github
/
cranelift
mirror of https://github.com/bytecodealliance/wasmtime.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity
Browse Source

Merge Lightbeam into Wasmtime.

pull/397/head
Dan Gohman 5 years ago
parent
4cf15bd8c4 5832eff76f
commit
9757f7194c
15 changed files with 10932 additions and 0 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Split View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 34
      lightbeam/Cargo.toml
  2. 220
      lightbeam/LICENSE
  3. 168
      lightbeam/README.md
  4. 36
      lightbeam/examples/test.rs
  5. 5553
      lightbeam/src/backend.rs
  6. 55
      lightbeam/src/disassemble.rs
  7. 27
      lightbeam/src/error.rs
  8. 858
      lightbeam/src/function_body.rs
  9. 41
      lightbeam/src/lib.rs
  10. 2103
      lightbeam/src/microwasm.rs
  11. 637
      lightbeam/src/module.rs
  12. 1067
      lightbeam/src/tests.rs
  13. 130
      lightbeam/src/translate_sections.rs
  14. BIN
      lightbeam/test.wasm
  15. 3
      lightbeam/test.wat

34
lightbeam/Cargo.toml
View File

@ -0,0 +1,34 @@
[package]
name = "lightbeam"
version = "0.0.0"
authors = ["The Lightbeam Project Developers"]
license = "Apache-2.0 WITH LLVM-exception"
readme = "README.md"
categories = ["wasm"]
keywords = ["webassembly", "wasm", "compile", "compiler", "jit"]
publish = false
edition = "2018"
[dependencies]
smallvec = "0.6"
dynasm = "0.3"
dynasmrt = "0.3"
wasmparser = "0.32"
memoffset = "0.2"
itertools = "0.8"
capstone = "0.5.0"
failure = "0.1.3"
failure_derive = "0.1.3"
cranelift-codegen = "0.44"
multi_mut = "0.1"
either = "1.5"
wabt = "0.7"
lazy_static = "1.2"
quickcheck = "0.7"
typemap = "0.3"
[badges]
maintenance = { status = "experimental" }
[features]
bench = []

220
lightbeam/LICENSE
View File

@ -0,0 +1,220 @@
Apache License
Version 2.0, January 2004
http://www.apache.org/licenses/
TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
1. Definitions.
"License" shall mean the terms and conditions for use, reproduction,
and distribution as defined by Sections 1 through 9 of this document.
"Licensor" shall mean the copyright owner or entity authorized by
the copyright owner that is granting the License.
"Legal Entity" shall mean the union of the acting entity and all
other entities that control, are controlled by, or are under common
control with that entity. For the purposes of this definition,
"control" means (i) the power, direct or indirect, to cause the
direction or management of such entity, whether by contract or
otherwise, or (ii) ownership of fifty percent (50%) or more of the
outstanding shares, or (iii) beneficial ownership of such entity.
"You" (or "Your") shall mean an individual or Legal Entity
exercising permissions granted by this License.
"Source" form shall mean the preferred form for making modifications,
including but not limited to software source code, documentation
source, and configuration files.
"Object" form shall mean any form resulting from mechanical
transformation or translation of a Source form, including but
not limited to compiled object code, generated documentation,
and conversions to other media types.
"Work" shall mean the work of authorship, whether in Source or
Object form, made available under the License, as indicated by a
copyright notice that is included in or attached to the work
(an example is provided in the Appendix below).
"Derivative Works" shall mean any work, whether in Source or Object
form, that is based on (or derived from) the Work and for which the
editorial revisions, annotations, elaborations, or other modifications
represent, as a whole, an original work of authorship. For the purposes
of this License, Derivative Works shall not include works that remain
separable from, or merely link (or bind by name) to the interfaces of,
the Work and Derivative Works thereof.
"Contribution" shall mean any work of authorship, including
the original version of the Work and any modifications or additions
to that Work or Derivative Works thereof, that is intentionally
submitted to Licensor for inclusion in the Work by the copyright owner
or by an individual or Legal Entity authorized to submit on behalf of
the copyright owner. For the purposes of this definition, "submitted"
means any form of electronic, verbal, or written communication sent
to the Licensor or its representatives, including but not limited to
communication on electronic mailing lists, source code control systems,
and issue tracking systems that are managed by, or on behalf of, the
Licensor for the purpose of discussing and improving the Work, but
excluding communication that is conspicuously marked or otherwise
designated in writing by the copyright owner as "Not a Contribution."
"Contributor" shall mean Licensor and any individual or Legal Entity
on behalf of whom a Contribution has been received by Licensor and
subsequently incorporated within the Work.
2. Grant of Copyright License. Subject to the terms and conditions of
this License, each Contributor hereby grants to You a perpetual,
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
copyright license to reproduce, prepare Derivative Works of,
publicly display, publicly perform, sublicense, and distribute the
Work and such Derivative Works in Source or Object form.
3. Grant of Patent License. Subject to the terms and conditions of
this License, each Contributor hereby grants to You a perpetual,
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
(except as stated in this section) patent license to make, have made,
use, offer to sell, sell, import, and otherwise transfer the Work,
where such license applies only to those patent claims licensable
by such Contributor that are necessarily infringed by their
Contribution(s) alone or by combination of their Contribution(s)
with the Work to which such Contribution(s) was submitted. If You
institute patent litigation against any entity (including a
cross-claim or counterclaim in a lawsuit) alleging that the Work
or a Contribution incorporated within the Work constitutes direct
or contributory patent infringement, then any patent licenses
granted to You under this License for that Work shall terminate
as of the date such litigation is filed.
4. Redistribution. You may reproduce and distribute copies of the
Work or Derivative Works thereof in any medium, with or without
modifications, and in Source or Object form, provided that You
meet the following conditions:
(a) You must give any other recipients of the Work or
Derivative Works a copy of this License; and
(b) You must cause any modified files to carry prominent notices
stating that You changed the files; and
(c) You must retain, in the Source form of any Derivative Works
that You distribute, all copyright, patent, trademark, and
attribution notices from the Source form of the Work,
excluding those notices that do not pertain to any part of
the Derivative Works; and
(d) If the Work includes a "NOTICE" text file as part of its
distribution, then any Derivative Works that You distribute must
include a readable copy of the attribution notices contained
within such NOTICE file, excluding those notices that do not
pertain to any part of the Derivative Works, in at least one
of the following places: within a NOTICE text file distributed
as part of the Derivative Works; within the Source form or
documentation, if provided along with the Derivative Works; or,
within a display generated by the Derivative Works, if and
wherever such third-party notices normally appear. The contents
of the NOTICE file are for informational purposes only and
do not modify the License. You may add Your own attribution
notices within Derivative Works that You distribute, alongside
or as an addendum to the NOTICE text from the Work, provided
that such additional attribution notices cannot be construed
as modifying the License.
You may add Your own copyright statement to Your modifications and
may provide additional or different license terms and conditions
for use, reproduction, or distribution of Your modifications, or
for any such Derivative Works as a whole, provided Your use,
reproduction, and distribution of the Work otherwise complies with
the conditions stated in this License.
5. Submission of Contributions. Unless You explicitly state otherwise,
any Contribution intentionally submitted for inclusion in the Work
by You to the Licensor shall be under the terms and conditions of
this License, without any additional terms or conditions.
Notwithstanding the above, nothing herein shall supersede or modify
the terms of any separate license agreement you may have executed
with Licensor regarding such Contributions.
6. Trademarks. This License does not grant permission to use the trade
names, trademarks, service marks, or product names of the Licensor,
except as required for reasonable and customary use in describing the
origin of the Work and reproducing the content of the NOTICE file.
7. Disclaimer of Warranty. Unless required by applicable law or
agreed to in writing, Licensor provides the Work (and each
Contributor provides its Contributions) on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
implied, including, without limitation, any warranties or conditions
of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
PARTICULAR PURPOSE. You are solely responsible for determining the
appropriateness of using or redistributing the Work and assume any
risks associated with Your exercise of permissions under this License.
8. Limitation of Liability. In no event and under no legal theory,
whether in tort (including negligence), contract, or otherwise,
unless required by applicable law (such as deliberate and grossly
negligent acts) or agreed to in writing, shall any Contributor be
liable to You for damages, including any direct, indirect, special,
incidental, or consequential damages of any character arising as a
result of this License or out of the use or inability to use the
Work (including but not limited to damages for loss of goodwill,
work stoppage, computer failure or malfunction, or any and all
other commercial damages or losses), even if such Contributor
has been advised of the possibility of such damages.
9. Accepting Warranty or Additional Liability. While redistributing
the Work or Derivative Works thereof, You may choose to offer,
and charge a fee for, acceptance of support, warranty, indemnity,
or other liability obligations and/or rights consistent with this
License. However, in accepting such obligations, You may act only
on Your own behalf and on Your sole responsibility, not on behalf
of any other Contributor, and only if You agree to indemnify,
defend, and hold each Contributor harmless for any liability
incurred by, or claims asserted against, such Contributor by reason
of your accepting any such warranty or additional liability.
END OF TERMS AND CONDITIONS
APPENDIX: How to apply the Apache License to your work.
To apply the Apache License to your work, attach the following
boilerplate notice, with the fields enclosed by brackets "[]"
replaced with your own identifying information. (Don't include
the brackets!) The text should be enclosed in the appropriate
comment syntax for the file format. We also recommend that a
file or class name and description of purpose be included on the
same "printed page" as the copyright notice for easier
identification within third-party archives.
Copyright [yyyy] [name of copyright owner]
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
--- LLVM Exceptions to the Apache 2.0 License ----
As an exception, if, as a result of your compiling your source code, portions
of this Software are embedded into an Object form of such source code, you
may redistribute such embedded portions in such Object form without complying
with the conditions of Sections 4(a), 4(b) and 4(d) of the License.
In addition, if you combine or link compiled forms of this Software with
software that is licensed under the GPLv2 ("Combined Software") and if a
court of competent jurisdiction determines that the patent provision (Section
3), the indemnity provision (Section 9) or other Section of the License
conflicts with the conditions of the GPLv2, you may retroactively and
prospectively choose to deem waived or otherwise exclude such Section(s) of
the License, but only in their entirety and only with respect to the Combined
Software.

168
lightbeam/README.md
View File

@ -0,0 +1,168 @@
# Lightbeam
Lightbeam is an optimising one-pass streaming compiler for WebAssembly, intended for use in [Wasmtime][wasmtime].
[wasmtime]: https://github.com/CraneStation/wasmtime
## Quality of output
Already - with a very small number of relatively simple optimisation rules - Lightbeam produces surprisingly high-quality output considering how restricted it is. It even produces better code than Cranelift, Firefox or both for some workloads. Here's a very simple example, this recursive fibonacci function in Rust:
```rust
fn fib(n: i32) -> i32 {
if n == 0 || n == 1 {
1
} else {
fib(n - 1) + fib(n - 2)
}
}
```
When compiled with optimisations enabled, rustc will produce the following WebAssembly:
```rust
(module
(func $fib (param $p0 i32) (result i32)
(local $l1 i32)
(set_local $l1
(i32.const 1))
(block $B0
(br_if $B0
(i32.lt_u
(get_local $p0)
(i32.const 2)))
(set_local $l1
(i32.const 1))
(loop $L1
(set_local $l1
(i32.add
(call $fib
(i32.add
(get_local $p0)
(i32.const -1)))
(get_local $l1)))
(br_if $L1
(i32.gt_u
(tee_local $p0
(i32.add
(get_local $p0)
(i32.const -2)))
(i32.const 1)))))
(get_local $l1)))
```
Firefox's optimising compiler produces the following assembly (labels cleaned up somewhat):
```asm
fib:
sub rsp, 0x18
cmp qword ptr [r14 + 0x28], rsp
jae stack_overflow
mov dword ptr [rsp + 0xc], edi
cmp edi, 2
jae .Lelse
mov eax, 1
mov dword ptr [rsp + 8], eax
jmp .Lreturn
.Lelse:
mov dword ptr [rsp + 0xc], edi
mov eax, 1
mov dword ptr [rsp + 8], eax
.Lloop:
mov edi, dword ptr [rsp + 0xc]
add edi, -1
call 0
mov ecx, dword ptr [rsp + 8]
add ecx, eax
mov dword ptr [rsp + 8], ecx
mov ecx, dword ptr [rsp + 0xc]
add ecx, -2
mov dword ptr [rsp + 0xc], ecx
cmp ecx, 1
ja .Lloop
.Lreturn:
mov eax, dword ptr [rsp + 8]
nop
add rsp, 0x18
ret
```
Cranelift with optimisations enabled produces similar:
```asm
fib:
push rbp
mov rbp, rsp
sub rsp, 0x20
mov qword ptr [rsp + 0x10], rdi
mov dword ptr [rsp + 0x1c], esi
mov eax, 1
mov dword ptr [rsp + 0x18], eax
mov eax, dword ptr [rsp + 0x1c]
cmp eax, 2
jb .Lreturn
movabs rax, 0
mov qword ptr [rsp + 8], rax
.Lloop:
mov eax, dword ptr [rsp + 0x1c]
add eax, -1
mov rcx, qword ptr [rsp + 8]
mov rdx, qword ptr [rsp + 0x10]
mov rdi, rdx
mov esi, eax
call rcx
mov ecx, dword ptr [rsp + 0x18]
add eax, ecx
mov dword ptr [rsp + 0x18], eax
mov eax, dword ptr [rsp + 0x1c]
add eax, -2
mov dword ptr [rsp + 0x1c], eax
mov eax, dword ptr [rsp + 0x1c]
cmp eax, 1
ja .Lloop
.Lreturn:
mov eax, dword ptr [rsp + 0x18]
add rsp, 0x20
pop rbp
ret
```
Whereas Lightbeam produces smaller code with far fewer memory accesses than both (and fewer blocks than Firefox's output):
```asm
fib:
cmp esi, 2
mov eax, 1
jb .Lreturn
mov eax, 1
.Lloop:
mov rcx, rsi
add ecx, 0xffffffff
push rsi
push rax
push rax
mov rsi, rcx
call fib
add eax, [rsp + 8]
mov rcx, [rsp + 0x10]
add ecx, 0xfffffffe
cmp ecx, 1
mov rsi, rcx
lea rsp, [rsp + 0x18]
ja .Lloop
.Lreturn:
ret
```
Now obviously I'm not advocating for replacing Firefox's optimising compiler with Lightbeam since the latter can only really produce better code when receiving optimised WebAssembly (and so debug-mode or hand-written WebAssembly may produce much worse output). However, this shows that even with the restrictions of a streaming compiler it's absolutely possible to produce high-quality assembly output. For the assembly above, the Lightbeam output runs within 15% of native speed. This is paramount for one of Lightbeam's intended usecases for real-time systems that want good runtime performance but cannot tolerate compiler bombs.
## Specification compliance
Lightbeam passes 100% of the specification test suite, but that doesn't necessarily mean that it's 100% specification-compliant. Hopefully as we run a fuzzer against it we can find any issues and get Lightbeam to a state where it can be used in production.
## Getting involved
Our [issue tracker][issue tracker] is pretty barren right now since this is currently more-or-less a one-person project, but if you want to get involved jump into the [CraneStation Gitter room][cranestation-gitter] and someone can direct you to the right place. I wish I could say "the most useful thing you can do is play with it and open issues where you find problems" but until it passes the spec suite that won't be very helpful.
[cranestation-gitter]: https://gitter.im/CraneStation/Lobby
[issue tracker]: https://github.com/CraneStation/lightbeam/issues

36
lightbeam/examples/test.rs
View File

@ -0,0 +1,36 @@
extern crate lightbeam;
use lightbeam::translate;
use std::fs::File;
use std::io;
use std::io::Read;
use std::path::Path;
fn read_to_end<P: AsRef<Path>>(path: P) -> io::Result<Vec<u8>> {
let mut buffer = Vec::new();
if path.as_ref() == Path::new("-") {
let stdin = io::stdin();
let mut stdin = stdin.lock();
stdin.read_to_end(&mut buffer)?;
} else {
let mut file = File::open(path)?;
file.read_to_end(&mut buffer)?;
}
Ok(buffer)
}
fn maybe_main() -> Result<(), String> {
let data = read_to_end("test.wasm").map_err(|e| e.to_string())?;
let translated = translate(&data).map_err(|e| e.to_string())?;
let result: u32 = translated.execute_func(0, (5u32, 3u32)).unwrap();
println!("f(5, 3) = {}", result);
Ok(())
}
fn main() {
match maybe_main() {
Ok(()) => (),
Err(e) => eprintln!("error: {}", e),
}
}

5553
lightbeam/src/backend.rs
View File

File diff suppressed because it is too large

55
lightbeam/src/disassemble.rs
View File

@ -0,0 +1,55 @@
use capstone::prelude::*;
use dynasmrt::AssemblyOffset;
use std::error::Error;
use std::fmt::{Display, Write};
pub fn disassemble(
mem: &[u8],
mut ops: &[(AssemblyOffset, impl Display)],
) -> Result<(), Box<dyn Error>> {
let mut cs = Capstone::new()
.x86()
.mode(arch::x86::ArchMode::Mode64)
.build()?;
println!("{} bytes:", mem.len());
let insns = cs.disasm_all(&mem, 0x0)?;
for i in insns.iter() {
let mut line = String::new();
let address = i.address();
loop {
if let Some((offset, op)) = ops.first() {
if offset.0 as u64 <= address {
ops = &ops[1..];
println!("{}", op);
} else {
break;
}
} else {
break;
}
}
write!(&mut line, "{:4x}:\t", i.address())?;
let mut bytes_str = String::new();
for b in i.bytes() {
write!(&mut bytes_str, "{:02x} ", b)?;
}
write!(&mut line, "{:24}\t", bytes_str)?;
if let Some(s) = i.mnemonic() {
write!(&mut line, "{}\t", s)?;
}
if let Some(s) = i.op_str() {
write!(&mut line, "{}", s)?;
}
println!("{}", line);
}
Ok(())
}

27
lightbeam/src/error.rs
View File

@ -0,0 +1,27 @@
use capstone;
use wasmparser::BinaryReaderError;
#[derive(Fail, PartialEq, Eq, Clone, Debug)]
pub enum Error {
#[fail(display = "Disassembler error: {}", _0)]
Disassembler(String),
#[fail(display = "Assembler error: {}", _0)]
Assembler(String),
#[fail(display = "Input error: {}", _0)]
Input(String),
}
impl From<BinaryReaderError> for Error {
fn from(e: BinaryReaderError) -> Self {
let BinaryReaderError { message, offset } = e;
Error::Input(format!("At wasm offset {}: {}", offset, message))
}
}
impl From<capstone::Error> for Error {
fn from(e: capstone::Error) -> Self {
Error::Disassembler(e.to_string())
}
}

858
lightbeam/src/function_body.rs
View File

@ -0,0 +1,858 @@
use crate::backend::{
ret_locs, BlockCallingConvention, CodeGenSession, Context, Label, Registers, ValueLocation,
VirtualCallingConvention,
};
use crate::error::Error;
use crate::microwasm::*;
use crate::module::{ModuleContext, SigType, Signature};
use cranelift_codegen::binemit;
use dynasmrt::DynasmApi;
use either::{Either, Left, Right};
use multi_mut::HashMapMultiMut;
use std::{collections::HashMap, fmt, hash::Hash, mem};
#[derive(Debug)]
struct Block {
label: BrTarget<Label>,
calling_convention: Option<Either<BlockCallingConvention, VirtualCallingConvention>>,
params: u32,
// TODO: Is there a cleaner way to do this? `has_backwards_callers` should always be set if `is_next`
// is false, so we should probably use an `enum` here.
is_next: bool,
num_callers: Option<u32>,
actual_num_callers: u32,
has_backwards_callers: bool,
}
impl Block {
fn should_serialize_args(&self) -> bool {
self.calling_convention.is_none()
&& (self.num_callers != Some(1) || self.has_backwards_callers)
}
}
const DISASSEMBLE: bool = false;
pub fn translate_wasm<M>(
session: &mut CodeGenSession<M>,
reloc_sink: &mut dyn binemit::RelocSink,
func_idx: u32,
body: &wasmparser::FunctionBody,
) -> Result<(), Error>
where
M: ModuleContext,
for<'any> &'any M::Signature: Into<OpSig>,
{
let ty = session.module_context.defined_func_type(func_idx);
if DISASSEMBLE {
let microwasm_conv = MicrowasmConv::new(
session.module_context,
ty.params().iter().map(SigType::to_microwasm_type),
ty.returns().iter().map(SigType::to_microwasm_type),
body,
);
let _ = crate::microwasm::dis(
std::io::stdout(),
func_idx,
microwasm_conv.flat_map(|ops| ops.unwrap()),
);
}
let microwasm_conv = MicrowasmConv::new(
session.module_context,
ty.params().iter().map(SigType::to_microwasm_type),
ty.returns().iter().map(SigType::to_microwasm_type),
body,
);
translate(
session,
reloc_sink,
func_idx,
microwasm_conv.flat_map(|i| i.expect("TODO: Make this not panic")),
)
}
pub fn translate<M, I, L: Send + Sync + 'static>(
session: &mut CodeGenSession<M>,
reloc_sink: &mut dyn binemit::RelocSink,
func_idx: u32,
body: I,
) -> Result<(), Error>
where
M: ModuleContext,
I: IntoIterator<Item = Operator<L>>,
L: Hash + Clone + Eq,
BrTarget<L>: std::fmt::Display,
{
fn drop_elements<T>(stack: &mut Vec<T>, depths: std::ops::RangeInclusive<u32>) {
let _ = (|| {
let start = stack
.len()
.checked_sub(1)?
.checked_sub(*depths.end() as usize)?;
let end = stack
.len()
.checked_sub(1)?
.checked_sub(*depths.start() as usize)?;
let real_range = start..=end;
stack.drain(real_range);
Some(())
})();
}
let func_type = session.module_context.defined_func_type(func_idx);
let mut body = body.into_iter().peekable();
let module_context = &*session.module_context;
let mut op_offset_map = mem::replace(&mut session.op_offset_map, vec![]);
let ctx = &mut session.new_context(func_idx, reloc_sink);
op_offset_map.push((
ctx.asm.offset(),
Box::new(format!("Function {}:", func_idx)),
));
let params = func_type
.params()
.iter()
.map(|t| t.to_microwasm_type())
.collect::<Vec<_>>();
ctx.start_function(params.iter().cloned());
let mut blocks = HashMap::<BrTarget<L>, Block>::new();
let num_returns = func_type.returns().len();
blocks.insert(
BrTarget::Return,
Block {
label: BrTarget::Return,
params: num_returns as u32,
calling_convention: Some(Left(BlockCallingConvention::function_start(ret_locs(
func_type.returns().iter().map(|t| t.to_microwasm_type()),
)))),
is_next: false,
has_backwards_callers: false,
actual_num_callers: 0,
num_callers: None,
},
);
while let Some(op) = body.next() {
if let Some(Operator::Label(label)) = body.peek() {
let block = blocks
.get_mut(&BrTarget::Label(label.clone()))
.expect("Label defined before being declared");
block.is_next = true;
}
macro_rules! assert_ge {
($left:expr, $right:expr) => ({
match (&$left, &$right) {
(left_val, right_val) => {
if !(*left_val >= *right_val) {
// The reborrows below are intentional. Without them, the stack slot for the
// borrow is initialized even before the values are compared, leading to a
// noticeable slow down.
panic!(r#"assertion failed: `(left >= right)`
left: `{:?}`,
right: `{:?}`"#, &*left_val, &*right_val)
}
}
}
});
($left:expr, $right:expr,) => ({
assert_ge!($left, $right)
});
}
// `cfg` on blocks doesn't work in the compiler right now, so we have to write a dummy macro
#[cfg(debug_assertions)]
macro_rules! assertions {
() => {
if let Operator::Label(label) = &op {
let block = &blocks[&BrTarget::Label(label.clone())];
let num_cc_params = block.calling_convention.as_ref().map(|cc| match cc {
Left(cc) => cc.arguments.len(),
Right(cc) => cc.stack.len(),
});
if let Some(num_cc_params) = num_cc_params {
assert_ge!(num_cc_params, block.params as usize);
}
} else {
let mut actual_regs = Registers::new();
for val in &ctx.block_state.stack {
if let ValueLocation::Reg(gpr) = val {
actual_regs.mark_used(*gpr);
}
}
assert_eq!(actual_regs, ctx.block_state.regs,);
}
};
}
#[cfg(not(debug_assertions))]
macro_rules! assertions {
() => {};
}
assertions!();
struct DisassemblyOpFormatter<Label>(Operator<Label>);
impl<Label> fmt::Display for DisassemblyOpFormatter<Label>
where
Operator<Label>: fmt::Display,
{
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
match self.0 {
Operator::Label(_) => write!(f, "{}", self.0),
Operator::Block { .. } => write!(f, "{:5}\t{}", "", self.0),
_ => write!(f, "{:5}\t {}", "", self.0),
}
}
}
op_offset_map.push((
ctx.asm.offset(),
Box::new(DisassemblyOpFormatter(op.clone())),
));
match op {
Operator::Unreachable => {
ctx.trap();
}
Operator::Label(label) => {
use std::collections::hash_map::Entry;
if let Entry::Occupied(mut entry) = blocks.entry(BrTarget::Label(label.clone())) {
let has_backwards_callers = {
let block = entry.get_mut();
// TODO: Maybe we want to restrict Microwasm so that at least one of its callers
// must be before the label. In an ideal world the restriction would be that
// blocks without callers are illegal, but that's not reasonably possible for
// Microwasm generated from Wasm.
if block.actual_num_callers == 0 {
loop {
let done = match body.peek() {
Some(Operator::Label(_)) | None => true,
Some(_) => false,
};
if done {
break;
}
let skipped = body.next();
// We still want to honour block definitions even in unreachable code
if let Some(Operator::Block {
label,
has_backwards_callers,
params,
num_callers,
}) = skipped
{
let asm_label = ctx.create_label();
blocks.insert(
BrTarget::Label(label),
Block {
label: BrTarget::Label(asm_label),
params: params.len() as _,
calling_convention: None,
is_next: false,
has_backwards_callers,
actual_num_callers: 0,
num_callers,
},
);
}
}
continue;
}
block.is_next = false;
// TODO: We can `take` this if it's a `Right`
match block.calling_convention.as_ref() {
Some(Left(cc)) => {
ctx.apply_cc(cc);
}
Some(Right(virt)) => {
ctx.set_state(virt.clone());
}
_ => assert_eq!(block.params as usize, ctx.block_state.stack.len()),
}
ctx.define_label(block.label.label().unwrap().clone());
block.has_backwards_callers
};
// To reduce memory overhead
if !has_backwards_callers {
entry.remove_entry();
}
} else {
panic!("Label defined before being declared");
}
}
Operator::Block {
label,
has_backwards_callers,
params,
num_callers,
} => {
let asm_label = ctx.create_label();
blocks.insert(
BrTarget::Label(label),
Block {
label: BrTarget::Label(asm_label),
params: params.len() as _,
calling_convention: None,
is_next: false,
has_backwards_callers,
actual_num_callers: 0,
num_callers,
},
);
}
Operator::Br { target } => {
// TODO: We should add the block to the hashmap if we don't have it already
let block = blocks.get_mut(&target).unwrap();
block.actual_num_callers += 1;
let should_serialize_args = block.should_serialize_args();
match block {
Block {
is_next,
label: BrTarget::Label(l),
calling_convention,
..
} => {
let cc = if should_serialize_args {
*calling_convention = Some(Left(ctx.serialize_args(block.params)));
None
} else {
calling_convention
.as_ref()
.map(Either::as_ref)
.and_then(Either::left)
};
if let Some(cc) = cc {
ctx.pass_block_args(cc);
}
if !*is_next {
ctx.br(*l);
}
}
Block {
label: BrTarget::Return,
calling_convention: Some(Left(cc)),
..
} => {
ctx.pass_block_args(cc);
ctx.ret();
}
_ => unimplemented!(),
}
}
Operator::BrIf { then, else_ } => {
let (then_block, else_block) = blocks.pair_mut(&then.target, &else_.target);
// TODO: If actual_num_callers == num_callers then we can remove this block from the hashmap.
// This frees memory and acts as a kind of verification that `num_callers` is set
// correctly. It doesn't help for loops and block ends generated from Wasm.
then_block.actual_num_callers += 1;
else_block.actual_num_callers += 1;
let then_block_parts = (then_block.is_next, then_block.label);
let else_block_parts = (else_block.is_next, else_block.label);
// TODO: The blocks should have compatible (one must be subset of other?) calling
// conventions or else at least one must have no calling convention. This
// should always be true for converting from WebAssembly AIUI.
let f = |ctx: &mut Context<_>| {
let then_block_should_serialize_args = then_block.should_serialize_args();
let else_block_should_serialize_args = else_block.should_serialize_args();
let max_params = then_block.params.max(else_block.params);
match (
(&mut then_block.calling_convention, &then.to_drop),
(&mut else_block.calling_convention, &else_.to_drop),
) {
((Some(Left(ref cc)), _), ref mut other @ (None, _))
| (ref mut other @ (None, _), (Some(Left(ref cc)), _)) => {
let mut cc = ctx.serialize_block_args(cc, max_params);
if let Some(to_drop) = other.1 {
drop_elements(&mut cc.arguments, to_drop.clone());
}
*other.0 = Some(Left(cc));
}
(
(ref mut then_cc @ None, then_to_drop),
(ref mut else_cc @ None, else_to_drop),
) => {
let virt_cc = if !then_block_should_serialize_args
|| !else_block_should_serialize_args
{
Some(ctx.virtual_calling_convention())
} else {
None
};
let cc = if then_block_should_serialize_args
|| else_block_should_serialize_args
{
Some(ctx.serialize_args(max_params))
} else {
None
};
**then_cc = if then_block_should_serialize_args {
let mut cc = cc.clone().unwrap();
if let Some(to_drop) = then_to_drop.clone() {
drop_elements(&mut cc.arguments, to_drop);
}
Some(Left(cc))
} else {
let mut cc = virt_cc.clone().unwrap();
if let Some(to_drop) = then_to_drop.clone() {
drop_elements(&mut cc.stack, to_drop);
}
Some(Right(cc))
};
**else_cc = if else_block_should_serialize_args {
let mut cc = cc.unwrap();
if let Some(to_drop) = else_to_drop.clone() {
drop_elements(&mut cc.arguments, to_drop);
}
Some(Left(cc))
} else {
let mut cc = virt_cc.unwrap();
if let Some(to_drop) = else_to_drop.clone() {
drop_elements(&mut cc.stack, to_drop);
}
Some(Right(cc))
};
}
_ => unimplemented!(
"Can't pass different params to different sides of `br_if` yet"
),
}
};
match (then_block_parts, else_block_parts) {
((true, _), (false, else_)) => {
ctx.br_if_false(else_, f);
}
((false, then), (true, _)) => {
ctx.br_if_true(then, f);
}
((false, then), (false, else_)) => {
ctx.br_if_true(then, f);
ctx.br(else_);
}
other => unimplemented!("{:#?}", other),
}
}
Operator::BrTable(BrTable { targets, default }) => {
use itertools::Itertools;
let (label, num_callers, params) = {
let def = &blocks[&default.target];
(
if def.is_next { None } else { Some(def.label) },
def.num_callers,
def.params
+ default
.to_drop
.as_ref()
.map(|t| t.clone().count())
.unwrap_or_default() as u32,
)
};
let target_labels = targets
.iter()
.map(|target| {
let block = &blocks[&target.target];
if block.is_next {
None
} else {
Some(block.label)
}
})
.collect::<Vec<_>>();
ctx.br_table(target_labels, label, |ctx| {
let mut cc = None;
let mut max_params = params;
let mut max_num_callers = num_callers;
for target in targets.iter().chain(std::iter::once(&default)).unique() {
let block = blocks.get_mut(&target.target).unwrap();
block.actual_num_callers += 1;
if block.calling_convention.is_some() {
let new_cc = block.calling_convention.clone();
assert!(cc.is_none() || cc == new_cc, "Can't pass different params to different elements of `br_table` yet");
cc = new_cc;
}
if let Some(max) = max_num_callers {
max_num_callers = block.num_callers.map(|n| max.max(n));
}
max_params = max_params.max(
block.params
+ target
.to_drop
.as_ref()
.map(|t| t.clone().count())
.unwrap_or_default() as u32
);
}
let cc = cc.map(|cc| {
match cc {
Left(cc) => Left(ctx.serialize_block_args(&cc, max_params)),
Right(cc) => Right(cc),
}
}).unwrap_or_else(||
if max_num_callers.map(|callers| callers <= 1).unwrap_or(false) {
Right(ctx.virtual_calling_convention())
} else {
Left(ctx.serialize_args(max_params))
}
);
for target in targets.iter().chain(std::iter::once(&default)).unique() {
let block = blocks.get_mut(&target.target).unwrap();
let mut cc = cc.clone();
if let Some(to_drop) = target.to_drop.clone() {
match &mut cc {
Left(cc) => drop_elements(&mut cc.arguments, to_drop),
Right(cc) => drop_elements(&mut cc.stack, to_drop),
}
}
block.calling_convention = Some(cc);
}
});
}
Operator::Swap(depth) => ctx.swap(depth),
Operator::Pick(depth) => ctx.pick(depth),
Operator::Eq(I32) => ctx.i32_eq(),
Operator::Eqz(Size::_32) => ctx.i32_eqz(),
Operator::Ne(I32) => ctx.i32_neq(),
Operator::Lt(SI32) => ctx.i32_lt_s(),
Operator::Le(SI32) => ctx.i32_le_s(),
Operator::Gt(SI32) => ctx.i32_gt_s(),
Operator::Ge(SI32) => ctx.i32_ge_s(),
Operator::Lt(SU32) => ctx.i32_lt_u(),
Operator::Le(SU32) => ctx.i32_le_u(),
Operator::Gt(SU32) => ctx.i32_gt_u(),
Operator::Ge(SU32) => ctx.i32_ge_u(),
Operator::Add(I32) => ctx.i32_add(),
Operator::Sub(I32) => ctx.i32_sub(),
Operator::And(Size::_32) => ctx.i32_and(),
Operator::Or(Size::_32) => ctx.i32_or(),
Operator::Xor(Size::_32) => ctx.i32_xor(),
Operator::Mul(I32) => ctx.i32_mul(),
Operator::Div(SU32) => ctx.i32_div_u(),
Operator::Div(SI32) => ctx.i32_div_s(),
Operator::Rem(sint::I32) => ctx.i32_rem_s(),
Operator::Rem(sint::U32) => ctx.i32_rem_u(),
Operator::Shl(Size::_32) => ctx.i32_shl(),
Operator::Shr(sint::I32) => ctx.i32_shr_s(),
Operator::Shr(sint::U32) => ctx.i32_shr_u(),
Operator::Rotl(Size::_32) => ctx.i32_rotl(),
Operator::Rotr(Size::_32) => ctx.i32_rotr(),
Operator::Clz(Size::_32) => ctx.i32_clz(),
Operator::Ctz(Size::_32) => ctx.i32_ctz(),
Operator::Popcnt(Size::_32) => ctx.i32_popcnt(),
Operator::Eq(I64) => ctx.i64_eq(),
Operator::Eqz(Size::_64) => ctx.i64_eqz(),
Operator::Ne(I64) => ctx.i64_neq(),
Operator::Lt(SI64) => ctx.i64_lt_s(),
Operator::Le(SI64) => ctx.i64_le_s(),
Operator::Gt(SI64) => ctx.i64_gt_s(),
Operator::Ge(SI64) => ctx.i64_ge_s(),
Operator::Lt(SU64) => ctx.i64_lt_u(),
Operator::Le(SU64) => ctx.i64_le_u(),
Operator::Gt(SU64) => ctx.i64_gt_u(),
Operator::Ge(SU64) => ctx.i64_ge_u(),
Operator::Add(I64) => ctx.i64_add(),
Operator::Sub(I64) => ctx.i64_sub(),
Operator::And(Size::_64) => ctx.i64_and(),
Operator::Or(Size::_64) => ctx.i64_or(),
Operator::Xor(Size::_64) => ctx.i64_xor(),
Operator::Mul(I64) => ctx.i64_mul(),
Operator::Div(SU64) => ctx.i64_div_u(),
Operator::Div(SI64) => ctx.i64_div_s(),
Operator::Rem(sint::I64) => ctx.i64_rem_s(),
Operator::Rem(sint::U64) => ctx.i64_rem_u(),
Operator::Shl(Size::_64) => ctx.i64_shl(),
Operator::Shr(sint::I64) => ctx.i64_shr_s(),
Operator::Shr(sint::U64) => ctx.i64_shr_u(),
Operator::Rotl(Size::_64) => ctx.i64_rotl(),
Operator::Rotr(Size::_64) => ctx.i64_rotr(),
Operator::Clz(Size::_64) => ctx.i64_clz(),
Operator::Ctz(Size::_64) => ctx.i64_ctz(),
Operator::Popcnt(Size::_64) => ctx.i64_popcnt(),
Operator::Add(F32) => ctx.f32_add(),
Operator::Mul(F32) => ctx.f32_mul(),
Operator::Sub(F32) => ctx.f32_sub(),
Operator::Div(SF32) => ctx.f32_div(),
Operator::Min(Size::_32) => ctx.f32_min(),
Operator::Max(Size::_32) => ctx.f32_max(),
Operator::Copysign(Size::_32) => ctx.f32_copysign(),
Operator::Sqrt(Size::_32) => ctx.f32_sqrt(),
Operator::Neg(Size::_32) => ctx.f32_neg(),
Operator::Abs(Size::_32) => ctx.f32_abs(),
Operator::Floor(Size::_32) => ctx.f32_floor(),
Operator::Ceil(Size::_32) => ctx.f32_ceil(),
Operator::Nearest(Size::_32) => ctx.f32_nearest(),
Operator::Trunc(Size::_32) => ctx.f32_trunc(),
Operator::Eq(F32) => ctx.f32_eq(),
Operator::Ne(F32) => ctx.f32_ne(),
Operator::Gt(SF32) => ctx.f32_gt(),
Operator::Ge(SF32) => ctx.f32_ge(),
Operator::Lt(SF32) => ctx.f32_lt(),
Operator::Le(SF32) => ctx.f32_le(),
Operator::Add(F64) => ctx.f64_add(),
Operator::Mul(F64) => ctx.f64_mul(),
Operator::Sub(F64) => ctx.f64_sub(),
Operator::Div(SF64) => ctx.f64_div(),
Operator::Min(Size::_64) => ctx.f64_min(),
Operator::Max(Size::_64) => ctx.f64_max(),
Operator::Copysign(Size::_64) => ctx.f64_copysign(),
Operator::Sqrt(Size::_64) => ctx.f64_sqrt(),
Operator::Neg(Size::_64) => ctx.f64_neg(),
Operator::Abs(Size::_64) => ctx.f64_abs(),
Operator::Floor(Size::_64) => ctx.f64_floor(),
Operator::Ceil(Size::_64) => ctx.f64_ceil(),
Operator::Nearest(Size::_64) => ctx.f64_nearest(),
Operator::Trunc(Size::_64) => ctx.f64_trunc(),
Operator::Eq(F64) => ctx.f64_eq(),
Operator::Ne(F64) => ctx.f64_ne(),
Operator::Gt(SF64) => ctx.f64_gt(),
Operator::Ge(SF64) => ctx.f64_ge(),
Operator::Lt(SF64) => ctx.f64_lt(),
Operator::Le(SF64) => ctx.f64_le(),
Operator::Drop(range) => ctx.drop(range),
Operator::Const(val) => ctx.const_(val),
Operator::I32WrapFromI64 => ctx.i32_wrap_from_i64(),
Operator::I32ReinterpretFromF32 => ctx.i32_reinterpret_from_f32(),
Operator::I64ReinterpretFromF64 => ctx.i64_reinterpret_from_f64(),
Operator::F32ReinterpretFromI32 => ctx.f32_reinterpret_from_i32(),
Operator::F64ReinterpretFromI64 => ctx.f64_reinterpret_from_i64(),
Operator::ITruncFromF {
input_ty: Size::_32,
output_ty: sint::I32,
} => {
ctx.i32_truncate_f32_s();
}
Operator::ITruncFromF {
input_ty: Size::_32,
output_ty: sint::U32,
} => {
ctx.i32_truncate_f32_u();
}
Operator::ITruncFromF {
input_ty: Size::_64,
output_ty: sint::I32,
} => {
ctx.i32_truncate_f64_s();
}
Operator::ITruncFromF {
input_ty: Size::_64,
output_ty: sint::U32,
} => {
ctx.i32_truncate_f64_u();
}
Operator::ITruncFromF {
input_ty: Size::_32,
output_ty: sint::I64,
} => {
ctx.i64_truncate_f32_s();
}
Operator::ITruncFromF {
input_ty: Size::_32,
output_ty: sint::U64,
} => {
ctx.i64_truncate_f32_u();
}
Operator::ITruncFromF {
input_ty: Size::_64,
output_ty: sint::I64,
} => {
ctx.i64_truncate_f64_s();
}
Operator::ITruncFromF {
input_ty: Size::_64,
output_ty: sint::U64,
} => {
ctx.i64_truncate_f64_u();
}
Operator::Extend {
sign: Signedness::Unsigned,
} => ctx.i32_extend_u(),
Operator::Extend {
sign: Signedness::Signed,
} => ctx.i32_extend_s(),
Operator::FConvertFromI {
input_ty: sint::I32,
output_ty: Size::_32,
} => ctx.f32_convert_from_i32_s(),
Operator::FConvertFromI {
input_ty: sint::I32,
output_ty: Size::_64,
} => ctx.f64_convert_from_i32_s(),
Operator::FConvertFromI {
input_ty: sint::I64,
output_ty: Size::_32,
} => ctx.f32_convert_from_i64_s(),
Operator::FConvertFromI {
input_ty: sint::I64,
output_ty: Size::_64,
} => ctx.f64_convert_from_i64_s(),
Operator::FConvertFromI {
input_ty: sint::U32,
output_ty: Size::_32,
} => ctx.f32_convert_from_i32_u(),
Operator::FConvertFromI {
input_ty: sint::U32,
output_ty: Size::_64,
} => ctx.f64_convert_from_i32_u(),
Operator::FConvertFromI {
input_ty: sint::U64,
output_ty: Size::_32,
} => ctx.f32_convert_from_i64_u(),
Operator::FConvertFromI {
input_ty: sint::U64,
output_ty: Size::_64,
} => ctx.f64_convert_from_i64_u(),
Operator::F64PromoteFromF32 => ctx.f64_from_f32(),
Operator::F32DemoteFromF64 => ctx.f32_from_f64(),
Operator::Load8 {
ty: sint::U32,
memarg,
} => ctx.i32_load8_u(memarg.offset),
Operator::Load16 {
ty: sint::U32,
memarg,
} => ctx.i32_load16_u(memarg.offset),
Operator::Load8 {
ty: sint::I32,
memarg,
} => ctx.i32_load8_s(memarg.offset),
Operator::Load16 {
ty: sint::I32,
memarg,
} => ctx.i32_load16_s(memarg.offset),
Operator::Load8 {
ty: sint::U64,
memarg,
} => ctx.i64_load8_u(memarg.offset),
Operator::Load16 {
ty: sint::U64,
memarg,
} => ctx.i64_load16_u(memarg.offset),
Operator::Load8 {
ty: sint::I64,
memarg,
} => ctx.i64_load8_s(memarg.offset),
Operator::Load16 {
ty: sint::I64,
memarg,
} => ctx.i64_load16_s(memarg.offset),
Operator::Load32 {
sign: Signedness::Unsigned,
memarg,
} => ctx.i64_load32_u(memarg.offset),
Operator::Load32 {
sign: Signedness::Signed,
memarg,
} => ctx.i64_load32_s(memarg.offset),
Operator::Load { ty: I32, memarg } => ctx.i32_load(memarg.offset),
Operator::Load { ty: F32, memarg } => ctx.f32_load(memarg.offset),
Operator::Load { ty: I64, memarg } => ctx.i64_load(memarg.offset),
Operator::Load { ty: F64, memarg } => ctx.f64_load(memarg.offset),
Operator::Store8 { ty: _, memarg } => ctx.store8(memarg.offset),
Operator::Store16 { ty: _, memarg } => ctx.store16(memarg.offset),
Operator::Store32 { memarg }
| Operator::Store { ty: I32, memarg }
| Operator::Store { ty: F32, memarg } => ctx.store32(memarg.offset),
Operator::Store { ty: I64, memarg } | Operator::Store { ty: F64, memarg } => {
ctx.store64(memarg.offset)
}
Operator::GetGlobal(idx) => ctx.get_global(idx),
Operator::SetGlobal(idx) => ctx.set_global(idx),
Operator::Select => {
ctx.select();
}
Operator::MemorySize { reserved: _ } => {
ctx.memory_size();
}
Operator::MemoryGrow { reserved: _ } => {
ctx.memory_grow();
}
Operator::Call { function_index } => {
let callee_ty = module_context.func_type(function_index);
if let Some(defined_index) = module_context.defined_func_index(function_index) {
if function_index == func_idx {
ctx.call_direct_self(
defined_index,
callee_ty.params().iter().map(|t| t.to_microwasm_type()),
callee_ty.returns().iter().map(|t| t.to_microwasm_type()),
);
} else {
ctx.call_direct(
function_index,
callee_ty.params().iter().map(|t| t.to_microwasm_type()),
callee_ty.returns().iter().map(|t| t.to_microwasm_type()),
);
}
} else {
ctx.call_direct_imported(
function_index,
callee_ty.params().iter().map(|t| t.to_microwasm_type()),
callee_ty.returns().iter().map(|t| t.to_microwasm_type()),
);
}
}
Operator::CallIndirect {
type_index,
table_index,
} => {
assert_eq!(table_index, 0);
let callee_ty = module_context.signature(type_index);
// TODO: this implementation assumes that this function is locally defined.
ctx.call_indirect(
type_index,
callee_ty.params().iter().map(|t| t.to_microwasm_type()),
callee_ty.returns().iter().map(|t| t.to_microwasm_type()),
);
}
}
}
ctx.epilogue();
mem::replace(&mut session.op_offset_map, op_offset_map);
Ok(())
}

41
lightbeam/src/lib.rs
View File

@ -0,0 +1,41 @@
#![cfg_attr(feature = "bench", feature(test))]
#![feature(proc_macro_hygiene)]
#[macro_use]
extern crate smallvec;
extern crate capstone;
extern crate either;
extern crate failure;
pub extern crate wasmparser;
#[macro_use]
extern crate failure_derive;
#[macro_use]
extern crate memoffset;
extern crate dynasm;
extern crate dynasmrt;
extern crate itertools;
#[cfg(test)]
#[macro_use]
extern crate lazy_static;
#[cfg(test)]
#[macro_use]
extern crate quickcheck;
extern crate wabt;
// Just so we can implement `Signature` for `cranelift_codegen::ir::Signature`
extern crate cranelift_codegen;
extern crate multi_mut;
mod backend;
mod disassemble;
mod error;
mod function_body;
mod microwasm;
mod module;
mod translate_sections;
#[cfg(test)]
mod tests;
pub use crate::backend::CodeGenSession;
pub use crate::function_body::translate_wasm as translate_function;
pub use crate::module::{translate, ExecutableModule, ModuleContext, Signature, TranslatedModule};

2103
lightbeam/src/microwasm.rs
View File

File diff suppressed because it is too large

637
lightbeam/src/module.rs
View File

@ -0,0 +1,637 @@
use crate::backend::TranslatedCodeSection;
use crate::error::Error;
use crate::microwasm;
use crate::translate_sections;
use cranelift_codegen::{
ir::{self, AbiParam, Signature as CraneliftSignature},
isa,
};
use std::{convert::TryInto, mem};
use wasmparser::{FuncType, MemoryType, ModuleReader, SectionCode, Type};
pub trait AsValueType {
const TYPE: Type;
}
pub trait TypeList {
const TYPE_LIST: &'static [Type];
}
impl<T> TypeList for T
where
T: AsValueType,
{
const TYPE_LIST: &'static [Type] = &[T::TYPE];
}
impl AsValueType for i32 {
const TYPE: Type = Type::I32;
}
impl AsValueType for i64 {
const TYPE: Type = Type::I64;
}
impl AsValueType for u32 {
const TYPE: Type = Type::I32;
}
impl AsValueType for u64 {
const TYPE: Type = Type::I64;
}
impl AsValueType for f32 {
const TYPE: Type = Type::F32;
}
impl AsValueType for f64 {
const TYPE: Type = Type::F64;
}
pub trait FunctionArgs<O> {
type FuncType;
unsafe fn call(self, func: Self::FuncType, vm_ctx: *const u8) -> O;
fn into_func(start: *const u8) -> Self::FuncType;
}
type VmCtxPtr = u64;
macro_rules! impl_function_args {
($first:ident $(, $rest:ident)*) => {
impl<Output, $first, $($rest),*> FunctionArgs<Output> for ($first, $($rest),*) {
type FuncType = unsafe extern "sysv64" fn(VmCtxPtr, $first $(, $rest)*) -> Output;
#[allow(non_snake_case)]
unsafe fn call(self, func: Self::FuncType, vm_ctx: *const u8) -> Output {
let ($first, $($rest),*) = self;
func(vm_ctx as VmCtxPtr, $first $(, $rest)*)
}
fn into_func(start: *const u8) -> Self::FuncType {
unsafe { mem::transmute(start) }
}
}
impl<$first: AsValueType, $($rest: AsValueType),*> TypeList for ($first, $($rest),*) {
const TYPE_LIST: &'static [Type] = &[$first::TYPE, $($rest::TYPE),*];
}
impl_function_args!($($rest),*);
};
() => {
impl<Output> FunctionArgs<Output> for () {
type FuncType = unsafe extern "sysv64" fn(VmCtxPtr) -> Output;
unsafe fn call(self, func: Self::FuncType, vm_ctx: *const u8) -> Output {
func(vm_ctx as VmCtxPtr)
}
fn into_func(start: *const u8) -> Self::FuncType {
unsafe { mem::transmute(start) }
}
}
impl TypeList for () {
const TYPE_LIST: &'static [Type] = &[];
}
};
}
impl_function_args!(A, B, C, D, E, F, G, H, I, J, K, L, M, N, O, P, Q, R, S);
#[derive(Default)]
pub struct TranslatedModule {
translated_code_section: Option<TranslatedCodeSection>,
ctx: SimpleContext,
// TODO: Should we wrap this in a `Mutex` so that calling functions from multiple
// threads doesn't cause data races?
memory: Option<MemoryType>,
}
impl TranslatedModule {
pub fn instantiate(self) -> ExecutableModule {
let mem_size = self.memory.map(|m| m.limits.initial).unwrap_or(0) as usize;
let mem: BoxSlice<_> = vec![0u8; mem_size * WASM_PAGE_SIZE]
.into_boxed_slice()
.into();
let ctx = if mem.len > 0 {
Some(Box::new(VmCtx { mem }) as Box<VmCtx>)
} else {
None
};
ExecutableModule {
module: self,
context: ctx,
}
}
pub fn disassemble(&self) {
self.translated_code_section
.as_ref()
.expect("no code section")
.disassemble();
}
}
#[derive(Debug, Copy, Clone, PartialEq, Eq)]
pub enum ExecutionError {
FuncIndexOutOfBounds,
TypeMismatch,
}
pub struct ExecutableModule {
module: TranslatedModule,
context: Option<Box<VmCtx>>,
}
impl ExecutableModule {
/// Executes the function _without checking types_. This can cause undefined
/// memory to be accessed.
pub unsafe fn execute_func_unchecked<Args: FunctionArgs<T>, T>(
&self,
func_idx: u32,
args: Args,
) -> T {
let code_section = self
.module
.translated_code_section
.as_ref()
.expect("no code section");
let start_buf = code_section.func_start(func_idx as usize);
args.call(
Args::into_func(start_buf),
self.context
.as_ref()
.map(|ctx| (&**ctx) as *const VmCtx as *const u8)
.unwrap_or(std::ptr::null()),
)
}
pub fn execute_func<Args: FunctionArgs<T> + TypeList, T: TypeList>(
&self,
func_idx: u32,
args: Args,
) -> Result<T, ExecutionError> {
let module = &self.module;
if func_idx as usize >= module.ctx.func_ty_indicies.len() {
return Err(ExecutionError::FuncIndexOutOfBounds);
}
let type_ = module.ctx.func_type(func_idx);
// TODO: Handle "compatible" types (i.e. f32 and i32)
if (&type_.params[..], &type_.returns[..]) != (Args::TYPE_LIST, T::TYPE_LIST) {
return Err(ExecutionError::TypeMismatch);
}
Ok(unsafe { self.execute_func_unchecked(func_idx, args) })
}
pub fn disassemble(&self) {
self.module.disassemble();
}
}
struct BoxSlice<T> {
len: usize,
ptr: *mut T,
}
impl<T> From<Box<[T]>> for BoxSlice<T> {
fn from(mut other: Box<[T]>) -> Self {
let out = BoxSlice {
len: other.len(),
ptr: other.as_mut_ptr(),
};
mem::forget(other);
out
}
}
unsafe impl<T: Send> Send for BoxSlice<T> {}
unsafe impl<T: Sync> Sync for BoxSlice<T> {}
impl<T> Drop for BoxSlice<T> {
fn drop(&mut self) {
unsafe { Vec::from_raw_parts(self.ptr, self.len, self.len) };
}
}
pub struct VmCtx {
mem: BoxSlice<u8>,
}
impl VmCtx {
pub fn offset_of_memory_ptr() -> u32 {
offset_of!(VmCtx, mem.ptr)
.try_into()
.expect("Offset exceeded size of u32")
}
pub fn offset_of_memory_len() -> u32 {
offset_of!(VmCtx, mem.len)
.try_into()
.expect("Offset exceeded size of u32")
}
}
#[derive(Default, Debug)]
pub struct SimpleContext {
types: Vec<FuncType>,
func_ty_indicies: Vec<u32>,
}
pub const WASM_PAGE_SIZE: usize = 65_536;
pub trait Signature {
type Type: SigType;
fn params(&self) -> &[Self::Type];
fn returns(&self) -> &[Self::Type];
}
pub trait SigType {
fn to_microwasm_type(&self) -> microwasm::SignlessType;
}
impl SigType for ir::Type {
fn to_microwasm_type(&self) -> microwasm::SignlessType {
use crate::microwasm::{Size::*, Type::*};
if self.is_int() {
match self.bits() {
32 => Int(_32),
64 => Int(_64),
_ => unimplemented!(),
}
} else if self.is_float() {
match self.bits() {
32 => Float(_32),
64 => Float(_64),
_ => unimplemented!(),
}
} else {
unimplemented!()
}
}
}
impl SigType for AbiParam {
fn to_microwasm_type(&self) -> microwasm::SignlessType {
self.value_type.to_microwasm_type()
}
}
impl Signature for CraneliftSignature {
type Type = AbiParam;
fn params(&self) -> &[Self::Type] {
// TODO: We want to instead add the `VMContext` to the signature used by
// cranelift, removing the special-casing from the internals.
assert_eq!(self.params[0].purpose, ir::ArgumentPurpose::VMContext);
assert_eq!(self.call_conv, isa::CallConv::SystemV);
&self.params[1..]
}
fn returns(&self) -> &[Self::Type] {
assert_eq!(self.call_conv, isa::CallConv::SystemV);
&self.returns
}
}
impl SigType for wasmparser::Type {
fn to_microwasm_type(&self) -> microwasm::SignlessType {
microwasm::Type::from_wasm(*self).unwrap()
}
}
impl Signature for FuncType {
type Type = wasmparser::Type;
fn params(&self) -> &[Self::Type] {
&*self.params
}
fn returns(&self) -> &[Self::Type] {
&*self.returns
}
}
pub trait ModuleContext {
type Signature: Signature;
type GlobalType: SigType;
fn vmctx_vmglobal_definition(&self, index: u32) -> u32;
fn vmctx_vmglobal_import_from(&self, index: u32) -> u32;
fn vmctx_vmmemory_import_from(&self, memory_index: u32) -> u32;
fn vmctx_vmmemory_definition(&self, defined_memory_index: u32) -> u32;
fn vmctx_vmmemory_definition_base(&self, defined_memory_index: u32) -> u32;
fn vmctx_vmmemory_definition_current_length(&self, defined_memory_index: u32) -> u32;
fn vmmemory_definition_base(&self) -> u8;
fn vmmemory_definition_current_length(&self) -> u8;
fn vmctx_vmtable_import_from(&self, table_index: u32) -> u32;
fn vmctx_vmtable_definition(&self, defined_table_index: u32) -> u32;
fn vmctx_vmtable_definition_base(&self, defined_table_index: u32) -> u32;
fn vmctx_vmtable_definition_current_elements(&self, defined_table_index: u32) -> u32;
fn vmctx_vmfunction_import_body(&self, func_index: u32) -> u32;
fn vmctx_vmfunction_import_vmctx(&self, func_index: u32) -> u32;
fn vmtable_definition_base(&self) -> u8;
fn vmtable_definition_current_elements(&self) -> u8;
fn vmctx_vmshared_signature_id(&self, signature_idx: u32) -> u32;
fn vmcaller_checked_anyfunc_type_index(&self) -> u8;
fn vmcaller_checked_anyfunc_func_ptr(&self) -> u8;
fn vmcaller_checked_anyfunc_vmctx(&self) -> u8;
fn size_of_vmcaller_checked_anyfunc(&self) -> u8;
fn defined_table_index(&self, table_index: u32) -> Option<u32>;
fn defined_memory_index(&self, index: u32) -> Option<u32>;
fn defined_global_index(&self, global_index: u32) -> Option<u32>;
fn global_type(&self, global_index: u32) -> &Self::GlobalType;
fn func_type_index(&self, func_idx: u32) -> u32;
fn signature(&self, index: u32) -> &Self::Signature;
fn func_index(&self, defined_func_index: u32) -> u32;
fn defined_func_index(&self, func_index: u32) -> Option<u32>;
fn defined_func_type(&self, func_idx: u32) -> &Self::Signature {
// TODO: This assumes that there are no imported functions.
self.func_type(self.func_index(func_idx))
}
fn func_type(&self, func_idx: u32) -> &Self::Signature {
self.signature(self.func_type_index(func_idx))
}
fn emit_memory_bounds_check(&self) -> bool {
true
}
}
impl ModuleContext for SimpleContext {
type Signature = FuncType;
type GlobalType = wasmparser::Type;
// TODO: We don't support external functions yet
fn func_index(&self, func_idx: u32) -> u32 {
func_idx
}
fn defined_func_index(&self, func_idx: u32) -> Option<u32> {
Some(func_idx)
}
fn func_type_index(&self, func_idx: u32) -> u32 {
self.func_ty_indicies[func_idx as usize]
}
fn defined_global_index(&self, _index: u32) -> Option<u32> {
unimplemented!()
}
fn global_type(&self, _global_index: u32) -> &Self::GlobalType {
unimplemented!()
}
fn signature(&self, index: u32) -> &Self::Signature {
&self.types[index as usize]
}
fn vmctx_vmglobal_definition(&self, _index: u32) -> u32 {
unimplemented!()
}
fn vmctx_vmglobal_import_from(&self, _index: u32) -> u32 {
unimplemented!()
}
fn defined_memory_index(&self, _index: u32) -> Option<u32> {
unimplemented!()
}
fn defined_table_index(&self, index: u32) -> Option<u32> {
Some(index)
}
fn vmctx_vmfunction_import_body(&self, _func_index: u32) -> u32 {
unimplemented!()
}
fn vmctx_vmfunction_import_vmctx(&self, _func_index: u32) -> u32 {
unimplemented!()
}
fn vmctx_vmtable_import_from(&self, _table_index: u32) -> u32 {
unimplemented!()
}
fn vmctx_vmmemory_definition(&self, _defined_memory_index: u32) -> u32 {
unimplemented!()
}
fn vmctx_vmmemory_import_from(&self, _memory_index: u32) -> u32 {
unimplemented!()
}
fn vmmemory_definition_base(&self) -> u8 {
unimplemented!()
}
fn vmmemory_definition_current_length(&self) -> u8 {
unimplemented!()
}
fn vmctx_vmmemory_definition_base(&self, defined_memory_index: u32) -> u32 {
assert_eq!(defined_memory_index, 0);
VmCtx::offset_of_memory_ptr()
}
fn vmctx_vmmemory_definition_current_length(&self, defined_memory_index: u32) -> u32 {
assert_eq!(defined_memory_index, 0);
VmCtx::offset_of_memory_len()
}
fn vmctx_vmtable_definition(&self, _defined_table_index: u32) -> u32 {
unimplemented!()
}
fn vmctx_vmtable_definition_base(&self, _defined_table_index: u32) -> u32 {
unimplemented!()
}
fn vmctx_vmtable_definition_current_elements(&self, _defined_table_index: u32) -> u32 {
unimplemented!()
}
fn vmtable_definition_base(&self) -> u8 {
unimplemented!()
}
fn vmtable_definition_current_elements(&self) -> u8 {
unimplemented!()
}
fn vmcaller_checked_anyfunc_vmctx(&self) -> u8 {
unimplemented!()
}
fn vmcaller_checked_anyfunc_type_index(&self) -> u8 {
unimplemented!()
}
fn vmcaller_checked_anyfunc_func_ptr(&self) -> u8 {
unimplemented!()
}
fn size_of_vmcaller_checked_anyfunc(&self) -> u8 {
unimplemented!()
}
fn vmctx_vmshared_signature_id(&self, _signature_idx: u32) -> u32 {
unimplemented!()
}
// TODO: type of a global
}
pub fn translate(data: &[u8]) -> Result<ExecutableModule, Error> {
translate_only(data).map(|m| m.instantiate())
}
/// Translate from a slice of bytes holding a wasm module.
pub fn translate_only(data: &[u8]) -> Result<TranslatedModule, Error> {
let mut reader = ModuleReader::new(data)?;
let mut output = TranslatedModule::default();
reader.skip_custom_sections()?;
if reader.eof() {
return Ok(output);
}
let mut section = reader.read()?;
if let SectionCode::Type = section.code {
let types_reader = section.get_type_section_reader()?;
output.ctx.types = translate_sections::type_(types_reader)?;
reader.skip_custom_sections()?;
if reader.eof() {
return Ok(output);
}
section = reader.read()?;
}
if let SectionCode::Import = section.code {
let imports = section.get_import_section_reader()?;
translate_sections::import(imports)?;
reader.skip_custom_sections()?;
if reader.eof() {
return Ok(output);
}
section = reader.read()?;
}
if let SectionCode::Function = section.code {
let functions = section.get_function_section_reader()?;
output.ctx.func_ty_indicies = translate_sections::function(functions)?;
reader.skip_custom_sections()?;
if reader.eof() {
return Ok(output);
}
section = reader.read()?;
}
if let SectionCode::Table = section.code {
let tables = section.get_table_section_reader()?;
translate_sections::table(tables)?;
reader.skip_custom_sections()?;
if reader.eof() {
return Ok(output);
}
section = reader.read()?;
}
if let SectionCode::Memory = section.code {
let memories = section.get_memory_section_reader()?;
let mem = translate_sections::memory(memories)?;
assert!(
mem.len() <= 1,
"Multiple memory sections not yet unimplemented"
);
if !mem.is_empty() {
let mem = mem[0];
assert_eq!(Some(mem.limits.initial), mem.limits.maximum);
output.memory = Some(mem);
}
reader.skip_custom_sections()?;
if reader.eof() {
return Ok(output);
}
section = reader.read()?;
}
if let SectionCode::Global = section.code {
let globals = section.get_global_section_reader()?;
translate_sections::global(globals)?;
reader.skip_custom_sections()?;
if reader.eof() {
return Ok(output);
}
section = reader.read()?;
}
if let SectionCode::Export = section.code {
let exports = section.get_export_section_reader()?;
translate_sections::export(exports)?;
reader.skip_custom_sections()?;
if reader.eof() {
return Ok(output);
}
section = reader.read()?;
}
if let SectionCode::Start = section.code {
let start = section.get_start_section_content()?;
translate_sections::start(start)?;
reader.skip_custom_sections()?;
if reader.eof() {
return Ok(output);
}
section = reader.read()?;
}
if let SectionCode::Element = section.code {
let elements = section.get_element_section_reader()?;
translate_sections::element(elements)?;
reader.skip_custom_sections()?;
if reader.eof() {
return Ok(output);
}
section = reader.read()?;
}
if let SectionCode::Code = section.code {
let code = section.get_code_section_reader()?;
output.translated_code_section = Some(translate_sections::code(code, &output.ctx)?);
reader.skip_custom_sections()?;
if reader.eof() {
return Ok(output);
}
section = reader.read()?;
}
if let SectionCode::Data = section.code {
let data = section.get_data_section_reader()?;
translate_sections::data(data)?;
}
assert!(reader.eof());
Ok(output)
}

1067
lightbeam/src/tests.rs
View File

File diff suppressed because it is too large

130
lightbeam/src/translate_sections.rs
View File

@ -0,0 +1,130 @@
use crate::backend::{CodeGenSession, TranslatedCodeSection};
use crate::error::Error;
use crate::function_body;
use crate::module::SimpleContext;
use cranelift_codegen::{binemit, ir};
use wasmparser::{
CodeSectionReader, DataSectionReader, ElementSectionReader, ExportSectionReader, FuncType,
FunctionSectionReader, GlobalSectionReader, ImportSectionReader, MemorySectionReader,
MemoryType, TableSectionReader, TableType, TypeSectionReader,
};
/// Parses the Type section of the wasm module.
pub fn type_(types_reader: TypeSectionReader) -> Result<Vec<FuncType>, Error> {
types_reader
.into_iter()
.map(|r| r.map_err(Into::into))
.collect()
}
/// Parses the Import section of the wasm module.
pub fn import(imports: ImportSectionReader) -> Result<(), Error> {
for entry in imports {
entry?; // TODO
}
Ok(())
}
/// Parses the Function section of the wasm module.
pub fn function(functions: FunctionSectionReader) -> Result<Vec<u32>, Error> {
functions
.into_iter()
.map(|r| r.map_err(Into::into))
.collect()
}
/// Parses the Table section of the wasm module.
pub fn table(tables: TableSectionReader) -> Result<Vec<TableType>, Error> {
tables.into_iter().map(|r| r.map_err(Into::into)).collect()
}
/// Parses the Memory section of the wasm module.
pub fn memory(memories: MemorySectionReader) -> Result<Vec<MemoryType>, Error> {
memories
.into_iter()
.map(|r| r.map_err(Into::into))
.collect()
}
/// Parses the Global section of the wasm module.
pub fn global(globals: GlobalSectionReader) -> Result<(), Error> {
for entry in globals {
entry?; // TODO
}
Ok(())
}
/// Parses the Export section of the wasm module.
pub fn export(exports: ExportSectionReader) -> Result<(), Error> {
for entry in exports {
entry?; // TODO
}
Ok(())
}
/// Parses the Start section of the wasm module.
pub fn start(_index: u32) -> Result<(), Error> {
// TODO
Ok(())
}
/// Parses the Element section of the wasm module.
pub fn element(elements: ElementSectionReader) -> Result<(), Error> {
for entry in elements {
entry?;
}
Ok(())
}
struct UnimplementedRelocSink;
impl binemit::RelocSink for UnimplementedRelocSink {
fn reloc_ebb(&mut self, _: binemit::CodeOffset, _: binemit::Reloc, _: binemit::CodeOffset) {
unimplemented!()
}
fn reloc_external(
&mut self,
_: binemit::CodeOffset,
_: binemit::Reloc,
_: &ir::ExternalName,
_: binemit::Addend,
) {
unimplemented!()
}
fn reloc_constant(&mut self, _: binemit::CodeOffset, _: binemit::Reloc, _: ir::ConstantOffset) {
unimplemented!()
}
fn reloc_jt(&mut self, _: binemit::CodeOffset, _: binemit::Reloc, _: ir::JumpTable) {
unimplemented!()
}
}
/// Parses the Code section of the wasm module.
pub fn code(
code: CodeSectionReader,
translation_ctx: &SimpleContext,
) -> Result<TranslatedCodeSection, Error> {
let func_count = code.get_count();
let mut session = CodeGenSession::new(func_count, translation_ctx);
for (idx, body) in code.into_iter().enumerate() {
let body = body?;
let mut relocs = UnimplementedRelocSink;
function_body::translate_wasm(&mut session, &mut relocs, idx as u32, &body)?;
}
Ok(session.into_translated_code_section()?)
}
/// Parses the Data section of the wasm module.
pub fn data(data: DataSectionReader) -> Result<(), Error> {
for entry in data {
entry?; // TODO
}
Ok(())
}

BIN
lightbeam/test.wasm
View File

Binary file not shown.

3
lightbeam/test.wat
View File

@ -0,0 +1,3 @@
(module
(func (param i32) (param i32) (result i32) (i32.add (get_local 0) (get_local 1)))
)
Write Preview
Loading…
Cancel
Save