Patch adds support for the perf jitdump file specification.
With this patch it should be possible to see profile data for code
generated and maped at runtime. Specifically the patch adds support
for the JIT_CODE_LOAD and the JIT_DEBUG_INFO record as described in
the specification. Dumping jitfiles is enabled with the --jitdump
flag. When the -g flag is also used there is an attempt to dump file
and line number information where this option would be most useful
when the WASM file already includes DWARF debug information.
The generation of the jitdump files has been tested on only a few wasm
files. This patch is expected to be useful/serviceable where currently
there is no means for jit profiling, but future patches may benefit
line mapping and add support for additional jitdump record types.
Usage Example:
Record
sudo perf record -k 1 -e instructions:u target/debug/wasmtime -g
--jitdump test.wasm
Combine
sudo perf inject -v -j -i perf.data -o perf.jit.data
Report
sudo perf report -i perf.jit.data -F+period,srcline
Crates used exclusively at build time, such as proc-macro crates and
their dependencies, don't benefit substantially from optimization; they
take far longer to optimize than time saved when running them during the
build. No machine code from these crates will appear in the final
compiled binary.
Use the new profile-overrides mechanism in Rust 1.41
(https://doc.rust-lang.org/cargo/reference/profiles.html#overrides) to
build such crates with opt-level 0.
On a 4-thread laptop, this brings build time from 6m27s to 4m21s, and
CPU time spent from 23m43s to 16m23s.
This commit deletes the old C implementation of the original
`wasi_unstable` module, instead only leaving around our single
`wasmtime-wasi` crate as the implementation for both
`wasi_snapshot_preview1` and `wasi_unstable`.
This hasn't been discussed (AFAIK) up until now, so this is also a
proposal! Some thoughts in favor of this deletion I would have are:
* This has been off-by-default for ages
* We don't build or test any of this on CI
* Published binaries with `wasmtime` do not have this possibility
enabled
* Future refactorings to the `wasmtime-wasi` crate will either need to
work around how the C implementation is different or bring it up to
speed.
This is motivated by the last bullet point where I was working on
getting `wasmtime-wasi` working purely as an implementation detail on
top of the `wasmtime` crate itself, but quickly ran into a case where
the CLI would need to multiplex all sorts of wasi implementations. In
any case I'm curious what others think, is this too soon? Is there
something remaining blocking this? (etc)
Currently the wasmtime binary (src/bin/wasmtime) and the wasmtime API
crate (crates/api) share the same output filename. This causes the
output files of the wasmtime binary to clobber the output files of the
wasmtime API crate. So running `cargo doc --all` will often not build
the wasmtime API docs, which is the more useful of the two.
This is a rustdoc bug, and can be worked around by disabling
documentation for the wasmtime binary.
* Move the C API to a separate crate
This commit moves the C API from `crates/api/src/wasm.rs` to
`crates/capi/src/lib.rs` to be located in a separate crate. There's a
number of reasons for this:
* When a Rust program depends on the `wasmtime` crate, there's no need
to compile in the C API.
* This should improve compile times of the `wasmtime` crate since it's
not producing artifacts which aren't always used.
* The development of the C API can be guaranteed to only use the public
API of the `wasmtime` crate itself.
Some CI pieces are tweaked and this overall shouldn't have much impact
on users, it's intended that it's a cleanup/speedup for developers!
* Disable rustdoc/tests for capi
* Review feedback
* Add back in accidentally deleted comment
* More renamings
* Try to fix dotnet build
* Update to the latest spec_testsuite and dependencies.
Update to target-lexicon 0.10, cranelift 0.54, wast 0.6, faerie 0.14,
and the latest spec_testsuite.
For wast and cranelift-wasm, update the code for API changes.
* Factor out the code for matching f32, f64, and v128.
This takes the idea from #802 to split out `f32_matches`, `f64_matches`,
and `v128_matches` functions, which better factor out the matching
functionality between scalar and vector.
This commit refactors the Wasmtime CLI tools to use `structopt` instead of
`docopt`.
The `wasmtime` tool now has the following subcommands:
* `config new` - creates a new Wasmtime configuration file.
* `run` - runs a WebAssembly module.
* `wasm2obj` - translates a Wasm module to native object file.
* `wast` - runs a test script file.
If no subcommand is specified, the `run` subcommand is used. Thus,
`wasmtime foo.wasm` should continue to function as expected.
The `wasm2obj` and `wast` tools still exist, but delegate to the same
implementation as the `wasmtime` subcommands. The standalone `wasm2obj` and
`wast` tools may be removed in the future in favor of simply using `wasmtime`.
Included in this commit is a breaking change to the default Wasmtime
configuration file: it has been renamed from `wasmtime-cache-config.toml` to
simply `config.toml`. The new name is less specific which will allow for
additional (non-cache-related) settings in the future.
There are some breaking changes to improve command line UX:
* The `--cache-config` option has been renamed to `--config`.
* The `--create-config-file` option has moved to the `config new` subcommand.
As a result, the `wasm2obj` and `wast` tools cannot be used to create a new
config file.
* The short form of the `--optimize` option has changed from
`-o` to `-O` for consistency.
* The `wasm2obj` command takes the output object file as a
required positional argument rather than the former required output *option*
(e.g. `wasmtime wasm2obj foo.wasm foo.obj`).
* Remove the `Flags` type from `Config` API
This commit removes the final foreign type from the `Config` API in the
`wasmtime` crate. The cranelift `Flags` type is now expanded into
various options on the `Config` structure itself, all prefixed with
`cranelift_` since they're only relevant to the Cranelift backend. The
various changes here were:
* The `avoid_div_traps` feature is enabled by default since it seemed
that was done anywhere anyway.
* Enabling the wasm SIMD feature enables the requisite features in
Cranelift as well.
* A method for enabling the debug verifier has been added.
* A method for configuring the Cranelift optimization level, as well as
a corresponding enumeration, has been added.
* Assert that `Config` is both `Send` and `Sync`
* Remove usage of `CompilationStrategy` from `Config`
This commit removes the public API usage of the internal
`CompilationStrategy` enumeration from the `Config` type in the
`wasmtime` crate. To do this the `enum` was copied locally into the
crate and renamed `Strategy`. The high-level description of this change
is:
* The `Config::strategy` method now takes a locally-defined `Strategy`
enumeration instead of an internal type.
* The contents of `Strategy` are always the same, not relying on Cargo
features to indicate which variants are present. This avoids
unnecessary downstream `#[cfg]`.
* A `lightbeam` feature was added to the `wasmtime` crate itself to
lightbeam compilation support.
* The `Config::strategy` method is now fallible. It returns a runtime
error if support for the selected strategy wasn't compiled in.
* The `Strategy` enum is listed as `#[non_exhaustive]` so we can safely
add variants over time to it.
This reduces the public crate dependencies of the `wasmtime` crate
itself, removing the need to reach into internal crates even more!
cc #708
* Fix fuzz targets
* Update nightly used to build releases
* Run rustfmt
This crate is intended to hold all of our various test case generators and
oracles. The fuzz targets we have at `wasmtime/fuzz/fuzz_targets/*` will
eventually be ~one-liner glue code calling into this crate.
Part of #611
* Fix feature-gating of test-programs
This commit fixes bugs in enabling feature-gating of `test-programs`
which was introduced in #600. It turns out, #600 accidentally
disabled `test-programs` from ever running, and this commit fixes
that.
* Fix the CI
This commit removes the usage of the `failure` crate and finishes up the
final pieces of the migration to `std::error::Error` and `anyhow`. The
`faerie` crate was updated to pull in its migration from `failure` to
`anyhow` as well.
* Fix fuzz target compilation.
* Bump version to 0.7.0
* Temporarily disable fuzz tests
Temporarily disable fuzz tests until https://github.com/bytecodealliance/cranelift/issues/1216 is resolved.
* Fix publish-all.sh to not modify the witx crate.
* Remove the "publish = false" attribute from Lightbeam.
* Add a README.md for wasmtime-interface-types.
* Remove the "rust" category.
This fixes the following warning:
warning: the following are not valid category slugs and were ignored: rust. Please see https://crates.io/category_slugs for the list of all category slugs.
* Mark wasmtime-cli as "publish = false".
* Sort the publishing rules in topological order.
Also, publish nightly-only crates with cargo +nightly.
* Reorganize wasi-misc-tests.
Move wasi-misc-tests out of wasi-common, to break a dependency cycle;
previously, wasmtime-* depended on wasi-common, but wasi-common
dev-dependended on wasmtime-*.
Now, wasi-common no longer dev-depends on wasmtime-*; instead, the
tests are in their own crate which depends on wasi-common and on
wasmtime-*.
Also, rename wasi-misc-tests to wasi-tests for simplicity.
This also removes the "wasm_tests" feature; it's replaced by the
"test-programs" feature.
* Update the CI script to use the new feature name.
* Update the CI script to use the new feature name in one more place.
* Change a `write!` to a `writeln!`.
* General Cargo.toml cleanup.
- Remove travis-ci attributes.
- Remove "experimental" badges from actively-developed crates.
- Reflow some long lines.
- Use dependency features consistently.
- Add readme attributes
* Update WASI to the latest trunk.
This notably adds a .gitignore file for the WASI directory.
This commit simplifies the build script slightly for generating tests by
doing a few dull refactorings:
* Leaves formatting to `rustfmt`
* Extract bulk of code execution into a top-level shared `run_wast`
function so each test is a one-liner
* Use `anyhow` for errors both in the script and in tests
We needed a workaround when using wabt on old Linux distros due to using
old versions of `strtof`. With the switch to `wat`, we no longer need
these workarounds.
* Initial checkin.
* Update to rust-lang libc.
* Add a .gitignore file.
* Factor out functions for cleaning up files and directories.
* Fix a typo in a comment.
* Print a "Success!" message if all tests passed.
* Factor out code for creating directories.
* Add wrappers around WASI functions.
These wrappers handle converting from &str to pointer+length and handle
unsafe.
* More refactoring.
* Refactor a fd_close helper.
* Move utility functions into a separate file.
* cargo update
* Add a basic test for random_get.
* Test that directories aren't resizable.
* Test clearing __WASI_RIGHT_PATH_FILESTAT_SET_SIZE.
Ensure that clearing __WASI_RIGHT_PATH_FILESTAT_SET_SIZE succeeds before
testing file truncation.
* cargo update
* Modularise tests for easier use with wasi-common crate
* Add a Code of Conduct and CONTRIBUTING.md.
* Fix typo
* Add testcase for fd_allocate
* Add positive test for fd_renumber
* Assert bufused in readlink_no_buffer testcase
* Add positive readlink testcase
* Add testcase for fd_seek and fd_tell
* Add fd_p{read, write} test
* Add README
* Add cases with trailing slashes to interesting_paths
* Split nofollow_errors testcase into two
* nofollow_errors now operators on symlinks to existing resources
* dangling_symlink covers danling symlinks tests
* Factor out a `create_file` helper function.
* Switch from the error crate to `std::io::Error::last_os_error()`.
* Use `create_file` in the readlink test too.
* Add a test for fd_filestat_set_*
* Minor refactoring
Add missing cleanup_file calls to file_pread_pwrite and
file_seek_tell.
* Add testcase for unbuffered fd_write; fixes#11
* Add testcase for path_rename
* Use the wasi crate.
Switch from depending on libc to depending on the new wasi crate to provide
the low-level WASI interfaces.
See also https://github.com/rust-lang/libc/pull/1461.
* Add a test for path_filestat_*
* Add a test for fd_readdir
* Use expect instead of unwrap
* Add a check for ino.
* Fix the build
* Don't assume a specific order of dirents
* Better test
* Test cookie value
* Fix file types
* Fix the test
* Fix the test
* Fix the test
* Cleanup
* Minor formatting tidying in README.md.
* Fix miscellaneous clippy warnings.
* Rename the crate to wasi-misc-tests.
* Update to wasi 0.7.0.
This switches from using the libc wasi bindings to using the wasi
crate's bindings. This eliminates a git dependency on libc, updates
to the new-style bindings which use Result where possible, and treats
functions that operate on raw file descriptors as unsafe.
* Add various tests for trailing-slash behavior.
* Sync new testcases with latest upstream
* Fix path_filestat testcase
* Add smoke test for fd_advise
This test is a true smoke test as it only tests whether issuing
an advise call to the host's kernel doesn't yield an error. The
consequence of issuing such a syscall is not tested.
* Check if CLOCK_MONOTONIC is actually monotonic
* Refactor the inequality assertions for more debuggable errors.
* Bump libc from 0.2.62 to 0.2.65
Bumps [libc](https://github.com/rust-lang/libc) from 0.2.62 to 0.2.65.
- [Release notes](https://github.com/rust-lang/libc/releases)
- [Commits](https://github.com/rust-lang/libc/compare/0.2.62...0.2.65)
Signed-off-by: dependabot-preview[bot] <support@dependabot.com>
* Fix compilation error
* Enable Actions and add rust.yml (#35)
* Enable Actions and add rust.yml
This commit enables Github Actions and adds corresponding configuration in rust.yml file.
* Update rust.yml
* Fix formatting
* Add empty .rustfmt.toml config file
* Add badge to README
* Update README
* Clean up Github Actions and README
* Add test case for `poll_oneoff` syscall (#38)
* Add test case for `poll_oneoff` syscall
This commit adds a test case for `poll_oneoff` syscall. In particular,
it builds on the excellent test use case provided by @dunnock in their
repo [poll_oneoff_tests] (thanks!), and tests:
* simple timeout
* stdin read with timeout
* fd read and fd write polls
[poll_oneoff_tests]: https://github.com/dunnock/poll_oneoff_tests
* Apply suggestions and negative test for bad fd
Co-authored-by: Maxim Vorobjov <maxim.vorobjov@gmail.com>
* Add smoke test for STDOUT/ERR readwrite poll
* Add comment on stdin/out/err
* Add a test for `*at`-style functions returning `ENOTDIR` when `dirfd` is not a dir.
* Remove misc_testsuite submodule
* Add "publish=false" to Cargo.toml; remove LICENSE
* fix Linux `isatty` implementation
* defer `WasiCtxBuilder` errors to `build()`; don't change API yet
This changes the fields on the builder to types that let the various `.arg()`, `.env()`, etc methods
infallible, so we don't have to worry about handling any errors till we actually build. This reduces
line noise when using a builder in a downstream application.
Deferring the processing of the builder fields also has the advantage of eliminating the opening and
closing of `/dev/null` for the default stdio file descriptors unless they're actually used by the
resulting `WasiCtx`.
Unicode errors when inheriting arguments and environment variables no longer cause a panic, but
instead go through `OsString`. We return `ENOTCAPABLE` at the end if there are NULs, or if UTF-8
conversion fails on Windows.
This also changes the bounds on some of the methods from `AsRef<str>` to `AsRef<[u8]>`. This
shouldn't break any existing code, but allows more flexibility when providing arguments. Depending
on the outcome of https://github.com/WebAssembly/WASI/issues/8 we may eventually want to require
these bytes be UTF-8, so we might want to revisit this later.
Finally, this fixes a tiny bug that could arise if we had exactly the maximum number of file
descriptors when populating the preopens.
* make `WasiCtxBuilder` method types less restrictive
This is a separate commit, since it changes the interface that downstream clients have to use, and
therefore requires a different commit of `wasmtime` for testing. That `wasmtime` commit is currently
on my private fork, so this will need to be amended before merging.
Now that failures are deferred until `WasiCtxBuilder::build()`, we don't need to have `Result` types
on the other methods any longer.
Additionally, using `IntoIterator` rather than `Iterator` as the trait bound for these methods is
slightly more general, and saves the client some typing.
* enforce that arguments and environment variables are valid UTF-8
* remove now-unnecessary platform-specific OsString handling
* `ENOTCAPABLE` -> `EILSEQ` for failed arg/env string conversions
* fix up comment style
* Apply @acfoltzer's fix to isatty on Linux to BSD
This commit bumps Cranelift to 0.49.0 and target-lexicon to 0.9.0 to fix the
failing build to due an updated faerie crate that violated semver with an
updated 0.9.0 target-lexicon dependency.
Fixes#491.
* Implement clock_time_get on Windows.
Also update misc_testsuite to include latest clock_time_get test
changes.
* Try to somehow implement clock_res_get on Windows.
* Fix 55ms
* Cache the perf counter resolution
* Fix integration tests
* Bump crate versions.
* Update dependency version numbers too.
This is a follow-up to f96b6c9e72ae50fcddd488be47a2d4b5ac7b926b to
update the version numbers in the local dependencies too.
* Use generated type bindings.
Use the witx API descriptions to generate the bulk of the contents of
host.rs, wasi.rs, and wasi32.rs.
This also prunes out many of the miscellaneous libc definitions from
those files which aren't currently in use by wasi-common. If there's
anything removed that's still needed by someone, it's easy to add things
back in.
* Remove unneeded iovec conversion routines.
* Migrate from failure to thiserror and anyhow
The failure crate invents its own traits that don't use
std::error::Error (because failure predates certain features added to
Error); this prevents using ? on an error from failure in a function
using Error. The thiserror and anyhow crates integrate with the standard
Error trait instead.
This change does not attempt to semantically change or refactor the
approach to error-handling in any portion of the code, to ensure that
the change remains straightforward to review. Modules using specific
differentiated error types move from failure_derive and derive(Fail) to
thiserror and derive(Error). Modules boxing all errors opaquely move
from failure::Error to anyhow. Modules using String as an error type
continue to do so. Code using unwrap or expect continues to do so.
Drop Display implementations when thiserror can easily derive an
identical instance.
Drop manual traversal of iter_causes; anyhow's Debug instance prints the
chain of causes by default.
Use anyhow's type alias anyhow::Result<T> in place of
std::result::Result<T, anyhow::Error> whenever possible.
* wasm2obj: Simplify error handling using existing messages
handle_module in wasm2obj manually maps
cranelift_codegen::isa::LookupError values to strings, but LookupError
values already have strings that say almost exactly the same thing.
Rely on the strings from cranelift.
* wasmtime: Rely on question-mark-in-main
The main() wrapper around rmain() completely matches the behavior of
question-mark-in-main (print error to stderr and return 1), so switch to
question-mark-in-main.
* Update to walrus 0.13 and wasm-webidl-bindings 0.6
Both crates switched from failure to anyhow; updating lets us avoid a
translation from failure to anyhow within wasmtime-interface-types.
Since `wasmtime` now uses `wasi` and `wasi32` modules, we can now
safely remove the `wasm32` module. This commit also updates `wasmtime`
to the latest upstream.
This commit syncs `wasmtime-wasi` crate with the latest refactoring
applied to `wasi-common` crate. Namely, `wasm32` is replaced with
two modules: `wasi` and `wasi32`. This change can be tracked via
CraneStation/wasi-common#151.
* Reorganize host.rs and wasm32.rs.
Reorganize host.rs and wasm32.rs into host.rs, wasi.rs, and wasi32.rs.
Most of the contents of host.rs was not actually host-specific, as most
of the types are fixed-size types like u32 or i64. These types are now
in wasi.rs.
The few types which do have pointer or usize-sized values now remain,
in two versions: host.rs has versions which use actual raw pointers and
usize, and wasi32.rs has versions which use u32 to represent them.
* Fix compilation on BSD
* Fix compilation on Windows
* Fully encapsulate endianness in memory.rs.
This refactors memory.rs to fully encapsulte endianness concerns, so
that outside that file, all values are in host-endian order.
This adds a dependency on the `num` crate, though it's only used for
the `PrimInt` trait, for handling endianness in a generic way.
* Use pub(crate).
* Fix some Windows warnings.
* Implement clock_time_get on Windows.
Also update misc_testsuite to include latest clock_time_get test
changes.
* improve comments
* Remove a leftover import.
Co-Authored-By: Jakub Konka <kubkon@jakubkonka.com>
This PR updates `wasmtime_wasi` crate by adjusting `poll_oneoff`'s
signature to that introduced in `wasi_common` in
CraneStation/wasi-common#137. This change is required in order to
fix#440.
This commit syncs tests with latest wasmtime revision.
As such, it now utilises the `wasmtime-api` crate for
runtime setup.
Closes#126, #127, #128, #129.
Johnnie Birch
Yury Delendik
Josh Triplett
Alex Crichton
Andrew Chin
Dan Gohman
Jakub Konka
Peter Huene
Nick Fitzgerald
Jakub Konka
Adam C. Foltzer
Marcin Mielniczuk