701            .local_slots
702            .iter()
703            .map(|_| LocalSlot {
704                value: VmValue::Nil,
705                initialized: false,
706                synced: false,
707            })
708            .collect()
709    }
710
711    pub(crate) fn bind_param_slots(
712        slots: &mut [LocalSlot],
713        func: &crate::chunk::CompiledFunction,
714        args: &[VmValue],
715        synced: bool,
716    ) {
717        Self::bind_param_slots_args(slots, func, &super::CallArgs::Slice(args), synced);
718    }
719
720    pub(crate) fn bind_param_slots_args(
721        slots: &mut [LocalSlot],
722        func: &crate::chunk::CompiledFunction,
723        args: &super::CallArgs<'_>,
724        synced: bool,
725    ) {
726        let param_count = func.params.len();
727        for (i, _param) in func.params.iter().enumerate() {
728            if i >= slots.len() {
729                break;
730            }
731            if func.has_rest_param && i == param_count - 1 {
732                let rest_args = args.to_vec_from(i);
733                slots[i].value = VmValue::List(std::sync::Arc::new(rest_args));
734                slots[i].initialized = true;
735                slots[i].synced = synced;
736            } else if let Some(arg) = args.get(i) {
737                slots[i].value = arg.clone();
738                slots[i].initialized = true;
739                slots[i].synced = synced;
740            }
741        }
742    }
743
744    pub(crate) fn visible_variables(&self) -> crate::value::DictMap {
745        let mut vars = self.env.all_variables();
746        let Some(frame) = self.frames.last() else {
747            return vars;
748        };
749        for (slot, info) in frame.local_slots.iter().zip(frame.chunk.local_slots.iter()) {
750            if slot.initialized && info.scope_depth <= frame.local_scope_depth {
751                vars.insert(crate::value::intern_key(&info.name), slot.value.clone());
752            }
753        }
754        vars
755    }
756
757    pub(crate) fn sync_current_frame_locals_to_env(&mut self) {
758        let frames = &mut self.frames;
759        let env = &mut self.env;
760        let Some(frame) = frames.last_mut() else {
761            return;
762        };
763        let local_scope_base = frame.local_scope_base;
764        let local_scope_depth = frame.local_scope_depth;
765        for (slot, info) in frame
766            .local_slots
767            .iter_mut()
768            .zip(frame.chunk.local_slots.iter())
769        {
770            if slot.initialized && !slot.synced && info.scope_depth <= local_scope_depth {
771                slot.synced = true;
772                let scope_idx = local_scope_base + info.scope_depth;
773                while env.scopes.len() <= scope_idx {
774                    env.push_scope();
775                }
776                // Slot-backed locals are never boxed captures (those resolve
777                // to `None` in the compiler and are defined straight into the
778                // env as `Cell`s), so syncing a slot always yields a `Value`.
779                Arc::make_mut(&mut env.scopes[scope_idx].vars).insert(
780                    info.name.clone(),
781                    crate::value::Binding::Value {
782                        value: slot.value.clone(),
783                        mutable: info.mutable,
784                    },
785                );
786            }
787        }
788    }
789
790    pub(crate) fn closure_call_env_for_current_frame(
791        &self,
792        closure: &crate::value::VmClosure,
793    ) -> VmEnv {
794        if closure.module_state().is_some() {
795            return closure.env.cloned_for_call();
796        }
797        let call_env = Self::closure_call_env(&self.env, closure);
798        // Same compile-time short-circuit as the env walk in
799        // `closure_call_env`: when the callee body never resolves an
800        // outer name through the env, injecting closure-typed *slot*
801        // locals from the caller's frame is wasted work too.
802        if !closure.func.chunk.references_outer_names {
803            return call_env;
804        }
805        let mut call_env = call_env;
806        let Some(frame) = self.frames.last() else {
807            return call_env;
808        };
809        for (slot, info) in frame
810            .local_slots
811            .iter()
812            .zip(frame.chunk.local_slots.iter())
813            .filter(|(slot, info)| slot.initialized && info.scope_depth <= frame.local_scope_depth)
814        {
815            if matches!(slot.value, VmValue::Closure(_)) && !call_env.contains(&info.name) {
816                let _ = call_env.define(&info.name, slot.value.clone(), info.mutable);
817            }
818        }
819        call_env
820    }
821
822    pub(crate) fn active_local_slot_value(&self, name: &str) -> Option<VmValue> {
823        let frame = self.frames.last()?;
824        let idx = self.active_local_slot_index(name)?;
825        frame.local_slots.get(idx).map(|slot| slot.value.clone())
826    }
827
828    /// Returns the slot index of an initialized active local with the given
829    /// name, walking from innermost to outermost scope. Used by legacy by-name
830    /// hot paths that still want to mutate the slot value in place without
831    /// paying a defensive `VmValue::clone` first.
832    pub(crate) fn active_local_slot_index(&self, name: &str) -> Option<usize> {
833        let frame = self.frames.last()?;
834        for (idx, info) in frame.chunk.local_slots.iter().enumerate().rev() {
835            if info.name == name && info.scope_depth <= frame.local_scope_depth {
836                if let Some(slot) = frame.local_slots.get(idx) {
837                    if slot.initialized {
838                        return Some(idx);
839                    }
840                }
841            }
842        }
843        None
844    }
845
846    pub(crate) fn assign_active_local_slot(
847        &mut self,
848        name: &str,
849        value: VmValue,
850        debug: bool,
851    ) -> Result<bool, VmError> {
852        let Some(frame) = self.frames.last_mut() else {
853            return Ok(false);
854        };
855        for (idx, info) in frame.chunk.local_slots.iter().enumerate().rev() {
856            if info.name == name && info.scope_depth <= frame.local_scope_depth {
857                if !debug && !info.mutable {
858                    return Err(VmError::ImmutableAssignment(name.to_string()));
859                }
860                if let Some(slot) = frame.local_slots.get_mut(idx) {
861                    crate::value::recursion::dismantle(std::mem::replace(&mut slot.value, value));
862                    slot.initialized = true;
863                    slot.synced = false;
864                    return Ok(true);
865                }
866            }
867        }
868        Ok(false)
869    }
870
871    pub fn new() -> Self {
872        crate::initialize_runtime_assets();
873        Self {
874            stack: Vec::with_capacity(256),
875            env: VmEnv::new(),
876            output: String::new(),
877            builtins: Arc::new(BTreeMap::new()),
878            async_builtins: Arc::new(BTreeMap::new()),
879            capability_methods: Arc::new(BTreeMap::new()),
880            builtin_metadata: Arc::new(BTreeMap::new()),
881            builtins_by_id: Arc::new(HashMap::new()),
882            builtin_id_collisions: Arc::new(HashSet::new()),
883            iterators: Vec::new(),
884            frames: Vec::new(),
885            exception_handlers: Vec::new(),
886            spawned_tasks: BTreeMap::new(),
887            process_exit_request: Arc::new(ProcessExitRequest::new()),
888            sync_runtime: Arc::new(crate::synchronization::VmSyncRuntime::new()),
889            shared_state_runtime: Arc::new(crate::shared_state::VmSharedStateRuntime::new()),
890            inline_cache_sets: Vec::new(),
891            inline_cache_set_by_chunk: HashMap::new(),
892            pool_registry: crate::stdlib::pool::new_pool_registry(),
893            llm_mock_context: crate::llm::mock::LlmMockContext::for_new_vm(),
894            package_snapshot_registry: Arc::new(Default::default()),
895            wait_for_graph: Arc::new(crate::wait_for_graph::VmWaitForGraph::new()),
896            held_sync_guards: Vec::new(),
897            inherited_held_keys: Arc::new(Vec::new()),
898            task_scopes: Vec::new(),
899            task_counter: 0,
900            runtime_context_counter: 0,
901            runtime_context: crate::runtime_context::RuntimeContext::root(),
902            deadlines: Vec::new(),
903            execution_deadline: super::execution::new_execution_deadline_state(None),
904            breakpoints: BTreeMap::new(),
905            function_breakpoints: std::collections::BTreeSet::new(),
906            pending_function_bp: None,
907            step_mode: false,
908            step_frame_depth: 0,
909            stopped: false,
910            last_line: 0,
911            source_dir: None,
912            package_execution_guard: None,
913            imported_paths: Vec::new(),
914            deferred_cyclic_imports: Vec::new(),
915            module_cache: Arc::new(BTreeMap::new()),
916            prepared_module_cache: crate::PreparedModuleCache::default(),
917            module_provenance: crate::module_artifact::ModuleProvenance::User,
918            module_phase_recorder: None,
919            lazy_callable_modules: Arc::new(crate::value::VmMutex::new(BTreeMap::new())),
920            source_cache: Arc::new(BTreeMap::new()),
921            graph_link_table: None,
922            source_file: None,
923            source_text: None,
924            coverage: crate::coverage::for_primary(None),
925            bridge: None,
926            denied_builtins: Arc::new(HashSet::new()),
927            cancel_token: None,
928            interrupt_signal_token: None,
929            cancel_grace_instructions_remaining: None,
930            interrupt_handlers: Vec::new(),
931            next_interrupt_handle: 1,
932            pending_interrupt_signal: None,
933            interrupted: false,
934            dispatching_interrupt: false,
935            interrupt_handler_deadline: None,
936            error_stack_trace: Vec::new(),
937            yield_sender: None,
938            project_root: None,
939            globals: Arc::new(crate::value::DictMap::new()),
940            root_harness: None,
941            runtime_effects: crate::orchestration::RuntimeEffectState::fresh(),
942            debug_hook: None,
943            runtime_limits: RuntimeLimits::default(),
944        }
945    }
946
947    pub fn baseline(&self) -> VmBaseline {
948        VmBaseline::from_vm(self)
949    }
950
951    /// Typed, runtime-resolved effects attempted by this VM execution tree.
952    pub fn executed_effects(&self) -> Vec<crate::orchestration::EffectRecord> {
953        self.runtime_effects.snapshot()
954    }
955
956    /// Clear the execution-local effect recorder and per-VM memo.
957    pub fn clear_executed_effects(&mut self) {
958        self.runtime_effects.clear();
959    }
960
961    pub(crate) fn record_capability_effects(
962        &mut self,
963        capability: harn_builtin_meta::CapabilityId,
964        method: &str,
965        args: &[VmValue],
966    ) {
967        self.runtime_effects
968            .record_capability(capability, method, args);
969    }
970
971    pub(crate) fn record_builtin_contract_effects(&mut self, name: &str, args: &[VmValue]) {
972        let Some(entry) = crate::stdlib::recorded_effect_builtin_manifest_entry(name) else {
973            return;
974        };
975        self.record_builtin_effect_specs(entry.contract.effects, args);
976    }
977
978    pub(crate) fn record_builtin_effect_specs(
979        &mut self,
980        specs: &'static [harn_builtin_meta::EffectSpec],
981        args: &[VmValue],
982    ) {
983        self.runtime_effects.record_specs(specs, args);
984    }
985
986    /// Replace the scoped immutable module-template cache used by this VM.
987    /// Fresh runtime module state is still instantiated for every VM.
988    pub fn set_prepared_module_cache(&mut self, cache: crate::PreparedModuleCache) {
989        self.prepared_module_cache = cache;
990    }
991
992    /// Hand this VM the link table for the import graph it is about to run.
993    ///
994    /// Only an entry-chunk cache hit can supply one, because only that path holds
995    /// a manifest proving the recorded digests still describe the files on disk.
996    /// A table never changes what a module resolves to: one the table does not
997    /// name, or whose artifact is no longer on disk, is read and compiled as
998    /// usual.
999    pub fn set_graph_link_table(
1000        &mut self,
1001        link_table: Option<Arc<crate::context_manifest::GraphLinkTable>>,
1002    ) {
1003        self.graph_link_table = link_table;
1004    }
1005
1006    /// Return the effective runtime limit profile for this VM.
1007    pub fn runtime_limits(&self) -> RuntimeLimits {
1008        self.runtime_limits
1009    }
1010
1011    /// Return a host/debug report describing the VM's effective runtime limits.
1012    pub fn runtime_limit_report(&self) -> crate::RuntimeLimitsReport {
1013        self.runtime_limits.report()
1014    }
1015
1016    /// Returns true if any debugging affordance is active — DAP hook,
1017    /// line breakpoints, or function breakpoints. Call-site code uses
1018    /// this to decide whether to capture per-frame restart snapshots
1019    /// (`initial_env`, `initial_local_slots`); without a debugger those
1020    /// snapshots are dead weight, so skipping them removes two
1021    /// allocations from every function call hot path.
1022    ///
1023    /// All three signals are stable across a function call's lifetime
1024    /// (they're set before pipeline execution starts), so the gate is
1025    /// consistent between frame creation and any later `restart_frame`
1026    /// invocation. The three `is_empty` checks compile to a handful of
1027    /// branch-predicted memory probes — cheaper than a single
1028    /// `BTreeMap` clone, which is what we're avoiding.
1029    #[inline]
1030    pub(crate) fn debugger_attached(&self) -> bool {
1031        self.debug_hook.is_some()
1032            || !self.breakpoints.is_empty()
1033            || !self.function_breakpoints.is_empty()
1034    }
1035
1036    /// Set the bridge for delegating unknown builtins in bridge mode.
1037    pub fn set_bridge(&mut self, bridge: Arc<crate::bridge::HostBridge>) {
1038        self.bridge = Some(bridge);
1039    }
1040
1041    /// Set builtins that are denied in sandbox mode.
1042    /// When called, the given builtin names will produce a permission error.
1043    pub fn set_denied_builtins(&mut self, mut denied: HashSet<String>) {
1044        // A builtin may have a private registry name plus one or more public
1045        // aliases. Normalize the whole family once so no dispatch path can
1046        // escape policy merely by using another name for the same handler.
1047        let denied_canonical_names = denied
1048            .iter()
1049            .filter_map(|name| crate::stdlib::builtin_manifest_entry(name))
1050            .map(|entry| entry.canonical_name)
1051            .collect::<HashSet<_>>();
1052        if !denied_canonical_names.is_empty() {
1053            denied.extend(
1054                crate::stdlib::all_builtin_manifest()
1055                    .iter()
1056                    .filter(|entry| denied_canonical_names.contains(entry.canonical_name))
1057                    .map(|entry| entry.name.to_string()),
1058            );
1059        }
1060        self.denied_builtins = Arc::new(denied);
1061    }
1062
1063    /// Set source info for error reporting (file path and source text).
1064    pub fn set_source_info(&mut self, file: &str, text: &str) {
1065        self.source_file = Some(file.to_string());
1066        self.source_text = Some(text.to_string());
1067        if let Some(cov) = self.coverage.as_mut() {
1068            cov.set_primary_file(file);
1069        }
1070        Arc::make_mut(&mut self.source_cache)
1071            .insert(std::path::PathBuf::from(file), Arc::from(text));
1072    }
1073
1074    /// Initialize execution (push the initial frame).
1075    pub fn start(&mut self, chunk: &Chunk) -> Result<(), VmError> {
1076        self.ensure_execution_available()?;
1077        // The top-level pipeline frame captures env at start so
1078        // restartFrame on the outermost frame rewinds to the
1079        // pre-pipeline state — basically "restart session" in
1080        // debugger terms. Skipped when no debugger is attached:
1081        // the snapshot is dead weight in that case and dominates
1082        // call-overhead bench numbers (~5-10%).
1083        let debugger = self.debugger_attached();
1084        let initial_env = if debugger {
1085            Some(self.env.clone())
1086        } else {
1087            None
1088        };
1089        let initial_local_slots = if debugger {
1090            Some(Self::fresh_local_slots(chunk))
1091        } else {
1092            None
1093        };
1094        let chunk = Arc::new(chunk.clone());
1095        let local_slots = Self::fresh_local_slots(&chunk);
1096        let inline_cache_set = self.inline_cache_set_index_for_chunk(&chunk);
1097        self.frames.push(CallFrame {
1098            chunk,
1099            inline_cache_set,
1100            ip: 0,
1101            stack_base: self.stack.len(),
1102            saved_env: self.env.clone(),
1103            initial_env,
1104            initial_local_slots,
1105            saved_iterator_depth: self.iterators.len(),
1106            fn_name: String::new(),
1107            argc: 0,
1108            saved_source_dir: None,
1109            module_functions: None,
1110            module_state: None,
1111            local_slots,
1112            local_scope_base: self.env.scope_depth().saturating_sub(1),
1113            local_scope_depth: 0,
1114        });
1115        Ok(())
1116    }
1117
1118    /// Create a child VM that shares builtins and env but has fresh execution state.
1119    /// Used for parallel/spawn to fork the VM for concurrent tasks.
1120    pub(crate) fn child_vm(&self) -> Vm {
1121        Vm {
1122            stack: Vec::with_capacity(64),
1123            env: self.env.clone(),
1124            output: String::new(),
1125            builtins: Arc::clone(&self.builtins),
1126            async_builtins: Arc::clone(&self.async_builtins),
1127            capability_methods: Arc::clone(&self.capability_methods),
1128            builtin_metadata: Arc::clone(&self.builtin_metadata),
1129            builtins_by_id: Arc::clone(&self.builtins_by_id),
1130            builtin_id_collisions: Arc::clone(&self.builtin_id_collisions),
1131            iterators: Vec::new(),
1132            frames: Vec::new(),
1133            exception_handlers: Vec::new(),
1134            spawned_tasks: BTreeMap::new(),
1135            process_exit_request: Arc::clone(&self.process_exit_request),
1136            sync_runtime: self.sync_runtime.clone(),
1137            shared_state_runtime: self.shared_state_runtime.clone(),
1138            inline_cache_sets: Vec::new(),
1139            inline_cache_set_by_chunk: HashMap::new(),
1140            pool_registry: self.pool_registry.clone(),
1141            llm_mock_context: self.llm_mock_context.clone(),
1142            package_snapshot_registry: self.package_snapshot_registry.clone(),
1143            wait_for_graph: self.wait_for_graph.clone(),
1144            held_sync_guards: Vec::new(),
1145            inherited_held_keys: Arc::new(Vec::new()),
1146            task_scopes: Vec::new(),
1147            task_counter: 0,
1148            runtime_context_counter: self.runtime_context_counter,
1149            runtime_context: self.runtime_context.clone(),
1150            deadlines: self.deadlines.clone(),
1151            execution_deadline: self.execution_deadline.fork(),
1152            breakpoints: BTreeMap::new(),
1153            function_breakpoints: std::collections::BTreeSet::new(),
1154            pending_function_bp: None,
1155            step_mode: false,
1156            step_frame_depth: 0,
1157            stopped: false,
1158            last_line: 0,
1159            source_dir: self.source_dir.clone(),
1160            package_execution_guard: self.package_execution_guard.clone(),
1161            imported_paths: Vec::new(),
1162            deferred_cyclic_imports: Vec::new(),
1163            module_cache: Arc::clone(&self.module_cache),
1164            prepared_module_cache: self.prepared_module_cache.clone(),
1165            module_provenance: self.module_provenance,
1166            module_phase_recorder: self.module_phase_recorder.clone(),
1167            lazy_callable_modules: Arc::clone(&self.lazy_callable_modules),
1168            source_cache: Arc::clone(&self.source_cache),
1169            graph_link_table: self.graph_link_table.clone(),
1170            source_file: self.source_file.clone(),
1171            source_text: self.source_text.clone(),
1172            coverage: crate::coverage::for_primary(self.source_file.as_deref()),
1173            bridge: self.bridge.clone(),
1174            denied_builtins: Arc::clone(&self.denied_builtins),
1175            cancel_token: self.cancel_token.clone(),
1176            interrupt_signal_token: self.interrupt_signal_token.clone(),
1177            cancel_grace_instructions_remaining: None,
1178            interrupt_handlers: Vec::new(),
1179            next_interrupt_handle: 1,
1180            pending_interrupt_signal: None,
1181            interrupted: self.interrupted,
1182            dispatching_interrupt: false,
1183            interrupt_handler_deadline: None,
1184            error_stack_trace: Vec::new(),
1185            yield_sender: None,
1186            project_root: self.project_root.clone(),
1187            globals: Arc::clone(&self.globals),
1188            root_harness: self.root_harness.clone(),
1189            runtime_effects: crate::orchestration::RuntimeEffectState::with_shared_recorder(
1190                Arc::clone(&self.runtime_effects.recorder),
1191            ),
1192            debug_hook: None,
1193            runtime_limits: self.runtime_limits,
1194        }
1195    }
1196
1197    /// Create a child VM for external adapters that need to invoke Harn
1198    /// closures while sharing the parent's builtins, globals, and module state.
1199    pub(crate) fn child_vm_for_host(&self) -> Vm {
1200        self.child_vm()
1201    }
1202
1203    pub(crate) fn request_process_exit(&self, code: i32) {
1204        self.process_exit_request.request(code);
1205    }
1206
1207    pub(crate) fn requested_process_exit(&self) -> Option<i32> {
1208        self.process_exit_request.code()
1209    }
1210
1211    /// Request cancellation for every outstanding child task owned by this VM
1212    /// and then abort the join handles. This prevents un-awaited spawned tasks
1213    /// from outliving their parent execution scope.
1214    pub(crate) fn cancel_spawned_tasks(&mut self) {
1215        for (_, task) in std::mem::take(&mut self.spawned_tasks) {
1216            task.cancel_token
1217                .store(true, std::sync::atomic::Ordering::SeqCst);
1218            task.handle.abort();
1219        }
1220    }
1221
1222    /// Set the source directory for import resolution and introspection.
1223    /// Also auto-detects the project root if not already set.
1224    pub fn set_source_dir(&mut self, dir: &std::path::Path) {
1225        let dir = crate::stdlib::process::normalize_context_path(dir);
1226        self.source_dir = Some(dir.clone());
1227        crate::stdlib::set_thread_source_dir(&dir);
1228        // Auto-detect project root if not explicitly set.
1229        if self.project_root.is_none() {
1230            self.project_root = crate::stdlib::process::find_project_root(&dir);
1231        }
1232    }
1233
1234    /// Explicitly set the project root directory.
1235    /// Used by ACP/CLI to override auto-detection.
1236    pub fn set_project_root(&mut self, root: &std::path::Path) {
1237        self.project_root = Some(root.to_path_buf());
1238    }
1239
1240    /// Get only the explicit or auto-detected project root, without falling
1241    /// back to `source_dir`.
1242    pub(crate) fn explicit_project_root(&self) -> Option<&std::path::Path> {
1243        self.project_root.as_deref()
1244    }
1245
1246    /// Get the project root directory, falling back to source_dir.
1247    pub fn project_root(&self) -> Option<&std::path::Path> {
1248        self.project_root.as_deref().or(self.source_dir.as_deref())
1249    }
1250
1251    /// Set a global constant (e.g. `pi`, `e`).
1252    /// Stored separately from the environment so user-defined variables can shadow them.
1253    pub fn set_global(&mut self, name: &str, value: VmValue) {
1254        Arc::make_mut(&mut self.globals).insert(crate::value::intern_key(name), value);
1255    }
1256
1257    /// Read a previously-installed ordinary global constant.
1258    pub fn global(&self, name: &str) -> Option<&VmValue> {
1259        self.globals.get(name)
1260    }
1261
1262    /// Install the script's root authority for the compiler-generated
1263    /// `main(harness: Harness)` call. It is kept out of all source-visible
1264    /// namespaces.
1265    pub fn set_harness(&mut self, harness: crate::harness::Harness) {
1266        self.root_harness = Some(harness.into_vm_value());
1267    }
1268
1269    pub(crate) fn harness(&self) -> Option<&crate::harness::VmHarness> {
1270        match self.root_harness.as_ref() {
1271            Some(VmValue::Harness(handle)) => Some(handle),
1272            _ => None,
1273        }
1274    }
1275
1276    /// Clone the embedder-owned root authority for an explicit exported
1277    /// function parameter. This does not install a source-visible binding.
1278    pub fn root_harness_value(&self) -> Option<VmValue> {
1279        self.root_harness.clone()
1280    }
1281
1282    /// Get the captured output.
1283    pub fn output(&self) -> &str {
1284        &self.output
1285    }
1286
1287    /// Drain and return the captured output, leaving the buffer empty.
1288    /// Used by the async-builtin dispatch path to forward closure output
1289    /// from a child VM back to its parent.
1290    pub fn take_output(&mut self) -> String {
1291        std::mem::take(&mut self.output)
1292    }
1293
1294    /// Append text to this VM's captured output. Used to forward output
1295    /// from child VMs (e.g. closures invoked via `call_closure_pub`)
1296    /// back into the parent stream.
1297    pub fn append_output(&mut self, text: &str) {
1298        self.output.push_str(text);
1299    }
1300
1301    pub(crate) fn pop(&mut self) -> Result<VmValue, VmError> {
1302        self.stack.pop().ok_or(VmError::StackUnderflow)
1303    }
1304
1305    pub(crate) fn peek(&self) -> Result<&VmValue, VmError> {
1306        self.stack.last().ok_or(VmError::StackUnderflow)
1307    }
1308
1309    pub(crate) fn const_str(c: &Constant) -> Result<&str, VmError> {
1310        match c {
1311            Constant::String(s) => Ok(s.as_str()),
1312            _ => Err(VmError::TypeError("expected string constant".into())),
1313        }
1314    }
1315
1316    pub(crate) fn release_sync_guards_for_current_scope(&mut self) {
1317        let depth = self.env.scope_depth();
1318        self.held_sync_guards
1319            .retain(|guard| guard.env_scope_depth < depth);
1320        // A `scope { }` torn down without a normal `TaskScopeExit` (break /
1321        // continue out of it) leaves a dangling nursery — cancel its tasks.
1322        self.cancel_task_scopes_where(|s| s.env_scope_depth >= depth);
1323    }
1324
1325    pub(crate) fn release_sync_guards_after_unwind(
1326        &mut self,
1327        frame_depth: usize,
1328        env_scope_depth: usize,
1329    ) {
1330        self.held_sync_guards.retain(|guard| {
1331            guard.frame_depth <= frame_depth && guard.env_scope_depth <= env_scope_depth
1332        });
1333        // Cancel nurseries opened above the catch handler (a `throw` unwound
1334        // past their `TaskScopeExit`).
1335        self.cancel_task_scopes_where(|s| {
1336            !(s.frame_depth <= frame_depth && s.env_scope_depth <= env_scope_depth)
1337        });
1338    }
1339
1340    pub(crate) fn release_sync_guards_for_frame(&mut self, frame_depth: usize) {
1341        self.held_sync_guards
1342            .retain(|guard| guard.frame_depth != frame_depth);
1343        // Cancel any nursery whose `scope {}` block belonged to the frame being
1344        // torn down (e.g. a `return` jumped past its `TaskScopeExit`).
1345        self.cancel_task_scopes_where(|s| s.frame_depth == frame_depth);
1346    }
1347
1348    pub(crate) fn adopt_sync_permit_for_current_scope(
1349        &mut self,
1350        permit: crate::value::VmSyncPermitHandle,
1351    ) {
1352        if permit.is_released()
1353            || self
1354                .held_sync_guards
1355                .iter()
1356                .any(|guard| guard._permit.same_lease(&permit))
1357        {
1358            return;
1359        }
1360        self.held_sync_guards
1361            .push(crate::synchronization::VmSyncHeldGuard {
1362                _permit: permit,
1363                frame_depth: self.frames.len(),
1364                env_scope_depth: self.env.scope_depth(),
1365            });
1366    }
1367
1368    /// Deregister a task id from every open nursery (it was explicitly
1369    /// `await`ed, so it must not be double-joined or cancelled at scope exit).
1370    pub(crate) fn deregister_task_from_scopes(&mut self, id: &str) {
1371        for scope in &mut self.task_scopes {
1372            scope.task_ids.retain(|t| t != id);
1373        }
1374    }
1375
1376    /// Cancel and remove every task scope matching `doomed`, aborting its bound
1377    /// tasks (used when a `scope {}` is torn down without a normal join).
1378    fn cancel_task_scopes_where<F: Fn(&TaskScope) -> bool>(&mut self, doomed: F) {
1379        let mut i = 0;
1380        while i < self.task_scopes.len() {
1381            if doomed(&self.task_scopes[i]) {
1382                let scope = self.task_scopes.remove(i);
1383                for id in &scope.task_ids {
1384                    if let Some(task) = self.spawned_tasks.remove(id) {
1385                        task.cancel_token
1386                            .store(true, std::sync::atomic::Ordering::SeqCst);
1387                        task.handle.abort();
1388                    }
1389                }
1390            } else {
1391                i += 1;
1392            }
1393        }
1394    }
1395
1396    /// Total live permits this VM already holds for `kind:key`. The held-set is
1397    /// tiny (bounded by lexical nesting and explicit sync acquisitions), so this
1398    /// scan is cheap and only runs on the rare blocking-acquire path.
1399    pub(crate) fn held_permits_for(&self, kind: &str, key: &str) -> u32 {
1400        let own: u32 = self
1401            .held_sync_guards
1402            .iter()
1403            .filter(|guard| {
1404                !guard._permit.is_released()
1405                    && guard._permit.kind() == kind
1406                    && guard._permit.key() == key
1407            })
1408            .map(|guard| guard._permit.permits())
1409            .sum();
1410        let inherited: u32 = self
1411            .inherited_held_keys
1412            .iter()
1413            .filter(|held| held.kind == kind && held.key == key)
1414            .map(|held| held.permits)
1415            .sum();
1416        own + inherited
1417    }
1418
1419    /// Every live sync permit held by this VM *and* its suspended ancestors: the
1420    /// transitive held-set seen by an inline child.
1421    pub(crate) fn combined_held_keys(&self) -> Vec<crate::synchronization::VmSyncHeldKey> {
1422        let mut keys: Vec<crate::synchronization::VmSyncHeldKey> = self
1423            .held_sync_guards
1424            .iter()
1425            .filter_map(|guard| crate::synchronization::VmSyncHeldKey::from_permit(&guard._permit))
1426            .collect();
1427        keys.extend(self.inherited_held_keys.iter().cloned());
1428        keys
1429    }
1430
1431    /// Clone a child VM for an **inline, same-task** execution (an async builtin
1432    /// awaited while this VM is parked, or a user closure that builtin runs and
1433    /// awaits). The child inherits this VM's transitive held-lock keys so a
1434    /// re-acquire of a parent-held lock is caught as a self-deadlock
1435    /// (HARN-ORC-011). Use plain `child_vm()` for new concurrent tasks.
1436    pub(crate) fn child_vm_inline(&self) -> Vm {
1437        let mut child = self.child_vm();
1438        child.inherited_held_keys = Arc::new(self.combined_held_keys());
1439        child
1440    }
1441}
1442
1443impl Drop for Vm {
1444    fn drop(&mut self) {
1445        if let Some(coverage) = self.coverage.take() {
1446            crate::coverage::merge_into_global(coverage);
1447        }
1448        self.cancel_spawned_tasks();
1449    }
1450}
1451
1452impl Default for Vm {
1453    fn default() -> Self {
1454        Self::new()
1455    }
1456}
1457
1458#[cfg(test)]
1459#[path = "state_tests.rs"]
1460mod tests;