1020: Turn a few more assert!s that should never fire into debug_assert!s. r=nlewycky a=nlewycky
These are here to protect against errors when refactoring more than anything else.
Co-authored-by: Nick Lewycky <nick@wasmer.io>
934: In LLVM backend, track which floats are guaranteed to be arithmetic, which makes the canonicalization a no-op. r=nlewycky a=nlewycky
# Description
This is a reimplementation of the patch in PR #651.
Extend state.rs ExtraInfo to track more information about floats. In addition to tracking whether the value has a pending canonicalization of NaNs, also track whether the value is known to be arithmetic (which includes infinities, regular values, and non-signalling NaNs (aka. "arithmetic NaNs" in the webassembly spec)). When the value is arithmetic, the correct sequence of operations to canonicalize the value is a no-op. Therefore, we create a lattice where pending+arithmetic=arithmetic.
Also, this extends the tracking to track all values, including non-SIMD integers. That's why there are more places where pending canonicalizations are applied.
Looking at c-wasm-simd128-example, this provides no performance change to the non-SIMD case (takes 58s on my noisy dev machine). The SIMD case drops from 46s to 29s.
# Review
- [ ] Add a short description of the the change to the CHANGELOG.md file
Co-authored-by: Nick Lewycky <nick@wasmer.io>
Fix a bug in Operator::Select and add a comment to explain the intention.
Use derived default for ExtraInfo.
Make ExtraInfo associated functions const.
Turn two asserts into debug_asserts.
We want to ignore the incoming pending NaN state (since the pending will propagate to the output if there was one on the input), and we want to add a new pending NaN state if we can (that is to say, if it isn't cancelled out by both inputs having arithmetic state). Do this by discarding the pending states on the inputs, intersecting them (to keep only the arithmetic state), then union in a pending nan state (which might do nothing, if it's arithmetic).
If the above sounds confusing, keep in mind that when a value is arithmetic, the act of performing the "NaN canonicalization" is a no-op. Thus, being arithmetic cancels out pending NaN states.
Unfortunately, this is quite buggy. For something as simple as F32Sub, to combine two ExtraInfos, we want to add a new pending_f32_nan(), unless both of the inputs are arithmetic_f32(). In this commit, we incorrectly calculate that we don't need a pending_f32_nan if either one of the inputs was arithmetic_f32().
It seemed like a good idea at the time, but in practice we discard the extra info all or almost all of the time.
This also introduces a new bug. In an operation like multiply, it's valid to multiply two values, one with a pending NaN and one without. As written, in the SIMD case (because of the two kinds of pending in play), we assert.
1017: Add a 'clear' method to map. r=nlewycky a=nlewycky
# Description
Motivated by work on the `nlewycky/inkwell-lifetimes` branch. No callers yet on master, but we use it on the branch and I'd like to merge in pieces.
Co-authored-by: Nick Lewycky <nick@wasmer.io>
1016: [fix 962] Activate pre_validation when llvm backend selected r=nlewycky a=pventuzelo
# Description
After analysis of issue https://github.com/wasmerio/wasmer/issues/962 by me and @nlewycky, it seems to be better to validate the wasm module before calling the llvm compiler.
This change will imply potential overhead in term of execution speed using llvm backend.
I will also suggest to set pre_validation to true for Singlepass by default:
5c1c786e35/lib/runtime-core/src/codegen.rs (L263-L269)
Additional improvement proposal for wasmer API & cli command options:
* add an API `compile_without_validation`.
* add a flag option in wasmer binary to deactivate `pre_validation` for all backend.
# Review
- [x] Add a short description of the the change to the CHANGELOG.md file
Co-authored-by: Patrick Ventuzelo <ventuzelo.patrick@gmail.com>
1002: Update the LLVM pass list. r=nlewycky a=nlewycky
# Description
Adds optimizations of loops, and inlinling and some simple interprocedural optimization.
Measured on the libsodium benchmarks, the new pass pipeline is a 2.35% geomean improvement. No major performance regressions known.
Co-authored-by: Nick Lewycky <nick@wasmer.io>
1009: Fix LLVM crash on `return` with a float on the stack and a crash with gas metering. r=nlewycky a=nlewycky
# Description
This PR fixes some matters of testing. The return+float bug is caught by the LLVM verifier, which was accidentally unconditionally disabled even when running "make test check". Add a new cargo feature named "test" which is enabled by all the test crates. When the "test" feature is passed to the llvm-backend, run the LLVM verifier.
Add a new crate `llvm-backend-test` which has one test so far, to ensure that this bug with the return+float does not crash LLVM. Include this new crate in runs of `make llvm`.
Fix LLVM crash on `return` when there is a float with pending NaN canonicalization on the stack.
Now that we run the verifier in testing, we discover another bug where `internal_field()` incorrectly labels a GEP with a TBAA label instead of the intended load. Fix this by labeling the correct instruction. No new test is introduced since this is already caught with `make bench-llvm`.
# Review
- [ ] Add a short description of the the change to the CHANGELOG.md file
Co-authored-by: Nick Lewycky <nick@wasmer.io>
713: Add AArch64 support for singlepass. r=syrusakbary a=losfair
This PR includes:
- Support for AArch64 (ARM64) in Singlepass backend. Implemented with a combination of x86_64 instruction translation and native code generation.
- State tracing and backtraces on AArch64/Singlepass. (Tiering is not implemented for this pair because there's no other backend supporting AArch64 yet)
- Debugging tools: `BlockTrace` middleware, and support for reading states of previous WebAssembly stack frames from middlewares.
Co-authored-by: losfair <zhy20000919@hotmail.com>
997: Compiler tests r=MarkMcCaskey a=syrusakbary
# Description
This PR simplifies the testing part, making `wasmer_runtime_core` the one responsible for picking the default backend.
This makes tests much simpler, and reliable when trying to run in different backends (this PR solves an issue when trying to run wasitests on ARM machines).
<!--
Provide details regarding the change including motivation,
links to related issues, and the context of the PR.
-->
# Review
- [x] Add a short description of the the change to the CHANGELOG.md file
Co-authored-by: Syrus <me@syrusakbary.com>
Co-authored-by: Mark McCaskey <mark@wasmer.io>
Co-authored-by: Mark McCaskey <5770194+markmccaskey@users.noreply.github.com>
