This adds benchmarks around module instantiation using criterion.
Both the default (i.e. on-demand) and pooling allocators are tested
sequentially and in parallel using a thread pool.
Instantiation is tested with an empty module, a module with a single page
linear memory, a larger linear memory with a data initializer, and a "hello
world" Rust WASI program.
On s390x, SIGILL and SIGFPE are delivered with the PSW address
pointing *after* the faulting instruction, while SIGSEGV and
SIGBUS are delivered with the PSW address pointing *to* the
faulting instruction. In order to support this, the common
code trap handler has to distinguish between those cases.
Also, enable SIGFPE on s390x (just like on x86).
* Bring back per-thread lazy initialization
Platforms Wasmtime supports may have per-thread initialization that
needs to run before WebAssembly. For example Unix needs to setup a
sigaltstack and macOS needs to set up mach ports. In #2757 this
per-thread setup was moved out of the invocation of a wasm function,
relying on the lack of Send for Store to initialize the thread at Store
creation time and never worry about it later.
This conflicted with [wasmtime's desired multithreading
story](https://github.com/bytecodealliance/wasmtime/pull/2812) so a new
[`Store::notify_switched_thread` was
added](https://github.com/bytecodealliance/wasmtime/pull/2822) to
explicitly indicate a Store has moved to another thread (if it unsafely
did so).
It turns out though that it's not always easy to determine when a
`Store` moves to a new thread. For example the Go bindings for Wasmtime
are generally unaware when a goroutine switches OS threads. This led to
https://github.com/bytecodealliance/wasmtime-go/issues/74 where a SIGILL
was left uncaught, making it appear that traps aren't working properly.
This commit revisits the decision in #2757 and moves per-thread
initialization back into the path of calling into WebAssembly. This is
differently from before, though, where there's still only one TLS access
on the path of calling into WebAssembly, unlike before where it was a
separate access. This allows us to get the speed benefits of #2757 as
well as the flexibility benefits of not having to explicitly move a
store between threads.
With this new ability this commit deletes the recently added
`Store::notify_switched_thread` method since it's no longer necessary.
* Fix a test compiling
* Optimize `table.init` instruction and instantiation
This commit optimizes table initialization as part of instance
instantiation and also applies the same optimization to the `table.init`
instruction. One part of this commit is to remove some preexisting
duplication between instance instantiation and the `table.init`
instruction itself, after this the actual implementation of `table.init`
is optimized to effectively have fewer bounds checks in fewer places and
have a much tighter loop for instantiation.
A big fallout from this change is that memory/table initializer offsets
are now stored as `u32` instead of `usize` to remove a few casts in a
few places. This ended up requiring moving some overflow checks that
happened in parsing to later in code itself because otherwise the wrong
spec test errors are emitted during testing. I've tried to trace where
these can possibly overflow but I think that I managed to get
everything.
In a local synthetic test where an empty module with a single 80,000
element initializer this improves total instantiation time by 4x (562us
=> 141us)
* Review comments
* Add resource limiting to the Wasmtime API.
This commit adds a `ResourceLimiter` trait to the Wasmtime API.
When used in conjunction with `Store::new_with_limiter`, this can be used to
monitor and prevent WebAssembly code from growing linear memories and tables.
This is particularly useful when hosts need to take into account host resource
usage to determine if WebAssembly code can consume more resources.
A simple `StaticResourceLimiter` is also included with these changes that will
simply limit the size of linear memories or tables for all instances created in
the store based on static values.
* Code review feedback.
* Implemented `StoreLimits` and `StoreLimitsBuilder`.
* Moved `max_instances`, `max_memories`, `max_tables` out of `Config` and into
`StoreLimits`.
* Moved storage of the limiter in the runtime into `Memory` and `Table`.
* Made `InstanceAllocationRequest` use a reference to the limiter.
* Updated docs.
* Made `ResourceLimiterProxy` generic to remove a level of indirection.
* Fixed the limiter not being used for `wasmtime::Memory` and
`wasmtime::Table`.
* Code review feedback and bug fix.
* `Memory::new` now returns `Result<Self>` so that an error can be returned if
the initial requested memory exceeds any limits placed on the store.
* Changed an `Arc` to `Rc` as the `Arc` wasn't necessary.
* Removed `Store` from the `ResourceLimiter` callbacks. Custom resource limiter
implementations are free to capture any context they want, so no need to
unnecessarily store a weak reference to `Store` from the proxy type.
* Fixed a bug in the pooling instance allocator where an instance would be
leaked from the pool. Previously, this would only have happened if the OS was
unable to make the necessary linear memory available for the instance. With
these changes, however, the instance might not be created due to limits
placed on the store. We now properly deallocate the instance on error.
* Added more tests, including one that covers the fix mentioned above.
* Code review feedback.
* Add another memory to `test_pooling_allocator_initial_limits_exceeded` to
ensure a partially created instance is successfully deallocated.
* Update some doc comments for better documentation of `Store` and
`ResourceLimiter`.
This commit uses a two-phase lookup of stack map information from modules
rather than giving back raw pointers to stack maps.
First the runtime looks up information about a module from a pc value, which
returns an `Arc` it keeps a reference on while completing the stack map lookup.
Second it then queries the module information for the stack map from a pc
value, getting a reference to the stack map (which is now safe because of the
`Arc` held by the runtime).
* Make `FunctionInfo` public and `CompiledModule::func_info` return it.
* Make the `StackMapLookup` trait unsafe.
* Add comments for the purpose of `EngineHostFuncs`.
* Rework ownership model of shared signatures: `SignatureCollection` in
conjunction with `SignatureRegistry` is now used so that the `Engine`,
`Store`, and `Module` don't need to worry about unregistering shared
signatures.
* Implement `Func::param_arity` and `Func::result_arity` in terms of
`Func::ty`.
* Make looking up a trampoline with the module registry more efficient by doing
a binary search on the function's starting PC value for the owning module and
then looking up the trampoline with only that module.
* Remove reference to the shared signatures from `GlobalRegisteredModule`.
This commit removes the stack map registry and instead uses the existing
information from the store's module registry to lookup stack maps.
A trait is now used to pass the lookup context to the runtime, implemented by
`Store` to do the lookup.
With this change, module registration in `Store` is now entirely limited to
inserting the module into the module registry.
This commit moves the shared signature registry out of `Store` and into
`Engine`.
This helps eliminate work that was performed whenever a `Module` was
instantiated into a `Store`.
Now a `Module` is registered with the shared signature registry upon creation,
storing the mapping from the module's signature index space to the shared index
space.
This also refactors the "frame info" registry into a general purpose "module
registry" that is used to look up trap information, signature information, and
(soon) stack map information.
This commit is intended to be a perf improvement for instantiation of
modules with lots of functions. Previously the `lookup_shared_signature`
callback was showing up quite high in profiles as part of instantiation.
As some background, this callback is used to translate from a module's
`SignatureIndex` to a `VMSharedSignatureIndex` which the instance
stores. This callback is called for two reasons, one is to translate all
of the module's own types into `VMSharedSignatureIndex` for the purposes
of `call_indirect` (the translation of that loads from this table to
compare indices). The second reason is that a `VMCallerCheckedAnyfunc`
is prepared for all functions and this embeds a `VMSharedSignatureIndex`
inside of it.
The slow part today is that the lookup callback was called
once-per-function and each lookup involved hashing a full
`WasmFuncType`. Albeit our hash algorithm is still Rust's default
SipHash algorithm which is quite slow, but we also shouldn't need to
re-hash each signature if we see it multiple times anyway.
The fix applied in this commit is to change this lookup callback to an
`enum` where one variant is that there's a table to lookup from. This
table is a `PrimaryMap` which means that lookup is quite fast. The only
thing we need to do is to prepare the table ahead of time. Currently
this happens on the instantiation path because in my measurments the
creation of the table is quite fast compared to the rest of
instantiation. If this becomes an issue, though, we can look into
creating the table as part of `SigRegistry::register_module` and caching
it somewhere (I'm not entirely sure where but I'm sure we can figure it
out).
There's in generally not a ton of efficiency around the `SigRegistry`
type. I'm hoping though that this fixes the next-lowest-hanging-fruit in
terms of performance without complicating the implementation too much. I
tried a few variants and this change seemed like the best balance
between simplicity and still a nice performance gain.
Locally I measured an improvement in instantiation time for a large-ish
module by reducing the time from ~3ms to ~2.6ms per instance.
This commit updates the implementation of `VMOffsets` to frontload all
checked arithmetic on construction of the `VMOffsets` which allows
eliding all checked arithmetic when accessing the fields of `VMOffsets`.
For testing and such this adds a new constructor as well from a new
`VMOffsetsFields` structure which is a clone of the old definition.
This should help speed up some profile hot spots I've been seeing where
with all the checked arithmetic on field sizes this was slowing down the
various accessors during instantiation (which uses `VMOffsets` to
initialize various fields of the `VMContext`).
Looking at some profiles these or their related functions were all
showing up, so this commit adds `#[inline]` to allow cross-crate
inlining by default.
- some tests don't pass because of bad interactions with the system's
libunwind; ignore them for now.
- the page size on mac aarch64 is 16K, not 4K; tweak some tests which
were expecting 4K or multiples of 4K pages to use a multiple of host page size
instead.
- a cranelift-native test needed an update for the new calling convention.
This commit fixes a bug where the wrong destination range was used when copying
data from the module's memory initialization upon instance initialization.
This affects the on-demand allocator only when using the `uffd` feature on
Linux and when the Wasm page being initialized is not the last in the module's
initial pages.
Fixes#2784.
* Use stable Rust on CI to test the x64 backend
This commit leverages the newly-released 1.51.0 compiler to test the
new backend on Windows and Linux with a stable compiler instead of a
nightly compiler. This isolates the nightly build to just the nightly
documentation generation and fuzzing, both of which rely on nightly for
the best results right now.
* Use updated stable in book build job
* Run rustfmt for new stable
* Silence new warnings for wasi-nn
* Allow some dead code in the x64 backend
Looks like new rustc is better about emitting some dead-code warnings
* Update rust in peepmatic job
* Fix a test in the pooling allocator
* Remove `package.metdata.docs.rs` temporarily
Needs resolution of https://github.com/rust-lang/cargo/pull/9300 first
* Fix a warning in a wasi-nn example
* Combine stack-based cleanups for faster wasm calls
This commit is an extension of #2757 where the goal is to optimize entry
into WebAssembly. Currently wasmtime has two stack-based cleanups when
entering wasm, one for the externref activation table and another for
stack limits getting reset. This commit fuses these two cleanups
together into one and moves some code around which enables less captures
for fewer closures and such to speed up calls in to wasm a bit more.
Overall this drops the execution time from 88ns to 80ns locally for me.
This also updates the atomic orderings when updating the stack limit
from `SeqCst` to `Relaxed`. While `SeqCst` is a reasonable starting
point the usage here should be safe to use `Relaxed` since we're not
using the atomics to actually protect any memory, it's simply receiving
signals from other threads.
* Determine whether a pc is wasm via a global map
The macOS implementation of traps recently changed to using mach ports
for handlers instead of signal handlers. This means that a previously
relied upon invariant, each thread fixes its own trap, was broken. The
macOS implementation worked around this by maintaining a global map from
thread id to thread local information, however, to solve the problem.
This global map is quite slow though. It involves taking a lock and
updating a hash map on all calls into WebAssembly. In my local testing
this accounts for >70% of the overhead of calling into WebAssembly on
macOS. Naturally it'd be great to remove this!
This commit fixes this issue and removes the global lock/map that is
updated on all calls into WebAssembly. The fix is to maintain a global
map of wasm modules and their trap addresses in the `wasmtime` crate.
Doing so is relatively simple since we're already tracking this
information at the `Store` level.
Once we've got a global map then the macOS implementation can use this
from a foreign thread and everything works out.
Locally this brings the overhead, on macOS specifically, of calling into
wasm from 80ns to ~20ns.
* Fix compiles
* Review comments
This commit implements a few optimizations, mainly inlining, that should
improve the performance of calling a WebAssembly function. This code
path can be quite hot depending on the embedding case and we hadn't
really put much effort into optimizing the nitty gritty.
The predominant optimization here is adding `#[inline]` to trivial
functions so performance is improved without having to compile with LTO.
Another optimization is to call `lazy_per_thread_init` when traps are
initialized per-thread (when a `Store` is created) rather than each time
a function is called. The next optimization is to change the unwind
reason in the `CallThreadState` to `MaybeUninit` to avoid extra checks
in the default case about whether we need to drop its variants (since in
the happy path we never need to drop it). The final optimization is to
optimize out a few checks when `async` support is disabled for a small
speed boost.
In a small benchmark where wasmtime calls a simple wasm function my
macOS computer dropped from 110ns to 86ns overhead, a 20% decrease. The
macOS overhead is still largely dominated by the global lock acquisition
and hash table management for traps right now, but I suspect the Linux
overhead is much better (should be on the order of ~30 or so ns).
We still have a long way to go to compete with SpiderMonkey which, in
testing, seem to have ~6ns overhead in calling the same wasm function on
my computer.
* Add assert to `StackPool::deallocate` to ensure the fiber stack given to it
comes from the pool.
* Remove outdated comment about windows and stacks as the allocator now returns
fiber stacks.
* Remove conditional compilation around `stack_size` in the allocators as it
was just clutter.
This commit splits out a `FiberStack` from `Fiber`, allowing the instance
allocator trait to return `FiberStack` rather than raw stack pointers. This
keeps the stack creation mostly in `wasmtime_fiber`, but now the on-demand
instance allocator can make use of it.
The instance allocators no longer have to return a "not supported" error to
indicate that the store should allocate its own fiber stack.
This includes a bunch of cleanup in the instance allocator to scope stacks to
the new "async" feature in the runtime.
Closes#2708.
* Switch macOS to using mach ports for trap handling
This commit moves macOS to using mach ports instead of signals for
handling traps. The motivation for this is listed in #2456, namely that
once mach ports are used in a process that means traditional UNIX signal
handlers won't get used. This means that if Wasmtime is integrated with
Breakpad, for example, then Wasmtime's trap handler never fires and
traps don't work.
The `traphandlers` module is refactored as part of this commit to split
the platform-specific bits into their own files (it was growing quite a
lot for one inline `cfg_if!`). The `unix.rs` and `windows.rs` files
remain the same as they were before with a few minor tweaks for some
refactored interfaces. The `macos.rs` file is brand new and lifts almost
its entire implementation from SpiderMonkey, adapted for Wasmtime
though.
The main gotcha with mach ports is that a separate thread is what
services the exception. Some unsafe magic allows this separate thread to
read non-`Send` and temporary state from other threads, but is hoped to
be safe in this context. The unfortunate downside is that calling wasm
on macOS now involves taking a global lock and modifying a global hash
map twice-per-call. I'm not entirely sure how to get out of this cost
for now, but hopefully for any embeddings on macOS it's not the end of
the world.
Closes#2456
* Add a sketch of arm64 apple support
* store: maintain CallThreadState mapping when switching fibers
* cranelift/aarch64: generate unwind directives to disable pointer auth
Aarch64 post ARMv8.3 has a feature called pointer authentication,
designed to fight ROP/JOP attacks: some pointers may be signed using new
instructions, adding payloads to the high (previously unused) bits of
the pointers. More on this here: https://lwn.net/Articles/718888/
Unwinders on aarch64 need to know if some pointers contained on the call
frame contain an authentication code or not, to be able to properly
authenticate them or use them directly. Since native code may have
enabled it by default (as is the case on the Mac M1), and the default is
that this configuration value is inherited, we need to explicitly
disable it, for the only kind of supported pointers (return addresses).
To do so, we set the value of a non-existing dwarf pseudo register (34)
to 0, as documented in
https://github.com/ARM-software/abi-aa/blob/master/aadwarf64/aadwarf64.rst#note-8.
This is done at the function granularity, in the spirit of Cranelift
compilation model. Alternatively, a single directive could be generated
in the CIE, generating less information per module.
* Make exception handling work on Mac aarch64 too
* fibers: use a breakpoint instruction after the final call in wasmtime_fiber_start
Co-authored-by: Alex Crichton <alex@alexcrichton.com>
This commit fixes a few issues around managing the thread-local state of
a wasmtime thread. We intentionally only have a singular TLS variable in
the whole world, and the problem is that when stack-switching off an
async thread we were not restoring the previous TLS state. This is
necessary in two cases:
* Futures aren't guaranteed to be polled/completed in a stack-like
fashion. If a poll sees that a future isn't ready then we may resume
execution in a previous wasm context that ends up needing the TLS
information.
* Futures can also cross threads (when the whole store crosses threads)
and we need to save/restore TLS state from the thread we're coming
from and the thread that we're going to.
The stack switching issue necessitates some more glue around suspension
and resumption of a stack to ensure we save/restore the TLS state on
both sides. The thread issue, however, also necessitates that we use
`#[inline(never)]` on TLS access functions and never have TLS borrows
live across a function which could result in running arbitrary code (as
was the case for the `tls::set` function.
* Implement defining host functions at the Config level.
This commit introduces defining host functions at the `Config` rather than with
`Func` tied to a `Store`.
The intention here is to enable a host to define all of the functions once
with a `Config` and then use a `Linker` (or directly with
`Store::get_host_func`) to use the functions when instantiating a module.
This should help improve the performance of use cases where a `Store` is
short-lived and redefining the functions at every module instantiation is a
noticeable performance hit.
This commit adds `add_to_config` to the code generation for Wasmtime's `Wasi`
type.
The new method adds the WASI functions to the given config as host functions.
This commit adds context functions to `Store`: `get` to get a context of a
particular type and `set` to set the context on the store.
For safety, `set` cannot replace an existing context value of the same type.
`Wasi::set_context` was added to set the WASI context for a `Store` when using
`Wasi::add_to_config`.
* Add `Config::define_host_func_async`.
* Make config "async" rather than store.
This commit moves the concept of "async-ness" to `Config` rather than `Store`.
Note: this is a breaking API change for anyone that's already adopted the new
async support in Wasmtime.
Now `Config::new_async` is used to create an "async" config and any `Store`
associated with that config is inherently "async".
This is needed for async shared host functions to have some sanity check during their
execution (async host functions, like "async" `Func`, need to be called with
the "async" variants).
* Update async function tests to smoke async shared host functions.
This commit updates the async function tests to also smoke the shared host
functions, plus `Func::wrap0_async`.
This also changes the "wrap async" method names on `Config` to
`wrap$N_host_func_async` to slightly better match what is on `Func`.
* Move the instance allocator into `Engine`.
This commit moves the instantiated instance allocator from `Config` into
`Engine`.
This makes certain settings in `Config` no longer order-dependent, which is how
`Config` should ideally be.
This also removes the confusing concept of the "default" instance allocator,
instead opting to construct the on-demand instance allocator when needed.
This does alter the semantics of the instance allocator as now each `Engine`
gets its own instance allocator rather than sharing a single one between all
engines created from a configuration.
* Make `Engine::new` return `Result`.
This is a breaking API change for anyone using `Engine::new`.
As creating the pooling instance allocator may fail (likely cause is not enough
memory for the provided limits), instead of panicking when creating an
`Engine`, `Engine::new` now returns a `Result`.
* Remove `Config::new_async`.
This commit removes `Config::new_async` in favor of treating "async support" as
any other setting on `Config`.
The setting is `Config::async_support`.
* Remove order dependency when defining async host functions in `Config`.
This commit removes the order dependency where async support must be enabled on
the `Config` prior to defining async host functions.
The check is now delayed to when an `Engine` is created from the config.
* Update WASI example to use shared `Wasi::add_to_config`.
This commit updates the WASI example to use `Wasi::add_to_config`.
As only a single store and instance are used in the example, it has no semantic
difference from the previous example, but the intention is to steer users
towards defining WASI on the config and only using `Wasi::add_to_linker` when
more explicit scoping of the WASI context is required.
This change makes the storage of `Table` more internally consistent.
Elements are stored as raw pointers for both static and dynamic table storage.
Explicitly storing elements as pointers removes assumptions being made by the
pooling allocator in terms of the size and default representation of the
elements.
However, care must be made to properly clone externrefs for table operations.
* Add more overflow checks in table/memory initialization.
* Comment for `with_allocation_strategy` to explain ignored `Config` options.
* Fix Wasmtime `Table` to not panic for type mismatches in `fill`/`copy`.
* Add tests for that fix.
* More use of `anyhow`.
* Change `make_accessible` into `protect_linear_memory` to better demonstrate
what it is used for; this will make the uffd implementation make a little
more sense.
* Remove `create_memory_map` in favor of just creating the `Mmap` instances in
the pooling allocator. This also removes the need for `MAP_NORESERVE` in the
uffd implementation.
* Moar comments.
* Remove `BasePointerIterator` in favor of `impl Iterator`.
* The uffd implementation now only monitors linear memory pages and will only
receive faults on pages that could potentially be accessible and never on a
statically known guard page.
* Stop allocating memory or table pools if the maximum limit of the memory or
table is 0.
* Add `anyhow` dependency to `wasmtime-runtime`.
* Revert `get_data` back to `fn`.
* Remove `DataInitializer` and box the data in `Module` translation instead.
* Improve comments on `MemoryInitialization`.
* Remove `MemoryInitialization::OutOfBounds` in favor of proper bulk memory
semantics.
* Use segmented memory initialization except for when the uffd feature is
enabled on Linux.
* Validate modules with the allocator after translation.
* Updated various functions in the runtime to return `anyhow::Result`.
* Use a slice when copying pages instead of `ptr::copy_nonoverlapping`.
* Remove unnecessary casts in `OnDemandAllocator::deallocate`.
* Better document the `uffd` feature.
* Use WebAssembly page-sized pages in the paged initialization.
* Remove the stack pool from the uffd handler and simply protect just the guard
pages.
Last minute code clean up to fix some comments and rename `address_space_size`
to `memory_reservation_size` to better describe what the option is doing.
This was originally written to support sourcing the table and memory
definitions differently for the pooling allocator.
However, both allocators do the exact same thing, so the closure arguments are
no longer necessary.
Additionally, this cleans up the code a bit to pass in the allocation request
rather than having individual parameters.
This commit implements copying paged initialization data upon a fault of a
linear memory page.
If the initialization data is "paged", then the appropriate pages are copied
into the Wasm page (or zeroed if the page is not present in the
initialization data).
If the initialization data is not "paged", the Wasm page is zeroed so that
module instantiation can initialize the pages.
As Windows uses the native fiber implementation, the stack tests should be
ignored on Windows as the implementation intentionally errors when handing out
stacks.
This commit implements the `uffd` feature which turns on support for utilizing
the `userfaultfd` system call on Linux for the pooling instance allocator.
By handling page faults in userland, we are able to detect guard page accesses
without having to constantly change memory page protections.
This should help reduce the number of syscalls as well as kernel lock
contentions when many threads are allocating and deallocating instances.
Additionally, the user fault handler can lazy initialize linear
memories of an instance (implementation to come).
This commit implements the pooling instance allocator.
The allocation strategy can be set with `Config::with_allocation_strategy`.
The pooling strategy uses the pooling instance allocator to preallocate a
contiguous region of memory for instantiating modules that adhere to various
limits.
The intention of the pooling instance allocator is to reserve as much of the
host address space needed for instantiating modules ahead of time and to reuse
committed memory pages wherever possible.
This commit implements allocating fiber stacks in an instance allocator.
The on-demand instance allocator doesn't support custom stacks, so the
implementation will use the allocation from `wasmtime-fiber` for the fiber
stacks.
In the future, the pooling instance allocator will return custom stacks to use
on Linux and macOS.
On Windows, the native fiber implementation will always be used.
This commit changes `Instance` such that memories can be stored statically,
with just a base pointer, size, maximum, and a callback to make memory
accessible.
Previously the memories were being stored as boxed trait objects, which would
require the pooling allocator to do some unpleasant things to avoid
allocations.
With this change, the pooling allocator can simply define a memory for the
instance without using a trait object.
