* Implement fused adapters for `(list T)` types This commit implements one of the two remaining types for adapter fusion, lists. This implementation is particularly tricky for a number of reasons: * Lists have a number of validity checks which need to be carefully implemented. For example the byte length of the list passed to allocation in the destination module could overflow the 32-bit index space. Additionally lists in 32-bit memories need a check that their final address is in-bounds in the address space. * In the effort to go ahead and support memory64 at the lowest layers this is where much of the magic happens. Lists are naturally always stored in memory and shifting between 64/32-bit address spaces is done here. This notably required plumbing an `Options` around during flattening/size/alignment calculations due to the size/types of lists changing depending on the memory configuration. I've also added a small `factc` program in this commit which should hopefully assist in exploring and debugging adapter modules. This takes as input a component (text or binary format) and then generates an adapter module for all component function signatures found internally. This commit notably does not include tests for lists. I tried to figure out a good way to add these but I felt like there were too many cases to test and the tests would otherwise be extremely verbose. Instead I think the best testing strategy for this commit will be through #4537 which should be relatively extensible to testing adapters between modules in addition to host-based lifting/lowering. * Improve handling of lists of 0-size types * Skip overflow checks on byte sizes for 0-size types * Skip the copy loop entirely when src/dst are both 0 * Skip the increments of src/dst pointers if either is 0-size * Update semantics for zero-sized lists/strings When a list/string has a 0-byte-size the base pointer is no longer verified to be in-bounds to match the supposedly desired adapter semantics where no trap happens because no turn of the loop happens.
194 lines
6.8 KiB
Rust
194 lines
6.8 KiB
Rust
use anyhow::{bail, Context, Result};
|
|
use clap::Parser;
|
|
use std::path::PathBuf;
|
|
use std::str;
|
|
use wasmparser::{Payload, Validator, WasmFeatures};
|
|
use wasmtime_environ::component::*;
|
|
use wasmtime_environ::fact::Module;
|
|
|
|
/// A small helper utility to explore generated adapter modules from Wasmtime's
|
|
/// adapter fusion compiler.
|
|
///
|
|
/// This utility takes a `*.wat` file as input which is expected to be a valid
|
|
/// WebAssembly component. The component is parsed and any type definition for a
|
|
/// component function gets a generated adapter for it as if the caller/callee
|
|
/// used that type as the adapter.
|
|
///
|
|
/// For example with an input that looks like:
|
|
///
|
|
/// (component
|
|
/// (type (func (param u32) (result (list u8))))
|
|
/// )
|
|
///
|
|
/// This tool can be used to generate an adapter for that signature.
|
|
#[derive(Parser)]
|
|
struct Factc {
|
|
/// Whether or not debug code is inserted into the generated adapter.
|
|
#[clap(long)]
|
|
debug: bool,
|
|
|
|
/// Whether or not the lifting options (the callee of the exported adapter)
|
|
/// uses a 64-bit memory as opposed to a 32-bit memory.
|
|
#[clap(long)]
|
|
lift64: bool,
|
|
|
|
/// Whether or not the lowering options (the caller of the exported adapter)
|
|
/// uses a 64-bit memory as opposed to a 32-bit memory.
|
|
#[clap(long)]
|
|
lower64: bool,
|
|
|
|
/// Whether or not a call to a `post-return` configured function is enabled
|
|
/// or not.
|
|
#[clap(long)]
|
|
post_return: bool,
|
|
|
|
/// Whether or not to skip validation of the generated adapter module.
|
|
#[clap(long)]
|
|
skip_validate: bool,
|
|
|
|
/// Where to place the generated adapter module. Standard output is used if
|
|
/// this is not specified.
|
|
#[clap(short, long)]
|
|
output: Option<PathBuf>,
|
|
|
|
/// Output the text format for WebAssembly instead of the binary format.
|
|
#[clap(short, long)]
|
|
text: bool,
|
|
|
|
/// TODO
|
|
input: PathBuf,
|
|
}
|
|
|
|
fn main() -> Result<()> {
|
|
Factc::parse().execute()
|
|
}
|
|
|
|
impl Factc {
|
|
fn execute(self) -> Result<()> {
|
|
env_logger::init();
|
|
|
|
let mut types = ComponentTypesBuilder::default();
|
|
|
|
// Manufactures a unique `CoreDef` so all function imports get unique
|
|
// function imports.
|
|
let mut next_def = 0;
|
|
let mut dummy_def = || {
|
|
next_def += 1;
|
|
CoreDef::Adapter(AdapterIndex::from_u32(next_def))
|
|
};
|
|
|
|
// Manufactures a `CoreExport` for a memory with the shape specified. Note
|
|
// that we can't import as many memories as functions so these are
|
|
// intentionally limited. Once a handful of memories are generated of each
|
|
// type then they start getting reused.
|
|
let mut next_memory = 0;
|
|
let mut memories32 = Vec::new();
|
|
let mut memories64 = Vec::new();
|
|
let mut dummy_memory = |memory64: bool| {
|
|
let dst = if memory64 {
|
|
&mut memories64
|
|
} else {
|
|
&mut memories32
|
|
};
|
|
let idx = if dst.len() < 5 {
|
|
next_memory += 1;
|
|
dst.push(next_memory - 1);
|
|
next_memory - 1
|
|
} else {
|
|
dst[0]
|
|
};
|
|
CoreExport {
|
|
instance: RuntimeInstanceIndex::from_u32(idx),
|
|
item: ExportItem::Name(String::new()),
|
|
}
|
|
};
|
|
|
|
let mut adapters = Vec::new();
|
|
let input = wat::parse_file(&self.input)?;
|
|
types.push_type_scope();
|
|
let mut validator = Validator::new_with_features(WasmFeatures {
|
|
component_model: true,
|
|
..Default::default()
|
|
});
|
|
for payload in wasmparser::Parser::new(0).parse_all(&input) {
|
|
let payload = payload?;
|
|
validator.payload(&payload)?;
|
|
let section = match payload {
|
|
Payload::ComponentTypeSection(s) => s,
|
|
_ => continue,
|
|
};
|
|
for ty in section {
|
|
let ty = types.intern_component_type(&ty?)?;
|
|
types.push_component_typedef(ty);
|
|
let ty = match ty {
|
|
TypeDef::ComponentFunc(ty) => ty,
|
|
_ => continue,
|
|
};
|
|
adapters.push(Adapter {
|
|
lift_ty: ty,
|
|
lower_ty: ty,
|
|
lower_options: AdapterOptions {
|
|
instance: RuntimeComponentInstanceIndex::from_u32(0),
|
|
string_encoding: StringEncoding::Utf8,
|
|
memory64: self.lower64,
|
|
// Pessimistically assume that memory/realloc are going to be
|
|
// required for this trampoline and provide it. Avoids doing
|
|
// calculations to figure out whether they're necessary and
|
|
// simplifies the fuzzer here without reducing coverage within FACT
|
|
// itself.
|
|
memory: Some(dummy_memory(self.lower64)),
|
|
realloc: Some(dummy_def()),
|
|
// Lowering never allows `post-return`
|
|
post_return: None,
|
|
},
|
|
lift_options: AdapterOptions {
|
|
instance: RuntimeComponentInstanceIndex::from_u32(1),
|
|
string_encoding: StringEncoding::Utf8,
|
|
memory64: self.lift64,
|
|
memory: Some(dummy_memory(self.lift64)),
|
|
realloc: Some(dummy_def()),
|
|
post_return: if self.post_return {
|
|
Some(dummy_def())
|
|
} else {
|
|
None
|
|
},
|
|
},
|
|
func: dummy_def(),
|
|
});
|
|
}
|
|
}
|
|
types.pop_type_scope();
|
|
|
|
let types = types.finish();
|
|
let mut fact_module = Module::new(&types, self.debug);
|
|
for (i, adapter) in adapters.iter().enumerate() {
|
|
fact_module.adapt(&format!("adapter{i}"), adapter);
|
|
}
|
|
let wasm = fact_module.encode();
|
|
Validator::new_with_features(WasmFeatures {
|
|
multi_memory: true,
|
|
memory64: true,
|
|
..WasmFeatures::default()
|
|
})
|
|
.validate_all(&wasm)
|
|
.context("failed to validate generated module")?;
|
|
|
|
let output = if self.text {
|
|
wasmprinter::print_bytes(&wasm)
|
|
.context("failed to convert binary wasm to text")?
|
|
.into_bytes()
|
|
} else if self.output.is_none() && atty::is(atty::Stream::Stdout) {
|
|
bail!("cannot print binary wasm output to a terminal unless `-t` flag is passed")
|
|
} else {
|
|
wasm
|
|
};
|
|
|
|
match &self.output {
|
|
Some(file) => std::fs::write(file, output).context("failed to write output file")?,
|
|
None => println!("{}", str::from_utf8(&output).unwrap()),
|
|
}
|
|
|
|
Ok(())
|
|
}
|
|
}
|