summaryrefslogtreecommitdiff
path: root/src
Commit message (Collapse)AuthorAgeFilesLines
* [Wasm GC][NFC] Optimize getStrictSubTypes() (#5130)Alon Zakai2022-10-121-1/+11
| | | | Avoid allocating there. This is both faster and also it ensures we never modify our internal data structure after our constructor.
* [Parser][NFC] Move `ParseInput` into the parser context (#5132)Thomas Lively2022-10-122-1346/+1188
| | | | | | | | | | | | | Rather than passing both a `Ctx` and a `ParseInput` to every parsing function, pass only a `Ctx` with a `ParseInput` inside of it. This significantly reduces verbosity in the parser. To handle cases where parsing needs to happen at specific locations, which used to be handled by constructing a new `ParseInput` independent from the ctx, introduce a new RAII utility for temporarily changing the location of the `ParseInput` inside a context. Also add a utility for generating an error at a particular location to avoid having to construct new `ParseInput` objects just for that purpose. This resolves a few TODOs about correcting error locations, but since we don't test those yet, I still consider this NFC.
* [Wasm GC] Fix the intersection of a bottom type null (#5128)Alon Zakai2022-10-121-2/+8
| | | | | When the heap types are not subtypes of each other, but a null is possible, the intersection exists and is a null. That null must be the shared bottom type.
* [Wasm GC] [GUFA] Add initial ConeType support (#5116)Alon Zakai2022-10-113-76/+296
| | | | | | | | | | | A cone type is a PossibleContents that has a base type and a depth, and it contains all subtypes up to that depth. So depth 0 is an exact type from before, etc. This only adds cone type computations when combining types, that is, when we combine two exact types we might get a cone, etc. This does not yet use the cone info in all places (like struct gets and sets), and it does not yet define roots of cone types, all of which is left for later. IOW this is the MVP of cone types that is just enough to add them + pass tests + test the new functionality.
* Support null characters in string.const expressions (#5123)Thomas Lively2022-10-112-16/+8
| | | | Remove an obsolete error about null characters and test both binary and text round tripping of a string constant containing an escaped zero byte.
* Make `Name` a pointer, length pair (#5122)Thomas Lively2022-10-1158-1244/+1251
| | | | | | | | | | | | | | | | | | | | | | | With the goal of supporting null characters (i.e. zero bytes) in strings. Rewrite the underlying interned `IString` to store a `std::string_view` rather than a `const char*`, reduce the number of map lookups necessary to intern a string, and present a more immutable interface. Most importantly, replace the `c_str()` method that returned a `const char*` with a `toString()` method that returns a `std::string`. This new method can correctly handle strings containing null characters. A `const char*` can still be had by calling `data()` on the `std::string_view`, although this usage should be discouraged. This change is NFC in spirit, although not in practice. It does not intend to support any particular new functionality, but it is probably now possible to use strings containing null characters in at least some cases. At least one parser bug is also incidentally fixed. Follow-on PRs will explicitly support and test strings containing nulls for particular use cases. The C API still uses `const char*` to represent strings. As strings containing nulls become better supported by the rest of Binaryen, this will no longer be sufficient. Updating the C and JS APIs to use pointer, length pairs is left as future work.
* [Parser] Parse folded instructions (#5124)Thomas Lively2022-10-111-3/+75
| | | | | | | | | | | | | | | Parse folded expressions as described in the spec: https://webassembly.github.io/spec/core/text/instructions.html#folded-instructions. The old binaryen parser _only_ parses folded expressions, and furthermore requires them to be folded such that a parent instruction consumes the values produced by its children and only those values. The standard format is much more general and allows folded instructions to have an arbitrary number of children independent of dataflow. To prevent the rest of the parser from having to know or care about the difference between folded and unfolded instructions, parse folded instructions after their children have been parsed. This means that a sequence of instructions is always parsed in the order they would appear in a binary no matter how they are folded (or not folded).
* [NFC] Remove more `cashew::` namespaces from IString (#5127)Thomas Lively2022-10-112-82/+81
| | | Finishes work missed in #5126.
* [NFC] Remove `cashew::` namespace from IString (#5126)Thomas Lively2022-10-105-127/+128
| | | | As an NFC preliminary change that will minimize the diff in #5122, which moves IString to the wasm namespace.
* Validate memory initial < max only if memory hasMax (#5125)Ashley Nelson2022-10-071-2/+4
| | | Making a change to wasm-validator so that Memory::kUnlimitedSize is treated properly like an unlimited case. The check for whether memory.initial < memory.max will only happen if memory.hasMax() — meaning if memory.max is not set to kUnlimitedSize.
* GUFA: Use SSA-style information (#5121)Alon Zakai2022-10-072-56/+59
| | | | | | | | | | | | | | | | | Previously we treated each local index as a location, and every local.set to that index could be read by every local.get. With this we connect only relevant sets to gets. Practically speaking, this removes LocalLocation which is what was just described, and instead there is ParamLocation for incoming parameter values. And local.get/set use normal ExpressionLocations to connect a set to a get. I was worried this would be slow, since computing LocalGraph takes time, but it actually more than makes up for itself on J2Wasm and we are faster actually rocket I guess since we do less updating after local.sets. This makes a noticeable change on the J2Wasm binary, and perhaps will help with benchmarks.
* [DWARF] Warn on unsupport DWARF versions and content (#5120)Alon Zakai2022-10-071-3/+10
| | | | | Unfortunately there isn't a single place where an error may occur. I tested on several files with different flags and added sufficient warnings so that we warn on them all.
* Implement bottom heap types (#5115)Thomas Lively2022-10-0728-372/+858
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | These types, `none`, `nofunc`, and `noextern` are uninhabited, so references to them can only possibly be null. To simplify the IR and increase type precision, introduce new invariants that all `ref.null` instructions must be typed with one of these new bottom types and that `Literals` have a bottom type iff they represent null values. These new invariants requires several additional changes. First, it is now possible that the `ref` or `target` child of a `StructGet`, `StructSet`, `ArrayGet`, `ArraySet`, or `CallRef` instruction has a bottom reference type, so it is not possible to determine what heap type annotation to emit in the binary or text formats. (The bottom types are not valid type annotations since they do not have indices in the type section.) To fix that problem, update the printer and binary emitter to emit unreachables instead of the instruction with undetermined type annotation. This is a valid transformation because the only possible value that could flow into those instructions in that case is null, and all of those instructions trap on nulls. That fix uncovered a latent bug in the binary parser in which new unreachables within unreachable code were handled incorrectly. This bug was not previously found by the fuzzer because we generally stop emitting code once we encounter an instruction with type `unreachable`. Now, however, it is possible to emit an `unreachable` for instructions that do not have type `unreachable` (but are known to trap at runtime), so we will continue emitting code. See the new test/lit/parse-double-unreachable.wast for details. Update other miscellaneous code that creates `RefNull` expressions and null `Literals` to maintain the new invariants as well.
* Warn on too many parameters for Web VMs (#5119)Alon Zakai2022-10-062-1/+7
| | | Fixes emscripten-core/emscripten#17988
* wasm2js: Support for flexible module import naming (#5114)Sam Clegg2022-10-051-40/+75
| | | | | | | | The previous code was making emscripten-specific assumptions about imports basically all coming from the `env` module. I can't find a way to make this backwards compatible so may do a combined roll with the emscripten-side change: https://github.com/emscripten-core/emscripten/pull/17806
* [GUFA] Fix call.without.effects's handling of the last parameter (#5111)Alon Zakai2022-10-051-0/+12
| | | | | | | | | The last parameter is the function to call, and we treated it like a normal parameter. This is mostly only an issue during debugging, but in theory sending this extra value could cause us to optimize less later (since it gets added to what the local of that index can contain). Also add assertions which would have caught this before.
* Strip em_js_deps exports (#5109)Sam Clegg2022-10-042-2/+8
| | | These are only needed for the metadata extraction in emcc.
* Simplify and fix heap type counting (#5110)Thomas Lively2022-10-041-15/+12
| | | | | Annotations on array.get and array.set were not being counted and the code could generally be simplified since `count` already ignores types that don't need to be counted.
* Make Asyncify work with wasm64 (#5105)Sam Clegg2022-10-041-57/+75
| | | | | The emscripten side is a little tricky but I've got some tests passing. Currently blocked on: https://github.com/emscripten-core/emscripten/issues/17969
* Refactor standardizeNaN (#5064)Max Graey2022-10-043-62/+55
| | | Change `standardizeNaN` to take a `Literal` to reduce usage verbosity.
* [Fuzzing] Allow recombine() to replace with a subtype (#5101)Alon Zakai2022-10-031-4/+43
| | | | Previously it would randomly replace an expression with another one with the exact same type. Allowing a subtype may give us more coverage.
* Fix ordering of visit() in MemoryGrow interpretation (#5108)Alon Zakai2022-10-031-4/+4
| | | | | | | This is a pretty subtle point that was missed in #4811 - we need to first visit the child, then compute the size, as the child may alter that size. Found by the fuzzer.
* Change `exit()` to `Fatal()`; make it possible to throw on `Fatal()`. (#5087)Brian Anderson2022-10-012-12/+27
| | | | | | | | | | | | | | | | This patch makes binaryen easier to call from other applications by making more errors recoverable instead of early-exiting. The main thing it does is change three calls to exit on I/O errors into calls to Fatal(), which is an existing custom abstraction for handling unrecoverable errors. Currently Fatal's destructor calls _Exit(1). My intent is to make it possible for Fatal to not exit, but to throw, allowing an embedding application to catch the exception. Because the previous early exits were exiting with error code EXIT_FAILURE, I also changed Fatal to exit with EXIT_FAILURE. The test suite continues to pass so I assume this is ok. Next I changed Fatal to buffer its error message until the destructor instead of immediately printing it to stderr. This is for ease of patching Fatal to throw instead. Finally, I also included the patch I need to make Fatal throw when THROW_ON_FATAL is defined at compile time. I can carry this patch out of tree, but it is a small patch, so perhaps you will be willing to take it. I am happy to remove it. Fixes #4938
* [Wasm GC] Fix .depth() and add testing (#5102)Alon Zakai2022-09-301-0/+30
|
* Refactor interaction between Pass and PassRunner (#5093)Thomas Lively2022-09-3093-336/+469
| | | | | | | | | | | | | | Previously only WalkerPasses had access to the `getPassRunner` and `getPassOptions` methods. Move those methods to `Pass` so all passes can use them. As a result, the `PassRunner` passed to `Pass::run` and `Pass::runOnFunction` is no longer necessary, so remove it. Also update `Pass::create` to return a unique_ptr, which is more efficient than having it return a raw pointer only to have the `PassRunner` wrap that raw pointer in a `unique_ptr`. Delete the unused template `PassRunner::getLast()`, which looks like it was intended to enable retrieving previous analyses and has been in the code base since 2015 but is not implemented anywhere.
* Fix bugs with copying expressions (#5099)Thomas Lively2022-09-302-1/+10
| | | | | | | | | | | | | | | | | It does not make sense to construct an `Expression` directly because all expressions must be specific expressions. However, we previously allowed constructing Expressions, and in particular we allowed them to be copy constructed. Unrelatedly, `Fatal::operator<<` took its argument by value. Together, these two facts produced UB when printing Expressions in fatal error messages because a new Expression would be copy constructed with the original expression ID but without any of the actual data from the original specific expression. For example, when trying to print a Block, the printing code would try to look at the expression list, but the expression list would be junk stack data because the copied Expression does not contain an expression list. Fix the problem by making Expression's constructors visible only to its subclasses and making `Fatal::operator<<` take its argument by forwarding reference instead of by value.
* Fix handling of unreachable selects in Directize (#5098)Alon Zakai2022-09-301-1/+1
| | | | We ignored only unreachable conditions, but we must ignore the arms as well, or else we could error.
* [GUFA] Improve hashing (#5091)Alon Zakai2022-09-282-12/+11
| | | | Avoid manually doing bitshifts etc. - leave combining to the core hash logic, which can do a better job.
* Fix nondeterminism in GlobalStructInference (#5092)Alon Zakai2022-09-281-1/+8
| | | | | We append to vectors of globals in a nondeterministically-ordered loop, which can lead to different orderings of the vectors. This happens quite frequently in very large J2Wasm files it turns out. As a solution, simply sort them after the nondeterministic stage.
* [GUFA] Prepare for cone types [NFC] (#5086)Alon Zakai2022-09-282-40/+69
| | | | | | | | | | | | This does not actually add cone types, but it does NFC refactoring towards that. Specifically it replaces the internal ExactType with ConeType, and the latter has a depth, so a cone type of depth 0 is the old exact type. Cone types with depth > 0 are not possible yet, keeping this NFC. I believe this design with a depth for cone types has little overhead. It does add to the size of ConeType, but the variant there is larger anyhow (it contains a Literal). And things like isSubType need to loop anyhow, so looping up to the depth etc. in checks won't make things slower.
* [GUFA] Fix haveIntersection on comparing nullable with non-nullable (#5089)Alon Zakai2022-09-281-2/+23
| | | | | We compared types and not heap types, so a difference in nullability confused us. But at that point in the code, we've ruled out nulls, so we should focus on heap types only.
* Memory64Lowering: Ignore data segments with non-const iniital offset (#5074)Sam Clegg2022-09-281-3/+31
| | | | This is the case for dynamic linking where the segment offset are derived from he `__memory_base` import.
* [GUFA] Simplify RefTest logic [NFC] (#5084)Alon Zakai2022-09-272-40/+30
| | | Move the logic to the GUFA pass.
* [GUFA] Optimize functions not taken by reference better (#5085)Alon Zakai2022-09-261-21/+14
| | | | | | | | | This moves the logic to add connections from signatures to functions from the top level into the RefFunc logic. That way we only add those connections to functions that actually have a RefFunc, which avoids us thinking that a function without one can be reached by a call_ref of its type. Has a small but non-zero benefit on j2wasm.
* [GUFA] Infer a RefEq value of 0 when possible (#5081)Alon Zakai2022-09-263-11/+115
| | | | If the PossibleContents for the two sides have no possible intersection then the result must be 0.
* [NFC] Simplify traversal code for setting the module (#5082)Alon Zakai2022-09-262-4/+3
| | | | Make walkModuleCode set the module automatically, like walkModule already does. Also remove some unneeded module settings when calling those methods.
* Emit call_ref with a type annotation (#5079)Thomas Lively2022-09-235-39/+33
| | | | | | | Emit call_ref instructions with type annotations and a temporary opcode. Also implement support for parsing optional type annotations on call_ref in the text and binary formats. This is part of a multi-part graceful update to switch Binaryen and all of its users over to using the type-annotated version of call_ref without there being any breakage.
* [C API] Make TypeBuilderSetSubType take a heap type (#5045)dcode2022-09-235-15/+15
| | | Fixes #5041
* Add a type annotation to return_call_ref (#5068)Thomas Lively2022-09-226-35/+88
| | | | | | The GC spec has been updated to have heap type annotations on call_ref and return_call_ref. To avoid breaking users, we will have a graceful, multi-step upgrade to the annotated version of call_ref, but since return_call_ref has no users yet, update it in a single step.
* Correctly handle escapes in string constants (#5070)Thomas Lively2022-09-222-66/+74
| | | | | | | Previously when we parsed `string.const` payloads in the text format we were using the text strings directly instead of un-escaping them. Fix that parsing, and while we're editing the code, also add support for the `\r` escape allowed by the spec. Remove a spurious nested anonymous namespace and spurious `static`s in Print.cpp as well.
* Fix some Closure warnings in Emscripten builds (#5075)dcode2022-09-222-0/+15
| | | | | See #5062 Also add a require() workaround, see https://github.com/emscripten-core/emscripten/pull/17851
* Remove some unused constants. NFC (#5072)Sam Clegg2022-09-222-6/+0
| | | | | TABLE_BASE usage was removed in #3211. MEMORY_BASE usage was removed in #3089. NEW_SIZE usage was removed in #3180.
* [GUFA] Optimize ref.test (#5067)Alon Zakai2022-09-221-12/+51
| | | | | | Similar to ref.cast slightly, but simpler. Also update some TODO text.
* [OptimizeInstruction] Prevent reordering for rule in #5034 (#5066)Max Graey2022-09-211-2/+3
|
* Add wasm64 support in OptimizeAddedConstants (#5043)Axis2022-09-211-9/+24
| | | This lets that pass optimize 64-bit offsets on memory64 loads and stores.
* Make closure errors into warnings in the Emscripten builds (#5063)Thomas Lively2022-09-201-1/+4
| | | | This should make the CI green again. Also fix one of the errors. I haven't fixed the other errors because I don't know how.
* [OptimizeInstructions] Simplify add / sub with negative on LHS or RHS for ↵Max Graey2022-09-201-0/+25
| | | | | | | | | floating points (#5034) ``` (-x) + y -> y - x x + (-y) -> x - y x - (-y) -> x + y ```
* [C-/JS-Api] Expose the multi memories feature (#4973)Max Graey2022-09-203-0/+5
| | | This finalizes the multi memories feature introduced in #4968.
* [Debugging] Fix compile error for dumping LocalGraph (#5055)Axis2022-09-201-4/+4
|
* [Strings] Add missing String effects + tests (#5057)Alon Zakai2022-09-191-3/+22
| | | Also fix some formatting issue in the file.
generated by cgit v1.2.3 (git 2.39.1) at 2025-08-15 17:59:54 +0000