(Proof-of-Concept/Draft) Pre-parsing feature #1636
Draft
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Robbepop
reviewed
Aug 14, 2025
Comment on lines
-139
to
+132
| const REVCOMP_INPUT: &[u8] = include_bytes!("rust/cases/reverse_complement/input.txt"); | ||
| const REVCOMP_OUTPUT: &[u8] = include_bytes!("rust/cases/reverse_complement/output.txt"); | ||
| const REVCOMP_INPUT: &[u8] = &[42]; //include_bytes!("rust/cases/reverse_complement/input.txt"); | ||
| const REVCOMP_OUTPUT: &[u8] = &[42]; //include_bytes!("rust/cases/reverse_complement/output.txt"); |
There was a problem hiding this comment.
Why have you effectively disabled this benchmark?
Comment on lines
+33
to
+36
| serde = { version = "1.0", default-features = false, features = [ | ||
| "derive", | ||
| "alloc", | ||
| ], optional = true } | ||
| postcard = { workspace = true, default-features = false, optional = true, features = [ | ||
| "alloc", | ||
| ] } |
There was a problem hiding this comment.
Please do not re-format code that is not associated to your PR diff.
Comment on lines
-14
to
+11
| criterion_group, | ||
| criterion_main, | ||
| measurement::WallTime, | ||
| Bencher, | ||
| BenchmarkGroup, | ||
| Criterion, | ||
| criterion_group, criterion_main, measurement::WallTime, Bencher, BenchmarkGroup, Criterion, |
There was a problem hiding this comment.
Please stick to Wasmi's formatting rules.
Comment on lines
-247
to
-273
| /// Initializes the [`EngineFunc`] for lazy translation. | ||
| /// | ||
| /// # Panics | ||
| /// | ||
| /// - If `func` is an invalid [`EngineFunc`] reference for this [`CodeMap`]. | ||
| /// - If `func` refers to an already initialized [`EngineFunc`]. | ||
| pub fn init_func_as_uncompiled( | ||
| &self, | ||
| func: EngineFunc, | ||
| func_idx: FuncIdx, | ||
| bytes: &[u8], | ||
| module: &ModuleHeader, | ||
| func_to_validate: Option<FuncToValidate<ValidatorResources>>, | ||
| ) { | ||
| let mut funcs = self.funcs.lock(); | ||
| let Some(func) = funcs.get_mut(func) else { | ||
| panic!("encountered invalid internal function: {func:?}") | ||
| }; | ||
| func.init_uncompiled(UncompiledFuncEntity::new( | ||
| func_idx, | ||
| bytes, | ||
| module.clone(), | ||
| func_to_validate, | ||
| )); | ||
| } | ||
|
|
There was a problem hiding this comment.
Please don't unnecessarily move functions around that have no effect on the PRs diff. This just makes it harder to review and blow-up the PR's diff.
crates/wasmi/src/engine/config.rs
Outdated
Comment on lines
86
to
141
| /// Sets the [`StackLimits`] for the [`Config`]. | ||
| pub fn set_stack_limits(&mut self, stack_limits: StackLimits) -> &mut Self { | ||
| self.stack_limits = stack_limits; | ||
| self | ||
| } | ||
|
|
||
| /// Returns the [`StackLimits`] of the [`Config`]. | ||
| pub(super) fn stack_limits(&self) -> StackLimits { | ||
| self.stack_limits | ||
| } | ||
|
|
||
| /// Sets the maximum amount of cached stacks for reuse for the [`Config`]. | ||
| /// | ||
| /// # Note | ||
| /// | ||
| /// Defaults to 2. | ||
| pub fn set_cached_stacks(&mut self, amount: usize) -> &mut Self { | ||
| self.cached_stacks = amount; | ||
| self | ||
| } | ||
|
|
||
| /// Returns the maximum amount of cached stacks for reuse of the [`Config`]. | ||
| pub(super) fn cached_stacks(&self) -> usize { | ||
| self.cached_stacks | ||
| } | ||
|
|
There was a problem hiding this comment.
Please rebase your work onto main. These methods already do not exist on main.
- new module in the wasmi crate - tests for (roundtrip) (de)serialization of Wasm modules
- introduce features to enable operation modes: - default (current way) - parser + serialization - deserialization
FedorSmirnov89
force-pushed
the
module-deser
branch
from
1957af0 to
fae56e1
Compare
Codecov Report❌ Patch coverage is Additional details and impacted files@@ Coverage Diff @@
## main #1636 +/- ##
==========================================
+ Coverage 70.39% 71.01% +0.61%
==========================================
Files 179 187 +8
Lines 15344 15799 +455
==========================================
+ Hits 10802 11220 +418
- Misses 4542 4579 +37 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Proof of Concept: Module Preparsing for Resource-Constrained Targets
This is a proof-of-concept implementation intended to demonstrate the concept and gather feedback. It is NOT intended for production use and contains several incomplete and unclean aspects that would need to be addressed in a proper implementation.
I'm happy to work with the Wasmi maintainers to implement this feature properly, starting from scratch if needed.
Overview
This PR demonstrates a module preparsing feature that enables Wasmi to run on resource-constrained embedded systems by allowing modules to be parsed and validated on development machines, then serialized to a compact format for lightweight runtime loading.
Problem Statement (also described in #1637 )
When deploying Wasmi on resource-constrained targets (e.g., nRF53 with ~1MB total memory), the current implementation consumes significant memory (~480 KiB when deployed with Embassy). This limits the ability to run complex WebAssembly workloads on embedded devices.
Proposed Solution
The feature adds module serialization and deserialization capabilities, allowing:
Implementation Details
New Features Added
serializationfeature for creating serialized modulesdeserializationfeature for lightweight runtime buildsparserfeature-gates the wasmi logic for parsing WebAssembly modules (compiling without this feature provides slim builds by excluding the parsing/validation/translation machinery)New Modules
preparsed/: Main module for serialization/deserializationpreparsed/serialized_module/: Data structures for serialized formatpreparsed/serialization/: Module serialization logicpreparsed/deserialization/: Module deserialization logicpreparsed/error.rs: Comprehensive error handlingSerialization Format
Performance Impact
Based on testing with nRF53 + Embassy + BLE stack:
What's Incomplete/Unclean
This proof-of-concept has several areas that need improvement:
1. Module Internals Visibility
pub(crate)visibility modifiers may be too permissive2. Error Handling
3. Feature Completeness
4. Architecture Integration
Example Repository
I've created an example repository demonstrating the usage and performance benefits: Preparsing Example
The repository shows:
Next Steps
If this approach is approved:
Questions for Maintainers