Files
regorus/src/rvm/vm/virtual_data.rs
Anand Krishnamoorthi 50c0215fdb Rvm optimizations (#620)
* perf(rvm): fix O(n²) comprehension yield by mutating in-place

Instead of cloning the entire accumulator collection on every yield
iteration, use take_register + Rc::make_mut to get exclusive ownership
and mutate in-place. This reduces comprehension yield from O(n²) to O(n)
for both run-to-completion and suspendable execution modes.

- Add RegoVM::take_register() helper that swaps register with Undefined
- Comprehension yield now takes the accumulator, mutates via Rc::make_mut,
  and writes back — avoiding deep clones when refcount == 1

Signed-off-by: Anand Krishnamoorthi <anakrish@microsoft.com>

* perf(rvm): use take_register for ObjectSet, ArrayPush, SetAdd

These instructions were cloning the container register (bumping Rc to 2),
then calling as_object_mut/as_array_mut/as_set_mut which invokes
Rc::make_mut — deep-cloning the entire collection since refcount > 1.

Use take_register instead so the Rc refcount stays at 1, making
Rc::make_mut a no-op and allowing in-place mutation.

Signed-off-by: Anand Krishnamoorthi <anakrish@microsoft.com>

* perf(rvm): remove unnecessary clones in rule caching

- execute_call_rule_common: move final_value into cache instead of
  cloning, since it is not used afterwards
- finalize_rule_frame_data: add comment clarifying the clone is needed
  because the value is both cached and returned
- Remove unnecessary .clone() on result_from_rule when setting register

Signed-off-by: Anand Krishnamoorthi <anakrish@microsoft.com>

* rvm: avoid RuleInfo clone per rule call

Replace RuleInfo.clone() (which heap-allocates name, destructuring_blocks, and
potentially function_info) with a cheap Arc<Program> clone (atomic refcount
bump) followed by borrowing &RuleInfo from the local Arc. This eliminates
per-rule-call heap allocations.

Sites changed:
- execute_call_rule_common: Arc clone + borrow
- execute_call_rule_suspendable: Arc clone + borrow
- finalize_rule_frame_data: Arc clone + borrow
- handle_rule_break_event: inline Arc clone + borrow (was get_rule_info)
- handle_rule_error_event: inline Arc clone + borrow (was get_rule_info)
- Removed now-unused get_rule_info method

Signed-off-by: Anand Krishnamoorthi <anakrish@microsoft.com>

* rvm: replace bincode with postcard for serialization

Remove unlinked bincode dependency. Use postcard (already a dep for rvm feature)
for all binary serialization/deserialization in program serialization and tests.

Also adds rvm_benchmark benchmark.

Signed-off-by: Anand Krishnamoorthi <anakrish@microsoft.com>

* perf(rvm): cache dummy Span/Expr for builtin calls

Every builtin call was allocating a Source (via from_contents), a Span, and
N Ref<Expr> wrappers just to satisfy the builtin function signature. These
dummy values are only used for error reporting context.

Cache the dummy Span and Vec<Ref<Expr>> on the RegoVM struct. The Source and
Span are created once on first builtin call; dummy Expr entries grow as
needed and are reused across calls via mem::take/put-back pattern.

This eliminates per-builtin-call heap allocations for Source (Rc + String +
Vec<lines>), Span clones, and Rc<Expr> wrappers.

Signed-off-by: Anand Krishnamoorthi <anakrish@microsoft.com>

* perf(rvm): round 2 allocation reduction in builtins, entry points, virtual data

- Cache builtin args Vec on RegoVM (mem::take/clear/put-back pattern)
- Restructure builtins_cache as two-level map for clone-free lookup
- Use IndexMap::get_index() in execute_entry_point_by_index
- Use mutable Vec path stack in traverse_rule_tree_subobject (push/pop)
- Walk data tree and rule-result paths by reference, clone only leaf
- Use mem::replace in resume() instead of cloning ExecutionState

Signed-off-by: Anand Krishnamoorthi <anakrish@microsoft.com>

* fix(rvm): address PR review feedback

- Restore cached_builtin_args on all error/early-return paths in
  execute_builtin_call to preserve allocation reuse
- Use 1-based line/col and \"<builtin>\" filename in dummy span for
  clearer diagnostics
- Restore result register before returning errors in comprehension
  mode-mismatch branches (both run-to-completion and suspendable)
- Avoid clone in resume() invalid-state error path by formatting
  debug string before moving state back

---------

Signed-off-by: Anand Krishnamoorthi <anakrish@microsoft.com>
2026-03-11 21:39:58 -05:00

313 lines
12 KiB
Rust

// Copyright (c) Microsoft Corporation.
// Licensed under the MIT License.
use crate::rvm::instructions::LiteralOrRegister;
use crate::value::Value;
use alloc::vec::Vec;
use core::convert::TryFrom as _;
use super::errors::{Result, VmError};
use super::machine::RegoVM;
impl RegoVM {
pub(super) fn execute_virtual_data_document_lookup_subobject(
&mut self,
path_components: &[LiteralOrRegister],
rule_tree_subobject: &Value,
) -> Result<Value> {
let mut root_path = Vec::new();
for component in path_components {
let key_value = self.literal_or_register_value(component)?;
root_path.push(key_value);
}
// Walk data tree by reference, only clone the leaf.
let mut data_ref = &self.data;
for path_component in &root_path {
data_ref = &data_ref[path_component];
}
let mut result_subobject = match *data_ref {
Value::Undefined => Value::new_object(),
_ => data_ref.clone(),
};
self.traverse_rule_tree_subobject(rule_tree_subobject, &mut result_subobject, &root_path)?;
Ok(result_subobject)
}
fn set_nested_value(target: &mut Value, path: &[Value], value: Value) -> Result<()> {
Self::set_nested_value_static(target, path, value)
}
fn set_nested_value_static(target: &mut Value, path: &[Value], value: Value) -> Result<()> {
let Some((head, tail)) = path.split_first() else {
*target = value;
return Ok(());
};
if *target == Value::Undefined {
*target = Value::new_object();
}
if let Value::Object(ref mut map) = *target {
if !map.contains_key(head) {
crate::Rc::make_mut(map).insert(head.clone(), Value::Undefined);
}
if let Some(next_target) = crate::Rc::make_mut(map).get_mut(head) {
Self::set_nested_value_static(next_target, tail, value)?;
}
} else {
return Err(VmError::InvalidRuleTreeEntry {
value: target.clone(),
pc: 0,
});
}
Ok(())
}
fn traverse_rule_tree_subobject(
&mut self,
rule_tree_node: &Value,
result_subobject: &mut Value,
root_path: &[Value],
) -> Result<()> {
let mut relative_path = Vec::new();
self.traverse_rule_tree_subobject_with_path(
rule_tree_node,
result_subobject,
root_path,
&mut relative_path,
)
}
fn traverse_rule_tree_subobject_with_path(
&mut self,
rule_tree_node: &Value,
result_subobject: &mut Value,
root_path: &[Value],
relative_path: &mut Vec<Value>,
) -> Result<()> {
match *rule_tree_node {
Value::Number(ref rule_idx) => {
if let Some(rule_index) = rule_idx.as_u64() {
// Walk the evaluated cache using root_path then relative_path,
// without allocating a combined full_cache_path Vec.
let cached_result = {
let mut cache_lookup = &self.evaluated;
let mut path_exists = true;
for path_component in root_path.iter().chain(relative_path.iter()) {
if let Value::Object(ref map) = *cache_lookup {
if let Some(next_value) = map.get(path_component) {
cache_lookup = next_value;
} else {
path_exists = false;
break;
}
} else {
path_exists = false;
break;
}
}
if path_exists {
if let Value::Object(ref map) = *cache_lookup {
map.get(&Value::Undefined).cloned()
} else {
None
}
} else {
None
}
};
let rule_result = if let Some(cached) = cached_result {
self.cache_hits =
self.checked_add_one(self.cache_hits, "cache hits counter")?;
cached
} else {
let temp_reg = u8::try_from(self.registers.len()).map_err(|_| {
VmError::RegisterIndexOutOfBounds {
index: u8::MAX,
pc: self.pc,
register_count: self.registers.len(),
}
})?;
self.registers.push(Value::Undefined);
let rule_index_u16 =
u16::try_from(rule_index).map_err(|_| VmError::InvalidRuleIndex {
rule_index: Value::Number(rule_idx.clone()),
pc: self.pc,
})?;
self.execute_call_rule_common(temp_reg, rule_index_u16, None)?;
let register_count = self.registers.len();
let result =
self.registers
.pop()
.ok_or(VmError::RegisterIndexOutOfBounds {
index: temp_reg,
pc: self.pc,
register_count,
})?;
// Build cache_path from root_path + relative_path + Undefined sentinel.
let mut cache_path: Vec<Value> = Vec::with_capacity(
root_path
.len()
.saturating_add(relative_path.len())
.saturating_add(1),
);
cache_path.extend_from_slice(root_path);
cache_path.extend_from_slice(relative_path);
cache_path.push(Value::Undefined);
Self::set_nested_value_static(
&mut self.evaluated,
&cache_path,
result.clone(),
)?;
result
};
Self::set_nested_value(result_subobject, relative_path, rule_result)?;
} else {
return Err(VmError::InvalidRuleIndex {
rule_index: Value::Number(rule_idx.clone()),
pc: self.pc,
});
}
}
Value::Object(ref obj) => {
for (key, value) in obj.iter() {
relative_path.push(key.clone());
self.traverse_rule_tree_subobject_with_path(
value,
result_subobject,
root_path,
relative_path,
)?;
relative_path.pop();
}
}
_ => {}
}
Ok(())
}
pub(super) fn execute_virtual_data_document_lookup(&mut self, params_index: u16) -> Result<()> {
let params = self
.program
.instruction_data
.get_virtual_data_document_lookup_params(params_index)
.ok_or(VmError::InvalidVirtualDataDocumentLookupParams {
index: params_index,
pc: self.pc,
available: self
.program
.instruction_data
.virtual_data_document_lookup_params
.len(),
})?
.clone();
let mut current_node = &self.program.rule_tree["data"];
let mut components_consumed = 0;
for (i, component) in params.path_components.iter().enumerate() {
let key_value = self.literal_or_register_value(component)?;
current_node = &current_node[&key_value];
components_consumed = self.checked_add_one(i, "path components traversed")?;
match *current_node {
Value::Undefined | Value::Number(_) => break,
_ => {}
}
}
match *current_node {
Value::Number(ref rule_index_value) => {
if let Some(rule_index) = rule_index_value.as_u64() {
let rule_index =
u16::try_from(rule_index).map_err(|_| VmError::InvalidRuleIndex {
rule_index: Value::Number(rule_index_value.clone()),
pc: self.pc,
})?;
self.execute_call_rule_common(params.dest, rule_index, None)?;
if components_consumed < params.path_components.len() {
// Walk remaining path by reference, clone only the leaf.
let mut ref_val = self.get_register(params.dest)?;
for component in params.path_components.iter().skip(components_consumed) {
let key_value = self.literal_or_register_value(component)?;
ref_val = &ref_val[&key_value];
}
let leaf = ref_val.clone();
self.set_register(params.dest, leaf)?;
}
} else {
return Err(VmError::InvalidRuleIndex {
rule_index: Value::Number(rule_index_value.clone()),
pc: self.pc,
});
}
}
Value::Undefined | Value::Object(_)
if components_consumed != params.path_components.len() =>
{
// Walk data tree by reference, clone only the leaf.
let mut data_ref = &self.data;
for component in &params.path_components {
let key_value = self.literal_or_register_value(component)?;
data_ref = &data_ref[&key_value];
}
let result = data_ref.clone();
self.set_register(params.dest, result)?;
}
Value::Object(_) => {
let rule_tree_subobject = current_node.clone();
let result = self.execute_virtual_data_document_lookup_subobject(
&params.path_components,
&rule_tree_subobject,
)?;
self.set_register(params.dest, result)?;
}
_ => {
return Err(VmError::InvalidRuleTreeEntry {
value: current_node.clone(),
pc: self.pc,
});
}
}
Ok(())
}
fn literal_or_register_value(&self, source: &LiteralOrRegister) -> Result<Value> {
let value = match *source {
LiteralOrRegister::Literal(ref idx) => self
.program
.literals
.get(usize::from(*idx))
.ok_or(VmError::LiteralIndexOutOfBounds {
index: *idx,
pc: self.pc,
})?
.clone(),
LiteralOrRegister::Register(ref reg) => self.get_register(*reg)?.clone(),
};
Ok(value)
}
}