923: Add probable fix of memory leak in trampoline code r=MarkMcCaskey a=MarkMcCaskey
might be what's needed for #810 ; but despite my best efforts I could not get asan working on osx, so I did not test it.
By my count this accounts for 40 bytes, so there may be another issue
edit: unless Rust is optimizing out turning a zero-sized type into a Box, in which case, this should account for 48 bytes
Co-authored-by: Mark McCaskey <mark@wasmer.io>
Host functions use `Context::External` with a `*mut vm::FuncCtx`
pointer, casted to `*mut vm::Ctx`. It creates a conflict with exports
that also use `Context::External`.
This patch introduces `Context::ExternalWithEnv` to create a specific
path in the code for an external context with `*mut vm::FuncEnv`.
This patch fixes all the `linking.wast` tests in the spectests.
In the `wrap` functions, we use `std::mem::transmute(&())` to get the
pointer to the value “around” `wrap` (`Fn` has a method `to_raw` which
declares a `wrap` function, which uses `transmute` to retrieve
`Fn`). This is an undefined behavior. It was working until the
`FuncCtx` is introduced. Since then, the undefined behavior was
causing an error with the Singlepass backend.
This patch stores the pointer to `Fn` in `func_env`, so that the
pointer to the user-defined host function is always predictable.
929: Update __wasi_rights_t and __wasi_signal_t with published changes r=MarkMcCaskey a=MarkMcCaskey
Follow up to #926 ; fixes everything else listed at https://github.com/WebAssembly/WASI/pull/135
Co-authored-by: Mark McCaskey <mark@wasmer.io>
911: Don't emit bounds checks when the offset is known at compile time to be less than the minimum memory size. r=nlewycky a=nlewycky
Co-authored-by: Nick Lewycky <nick@wasmer.io>
915: fix(runtime-core) Share the definition of `Trampoline` across all the backends r=Hywan a=Hywan
Extracted from https://github.com/wasmerio/wasmer/pull/882.
This patch updates all the backends to use the definition of
`Trampoline` as defined in the `wasmer_runtime_core::typed_func`
module. That way, there is no copy of that type, and as such, it is
easier to avoid regression (a simple `cargo check` does the job).
This patch also formats the `use` statements in the updated files.
Co-authored-by: Ivan Enderlin <ivan.enderlin@hoa-project.net>
This patch allows host functions to get a signature without an
explicit `vm::Ctx` argument.
It is for Rust only. The C API receives a function pointer and has no
clue whether a `vm::Ctx` argument is present or not, so it assumes it
is always declared.
From the backend point of view, the pointer to `vm::Ctx` is always
inserted in the stack, but it is not used by the function when the
argument is absent.
916: feat(runtime-core-tests) Introduce the new `wasmer-runtime-core-tests` crate r=Hywan a=Hywan
Extracted from https://github.com/wasmerio/wasmer/pull/882.
This non-publishable new crate contains a test suite for the
`wasmer-runtime-core` crate. So far, the test suite is rather small,
but it aims to be extended in a close future.
Co-authored-by: Ivan Enderlin <ivan.enderlin@hoa-project.net>
This non-publishable new crate contains a test suite for the
`wasmer-runtime-core` crate. So far, the test suite is rather small,
but it aims to be extended in a close future.
This patch updates all the backends to use the definition of
`Trampoline` as defined in the `wasmer_runtime_core::typed_func`
module. That way, there is no copy of that type, and as such, it is
easier to avoid regression (a simple `cargo check` does the job).
This patch also formats the `use` statements in the updated files.
914: Fix bug in getting a Memory from export in C API r=Hywan a=MarkMcCaskey
I believe this fixes the non-deterministic crashing on OSX in `go-ext-wasm`
Co-authored-by: Mark McCaskey <mark@wasmer.io>
856: Add WASI support to C API r=MarkMcCaskey a=MarkMcCaskey
This is an additive change (with one unrelated clean up of `*mut ptr -> *const ptr`). It exposes the functions to get a WASI import object. It also implements a function on import object to get an `import` with namespace and name.
These changes should be okay to ship now, we can follow up to finish adding methods to ImportObject on the C API side and start migrating away from `*imports`.
# Review
- [x] Add a short description of the the change to the CHANGELOG.md file
Co-authored-by: Mark McCaskey <mark@wasmer.io>
Co-authored-by: Ivan Enderlin <ivan.enderlin@hoa-project.net>
Co-authored-by: Mark McCaskey <markmccaskey@users.noreply.github.com>
Co-authored-by: nlewycky <nick@wasmer.io>
901: Set target triple and datalayout when creating the LLVM module. r=nlewycky a=nlewycky
We were giving LLVM a triple and datalayout only when producing native code from the LLVM IR. With this change, we tell LLVM as early as possible so that the entire optimization stack knows that it's safe to use target-specific constructs (including target intrinsics `@llvm.x86.sse2.ucomieq.sd`) as well as cost models (for autovectorization) and knowing the bitwidth of the registers so that we can know it's profitable to eliminate redundant extend/truncate casts.
Co-authored-by: Nick Lewycky <nick@wasmer.io>
904: Use getelementptr instruction instead of int_to_ptr and ptr_to_int. r=nlewycky a=nlewycky
The main part of this change is that we no longer turn pointers into integers to do arithmetic on them, then turn them back into pointers. Doing so is a signal to LLVM that it should not attempt to analyze the provenance of the pointers, disabling some optimizations. Using getelementptr allows us to perform arithmetic on pointers while keeping them in pointer types, which LLVM can then analyze.
Most of the textual change is a refactoring to reorder the operations. Previously the bounds checking and determining of the base and bounds were combined because you could put both into the same match, since both actions are performed differently depending on whether the memory is static or dynamic. In this case, we simply check the type twice and do two different things, with comments labelling the steps.
Co-authored-by: Nick Lewycky <nick@wasmer.io>
ptr_to_int and int_to_ptr are not treated the same as type-safe pointer operations. LLVM intentionally performs fewer optimizations to code using the former as compared with the latter. We don't need that in our code, standard pointer optimizations are safe for us.
877: Reimplement F32Min, F32Max, F64Min and F64Max. r=nlewycky a=nlewycky
# Description
Reimplement F32Min, F32Max, F64Min and F64Max.
Adds XMM8--15 registers. Adds VMOVA, VBLEND and VXORP, and the VCMPUNORD and VCMPORD comparisons.
Fixes 419 test failures.
Co-authored-by: Nick Lewycky <nick@wasmer.io>
887: Get aggressive about running cargo check over all packages with all features. r=nlewycky a=nlewycky
Fixes the one issue uncovered. The capstone disassembling support in the LLVM backend was broken. Fixed by removing it. Instead, use the `--llvm-object-file` flag to get a finished object file to disassemble with any disassembler.
Co-authored-by: Nick Lewycky <nick@wasmer.io>
Co-authored-by: Mark McCaskey <mark@wasmer.io>
889: fix(singlepass-backend) Use wasmparser from `runtime-core` r=Hywan a=Hywan
The `wasmer-runtime-core` crate re-exports the `wasmparser`
crate. This patch updates the `singlepass-backend` crate to use
`wasmparser` through the `wasmer-runtime-core` crate, which removes a
direct dependency for this crate.
Co-authored-by: Ivan Enderlin <ivan.enderlin@hoa-project.net>
The `wasmer-runtime-core` crate re-exports the `wasmparser`
crate. This patch updates the `singlepass-backend` crate to use
`wasmparser` through the `wasmer-runtime-core` crate, which removes a
direct dependency for this crate.
Fixes the one issue uncovered. The capstone disassembling support in the LLVM backend was broken. Fixed by removing it. Instead, use the `--llvm-object-file` flag to get a finished object file to disassemble with any disassembler.
885: fix(cranelift-backend) Remove broken (and useless?) debug code r=nlewycky a=Hywan
This debug code is broken (it doesn't compile). It can be assumed
nobody uses it, and can be considered as dead code. As such, this PR
removes it.
The `debug` feature is kept in `Cargo.toml` to activate `wasmer-runtime-core/debug`.
Co-authored-by: Ivan Enderlin <ivan.enderlin@hoa-project.net>
Instead of ensuring outputs are arithmetic NaNs on every function, we tag them as pending such a check, so that a sequence of computation can have a single canonicalization step at the end.
There's an extra wriggle for SIMD. The Wasm type system only indicates them as V128, so it's possible that we might do computations as F32x4Add, I8x16Add, F64x2Add in a row with no other computations in between. Thus, most SIMD functions apply pending canonicalizations to their inputs, even integer SIMD operations.
After a long time, the macros are easy to read, but not at first
glance. I hope this PR will improve the situation: Same syntax used
everywhere, more spaces…
863: Rewrite Min/Max to handle all cases correctly. Fixes 545 spectest failures. r=nlewycky a=nlewycky
# Description
The llvm backend was not quite following the Wasm spec for {F32,F64,F32x4xF64x2}{Min,Max}. We used the `@llvm.minnum` and `@llvm.maxnum` intrinsics which don't handle the corner cases the same. When we tried to use `@llvm.minimum` and `@llvm.maximum` which do, we get an internal error from the x86 backend. I was hoping that crash would go away with the upgrade to LLVM 9, but it does not.
Reimplement these operations using plain LLVM instructions.
# Review
- [x] Add a short description of the the change to the CHANGELOG.md file
Co-authored-by: Nick Lewycky <nick@wasmer.io>
870: Fix unused value warning due to inkwell API change. NFC. r=nlewycky a=nlewycky
# Description
Fix unused value warning due to inkwell API change. No functionality change.
Co-authored-by: Nick Lewycky <nick@wasmer.io>
