summaryrefslogtreecommitdiff
path: root/src/wasm2js.h
Commit message (Collapse)AuthorAgeFilesLines
* Add --deterministic flag to wasm2js, for fuzzing (#2757)Alon Zakai2020-04-131-0/+2
| | | | | | | | | | | | | | | | | | | | | | In wasm2js we ignore things that trap in wasm that we can't really handle, like a load from memory out of bounds would trap in wasm, but in JS we don't want to emit a bounds check on each load. So wasm2js focuses on programs that don't trap. However, this is annoying in the fuzzer as it turns out that our behavior for places where wasm would trap was not deterministic. That is, wasm would trap, wasm2js would not trap and do behavior X, and wasm2js with optimizations would also not trap but do behavior Y != X. This produced false positives in the fuzzer (and might be annoying in manual debugging too). As a workaround, this adds a --deterministic flag to wasm2js, which tries to be deterministic about what it does for cases where wasm would trap. This handles the case of an int division by 0 which traps in wasm but without this flag could have different behavior in wasm2js with or without opts (see details in the patch).
* Remove duplicate Type:: prefixes (NFC) (#2753)Heejin Ahn2020-04-121-4/+4
| | | | | | | | | These seem to be accidentally introduced in when we enforced use of `Type::` on type names in #2434. By the way TIL this actually compiles, and don't know why: ``` Type::Type::Type::Type::Type::Type::Type::Type::none ```
* Run reorder-locals more in wasm2js (#2729)Alon Zakai2020-04-081-0/+1
| | | | | | | | | | | coalesce-locals is nonlinear in the number of locals, so it is greatly beneficial to reorder the locals (which then drops the unused ones at the end automatically). The default passes do this already, but wasm2js does some custom work, and this was missing. With this change that pass takes 10x less time on poppler with --flatten --flatten --simplify-locals-notee-nostructure which approximates what wasm2js does.
* Avoid fp$ access in MAIN_MODULES (#2704)Alon Zakai2020-03-271-1/+1
| | | | | | | | | | | | | | | | Depends on emscripten-core/emscripten#10741 which ensures that table indexes are unique. With that guarantee, a main module can just add its function pointers into the table, and use them based on that index. The loader will then see them in the table and then give other modules the identical function pointer for a function, ensuring function pointer equality. This avoids calling fp$ functions during startup for the main module's own functions (which are slow). We do still call fp$s of things we import from outside, as we don't have anything to put in the table for them, we depend on the loader for that. I suspect this can also be done with SIDE_MODULES, but did not want to try too much at once.
* Initial multivalue support (#2675)Thomas Lively2020-03-051-0/+8
| | | | | | | | | Implements parsing and emitting of tuple creation and extraction and tuple-typed control flow for both the text and binary formats. TODO: - Extend Precompute/interpreter to handle tuple values - C and JS API support/testing - Figure out how to lower in stack IR - Fuzzing
* Fix missing newline after // EMSCRIPTEN_START_FUNCS and // ↵juj2020-02-101-2/+3
| | | | | | | | | | EMSCRIPTEN_END_FUNCS markers. (#2626) * Fix missing newline after // EMSCRIPTEN_START_FUNCS and // EMSCRIPTEN_END_FUNCS markers. * Flake * Update tests
* Optimize base64 decoding (#2623)juj2020-02-091-22/+26
| | | | | | | | | | | | | | * Optimize base64 decoding (about 7x-10x faster and temporary garbage-free compared to the original version) * new Uint8Array * Reuse Uint8Array view * Fix end handling * Code format * Update tests
* Add EH support for EffectAnalyzer (#2631)Heejin Ahn2020-02-031-5/+9
| | | | | | | | | | | | | | | | | | | | This adds EH support to `EffectAnalyzer`. Before `throw` and `rethrow` conservatively set property. Now `EffectAnalyzer` has a new property `throws` to represent an expression that can throw, and expression that can throw sets `throws` correctly. When EH is enabled, any calls can throw too, so we cannot reorder them with another expression with any side effects, meaning all calls should be treated in the same way as branches when evaluating `invalidate`. This prevents many reorderings, so this patch sets `throws` for calls only when the exception handling features is enabled. This is also why I passed `--disable-exception-handling` to `wasm2js` tests. Most of code changes outside of `EffectAnalyzer` class was made in order to pass `FeatureSet` to it. `throws` isn't always set whenever an expression contains a throwable instruction. When an throwable instruction is within an inner try, it will be caught by the corresponding inner catch, so it does not set `throws`.
* Remove implicit conversion operators from Type (#2577)Thomas Lively2020-01-081-11/+13
| | | | | | | | | | * Remove implicit conversion operators from Type Now types must be explicitly converted to uint32_t with Type::getID or to ValueType with Type::getVT. This fixes #2572 for switches that use Type::getVT. * getVT => getSingle
* [NFC] Enforce use of `Type::` on type names (#2434)Thomas Lively2020-01-071-36/+36
|
* [NFC] Clean up unnecessary `template`s in calls 🧹🧹🧹 (#2394)Thomas Lively2020-01-071-3/+2
|
* DCE at the end of wasm2js (#2574)Alon Zakai2020-01-061-0/+3
| | | | | | By doing so we ensure that our calls to convert wasm types to JS types never try to convert an unreachable. Fixes #2558
* Add support for reference types proposal (#2451)Heejin Ahn2019-12-301-0/+12
| | | | | | | | | | | | This adds support for the reference type proposal. This includes support for all reference types (`anyref`, `funcref`(=`anyfunc`), and `nullref`) and four new instructions: `ref.null`, `ref.is_null`, `ref.func`, and new typed `select`. This also adds subtype relationship support between reference types. This does not include table instructions yet. This also does not include wasm2js support. Fixes #2444 and fixes #2447.
* Remove FunctionType (#2510)Thomas Lively2019-12-111-5/+2
| | | | | | | | | | | | | | | | | Function signatures were previously redundantly stored on Function objects as well as on FunctionType objects. These two signature representations had to always be kept in sync, which was error-prone and needlessly complex. This PR takes advantage of the new ability of Type to represent multiple value types by consolidating function signatures as a pair of Types (params and results) stored on the Function object. Since there are no longer module-global named function types, significant changes had to be made to the printing and emitting of function types, as well as their parsing and manipulation in various passes. The C and JS APIs and their tests also had to be updated to remove named function types.
* Add string parameter to WASM_UNREACHABLE (#2499)Sam Clegg2019-12-051-39/+29
| | | | | This works more like llvm's unreachable handler in that is preserves information even in release builds.
* Convert to using DEBUG macros (#2497)Sam Clegg2019-12-041-1/+1
| | | | | | This means that debugging/tracing can now be enabled and controlled centrally without managing and passing state around the codebase.
* Multivalue type creation and inspection (#2459)Thomas Lively2019-11-221-1/+1
| | | | | | | | | | | | | Adds the ability to create multivalue types from vectors of concrete value types. All types are transparently interned, so their representation is still a single uint32_t. Types can be extracted into vectors of their component parts, and all the single value types expand into vectors containing themselves. Multivalue types are not yet used in the IR, but their creation and inspection functionality is exposed and tested in the C and JS APIs. Also makes common type predicates methods of Type and improves the ergonomics of type printing.
* vNxM.load_splat instructions (#2350)Thomas Lively2019-09-231-0/+4
| | | | | | | Introduces a new instruction class, `SIMDLoad`. Implements encoding, decoding, parsing, printing, and interpretation of the load and splat instructions, including in the C and JS APIs. `v128.load` remains in the `Load` instruction class for now because the interpreter code expects a `Load` to be able to load any memory value type.
* [wasm2js] Fix memory.size (#2330)Alon Zakai2019-09-051-11/+19
| | | | | | | We emitted the __wasm_memory_size function only when memory growth was enabled, but it can be used without that too. In theory we could only emit it if either memory growth or memory.size is used, but I think we can expect JS minifiers to do that later. Also fix a test suite bug - the check/auto_update script didn't run all the wasm2js tests when you run it with argument wasm2js (it used that as the list of tests, instead of the list of files, which confused me here for a while...).
* QFMA/QFMS instructions (#2328)Thomas Lively2019-09-031-1/+1
| | | | | | | | | Renames the SIMDBitselect class to SIMDTernary and adds the new {f32x4,f64x2}.qfm{a,s} ternary instructions. Because the SIMDBitselect class is no more, this is a backwards-incompatible change to the C interface. The new instructions are not yet used in the fuzzer because they are not yet implemented in V8. The corresponding LLVM commit is https://reviews.llvm.org/rL370556.
* Followup to workaround for minification of wasm2js mem init (#2318)Brion Vibber2019-08-301-3/+3
| | | | | | | | | | | | | Emscripten's minifier mis-minifies a couple bits in the memory init function that's used with wasm2js when not using an external memory init file: https://github.com/emscripten-core/emscripten/issues/8886 Previous fix worked around the bug in one place but failed to account for another. Have now confirmed that it works with this change in place. Updated test cases to match.
* Add atomic.fence instruction (#2307)Heejin Ahn2019-08-271-0/+4
| | | | | | | This adds `atomic.fence` instruction: https://github.com/WebAssembly/threads/blob/master/proposals/threads/Overview.md#fence-operator This also fix bugs in `atomic.wait` and `atomic.notify` instructions in binaryen.js and adds tests for them.
* wasm2js: Fix switch lowering, don't fall through after the hoisted parts (#2301)Alon Zakai2019-08-161-38/+18
| | | | | The switch lowering will "hoist" blocks of code into the JS switch when it can. If it can hoist some but not others, it must not fall through into those others (while it can fall through the hoisted ones - they began as nested blocks with falling-through between them). To fix this, after the hoisted ones issue a break out of the switch (which now contains all the hoisted code, so breaking out of it gets to the code right after the hoisted ones). fixes #2300
* Add basic exception handling support (#2282)Heejin Ahn2019-08-131-0/+16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This adds basic support for exception handling instructions, according to the spec: https://github.com/WebAssembly/exception-handling/blob/master/proposals/Exceptions.md This PR includes support for: - Binary reading/writing - Wast reading/writing - Stack IR - Validation - binaryen.js + C API - Few IR routines: branch-utils, type-updating, etc - Few passes: just enough to make `wasm-opt -O` pass - Tests This PR does not include support for many optimization passes, fuzzer, or interpreter. They will be follow-up PRs. Try-catch construct is modeled in Binaryen IR in a similar manner to that of if-else: each of try body and catch body will contain a block, which can be omitted if there is only a single instruction. This block will not be emitted in wast or binary, as in if-else. As in if-else, `class Try` contains two expressions each for try body and catch body, and `catch` is not modeled as an instruction. `exnref` value pushed by `catch` is get by `pop` instruction. `br_on_exn` is special: it returns different types of values when taken and not taken. We make `exnref`, the type `br_on_exn` pushes if not taken, as `br_on_exn`'s type.
* wasm2js: Mangle import names for JS (#2267)Alon Zakai2019-07-281-2/+3
| | | | | | | This fixes names that would be invalid in JS, like a.b. Turns out the Go compiler emits wasm with such imports. Also add some docs on how to use wasm2js. Fixes #2263
* Finalize tail call support (#2246)Thomas Lively2019-07-231-0/+6
| | | | Adds tail call support to fuzzer and makes small changes to handle return calls in multiple utilities and passes. Makes larger changes to DAE and inlining passes to properly handle tail calls.
* Simpify PassRunner.add() and automatically parallelize parallel functions ↵Alon Zakai2019-07-191-1/+1
| | | | | | | | | (#2242) Main change here is in pass.h, everything else is changes to work with the new API. The add("name") remains as before, while the weird variadic add(..) which constructed the pass now just gets a std::unique_ptr of a pass. This also makes the memory management internally fully automatic. And it makes it trivial to parallelize WalkerPass::run on parallel passes. As a benefit, this allows removing a lot of code since in many cases there is no need to create a new pass runner, and running a pass can be just a single line.
* Stop emitting "almost asm" in wasm2js output (#2221)Alon Zakai2019-07-121-2/+0
| | | We don't ever emit "use asm" anymore, so this similar annotation is not really useful, it just increases size.
* Add an option to emit a symbols file from wasm2js. (#2214)Alon Zakai2019-07-111-0/+10
| | | This can't use the normal wasm-opt mechanism because we modify the discard the wasm as part of running wasm2js, so we need to emit it in the proper place in the middle.
* fix wasm2js compilation after conflicting landings (#2209)Alon Zakai2019-07-081-0/+8
|
* wasm2js: use OverriddenVisitor, so we show a clear error on unsupported ↵Alon Zakai2019-07-081-2/+62
| | | | instructions (#2199)
* wasm2js: export memory growth function only if memory growth is enabled (#2194)Alon Zakai2019-07-031-2/+4
| | | Previously we tried to export it if the memory was exported, even if growth was not on, which caused an error.
* Workaround for wasm2js output minification issue with emscripten (#2185)Brion Vibber2019-07-011-2/+3
| | | | | | | | | | | | | | * Workaround for wasm2js output minification issue with emscripten When using emscripten with -O2 and --memory-init-file 0, the JS minification breaks on this function for memory initialization setup, causing an exception to be thrown during module setup. Moving from two 'var' declarations for the same variable to one should avoid hitting this with no change in functionality (the var gets hoisted anyway). https://github.com/emscripten-core/emscripten/issues/8886
* Bysyncify: async transform for wasm (#2172)Alon Zakai2019-06-151-1/+1
| | | | | | | | | This adds a new pass, Bysyncify, which transforms code to allow unwind and rewinding the call stack and local state. This allows things like coroutines, turning synchronous code asynchronous, etc. The new pass file itself has a large comment on top with docs. So far the tests here seem to show this works, but this hasn't been tested heavily yet. My next step is to hook this up to emscripten as a replacement for asyncify/emterpreter, see emscripten-core/emscripten#8561 Note that this is completely usable by itself, so it could be useful for any language that needs coroutines etc., and not just ones using LLVM and/or emscripten. See docs on the ABI in the pass source.
* wasm2js: Switch optimizations (#2141)Alon Zakai2019-05-281-8/+202
| | | | | This pattern-matches towers of blocks + a br_table into a JS switch. This is much smaller in code size and also avoids heavy nesting that can exceed the recursion limits of JS parsers. This is not enough yet, because it pattern-matches very specifically. In reality, switches can look slightly different. Followup PRs will extend this. For now, this passes the test suite (what passed before - not including the massive-switch tests) + fuzzing so it's a good start.
* Reflect instruction renaming in code (#2128)Heejin Ahn2019-05-211-27/+27
| | | | | | | | | | | | | | | | | | | | | | | | | | | | - Reflected new renamed instruction names in code and tests: - `get_local` -> `local.get` - `set_local` -> `local.set` - `tee_local` -> `local.tee` - `get_global` -> `global.get` - `set_global` -> `global.set` - `current_memory` -> `memory.size` - `grow_memory` -> `memory.grow` - Removed APIs related to old instruction names in Binaryen.js and added APIs with new names if they are missing. - Renamed `typedef SortedVector LocalSet` to `SetsOfLocals` to prevent name clashes. - Resolved several TODO renaming items in wasm-binary.h: - `TableSwitch` -> `BrTable` - `I32ConvertI64` -> `I32WrapI64` - `I64STruncI32` -> `I64SExtendI32` - `I64UTruncI32` -> `I64UExtendI32` - `F32ConvertF64` -> `F32DemoteI64` - `F64ConvertF32` -> `F64PromoteF32` - Renamed `BinaryenGetFeatures` and `BinaryenSetFeatures` to `BinaryenModuleGetFeatures` and `BinaryenModuleSetFeatures` for consistency.
* wasm2js: optimize away unneeded load coercions (#2107)Alon Zakai2019-05-151-0/+3
|
* wasm2js: Emit table in a way that is friendly to emscripten minification (#2102)Alon Zakai2019-05-131-2/+9
| | | Set it to a local in the asmFunc scope, so that minifiers can easily see it's a simple local value (instead of using it as an upvar from the parameters higher up, which was how the emscripten glue was emitting it).
* wasm2js: avoid reinterprets (#2094)Alon Zakai2019-05-101-0/+11
| | | | | In JS a reinterpret is especially expensive, as we implement it as a write to a temp buffer and a read using another view. This finds places where we load a value from memory, then reinterpret it later - in that case, we can load it using another view, at the cost of another load and another local. This is helpful on things like Box2D, where there are many reinterprets due to the main 2D vector class being an union over two floats/ints, and LLVM likes to do a single i64 load of them.
* wasm2js: 'handle' saturating float to int conversions - we don't handle any ↵Alon Zakai2019-05-061-5/+12
| | | | of them precisely anyhow (#2087)
* wasm2js: optimize loads (#2085)Alon Zakai2019-05-031-2/+0
| | | | When loading a boolean, prefer the signed heap (which is more commonly used, and may be faster). We never use HEAPU32 (HEAP32 is always enough), just remove it.
* wasm2js: avoid some slow operations when not optimizing (#2082)Alon Zakai2019-05-031-4/+6
| | | Without this PR, wasm2js0.test_printf in emscripten took an extremely long time to compile.
* wasm2js: ignore implicit traps (#2079)Alon Zakai2019-05-021-1/+9
| | | | | We don't actually try to emit traps for loads, stores, invalid float to ints, etc., so when optimizing we may as well do so under the assumption those traps do not exist. This lets us emit nice code for a select whose operands are loads, for example - otherwise, the values seem to have side effects.
* Add a pass to lower unaligned loads and stores (#2078)Alon Zakai2019-05-021-97/+5
| | | | | This replaces the wasm2js code that lowered them to pessimistic (1-byte aligned) loads and stores. The new pass will do the optimal thing, keeping 2-byte alignment where possible. This is also nicer as a standalone pass, which has the simple property that after it runs all loads and stores are aligned, instead of some code scattered inside wasm2js.
* wasm2js: don't run coalesce-locals if not optimizing (#2076)Alon Zakai2019-05-011-1/+5
| | | That pass is very slow on unoptimized code (super-linear on the number of locals, which if unoptimized can be massive due to flatten).
* clang-tidy braces changes (#2075)Alon Zakai2019-05-011-6/+10
| | | Applies the changes in #2065, and temprarily disables the hook since it's too slow to run on a change this large. We should re-enable it in a later commit.
* wasm2js: run more optimizations (#2073)Alon Zakai2019-05-011-2/+3
| | | In particular, coalesce-locals is useful even if closure is run later (apparently it finds stuff closure can't).
* wasm2js: run full optimizations during the pipeline (#2071)Alon Zakai2019-04-301-23/+34
| | | | | We flatten for the i64 lowering etc. passes, and it is worth optimizing afterwards, to clean up stuff they created. That is run if the user ran wasm2js with an optimization level (like wasm2js -O3). Split the test files to check both optimized and unoptimized code.
* wasm2js: optimize switches (#2067)Alon Zakai2019-04-301-3/+22
| | | | | | | | | | Don't emit unneeded breaks in switch cases, instead do case X: case Y: .. case W: break .. for each group. Also, the group with the default doesn't need any cases but the default itself.
* Apply format changes from #2048 (#2059)Alon Zakai2019-04-261-531/+491
| | | Mass change to apply clang-format to everything. We are applying this in a PR by me so the (git) blame is all mine ;) but @aheejin did all the work to get clang-format set up and all the manual work to tidy up some things to make the output nicer in #2048
generated by cgit v1.2.3 (git 2.39.1) at 2025-08-06 01:32:00 +0000