portal/waffle
1
0
Fork 0
Code Issues Pull requests Actions Packages Projects Releases Wiki Activity

Merge branch 'main' of github.com:cfallin/waffle

Browse source
This commit is contained in:
Chris Fallin 2023-02-07 17:36:03 -08:00
parent 56fa6308b4 8f75137f0f
commit 332c0c6f3c
9 changed files with 264 additions and 47 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

1
fuzz/fuzz_targets/roundtrip.rs
View file

@ -24,6 +24,7 @@ fuzz_target!(|module: wasm_smith::Module| {
}
}
};
parsed_module.expand_all_funcs().unwrap();
parsed_module.optimize();
let _ = parsed_module.to_wasm_bytes();
});

34
src/backend/mod.rs
View file

@ -585,7 +585,7 @@ pub fn compile(module: &Module<'_>) -> anyhow::Result<Vec<u8>> {
for (func, func_decl) in module.funcs.entries().skip(num_func_imports) {
match func_decl {
FuncDecl::Import(_) => anyhow::bail!("Import comes after func with body: {}", func),
FuncDecl::Body(sig, _) => {
FuncDecl::Lazy(sig, _) | FuncDecl::Body(sig, _) => {
funcs.function(sig.index() as u32);
}
}
@ -689,21 +689,43 @@ pub fn compile(module: &Module<'_>) -> anyhow::Result<Vec<u8>> {
into_mod.section(&elem);
let mut code = wasm_encoder::CodeSection::new();
enum FuncOrRawBytes<'a> {
Func(wasm_encoder::Function),
Raw(&'a [u8]),
}
let bodies = module
.funcs
.entries()
.skip(num_func_imports)
.collect::<Vec<_>>()
.par_iter()
.map(|(func, func_decl)| -> Result<wasm_encoder::Function> {
let body = func_decl.body().unwrap();
.map(|(func, func_decl)| -> Result<FuncOrRawBytes> {
match func_decl {
FuncDecl::Lazy(_, reader) => {
let data = &module.orig_bytes[reader.range()];
Ok(FuncOrRawBytes::Raw(data))
}
FuncDecl::Body(_, body) => {
log::debug!("Compiling {}", func);
WasmFuncBackend::new(body)?.compile()
WasmFuncBackend::new(body)?
.compile()
.map(|f| FuncOrRawBytes::Func(f))
}
FuncDecl::Import(_) => unreachable!("Should have skipped imports"),
}
})
.collect::<Result<Vec<_>>>()?;
.collect::<Result<Vec<FuncOrRawBytes<'_>>>>()?;
for body in bodies {
code.function(&body);
match body {
FuncOrRawBytes::Func(f) => {
code.function(&f);
}
FuncOrRawBytes::Raw(bytes) => {
code.raw(bytes);
}
}
}
into_mod.section(&code);

29
src/bin/waffle-util.rs
View file

@ -41,6 +41,19 @@ enum Command {
},
}
fn apply_options(opts: &Options, module: &mut Module) -> Result<()> {
if opts.basic_opts || opts.max_ssa {
module.expand_all_funcs()?;
}
if opts.basic_opts {
module.per_func_body(|body| body.optimize());
}
if opts.max_ssa {
module.per_func_body(|body| body.convert_to_max_ssa());
}
Ok(())
}
fn main() -> Result<()> {
let opts = Options::from_args();
@ -50,29 +63,19 @@ fn main() -> Result<()> {
}
let _ = logger.try_init();
match opts.command {
match &opts.command {
Command::PrintIR { wasm } => {
let bytes = std::fs::read(wasm)?;
debug!("Loaded {} bytes of Wasm data", bytes.len());
let mut module = Module::from_wasm_bytes(&bytes[..])?;
if opts.basic_opts {
module.optimize();
}
if opts.max_ssa {
module.convert_to_max_ssa();
}
apply_options(&opts, &mut module)?;
println!("{}", module.display());
}
Command::RoundTrip { input, output } => {
let bytes = std::fs::read(input)?;
debug!("Loaded {} bytes of Wasm data", bytes.len());
let mut module = Module::from_wasm_bytes(&bytes[..])?;
if opts.basic_opts {
module.optimize();
}
if opts.max_ssa {
module.convert_to_max_ssa();
}
apply_options(&opts, &mut module)?;
let produced = module.to_wasm_bytes()?;
std::fs::write(output, &produced[..])?;
}

9
src/frontend.rs
View file

@ -146,10 +146,7 @@ fn handle_payload<'a>(
*next_func += 1;
let my_sig = module.funcs[func_idx].sig();
let body = parse_body(module, my_sig, body)?;
let existing_body = module.funcs[func_idx].body_mut().unwrap();
*existing_body = body;
module.funcs[func_idx] = FuncDecl::Lazy(my_sig, body);
}
Payload::ExportSection(reader) => {
for export in reader {
@ -269,10 +266,10 @@ fn handle_payload<'a>(
Ok(())
}
fn parse_body<'a>(
pub(crate) fn parse_body<'a>(
module: &'a Module,
my_sig: Signature,
body: wasmparser::FunctionBody,
body: &mut wasmparser::FunctionBody,
) -> Result<FunctionBody> {
let mut ret: FunctionBody = FunctionBody::default();

12
src/ir/display.rs
View file

@ -190,14 +190,10 @@ impl<'a> Display for ModuleDisplay<'a> {
for seg in &memory_data.segments {
writeln!(
f,
" {} offset {}: [{}]",
" {} offset {}: # {} bytes",
memory,
seg.offset,
seg.data
.iter()
.map(|&byte| format!("0x{:02x}", byte))
.collect::<Vec<_>>()
.join(", ")
seg.data.len()
)?;
}
}
@ -217,6 +213,10 @@ impl<'a> Display for ModuleDisplay<'a> {
writeln!(f, " {}: {} = # {}", func, sig, sig_strs.get(&sig).unwrap())?;
writeln!(f, "{}", body.display(" "))?;
}
FuncDecl::Lazy(sig, reader) => {
writeln!(f, " {}: {} = # {}", func, sig, sig_strs.get(&sig).unwrap())?;
writeln!(f, " # raw bytes (length {})", reader.range().len())?;
}
FuncDecl::Import(sig) => {
writeln!(f, " {}: {} # {}", func, sig, sig_strs.get(&sig).unwrap())?;
}

66
src/ir/func.rs
View file

@ -1,21 +1,54 @@
use super::{Block, FunctionBodyDisplay, Local, Module, Signature, Type, Value, ValueDef};
use crate::cfg::CFGInfo;
use crate::entity::{EntityRef, EntityVec, PerEntity};
use crate::frontend::parse_body;
use anyhow::Result;
#[derive(Clone, Debug)]
pub enum FuncDecl {
pub enum FuncDecl<'a> {
Import(Signature),
Lazy(Signature, wasmparser::FunctionBody<'a>),
Body(Signature, FunctionBody),
}
impl FuncDecl {
impl<'a> FuncDecl<'a> {
pub fn sig(&self) -> Signature {
match self {
FuncDecl::Import(sig) => *sig,
FuncDecl::Lazy(sig, ..) => *sig,
FuncDecl::Body(sig, ..) => *sig,
}
}
pub fn parse(&mut self, module: &Module) -> Result<()> {
match self {
FuncDecl::Lazy(sig, body) => {
let body = parse_body(module, *sig, body)?;
*self = FuncDecl::Body(*sig, body);
Ok(())
}
_ => Ok(()),
}
}
pub fn optimize(&mut self) {
match self {
FuncDecl::Body(_, body) => {
body.optimize();
}
_ => {}
}
}
pub fn convert_to_max_ssa(&mut self) {
match self {
FuncDecl::Body(_, body) => {
body.convert_to_max_ssa();
}
_ => {}
}
}
pub fn body(&self) -> Option<&FunctionBody> {
match self {
FuncDecl::Body(_, body) => Some(body),
@ -78,6 +111,18 @@ impl FunctionBody {
}
}
pub fn optimize(&mut self) {
let cfg = crate::cfg::CFGInfo::new(self);
crate::passes::basic_opt::gvn(self, &cfg);
crate::passes::resolve_aliases::run(self);
crate::passes::empty_blocks::run(self);
}
pub fn convert_to_max_ssa(&mut self) {
let cfg = crate::cfg::CFGInfo::new(self);
crate::passes::maxssa::run(self, &cfg);
}
pub fn add_block(&mut self) -> Block {
let id = self.blocks.push(BlockDef::default());
log::trace!("add_block: block {}", id);
@ -94,6 +139,23 @@ impl FunctionBody {
log::trace!("add_edge: from {} to {}", from, to);
}
pub fn recompute_edges(&mut self) {
for block in self.blocks.values_mut() {
block.preds.clear();
block.succs.clear();
block.pos_in_succ_pred.clear();
block.pos_in_pred_succ.clear();
}
for block in 0..self.blocks.len() {
let block = Block::new(block);
let terminator = self.blocks[block].terminator.clone();
terminator.visit_successors(|succ| {
self.add_edge(block, succ);
});
}
}
pub fn add_value(&mut self, value: ValueDef) -> Value {
log::trace!("add_value: def {:?}", value);
let value = self.values.push(value);

29
src/ir/module.rs
View file

@ -1,5 +1,5 @@
use super::{Func, FuncDecl, Global, Memory, ModuleDisplay, Signature, Table, Type};
use crate::entity::EntityVec;
use crate::entity::{EntityRef, EntityVec};
use crate::ir::FunctionBody;
use crate::{backend, frontend};
use anyhow::Result;
@ -7,7 +7,7 @@ use anyhow::Result;
#[derive(Clone, Debug)]
pub struct Module<'a> {
pub orig_bytes: &'a [u8],
pub funcs: EntityVec<Func, FuncDecl>,
pub funcs: EntityVec<Func, FuncDecl<'a>>,
pub signatures: EntityVec<Signature, SignatureData>,
pub globals: EntityVec<Global, GlobalData>,
pub tables: EntityVec<Table, TableData>,
@ -171,19 +171,22 @@ impl<'a> Module<'a> {
}
}
pub fn optimize(&mut self) {
self.per_func_body(|body| {
let cfg = crate::cfg::CFGInfo::new(body);
crate::passes::basic_opt::gvn(body, &cfg);
crate::passes::resolve_aliases::run(body);
});
pub fn expand_func<'b>(&'b mut self, id: Func) -> Result<&'b mut FuncDecl<'a>> {
if let FuncDecl::Lazy(..) = self.funcs[id] {
// End the borrow. This is cheap (a slice copy).
let mut func = self.funcs[id].clone();
func.parse(self)?;
self.funcs[id] = func;
}
Ok(&mut self.funcs[id])
}
pub fn convert_to_max_ssa(&mut self) {
self.per_func_body(|body| {
let cfg = crate::cfg::CFGInfo::new(body);
crate::passes::maxssa::run(body, &cfg);
});
pub fn expand_all_funcs(&mut self) -> Result<()> {
for id in 0..self.funcs.len() {
let id = Func::new(id);
self.expand_func(id)?;
}
Ok(())
}
pub fn display<'b>(&'b self) -> ModuleDisplay<'b>

1
src/passes.rs
View file

@ -2,5 +2,6 @@
pub mod basic_opt;
pub mod dom_pass;
pub mod empty_blocks;
pub mod maxssa;
pub mod resolve_aliases;

128
src/passes/empty_blocks.rs Normal file
View file

@ -0,0 +1,128 @@
//! Pass to remove empty blocks.
use crate::entity::EntityRef;
use crate::ir::{Block, BlockTarget, FunctionBody, Terminator, Value, ValueDef};
use std::borrow::Cow;
use std::collections::HashSet;
#[derive(Clone, Debug)]
struct Forwarding {
to: Block,
args: Vec<ForwardingArg>,
}
#[derive(Clone, Copy, Debug)]
enum ForwardingArg {
BlockParam(usize),
Value(Value),
}
impl Forwarding {
fn compose(a: &Forwarding, b: &Forwarding) -> Forwarding {
// `b` should be the target of `a.to`, but we can't assert
// that here. The composed target is thus `b.to`.
let to = b.to;
// For each arg in `b.args`, evaluate, replacing any
// `BlockParam` with the corresponding value from `a.args`.
let args = b
.args
.iter()
.map(|&arg| match arg {
ForwardingArg::BlockParam(idx) => a.args[idx].clone(),
ForwardingArg::Value(v) => ForwardingArg::Value(v),
})
.collect::<Vec<_>>();
Forwarding { to, args }
}
}
fn block_to_forwarding(body: &FunctionBody, block: Block) -> Option<Forwarding> {
// Must be empty except for terminator, and must have an
// unconditional-branch terminator.
if body.blocks[block].insts.len() > 0 {
return None;
}
let target = match &body.blocks[block].terminator {
&Terminator::Br { ref target } => target,
_ => return None,
};
// If conditions met, then gather ForwardingArgs.
let args = target
.args
.iter()
.map(|&arg| {
let arg = body.resolve_alias(arg);
match &body.values[arg] {
&ValueDef::BlockParam(param_block, index, _) if param_block == block => {
ForwardingArg::BlockParam(index)
}
_ => ForwardingArg::Value(arg),
}
})
.collect::<Vec<_>>();
Some(Forwarding {
to: target.block,
args,
})
}
fn rewrite_target(forwardings: &[Option<Forwarding>], target: &BlockTarget) -> Option<BlockTarget> {
if !forwardings[target.block.index()].is_some() {
return None;
}
let mut forwarding = Cow::Borrowed(forwardings[target.block.index()].as_ref().unwrap());
let mut seen = HashSet::new();
while forwardings[forwarding.to.index()].is_some() && seen.insert(forwarding.to.index()) {
forwarding = Cow::Owned(Forwarding::compose(
&forwarding,
forwardings[forwarding.to.index()].as_ref().unwrap(),
));
}
let args = forwarding
.args
.iter()
.map(|arg| match arg {
&ForwardingArg::Value(v) => v,
&ForwardingArg::BlockParam(idx) => target.args[idx],
})
.collect::<Vec<_>>();
Some(BlockTarget {
block: forwarding.to,
args,
})
}
pub fn run(body: &mut FunctionBody) {
// Identify empty blocks, and to where they should forward.
let forwardings = body
.blocks
.iter()
.map(|block| {
if block != body.entry {
block_to_forwarding(body, block)
} else {
None
}
})
.collect::<Vec<_>>();
// Rewrite every target according to a forwarding (or potentially
// a chain of composed forwardings).
for block_data in body.blocks.values_mut() {
block_data.terminator.update_targets(|target| {
if let Some(new_target) = rewrite_target(&forwardings[..], target) {
*target = new_target;
}
});
}
// Recompute preds/succs.
body.recompute_edges();
}