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    /// Verified, closed-program module templates for this execution tree.
500    /// Unlike the generic graph link table, absence is never interpreted as a
501    /// source-cache miss by the linked-program path.
502    pub(crate) linked_program_repository:
503        Option<Arc<crate::linked_program::LinkedProgramRepository>>,
504    /// Source file path for error reporting.
505    pub(crate) source_file: Option<String>,
506    /// Source text for error reporting.
507    pub(crate) source_text: Option<String>,
508    /// Line-coverage accumulator. `Some` only while a coverage session is
509    /// active (see [`crate::coverage`]); folded into the global report on drop.
510    pub(crate) coverage: Option<crate::coverage::Coverage>,
511    /// Optional bridge for delegating unknown builtins in bridge mode.
512    pub(crate) bridge: Option<Arc<crate::bridge::HostBridge>>,
513    /// Builtins denied by sandbox mode (`--deny` / `--allow` flags).
514    pub(crate) denied_builtins: Arc<HashSet<String>>,
515    /// Cancellation token for cooperative graceful shutdown (set by parent).
516    pub(crate) cancel_token: Option<std::sync::Arc<std::sync::atomic::AtomicBool>>,
517    pub(crate) interrupt_signal_token: Option<std::sync::Arc<std::sync::Mutex<Option<String>>>>,
518    /// Remaining instruction-boundary checks before a requested host
519    /// cancellation is forcefully raised. This gives `is_cancelled()` loops a
520    /// deterministic chance to return cleanly without letting non-cooperative
521    /// CPU-bound code run forever.
522    pub(crate) cancel_grace_instructions_remaining: Option<usize>,
523    /// User-visible interrupt handlers registered through `std/signal`.
524    pub(crate) interrupt_handlers: Vec<InterruptHandler>,
525    pub(crate) next_interrupt_handle: i64,
526    pub(crate) pending_interrupt_signal: Option<String>,
527    pub(crate) interrupted: bool,
528    pub(crate) dispatching_interrupt: bool,
529    pub(crate) interrupt_handler_deadline: Option<Instant>,
530    /// Captured stack trace from the most recent error (fn_name, line, col).
531    pub(crate) error_stack_trace: Vec<(String, usize, usize, Option<String>)>,
532    /// Yield channel sender for generator execution. When set, `Op::Yield`
533    /// sends values through this channel instead of being a no-op.
534    pub(crate) yield_sender: Option<tokio::sync::mpsc::Sender<Result<VmValue, VmError>>>,
535    /// Project root directory (detected via harn.toml).
536    /// Used as base directory for metadata, store, and checkpoint operations.
537    pub(crate) project_root: Option<std::path::PathBuf>,
538    /// Global constants (e.g. `pi`, `e`). Checked as a fallback in `GetVar`
539    /// after the environment, so user-defined variables can shadow them.
540    pub(crate) globals: Arc<crate::value::DictMap>,
541    /// Root authority injected by the embedder for the compiler-generated
542    /// `main(harness: Harness)` call. This is deliberately not an environment
543    /// binding or global: user code can receive it only through an explicit
544    /// parameter.
545    pub(crate) root_harness: Option<VmValue>,
546    /// Runtime effect receipt state (shared recorder + per-VM call memo).
547    pub(crate) runtime_effects: crate::orchestration::RuntimeEffectState,
548    /// Optional debugger hook invoked when execution advances to a new source line.
549    pub(crate) debug_hook: Option<parking_lot::Mutex<Box<DebugHook>>>,
550    /// Effective runtime ceilings for this VM execution.
551    pub(crate) runtime_limits: RuntimeLimits,
552}
553
554/// Reusable VM baseline for hosts that need many clean executions with the
555/// same stable builtin/source setup.
556///
557/// The baseline intentionally does not snapshot execution state. Each
558/// instantiation gets fresh stacks, frames, tasks, cancellation fields, sync
559/// primitives, shared cells/maps/mailboxes, and debug state. Builtin tables are
560/// shared through `Arc` until a per-execution rebind needs copy-on-write.
561#[derive(Clone)]
562pub struct VmBaseline {
563    builtins: Arc<BTreeMap<String, VmBuiltinFn>>,
564    async_builtins: Arc<BTreeMap<String, VmAsyncBuiltinFn>>,
565    capability_methods:
566        Arc<BTreeMap<harn_builtin_meta::CapabilityId, BTreeMap<String, VmBuiltinDispatch>>>,
567    builtin_metadata: Arc<BTreeMap<String, VmBuiltinMetadata>>,
568    builtins_by_id: Arc<HashMap<BuiltinId, VmBuiltinEntry>>,
569    builtin_id_collisions: Arc<HashSet<BuiltinId>>,
570    source_dir: Option<std::path::PathBuf>,
571    source_file: Option<String>,
572    source_text: Option<String>,
573    project_root: Option<std::path::PathBuf>,
574    globals: Arc<crate::value::DictMap>,
575    root_harness: Option<VmValue>,
576    denied_builtins: Arc<HashSet<String>>,
577    prepared_module_cache: crate::PreparedModuleCache,
578    module_provenance: crate::module_artifact::ModuleProvenance,
579    /// Carried so a VM rebuilt from this baseline keeps resolving modules
580    /// through the same link table the original spawn was handed.
581    graph_link_table: Option<Arc<crate::context_manifest::GraphLinkTable>>,
582    linked_program_repository: Option<Arc<crate::linked_program::LinkedProgramRepository>>,
583    runtime_limits: RuntimeLimits,
584}
585
586impl VmBaseline {
587    pub fn from_vm(vm: &Vm) -> Self {
588        Self {
589            builtins: Arc::clone(&vm.builtins),
590            async_builtins: Arc::clone(&vm.async_builtins),
591            capability_methods: Arc::clone(&vm.capability_methods),
592            builtin_metadata: Arc::clone(&vm.builtin_metadata),
593            builtins_by_id: Arc::clone(&vm.builtins_by_id),
594            builtin_id_collisions: Arc::clone(&vm.builtin_id_collisions),
595            source_dir: vm.source_dir.clone(),
596            source_file: vm.source_file.clone(),
597            source_text: vm.source_text.clone(),
598            project_root: vm.project_root.clone(),
599            globals: Arc::clone(&vm.globals),
600            root_harness: vm.root_harness.clone(),
601            denied_builtins: Arc::clone(&vm.denied_builtins),
602            prepared_module_cache: vm.prepared_module_cache.clone(),
603            module_provenance: vm.module_provenance,
604            graph_link_table: vm.graph_link_table.clone(),
605            linked_program_repository: vm.linked_program_repository.clone(),
606            runtime_limits: vm.runtime_limits,
607        }
608    }
609
610    pub fn instantiate(&self) -> Vm {
611        crate::initialize_runtime_assets();
612        let mut source_cache = BTreeMap::new();
613        if let (Some(file), Some(text)) = (&self.source_file, &self.source_text) {
614            source_cache.insert(std::path::PathBuf::from(file), Arc::from(text.as_str()));
615        }
616        if let Some(dir) = &self.source_dir {
617            crate::stdlib::set_thread_source_dir(dir);
618        }
619
620        let mut vm = Vm {
621            stack: Vec::with_capacity(256),
622            env: VmEnv::new(),
623            output: String::new(),
624            builtins: Arc::clone(&self.builtins),
625            async_builtins: Arc::clone(&self.async_builtins),
626            capability_methods: Arc::clone(&self.capability_methods),
627            builtin_metadata: Arc::clone(&self.builtin_metadata),
628            builtins_by_id: Arc::clone(&self.builtins_by_id),
629            builtin_id_collisions: Arc::clone(&self.builtin_id_collisions),
630            iterators: Vec::new(),
631            frames: Vec::new(),
632            exception_handlers: Vec::new(),
633            spawned_tasks: BTreeMap::new(),
634            process_exit_request: Arc::new(ProcessExitRequest::new()),
635            sync_runtime: Arc::new(crate::synchronization::VmSyncRuntime::new()),
636            shared_state_runtime: Arc::new(crate::shared_state::VmSharedStateRuntime::new()),
637            inline_cache_sets: Vec::new(),
638            inline_cache_set_by_chunk: HashMap::new(),
639            pool_registry: crate::stdlib::pool::new_pool_registry(),
640            llm_mock_context: crate::llm::mock::LlmMockContext::for_new_vm(),
641            package_snapshot_registry: Arc::new(Default::default()),
642            wait_for_graph: Arc::new(crate::wait_for_graph::VmWaitForGraph::new()),
643            held_sync_guards: Vec::new(),
644            inherited_held_keys: Arc::new(Vec::new()),
645            task_scopes: Vec::new(),
646            task_counter: 0,
647            runtime_context_counter: 0,
648            runtime_context: crate::runtime_context::RuntimeContext::root(),
649            deadlines: Vec::new(),
650            execution_deadline: super::execution::new_execution_deadline_state(None),
651            breakpoints: BTreeMap::new(),
652            function_breakpoints: std::collections::BTreeSet::new(),
653            pending_function_bp: None,
654            step_mode: false,
655            step_frame_depth: 0,
656            stopped: false,
657            last_line: 0,
658            source_dir: self.source_dir.clone(),
659            package_execution_guard: None,
660            imported_paths: Vec::new(),
661            deferred_cyclic_imports: Vec::new(),
662            module_cache: Arc::new(BTreeMap::new()),
663            prepared_module_cache: self.prepared_module_cache.clone(),
664            module_provenance: self.module_provenance,
665            module_phase_recorder: None,
666            lazy_callable_modules: Arc::new(crate::value::VmMutex::new(BTreeMap::new())),
667            source_cache: Arc::new(source_cache),
668            graph_link_table: self.graph_link_table.clone(),
669            linked_program_repository: self.linked_program_repository.clone(),
670            source_file: self.source_file.clone(),
671            source_text: self.source_text.clone(),
672            coverage: crate::coverage::for_primary(self.source_file.as_deref()),
673            bridge: None,
674            denied_builtins: Arc::clone(&self.denied_builtins),
675            cancel_token: None,
676            interrupt_signal_token: None,
677            cancel_grace_instructions_remaining: None,
678            interrupt_handlers: Vec::new(),
679            next_interrupt_handle: 1,
680            pending_interrupt_signal: None,
681            interrupted: false,
682            dispatching_interrupt: false,
683            interrupt_handler_deadline: None,
684            error_stack_trace: Vec::new(),
685            yield_sender: None,
686            project_root: self.project_root.clone(),
687            globals: Arc::clone(&self.globals),
688            root_harness: self.root_harness.clone(),
689            runtime_effects: crate::orchestration::RuntimeEffectState::fresh(),
690            debug_hook: None,
691            runtime_limits: self.runtime_limits,
692        };
693
694        crate::stdlib::rebind_execution_state_builtins(&mut vm);
695        vm
696    }
697}
698
699impl Vm {
700    pub(crate) fn ensure_execution_available(&self) -> Result<(), VmError> {
701        if self.execution_deadline.is_abandoned() {
702            return Err(VmError::AbandonedExecution);
703        }
704        Ok(())
705    }
706
707    pub(crate) fn fresh_local_slots(chunk: &Chunk) -> Vec<LocalSlot> {
708        chunk
709            .local_slots
710            .iter()
711            .map(|_| LocalSlot {
712                value: VmValue::Nil,
713                initialized: false,
714                synced: false,
715            })
716            .collect()
717    }
718
719    pub(crate) fn bind_param_slots(
720        slots: &mut [LocalSlot],
721        func: &crate::chunk::CompiledFunction,
722        args: &[VmValue],
723        synced: bool,
724    ) {
725        Self::bind_param_slots_args(slots, func, &super::CallArgs::Slice(args), synced);
726    }
727
728    pub(crate) fn bind_param_slots_args(
729        slots: &mut [LocalSlot],
730        func: &crate::chunk::CompiledFunction,
731        args: &super::CallArgs<'_>,
732        synced: bool,
733    ) {
734        let param_count = func.params.len();
735        for (i, _param) in func.params.iter().enumerate() {
736            if i >= slots.len() {
737                break;
738            }
739            if func.has_rest_param && i == param_count - 1 {
740                let rest_args = args.to_vec_from(i);
741                slots[i].value = VmValue::List(std::sync::Arc::new(rest_args));
742                slots[i].initialized = true;
743                slots[i].synced = synced;
744            } else if let Some(arg) = args.get(i) {
745                slots[i].value = arg.clone();
746                slots[i].initialized = true;
747                slots[i].synced = synced;
748            }
749        }
750    }
751
752    pub(crate) fn visible_variables(&self) -> crate::value::DictMap {
753        let mut vars = self.env.all_variables();
754        let Some(frame) = self.frames.last() else {
755            return vars;
756        };
757        for (slot, info) in frame.local_slots.iter().zip(frame.chunk.local_slots.iter()) {
758            if slot.initialized && info.scope_depth <= frame.local_scope_depth {
759                vars.insert(crate::value::intern_key(&info.name), slot.value.clone());
760            }
761        }
762        vars
763    }
764
765    pub(crate) fn sync_current_frame_locals_to_env(&mut self) {
766        let frames = &mut self.frames;
767        let env = &mut self.env;
768        let Some(frame) = frames.last_mut() else {
769            return;
770        };
771        let local_scope_base = frame.local_scope_base;
772        let local_scope_depth = frame.local_scope_depth;
773        for (slot, info) in frame
774            .local_slots
775            .iter_mut()
776            .zip(frame.chunk.local_slots.iter())
777        {
778            if slot.initialized && !slot.synced && info.scope_depth <= local_scope_depth {
779                slot.synced = true;
780                let scope_idx = local_scope_base + info.scope_depth;
781                while env.scopes.len() <= scope_idx {
782                    env.push_scope();
783                }
784                // Slot-backed locals are never boxed captures (those resolve
785                // to `None` in the compiler and are defined straight into the
786                // env as `Cell`s), so syncing a slot always yields a `Value`.
787                Arc::make_mut(&mut env.scopes[scope_idx].vars).insert(
788                    info.name.clone(),
789                    crate::value::Binding::Value {
790                        value: slot.value.clone(),
791                        mutable: info.mutable,
792                    },
793                );
794            }
795        }
796    }
797
798    pub(crate) fn closure_call_env_for_current_frame(
799        &self,
800        closure: &crate::value::VmClosure,
801    ) -> VmEnv {
802        if closure.module_state().is_some() {
803            return closure.env.cloned_for_call();
804        }
805        let call_env = Self::closure_call_env(&self.env, closure);
806        // Same compile-time short-circuit as the env walk in
807        // `closure_call_env`: when the callee body never resolves an
808        // outer name through the env, injecting closure-typed *slot*
809        // locals from the caller's frame is wasted work too.
810        if !closure.func.chunk.references_outer_names {
811            return call_env;
812        }
813        let mut call_env = call_env;
814        let Some(frame) = self.frames.last() else {
815            return call_env;
816        };
817        for (slot, info) in frame
818            .local_slots
819            .iter()
820            .zip(frame.chunk.local_slots.iter())
821            .filter(|(slot, info)| slot.initialized && info.scope_depth <= frame.local_scope_depth)
822        {
823            if matches!(slot.value, VmValue::Closure(_)) && !call_env.contains(&info.name) {
824                let _ = call_env.define(&info.name, slot.value.clone(), info.mutable);
825            }
826        }
827        call_env
828    }
829
830    pub(crate) fn active_local_slot_value(&self, name: &str) -> Option<VmValue> {
831        let frame = self.frames.last()?;
832        let idx = self.active_local_slot_index(name)?;
833        frame.local_slots.get(idx).map(|slot| slot.value.clone())
834    }
835
836    /// Returns the slot index of an initialized active local with the given
837    /// name, walking from innermost to outermost scope. Used by legacy by-name
838    /// hot paths that still want to mutate the slot value in place without
839    /// paying a defensive `VmValue::clone` first.
840    pub(crate) fn active_local_slot_index(&self, name: &str) -> Option<usize> {
841        let frame = self.frames.last()?;
842        for (idx, info) in frame.chunk.local_slots.iter().enumerate().rev() {
843            if info.name == name && info.scope_depth <= frame.local_scope_depth {
844                if let Some(slot) = frame.local_slots.get(idx) {
845                    if slot.initialized {
846                        return Some(idx);
847                    }
848                }
849            }
850        }
851        None
852    }
853
854    pub(crate) fn assign_active_local_slot(
855        &mut self,
856        name: &str,
857        value: VmValue,
858        debug: bool,
859    ) -> Result<bool, VmError> {
860        let Some(frame) = self.frames.last_mut() else {
861            return Ok(false);
862        };
863        for (idx, info) in frame.chunk.local_slots.iter().enumerate().rev() {
864            if info.name == name && info.scope_depth <= frame.local_scope_depth {
865                if !debug && !info.mutable {
866                    return Err(VmError::ImmutableAssignment(name.to_string()));
867                }
868                if let Some(slot) = frame.local_slots.get_mut(idx) {
869                    crate::value::recursion::dismantle(std::mem::replace(&mut slot.value, value));
870                    slot.initialized = true;
871                    slot.synced = false;
872                    return Ok(true);
873                }
874            }
875        }
876        Ok(false)
877    }
878
879    pub fn new() -> Self {
880        crate::initialize_runtime_assets();
881        Self {
882            stack: Vec::with_capacity(256),
883            env: VmEnv::new(),
884            output: String::new(),
885            builtins: Arc::new(BTreeMap::new()),
886            async_builtins: Arc::new(BTreeMap::new()),
887            capability_methods: Arc::new(BTreeMap::new()),
888            builtin_metadata: Arc::new(BTreeMap::new()),
889            builtins_by_id: Arc::new(HashMap::new()),
890            builtin_id_collisions: Arc::new(HashSet::new()),
891            iterators: Vec::new(),
892            frames: Vec::new(),
893            exception_handlers: Vec::new(),
894            spawned_tasks: BTreeMap::new(),
895            process_exit_request: Arc::new(ProcessExitRequest::new()),
896            sync_runtime: Arc::new(crate::synchronization::VmSyncRuntime::new()),
897            shared_state_runtime: Arc::new(crate::shared_state::VmSharedStateRuntime::new()),
898            inline_cache_sets: Vec::new(),
899            inline_cache_set_by_chunk: HashMap::new(),
900            pool_registry: crate::stdlib::pool::new_pool_registry(),
901            llm_mock_context: crate::llm::mock::LlmMockContext::for_new_vm(),
902            package_snapshot_registry: Arc::new(Default::default()),
903            wait_for_graph: Arc::new(crate::wait_for_graph::VmWaitForGraph::new()),
904            held_sync_guards: Vec::new(),
905            inherited_held_keys: Arc::new(Vec::new()),
906            task_scopes: Vec::new(),
907            task_counter: 0,
908            runtime_context_counter: 0,
909            runtime_context: crate::runtime_context::RuntimeContext::root(),
910            deadlines: Vec::new(),
911            execution_deadline: super::execution::new_execution_deadline_state(None),
912            breakpoints: BTreeMap::new(),
913            function_breakpoints: std::collections::BTreeSet::new(),
914            pending_function_bp: None,
915            step_mode: false,
916            step_frame_depth: 0,
917            stopped: false,
918            last_line: 0,
919            source_dir: None,
920            package_execution_guard: None,
921            imported_paths: Vec::new(),
922            deferred_cyclic_imports: Vec::new(),
923            module_cache: Arc::new(BTreeMap::new()),
924            prepared_module_cache: crate::PreparedModuleCache::default(),
925            module_provenance: crate::module_artifact::ModuleProvenance::User,
926            module_phase_recorder: None,
927            lazy_callable_modules: Arc::new(crate::value::VmMutex::new(BTreeMap::new())),
928            source_cache: Arc::new(BTreeMap::new()),
929            graph_link_table: None,
930            linked_program_repository: None,
931            source_file: None,
932            source_text: None,
933            coverage: crate::coverage::for_primary(None),
934            bridge: None,
935            denied_builtins: Arc::new(HashSet::new()),
936            cancel_token: None,
937            interrupt_signal_token: None,
938            cancel_grace_instructions_remaining: None,
939            interrupt_handlers: Vec::new(),
940            next_interrupt_handle: 1,
941            pending_interrupt_signal: None,
942            interrupted: false,
943            dispatching_interrupt: false,
944            interrupt_handler_deadline: None,
945            error_stack_trace: Vec::new(),
946            yield_sender: None,
947            project_root: None,
948            globals: Arc::new(crate::value::DictMap::new()),
949            root_harness: None,
950            runtime_effects: crate::orchestration::RuntimeEffectState::fresh(),
951            debug_hook: None,
952            runtime_limits: RuntimeLimits::default(),
953        }
954    }
955
956    pub fn baseline(&self) -> VmBaseline {
957        VmBaseline::from_vm(self)
958    }
959
960    /// Typed, runtime-resolved effects attempted by this VM execution tree.
961    pub fn executed_effects(&self) -> Vec<crate::orchestration::EffectRecord> {
962        self.runtime_effects.snapshot()
963    }
964
965    /// Clear the execution-local effect recorder and per-VM memo.
966    pub fn clear_executed_effects(&mut self) {
967        self.runtime_effects.clear();
968    }
969
970    pub(crate) fn record_capability_effects(
971        &mut self,
972        capability: harn_builtin_meta::CapabilityId,
973        method: &str,
974        args: &[VmValue],
975    ) {
976        self.runtime_effects
977            .record_capability(capability, method, args);
978    }
979
980    pub(crate) fn record_builtin_contract_effects(&mut self, name: &str, args: &[VmValue]) {
981        let Some(entry) = crate::stdlib::recorded_effect_builtin_manifest_entry(name) else {
982            return;
983        };
984        self.record_builtin_effect_specs(entry.contract.effects, args);
985    }
986
987    pub(crate) fn record_builtin_effect_specs(
988        &mut self,
989        specs: &'static [harn_builtin_meta::EffectSpec],
990        args: &[VmValue],
991    ) {
992        self.runtime_effects.record_specs(specs, args);
993    }
994
995    /// Replace the scoped immutable module-template cache used by this VM.
996    /// Fresh runtime module state is still instantiated for every VM.
997    pub fn set_prepared_module_cache(&mut self, cache: crate::PreparedModuleCache) {
998        self.prepared_module_cache = cache;
999    }
1000
1001    /// Hand this VM the link table for the import graph it is about to run.
1002    ///
1003    /// Only an entry-chunk cache hit can supply one, because only that path holds
1004    /// a manifest proving the recorded digests still describe the files on disk.
1005    /// A table never changes what a module resolves to: one the table does not
1006    /// name, or whose artifact is no longer on disk, is read and compiled as
1007    /// usual.
1008    pub fn set_graph_link_table(
1009        &mut self,
1010        link_table: Option<Arc<crate::context_manifest::GraphLinkTable>>,
1011    ) {
1012        self.graph_link_table = link_table;
1013    }
1014
1015    /// Install the verified module-template repository carried by one linked
1016    /// program. Templates remain scoped to this VM tree and never enter the
1017    /// ordinary caller-independent cache.
1018    pub fn set_linked_program_runtime(
1019        &mut self,
1020        runtime: &crate::linked_program::LinkedProgramRuntime,
1021    ) {
1022        self.linked_program_repository = Some(Arc::clone(&runtime.repository));
1023        self.graph_link_table = None;
1024    }
1025
1026    /// Return the effective runtime limit profile for this VM.
1027    pub fn runtime_limits(&self) -> RuntimeLimits {
1028        self.runtime_limits
1029    }
1030
1031    /// Return a host/debug report describing the VM's effective runtime limits.
1032    pub fn runtime_limit_report(&self) -> crate::RuntimeLimitsReport {
1033        self.runtime_limits.report()
1034    }
1035
1036    /// Returns true if any debugging affordance is active — DAP hook,
1037    /// line breakpoints, or function breakpoints. Call-site code uses
1038    /// this to decide whether to capture per-frame restart snapshots
1039    /// (`initial_env`, `initial_local_slots`); without a debugger those
1040    /// snapshots are dead weight, so skipping them removes two
1041    /// allocations from every function call hot path.
1042    ///
1043    /// All three signals are stable across a function call's lifetime
1044    /// (they're set before pipeline execution starts), so the gate is
1045    /// consistent between frame creation and any later `restart_frame`
1046    /// invocation. The three `is_empty` checks compile to a handful of
1047    /// branch-predicted memory probes — cheaper than a single
1048    /// `BTreeMap` clone, which is what we're avoiding.
1049    #[inline]
1050    pub(crate) fn debugger_attached(&self) -> bool {
1051        self.debug_hook.is_some()
1052            || !self.breakpoints.is_empty()
1053            || !self.function_breakpoints.is_empty()
1054    }
1055
1056    /// Set the bridge for delegating unknown builtins in bridge mode.
1057    pub fn set_bridge(&mut self, bridge: Arc<crate::bridge::HostBridge>) {
1058        self.bridge = Some(bridge);
1059    }
1060
1061    /// Set builtins that are denied in sandbox mode.
1062    /// When called, the given builtin names will produce a permission error.
1063    pub fn set_denied_builtins(&mut self, mut denied: HashSet<String>) {
1064        // A builtin may have a private registry name plus one or more public
1065        // aliases. Normalize the whole family once so no dispatch path can
1066        // escape policy merely by using another name for the same handler.
1067        let denied_canonical_names = denied
1068            .iter()
1069            .filter_map(|name| crate::stdlib::builtin_manifest_entry(name))
1070            .map(|entry| entry.canonical_name)
1071            .collect::<HashSet<_>>();
1072        if !denied_canonical_names.is_empty() {
1073            denied.extend(
1074                crate::stdlib::all_builtin_manifest()
1075                    .iter()
1076                    .filter(|entry| denied_canonical_names.contains(entry.canonical_name))
1077                    .map(|entry| entry.name.to_string()),
1078            );
1079        }
1080        self.denied_builtins = Arc::new(denied);
1081    }
1082
1083    /// Set source info for error reporting (file path and source text).
1084    pub fn set_source_info(&mut self, file: &str, text: &str) {
1085        self.source_file = Some(file.to_string());
1086        self.source_text = Some(text.to_string());
1087        if let Some(cov) = self.coverage.as_mut() {
1088            cov.set_primary_file(file);
1089        }
1090        Arc::make_mut(&mut self.source_cache)
1091            .insert(std::path::PathBuf::from(file), Arc::from(text));
1092    }
1093
1094    /// Initialize execution (push the initial frame).
1095    pub fn start(&mut self, chunk: &Chunk) -> Result<(), VmError> {
1096        self.ensure_execution_available()?;
1097        // The top-level pipeline frame captures env at start so
1098        // restartFrame on the outermost frame rewinds to the
1099        // pre-pipeline state — basically "restart session" in
1100        // debugger terms. Skipped when no debugger is attached:
1101        // the snapshot is dead weight in that case and dominates
1102        // call-overhead bench numbers (~5-10%).
1103        let debugger = self.debugger_attached();
1104        let initial_env = if debugger {
1105            Some(self.env.clone())
1106        } else {
1107            None
1108        };
1109        let initial_local_slots = if debugger {
1110            Some(Self::fresh_local_slots(chunk))
1111        } else {
1112            None
1113        };
1114        let chunk = Arc::new(chunk.clone());
1115        let local_slots = Self::fresh_local_slots(&chunk);
1116        let inline_cache_set = self.inline_cache_set_index_for_chunk(&chunk);
1117        self.frames.push(CallFrame {
1118            chunk,
1119            inline_cache_set,
1120            ip: 0,
1121            stack_base: self.stack.len(),
1122            saved_env: self.env.clone(),
1123            initial_env,
1124            initial_local_slots,
1125            saved_iterator_depth: self.iterators.len(),
1126            fn_name: String::new(),
1127            argc: 0,
1128            saved_source_dir: None,
1129            module_functions: None,
1130            module_state: None,
1131            local_slots,
1132            local_scope_base: self.env.scope_depth().saturating_sub(1),
1133            local_scope_depth: 0,
1134        });
1135        Ok(())
1136    }
1137
1138    /// Create a child VM that shares builtins and env but has fresh execution state.
1139    /// Used for parallel/spawn to fork the VM for concurrent tasks.
1140    pub(crate) fn child_vm(&self) -> Vm {
1141        Vm {
1142            stack: Vec::with_capacity(64),
1143            env: self.env.clone(),
1144            output: String::new(),
1145            builtins: Arc::clone(&self.builtins),
1146            async_builtins: Arc::clone(&self.async_builtins),
1147            capability_methods: Arc::clone(&self.capability_methods),
1148            builtin_metadata: Arc::clone(&self.builtin_metadata),
1149            builtins_by_id: Arc::clone(&self.builtins_by_id),
1150            builtin_id_collisions: Arc::clone(&self.builtin_id_collisions),
1151            iterators: Vec::new(),
1152            frames: Vec::new(),
1153            exception_handlers: Vec::new(),
1154            spawned_tasks: BTreeMap::new(),
1155            process_exit_request: Arc::clone(&self.process_exit_request),
1156            sync_runtime: self.sync_runtime.clone(),
1157            shared_state_runtime: self.shared_state_runtime.clone(),
1158            inline_cache_sets: Vec::new(),
1159            inline_cache_set_by_chunk: HashMap::new(),
1160            pool_registry: self.pool_registry.clone(),
1161            llm_mock_context: self.llm_mock_context.clone(),
1162            package_snapshot_registry: self.package_snapshot_registry.clone(),
1163            wait_for_graph: self.wait_for_graph.clone(),
1164            held_sync_guards: Vec::new(),
1165            inherited_held_keys: Arc::new(Vec::new()),
1166            task_scopes: Vec::new(),
1167            task_counter: 0,
1168            runtime_context_counter: self.runtime_context_counter,
1169            runtime_context: self.runtime_context.clone(),
1170            deadlines: self.deadlines.clone(),
1171            execution_deadline: self.execution_deadline.fork(),
1172            breakpoints: BTreeMap::new(),
1173            function_breakpoints: std::collections::BTreeSet::new(),
1174            pending_function_bp: None,
1175            step_mode: false,
1176            step_frame_depth: 0,
1177            stopped: false,
1178            last_line: 0,
1179            source_dir: self.source_dir.clone(),
1180            package_execution_guard: self.package_execution_guard.clone(),
1181            imported_paths: Vec::new(),
1182            deferred_cyclic_imports: Vec::new(),
1183            module_cache: Arc::clone(&self.module_cache),
1184            prepared_module_cache: self.prepared_module_cache.clone(),
1185            module_provenance: self.module_provenance,
1186            module_phase_recorder: self.module_phase_recorder.clone(),
1187            lazy_callable_modules: Arc::clone(&self.lazy_callable_modules),
1188            source_cache: Arc::clone(&self.source_cache),
1189            graph_link_table: self.graph_link_table.clone(),
1190            linked_program_repository: self.linked_program_repository.clone(),
1191            source_file: self.source_file.clone(),
1192            source_text: self.source_text.clone(),
1193            coverage: crate::coverage::for_primary(self.source_file.as_deref()),
1194            bridge: self.bridge.clone(),
1195            denied_builtins: Arc::clone(&self.denied_builtins),
1196            cancel_token: self.cancel_token.clone(),
1197            interrupt_signal_token: self.interrupt_signal_token.clone(),
1198            cancel_grace_instructions_remaining: None,
1199            interrupt_handlers: Vec::new(),
1200            next_interrupt_handle: 1,
1201            pending_interrupt_signal: None,
1202            interrupted: self.interrupted,
1203            dispatching_interrupt: false,
1204            interrupt_handler_deadline: None,
1205            error_stack_trace: Vec::new(),
1206            yield_sender: None,
1207            project_root: self.project_root.clone(),
1208            globals: Arc::clone(&self.globals),
1209            root_harness: self.root_harness.clone(),
1210            runtime_effects: crate::orchestration::RuntimeEffectState::with_shared_recorder(
1211                Arc::clone(&self.runtime_effects.recorder),
1212            ),
1213            debug_hook: None,
1214            runtime_limits: self.runtime_limits,
1215        }
1216    }
1217
1218    /// Create a child VM for external adapters that need to invoke Harn
1219    /// closures while sharing the parent's builtins, globals, and module state.
1220    pub(crate) fn child_vm_for_host(&self) -> Vm {
1221        self.child_vm()
1222    }
1223
1224    /// Return the cooperative cancellation/deadline sources that a blocking
1225    /// host adapter must install on its worker thread. The values are cloned
1226    /// once at the async-boundary; the token remains shared with the owning VM
1227    /// so `parallel` aborts and host cancellation reach an in-flight wait.
1228    pub(crate) fn interrupt_sources(
1229        &self,
1230    ) -> (
1231        Option<std::sync::Arc<std::sync::atomic::AtomicBool>>,
1232        Option<std::time::Instant>,
1233    ) {
1234        let scope_deadline = self.deadlines.last().map(|(deadline, _)| *deadline);
1235        let deadline = match (scope_deadline, self.interrupt_handler_deadline) {
1236            (Some(scope), Some(interrupt)) => Some(scope.min(interrupt)),
1237            (scope, interrupt) => scope.or(interrupt),
1238        };
1239        (self.cancel_token.clone(), deadline)
1240    }
1241
1242    pub(crate) fn request_process_exit(&self, code: i32) {
1243        self.process_exit_request.request(code);
1244    }
1245
1246    pub(crate) fn requested_process_exit(&self) -> Option<i32> {
1247        self.process_exit_request.code()
1248    }
1249
1250    /// Request cancellation for every outstanding child task owned by this VM
1251    /// and then abort the join handles. This prevents un-awaited spawned tasks
1252    /// from outliving their parent execution scope.
1253    pub(crate) fn cancel_spawned_tasks(&mut self) {
1254        for (_, task) in std::mem::take(&mut self.spawned_tasks) {
1255            task.cancel_token
1256                .store(true, std::sync::atomic::Ordering::SeqCst);
1257            task.handle.abort();
1258        }
1259    }
1260
1261    /// Set the source directory for import resolution and introspection.
1262    /// Also auto-detects the project root if not already set.
1263    pub fn set_source_dir(&mut self, dir: &std::path::Path) {
1264        let dir = crate::stdlib::process::normalize_context_path(dir);
1265        self.source_dir = Some(dir.clone());
1266        crate::stdlib::set_thread_source_dir(&dir);
1267        // Auto-detect project root if not explicitly set.
1268        if self.project_root.is_none() {
1269            self.project_root = crate::stdlib::process::find_project_root(&dir);
1270        }
1271    }
1272
1273    /// Explicitly set the project root directory.
1274    /// Used by ACP/CLI to override auto-detection.
1275    pub fn set_project_root(&mut self, root: &std::path::Path) {
1276        self.project_root = Some(root.to_path_buf());
1277    }
1278
1279    /// Get only the explicit or auto-detected project root, without falling
1280    /// back to `source_dir`.
1281    pub(crate) fn explicit_project_root(&self) -> Option<&std::path::Path> {
1282        self.project_root.as_deref()
1283    }
1284
1285    /// Get the project root directory, falling back to source_dir.
1286    pub fn project_root(&self) -> Option<&std::path::Path> {
1287        self.project_root.as_deref().or(self.source_dir.as_deref())
1288    }
1289
1290    /// Set a global constant (e.g. `pi`, `e`).
1291    /// Stored separately from the environment so user-defined variables can shadow them.
1292    pub fn set_global(&mut self, name: &str, value: VmValue) {
1293        Arc::make_mut(&mut self.globals).insert(crate::value::intern_key(name), value);
1294    }
1295
1296    /// Read a previously-installed ordinary global constant.
1297    pub fn global(&self, name: &str) -> Option<&VmValue> {
1298        self.globals.get(name)
1299    }
1300
1301    /// Install the script's root authority for the compiler-generated
1302    /// `main(harness: Harness)` call. It is kept out of all source-visible
1303    /// namespaces.
1304    pub fn set_harness(&mut self, harness: crate::harness::Harness) {
1305        self.root_harness = Some(harness.into_vm_value());
1306    }
1307
1308    pub(crate) fn harness(&self) -> Option<&crate::harness::VmHarness> {
1309        match self.root_harness.as_ref() {
1310            Some(VmValue::Harness(handle)) => Some(handle),
1311            _ => None,
1312        }
1313    }
1314
1315    /// Clone the embedder-owned root authority for an explicit exported
1316    /// function parameter. This does not install a source-visible binding.
1317    pub fn root_harness_value(&self) -> Option<VmValue> {
1318        self.root_harness.clone()
1319    }
1320
1321    /// Get the captured output.
1322    pub fn output(&self) -> &str {
1323        &self.output
1324    }
1325
1326    /// Drain and return the captured output, leaving the buffer empty.
1327    /// Used by the async-builtin dispatch path to forward closure output
1328    /// from a child VM back to its parent.
1329    pub fn take_output(&mut self) -> String {
1330        std::mem::take(&mut self.output)
1331    }
1332
1333    /// Append text to this VM's captured output. Used to forward output
1334    /// from child VMs (e.g. closures invoked via `call_closure_pub`)
1335    /// back into the parent stream.
1336    pub fn append_output(&mut self, text: &str) {
1337        self.output.push_str(text);
1338    }
1339
1340    pub(crate) fn pop(&mut self) -> Result<VmValue, VmError> {
1341        self.stack.pop().ok_or(VmError::StackUnderflow)
1342    }
1343
1344    pub(crate) fn peek(&self) -> Result<&VmValue, VmError> {
1345        self.stack.last().ok_or(VmError::StackUnderflow)
1346    }
1347
1348    pub(crate) fn const_str(c: &Constant) -> Result<&str, VmError> {
1349        match c {
1350            Constant::String(s) => Ok(s.as_str()),
1351            _ => Err(VmError::TypeError("expected string constant".into())),
1352        }
1353    }
1354
1355    pub(crate) fn release_sync_guards_for_current_scope(&mut self) {
1356        let depth = self.env.scope_depth();
1357        self.held_sync_guards
1358            .retain(|guard| guard.env_scope_depth < depth);
1359        // A `scope { }` torn down without a normal `TaskScopeExit` (break /
1360        // continue out of it) leaves a dangling nursery — cancel its tasks.
1361        self.cancel_task_scopes_where(|s| s.env_scope_depth >= depth);
1362    }
1363
1364    pub(crate) fn release_sync_guards_after_unwind(
1365        &mut self,
1366        frame_depth: usize,
1367        env_scope_depth: usize,
1368    ) {
1369        self.held_sync_guards.retain(|guard| {
1370            guard.frame_depth <= frame_depth && guard.env_scope_depth <= env_scope_depth
1371        });
1372        // Cancel nurseries opened above the catch handler (a `throw` unwound
1373        // past their `TaskScopeExit`).
1374        self.cancel_task_scopes_where(|s| {
1375            !(s.frame_depth <= frame_depth && s.env_scope_depth <= env_scope_depth)
1376        });
1377    }
1378
1379    pub(crate) fn release_sync_guards_for_frame(&mut self, frame_depth: usize) {
1380        self.held_sync_guards
1381            .retain(|guard| guard.frame_depth != frame_depth);
1382        // Cancel any nursery whose `scope {}` block belonged to the frame being
1383        // torn down (e.g. a `return` jumped past its `TaskScopeExit`).
1384        self.cancel_task_scopes_where(|s| s.frame_depth == frame_depth);
1385    }
1386
1387    pub(crate) fn adopt_sync_permit_for_current_scope(
1388        &mut self,
1389        permit: crate::value::VmSyncPermitHandle,
1390    ) {
1391        if permit.is_released()
1392            || self
1393                .held_sync_guards
1394                .iter()
1395                .any(|guard| guard._permit.same_lease(&permit))
1396        {
1397            return;
1398        }
1399        self.held_sync_guards
1400            .push(crate::synchronization::VmSyncHeldGuard {
1401                _permit: permit,
1402                frame_depth: self.frames.len(),
1403                env_scope_depth: self.env.scope_depth(),
1404            });
1405    }
1406
1407    /// Deregister a task id from every open nursery (it was explicitly
1408    /// `await`ed, so it must not be double-joined or cancelled at scope exit).
1409    pub(crate) fn deregister_task_from_scopes(&mut self, id: &str) {
1410        for scope in &mut self.task_scopes {
1411            scope.task_ids.retain(|t| t != id);
1412        }
1413    }
1414
1415    /// Cancel and remove every task scope matching `doomed`, aborting its bound
1416    /// tasks (used when a `scope {}` is torn down without a normal join).
1417    fn cancel_task_scopes_where<F: Fn(&TaskScope) -> bool>(&mut self, doomed: F) {
1418        let mut i = 0;
1419        while i < self.task_scopes.len() {
1420            if doomed(&self.task_scopes[i]) {
1421                let scope = self.task_scopes.remove(i);
1422                for id in &scope.task_ids {
1423                    if let Some(task) = self.spawned_tasks.remove(id) {
1424                        task.cancel_token
1425                            .store(true, std::sync::atomic::Ordering::SeqCst);
1426                        task.handle.abort();
1427                    }
1428                }
1429            } else {
1430                i += 1;
1431            }
1432        }
1433    }
1434
1435    /// Total live permits this VM already holds for `kind:key`. The held-set is
1436    /// tiny (bounded by lexical nesting and explicit sync acquisitions), so this
1437    /// scan is cheap and only runs on the rare blocking-acquire path.
1438    pub(crate) fn held_permits_for(&self, kind: &str, key: &str) -> u32 {
1439        let own: u32 = self
1440            .held_sync_guards
1441            .iter()
1442            .filter(|guard| {
1443                !guard._permit.is_released()
1444                    && guard._permit.kind() == kind
1445                    && guard._permit.key() == key
1446            })
1447            .map(|guard| guard._permit.permits())
1448            .sum();
1449        let inherited: u32 = self
1450            .inherited_held_keys
1451            .iter()
1452            .filter(|held| held.kind == kind && held.key == key)
1453            .map(|held| held.permits)
1454            .sum();
1455        own + inherited
1456    }
1457
1458    /// Every live sync permit held by this VM *and* its suspended ancestors: the
1459    /// transitive held-set seen by an inline child.
1460    pub(crate) fn combined_held_keys(&self) -> Vec<crate::synchronization::VmSyncHeldKey> {
1461        let mut keys: Vec<crate::synchronization::VmSyncHeldKey> = self
1462            .held_sync_guards
1463            .iter()
1464            .filter_map(|guard| crate::synchronization::VmSyncHeldKey::from_permit(&guard._permit))
1465            .collect();
1466        keys.extend(self.inherited_held_keys.iter().cloned());
1467        keys
1468    }
1469
1470    /// Clone a child VM for an **inline, same-task** execution (an async builtin
1471    /// awaited while this VM is parked, or a user closure that builtin runs and
1472    /// awaits). The child inherits this VM's transitive held-lock keys so a
1473    /// re-acquire of a parent-held lock is caught as a self-deadlock
1474    /// (HARN-ORC-011). Use plain `child_vm()` for new concurrent tasks.
1475    pub(crate) fn child_vm_inline(&self) -> Vm {
1476        let mut child = self.child_vm();
1477        child.inherited_held_keys = Arc::new(self.combined_held_keys());
1478        child
1479    }
1480}
1481
1482impl Drop for Vm {
1483    fn drop(&mut self) {
1484        if let Some(coverage) = self.coverage.take() {
1485            crate::coverage::merge_into_global(coverage);
1486        }
1487        self.cancel_spawned_tasks();
1488    }
1489}
1490
1491impl Default for Vm {
1492    fn default() -> Self {
1493        Self::new()
1494    }
1495}
1496
1497#[cfg(test)]
1498#[path = "state_tests.rs"]
1499mod tests;