If you aren't expecting `VMExternRef`'s pointer-equality semantics, then these
trait implementations can be foot guns. Instead of implementing the trait, make
free functions in the `VMExternRef` namespace. This way, callers have to be a
little more explicit.
This is enough to get an `externref -> externref` identity function
passing.
However, `externref`s that are dropped by compiled Wasm code are (safely)
leaked. Follow up work will leverage cranelift's stack maps to resolve this
issue.
In the `ModuleEnvironment::declare_signature` callback, also pass the original
Wasm function signature, so that consumers may associate this information with
each compiled function. This is often necessary because while each Wasm
signature gets compiled down into a single native signature, multiple Wasm
signatures might compile down into the same native signature, and in these cases
the original Wasm signature is required for dynamic type checking of calls.
`VMExternRef` is a reference-counted box for any kind of data that is
external and opaque to running Wasm. Sometimes it might hold a Wasmtime
thing, other times it might hold something from a Wasmtime embedder and is
opaque even to us. It is morally equivalent to `Rc<dyn Any>` in Rust, but
additionally always fits in a pointer-sized word. `VMExternRef` is
non-nullable, but `Option<VMExternRef>` is a null pointer.
The one part of `VMExternRef` that can't ever be opaque to us is the
reference count. Even when we don't know what's inside an `VMExternRef`, we
need to be able to manipulate its reference count as we add and remove
references to it. And we need to do this from compiled Wasm code, so it must
be `repr(C)`!
`VMExternRef` itself is just a pointer to an `VMExternData`, which holds the
opaque, boxed value, its reference count, and its vtable pointer.
The `VMExternData` struct is *preceded* by the dynamically-sized value boxed
up and referenced by one or more `VMExternRef`s:
```ignore
,-------------------------------------------------------.
| |
V |
+----------------------------+-----------+-----------+ |
| dynamically-sized value... | ref_count | value_ptr |---'
+----------------------------+-----------+-----------+
| VMExternData |
+-----------------------+
^
+-------------+ |
| VMExternRef |-------------------+
+-------------+ |
|
+-------------+ |
| VMExternRef |-------------------+
+-------------+ |
|
... ===
|
+-------------+ |
| VMExternRef |-------------------'
+-------------+
```
The `value_ptr` member always points backwards to the start of the
dynamically-sized value (which is also the start of the heap allocation for
this value-and-`VMExternData` pair). Because it is a `dyn` pointer, it is
fat, and also points to the value's `Any` vtable.
The boxed value and the `VMExternRef` footer are held a single heap
allocation. The layout described above is used to make satisfying the
value's alignment easy: we just need to ensure that the heap allocation used
to hold everything satisfies its alignment. It also ensures that we don't
need a ton of excess padding between the `VMExternData` and the value for
values with large alignment.
About half of the `FuncEnvironment::translate_table_*` methods were using the
`TableIndex` newtype, while the other half were using raw `u32`s. This commit
makes everything use `TableIndex`.
# Overview
This commit makes changes to the `wiggle::from_witx` procedural in order
to allow for escaping strict and reserved Rust keywords.
Additionally, this commit introduces the ability to use a `witx_literal`
field in the `{..}` object provided as an argument to
`wiggle::from_witx`. This field allows for witx documents to be provided
as inline string literals.
Documentation comments are added to the methods of
`wiggle_generate::names::Names` struct responsible for generating
`proc_macro2::Ident` words.
## Keyword Escaping
Today, an interface that includes witx identifiers that conflict with
with Rust syntax will cause the `from_witx` macro to panic at
compilation time.
Here is a small example (adapted from
`/crates/wiggle/tests/keywords.rs`) that demonstrates this issue:
```
;; Attempts to define a module `self`, containing a trait `Self`. Both
;; of these are reserved keywords, and will thus cause a compilation
;; error.
(module $self
(@interface func (export "betchya_cant_implement_this")
)
)
```
Building off of code that (as of `master` today)
[demonstrates a strategy][esc] for escaping keywords, we introduce an
internal `escaping` module to `generate/src/config.rs` that contains
code responsible for escaping Rust keywords in a generalized manner.
[esc]: 0dd77d36f8/crates/wiggle/generate/src/names.rs (L106)
Some code related to special cases, such as accounting for
[`errno::2big`][err] while generating names for enum variants, is moved
into this module as well.
[err]: https://github.com/WebAssembly/WASI/blob/master/phases/snapshot/docs.md#-errno-enumu16
As mentioned in the document comments of this diff, we do not include
weak keywords like `'static` or `union`. Their semantics do not impact
us in the same way from a code generation perspective.
## witx_literal
First, some background. Trait names, type names, and so on use a
camel-cased naming convention. As such, `Self` is the only keyword that
can potentially conflict with these identifiers. (See the [Rust
Reference][key] for a complete list of strict, reserved, and weak
keywords.)
When writing tests, this meant that many tests had to be outlined into
separate files, as items with the name `$self` could not be defined in
the same namespace. As such, it seemed like a worthwhile feature to
implement while the above work was being developed.
The most important function to note is the `load_document` inherent
method added to `WitxConf`, and that `WitxConf` is now an enum
containing either (a) a collection of paths, identical to its current
functionality, or (b) a single string literal.
Note that a witx document given to `from_witx` using a string literal
provided to `from_witx` cannot include `use (..)` directives, per
the `witx::parse` documentation.
(See: https://docs.rs/witx/0.8.5/witx/fn.parse.html)
Two newtypes, `Paths` and `Literal`, are introduced to facilitate the
parsing of `WitxConf` values. Their public API and trait implementations
has been kept to the minimum required to satisfy compilation in order to
limit the scope of this diff. Additional surface for external consumers
can be added in follow-up commits if deemed necessary in review.
* Minor code tidying.
* Document that `Linker::iter`'s iteration order is arbitrary.
* Add a few more tests for `wasmtime::Linker`.
* Refactor `Linker::compute_imports`.
- Extract the error message generation into a separate function.
- In the error message, sort the candidates.
* Fix a typo in a comment.
* Add `__rtti_base` to the list of allowed but deprecated exports.
* Don't print an Error message when a program exits normally.
* Update comments to reflect the current code.
* Also allow "table" as an exported table, which is used by AssemblyScript.
* Reactor support.
This implements the new WASI ABI described here:
https://github.com/WebAssembly/WASI/blob/master/design/application-abi.md
It adds APIs to `Instance` and `Linker` with support for running
WASI programs, and also simplifies the process of instantiating
WASI API modules.
This currently only includes Rust API support.
* Add comments and fix a typo in a comment.
* Fix a rustdoc warning.
* Tidy an unneeded `mut`.
* Factor out instance initialization with `NewInstance`.
This also separates instantiation from initialization in a manner
similar to https://github.com/bytecodealliance/lucet/pull/506.
* Update fuzzing oracles for the API changes.
* Remove `wasi_linker` and clarify that Commands/Reactors aren't connected to WASI.
* Move Command/Reactor semantics into the Linker.
* C API support.
* Fix fuzzer build.
* Update usage syntax from "::" to "=".
* Remove `NewInstance` and `start()`.
* Elaborate on Commands and Reactors and add a spec link.
* Add more comments.
* Fix wat syntax.
* Fix wat.
* Use the `Debug` formatter to format an anyhow::Error.
* Fix wat.
The documentation for RtlNtStatusToDosError explicitly tells to call
it via GetProcAddress. The documentation for NtQueryInformationFile
does not, but similar considerations apply because there is normally
no import library for ntdll.
The main reason to use GetProcAddress though is because MinGW does
include an import library for ntdll, this import library contains
a definition of setjmp, and because of the way rustc orders linker
arguments, this definition of setjmp conflicts with and gets picked
over the one in msvcrt. Using setjmp from ntdll is undesirable as it
is an undocumented API and it is missing from wine, making it harder
to develop wasmtime using a cross-compiler.
Fixes#1738.
* Revert fstatat on *nix and test symlinks in path_filestat calls
This commit effectively reverts too eager refactoring on my part which
resulted in incorrect `path_filestat_{get, set_times}` behaviour on
*nix hosts. In the presence of symlinks, neither of the calls would
work properly.
In order to shield ourselves from similar errors in the future, I've
augmented the `path_filestat` test cases with symlink checks as well.
* Pass appropriate flags to fstatat and utimensat
* Fix formatting
* Fix Windows build
* Expand final symlinks if follow is set on Windows
* Fix formatting
* Do not follow symlinks unless specified on Windows
* Update comments and restart CI
* Skip testing volatile atim field
This patch includes:
- A complete rework of the way that CLIF blocks and edge blocks are
lowered into VCode blocks. The new mechanism in `BlockLoweringOrder`
computes RPO over the CFG, but with a twist: it merges edge blocks intto
heads or tails of original CLIF blocks wherever possible, and it does
this without ever actually materializing the full nodes-plus-edges
graph first. The backend driver lowers blocks in final order so
there's no need to reshuffle later.
- A new `MachBuffer` that replaces the `MachSection`. This is a special
version of a code-sink that is far more than a humble `Vec<u8>`. In
particular, it keeps a record of label definitions and label uses,
with a machine-pluggable `LabelUse` trait that defines various types
of fixups (basically internal relocations).
Importantly, it implements some simple peephole-style branch rewrites
*inline in the emission pass*, without any separate traversals over
the code to use fallthroughs, swap taken/not-taken arms, etc. It
tracks branches at the tail of the buffer and can (i) remove blocks
that are just unconditional branches (by redirecting the label), (ii)
understand a conditional/unconditional pair and swap the conditional
polarity when it's helpful; and (iii) remove branches that branch to
the fallthrough PC.
The `MachBuffer` also implements branch-island support. On
architectures like AArch64, this is needed to allow conditional
branches within plausibly-attainable ranges (+/- 1MB on AArch64
specifically). It also does this inline while streaming through the
emission, without any sort of fixpoint algorithm or later moving of
code, by simply tracking outstanding references and "deadlines" and
emitting an island just-in-time when we're in danger of going out of
range.
- A rework of the instruction selector driver. This is largely following
the same algorithm as before, but is cleaned up significantly, in
particular in the API: the machine backend can ask for an input arg
and get any of three forms (constant, register, producing
instruction), indicating it needs the register or can merge the
constant or producing instruction as appropriate. This new driver
takes special care to emit constants right at use-sites (and at phi
inputs), minimizing their live-ranges, and also special-cases the
"pinned register" to avoid superfluous moves.
Overall, on `bz2.wasm`, the results are:
wasmtime full run (compile + runtime) of bz2:
baseline: 9774M insns, 9742M cycles, 3.918s
w/ changes: 7012M insns, 6888M cycles, 2.958s (24.5% faster, 28.3% fewer insns)
clif-util wasm compile bz2:
baseline: 2633M insns, 3278M cycles, 1.034s
w/ changes: 2366M insns, 2920M cycles, 0.923s (10.7% faster, 10.1% fewer insns)
All numbers are averages of two runs on an Ampere eMAG.
This commit adds a suite of `wasmtime_funcref_table_*` APIs which mirror
the standard APIs but have a few differences:
* More errors are returned. For example error messages are communicated
through `wasmtime_error_t` and out-of-bounds vs load of null can be
differentiated in the `get` API.
* APIs take `wasm_func_t` instead of `wasm_ref_t`. Given the recent
decision to remove subtyping from the anyref proposal it's not clear
how the C API for tables will be affected, so for now these APIs are
all specialized to only funcref tables.
* Growth now allows access to the previous size of the table, if
desired, which mirrors the `table.grow` instruction.
This was originally motivated by bytecodealliance/wasmtime-go#5 where
the current APIs we have for working with tables don't quite work. We
don't have a great way to take an anyref constructed from a `Func` and
get the `Func` back out, so for now this sidesteps those concerns while
we sort out the anyref story.
It's intended that once the anyref story has settled and the official C
API has updated we'll likely delete these wasmtime-specific APIs or
implement them as trivial wrappers around the official ones.
