From 1cca77ec37fe2d1e344ecc338bac4d69b8119f21 Mon Sep 17 00:00:00 2001
From: Chris Fallin <chris@cfallin.org>
Date: Sat, 13 Nov 2021 00:52:35 -0800
Subject: [PATCH] WIP: command line utility and start of Wasm-parsing frontend.

---
 Cargo.toml             |  6 +++++-
 src/bin/waffle-util.rs | 47 ++++++++++++++++++++++++++++++++++++++++++
 src/frontend.rs        | 39 +++++++++++++++++++++++++++++++++++
 src/ir.rs              | 39 ++++++++++++++++++++++++++++++++---
 src/lib.rs             |  3 ++-
 5 files changed, 129 insertions(+), 5 deletions(-)
 create mode 100644 src/bin/waffle-util.rs
 create mode 100644 src/frontend.rs

diff --git a/Cargo.toml b/Cargo.toml
index a9c1fed..b29e6c8 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -2,10 +2,14 @@
 name = "waffle"
 version = "0.0.1"
 description = "Wasm Analysis Framework For Lightweight Experiments"
-author = ["Chris Fallin <chris@cfallin.org>"]
+authors = ["Chris Fallin <chris@cfallin.org>"]
 license = "Apache-2.0 WITH LLVM-exception"
 edition = "2018"
 
 [dependencies]
 wasmparser = "0.81"
 wasm-encoder = "0.3"
+anyhow = "1.0"
+structopt = "0.3"
+log = "0.4"
+env_logger = "0.9"
\ No newline at end of file
diff --git a/src/bin/waffle-util.rs b/src/bin/waffle-util.rs
new file mode 100644
index 0000000..eec4fb4
--- /dev/null
+++ b/src/bin/waffle-util.rs
@@ -0,0 +1,47 @@
+//! WAFFLE command-line tool.
+
+use anyhow::Result;
+use log::debug;
+use std::path::PathBuf;
+use structopt::StructOpt;
+use waffle::frontend;
+
+#[derive(Debug, StructOpt)]
+#[structopt(name = "waffle-util", about = "WAFFLE utility.")]
+struct Options {
+    #[structopt(short, long)]
+    debug: bool,
+
+    #[structopt(subcommand)]
+    command: Command,
+}
+
+#[derive(Debug, StructOpt)]
+enum Command {
+    #[structopt(name = "print-ir", about = "Parse Wasm and print resulting IR")]
+    PrintIR {
+        #[structopt(help = "Wasm file to parse")]
+        wasm: PathBuf,
+    },
+}
+
+fn main() -> Result<()> {
+    let opts = Options::from_args();
+
+    let mut logger = env_logger::Builder::from_default_env();
+    if opts.debug {
+        logger.filter_level(log::LevelFilter::Debug);
+    }
+    let _ = logger.try_init();
+
+    match opts.command {
+        Command::PrintIR { wasm } => {
+            let bytes = std::fs::read(wasm)?;
+            debug!("Loaded {} bytes of Wasm data", bytes.len());
+            let module = frontend::wasm_to_ir(&bytes[..])?;
+            println!("{:?}", module);
+        }
+    }
+
+    Ok(())
+}
diff --git a/src/frontend.rs b/src/frontend.rs
new file mode 100644
index 0000000..8670bee
--- /dev/null
+++ b/src/frontend.rs
@@ -0,0 +1,39 @@
+//! Frontend: convert Wasm to IR.
+
+use crate::ir::*;
+use anyhow::{bail, Result};
+use log::debug;
+use wasmparser::{Chunk, Parser, Payload, TypeDef};
+
+pub fn wasm_to_ir(bytes: &[u8]) -> Result<Module> {
+    let mut module = Module::default();
+    let mut parser = Parser::new(0);
+    for chunk in parser.parse(bytes, /* eof = */ true) {
+        match chunk {
+            Chunk::NeedMoreData(_) => bail!("Unexpected EOF in Wasm"),
+            Chunk::Parsed { payload, .. } => handle_payload(&mut module, payload)?,
+        }
+    }
+
+    Ok(module)
+}
+
+fn handle_payload<'a>(module: &mut Module, payload: Payload<'a>) -> Result<()> {
+    debug!("Wasm parser item: {:?}", payload);
+    match payload {
+        Payload::TypeSection(mut reader) => {
+            for _ in 0..reader.get_count() {
+                let ty = reader.read()?;
+                match ty {
+                    TypeDef::Func(fty) => {
+                        module.signatures.push(fty);
+                    }
+                    _ => {}
+                }
+            }
+        }
+        _ => {}
+    }
+
+    Ok(())
+}
diff --git a/src/ir.rs b/src/ir.rs
index 75329de..696bee1 100644
--- a/src/ir.rs
+++ b/src/ir.rs
@@ -1,13 +1,14 @@
 //! Intermediate representation for Wasm.
 
-use wasmparser::{FuncType, Type};
+use wasmparser::{FuncType, Operator, Type};
 
 pub type SignatureId = usize;
 pub type FuncId = usize;
 pub type BlockId = usize;
 pub type InstId = usize;
+pub type ValueId = usize;
 
-#[derive(Clone, Debug)]
+#[derive(Clone, Debug, Default)]
 pub struct Module {
     pub funcs: Vec<FuncDecl>,
     pub signatures: Vec<FuncType>,
@@ -19,8 +20,40 @@ pub enum FuncDecl {
     Body(SignatureId, FunctionBody),
 }
 
-#[derive(Clone, Debug)]
+#[derive(Clone, Debug, Default)]
 pub struct FunctionBody {
     pub locals: Vec<Type>,
     pub blocks: Vec<Block>,
+    pub values: Vec<ValueDef>,
+}
+
+#[derive(Clone, Debug)]
+pub struct ValueDef {
+    pub kind: ValueKind,
+    pub ty: Type,
+}
+
+#[derive(Clone, Debug)]
+pub enum ValueKind {
+    BlockParam(Block),
+    Inst(Block, Inst),
+}
+
+#[derive(Clone, Debug, Default)]
+pub struct Block {
+    pub params: Vec<Type>,
+    pub insts: Vec<Inst>,
+}
+
+#[derive(Clone, Debug)]
+pub struct Inst {
+    pub operator: Operator<'static>,
+    pub outputs: Vec<ValueId>,
+    pub inputs: Vec<Operand>,
+}
+
+#[derive(Clone, Debug)]
+pub enum Operand {
+    Value(ValueId),
+    Sub(Box<Inst>),
 }
diff --git a/src/lib.rs b/src/lib.rs
index 01e38da..7a84d4b 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -2,7 +2,8 @@
 
 // Re-export wasmparser and wasmencoder for easier use of the right
 // version by our embedders.
-pub use wasmparser;
 pub use wasm_encoder;
+pub use wasmparser;
 
+pub mod frontend;
 pub mod ir;