* Pull `Module` out of `ModuleTextBuilder` This commit is the first in what will likely be a number towards preparing for serializing a compiled component to bytes, a precompiled artifact. To that end my rough plan is to merge all of the compiled artifacts for a component into one large object file instead of having lots of separate object files and lots of separate mmaps to manage. To that end I plan on eventually using `ModuleTextBuilder` to build one large text section for all core wasm modules and trampolines, meaning that `ModuleTextBuilder` is no longer specific to one module. I've extracted out functionality such as function name calculation as well as relocation resolving (now a closure passed in) in preparation for this. For now this just keeps tests passing, and the trajectory for this should become more clear over the following commits. * Remove component-specific object emission This commit removes the `ComponentCompiler::emit_obj` function in favor of `Compiler::emit_obj`, now renamed `append_code`. This involved significantly refactoring code emission to take a flat list of functions into `append_code` and the caller is responsible for weaving together various "families" of functions and un-weaving them afterwards. * Consolidate ELF parsing in `CodeMemory` This commit moves the ELF file parsing and section iteration from `CompiledModule` into `CodeMemory` so one location keeps track of section ranges and such. This is in preparation for sharing much of this code with components which needs all the same sections to get tracked but won't be using `CompiledModule`. A small side benefit from this is that the section parsing done in `CodeMemory` and `CompiledModule` is no longer duplicated. * Remove separately tracked traps in components Previously components would generate an "always trapping" function and the metadata around which pc was allowed to trap was handled manually for components. With recent refactorings the Wasmtime-standard trap section in object files is now being generated for components as well which means that can be reused instead of custom-tracking this metadata. This commit removes the manual tracking for the `always_trap` functions and plumbs the necessary bits around to make components look more like modules. * Remove a now-unnecessary `Arc` in `Module` Not expected to have any measurable impact on performance, but complexity-wise this should make it a bit easier to understand the internals since there's no longer any need to store this somewhere else than its owner's location. * Merge compilation artifacts of components This commit is a large refactoring of the component compilation process to produce a single artifact instead of multiple binary artifacts. The core wasm compilation process is refactored as well to share as much code as necessary with the component compilation process. This method of representing a compiled component necessitated a few medium-sized changes internally within Wasmtime: * A new data structure was created, `CodeObject`, which represents metadata about a single compiled artifact. This is then stored as an `Arc` within a component and a module. For `Module` this is always uniquely owned and represents a shuffling around of data from one owner to another. For a `Component`, however, this is shared amongst all loaded modules and the top-level component. * The "module registry" which is used for symbolicating backtraces and for trap information has been updated to account for a single region of loaded code holding possibly multiple modules. This involved adding a second-level `BTreeMap` for now. This will likely slow down instantiation slightly but if it poses an issue in the future this should be able to be represented with a more clever data structure. This commit additionally solves a number of longstanding issues with components such as compiling only one host-to-wasm trampoline per signature instead of possibly once-per-module. Additionally the `SignatureCollection` registration now happens once-per-component instead of once-per-module-within-a-component. * Fix compile errors from prior commits * Support AOT-compiling components This commit adds support for AOT-compiled components in the same manner as `Module`, specifically adding: * `Engine::precompile_component` * `Component::serialize` * `Component::deserialize` * `Component::deserialize_file` Internally the support for components looks quite similar to `Module`. All the prior commits to this made adding the support here (unsurprisingly) easy. Components are represented as a single object file as are modules, and the functions for each module are all piled into the same object file next to each other (as are areas such as data sections). Support was also added here to quickly differentiate compiled components vs compiled modules via the `e_flags` field in the ELF header. * Prevent serializing exported modules on components The current representation of a module within a component means that the implementation of `Module::serialize` will not work if the module is exported from a component. The reason for this is that `serialize` doesn't actually do anything and simply returns the underlying mmap as a list of bytes. The mmap, however, has `.wasmtime.info` describing component metadata as opposed to this module's metadata. While rewriting this section could be implemented it's not so easy to do so and is otherwise seen as not super important of a feature right now anyway. * Fix windows build * Fix an unused function warning * Update crates/environ/src/compilation.rs Co-authored-by: Nick Fitzgerald <fitzgen@gmail.com> Co-authored-by: Nick Fitzgerald <fitzgen@gmail.com>
222 lines
8.2 KiB
Rust
222 lines
8.2 KiB
Rust
//! Support for compiling with Cranelift.
|
|
//!
|
|
//! This crate provides an implementation of the `wasmtime_environ::Compiler`
|
|
//! and `wasmtime_environ::CompilerBuilder` traits.
|
|
|
|
use cranelift_codegen::binemit;
|
|
use cranelift_codegen::ir;
|
|
use cranelift_codegen::isa::{unwind::UnwindInfo, CallConv, TargetIsa};
|
|
use cranelift_entity::PrimaryMap;
|
|
use cranelift_wasm::{DefinedFuncIndex, FuncIndex, WasmFuncType, WasmType};
|
|
use target_lexicon::{Architecture, CallingConvention};
|
|
use wasmtime_environ::{
|
|
FilePos, InstructionAddressMap, ModuleTranslation, ModuleTypes, TrapInformation,
|
|
};
|
|
|
|
pub use builder::builder;
|
|
|
|
mod builder;
|
|
mod compiler;
|
|
mod debug;
|
|
mod func_environ;
|
|
mod obj;
|
|
|
|
type CompiledFunctions<'a> = PrimaryMap<DefinedFuncIndex, &'a CompiledFunction>;
|
|
|
|
/// Compiled function: machine code body, jump table offsets, and unwind information.
|
|
#[derive(Default)]
|
|
pub struct CompiledFunction {
|
|
/// The machine code for this function.
|
|
body: Vec<u8>,
|
|
|
|
/// The unwind information.
|
|
unwind_info: Option<UnwindInfo>,
|
|
|
|
/// Information used to translate from binary offsets back to the original
|
|
/// location found in the wasm input.
|
|
address_map: FunctionAddressMap,
|
|
|
|
/// Metadata about traps in this module, mapping code offsets to the trap
|
|
/// that they may cause.
|
|
traps: Vec<TrapInformation>,
|
|
|
|
relocations: Vec<Relocation>,
|
|
value_labels_ranges: cranelift_codegen::ValueLabelsRanges,
|
|
sized_stack_slots: ir::StackSlots,
|
|
alignment: u32,
|
|
}
|
|
|
|
/// Function and its instructions addresses mappings.
|
|
#[derive(Debug, Clone, PartialEq, Eq, Default)]
|
|
struct FunctionAddressMap {
|
|
/// An array of data for the instructions in this function, indicating where
|
|
/// each instruction maps back to in the original function.
|
|
///
|
|
/// This array is sorted least-to-greatest by the `code_offset` field.
|
|
/// Additionally the span of each `InstructionAddressMap` is implicitly the
|
|
/// gap between it and the next item in the array.
|
|
instructions: Box<[InstructionAddressMap]>,
|
|
|
|
/// Function's initial offset in the source file, specified in bytes from
|
|
/// the front of the file.
|
|
start_srcloc: FilePos,
|
|
|
|
/// Function's end offset in the source file, specified in bytes from
|
|
/// the front of the file.
|
|
end_srcloc: FilePos,
|
|
|
|
/// Generated function body offset if applicable, otherwise 0.
|
|
body_offset: usize,
|
|
|
|
/// Generated function body length.
|
|
body_len: u32,
|
|
}
|
|
|
|
/// A record of a relocation to perform.
|
|
#[derive(Debug, Clone, PartialEq, Eq)]
|
|
struct Relocation {
|
|
/// The relocation code.
|
|
reloc: binemit::Reloc,
|
|
/// Relocation target.
|
|
reloc_target: RelocationTarget,
|
|
/// The offset where to apply the relocation.
|
|
offset: binemit::CodeOffset,
|
|
/// The addend to add to the relocation value.
|
|
addend: binemit::Addend,
|
|
}
|
|
|
|
/// Destination function. Can be either user function or some special one, like `memory.grow`.
|
|
#[derive(Debug, Copy, Clone, PartialEq, Eq)]
|
|
enum RelocationTarget {
|
|
/// The user function index.
|
|
UserFunc(FuncIndex),
|
|
/// A compiler-generated libcall.
|
|
LibCall(ir::LibCall),
|
|
}
|
|
|
|
/// Creates a new cranelift `Signature` with no wasm params/results for the
|
|
/// given calling convention.
|
|
///
|
|
/// This will add the default vmctx/etc parameters to the signature returned.
|
|
fn blank_sig(isa: &dyn TargetIsa, call_conv: CallConv) -> ir::Signature {
|
|
let pointer_type = isa.pointer_type();
|
|
let mut sig = ir::Signature::new(call_conv);
|
|
// Add the caller/callee `vmctx` parameters.
|
|
sig.params.push(ir::AbiParam::special(
|
|
pointer_type,
|
|
ir::ArgumentPurpose::VMContext,
|
|
));
|
|
sig.params.push(ir::AbiParam::new(pointer_type));
|
|
return sig;
|
|
}
|
|
|
|
/// Returns the default calling convention for the `isa` provided.
|
|
///
|
|
/// Note that this calling convention is used for exported functions.
|
|
fn wasmtime_call_conv(isa: &dyn TargetIsa) -> CallConv {
|
|
match isa.triple().default_calling_convention() {
|
|
Ok(CallingConvention::AppleAarch64) => CallConv::WasmtimeAppleAarch64,
|
|
Ok(CallingConvention::SystemV) | Err(()) => CallConv::WasmtimeSystemV,
|
|
Ok(CallingConvention::WindowsFastcall) => CallConv::WasmtimeFastcall,
|
|
Ok(unimp) => unimplemented!("calling convention: {:?}", unimp),
|
|
}
|
|
}
|
|
|
|
/// Appends the types of the `wasm` function signature into the `sig` signature
|
|
/// provided.
|
|
///
|
|
/// Typically the `sig` signature will have been created from [`blank_sig`]
|
|
/// above.
|
|
fn push_types(isa: &dyn TargetIsa, sig: &mut ir::Signature, wasm: &WasmFuncType) {
|
|
let cvt = |ty: &WasmType| ir::AbiParam::new(value_type(isa, *ty));
|
|
sig.params.extend(wasm.params().iter().map(&cvt));
|
|
sig.returns.extend(wasm.returns().iter().map(&cvt));
|
|
}
|
|
|
|
/// Returns the corresponding cranelift type for the provided wasm type.
|
|
fn value_type(isa: &dyn TargetIsa, ty: WasmType) -> ir::types::Type {
|
|
match ty {
|
|
WasmType::I32 => ir::types::I32,
|
|
WasmType::I64 => ir::types::I64,
|
|
WasmType::F32 => ir::types::F32,
|
|
WasmType::F64 => ir::types::F64,
|
|
WasmType::V128 => ir::types::I8X16,
|
|
WasmType::FuncRef | WasmType::ExternRef => reference_type(ty, isa.pointer_type()),
|
|
}
|
|
}
|
|
|
|
/// Returns a cranelift signature suitable to indirectly call the wasm signature
|
|
/// specified by `wasm`.
|
|
///
|
|
/// This will implicitly use the default calling convention for `isa` since to
|
|
/// indirectly call a wasm function it must be possibly exported somehow (e.g.
|
|
/// this assumes the function target to call doesn't use the "fast" calling
|
|
/// convention).
|
|
fn indirect_signature(isa: &dyn TargetIsa, wasm: &WasmFuncType) -> ir::Signature {
|
|
let mut sig = blank_sig(isa, wasmtime_call_conv(isa));
|
|
push_types(isa, &mut sig, wasm);
|
|
return sig;
|
|
}
|
|
|
|
/// Returns the cranelift fucntion signature of the function specified.
|
|
///
|
|
/// Note that this will determine the calling convention for the function, and
|
|
/// namely includes an optimization where functions never exported from a module
|
|
/// use a custom theoretically faster calling convention instead of the default.
|
|
fn func_signature(
|
|
isa: &dyn TargetIsa,
|
|
translation: &ModuleTranslation,
|
|
types: &ModuleTypes,
|
|
index: FuncIndex,
|
|
) -> ir::Signature {
|
|
let func = &translation.module.functions[index];
|
|
let call_conv = match translation.module.defined_func_index(index) {
|
|
// If this is a defined function in the module and it doesn't escape
|
|
// then we can optimize this function to use the fastest calling
|
|
// convention since it's purely an internal implementation detail of
|
|
// the module itself.
|
|
Some(_idx) if !func.is_escaping() => {
|
|
let on_apple_aarch64 = isa
|
|
.triple()
|
|
.default_calling_convention()
|
|
.unwrap_or(CallingConvention::SystemV)
|
|
== CallingConvention::AppleAarch64;
|
|
|
|
if on_apple_aarch64 {
|
|
// FIXME: We need an Apple-specific calling convention, so that
|
|
// Cranelift's ABI implementation generates unwinding directives
|
|
// about pointer authentication usage, so we can't just use
|
|
// `CallConv::Fast`.
|
|
CallConv::WasmtimeAppleAarch64
|
|
} else if isa.triple().architecture == Architecture::S390x {
|
|
// On S390x we need a Wasmtime calling convention to ensure
|
|
// we're using little-endian vector lane order.
|
|
wasmtime_call_conv(isa)
|
|
} else {
|
|
CallConv::Fast
|
|
}
|
|
}
|
|
|
|
// ... otherwise if it's an imported function or if it's a possibly
|
|
// exported function then we use the default ABI wasmtime would
|
|
// otherwise select.
|
|
_ => wasmtime_call_conv(isa),
|
|
};
|
|
let mut sig = blank_sig(isa, call_conv);
|
|
push_types(isa, &mut sig, &types[func.signature]);
|
|
return sig;
|
|
}
|
|
|
|
/// Returns the reference type to use for the provided wasm type.
|
|
fn reference_type(wasm_ty: cranelift_wasm::WasmType, pointer_type: ir::Type) -> ir::Type {
|
|
match wasm_ty {
|
|
cranelift_wasm::WasmType::FuncRef => pointer_type,
|
|
cranelift_wasm::WasmType::ExternRef => match pointer_type {
|
|
ir::types::I32 => ir::types::R32,
|
|
ir::types::I64 => ir::types::R64,
|
|
_ => panic!("unsupported pointer type"),
|
|
},
|
|
_ => panic!("unsupported Wasm reference type"),
|
|
}
|
|
}
|