diff options
Diffstat (limited to 'third_party/rust/wasmparser/src')
38 files changed, 18682 insertions, 0 deletions
diff --git a/third_party/rust/wasmparser/src/binary_reader.rs b/third_party/rust/wasmparser/src/binary_reader.rs new file mode 100644 index 0000000000..43fef14cdd --- /dev/null +++ b/third_party/rust/wasmparser/src/binary_reader.rs @@ -0,0 +1,1682 @@ +/* Copyright 2018 Mozilla Foundation + * + * 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. + */ + +use crate::{limits::*, *}; +use std::convert::TryInto; +use std::error::Error; +use std::fmt; +use std::marker; +use std::ops::Range; +use std::str; + +const WASM_MAGIC_NUMBER: &[u8; 4] = b"\0asm"; + +/// A binary reader for WebAssembly modules. +#[derive(Debug, Clone)] +pub struct BinaryReaderError { + // Wrap the actual error data in a `Box` so that the error is just one + // word. This means that we can continue returning small `Result`s in + // registers. + pub(crate) inner: Box<BinaryReaderErrorInner>, +} + +#[derive(Debug, Clone)] +pub(crate) struct BinaryReaderErrorInner { + pub(crate) message: String, + pub(crate) offset: usize, + pub(crate) needed_hint: Option<usize>, +} + +/// The result for `BinaryReader` operations. +pub type Result<T, E = BinaryReaderError> = std::result::Result<T, E>; + +impl Error for BinaryReaderError {} + +impl fmt::Display for BinaryReaderError { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + write!( + f, + "{} (at offset 0x{:x})", + self.inner.message, self.inner.offset + ) + } +} + +impl BinaryReaderError { + #[cold] + pub(crate) fn new(message: impl Into<String>, offset: usize) -> Self { + let message = message.into(); + BinaryReaderError { + inner: Box::new(BinaryReaderErrorInner { + message, + offset, + needed_hint: None, + }), + } + } + + #[cold] + pub(crate) fn fmt(args: fmt::Arguments<'_>, offset: usize) -> Self { + BinaryReaderError::new(args.to_string(), offset) + } + + #[cold] + pub(crate) fn eof(offset: usize, needed_hint: usize) -> Self { + BinaryReaderError { + inner: Box::new(BinaryReaderErrorInner { + message: "unexpected end-of-file".to_string(), + offset, + needed_hint: Some(needed_hint), + }), + } + } + + /// Get this error's message. + pub fn message(&self) -> &str { + &self.inner.message + } + + /// Get the offset within the Wasm binary where the error occurred. + pub fn offset(&self) -> usize { + self.inner.offset + } +} + +/// A binary reader of the WebAssembly structures and types. +#[derive(Clone, Debug, Hash)] +pub struct BinaryReader<'a> { + pub(crate) buffer: &'a [u8], + pub(crate) position: usize, + original_offset: usize, + allow_memarg64: bool, +} + +impl<'a> BinaryReader<'a> { + /// Constructs `BinaryReader` type. + /// + /// # Examples + /// ``` + /// let fn_body = &vec![0x41, 0x00, 0x10, 0x00, 0x0B]; + /// let mut reader = wasmparser::BinaryReader::new(fn_body); + /// while !reader.eof() { + /// let op = reader.read_operator(); + /// println!("{:?}", op) + /// } + /// ``` + pub fn new(data: &[u8]) -> BinaryReader { + BinaryReader { + buffer: data, + position: 0, + original_offset: 0, + allow_memarg64: false, + } + } + + /// Constructs a `BinaryReader` with an explicit starting offset. + pub fn new_with_offset(data: &[u8], original_offset: usize) -> BinaryReader { + BinaryReader { + buffer: data, + position: 0, + original_offset, + allow_memarg64: false, + } + } + + /// Gets the original position of the binary reader. + #[inline] + pub fn original_position(&self) -> usize { + self.original_offset + self.position + } + + /// Whether or not to allow 64-bit memory arguments in functions. + /// + /// This is intended to be `true` when support for the memory64 + /// WebAssembly proposal is also enabled. + pub fn allow_memarg64(&mut self, allow: bool) { + self.allow_memarg64 = allow; + } + + /// Returns a range from the starting offset to the end of the buffer. + pub fn range(&self) -> Range<usize> { + self.original_offset..self.original_offset + self.buffer.len() + } + + pub(crate) fn remaining_buffer(&self) -> &'a [u8] { + &self.buffer[self.position..] + } + + fn ensure_has_byte(&self) -> Result<()> { + if self.position < self.buffer.len() { + Ok(()) + } else { + Err(BinaryReaderError::eof(self.original_position(), 1)) + } + } + + pub(crate) fn ensure_has_bytes(&self, len: usize) -> Result<()> { + if self.position + len <= self.buffer.len() { + Ok(()) + } else { + let hint = self.position + len - self.buffer.len(); + Err(BinaryReaderError::eof(self.original_position(), hint)) + } + } + + /// Reads a value of type `T` from this binary reader, advancing the + /// internal position in this reader forward as data is read. + #[inline] + pub fn read<T>(&mut self) -> Result<T> + where + T: FromReader<'a>, + { + T::from_reader(self) + } + + pub(crate) fn read_u7(&mut self) -> Result<u8> { + let b = self.read_u8()?; + if (b & 0x80) != 0 { + return Err(BinaryReaderError::new( + "invalid u7", + self.original_position() - 1, + )); + } + Ok(b) + } + + pub(crate) fn external_kind_from_byte(byte: u8, offset: usize) -> Result<ExternalKind> { + match byte { + 0x00 => Ok(ExternalKind::Func), + 0x01 => Ok(ExternalKind::Table), + 0x02 => Ok(ExternalKind::Memory), + 0x03 => Ok(ExternalKind::Global), + 0x04 => Ok(ExternalKind::Tag), + x => Err(Self::invalid_leading_byte_error(x, "external kind", offset)), + } + } + + /// Reads a variable-length 32-bit size from the byte stream while checking + /// against a limit. + pub fn read_size(&mut self, limit: usize, desc: &str) -> Result<usize> { + let pos = self.original_position(); + let size = self.read_var_u32()? as usize; + if size > limit { + bail!(pos, "{desc} size is out of bounds"); + } + Ok(size) + } + + /// Reads a variable-length 32-bit size from the byte stream while checking + /// against a limit. + /// + /// Then reads that many values of type `T` and returns them as an iterator. + /// + /// Note that regardless of how many items are read from the returned + /// iterator the items will still be parsed from this reader. + pub fn read_iter<'me, T>( + &'me mut self, + limit: usize, + desc: &str, + ) -> Result<BinaryReaderIter<'a, 'me, T>> + where + T: FromReader<'a>, + { + let size = self.read_size(limit, desc)?; + Ok(BinaryReaderIter { + remaining: size, + reader: self, + _marker: marker::PhantomData, + }) + } + + fn read_first_byte_and_var_u32(&mut self) -> Result<(u8, u32)> { + let pos = self.position; + let val = self.read_var_u32()?; + Ok((self.buffer[pos], val)) + } + + fn read_memarg(&mut self, max_align: u8) -> Result<MemArg> { + let flags_pos = self.original_position(); + let mut flags = self.read_var_u32()?; + let memory = if flags & (1 << 6) != 0 { + flags ^= 1 << 6; + self.read_var_u32()? + } else { + 0 + }; + let align = if flags >= (1 << 6) { + return Err(BinaryReaderError::new("alignment too large", flags_pos)); + } else { + flags as u8 + }; + let offset = if self.allow_memarg64 { + self.read_var_u64()? + } else { + u64::from(self.read_var_u32()?) + }; + Ok(MemArg { + align, + max_align, + offset, + memory, + }) + } + + fn read_br_table(&mut self) -> Result<BrTable<'a>> { + let cnt = self.read_size(MAX_WASM_BR_TABLE_SIZE, "br_table")?; + let start = self.position; + for _ in 0..cnt { + self.read_var_u32()?; + } + let end = self.position; + let default = self.read_var_u32()?; + Ok(BrTable { + reader: BinaryReader::new_with_offset(&self.buffer[start..end], start), + cnt: cnt as u32, + default, + }) + } + + /// Returns whether the `BinaryReader` has reached the end of the file. + #[inline] + pub fn eof(&self) -> bool { + self.position >= self.buffer.len() + } + + /// Returns the `BinaryReader`'s current position. + #[inline] + pub fn current_position(&self) -> usize { + self.position + } + + /// Returns the number of bytes remaining in the `BinaryReader`. + #[inline] + pub fn bytes_remaining(&self) -> usize { + self.buffer.len() - self.position + } + + /// Advances the `BinaryReader` `size` bytes, and returns a slice from the + /// current position of `size` length. + /// + /// # Errors + /// If `size` exceeds the remaining length in `BinaryReader`. + pub fn read_bytes(&mut self, size: usize) -> Result<&'a [u8]> { + self.ensure_has_bytes(size)?; + let start = self.position; + self.position += size; + Ok(&self.buffer[start..self.position]) + } + + /// Reads a length-prefixed list of bytes from this reader and returns a + /// new `BinaryReader` to read that list of bytes. + /// + /// Advances the position of this reader by the number of bytes read. + pub fn read_reader(&mut self, err: &str) -> Result<BinaryReader<'a>> { + let size = self.read_var_u32()? as usize; + let body_start = self.position; + let buffer = match self.buffer.get(self.position..).and_then(|s| s.get(..size)) { + Some(buf) => buf, + None => { + return Err(BinaryReaderError::new( + err, + self.original_offset + self.buffer.len(), + )) + } + }; + self.position += size; + Ok(BinaryReader::new_with_offset( + buffer, + self.original_offset + body_start, + )) + } + + /// Advances the `BinaryReader` four bytes and returns a `u32`. + /// # Errors + /// If `BinaryReader` has less than four bytes remaining. + pub fn read_u32(&mut self) -> Result<u32> { + self.ensure_has_bytes(4)?; + let word = u32::from_le_bytes( + self.buffer[self.position..self.position + 4] + .try_into() + .unwrap(), + ); + self.position += 4; + Ok(word) + } + + /// Advances the `BinaryReader` eight bytes and returns a `u64`. + /// # Errors + /// If `BinaryReader` has less than eight bytes remaining. + pub fn read_u64(&mut self) -> Result<u64> { + self.ensure_has_bytes(8)?; + let word = u64::from_le_bytes( + self.buffer[self.position..self.position + 8] + .try_into() + .unwrap(), + ); + self.position += 8; + Ok(word) + } + + /// Advances the `BinaryReader` a single byte. + /// + /// # Errors + /// + /// If `BinaryReader` has no bytes remaining. + #[inline] + pub fn read_u8(&mut self) -> Result<u8> { + let b = match self.buffer.get(self.position) { + Some(b) => *b, + None => return Err(self.eof_err()), + }; + self.position += 1; + Ok(b) + } + + #[cold] + fn eof_err(&self) -> BinaryReaderError { + BinaryReaderError::eof(self.original_position(), 1) + } + + /// Advances the `BinaryReader` up to four bytes to parse a variable + /// length integer as a `u32`. + /// + /// # Errors + /// + /// If `BinaryReader` has less than one or up to four bytes remaining, or + /// the integer is larger than 32 bits. + #[inline] + pub fn read_var_u32(&mut self) -> Result<u32> { + // Optimization for single byte i32. + let byte = self.read_u8()?; + if (byte & 0x80) == 0 { + Ok(u32::from(byte)) + } else { + self.read_var_u32_big(byte) + } + } + + fn read_var_u32_big(&mut self, byte: u8) -> Result<u32> { + let mut result = (byte & 0x7F) as u32; + let mut shift = 7; + loop { + let byte = self.read_u8()?; + result |= ((byte & 0x7F) as u32) << shift; + if shift >= 25 && (byte >> (32 - shift)) != 0 { + let msg = if byte & 0x80 != 0 { + "invalid var_u32: integer representation too long" + } else { + "invalid var_u32: integer too large" + }; + // The continuation bit or unused bits are set. + return Err(BinaryReaderError::new(msg, self.original_position() - 1)); + } + shift += 7; + if (byte & 0x80) == 0 { + break; + } + } + Ok(result) + } + + /// Advances the `BinaryReader` up to four bytes to parse a variable + /// length integer as a `u64`. + /// + /// # Errors + /// + /// If `BinaryReader` has less than one or up to eight bytes remaining, or + /// the integer is larger than 64 bits. + #[inline] + pub fn read_var_u64(&mut self) -> Result<u64> { + // Optimization for single byte u64. + let byte = u64::from(self.read_u8()?); + if (byte & 0x80) == 0 { + Ok(byte) + } else { + self.read_var_u64_big(byte) + } + } + + fn read_var_u64_big(&mut self, byte: u64) -> Result<u64> { + let mut result = byte & 0x7F; + let mut shift = 7; + loop { + let byte = u64::from(self.read_u8()?); + result |= (byte & 0x7F) << shift; + if shift >= 57 && (byte >> (64 - shift)) != 0 { + let msg = if byte & 0x80 != 0 { + "invalid var_u64: integer representation too long" + } else { + "invalid var_u64: integer too large" + }; + // The continuation bit or unused bits are set. + return Err(BinaryReaderError::new(msg, self.original_position() - 1)); + } + shift += 7; + if (byte & 0x80) == 0 { + break; + } + } + Ok(result) + } + + /// Executes `f` to skip some data in this binary reader and then returns a + /// reader which will read the skipped data. + pub fn skip(&mut self, f: impl FnOnce(&mut Self) -> Result<()>) -> Result<Self> { + let start = self.position; + f(self)?; + Ok(BinaryReader::new_with_offset( + &self.buffer[start..self.position], + self.original_offset + start, + )) + } + + /// Advances the `BinaryReader` past a WebAssembly string. This method does + /// not perform any utf-8 validation. + /// # Errors + /// If `BinaryReader` has less than four bytes, the string's length exceeds + /// the remaining bytes, or the string length + /// exceeds `limits::MAX_WASM_STRING_SIZE`. + pub fn skip_string(&mut self) -> Result<()> { + let len = self.read_var_u32()? as usize; + if len > MAX_WASM_STRING_SIZE { + return Err(BinaryReaderError::new( + "string size out of bounds", + self.original_position() - 1, + )); + } + self.ensure_has_bytes(len)?; + self.position += len; + Ok(()) + } + + /// Advances the `BinaryReader` up to four bytes to parse a variable + /// length integer as a `i32`. + /// # Errors + /// If `BinaryReader` has less than one or up to four bytes remaining, or + /// the integer is larger than 32 bits. + #[inline] + pub fn read_var_i32(&mut self) -> Result<i32> { + // Optimization for single byte i32. + let byte = self.read_u8()?; + if (byte & 0x80) == 0 { + Ok(((byte as i32) << 25) >> 25) + } else { + self.read_var_i32_big(byte) + } + } + + fn read_var_i32_big(&mut self, byte: u8) -> Result<i32> { + let mut result = (byte & 0x7F) as i32; + let mut shift = 7; + loop { + let byte = self.read_u8()?; + result |= ((byte & 0x7F) as i32) << shift; + if shift >= 25 { + let continuation_bit = (byte & 0x80) != 0; + let sign_and_unused_bit = (byte << 1) as i8 >> (32 - shift); + if continuation_bit || (sign_and_unused_bit != 0 && sign_and_unused_bit != -1) { + let msg = if continuation_bit { + "invalid var_i32: integer representation too long" + } else { + "invalid var_i32: integer too large" + }; + return Err(BinaryReaderError::new(msg, self.original_position() - 1)); + } + return Ok(result); + } + shift += 7; + if (byte & 0x80) == 0 { + break; + } + } + let ashift = 32 - shift; + Ok((result << ashift) >> ashift) + } + + /// Advances the `BinaryReader` up to four bytes to parse a variable + /// length integer as a signed 33 bit integer, returned as a `i64`. + /// # Errors + /// If `BinaryReader` has less than one or up to five bytes remaining, or + /// the integer is larger than 33 bits. + pub fn read_var_s33(&mut self) -> Result<i64> { + // Optimization for single byte. + let byte = self.read_u8()?; + if (byte & 0x80) == 0 { + return Ok(((byte as i8) << 1) as i64 >> 1); + } + + let mut result = (byte & 0x7F) as i64; + let mut shift = 7; + loop { + let byte = self.read_u8()?; + result |= ((byte & 0x7F) as i64) << shift; + if shift >= 25 { + let continuation_bit = (byte & 0x80) != 0; + let sign_and_unused_bit = (byte << 1) as i8 >> (33 - shift); + if continuation_bit || (sign_and_unused_bit != 0 && sign_and_unused_bit != -1) { + return Err(BinaryReaderError::new( + "invalid var_s33: integer representation too long", + self.original_position() - 1, + )); + } + return Ok(result); + } + shift += 7; + if (byte & 0x80) == 0 { + break; + } + } + let ashift = 64 - shift; + Ok((result << ashift) >> ashift) + } + + /// Advances the `BinaryReader` up to eight bytes to parse a variable + /// length integer as a 64 bit integer, returned as a `i64`. + /// # Errors + /// If `BinaryReader` has less than one or up to eight bytes remaining, or + /// the integer is larger than 64 bits. + pub fn read_var_i64(&mut self) -> Result<i64> { + let mut result: i64 = 0; + let mut shift = 0; + loop { + let byte = self.read_u8()?; + result |= i64::from(byte & 0x7F) << shift; + if shift >= 57 { + let continuation_bit = (byte & 0x80) != 0; + let sign_and_unused_bit = ((byte << 1) as i8) >> (64 - shift); + if continuation_bit || (sign_and_unused_bit != 0 && sign_and_unused_bit != -1) { + let msg = if continuation_bit { + "invalid var_i64: integer representation too long" + } else { + "invalid var_i64: integer too large" + }; + return Err(BinaryReaderError::new(msg, self.original_position() - 1)); + } + return Ok(result); + } + shift += 7; + if (byte & 0x80) == 0 { + break; + } + } + let ashift = 64 - shift; + Ok((result << ashift) >> ashift) + } + + /// Advances the `BinaryReader` up to four bytes to parse a variable + /// length integer as a 32 bit floating point integer, returned as `Ieee32`. + /// # Errors + /// If `BinaryReader` has less than one or up to four bytes remaining, or + /// the integer is larger than 32 bits. + pub fn read_f32(&mut self) -> Result<Ieee32> { + let value = self.read_u32()?; + Ok(Ieee32(value)) + } + + /// Advances the `BinaryReader` up to four bytes to parse a variable + /// length integer as a 32 bit floating point integer, returned as `Ieee32`. + /// # Errors + /// If `BinaryReader` has less than one or up to four bytes remaining, or + /// the integer is larger than 32 bits. + pub fn read_f64(&mut self) -> Result<Ieee64> { + let value = self.read_u64()?; + Ok(Ieee64(value)) + } + + /// Reads a WebAssembly string from the module. + /// # Errors + /// If `BinaryReader` has less than up to four bytes remaining, the string's + /// length exceeds the remaining bytes, the string's length exceeds + /// `limits::MAX_WASM_STRING_SIZE`, or the string contains invalid utf-8. + pub fn read_string(&mut self) -> Result<&'a str> { + let len = self.read_var_u32()? as usize; + if len > MAX_WASM_STRING_SIZE { + return Err(BinaryReaderError::new( + "string size out of bounds", + self.original_position() - 1, + )); + } + let bytes = self.read_bytes(len)?; + str::from_utf8(bytes).map_err(|_| { + BinaryReaderError::new("invalid UTF-8 encoding", self.original_position() - 1) + }) + } + + #[cold] + pub(crate) fn invalid_leading_byte<T>(&self, byte: u8, desc: &str) -> Result<T> { + Err(Self::invalid_leading_byte_error( + byte, + desc, + self.original_position() - 1, + )) + } + + pub(crate) fn invalid_leading_byte_error( + byte: u8, + desc: &str, + offset: usize, + ) -> BinaryReaderError { + format_err!(offset, "invalid leading byte (0x{byte:x}) for {desc}") + } + + pub(crate) fn peek(&self) -> Result<u8> { + self.ensure_has_byte()?; + Ok(self.buffer[self.position]) + } + + fn read_block_type(&mut self) -> Result<BlockType> { + let b = self.peek()?; + + // Check for empty block + if b == 0x40 { + self.position += 1; + return Ok(BlockType::Empty); + } + + // Check for a block type of form [] -> [t]. + if ValType::is_valtype_byte(b) { + return Ok(BlockType::Type(self.read()?)); + } + + // Not empty or a singular type, so read the function type index + let idx = self.read_var_s33()?; + match u32::try_from(idx) { + Ok(idx) => Ok(BlockType::FuncType(idx)), + Err(_) => { + return Err(BinaryReaderError::new( + "invalid function type", + self.original_position(), + )); + } + } + } + + /// Visit the next available operator with the specified [`VisitOperator`] instance. + /// + /// Note that this does not implicitly propagate any additional information such as instruction + /// offsets. In order to do so, consider storing such data within the visitor before visiting. + /// + /// # Errors + /// + /// If `BinaryReader` has less bytes remaining than required to parse the `Operator`. + /// + /// # Examples + /// + /// Store an offset for use in diagnostics or any other purposes: + /// + /// ``` + /// # use wasmparser::{BinaryReader, VisitOperator, Result, for_each_operator}; + /// + /// pub fn dump(mut reader: BinaryReader) -> Result<()> { + /// let mut visitor = Dumper { offset: 0 }; + /// while !reader.eof() { + /// visitor.offset = reader.original_position(); + /// reader.visit_operator(&mut visitor)?; + /// } + /// Ok(()) + /// } + /// + /// struct Dumper { + /// offset: usize + /// } + /// + /// macro_rules! define_visit_operator { + /// ($(@$proposal:ident $op:ident $({ $($arg:ident: $argty:ty),* })? => $visit:ident)*) => { + /// $( + /// fn $visit(&mut self $($(,$arg: $argty)*)?) -> Self::Output { + /// println!("{}: {}", self.offset, stringify!($visit)); + /// } + /// )* + /// } + /// } + /// + /// impl<'a> VisitOperator<'a> for Dumper { + /// type Output = (); + /// for_each_operator!(define_visit_operator); + /// } + /// + /// ``` + pub fn visit_operator<T>(&mut self, visitor: &mut T) -> Result<<T as VisitOperator<'a>>::Output> + where + T: VisitOperator<'a>, + { + let pos = self.original_position(); + let code = self.read_u8()? as u8; + Ok(match code { + 0x00 => visitor.visit_unreachable(), + 0x01 => visitor.visit_nop(), + 0x02 => visitor.visit_block(self.read_block_type()?), + 0x03 => visitor.visit_loop(self.read_block_type()?), + 0x04 => visitor.visit_if(self.read_block_type()?), + 0x05 => visitor.visit_else(), + 0x06 => visitor.visit_try(self.read_block_type()?), + 0x07 => visitor.visit_catch(self.read_var_u32()?), + 0x08 => visitor.visit_throw(self.read_var_u32()?), + 0x09 => visitor.visit_rethrow(self.read_var_u32()?), + 0x0b => visitor.visit_end(), + 0x0c => visitor.visit_br(self.read_var_u32()?), + 0x0d => visitor.visit_br_if(self.read_var_u32()?), + 0x0e => visitor.visit_br_table(self.read_br_table()?), + 0x0f => visitor.visit_return(), + 0x10 => visitor.visit_call(self.read_var_u32()?), + 0x11 => { + let index = self.read_var_u32()?; + let (table_byte, table_index) = self.read_first_byte_and_var_u32()?; + visitor.visit_call_indirect(index, table_index, table_byte) + } + 0x12 => visitor.visit_return_call(self.read_var_u32()?), + 0x13 => visitor.visit_return_call_indirect(self.read_var_u32()?, self.read_var_u32()?), + 0x14 => visitor.visit_call_ref(self.read()?), + 0x15 => visitor.visit_return_call_ref(self.read()?), + 0x18 => visitor.visit_delegate(self.read_var_u32()?), + 0x19 => visitor.visit_catch_all(), + 0x1a => visitor.visit_drop(), + 0x1b => visitor.visit_select(), + 0x1c => { + let results = self.read_var_u32()?; + if results != 1 { + return Err(BinaryReaderError::new( + "invalid result arity", + self.position, + )); + } + visitor.visit_typed_select(self.read()?) + } + + 0x20 => visitor.visit_local_get(self.read_var_u32()?), + 0x21 => visitor.visit_local_set(self.read_var_u32()?), + 0x22 => visitor.visit_local_tee(self.read_var_u32()?), + 0x23 => visitor.visit_global_get(self.read_var_u32()?), + 0x24 => visitor.visit_global_set(self.read_var_u32()?), + 0x25 => visitor.visit_table_get(self.read_var_u32()?), + 0x26 => visitor.visit_table_set(self.read_var_u32()?), + + 0x28 => visitor.visit_i32_load(self.read_memarg(2)?), + 0x29 => visitor.visit_i64_load(self.read_memarg(3)?), + 0x2a => visitor.visit_f32_load(self.read_memarg(2)?), + 0x2b => visitor.visit_f64_load(self.read_memarg(3)?), + 0x2c => visitor.visit_i32_load8_s(self.read_memarg(0)?), + 0x2d => visitor.visit_i32_load8_u(self.read_memarg(0)?), + 0x2e => visitor.visit_i32_load16_s(self.read_memarg(1)?), + 0x2f => visitor.visit_i32_load16_u(self.read_memarg(1)?), + 0x30 => visitor.visit_i64_load8_s(self.read_memarg(0)?), + 0x31 => visitor.visit_i64_load8_u(self.read_memarg(0)?), + 0x32 => visitor.visit_i64_load16_s(self.read_memarg(1)?), + 0x33 => visitor.visit_i64_load16_u(self.read_memarg(1)?), + 0x34 => visitor.visit_i64_load32_s(self.read_memarg(2)?), + 0x35 => visitor.visit_i64_load32_u(self.read_memarg(2)?), + 0x36 => visitor.visit_i32_store(self.read_memarg(2)?), + 0x37 => visitor.visit_i64_store(self.read_memarg(3)?), + 0x38 => visitor.visit_f32_store(self.read_memarg(2)?), + 0x39 => visitor.visit_f64_store(self.read_memarg(3)?), + 0x3a => visitor.visit_i32_store8(self.read_memarg(0)?), + 0x3b => visitor.visit_i32_store16(self.read_memarg(1)?), + 0x3c => visitor.visit_i64_store8(self.read_memarg(0)?), + 0x3d => visitor.visit_i64_store16(self.read_memarg(1)?), + 0x3e => visitor.visit_i64_store32(self.read_memarg(2)?), + 0x3f => { + let (mem_byte, mem) = self.read_first_byte_and_var_u32()?; + visitor.visit_memory_size(mem, mem_byte) + } + 0x40 => { + let (mem_byte, mem) = self.read_first_byte_and_var_u32()?; + visitor.visit_memory_grow(mem, mem_byte) + } + + 0x41 => visitor.visit_i32_const(self.read_var_i32()?), + 0x42 => visitor.visit_i64_const(self.read_var_i64()?), + 0x43 => visitor.visit_f32_const(self.read_f32()?), + 0x44 => visitor.visit_f64_const(self.read_f64()?), + + 0x45 => visitor.visit_i32_eqz(), + 0x46 => visitor.visit_i32_eq(), + 0x47 => visitor.visit_i32_ne(), + 0x48 => visitor.visit_i32_lt_s(), + 0x49 => visitor.visit_i32_lt_u(), + 0x4a => visitor.visit_i32_gt_s(), + 0x4b => visitor.visit_i32_gt_u(), + 0x4c => visitor.visit_i32_le_s(), + 0x4d => visitor.visit_i32_le_u(), + 0x4e => visitor.visit_i32_ge_s(), + 0x4f => visitor.visit_i32_ge_u(), + 0x50 => visitor.visit_i64_eqz(), + 0x51 => visitor.visit_i64_eq(), + 0x52 => visitor.visit_i64_ne(), + 0x53 => visitor.visit_i64_lt_s(), + 0x54 => visitor.visit_i64_lt_u(), + 0x55 => visitor.visit_i64_gt_s(), + 0x56 => visitor.visit_i64_gt_u(), + 0x57 => visitor.visit_i64_le_s(), + 0x58 => visitor.visit_i64_le_u(), + 0x59 => visitor.visit_i64_ge_s(), + 0x5a => visitor.visit_i64_ge_u(), + 0x5b => visitor.visit_f32_eq(), + 0x5c => visitor.visit_f32_ne(), + 0x5d => visitor.visit_f32_lt(), + 0x5e => visitor.visit_f32_gt(), + 0x5f => visitor.visit_f32_le(), + 0x60 => visitor.visit_f32_ge(), + 0x61 => visitor.visit_f64_eq(), + 0x62 => visitor.visit_f64_ne(), + 0x63 => visitor.visit_f64_lt(), + 0x64 => visitor.visit_f64_gt(), + 0x65 => visitor.visit_f64_le(), + 0x66 => visitor.visit_f64_ge(), + 0x67 => visitor.visit_i32_clz(), + 0x68 => visitor.visit_i32_ctz(), + 0x69 => visitor.visit_i32_popcnt(), + 0x6a => visitor.visit_i32_add(), + 0x6b => visitor.visit_i32_sub(), + 0x6c => visitor.visit_i32_mul(), + 0x6d => visitor.visit_i32_div_s(), + 0x6e => visitor.visit_i32_div_u(), + 0x6f => visitor.visit_i32_rem_s(), + 0x70 => visitor.visit_i32_rem_u(), + 0x71 => visitor.visit_i32_and(), + 0x72 => visitor.visit_i32_or(), + 0x73 => visitor.visit_i32_xor(), + 0x74 => visitor.visit_i32_shl(), + 0x75 => visitor.visit_i32_shr_s(), + 0x76 => visitor.visit_i32_shr_u(), + 0x77 => visitor.visit_i32_rotl(), + 0x78 => visitor.visit_i32_rotr(), + 0x79 => visitor.visit_i64_clz(), + 0x7a => visitor.visit_i64_ctz(), + 0x7b => visitor.visit_i64_popcnt(), + 0x7c => visitor.visit_i64_add(), + 0x7d => visitor.visit_i64_sub(), + 0x7e => visitor.visit_i64_mul(), + 0x7f => visitor.visit_i64_div_s(), + 0x80 => visitor.visit_i64_div_u(), + 0x81 => visitor.visit_i64_rem_s(), + 0x82 => visitor.visit_i64_rem_u(), + 0x83 => visitor.visit_i64_and(), + 0x84 => visitor.visit_i64_or(), + 0x85 => visitor.visit_i64_xor(), + 0x86 => visitor.visit_i64_shl(), + 0x87 => visitor.visit_i64_shr_s(), + 0x88 => visitor.visit_i64_shr_u(), + 0x89 => visitor.visit_i64_rotl(), + 0x8a => visitor.visit_i64_rotr(), + 0x8b => visitor.visit_f32_abs(), + 0x8c => visitor.visit_f32_neg(), + 0x8d => visitor.visit_f32_ceil(), + 0x8e => visitor.visit_f32_floor(), + 0x8f => visitor.visit_f32_trunc(), + 0x90 => visitor.visit_f32_nearest(), + 0x91 => visitor.visit_f32_sqrt(), + 0x92 => visitor.visit_f32_add(), + 0x93 => visitor.visit_f32_sub(), + 0x94 => visitor.visit_f32_mul(), + 0x95 => visitor.visit_f32_div(), + 0x96 => visitor.visit_f32_min(), + 0x97 => visitor.visit_f32_max(), + 0x98 => visitor.visit_f32_copysign(), + 0x99 => visitor.visit_f64_abs(), + 0x9a => visitor.visit_f64_neg(), + 0x9b => visitor.visit_f64_ceil(), + 0x9c => visitor.visit_f64_floor(), + 0x9d => visitor.visit_f64_trunc(), + 0x9e => visitor.visit_f64_nearest(), + 0x9f => visitor.visit_f64_sqrt(), + 0xa0 => visitor.visit_f64_add(), + 0xa1 => visitor.visit_f64_sub(), + 0xa2 => visitor.visit_f64_mul(), + 0xa3 => visitor.visit_f64_div(), + 0xa4 => visitor.visit_f64_min(), + 0xa5 => visitor.visit_f64_max(), + 0xa6 => visitor.visit_f64_copysign(), + 0xa7 => visitor.visit_i32_wrap_i64(), + 0xa8 => visitor.visit_i32_trunc_f32_s(), + 0xa9 => visitor.visit_i32_trunc_f32_u(), + 0xaa => visitor.visit_i32_trunc_f64_s(), + 0xab => visitor.visit_i32_trunc_f64_u(), + 0xac => visitor.visit_i64_extend_i32_s(), + 0xad => visitor.visit_i64_extend_i32_u(), + 0xae => visitor.visit_i64_trunc_f32_s(), + 0xaf => visitor.visit_i64_trunc_f32_u(), + 0xb0 => visitor.visit_i64_trunc_f64_s(), + 0xb1 => visitor.visit_i64_trunc_f64_u(), + 0xb2 => visitor.visit_f32_convert_i32_s(), + 0xb3 => visitor.visit_f32_convert_i32_u(), + 0xb4 => visitor.visit_f32_convert_i64_s(), + 0xb5 => visitor.visit_f32_convert_i64_u(), + 0xb6 => visitor.visit_f32_demote_f64(), + 0xb7 => visitor.visit_f64_convert_i32_s(), + 0xb8 => visitor.visit_f64_convert_i32_u(), + 0xb9 => visitor.visit_f64_convert_i64_s(), + 0xba => visitor.visit_f64_convert_i64_u(), + 0xbb => visitor.visit_f64_promote_f32(), + 0xbc => visitor.visit_i32_reinterpret_f32(), + 0xbd => visitor.visit_i64_reinterpret_f64(), + 0xbe => visitor.visit_f32_reinterpret_i32(), + 0xbf => visitor.visit_f64_reinterpret_i64(), + + 0xc0 => visitor.visit_i32_extend8_s(), + 0xc1 => visitor.visit_i32_extend16_s(), + 0xc2 => visitor.visit_i64_extend8_s(), + 0xc3 => visitor.visit_i64_extend16_s(), + 0xc4 => visitor.visit_i64_extend32_s(), + + 0xd0 => visitor.visit_ref_null(self.read()?), + 0xd1 => visitor.visit_ref_is_null(), + 0xd2 => visitor.visit_ref_func(self.read_var_u32()?), + 0xd3 => visitor.visit_ref_as_non_null(), + 0xd4 => visitor.visit_br_on_null(self.read_var_u32()?), + 0xd6 => visitor.visit_br_on_non_null(self.read_var_u32()?), + + 0xfc => self.visit_0xfc_operator(pos, visitor)?, + 0xfd => self.visit_0xfd_operator(pos, visitor)?, + 0xfe => self.visit_0xfe_operator(pos, visitor)?, + + _ => bail!(pos, "illegal opcode: 0x{code:x}"), + }) + } + + fn visit_0xfc_operator<T>( + &mut self, + pos: usize, + visitor: &mut T, + ) -> Result<<T as VisitOperator<'a>>::Output> + where + T: VisitOperator<'a>, + { + let code = self.read_var_u32()?; + Ok(match code { + 0x00 => visitor.visit_i32_trunc_sat_f32_s(), + 0x01 => visitor.visit_i32_trunc_sat_f32_u(), + 0x02 => visitor.visit_i32_trunc_sat_f64_s(), + 0x03 => visitor.visit_i32_trunc_sat_f64_u(), + 0x04 => visitor.visit_i64_trunc_sat_f32_s(), + 0x05 => visitor.visit_i64_trunc_sat_f32_u(), + 0x06 => visitor.visit_i64_trunc_sat_f64_s(), + 0x07 => visitor.visit_i64_trunc_sat_f64_u(), + + 0x08 => { + let segment = self.read_var_u32()?; + let mem = self.read_var_u32()?; + visitor.visit_memory_init(segment, mem) + } + 0x09 => { + let segment = self.read_var_u32()?; + visitor.visit_data_drop(segment) + } + 0x0a => { + let dst = self.read_var_u32()?; + let src = self.read_var_u32()?; + visitor.visit_memory_copy(dst, src) + } + 0x0b => { + let mem = self.read_var_u32()?; + visitor.visit_memory_fill(mem) + } + 0x0c => { + let segment = self.read_var_u32()?; + let table = self.read_var_u32()?; + visitor.visit_table_init(segment, table) + } + 0x0d => { + let segment = self.read_var_u32()?; + visitor.visit_elem_drop(segment) + } + 0x0e => { + let dst_table = self.read_var_u32()?; + let src_table = self.read_var_u32()?; + visitor.visit_table_copy(dst_table, src_table) + } + + 0x0f => { + let table = self.read_var_u32()?; + visitor.visit_table_grow(table) + } + 0x10 => { + let table = self.read_var_u32()?; + visitor.visit_table_size(table) + } + + 0x11 => { + let table = self.read_var_u32()?; + visitor.visit_table_fill(table) + } + + 0x12 => { + let mem = self.read_var_u32()?; + visitor.visit_memory_discard(mem) + } + + _ => bail!(pos, "unknown 0xfc subopcode: 0x{code:x}"), + }) + } + + fn visit_0xfd_operator<T>( + &mut self, + pos: usize, + visitor: &mut T, + ) -> Result<<T as VisitOperator<'a>>::Output> + where + T: VisitOperator<'a>, + { + let code = self.read_var_u32()?; + Ok(match code { + 0x00 => visitor.visit_v128_load(self.read_memarg(4)?), + 0x01 => visitor.visit_v128_load8x8_s(self.read_memarg(3)?), + 0x02 => visitor.visit_v128_load8x8_u(self.read_memarg(3)?), + 0x03 => visitor.visit_v128_load16x4_s(self.read_memarg(3)?), + 0x04 => visitor.visit_v128_load16x4_u(self.read_memarg(3)?), + 0x05 => visitor.visit_v128_load32x2_s(self.read_memarg(3)?), + 0x06 => visitor.visit_v128_load32x2_u(self.read_memarg(3)?), + 0x07 => visitor.visit_v128_load8_splat(self.read_memarg(0)?), + 0x08 => visitor.visit_v128_load16_splat(self.read_memarg(1)?), + 0x09 => visitor.visit_v128_load32_splat(self.read_memarg(2)?), + 0x0a => visitor.visit_v128_load64_splat(self.read_memarg(3)?), + + 0x0b => visitor.visit_v128_store(self.read_memarg(4)?), + 0x0c => visitor.visit_v128_const(self.read_v128()?), + 0x0d => { + let mut lanes: [u8; 16] = [0; 16]; + for lane in &mut lanes { + *lane = self.read_lane_index(32)? + } + visitor.visit_i8x16_shuffle(lanes) + } + + 0x0e => visitor.visit_i8x16_swizzle(), + 0x0f => visitor.visit_i8x16_splat(), + 0x10 => visitor.visit_i16x8_splat(), + 0x11 => visitor.visit_i32x4_splat(), + 0x12 => visitor.visit_i64x2_splat(), + 0x13 => visitor.visit_f32x4_splat(), + 0x14 => visitor.visit_f64x2_splat(), + + 0x15 => visitor.visit_i8x16_extract_lane_s(self.read_lane_index(16)?), + 0x16 => visitor.visit_i8x16_extract_lane_u(self.read_lane_index(16)?), + 0x17 => visitor.visit_i8x16_replace_lane(self.read_lane_index(16)?), + 0x18 => visitor.visit_i16x8_extract_lane_s(self.read_lane_index(8)?), + 0x19 => visitor.visit_i16x8_extract_lane_u(self.read_lane_index(8)?), + 0x1a => visitor.visit_i16x8_replace_lane(self.read_lane_index(8)?), + 0x1b => visitor.visit_i32x4_extract_lane(self.read_lane_index(4)?), + + 0x1c => visitor.visit_i32x4_replace_lane(self.read_lane_index(4)?), + 0x1d => visitor.visit_i64x2_extract_lane(self.read_lane_index(2)?), + 0x1e => visitor.visit_i64x2_replace_lane(self.read_lane_index(2)?), + 0x1f => visitor.visit_f32x4_extract_lane(self.read_lane_index(4)?), + 0x20 => visitor.visit_f32x4_replace_lane(self.read_lane_index(4)?), + 0x21 => visitor.visit_f64x2_extract_lane(self.read_lane_index(2)?), + 0x22 => visitor.visit_f64x2_replace_lane(self.read_lane_index(2)?), + + 0x23 => visitor.visit_i8x16_eq(), + 0x24 => visitor.visit_i8x16_ne(), + 0x25 => visitor.visit_i8x16_lt_s(), + 0x26 => visitor.visit_i8x16_lt_u(), + 0x27 => visitor.visit_i8x16_gt_s(), + 0x28 => visitor.visit_i8x16_gt_u(), + 0x29 => visitor.visit_i8x16_le_s(), + 0x2a => visitor.visit_i8x16_le_u(), + 0x2b => visitor.visit_i8x16_ge_s(), + 0x2c => visitor.visit_i8x16_ge_u(), + 0x2d => visitor.visit_i16x8_eq(), + 0x2e => visitor.visit_i16x8_ne(), + 0x2f => visitor.visit_i16x8_lt_s(), + 0x30 => visitor.visit_i16x8_lt_u(), + 0x31 => visitor.visit_i16x8_gt_s(), + 0x32 => visitor.visit_i16x8_gt_u(), + 0x33 => visitor.visit_i16x8_le_s(), + 0x34 => visitor.visit_i16x8_le_u(), + 0x35 => visitor.visit_i16x8_ge_s(), + 0x36 => visitor.visit_i16x8_ge_u(), + 0x37 => visitor.visit_i32x4_eq(), + 0x38 => visitor.visit_i32x4_ne(), + 0x39 => visitor.visit_i32x4_lt_s(), + 0x3a => visitor.visit_i32x4_lt_u(), + 0x3b => visitor.visit_i32x4_gt_s(), + 0x3c => visitor.visit_i32x4_gt_u(), + 0x3d => visitor.visit_i32x4_le_s(), + 0x3e => visitor.visit_i32x4_le_u(), + 0x3f => visitor.visit_i32x4_ge_s(), + 0x40 => visitor.visit_i32x4_ge_u(), + 0x41 => visitor.visit_f32x4_eq(), + 0x42 => visitor.visit_f32x4_ne(), + 0x43 => visitor.visit_f32x4_lt(), + 0x44 => visitor.visit_f32x4_gt(), + 0x45 => visitor.visit_f32x4_le(), + 0x46 => visitor.visit_f32x4_ge(), + 0x47 => visitor.visit_f64x2_eq(), + 0x48 => visitor.visit_f64x2_ne(), + 0x49 => visitor.visit_f64x2_lt(), + 0x4a => visitor.visit_f64x2_gt(), + 0x4b => visitor.visit_f64x2_le(), + 0x4c => visitor.visit_f64x2_ge(), + 0x4d => visitor.visit_v128_not(), + 0x4e => visitor.visit_v128_and(), + 0x4f => visitor.visit_v128_andnot(), + 0x50 => visitor.visit_v128_or(), + 0x51 => visitor.visit_v128_xor(), + 0x52 => visitor.visit_v128_bitselect(), + 0x53 => visitor.visit_v128_any_true(), + + 0x54 => { + let memarg = self.read_memarg(0)?; + let lane = self.read_lane_index(16)?; + visitor.visit_v128_load8_lane(memarg, lane) + } + 0x55 => { + let memarg = self.read_memarg(1)?; + let lane = self.read_lane_index(8)?; + visitor.visit_v128_load16_lane(memarg, lane) + } + 0x56 => { + let memarg = self.read_memarg(2)?; + let lane = self.read_lane_index(4)?; + visitor.visit_v128_load32_lane(memarg, lane) + } + 0x57 => { + let memarg = self.read_memarg(3)?; + let lane = self.read_lane_index(2)?; + visitor.visit_v128_load64_lane(memarg, lane) + } + 0x58 => { + let memarg = self.read_memarg(0)?; + let lane = self.read_lane_index(16)?; + visitor.visit_v128_store8_lane(memarg, lane) + } + 0x59 => { + let memarg = self.read_memarg(1)?; + let lane = self.read_lane_index(8)?; + visitor.visit_v128_store16_lane(memarg, lane) + } + 0x5a => { + let memarg = self.read_memarg(2)?; + let lane = self.read_lane_index(4)?; + visitor.visit_v128_store32_lane(memarg, lane) + } + 0x5b => { + let memarg = self.read_memarg(3)?; + let lane = self.read_lane_index(2)?; + visitor.visit_v128_store64_lane(memarg, lane) + } + + 0x5c => visitor.visit_v128_load32_zero(self.read_memarg(2)?), + 0x5d => visitor.visit_v128_load64_zero(self.read_memarg(3)?), + 0x5e => visitor.visit_f32x4_demote_f64x2_zero(), + 0x5f => visitor.visit_f64x2_promote_low_f32x4(), + 0x60 => visitor.visit_i8x16_abs(), + 0x61 => visitor.visit_i8x16_neg(), + 0x62 => visitor.visit_i8x16_popcnt(), + 0x63 => visitor.visit_i8x16_all_true(), + 0x64 => visitor.visit_i8x16_bitmask(), + 0x65 => visitor.visit_i8x16_narrow_i16x8_s(), + 0x66 => visitor.visit_i8x16_narrow_i16x8_u(), + 0x67 => visitor.visit_f32x4_ceil(), + 0x68 => visitor.visit_f32x4_floor(), + 0x69 => visitor.visit_f32x4_trunc(), + 0x6a => visitor.visit_f32x4_nearest(), + 0x6b => visitor.visit_i8x16_shl(), + 0x6c => visitor.visit_i8x16_shr_s(), + 0x6d => visitor.visit_i8x16_shr_u(), + 0x6e => visitor.visit_i8x16_add(), + 0x6f => visitor.visit_i8x16_add_sat_s(), + 0x70 => visitor.visit_i8x16_add_sat_u(), + 0x71 => visitor.visit_i8x16_sub(), + 0x72 => visitor.visit_i8x16_sub_sat_s(), + 0x73 => visitor.visit_i8x16_sub_sat_u(), + 0x74 => visitor.visit_f64x2_ceil(), + 0x75 => visitor.visit_f64x2_floor(), + 0x76 => visitor.visit_i8x16_min_s(), + 0x77 => visitor.visit_i8x16_min_u(), + 0x78 => visitor.visit_i8x16_max_s(), + 0x79 => visitor.visit_i8x16_max_u(), + 0x7a => visitor.visit_f64x2_trunc(), + 0x7b => visitor.visit_i8x16_avgr_u(), + 0x7c => visitor.visit_i16x8_extadd_pairwise_i8x16_s(), + 0x7d => visitor.visit_i16x8_extadd_pairwise_i8x16_u(), + 0x7e => visitor.visit_i32x4_extadd_pairwise_i16x8_s(), + 0x7f => visitor.visit_i32x4_extadd_pairwise_i16x8_u(), + 0x80 => visitor.visit_i16x8_abs(), + 0x81 => visitor.visit_i16x8_neg(), + 0x82 => visitor.visit_i16x8_q15mulr_sat_s(), + 0x83 => visitor.visit_i16x8_all_true(), + 0x84 => visitor.visit_i16x8_bitmask(), + 0x85 => visitor.visit_i16x8_narrow_i32x4_s(), + 0x86 => visitor.visit_i16x8_narrow_i32x4_u(), + 0x87 => visitor.visit_i16x8_extend_low_i8x16_s(), + 0x88 => visitor.visit_i16x8_extend_high_i8x16_s(), + 0x89 => visitor.visit_i16x8_extend_low_i8x16_u(), + 0x8a => visitor.visit_i16x8_extend_high_i8x16_u(), + 0x8b => visitor.visit_i16x8_shl(), + 0x8c => visitor.visit_i16x8_shr_s(), + 0x8d => visitor.visit_i16x8_shr_u(), + 0x8e => visitor.visit_i16x8_add(), + 0x8f => visitor.visit_i16x8_add_sat_s(), + 0x90 => visitor.visit_i16x8_add_sat_u(), + 0x91 => visitor.visit_i16x8_sub(), + 0x92 => visitor.visit_i16x8_sub_sat_s(), + 0x93 => visitor.visit_i16x8_sub_sat_u(), + 0x94 => visitor.visit_f64x2_nearest(), + 0x95 => visitor.visit_i16x8_mul(), + 0x96 => visitor.visit_i16x8_min_s(), + 0x97 => visitor.visit_i16x8_min_u(), + 0x98 => visitor.visit_i16x8_max_s(), + 0x99 => visitor.visit_i16x8_max_u(), + 0x9b => visitor.visit_i16x8_avgr_u(), + 0x9c => visitor.visit_i16x8_extmul_low_i8x16_s(), + 0x9d => visitor.visit_i16x8_extmul_high_i8x16_s(), + 0x9e => visitor.visit_i16x8_extmul_low_i8x16_u(), + 0x9f => visitor.visit_i16x8_extmul_high_i8x16_u(), + 0xa0 => visitor.visit_i32x4_abs(), + 0xa1 => visitor.visit_i32x4_neg(), + 0xa3 => visitor.visit_i32x4_all_true(), + 0xa4 => visitor.visit_i32x4_bitmask(), + 0xa7 => visitor.visit_i32x4_extend_low_i16x8_s(), + 0xa8 => visitor.visit_i32x4_extend_high_i16x8_s(), + 0xa9 => visitor.visit_i32x4_extend_low_i16x8_u(), + 0xaa => visitor.visit_i32x4_extend_high_i16x8_u(), + 0xab => visitor.visit_i32x4_shl(), + 0xac => visitor.visit_i32x4_shr_s(), + 0xad => visitor.visit_i32x4_shr_u(), + 0xae => visitor.visit_i32x4_add(), + 0xb1 => visitor.visit_i32x4_sub(), + 0xb5 => visitor.visit_i32x4_mul(), + 0xb6 => visitor.visit_i32x4_min_s(), + 0xb7 => visitor.visit_i32x4_min_u(), + 0xb8 => visitor.visit_i32x4_max_s(), + 0xb9 => visitor.visit_i32x4_max_u(), + 0xba => visitor.visit_i32x4_dot_i16x8_s(), + 0xbc => visitor.visit_i32x4_extmul_low_i16x8_s(), + 0xbd => visitor.visit_i32x4_extmul_high_i16x8_s(), + 0xbe => visitor.visit_i32x4_extmul_low_i16x8_u(), + 0xbf => visitor.visit_i32x4_extmul_high_i16x8_u(), + 0xc0 => visitor.visit_i64x2_abs(), + 0xc1 => visitor.visit_i64x2_neg(), + 0xc3 => visitor.visit_i64x2_all_true(), + 0xc4 => visitor.visit_i64x2_bitmask(), + 0xc7 => visitor.visit_i64x2_extend_low_i32x4_s(), + 0xc8 => visitor.visit_i64x2_extend_high_i32x4_s(), + 0xc9 => visitor.visit_i64x2_extend_low_i32x4_u(), + 0xca => visitor.visit_i64x2_extend_high_i32x4_u(), + 0xcb => visitor.visit_i64x2_shl(), + 0xcc => visitor.visit_i64x2_shr_s(), + 0xcd => visitor.visit_i64x2_shr_u(), + 0xce => visitor.visit_i64x2_add(), + 0xd1 => visitor.visit_i64x2_sub(), + 0xd5 => visitor.visit_i64x2_mul(), + 0xd6 => visitor.visit_i64x2_eq(), + 0xd7 => visitor.visit_i64x2_ne(), + 0xd8 => visitor.visit_i64x2_lt_s(), + 0xd9 => visitor.visit_i64x2_gt_s(), + 0xda => visitor.visit_i64x2_le_s(), + 0xdb => visitor.visit_i64x2_ge_s(), + 0xdc => visitor.visit_i64x2_extmul_low_i32x4_s(), + 0xdd => visitor.visit_i64x2_extmul_high_i32x4_s(), + 0xde => visitor.visit_i64x2_extmul_low_i32x4_u(), + 0xdf => visitor.visit_i64x2_extmul_high_i32x4_u(), + 0xe0 => visitor.visit_f32x4_abs(), + 0xe1 => visitor.visit_f32x4_neg(), + 0xe3 => visitor.visit_f32x4_sqrt(), + 0xe4 => visitor.visit_f32x4_add(), + 0xe5 => visitor.visit_f32x4_sub(), + 0xe6 => visitor.visit_f32x4_mul(), + 0xe7 => visitor.visit_f32x4_div(), + 0xe8 => visitor.visit_f32x4_min(), + 0xe9 => visitor.visit_f32x4_max(), + 0xea => visitor.visit_f32x4_pmin(), + 0xeb => visitor.visit_f32x4_pmax(), + 0xec => visitor.visit_f64x2_abs(), + 0xed => visitor.visit_f64x2_neg(), + 0xef => visitor.visit_f64x2_sqrt(), + 0xf0 => visitor.visit_f64x2_add(), + 0xf1 => visitor.visit_f64x2_sub(), + 0xf2 => visitor.visit_f64x2_mul(), + 0xf3 => visitor.visit_f64x2_div(), + 0xf4 => visitor.visit_f64x2_min(), + 0xf5 => visitor.visit_f64x2_max(), + 0xf6 => visitor.visit_f64x2_pmin(), + 0xf7 => visitor.visit_f64x2_pmax(), + 0xf8 => visitor.visit_i32x4_trunc_sat_f32x4_s(), + 0xf9 => visitor.visit_i32x4_trunc_sat_f32x4_u(), + 0xfa => visitor.visit_f32x4_convert_i32x4_s(), + 0xfb => visitor.visit_f32x4_convert_i32x4_u(), + 0xfc => visitor.visit_i32x4_trunc_sat_f64x2_s_zero(), + 0xfd => visitor.visit_i32x4_trunc_sat_f64x2_u_zero(), + 0xfe => visitor.visit_f64x2_convert_low_i32x4_s(), + 0xff => visitor.visit_f64x2_convert_low_i32x4_u(), + 0x100 => visitor.visit_i8x16_relaxed_swizzle(), + 0x101 => visitor.visit_i32x4_relaxed_trunc_f32x4_s(), + 0x102 => visitor.visit_i32x4_relaxed_trunc_f32x4_u(), + 0x103 => visitor.visit_i32x4_relaxed_trunc_f64x2_s_zero(), + 0x104 => visitor.visit_i32x4_relaxed_trunc_f64x2_u_zero(), + 0x105 => visitor.visit_f32x4_relaxed_madd(), + 0x106 => visitor.visit_f32x4_relaxed_nmadd(), + 0x107 => visitor.visit_f64x2_relaxed_madd(), + 0x108 => visitor.visit_f64x2_relaxed_nmadd(), + 0x109 => visitor.visit_i8x16_relaxed_laneselect(), + 0x10a => visitor.visit_i16x8_relaxed_laneselect(), + 0x10b => visitor.visit_i32x4_relaxed_laneselect(), + 0x10c => visitor.visit_i64x2_relaxed_laneselect(), + 0x10d => visitor.visit_f32x4_relaxed_min(), + 0x10e => visitor.visit_f32x4_relaxed_max(), + 0x10f => visitor.visit_f64x2_relaxed_min(), + 0x110 => visitor.visit_f64x2_relaxed_max(), + 0x111 => visitor.visit_i16x8_relaxed_q15mulr_s(), + 0x112 => visitor.visit_i16x8_relaxed_dot_i8x16_i7x16_s(), + 0x113 => visitor.visit_i32x4_relaxed_dot_i8x16_i7x16_add_s(), + + _ => bail!(pos, "unknown 0xfd subopcode: 0x{code:x}"), + }) + } + + fn visit_0xfe_operator<T>( + &mut self, + pos: usize, + visitor: &mut T, + ) -> Result<<T as VisitOperator<'a>>::Output> + where + T: VisitOperator<'a>, + { + let code = self.read_var_u32()?; + Ok(match code { + 0x00 => visitor.visit_memory_atomic_notify(self.read_memarg(2)?), + 0x01 => visitor.visit_memory_atomic_wait32(self.read_memarg(2)?), + 0x02 => visitor.visit_memory_atomic_wait64(self.read_memarg(3)?), + 0x03 => { + if self.read_u8()? != 0 { + bail!(pos, "nonzero byte after `atomic.fence`"); + } + visitor.visit_atomic_fence() + } + 0x10 => visitor.visit_i32_atomic_load(self.read_memarg(2)?), + 0x11 => visitor.visit_i64_atomic_load(self.read_memarg(3)?), + 0x12 => visitor.visit_i32_atomic_load8_u(self.read_memarg(0)?), + 0x13 => visitor.visit_i32_atomic_load16_u(self.read_memarg(1)?), + 0x14 => visitor.visit_i64_atomic_load8_u(self.read_memarg(0)?), + 0x15 => visitor.visit_i64_atomic_load16_u(self.read_memarg(1)?), + 0x16 => visitor.visit_i64_atomic_load32_u(self.read_memarg(2)?), + 0x17 => visitor.visit_i32_atomic_store(self.read_memarg(2)?), + 0x18 => visitor.visit_i64_atomic_store(self.read_memarg(3)?), + 0x19 => visitor.visit_i32_atomic_store8(self.read_memarg(0)?), + 0x1a => visitor.visit_i32_atomic_store16(self.read_memarg(1)?), + 0x1b => visitor.visit_i64_atomic_store8(self.read_memarg(0)?), + 0x1c => visitor.visit_i64_atomic_store16(self.read_memarg(1)?), + 0x1d => visitor.visit_i64_atomic_store32(self.read_memarg(2)?), + 0x1e => visitor.visit_i32_atomic_rmw_add(self.read_memarg(2)?), + 0x1f => visitor.visit_i64_atomic_rmw_add(self.read_memarg(3)?), + 0x20 => visitor.visit_i32_atomic_rmw8_add_u(self.read_memarg(0)?), + 0x21 => visitor.visit_i32_atomic_rmw16_add_u(self.read_memarg(1)?), + 0x22 => visitor.visit_i64_atomic_rmw8_add_u(self.read_memarg(0)?), + 0x23 => visitor.visit_i64_atomic_rmw16_add_u(self.read_memarg(1)?), + 0x24 => visitor.visit_i64_atomic_rmw32_add_u(self.read_memarg(2)?), + 0x25 => visitor.visit_i32_atomic_rmw_sub(self.read_memarg(2)?), + 0x26 => visitor.visit_i64_atomic_rmw_sub(self.read_memarg(3)?), + 0x27 => visitor.visit_i32_atomic_rmw8_sub_u(self.read_memarg(0)?), + 0x28 => visitor.visit_i32_atomic_rmw16_sub_u(self.read_memarg(1)?), + 0x29 => visitor.visit_i64_atomic_rmw8_sub_u(self.read_memarg(0)?), + 0x2a => visitor.visit_i64_atomic_rmw16_sub_u(self.read_memarg(1)?), + 0x2b => visitor.visit_i64_atomic_rmw32_sub_u(self.read_memarg(2)?), + 0x2c => visitor.visit_i32_atomic_rmw_and(self.read_memarg(2)?), + 0x2d => visitor.visit_i64_atomic_rmw_and(self.read_memarg(3)?), + 0x2e => visitor.visit_i32_atomic_rmw8_and_u(self.read_memarg(0)?), + 0x2f => visitor.visit_i32_atomic_rmw16_and_u(self.read_memarg(1)?), + 0x30 => visitor.visit_i64_atomic_rmw8_and_u(self.read_memarg(0)?), + 0x31 => visitor.visit_i64_atomic_rmw16_and_u(self.read_memarg(1)?), + 0x32 => visitor.visit_i64_atomic_rmw32_and_u(self.read_memarg(2)?), + 0x33 => visitor.visit_i32_atomic_rmw_or(self.read_memarg(2)?), + 0x34 => visitor.visit_i64_atomic_rmw_or(self.read_memarg(3)?), + 0x35 => visitor.visit_i32_atomic_rmw8_or_u(self.read_memarg(0)?), + 0x36 => visitor.visit_i32_atomic_rmw16_or_u(self.read_memarg(1)?), + 0x37 => visitor.visit_i64_atomic_rmw8_or_u(self.read_memarg(0)?), + 0x38 => visitor.visit_i64_atomic_rmw16_or_u(self.read_memarg(1)?), + 0x39 => visitor.visit_i64_atomic_rmw32_or_u(self.read_memarg(2)?), + 0x3a => visitor.visit_i32_atomic_rmw_xor(self.read_memarg(2)?), + 0x3b => visitor.visit_i64_atomic_rmw_xor(self.read_memarg(3)?), + 0x3c => visitor.visit_i32_atomic_rmw8_xor_u(self.read_memarg(0)?), + 0x3d => visitor.visit_i32_atomic_rmw16_xor_u(self.read_memarg(1)?), + 0x3e => visitor.visit_i64_atomic_rmw8_xor_u(self.read_memarg(0)?), + 0x3f => visitor.visit_i64_atomic_rmw16_xor_u(self.read_memarg(1)?), + 0x40 => visitor.visit_i64_atomic_rmw32_xor_u(self.read_memarg(2)?), + 0x41 => visitor.visit_i32_atomic_rmw_xchg(self.read_memarg(2)?), + 0x42 => visitor.visit_i64_atomic_rmw_xchg(self.read_memarg(3)?), + 0x43 => visitor.visit_i32_atomic_rmw8_xchg_u(self.read_memarg(0)?), + 0x44 => visitor.visit_i32_atomic_rmw16_xchg_u(self.read_memarg(1)?), + 0x45 => visitor.visit_i64_atomic_rmw8_xchg_u(self.read_memarg(0)?), + 0x46 => visitor.visit_i64_atomic_rmw16_xchg_u(self.read_memarg(1)?), + 0x47 => visitor.visit_i64_atomic_rmw32_xchg_u(self.read_memarg(2)?), + 0x48 => visitor.visit_i32_atomic_rmw_cmpxchg(self.read_memarg(2)?), + 0x49 => visitor.visit_i64_atomic_rmw_cmpxchg(self.read_memarg(3)?), + 0x4a => visitor.visit_i32_atomic_rmw8_cmpxchg_u(self.read_memarg(0)?), + 0x4b => visitor.visit_i32_atomic_rmw16_cmpxchg_u(self.read_memarg(1)?), + 0x4c => visitor.visit_i64_atomic_rmw8_cmpxchg_u(self.read_memarg(0)?), + 0x4d => visitor.visit_i64_atomic_rmw16_cmpxchg_u(self.read_memarg(1)?), + 0x4e => visitor.visit_i64_atomic_rmw32_cmpxchg_u(self.read_memarg(2)?), + + _ => bail!(pos, "unknown 0xfe subopcode: 0x{code:x}"), + }) + } + + /// Reads the next available `Operator`. + /// + /// # Errors + /// + /// If `BinaryReader` has less bytes remaining than required to parse + /// the `Operator`. + pub fn read_operator(&mut self) -> Result<Operator<'a>> { + self.visit_operator(&mut OperatorFactory::new()) + } + + fn read_lane_index(&mut self, max: u8) -> Result<u8> { + let index = self.read_u8()?; + if index >= max { + return Err(BinaryReaderError::new( + "invalid lane index", + self.original_position() - 1, + )); + } + Ok(index) + } + + fn read_v128(&mut self) -> Result<V128> { + let mut bytes = [0; 16]; + bytes.clone_from_slice(self.read_bytes(16)?); + Ok(V128(bytes)) + } + + pub(crate) fn read_header_version(&mut self) -> Result<u32> { + let magic_number = self.read_bytes(4)?; + if magic_number != WASM_MAGIC_NUMBER { + return Err(BinaryReaderError::new( + "magic header not detected: bad magic number", + self.original_position() - 4, + )); + } + self.read_u32() + } + + pub(crate) fn skip_const_expr(&mut self) -> Result<()> { + // TODO add skip_operator() method and/or validate ConstExpr operators. + loop { + if let Operator::End = self.read_operator()? { + return Ok(()); + } + } + } +} + +impl<'a> BrTable<'a> { + /// Returns the number of `br_table` entries, not including the default + /// label + pub fn len(&self) -> u32 { + self.cnt + } + + /// Returns whether `BrTable` doesn't have any labels apart from the default one. + pub fn is_empty(&self) -> bool { + self.len() == 0 + } + + /// Returns the default target of this `br_table` instruction. + pub fn default(&self) -> u32 { + self.default + } + + /// Returns the list of targets that this `br_table` instruction will be + /// jumping to. + /// + /// This method will return an iterator which parses each target of this + /// `br_table` except the default target. The returned iterator will + /// yield `self.len()` elements. + /// + /// # Examples + /// + /// ```rust + /// let buf = [0x0e, 0x02, 0x01, 0x02, 0x00]; + /// let mut reader = wasmparser::BinaryReader::new(&buf); + /// let op = reader.read_operator().unwrap(); + /// if let wasmparser::Operator::BrTable { targets } = op { + /// let targets = targets.targets().collect::<Result<Vec<_>, _>>().unwrap(); + /// assert_eq!(targets, [1, 2]); + /// } + /// ``` + pub fn targets(&self) -> BrTableTargets { + BrTableTargets { + reader: self.reader.clone(), + remaining: self.cnt, + } + } +} + +/// An iterator over the targets of a [`BrTable`]. +/// +/// # Note +/// +/// This iterator parses each target of the underlying `br_table` +/// except for the default target. +/// The iterator will yield exactly as many targets as the `br_table` has. +pub struct BrTableTargets<'a> { + reader: crate::BinaryReader<'a>, + remaining: u32, +} + +impl<'a> Iterator for BrTableTargets<'a> { + type Item = Result<u32>; + + fn size_hint(&self) -> (usize, Option<usize>) { + let remaining = usize::try_from(self.remaining).unwrap_or_else(|error| { + panic!("could not convert remaining `u32` into `usize`: {}", error) + }); + (remaining, Some(remaining)) + } + + fn next(&mut self) -> Option<Self::Item> { + if self.remaining == 0 { + if !self.reader.eof() { + return Some(Err(BinaryReaderError::new( + "trailing data in br_table", + self.reader.original_position(), + ))); + } + return None; + } + self.remaining -= 1; + Some(self.reader.read_var_u32()) + } +} + +impl fmt::Debug for BrTable<'_> { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + let mut f = f.debug_struct("BrTable"); + f.field("count", &self.cnt); + f.field("default", &self.default); + match self.targets().collect::<Result<Vec<_>>>() { + Ok(targets) => { + f.field("targets", &targets); + } + Err(_) => { + f.field("reader", &self.reader); + } + } + f.finish() + } +} + +/// A factory to construct [`Operator`] instances via the [`VisitOperator`] trait. +struct OperatorFactory<'a> { + marker: core::marker::PhantomData<fn() -> &'a ()>, +} + +impl<'a> OperatorFactory<'a> { + /// Creates a new [`OperatorFactory`]. + fn new() -> Self { + Self { + marker: core::marker::PhantomData, + } + } +} + +macro_rules! define_visit_operator { + ($(@$proposal:ident $op:ident $({ $($arg:ident: $argty:ty),* })? => $visit:ident)*) => { + $( + fn $visit(&mut self $($(,$arg: $argty)*)?) -> Operator<'a> { + Operator::$op $({ $($arg),* })? + } + )* + } +} + +impl<'a> VisitOperator<'a> for OperatorFactory<'a> { + type Output = Operator<'a>; + + for_each_operator!(define_visit_operator); +} + +/// Iterator returned from [`BinaryReader::read_iter`]. +pub struct BinaryReaderIter<'a, 'me, T: FromReader<'a>> { + remaining: usize, + reader: &'me mut BinaryReader<'a>, + _marker: marker::PhantomData<T>, +} + +impl<'a, T> Iterator for BinaryReaderIter<'a, '_, T> +where + T: FromReader<'a>, +{ + type Item = Result<T>; + + fn next(&mut self) -> Option<Result<T>> { + if self.remaining == 0 { + None + } else { + let ret = self.reader.read::<T>(); + if ret.is_err() { + self.remaining = 0; + } else { + self.remaining -= 1; + } + Some(ret) + } + } + + fn size_hint(&self) -> (usize, Option<usize>) { + (self.remaining, Some(self.remaining)) + } +} + +impl<'a, T> Drop for BinaryReaderIter<'a, '_, T> +where + T: FromReader<'a>, +{ + fn drop(&mut self) { + while self.next().is_some() { + // ... + } + } +} diff --git a/third_party/rust/wasmparser/src/lib.rs b/third_party/rust/wasmparser/src/lib.rs new file mode 100644 index 0000000000..708809118d --- /dev/null +++ b/third_party/rust/wasmparser/src/lib.rs @@ -0,0 +1,712 @@ +/* Copyright 2017 Mozilla Foundation + * + * 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. + */ + +//! A simple event-driven library for parsing WebAssembly binary files +//! (or streams). +//! +//! The parser library reports events as they happen and only stores +//! parsing information for a brief period of time, making it very fast +//! and memory-efficient. The event-driven model, however, has some drawbacks. +//! If you need random access to the entire WebAssembly data-structure, +//! this is not the right library for you. You could however, build such +//! a data-structure using this library. + +#![deny(missing_docs)] + +/// A helper macro to conveniently iterate over all opcodes recognized by this +/// crate. This can be used to work with either the [`Operator`] enumeration or +/// the [`VisitOperator`] trait if your use case uniformly handles all operators +/// the same way. +/// +/// It is also possible to specialize handling of operators depending on the +/// Wasm proposal from which they are originating. +/// +/// This is an "iterator macro" where this macro is invoked with the name of +/// another macro, and then that macro is invoked with the list of all +/// operators. An example invocation of this looks like: +/// +/// The list of specializable Wasm proposals is as follows: +/// +/// - `@mvp`: Denoting a Wasm operator from the initial Wasm MVP version. +/// - `@exceptions`: [Wasm `expection-handling` proposal] +/// - `@tail_call`: [Wasm `tail-calls` proposal] +/// - `@reference_types`: [Wasm `reference-types` proposal] +/// - `@sign_extension`: [Wasm `sign-extension-ops` proposal] +/// - `@saturating_float_to_int`: [Wasm `non_trapping_float-to-int-conversions` proposal] +/// - `@bulk_memory `:[Wasm `bulk-memory` proposal] +/// - `@threads`: [Wasm `threads` proposal] +/// - `@simd`: [Wasm `simd` proposal] +/// - `@relaxed_simd`: [Wasm `relaxed-simd` proposal] +/// +/// [Wasm `expection-handling` proposal]: +/// https://github.com/WebAssembly/exception-handling +/// +/// [Wasm `tail-calls` proposal]: +/// https://github.com/WebAssembly/tail-call +/// +/// [Wasm `reference-types` proposal]: +/// https://github.com/WebAssembly/reference-types +/// +/// [Wasm `sign-extension-ops` proposal]: +/// https://github.com/WebAssembly/sign-extension-ops +/// +/// [Wasm `non_trapping_float-to-int-conversions` proposal]: +/// https://github.com/WebAssembly/nontrapping-float-to-int-conversions +/// +/// [Wasm `bulk-memory` proposal]: +/// https://github.com/WebAssembly/bulk-memory-operations +/// +/// [Wasm `threads` proposal]: +/// https://github.com/webassembly/threads +/// +/// [Wasm `simd` proposal]: +/// https://github.com/webassembly/simd +/// +/// [Wasm `relaxed-simd` proposal]: +/// https://github.com/WebAssembly/relaxed-simd +/// +/// ``` +/// macro_rules! define_visit_operator { +/// // The outer layer of repetition represents how all operators are +/// // provided to the macro at the same time. +/// // +/// // The `$proposal` identifier indicates the Wasm proposals from which +/// // the Wasm operator is originating. +/// // For example to specialize the macro match arm for Wasm SIMD proposal +/// // operators you could write `@simd` instead of `@$proposal:ident` to +/// // only catch those operators. +/// // +/// // The `$op` name is bound to the `Operator` variant name. The +/// // payload of the operator is optionally specified (the `$(...)?` +/// // clause) since not all instructions have payloads. Within the payload +/// // each argument is named and has its type specified. +/// // +/// // The `$visit` name is bound to the corresponding name in the +/// // `VisitOperator` trait that this corresponds to. +/// ($( @$proposal:ident $op:ident $({ $($arg:ident: $argty:ty),* })? => $visit:ident)*) => { +/// $( +/// fn $visit(&mut self $($(,$arg: $argty)*)?) { +/// // do nothing for this example +/// } +/// )* +/// } +/// } +/// +/// pub struct VisitAndDoNothing; +/// +/// impl<'a> wasmparser::VisitOperator<'a> for VisitAndDoNothing { +/// type Output = (); +/// +/// wasmparser::for_each_operator!(define_visit_operator); +/// } +/// ``` +#[macro_export] +macro_rules! for_each_operator { + ($mac:ident) => { + $mac! { + @mvp Unreachable => visit_unreachable + @mvp Nop => visit_nop + @mvp Block { blockty: $crate::BlockType } => visit_block + @mvp Loop { blockty: $crate::BlockType } => visit_loop + @mvp If { blockty: $crate::BlockType } => visit_if + @mvp Else => visit_else + @exceptions Try { blockty: $crate::BlockType } => visit_try + @exceptions Catch { tag_index: u32 } => visit_catch + @exceptions Throw { tag_index: u32 } => visit_throw + @exceptions Rethrow { relative_depth: u32 } => visit_rethrow + @mvp End => visit_end + @mvp Br { relative_depth: u32 } => visit_br + @mvp BrIf { relative_depth: u32 } => visit_br_if + @mvp BrTable { targets: $crate::BrTable<'a> } => visit_br_table + @mvp Return => visit_return + @mvp Call { function_index: u32 } => visit_call + @mvp CallIndirect { type_index: u32, table_index: u32, table_byte: u8 } => visit_call_indirect + @tail_call ReturnCall { function_index: u32 } => visit_return_call + @tail_call ReturnCallIndirect { type_index: u32, table_index: u32 } => visit_return_call_indirect + @exceptions Delegate { relative_depth: u32 } => visit_delegate + @exceptions CatchAll => visit_catch_all + @mvp Drop => visit_drop + @mvp Select => visit_select + @reference_types TypedSelect { ty: $crate::ValType } => visit_typed_select + @mvp LocalGet { local_index: u32 } => visit_local_get + @mvp LocalSet { local_index: u32 } => visit_local_set + @mvp LocalTee { local_index: u32 } => visit_local_tee + @mvp GlobalGet { global_index: u32 } => visit_global_get + @mvp GlobalSet { global_index: u32 } => visit_global_set + @mvp I32Load { memarg: $crate::MemArg } => visit_i32_load + @mvp I64Load { memarg: $crate::MemArg } => visit_i64_load + @mvp F32Load { memarg: $crate::MemArg } => visit_f32_load + @mvp F64Load { memarg: $crate::MemArg } => visit_f64_load + @mvp I32Load8S { memarg: $crate::MemArg } => visit_i32_load8_s + @mvp I32Load8U { memarg: $crate::MemArg } => visit_i32_load8_u + @mvp I32Load16S { memarg: $crate::MemArg } => visit_i32_load16_s + @mvp I32Load16U { memarg: $crate::MemArg } => visit_i32_load16_u + @mvp I64Load8S { memarg: $crate::MemArg } => visit_i64_load8_s + @mvp I64Load8U { memarg: $crate::MemArg } => visit_i64_load8_u + @mvp I64Load16S { memarg: $crate::MemArg } => visit_i64_load16_s + @mvp I64Load16U { memarg: $crate::MemArg } => visit_i64_load16_u + @mvp I64Load32S { memarg: $crate::MemArg } => visit_i64_load32_s + @mvp I64Load32U { memarg: $crate::MemArg } => visit_i64_load32_u + @mvp I32Store { memarg: $crate::MemArg } => visit_i32_store + @mvp I64Store { memarg: $crate::MemArg } => visit_i64_store + @mvp F32Store { memarg: $crate::MemArg } => visit_f32_store + @mvp F64Store { memarg: $crate::MemArg } => visit_f64_store + @mvp I32Store8 { memarg: $crate::MemArg } => visit_i32_store8 + @mvp I32Store16 { memarg: $crate::MemArg } => visit_i32_store16 + @mvp I64Store8 { memarg: $crate::MemArg } => visit_i64_store8 + @mvp I64Store16 { memarg: $crate::MemArg } => visit_i64_store16 + @mvp I64Store32 { memarg: $crate::MemArg } => visit_i64_store32 + @mvp MemorySize { mem: u32, mem_byte: u8 } => visit_memory_size + @mvp MemoryGrow { mem: u32, mem_byte: u8 } => visit_memory_grow + @mvp I32Const { value: i32 } => visit_i32_const + @mvp I64Const { value: i64 } => visit_i64_const + @mvp F32Const { value: $crate::Ieee32 } => visit_f32_const + @mvp F64Const { value: $crate::Ieee64 } => visit_f64_const + @reference_types RefNull { hty: $crate::HeapType } => visit_ref_null + @reference_types RefIsNull => visit_ref_is_null + @reference_types RefFunc { function_index: u32 } => visit_ref_func + @mvp I32Eqz => visit_i32_eqz + @mvp I32Eq => visit_i32_eq + @mvp I32Ne => visit_i32_ne + @mvp I32LtS => visit_i32_lt_s + @mvp I32LtU => visit_i32_lt_u + @mvp I32GtS => visit_i32_gt_s + @mvp I32GtU => visit_i32_gt_u + @mvp I32LeS => visit_i32_le_s + @mvp I32LeU => visit_i32_le_u + @mvp I32GeS => visit_i32_ge_s + @mvp I32GeU => visit_i32_ge_u + @mvp I64Eqz => visit_i64_eqz + @mvp I64Eq => visit_i64_eq + @mvp I64Ne => visit_i64_ne + @mvp I64LtS => visit_i64_lt_s + @mvp I64LtU => visit_i64_lt_u + @mvp I64GtS => visit_i64_gt_s + @mvp I64GtU => visit_i64_gt_u + @mvp I64LeS => visit_i64_le_s + @mvp I64LeU => visit_i64_le_u + @mvp I64GeS => visit_i64_ge_s + @mvp I64GeU => visit_i64_ge_u + @mvp F32Eq => visit_f32_eq + @mvp F32Ne => visit_f32_ne + @mvp F32Lt => visit_f32_lt + @mvp F32Gt => visit_f32_gt + @mvp F32Le => visit_f32_le + @mvp F32Ge => visit_f32_ge + @mvp F64Eq => visit_f64_eq + @mvp F64Ne => visit_f64_ne + @mvp F64Lt => visit_f64_lt + @mvp F64Gt => visit_f64_gt + @mvp F64Le => visit_f64_le + @mvp F64Ge => visit_f64_ge + @mvp I32Clz => visit_i32_clz + @mvp I32Ctz => visit_i32_ctz + @mvp I32Popcnt => visit_i32_popcnt + @mvp I32Add => visit_i32_add + @mvp I32Sub => visit_i32_sub + @mvp I32Mul => visit_i32_mul + @mvp I32DivS => visit_i32_div_s + @mvp I32DivU => visit_i32_div_u + @mvp I32RemS => visit_i32_rem_s + @mvp I32RemU => visit_i32_rem_u + @mvp I32And => visit_i32_and + @mvp I32Or => visit_i32_or + @mvp I32Xor => visit_i32_xor + @mvp I32Shl => visit_i32_shl + @mvp I32ShrS => visit_i32_shr_s + @mvp I32ShrU => visit_i32_shr_u + @mvp I32Rotl => visit_i32_rotl + @mvp I32Rotr => visit_i32_rotr + @mvp I64Clz => visit_i64_clz + @mvp I64Ctz => visit_i64_ctz + @mvp I64Popcnt => visit_i64_popcnt + @mvp I64Add => visit_i64_add + @mvp I64Sub => visit_i64_sub + @mvp I64Mul => visit_i64_mul + @mvp I64DivS => visit_i64_div_s + @mvp I64DivU => visit_i64_div_u + @mvp I64RemS => visit_i64_rem_s + @mvp I64RemU => visit_i64_rem_u + @mvp I64And => visit_i64_and + @mvp I64Or => visit_i64_or + @mvp I64Xor => visit_i64_xor + @mvp I64Shl => visit_i64_shl + @mvp I64ShrS => visit_i64_shr_s + @mvp I64ShrU => visit_i64_shr_u + @mvp I64Rotl => visit_i64_rotl + @mvp I64Rotr => visit_i64_rotr + @mvp F32Abs => visit_f32_abs + @mvp F32Neg => visit_f32_neg + @mvp F32Ceil => visit_f32_ceil + @mvp F32Floor => visit_f32_floor + @mvp F32Trunc => visit_f32_trunc + @mvp F32Nearest => visit_f32_nearest + @mvp F32Sqrt => visit_f32_sqrt + @mvp F32Add => visit_f32_add + @mvp F32Sub => visit_f32_sub + @mvp F32Mul => visit_f32_mul + @mvp F32Div => visit_f32_div + @mvp F32Min => visit_f32_min + @mvp F32Max => visit_f32_max + @mvp F32Copysign => visit_f32_copysign + @mvp F64Abs => visit_f64_abs + @mvp F64Neg => visit_f64_neg + @mvp F64Ceil => visit_f64_ceil + @mvp F64Floor => visit_f64_floor + @mvp F64Trunc => visit_f64_trunc + @mvp F64Nearest => visit_f64_nearest + @mvp F64Sqrt => visit_f64_sqrt + @mvp F64Add => visit_f64_add + @mvp F64Sub => visit_f64_sub + @mvp F64Mul => visit_f64_mul + @mvp F64Div => visit_f64_div + @mvp F64Min => visit_f64_min + @mvp F64Max => visit_f64_max + @mvp F64Copysign => visit_f64_copysign + @mvp I32WrapI64 => visit_i32_wrap_i64 + @mvp I32TruncF32S => visit_i32_trunc_f32_s + @mvp I32TruncF32U => visit_i32_trunc_f32_u + @mvp I32TruncF64S => visit_i32_trunc_f64_s + @mvp I32TruncF64U => visit_i32_trunc_f64_u + @mvp I64ExtendI32S => visit_i64_extend_i32_s + @mvp I64ExtendI32U => visit_i64_extend_i32_u + @mvp I64TruncF32S => visit_i64_trunc_f32_s + @mvp I64TruncF32U => visit_i64_trunc_f32_u + @mvp I64TruncF64S => visit_i64_trunc_f64_s + @mvp I64TruncF64U => visit_i64_trunc_f64_u + @mvp F32ConvertI32S => visit_f32_convert_i32_s + @mvp F32ConvertI32U => visit_f32_convert_i32_u + @mvp F32ConvertI64S => visit_f32_convert_i64_s + @mvp F32ConvertI64U => visit_f32_convert_i64_u + @mvp F32DemoteF64 => visit_f32_demote_f64 + @mvp F64ConvertI32S => visit_f64_convert_i32_s + @mvp F64ConvertI32U => visit_f64_convert_i32_u + @mvp F64ConvertI64S => visit_f64_convert_i64_s + @mvp F64ConvertI64U => visit_f64_convert_i64_u + @mvp F64PromoteF32 => visit_f64_promote_f32 + @mvp I32ReinterpretF32 => visit_i32_reinterpret_f32 + @mvp I64ReinterpretF64 => visit_i64_reinterpret_f64 + @mvp F32ReinterpretI32 => visit_f32_reinterpret_i32 + @mvp F64ReinterpretI64 => visit_f64_reinterpret_i64 + @sign_extension I32Extend8S => visit_i32_extend8_s + @sign_extension I32Extend16S => visit_i32_extend16_s + @sign_extension I64Extend8S => visit_i64_extend8_s + @sign_extension I64Extend16S => visit_i64_extend16_s + @sign_extension I64Extend32S => visit_i64_extend32_s + + // 0xFC operators + // Non-trapping Float-to-int Conversions + // https://github.com/WebAssembly/nontrapping-float-to-int-conversions + @saturating_float_to_int I32TruncSatF32S => visit_i32_trunc_sat_f32_s + @saturating_float_to_int I32TruncSatF32U => visit_i32_trunc_sat_f32_u + @saturating_float_to_int I32TruncSatF64S => visit_i32_trunc_sat_f64_s + @saturating_float_to_int I32TruncSatF64U => visit_i32_trunc_sat_f64_u + @saturating_float_to_int I64TruncSatF32S => visit_i64_trunc_sat_f32_s + @saturating_float_to_int I64TruncSatF32U => visit_i64_trunc_sat_f32_u + @saturating_float_to_int I64TruncSatF64S => visit_i64_trunc_sat_f64_s + @saturating_float_to_int I64TruncSatF64U => visit_i64_trunc_sat_f64_u + + // 0xFC prefixed operators + // bulk memory operations + // https://github.com/WebAssembly/bulk-memory-operations + @bulk_memory MemoryInit { data_index: u32, mem: u32 } => visit_memory_init + @bulk_memory DataDrop { data_index: u32 } => visit_data_drop + @bulk_memory MemoryCopy { dst_mem: u32, src_mem: u32 } => visit_memory_copy + @bulk_memory MemoryFill { mem: u32 } => visit_memory_fill + @bulk_memory TableInit { elem_index: u32, table: u32 } => visit_table_init + @bulk_memory ElemDrop { elem_index: u32 } => visit_elem_drop + @bulk_memory TableCopy { dst_table: u32, src_table: u32 } => visit_table_copy + + // 0xFC prefixed operators + // reference-types + // https://github.com/WebAssembly/reference-types + @reference_types TableFill { table: u32 } => visit_table_fill + @reference_types TableGet { table: u32 } => visit_table_get + @reference_types TableSet { table: u32 } => visit_table_set + @reference_types TableGrow { table: u32 } => visit_table_grow + @reference_types TableSize { table: u32 } => visit_table_size + + // OxFC prefixed operators + // memory control (experimental) + // https://github.com/WebAssembly/design/issues/1439 + @memory_control MemoryDiscard { mem: u32 } => visit_memory_discard + + // 0xFE prefixed operators + // threads + // https://github.com/WebAssembly/threads + @threads MemoryAtomicNotify { memarg: $crate::MemArg } => visit_memory_atomic_notify + @threads MemoryAtomicWait32 { memarg: $crate::MemArg } => visit_memory_atomic_wait32 + @threads MemoryAtomicWait64 { memarg: $crate::MemArg } => visit_memory_atomic_wait64 + @threads AtomicFence => visit_atomic_fence + @threads I32AtomicLoad { memarg: $crate::MemArg } => visit_i32_atomic_load + @threads I64AtomicLoad { memarg: $crate::MemArg } => visit_i64_atomic_load + @threads I32AtomicLoad8U { memarg: $crate::MemArg } => visit_i32_atomic_load8_u + @threads I32AtomicLoad16U { memarg: $crate::MemArg } => visit_i32_atomic_load16_u + @threads I64AtomicLoad8U { memarg: $crate::MemArg } => visit_i64_atomic_load8_u + @threads I64AtomicLoad16U { memarg: $crate::MemArg } => visit_i64_atomic_load16_u + @threads I64AtomicLoad32U { memarg: $crate::MemArg } => visit_i64_atomic_load32_u + @threads I32AtomicStore { memarg: $crate::MemArg } => visit_i32_atomic_store + @threads I64AtomicStore { memarg: $crate::MemArg } => visit_i64_atomic_store + @threads I32AtomicStore8 { memarg: $crate::MemArg } => visit_i32_atomic_store8 + @threads I32AtomicStore16 { memarg: $crate::MemArg } => visit_i32_atomic_store16 + @threads I64AtomicStore8 { memarg: $crate::MemArg } => visit_i64_atomic_store8 + @threads I64AtomicStore16 { memarg: $crate::MemArg } => visit_i64_atomic_store16 + @threads I64AtomicStore32 { memarg: $crate::MemArg } => visit_i64_atomic_store32 + @threads I32AtomicRmwAdd { memarg: $crate::MemArg } => visit_i32_atomic_rmw_add + @threads I64AtomicRmwAdd { memarg: $crate::MemArg } => visit_i64_atomic_rmw_add + @threads I32AtomicRmw8AddU { memarg: $crate::MemArg } => visit_i32_atomic_rmw8_add_u + @threads I32AtomicRmw16AddU { memarg: $crate::MemArg } => visit_i32_atomic_rmw16_add_u + @threads I64AtomicRmw8AddU { memarg: $crate::MemArg } => visit_i64_atomic_rmw8_add_u + @threads I64AtomicRmw16AddU { memarg: $crate::MemArg } => visit_i64_atomic_rmw16_add_u + @threads I64AtomicRmw32AddU { memarg: $crate::MemArg } => visit_i64_atomic_rmw32_add_u + @threads I32AtomicRmwSub { memarg: $crate::MemArg } => visit_i32_atomic_rmw_sub + @threads I64AtomicRmwSub { memarg: $crate::MemArg } => visit_i64_atomic_rmw_sub + @threads I32AtomicRmw8SubU { memarg: $crate::MemArg } => visit_i32_atomic_rmw8_sub_u + @threads I32AtomicRmw16SubU { memarg: $crate::MemArg } => visit_i32_atomic_rmw16_sub_u + @threads I64AtomicRmw8SubU { memarg: $crate::MemArg } => visit_i64_atomic_rmw8_sub_u + @threads I64AtomicRmw16SubU { memarg: $crate::MemArg } => visit_i64_atomic_rmw16_sub_u + @threads I64AtomicRmw32SubU { memarg: $crate::MemArg } => visit_i64_atomic_rmw32_sub_u + @threads I32AtomicRmwAnd { memarg: $crate::MemArg } => visit_i32_atomic_rmw_and + @threads I64AtomicRmwAnd { memarg: $crate::MemArg } => visit_i64_atomic_rmw_and + @threads I32AtomicRmw8AndU { memarg: $crate::MemArg } => visit_i32_atomic_rmw8_and_u + @threads I32AtomicRmw16AndU { memarg: $crate::MemArg } => visit_i32_atomic_rmw16_and_u + @threads I64AtomicRmw8AndU { memarg: $crate::MemArg } => visit_i64_atomic_rmw8_and_u + @threads I64AtomicRmw16AndU { memarg: $crate::MemArg } => visit_i64_atomic_rmw16_and_u + @threads I64AtomicRmw32AndU { memarg: $crate::MemArg } => visit_i64_atomic_rmw32_and_u + @threads I32AtomicRmwOr { memarg: $crate::MemArg } => visit_i32_atomic_rmw_or + @threads I64AtomicRmwOr { memarg: $crate::MemArg } => visit_i64_atomic_rmw_or + @threads I32AtomicRmw8OrU { memarg: $crate::MemArg } => visit_i32_atomic_rmw8_or_u + @threads I32AtomicRmw16OrU { memarg: $crate::MemArg } => visit_i32_atomic_rmw16_or_u + @threads I64AtomicRmw8OrU { memarg: $crate::MemArg } => visit_i64_atomic_rmw8_or_u + @threads I64AtomicRmw16OrU { memarg: $crate::MemArg } => visit_i64_atomic_rmw16_or_u + @threads I64AtomicRmw32OrU { memarg: $crate::MemArg } => visit_i64_atomic_rmw32_or_u + @threads I32AtomicRmwXor { memarg: $crate::MemArg } => visit_i32_atomic_rmw_xor + @threads I64AtomicRmwXor { memarg: $crate::MemArg } => visit_i64_atomic_rmw_xor + @threads I32AtomicRmw8XorU { memarg: $crate::MemArg } => visit_i32_atomic_rmw8_xor_u + @threads I32AtomicRmw16XorU { memarg: $crate::MemArg } => visit_i32_atomic_rmw16_xor_u + @threads I64AtomicRmw8XorU { memarg: $crate::MemArg } => visit_i64_atomic_rmw8_xor_u + @threads I64AtomicRmw16XorU { memarg: $crate::MemArg } => visit_i64_atomic_rmw16_xor_u + @threads I64AtomicRmw32XorU { memarg: $crate::MemArg } => visit_i64_atomic_rmw32_xor_u + @threads I32AtomicRmwXchg { memarg: $crate::MemArg } => visit_i32_atomic_rmw_xchg + @threads I64AtomicRmwXchg { memarg: $crate::MemArg } => visit_i64_atomic_rmw_xchg + @threads I32AtomicRmw8XchgU { memarg: $crate::MemArg } => visit_i32_atomic_rmw8_xchg_u + @threads I32AtomicRmw16XchgU { memarg: $crate::MemArg } => visit_i32_atomic_rmw16_xchg_u + @threads I64AtomicRmw8XchgU { memarg: $crate::MemArg } => visit_i64_atomic_rmw8_xchg_u + @threads I64AtomicRmw16XchgU { memarg: $crate::MemArg } => visit_i64_atomic_rmw16_xchg_u + @threads I64AtomicRmw32XchgU { memarg: $crate::MemArg } => visit_i64_atomic_rmw32_xchg_u + @threads I32AtomicRmwCmpxchg { memarg: $crate::MemArg } => visit_i32_atomic_rmw_cmpxchg + @threads I64AtomicRmwCmpxchg { memarg: $crate::MemArg } => visit_i64_atomic_rmw_cmpxchg + @threads I32AtomicRmw8CmpxchgU { memarg: $crate::MemArg } => visit_i32_atomic_rmw8_cmpxchg_u + @threads I32AtomicRmw16CmpxchgU { memarg: $crate::MemArg } => visit_i32_atomic_rmw16_cmpxchg_u + @threads I64AtomicRmw8CmpxchgU { memarg: $crate::MemArg } => visit_i64_atomic_rmw8_cmpxchg_u + @threads I64AtomicRmw16CmpxchgU { memarg: $crate::MemArg } => visit_i64_atomic_rmw16_cmpxchg_u + @threads I64AtomicRmw32CmpxchgU { memarg: $crate::MemArg } => visit_i64_atomic_rmw32_cmpxchg_u + + // 0xFD operators + // 128-bit SIMD + // - https://github.com/webassembly/simd + // - https://webassembly.github.io/simd/core/binary/instructions.html + @simd V128Load { memarg: $crate::MemArg } => visit_v128_load + @simd V128Load8x8S { memarg: $crate::MemArg } => visit_v128_load8x8_s + @simd V128Load8x8U { memarg: $crate::MemArg } => visit_v128_load8x8_u + @simd V128Load16x4S { memarg: $crate::MemArg } => visit_v128_load16x4_s + @simd V128Load16x4U { memarg: $crate::MemArg } => visit_v128_load16x4_u + @simd V128Load32x2S { memarg: $crate::MemArg } => visit_v128_load32x2_s + @simd V128Load32x2U { memarg: $crate::MemArg } => visit_v128_load32x2_u + @simd V128Load8Splat { memarg: $crate::MemArg } => visit_v128_load8_splat + @simd V128Load16Splat { memarg: $crate::MemArg } => visit_v128_load16_splat + @simd V128Load32Splat { memarg: $crate::MemArg } => visit_v128_load32_splat + @simd V128Load64Splat { memarg: $crate::MemArg } => visit_v128_load64_splat + @simd V128Load32Zero { memarg: $crate::MemArg } => visit_v128_load32_zero + @simd V128Load64Zero { memarg: $crate::MemArg } => visit_v128_load64_zero + @simd V128Store { memarg: $crate::MemArg } => visit_v128_store + @simd V128Load8Lane { memarg: $crate::MemArg, lane: u8 } => visit_v128_load8_lane + @simd V128Load16Lane { memarg: $crate::MemArg, lane: u8 } => visit_v128_load16_lane + @simd V128Load32Lane { memarg: $crate::MemArg, lane: u8 } => visit_v128_load32_lane + @simd V128Load64Lane { memarg: $crate::MemArg, lane: u8 } => visit_v128_load64_lane + @simd V128Store8Lane { memarg: $crate::MemArg, lane: u8 } => visit_v128_store8_lane + @simd V128Store16Lane { memarg: $crate::MemArg, lane: u8 } => visit_v128_store16_lane + @simd V128Store32Lane { memarg: $crate::MemArg, lane: u8 } => visit_v128_store32_lane + @simd V128Store64Lane { memarg: $crate::MemArg, lane: u8 } => visit_v128_store64_lane + @simd V128Const { value: $crate::V128 } => visit_v128_const + @simd I8x16Shuffle { lanes: [u8; 16] } => visit_i8x16_shuffle + @simd I8x16ExtractLaneS { lane: u8 } => visit_i8x16_extract_lane_s + @simd I8x16ExtractLaneU { lane: u8 } => visit_i8x16_extract_lane_u + @simd I8x16ReplaceLane { lane: u8 } => visit_i8x16_replace_lane + @simd I16x8ExtractLaneS { lane: u8 } => visit_i16x8_extract_lane_s + @simd I16x8ExtractLaneU { lane: u8 } => visit_i16x8_extract_lane_u + @simd I16x8ReplaceLane { lane: u8 } => visit_i16x8_replace_lane + @simd I32x4ExtractLane { lane: u8 } => visit_i32x4_extract_lane + @simd I32x4ReplaceLane { lane: u8 } => visit_i32x4_replace_lane + @simd I64x2ExtractLane { lane: u8 } => visit_i64x2_extract_lane + @simd I64x2ReplaceLane { lane: u8 } => visit_i64x2_replace_lane + @simd F32x4ExtractLane { lane: u8 } => visit_f32x4_extract_lane + @simd F32x4ReplaceLane { lane: u8 } => visit_f32x4_replace_lane + @simd F64x2ExtractLane { lane: u8 } => visit_f64x2_extract_lane + @simd F64x2ReplaceLane { lane: u8 } => visit_f64x2_replace_lane + @simd I8x16Swizzle => visit_i8x16_swizzle + @simd I8x16Splat => visit_i8x16_splat + @simd I16x8Splat => visit_i16x8_splat + @simd I32x4Splat => visit_i32x4_splat + @simd I64x2Splat => visit_i64x2_splat + @simd F32x4Splat => visit_f32x4_splat + @simd F64x2Splat => visit_f64x2_splat + @simd I8x16Eq => visit_i8x16_eq + @simd I8x16Ne => visit_i8x16_ne + @simd I8x16LtS => visit_i8x16_lt_s + @simd I8x16LtU => visit_i8x16_lt_u + @simd I8x16GtS => visit_i8x16_gt_s + @simd I8x16GtU => visit_i8x16_gt_u + @simd I8x16LeS => visit_i8x16_le_s + @simd I8x16LeU => visit_i8x16_le_u + @simd I8x16GeS => visit_i8x16_ge_s + @simd I8x16GeU => visit_i8x16_ge_u + @simd I16x8Eq => visit_i16x8_eq + @simd I16x8Ne => visit_i16x8_ne + @simd I16x8LtS => visit_i16x8_lt_s + @simd I16x8LtU => visit_i16x8_lt_u + @simd I16x8GtS => visit_i16x8_gt_s + @simd I16x8GtU => visit_i16x8_gt_u + @simd I16x8LeS => visit_i16x8_le_s + @simd I16x8LeU => visit_i16x8_le_u + @simd I16x8GeS => visit_i16x8_ge_s + @simd I16x8GeU => visit_i16x8_ge_u + @simd I32x4Eq => visit_i32x4_eq + @simd I32x4Ne => visit_i32x4_ne + @simd I32x4LtS => visit_i32x4_lt_s + @simd I32x4LtU => visit_i32x4_lt_u + @simd I32x4GtS => visit_i32x4_gt_s + @simd I32x4GtU => visit_i32x4_gt_u + @simd I32x4LeS => visit_i32x4_le_s + @simd I32x4LeU => visit_i32x4_le_u + @simd I32x4GeS => visit_i32x4_ge_s + @simd I32x4GeU => visit_i32x4_ge_u + @simd I64x2Eq => visit_i64x2_eq + @simd I64x2Ne => visit_i64x2_ne + @simd I64x2LtS => visit_i64x2_lt_s + @simd I64x2GtS => visit_i64x2_gt_s + @simd I64x2LeS => visit_i64x2_le_s + @simd I64x2GeS => visit_i64x2_ge_s + @simd F32x4Eq => visit_f32x4_eq + @simd F32x4Ne => visit_f32x4_ne + @simd F32x4Lt => visit_f32x4_lt + @simd F32x4Gt => visit_f32x4_gt + @simd F32x4Le => visit_f32x4_le + @simd F32x4Ge => visit_f32x4_ge + @simd F64x2Eq => visit_f64x2_eq + @simd F64x2Ne => visit_f64x2_ne + @simd F64x2Lt => visit_f64x2_lt + @simd F64x2Gt => visit_f64x2_gt + @simd F64x2Le => visit_f64x2_le + @simd F64x2Ge => visit_f64x2_ge + @simd V128Not => visit_v128_not + @simd V128And => visit_v128_and + @simd V128AndNot => visit_v128_andnot + @simd V128Or => visit_v128_or + @simd V128Xor => visit_v128_xor + @simd V128Bitselect => visit_v128_bitselect + @simd V128AnyTrue => visit_v128_any_true + @simd I8x16Abs => visit_i8x16_abs + @simd I8x16Neg => visit_i8x16_neg + @simd I8x16Popcnt => visit_i8x16_popcnt + @simd I8x16AllTrue => visit_i8x16_all_true + @simd I8x16Bitmask => visit_i8x16_bitmask + @simd I8x16NarrowI16x8S => visit_i8x16_narrow_i16x8_s + @simd I8x16NarrowI16x8U => visit_i8x16_narrow_i16x8_u + @simd I8x16Shl => visit_i8x16_shl + @simd I8x16ShrS => visit_i8x16_shr_s + @simd I8x16ShrU => visit_i8x16_shr_u + @simd I8x16Add => visit_i8x16_add + @simd I8x16AddSatS => visit_i8x16_add_sat_s + @simd I8x16AddSatU => visit_i8x16_add_sat_u + @simd I8x16Sub => visit_i8x16_sub + @simd I8x16SubSatS => visit_i8x16_sub_sat_s + @simd I8x16SubSatU => visit_i8x16_sub_sat_u + @simd I8x16MinS => visit_i8x16_min_s + @simd I8x16MinU => visit_i8x16_min_u + @simd I8x16MaxS => visit_i8x16_max_s + @simd I8x16MaxU => visit_i8x16_max_u + @simd I8x16AvgrU => visit_i8x16_avgr_u + @simd I16x8ExtAddPairwiseI8x16S => visit_i16x8_extadd_pairwise_i8x16_s + @simd I16x8ExtAddPairwiseI8x16U => visit_i16x8_extadd_pairwise_i8x16_u + @simd I16x8Abs => visit_i16x8_abs + @simd I16x8Neg => visit_i16x8_neg + @simd I16x8Q15MulrSatS => visit_i16x8_q15mulr_sat_s + @simd I16x8AllTrue => visit_i16x8_all_true + @simd I16x8Bitmask => visit_i16x8_bitmask + @simd I16x8NarrowI32x4S => visit_i16x8_narrow_i32x4_s + @simd I16x8NarrowI32x4U => visit_i16x8_narrow_i32x4_u + @simd I16x8ExtendLowI8x16S => visit_i16x8_extend_low_i8x16_s + @simd I16x8ExtendHighI8x16S => visit_i16x8_extend_high_i8x16_s + @simd I16x8ExtendLowI8x16U => visit_i16x8_extend_low_i8x16_u + @simd I16x8ExtendHighI8x16U => visit_i16x8_extend_high_i8x16_u + @simd I16x8Shl => visit_i16x8_shl + @simd I16x8ShrS => visit_i16x8_shr_s + @simd I16x8ShrU => visit_i16x8_shr_u + @simd I16x8Add => visit_i16x8_add + @simd I16x8AddSatS => visit_i16x8_add_sat_s + @simd I16x8AddSatU => visit_i16x8_add_sat_u + @simd I16x8Sub => visit_i16x8_sub + @simd I16x8SubSatS => visit_i16x8_sub_sat_s + @simd I16x8SubSatU => visit_i16x8_sub_sat_u + @simd I16x8Mul => visit_i16x8_mul + @simd I16x8MinS => visit_i16x8_min_s + @simd I16x8MinU => visit_i16x8_min_u + @simd I16x8MaxS => visit_i16x8_max_s + @simd I16x8MaxU => visit_i16x8_max_u + @simd I16x8AvgrU => visit_i16x8_avgr_u + @simd I16x8ExtMulLowI8x16S => visit_i16x8_extmul_low_i8x16_s + @simd I16x8ExtMulHighI8x16S => visit_i16x8_extmul_high_i8x16_s + @simd I16x8ExtMulLowI8x16U => visit_i16x8_extmul_low_i8x16_u + @simd I16x8ExtMulHighI8x16U => visit_i16x8_extmul_high_i8x16_u + @simd I32x4ExtAddPairwiseI16x8S => visit_i32x4_extadd_pairwise_i16x8_s + @simd I32x4ExtAddPairwiseI16x8U => visit_i32x4_extadd_pairwise_i16x8_u + @simd I32x4Abs => visit_i32x4_abs + @simd I32x4Neg => visit_i32x4_neg + @simd I32x4AllTrue => visit_i32x4_all_true + @simd I32x4Bitmask => visit_i32x4_bitmask + @simd I32x4ExtendLowI16x8S => visit_i32x4_extend_low_i16x8_s + @simd I32x4ExtendHighI16x8S => visit_i32x4_extend_high_i16x8_s + @simd I32x4ExtendLowI16x8U => visit_i32x4_extend_low_i16x8_u + @simd I32x4ExtendHighI16x8U => visit_i32x4_extend_high_i16x8_u + @simd I32x4Shl => visit_i32x4_shl + @simd I32x4ShrS => visit_i32x4_shr_s + @simd I32x4ShrU => visit_i32x4_shr_u + @simd I32x4Add => visit_i32x4_add + @simd I32x4Sub => visit_i32x4_sub + @simd I32x4Mul => visit_i32x4_mul + @simd I32x4MinS => visit_i32x4_min_s + @simd I32x4MinU => visit_i32x4_min_u + @simd I32x4MaxS => visit_i32x4_max_s + @simd I32x4MaxU => visit_i32x4_max_u + @simd I32x4DotI16x8S => visit_i32x4_dot_i16x8_s + @simd I32x4ExtMulLowI16x8S => visit_i32x4_extmul_low_i16x8_s + @simd I32x4ExtMulHighI16x8S => visit_i32x4_extmul_high_i16x8_s + @simd I32x4ExtMulLowI16x8U => visit_i32x4_extmul_low_i16x8_u + @simd I32x4ExtMulHighI16x8U => visit_i32x4_extmul_high_i16x8_u + @simd I64x2Abs => visit_i64x2_abs + @simd I64x2Neg => visit_i64x2_neg + @simd I64x2AllTrue => visit_i64x2_all_true + @simd I64x2Bitmask => visit_i64x2_bitmask + @simd I64x2ExtendLowI32x4S => visit_i64x2_extend_low_i32x4_s + @simd I64x2ExtendHighI32x4S => visit_i64x2_extend_high_i32x4_s + @simd I64x2ExtendLowI32x4U => visit_i64x2_extend_low_i32x4_u + @simd I64x2ExtendHighI32x4U => visit_i64x2_extend_high_i32x4_u + @simd I64x2Shl => visit_i64x2_shl + @simd I64x2ShrS => visit_i64x2_shr_s + @simd I64x2ShrU => visit_i64x2_shr_u + @simd I64x2Add => visit_i64x2_add + @simd I64x2Sub => visit_i64x2_sub + @simd I64x2Mul => visit_i64x2_mul + @simd I64x2ExtMulLowI32x4S => visit_i64x2_extmul_low_i32x4_s + @simd I64x2ExtMulHighI32x4S => visit_i64x2_extmul_high_i32x4_s + @simd I64x2ExtMulLowI32x4U => visit_i64x2_extmul_low_i32x4_u + @simd I64x2ExtMulHighI32x4U => visit_i64x2_extmul_high_i32x4_u + @simd F32x4Ceil => visit_f32x4_ceil + @simd F32x4Floor => visit_f32x4_floor + @simd F32x4Trunc => visit_f32x4_trunc + @simd F32x4Nearest => visit_f32x4_nearest + @simd F32x4Abs => visit_f32x4_abs + @simd F32x4Neg => visit_f32x4_neg + @simd F32x4Sqrt => visit_f32x4_sqrt + @simd F32x4Add => visit_f32x4_add + @simd F32x4Sub => visit_f32x4_sub + @simd F32x4Mul => visit_f32x4_mul + @simd F32x4Div => visit_f32x4_div + @simd F32x4Min => visit_f32x4_min + @simd F32x4Max => visit_f32x4_max + @simd F32x4PMin => visit_f32x4_pmin + @simd F32x4PMax => visit_f32x4_pmax + @simd F64x2Ceil => visit_f64x2_ceil + @simd F64x2Floor => visit_f64x2_floor + @simd F64x2Trunc => visit_f64x2_trunc + @simd F64x2Nearest => visit_f64x2_nearest + @simd F64x2Abs => visit_f64x2_abs + @simd F64x2Neg => visit_f64x2_neg + @simd F64x2Sqrt => visit_f64x2_sqrt + @simd F64x2Add => visit_f64x2_add + @simd F64x2Sub => visit_f64x2_sub + @simd F64x2Mul => visit_f64x2_mul + @simd F64x2Div => visit_f64x2_div + @simd F64x2Min => visit_f64x2_min + @simd F64x2Max => visit_f64x2_max + @simd F64x2PMin => visit_f64x2_pmin + @simd F64x2PMax => visit_f64x2_pmax + @simd I32x4TruncSatF32x4S => visit_i32x4_trunc_sat_f32x4_s + @simd I32x4TruncSatF32x4U => visit_i32x4_trunc_sat_f32x4_u + @simd F32x4ConvertI32x4S => visit_f32x4_convert_i32x4_s + @simd F32x4ConvertI32x4U => visit_f32x4_convert_i32x4_u + @simd I32x4TruncSatF64x2SZero => visit_i32x4_trunc_sat_f64x2_s_zero + @simd I32x4TruncSatF64x2UZero => visit_i32x4_trunc_sat_f64x2_u_zero + @simd F64x2ConvertLowI32x4S => visit_f64x2_convert_low_i32x4_s + @simd F64x2ConvertLowI32x4U => visit_f64x2_convert_low_i32x4_u + @simd F32x4DemoteF64x2Zero => visit_f32x4_demote_f64x2_zero + @simd F64x2PromoteLowF32x4 => visit_f64x2_promote_low_f32x4 + + // Relaxed SIMD operators + // https://github.com/WebAssembly/relaxed-simd + @relaxed_simd I8x16RelaxedSwizzle => visit_i8x16_relaxed_swizzle + @relaxed_simd I32x4RelaxedTruncF32x4S => visit_i32x4_relaxed_trunc_f32x4_s + @relaxed_simd I32x4RelaxedTruncF32x4U => visit_i32x4_relaxed_trunc_f32x4_u + @relaxed_simd I32x4RelaxedTruncF64x2SZero => visit_i32x4_relaxed_trunc_f64x2_s_zero + @relaxed_simd I32x4RelaxedTruncF64x2UZero => visit_i32x4_relaxed_trunc_f64x2_u_zero + @relaxed_simd F32x4RelaxedMadd => visit_f32x4_relaxed_madd + @relaxed_simd F32x4RelaxedNmadd => visit_f32x4_relaxed_nmadd + @relaxed_simd F64x2RelaxedMadd => visit_f64x2_relaxed_madd + @relaxed_simd F64x2RelaxedNmadd => visit_f64x2_relaxed_nmadd + @relaxed_simd I8x16RelaxedLaneselect => visit_i8x16_relaxed_laneselect + @relaxed_simd I16x8RelaxedLaneselect => visit_i16x8_relaxed_laneselect + @relaxed_simd I32x4RelaxedLaneselect => visit_i32x4_relaxed_laneselect + @relaxed_simd I64x2RelaxedLaneselect => visit_i64x2_relaxed_laneselect + @relaxed_simd F32x4RelaxedMin => visit_f32x4_relaxed_min + @relaxed_simd F32x4RelaxedMax => visit_f32x4_relaxed_max + @relaxed_simd F64x2RelaxedMin => visit_f64x2_relaxed_min + @relaxed_simd F64x2RelaxedMax => visit_f64x2_relaxed_max + @relaxed_simd I16x8RelaxedQ15mulrS => visit_i16x8_relaxed_q15mulr_s + @relaxed_simd I16x8RelaxedDotI8x16I7x16S => visit_i16x8_relaxed_dot_i8x16_i7x16_s + @relaxed_simd I32x4RelaxedDotI8x16I7x16AddS => visit_i32x4_relaxed_dot_i8x16_i7x16_add_s + + // Typed Function references + @function_references CallRef { hty: $crate::HeapType } => visit_call_ref + @function_references ReturnCallRef { hty: $crate::HeapType } => visit_return_call_ref + @function_references RefAsNonNull => visit_ref_as_non_null + @function_references BrOnNull { relative_depth: u32 } => visit_br_on_null + @function_references BrOnNonNull { relative_depth: u32 } => visit_br_on_non_null + } + }; +} + +macro_rules! format_err { + ($offset:expr, $($arg:tt)*) => { + crate::BinaryReaderError::fmt(format_args!($($arg)*), $offset) + } +} + +macro_rules! bail { + ($($arg:tt)*) => {return Err(format_err!($($arg)*))} +} + +pub use crate::binary_reader::{BinaryReader, BinaryReaderError, Result}; +pub use crate::parser::*; +pub use crate::readers::*; +pub use crate::resources::*; +pub use crate::validator::*; + +mod binary_reader; +mod limits; +mod parser; +mod readers; +mod resources; +mod validator; diff --git a/third_party/rust/wasmparser/src/limits.rs b/third_party/rust/wasmparser/src/limits.rs new file mode 100644 index 0000000000..e9ab7d06a6 --- /dev/null +++ b/third_party/rust/wasmparser/src/limits.rs @@ -0,0 +1,57 @@ +/* Copyright 2017 Mozilla Foundation + * + * 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. + */ + +// The following limits are imposed by wasmparser on WebAssembly modules. +// The limits are agreed upon with other engines for consistency. +pub const MAX_WASM_TYPES: usize = 1_000_000; +pub const MAX_WASM_FUNCTIONS: usize = 1_000_000; +pub const MAX_WASM_EXPORTS: usize = 100_000; +pub const MAX_WASM_GLOBALS: usize = 1_000_000; +pub const MAX_WASM_ELEMENT_SEGMENTS: usize = 100_000; +pub const MAX_WASM_DATA_SEGMENTS: usize = 100_000; +pub const MAX_WASM_MEMORY32_PAGES: u64 = 65536; +pub const MAX_WASM_MEMORY64_PAGES: u64 = 1 << 48; +pub const MAX_WASM_STRING_SIZE: usize = 100_000; +pub const MAX_WASM_FUNCTION_SIZE: usize = 128 * 1024; +pub const MAX_WASM_FUNCTION_LOCALS: usize = 50000; +pub const MAX_WASM_FUNCTION_PARAMS: usize = 1000; +pub const MAX_WASM_FUNCTION_RETURNS: usize = 1000; +pub const _MAX_WASM_TABLE_SIZE: usize = 10_000_000; +pub const MAX_WASM_TABLE_ENTRIES: usize = 10_000_000; +pub const MAX_WASM_TABLES: usize = 100; +pub const MAX_WASM_MEMORIES: usize = 100; +pub const MAX_WASM_TAGS: usize = 1_000_000; +pub const MAX_WASM_BR_TABLE_SIZE: usize = MAX_WASM_FUNCTION_SIZE; + +// Component-related limits +pub const MAX_WASM_MODULE_SIZE: usize = 1024 * 1024 * 1024; //= 1 GiB +pub const MAX_WASM_MODULE_TYPE_DECLS: usize = 100_000; +pub const MAX_WASM_COMPONENT_TYPE_DECLS: usize = 100_000; +pub const MAX_WASM_INSTANCE_TYPE_DECLS: usize = 100_000; +pub const MAX_WASM_RECORD_FIELDS: usize = 1000; +pub const MAX_WASM_VARIANT_CASES: usize = 1000; +pub const MAX_WASM_TUPLE_TYPES: usize = 1000; +pub const MAX_WASM_FLAG_NAMES: usize = 1000; +pub const MAX_WASM_ENUM_CASES: usize = 1000; +pub const MAX_WASM_UNION_TYPES: usize = 1000; +pub const MAX_WASM_INSTANTIATION_EXPORTS: usize = 1000; +pub const MAX_WASM_CANONICAL_OPTIONS: usize = 10; +pub const MAX_WASM_INSTANTIATION_ARGS: usize = 1000; +pub const MAX_WASM_START_ARGS: usize = 1000; +pub const MAX_WASM_TYPE_SIZE: u32 = 1_000_000; +pub const MAX_WASM_MODULES: usize = 1_000; +pub const MAX_WASM_COMPONENTS: usize = 1_000; +pub const MAX_WASM_INSTANCES: usize = 1_000; +pub const MAX_WASM_VALUES: usize = 1_000; diff --git a/third_party/rust/wasmparser/src/parser.rs b/third_party/rust/wasmparser/src/parser.rs new file mode 100644 index 0000000000..227530abba --- /dev/null +++ b/third_party/rust/wasmparser/src/parser.rs @@ -0,0 +1,1496 @@ +use crate::CoreTypeSectionReader; +use crate::{ + limits::MAX_WASM_MODULE_SIZE, BinaryReader, BinaryReaderError, ComponentCanonicalSectionReader, + ComponentExportSectionReader, ComponentImportSectionReader, ComponentInstanceSectionReader, + ComponentStartFunction, ComponentTypeSectionReader, CustomSectionReader, DataSectionReader, + ElementSectionReader, ExportSectionReader, FromReader, FunctionBody, FunctionSectionReader, + GlobalSectionReader, ImportSectionReader, InstanceSectionReader, MemorySectionReader, Result, + SectionLimited, TableSectionReader, TagSectionReader, TypeSectionReader, +}; +use std::convert::TryInto; +use std::fmt; +use std::iter; +use std::ops::Range; + +pub(crate) const WASM_MODULE_VERSION: u16 = 0x1; + +// Note that this started at `0xa` and we're incrementing up from there. When +// the component model is stabilized this will become 0x1. The changes here are: +// +// * [????-??-??] 0xa - original version +// * [2022-01-05] 0xb - `export` introduces an alias +// * [2022-02-06] 0xc - `export` has an optional type ascribed to it +pub(crate) const WASM_COMPONENT_VERSION: u16 = 0xc; + +/// The supported encoding formats for the parser. +#[derive(Debug, Clone, Copy, Eq, PartialEq)] +pub enum Encoding { + /// The encoding format is a WebAssembly module. + Module, + /// The encoding format is a WebAssembly component. + Component, +} + +/// An incremental parser of a binary WebAssembly module or component. +/// +/// This type is intended to be used to incrementally parse a WebAssembly module +/// or component as bytes become available for the module. This can also be used +/// to parse modules or components that are already entirely resident within memory. +/// +/// This primary function for a parser is the [`Parser::parse`] function which +/// will incrementally consume input. You can also use the [`Parser::parse_all`] +/// function to parse a module or component that is entirely resident in memory. +#[derive(Debug, Clone)] +pub struct Parser { + state: State, + offset: u64, + max_size: u64, + encoding: Encoding, +} + +#[derive(Debug, Clone)] +enum State { + Header, + SectionStart, + FunctionBody { remaining: u32, len: u32 }, +} + +/// A successful return payload from [`Parser::parse`]. +/// +/// On success one of two possible values can be returned, either that more data +/// is needed to continue parsing or a chunk of the input was parsed, indicating +/// how much of it was parsed. +#[derive(Debug)] +pub enum Chunk<'a> { + /// This can be returned at any time and indicates that more data is needed + /// to proceed with parsing. Zero bytes were consumed from the input to + /// [`Parser::parse`]. The `usize` value here is a hint as to how many more + /// bytes are needed to continue parsing. + NeedMoreData(u64), + + /// A chunk was successfully parsed. + Parsed { + /// This many bytes of the `data` input to [`Parser::parse`] were + /// consumed to produce `payload`. + consumed: usize, + /// The value that we actually parsed. + payload: Payload<'a>, + }, +} + +/// Values that can be parsed from a WebAssembly module or component. +/// +/// This enumeration is all possible chunks of pieces that can be parsed by a +/// [`Parser`] from a binary WebAssembly module or component. Note that for many +/// sections the entire section is parsed all at once, whereas other functions, +/// like the code section, are parsed incrementally. This is a distinction where some +/// sections, like the type section, are required to be fully resident in memory +/// (fully downloaded) before proceeding. Other sections, like the code section, +/// can be processed in a streaming fashion where each function is extracted +/// individually so it can possibly be shipped to another thread while you wait +/// for more functions to get downloaded. +/// +/// Note that payloads, when returned, do not indicate that the module or component +/// is valid. For example when you receive a `Payload::TypeSection` the type +/// section itself has not yet actually been parsed. The reader returned will be +/// able to parse it, but you'll have to actually iterate the reader to do the +/// full parse. Each payload returned is intended to be a *window* into the +/// original `data` passed to [`Parser::parse`] which can be further processed +/// if necessary. +pub enum Payload<'a> { + /// Indicates the header of a WebAssembly module or component. + Version { + /// The version number found in the header. + num: u16, + /// The encoding format being parsed. + encoding: Encoding, + /// The range of bytes that were parsed to consume the header of the + /// module or component. Note that this range is relative to the start + /// of the byte stream. + range: Range<usize>, + }, + + /// A module type section was received and the provided reader can be + /// used to parse the contents of the type section. + TypeSection(TypeSectionReader<'a>), + /// A module import section was received and the provided reader can be + /// used to parse the contents of the import section. + ImportSection(ImportSectionReader<'a>), + /// A module function section was received and the provided reader can be + /// used to parse the contents of the function section. + FunctionSection(FunctionSectionReader<'a>), + /// A module table section was received and the provided reader can be + /// used to parse the contents of the table section. + TableSection(TableSectionReader<'a>), + /// A module memory section was received and the provided reader can be + /// used to parse the contents of the memory section. + MemorySection(MemorySectionReader<'a>), + /// A module tag section was received, and the provided reader can be + /// used to parse the contents of the tag section. + TagSection(TagSectionReader<'a>), + /// A module global section was received and the provided reader can be + /// used to parse the contents of the global section. + GlobalSection(GlobalSectionReader<'a>), + /// A module export section was received, and the provided reader can be + /// used to parse the contents of the export section. + ExportSection(ExportSectionReader<'a>), + /// A module start section was received. + StartSection { + /// The start function index + func: u32, + /// The range of bytes that specify the `func` field, specified in + /// offsets relative to the start of the byte stream. + range: Range<usize>, + }, + /// A module element section was received and the provided reader can be + /// used to parse the contents of the element section. + ElementSection(ElementSectionReader<'a>), + /// A module data count section was received. + DataCountSection { + /// The number of data segments. + count: u32, + /// The range of bytes that specify the `count` field, specified in + /// offsets relative to the start of the byte stream. + range: Range<usize>, + }, + /// A module data section was received and the provided reader can be + /// used to parse the contents of the data section. + DataSection(DataSectionReader<'a>), + /// Indicator of the start of the code section of a WebAssembly module. + /// + /// This entry is returned whenever the code section starts. The `count` + /// field indicates how many entries are in this code section. After + /// receiving this start marker you're guaranteed that the next `count` + /// items will be either `CodeSectionEntry` or an error will be returned. + /// + /// This, unlike other sections, is intended to be used for streaming the + /// contents of the code section. The code section is not required to be + /// fully resident in memory when we parse it. Instead a [`Parser`] is + /// capable of parsing piece-by-piece of a code section. + CodeSectionStart { + /// The number of functions in this section. + count: u32, + /// The range of bytes that represent this section, specified in + /// offsets relative to the start of the byte stream. + range: Range<usize>, + /// The size, in bytes, of the remaining contents of this section. + /// + /// This can be used in combination with [`Parser::skip_section`] + /// where the caller will know how many bytes to skip before feeding + /// bytes into `Parser` again. + size: u32, + }, + /// An entry of the code section, a function, was parsed from a WebAssembly + /// module. + /// + /// This entry indicates that a function was successfully received from the + /// code section, and the payload here is the window into the original input + /// where the function resides. Note that the function itself has not been + /// parsed, it's only been outlined. You'll need to process the + /// `FunctionBody` provided to test whether it parses and/or is valid. + CodeSectionEntry(FunctionBody<'a>), + + /// A core module section was received and the provided parser can be + /// used to parse the nested module. + /// + /// This variant is special in that it returns a sub-`Parser`. Upon + /// receiving a `ModuleSection` it is expected that the returned + /// `Parser` will be used instead of the parent `Parser` until the parse has + /// finished. You'll need to feed data into the `Parser` returned until it + /// returns `Payload::End`. After that you'll switch back to the parent + /// parser to resume parsing the rest of the current component. + /// + /// Note that binaries will not be parsed correctly if you feed the data for + /// a nested module into the parent [`Parser`]. + ModuleSection { + /// The parser for the nested module. + parser: Parser, + /// The range of bytes that represent the nested module in the + /// original byte stream. + range: Range<usize>, + }, + /// A core instance section was received and the provided parser can be + /// used to parse the contents of the core instance section. + /// + /// Currently this section is only parsed in a component. + InstanceSection(InstanceSectionReader<'a>), + /// A core type section was received and the provided parser can be + /// used to parse the contents of the core type section. + /// + /// Currently this section is only parsed in a component. + CoreTypeSection(CoreTypeSectionReader<'a>), + /// A component section from a WebAssembly component was received and the + /// provided parser can be used to parse the nested component. + /// + /// This variant is special in that it returns a sub-`Parser`. Upon + /// receiving a `ComponentSection` it is expected that the returned + /// `Parser` will be used instead of the parent `Parser` until the parse has + /// finished. You'll need to feed data into the `Parser` returned until it + /// returns `Payload::End`. After that you'll switch back to the parent + /// parser to resume parsing the rest of the current component. + /// + /// Note that binaries will not be parsed correctly if you feed the data for + /// a nested component into the parent [`Parser`]. + ComponentSection { + /// The parser for the nested component. + parser: Parser, + /// The range of bytes that represent the nested component in the + /// original byte stream. + range: Range<usize>, + }, + /// A component instance section was received and the provided reader can be + /// used to parse the contents of the component instance section. + ComponentInstanceSection(ComponentInstanceSectionReader<'a>), + /// A component alias section was received and the provided reader can be + /// used to parse the contents of the component alias section. + ComponentAliasSection(SectionLimited<'a, crate::ComponentAlias<'a>>), + /// A component type section was received and the provided reader can be + /// used to parse the contents of the component type section. + ComponentTypeSection(ComponentTypeSectionReader<'a>), + /// A component canonical section was received and the provided reader can be + /// used to parse the contents of the component canonical section. + ComponentCanonicalSection(ComponentCanonicalSectionReader<'a>), + /// A component start section was received. + ComponentStartSection { + /// The start function description. + start: ComponentStartFunction, + /// The range of bytes that specify the `start` field. + range: Range<usize>, + }, + /// A component import section was received and the provided reader can be + /// used to parse the contents of the component import section. + ComponentImportSection(ComponentImportSectionReader<'a>), + /// A component export section was received, and the provided reader can be + /// used to parse the contents of the component export section. + ComponentExportSection(ComponentExportSectionReader<'a>), + + /// A module or component custom section was received. + CustomSection(CustomSectionReader<'a>), + + /// An unknown section was found. + /// + /// This variant is returned for all unknown sections encountered. This + /// likely wants to be interpreted as an error by consumers of the parser, + /// but this can also be used to parse sections currently unsupported by + /// the parser. + UnknownSection { + /// The 8-bit identifier for this section. + id: u8, + /// The contents of this section. + contents: &'a [u8], + /// The range of bytes, relative to the start of the original data + /// stream, that the contents of this section reside in. + range: Range<usize>, + }, + + /// The end of the WebAssembly module or component was reached. + /// + /// The value is the offset in the input byte stream where the end + /// was reached. + End(usize), +} + +const CUSTOM_SECTION: u8 = 0; +const TYPE_SECTION: u8 = 1; +const IMPORT_SECTION: u8 = 2; +const FUNCTION_SECTION: u8 = 3; +const TABLE_SECTION: u8 = 4; +const MEMORY_SECTION: u8 = 5; +const GLOBAL_SECTION: u8 = 6; +const EXPORT_SECTION: u8 = 7; +const START_SECTION: u8 = 8; +const ELEMENT_SECTION: u8 = 9; +const CODE_SECTION: u8 = 10; +const DATA_SECTION: u8 = 11; +const DATA_COUNT_SECTION: u8 = 12; +const TAG_SECTION: u8 = 13; + +const COMPONENT_MODULE_SECTION: u8 = 1; +const COMPONENT_CORE_INSTANCE_SECTION: u8 = 2; +const COMPONENT_CORE_TYPE_SECTION: u8 = 3; +const COMPONENT_SECTION: u8 = 4; +const COMPONENT_INSTANCE_SECTION: u8 = 5; +const COMPONENT_ALIAS_SECTION: u8 = 6; +const COMPONENT_TYPE_SECTION: u8 = 7; +const COMPONENT_CANONICAL_SECTION: u8 = 8; +const COMPONENT_START_SECTION: u8 = 9; +const COMPONENT_IMPORT_SECTION: u8 = 10; +const COMPONENT_EXPORT_SECTION: u8 = 11; + +impl Parser { + /// Creates a new parser. + /// + /// Reports errors and ranges relative to `offset` provided, where `offset` + /// is some logical offset within the input stream that we're parsing. + pub fn new(offset: u64) -> Parser { + Parser { + state: State::Header, + offset, + max_size: u64::MAX, + // Assume the encoding is a module until we know otherwise + encoding: Encoding::Module, + } + } + + /// Attempts to parse a chunk of data. + /// + /// This method will attempt to parse the next incremental portion of a + /// WebAssembly binary. Data available for the module or component is + /// provided as `data`, and the data can be incomplete if more data has yet + /// to arrive. The `eof` flag indicates whether more data will ever be received. + /// + /// There are two ways parsing can succeed with this method: + /// + /// * `Chunk::NeedMoreData` - this indicates that there is not enough bytes + /// in `data` to parse a payload. The caller needs to wait for more data to + /// be available in this situation before calling this method again. It is + /// guaranteed that this is only returned if `eof` is `false`. + /// + /// * `Chunk::Parsed` - this indicates that a chunk of the input was + /// successfully parsed. The payload is available in this variant of what + /// was parsed, and this also indicates how many bytes of `data` was + /// consumed. It's expected that the caller will not provide these bytes + /// back to the [`Parser`] again. + /// + /// Note that all `Chunk` return values are connected, with a lifetime, to + /// the input buffer. Each parsed chunk borrows the input buffer and is a + /// view into it for successfully parsed chunks. + /// + /// It is expected that you'll call this method until `Payload::End` is + /// reached, at which point you're guaranteed that the parse has completed. + /// Note that complete parsing, for the top-level module or component, + /// implies that `data` is empty and `eof` is `true`. + /// + /// # Errors + /// + /// Parse errors are returned as an `Err`. Errors can happen when the + /// structure of the data is unexpected or if sections are too large for + /// example. Note that errors are not returned for malformed *contents* of + /// sections here. Sections are generally not individually parsed and each + /// returned [`Payload`] needs to be iterated over further to detect all + /// errors. + /// + /// # Examples + /// + /// An example of reading a wasm file from a stream (`std::io::Read`) and + /// incrementally parsing it. + /// + /// ``` + /// use std::io::Read; + /// use anyhow::Result; + /// use wasmparser::{Parser, Chunk, Payload::*}; + /// + /// fn parse(mut reader: impl Read) -> Result<()> { + /// let mut buf = Vec::new(); + /// let mut parser = Parser::new(0); + /// let mut eof = false; + /// let mut stack = Vec::new(); + /// + /// loop { + /// let (payload, consumed) = match parser.parse(&buf, eof)? { + /// Chunk::NeedMoreData(hint) => { + /// assert!(!eof); // otherwise an error would be returned + /// + /// // Use the hint to preallocate more space, then read + /// // some more data into our buffer. + /// // + /// // Note that the buffer management here is not ideal, + /// // but it's compact enough to fit in an example! + /// let len = buf.len(); + /// buf.extend((0..hint).map(|_| 0u8)); + /// let n = reader.read(&mut buf[len..])?; + /// buf.truncate(len + n); + /// eof = n == 0; + /// continue; + /// } + /// + /// Chunk::Parsed { consumed, payload } => (payload, consumed), + /// }; + /// + /// match payload { + /// // Sections for WebAssembly modules + /// Version { .. } => { /* ... */ } + /// TypeSection(_) => { /* ... */ } + /// ImportSection(_) => { /* ... */ } + /// FunctionSection(_) => { /* ... */ } + /// TableSection(_) => { /* ... */ } + /// MemorySection(_) => { /* ... */ } + /// TagSection(_) => { /* ... */ } + /// GlobalSection(_) => { /* ... */ } + /// ExportSection(_) => { /* ... */ } + /// StartSection { .. } => { /* ... */ } + /// ElementSection(_) => { /* ... */ } + /// DataCountSection { .. } => { /* ... */ } + /// DataSection(_) => { /* ... */ } + /// + /// // Here we know how many functions we'll be receiving as + /// // `CodeSectionEntry`, so we can prepare for that, and + /// // afterwards we can parse and handle each function + /// // individually. + /// CodeSectionStart { .. } => { /* ... */ } + /// CodeSectionEntry(body) => { + /// // here we can iterate over `body` to parse the function + /// // and its locals + /// } + /// + /// // Sections for WebAssembly components + /// ModuleSection { .. } => { /* ... */ } + /// InstanceSection(_) => { /* ... */ } + /// CoreTypeSection(_) => { /* ... */ } + /// ComponentSection { .. } => { /* ... */ } + /// ComponentInstanceSection(_) => { /* ... */ } + /// ComponentAliasSection(_) => { /* ... */ } + /// ComponentTypeSection(_) => { /* ... */ } + /// ComponentCanonicalSection(_) => { /* ... */ } + /// ComponentStartSection { .. } => { /* ... */ } + /// ComponentImportSection(_) => { /* ... */ } + /// ComponentExportSection(_) => { /* ... */ } + /// + /// CustomSection(_) => { /* ... */ } + /// + /// // most likely you'd return an error here + /// UnknownSection { id, .. } => { /* ... */ } + /// + /// // Once we've reached the end of a parser we either resume + /// // at the parent parser or we break out of the loop because + /// // we're done. + /// End(_) => { + /// if let Some(parent_parser) = stack.pop() { + /// parser = parent_parser; + /// } else { + /// break; + /// } + /// } + /// } + /// + /// // once we're done processing the payload we can forget the + /// // original. + /// buf.drain(..consumed); + /// } + /// + /// Ok(()) + /// } + /// + /// # parse(&b"\0asm\x01\0\0\0"[..]).unwrap(); + /// ``` + pub fn parse<'a>(&mut self, data: &'a [u8], eof: bool) -> Result<Chunk<'a>> { + let (data, eof) = if usize_to_u64(data.len()) > self.max_size { + (&data[..(self.max_size as usize)], true) + } else { + (data, eof) + }; + // TODO: thread through `offset: u64` to `BinaryReader`, remove + // the cast here. + let mut reader = BinaryReader::new_with_offset(data, self.offset as usize); + match self.parse_reader(&mut reader, eof) { + Ok(payload) => { + // Be sure to update our offset with how far we got in the + // reader + self.offset += usize_to_u64(reader.position); + self.max_size -= usize_to_u64(reader.position); + Ok(Chunk::Parsed { + consumed: reader.position, + payload, + }) + } + Err(e) => { + // If we're at EOF then there's no way we can recover from any + // error, so continue to propagate it. + if eof { + return Err(e); + } + + // If our error doesn't look like it can be resolved with more + // data being pulled down, then propagate it, otherwise switch + // the error to "feed me please" + match e.inner.needed_hint { + Some(hint) => Ok(Chunk::NeedMoreData(usize_to_u64(hint))), + None => Err(e), + } + } + } + } + + fn parse_reader<'a>( + &mut self, + reader: &mut BinaryReader<'a>, + eof: bool, + ) -> Result<Payload<'a>> { + use Payload::*; + + match self.state { + State::Header => { + const KIND_MODULE: u16 = 0x00; + const KIND_COMPONENT: u16 = 0x01; + + let start = reader.original_position(); + let header_version = reader.read_header_version()?; + self.encoding = match (header_version >> 16) as u16 { + KIND_MODULE => Encoding::Module, + KIND_COMPONENT => Encoding::Component, + _ => bail!(start + 4, "unknown binary version: {header_version:#10x}"), + }; + let num = header_version as u16; + self.state = State::SectionStart; + Ok(Version { + num, + encoding: self.encoding, + range: start..reader.original_position(), + }) + } + State::SectionStart => { + // If we're at eof and there are no bytes in our buffer, then + // that means we reached the end of the data since it's + // just a bunch of sections concatenated after the header. + if eof && reader.bytes_remaining() == 0 { + return Ok(Payload::End(reader.original_position())); + } + + let id_pos = reader.position; + let id = reader.read_u8()?; + if id & 0x80 != 0 { + return Err(BinaryReaderError::new("malformed section id", id_pos)); + } + let len_pos = reader.original_position(); + let mut len = reader.read_var_u32()?; + + // Test to make sure that this section actually fits within + // `Parser::max_size`. This doesn't matter for top-level modules + // but it is required for nested modules/components to correctly ensure + // that all sections live entirely within their section of the + // file. + let section_overflow = self + .max_size + .checked_sub(usize_to_u64(reader.position)) + .and_then(|s| s.checked_sub(len.into())) + .is_none(); + if section_overflow { + return Err(BinaryReaderError::new("section too large", len_pos)); + } + + match (self.encoding, id) { + // Sections for both modules and components. + (_, 0) => section(reader, len, CustomSectionReader::new, CustomSection), + + // Module sections + (Encoding::Module, TYPE_SECTION) => { + section(reader, len, TypeSectionReader::new, TypeSection) + } + (Encoding::Module, IMPORT_SECTION) => { + section(reader, len, ImportSectionReader::new, ImportSection) + } + (Encoding::Module, FUNCTION_SECTION) => { + section(reader, len, FunctionSectionReader::new, FunctionSection) + } + (Encoding::Module, TABLE_SECTION) => { + section(reader, len, TableSectionReader::new, TableSection) + } + (Encoding::Module, MEMORY_SECTION) => { + section(reader, len, MemorySectionReader::new, MemorySection) + } + (Encoding::Module, GLOBAL_SECTION) => { + section(reader, len, GlobalSectionReader::new, GlobalSection) + } + (Encoding::Module, EXPORT_SECTION) => { + section(reader, len, ExportSectionReader::new, ExportSection) + } + (Encoding::Module, START_SECTION) => { + let (func, range) = single_item(reader, len, "start")?; + Ok(StartSection { func, range }) + } + (Encoding::Module, ELEMENT_SECTION) => { + section(reader, len, ElementSectionReader::new, ElementSection) + } + (Encoding::Module, CODE_SECTION) => { + let start = reader.original_position(); + let count = delimited(reader, &mut len, |r| r.read_var_u32())?; + let range = start..reader.original_position() + len as usize; + self.state = State::FunctionBody { + remaining: count, + len, + }; + Ok(CodeSectionStart { + count, + range, + size: len, + }) + } + (Encoding::Module, DATA_SECTION) => { + section(reader, len, DataSectionReader::new, DataSection) + } + (Encoding::Module, DATA_COUNT_SECTION) => { + let (count, range) = single_item(reader, len, "data count")?; + Ok(DataCountSection { count, range }) + } + (Encoding::Module, TAG_SECTION) => { + section(reader, len, TagSectionReader::new, TagSection) + } + + // Component sections + (Encoding::Component, COMPONENT_MODULE_SECTION) + | (Encoding::Component, COMPONENT_SECTION) => { + if len as usize > MAX_WASM_MODULE_SIZE { + bail!( + len_pos, + "{} section is too large", + if id == 1 { "module" } else { "component " } + ); + } + + let range = + reader.original_position()..reader.original_position() + len as usize; + self.max_size -= u64::from(len); + self.offset += u64::from(len); + let mut parser = Parser::new(usize_to_u64(reader.original_position())); + parser.max_size = len.into(); + + Ok(match id { + 1 => ModuleSection { parser, range }, + 4 => ComponentSection { parser, range }, + _ => unreachable!(), + }) + } + (Encoding::Component, COMPONENT_CORE_INSTANCE_SECTION) => { + section(reader, len, InstanceSectionReader::new, InstanceSection) + } + (Encoding::Component, COMPONENT_CORE_TYPE_SECTION) => { + section(reader, len, CoreTypeSectionReader::new, CoreTypeSection) + } + (Encoding::Component, COMPONENT_INSTANCE_SECTION) => section( + reader, + len, + ComponentInstanceSectionReader::new, + ComponentInstanceSection, + ), + (Encoding::Component, COMPONENT_ALIAS_SECTION) => { + section(reader, len, SectionLimited::new, ComponentAliasSection) + } + (Encoding::Component, COMPONENT_TYPE_SECTION) => section( + reader, + len, + ComponentTypeSectionReader::new, + ComponentTypeSection, + ), + (Encoding::Component, COMPONENT_CANONICAL_SECTION) => section( + reader, + len, + ComponentCanonicalSectionReader::new, + ComponentCanonicalSection, + ), + (Encoding::Component, COMPONENT_START_SECTION) => { + let (start, range) = single_item(reader, len, "component start")?; + Ok(ComponentStartSection { start, range }) + } + (Encoding::Component, COMPONENT_IMPORT_SECTION) => section( + reader, + len, + ComponentImportSectionReader::new, + ComponentImportSection, + ), + (Encoding::Component, COMPONENT_EXPORT_SECTION) => section( + reader, + len, + ComponentExportSectionReader::new, + ComponentExportSection, + ), + (_, id) => { + let offset = reader.original_position(); + let contents = reader.read_bytes(len as usize)?; + let range = offset..offset + len as usize; + Ok(UnknownSection { + id, + contents, + range, + }) + } + } + } + + // Once we hit 0 remaining incrementally parsed items, with 0 + // remaining bytes in each section, we're done and can switch back + // to parsing sections. + State::FunctionBody { + remaining: 0, + len: 0, + } => { + self.state = State::SectionStart; + self.parse_reader(reader, eof) + } + + // ... otherwise trailing bytes with no remaining entries in these + // sections indicates an error. + State::FunctionBody { remaining: 0, len } => { + debug_assert!(len > 0); + let offset = reader.original_position(); + Err(BinaryReaderError::new( + "trailing bytes at end of section", + offset, + )) + } + + // Functions are relatively easy to parse when we know there's at + // least one remaining and at least one byte available to read + // things. + // + // We use the remaining length try to read a u32 size of the + // function, and using that size we require the entire function be + // resident in memory. This means that we're reading whole chunks of + // functions at a time. + // + // Limiting via `Parser::max_size` (nested parsing) happens above in + // `fn parse`, and limiting by our section size happens via + // `delimited`. Actual parsing of the function body is delegated to + // the caller to iterate over the `FunctionBody` structure. + State::FunctionBody { remaining, mut len } => { + let body = delimited(reader, &mut len, |r| { + let size = r.read_var_u32()?; + let offset = r.original_position(); + Ok(FunctionBody::new(offset, r.read_bytes(size as usize)?)) + })?; + self.state = State::FunctionBody { + remaining: remaining - 1, + len, + }; + Ok(CodeSectionEntry(body)) + } + } + } + + /// Convenience function that can be used to parse a module or component + /// that is entirely resident in memory. + /// + /// This function will parse the `data` provided as a WebAssembly module + /// or component. + /// + /// Note that when this function yields sections that provide parsers, + /// no further action is required for those sections as payloads from + /// those parsers will be automatically returned. + pub fn parse_all(self, mut data: &[u8]) -> impl Iterator<Item = Result<Payload>> { + let mut stack = Vec::new(); + let mut cur = self; + let mut done = false; + iter::from_fn(move || { + if done { + return None; + } + let payload = match cur.parse(data, true) { + // Propagate all errors + Err(e) => { + done = true; + return Some(Err(e)); + } + + // This isn't possible because `eof` is always true. + Ok(Chunk::NeedMoreData(_)) => unreachable!(), + + Ok(Chunk::Parsed { payload, consumed }) => { + data = &data[consumed..]; + payload + } + }; + + match &payload { + Payload::ModuleSection { parser, .. } + | Payload::ComponentSection { parser, .. } => { + stack.push(cur.clone()); + cur = parser.clone(); + } + Payload::End(_) => match stack.pop() { + Some(p) => cur = p, + None => done = true, + }, + + _ => {} + } + + Some(Ok(payload)) + }) + } + + /// Skip parsing the code section entirely. + /// + /// This function can be used to indicate, after receiving + /// `CodeSectionStart`, that the section will not be parsed. + /// + /// The caller will be responsible for skipping `size` bytes (found in the + /// `CodeSectionStart` payload). Bytes should only be fed into `parse` + /// after the `size` bytes have been skipped. + /// + /// # Panics + /// + /// This function will panic if the parser is not in a state where it's + /// parsing the code section. + /// + /// # Examples + /// + /// ``` + /// use wasmparser::{Result, Parser, Chunk, Payload::*}; + /// use std::ops::Range; + /// + /// fn objdump_headers(mut wasm: &[u8]) -> Result<()> { + /// let mut parser = Parser::new(0); + /// loop { + /// let payload = match parser.parse(wasm, true)? { + /// Chunk::Parsed { consumed, payload } => { + /// wasm = &wasm[consumed..]; + /// payload + /// } + /// // this state isn't possible with `eof = true` + /// Chunk::NeedMoreData(_) => unreachable!(), + /// }; + /// match payload { + /// TypeSection(s) => print_range("type section", &s.range()), + /// ImportSection(s) => print_range("import section", &s.range()), + /// // .. other sections + /// + /// // Print the range of the code section we see, but don't + /// // actually iterate over each individual function. + /// CodeSectionStart { range, size, .. } => { + /// print_range("code section", &range); + /// parser.skip_section(); + /// wasm = &wasm[size as usize..]; + /// } + /// End(_) => break, + /// _ => {} + /// } + /// } + /// Ok(()) + /// } + /// + /// fn print_range(section: &str, range: &Range<usize>) { + /// println!("{:>40}: {:#010x} - {:#010x}", section, range.start, range.end); + /// } + /// ``` + pub fn skip_section(&mut self) { + let skip = match self.state { + State::FunctionBody { remaining: _, len } => len, + _ => panic!("wrong state to call `skip_section`"), + }; + self.offset += u64::from(skip); + self.max_size -= u64::from(skip); + self.state = State::SectionStart; + } +} + +fn usize_to_u64(a: usize) -> u64 { + a.try_into().unwrap() +} + +/// Parses an entire section resident in memory into a `Payload`. +/// +/// Requires that `len` bytes are resident in `reader` and uses `ctor`/`variant` +/// to construct the section to return. +fn section<'a, T>( + reader: &mut BinaryReader<'a>, + len: u32, + ctor: fn(&'a [u8], usize) -> Result<T>, + variant: fn(T) -> Payload<'a>, +) -> Result<Payload<'a>> { + let offset = reader.original_position(); + let payload = reader.read_bytes(len as usize)?; + // clear the hint for "need this many more bytes" here because we already + // read all the bytes, so it's not possible to read more bytes if this + // fails. + let reader = ctor(payload, offset).map_err(clear_hint)?; + Ok(variant(reader)) +} + +/// Reads a section that is represented by a single uleb-encoded `u32`. +fn single_item<'a, T>( + reader: &mut BinaryReader<'a>, + len: u32, + desc: &str, +) -> Result<(T, Range<usize>)> +where + T: FromReader<'a>, +{ + let range = reader.original_position()..reader.original_position() + len as usize; + let mut content = BinaryReader::new_with_offset(reader.read_bytes(len as usize)?, range.start); + // We can't recover from "unexpected eof" here because our entire section is + // already resident in memory, so clear the hint for how many more bytes are + // expected. + let ret = content.read().map_err(clear_hint)?; + if !content.eof() { + bail!( + content.original_position(), + "unexpected content in the {desc} section", + ); + } + Ok((ret, range)) +} + +/// Attempts to parse using `f`. +/// +/// This will update `*len` with the number of bytes consumed, and it will cause +/// a failure to be returned instead of the number of bytes consumed exceeds +/// what `*len` currently is. +fn delimited<'a, T>( + reader: &mut BinaryReader<'a>, + len: &mut u32, + f: impl FnOnce(&mut BinaryReader<'a>) -> Result<T>, +) -> Result<T> { + let start = reader.position; + let ret = f(reader)?; + *len = match (reader.position - start) + .try_into() + .ok() + .and_then(|i| len.checked_sub(i)) + { + Some(i) => i, + None => return Err(BinaryReaderError::new("unexpected end-of-file", start)), + }; + Ok(ret) +} + +impl Default for Parser { + fn default() -> Parser { + Parser::new(0) + } +} + +impl Payload<'_> { + /// If this `Payload` represents a section in the original wasm module then + /// the section's id and range within the original wasm binary are returned. + /// + /// Not all payloads refer to entire sections, such as the `Version` and + /// `CodeSectionEntry` variants. These variants will return `None` from this + /// function. + /// + /// Otherwise this function will return `Some` where the first element is + /// the byte identifier for the section and the second element is the range + /// of the contents of the section within the original wasm binary. + /// + /// The purpose of this method is to enable tools to easily iterate over + /// entire sections if necessary and handle sections uniformly, for example + /// dropping custom sections while preserving all other sections. + pub fn as_section(&self) -> Option<(u8, Range<usize>)> { + use Payload::*; + + match self { + Version { .. } => None, + TypeSection(s) => Some((TYPE_SECTION, s.range())), + ImportSection(s) => Some((IMPORT_SECTION, s.range())), + FunctionSection(s) => Some((FUNCTION_SECTION, s.range())), + TableSection(s) => Some((TABLE_SECTION, s.range())), + MemorySection(s) => Some((MEMORY_SECTION, s.range())), + TagSection(s) => Some((TAG_SECTION, s.range())), + GlobalSection(s) => Some((GLOBAL_SECTION, s.range())), + ExportSection(s) => Some((EXPORT_SECTION, s.range())), + ElementSection(s) => Some((ELEMENT_SECTION, s.range())), + DataSection(s) => Some((DATA_SECTION, s.range())), + StartSection { range, .. } => Some((START_SECTION, range.clone())), + DataCountSection { range, .. } => Some((DATA_COUNT_SECTION, range.clone())), + CodeSectionStart { range, .. } => Some((CODE_SECTION, range.clone())), + CodeSectionEntry(_) => None, + + ModuleSection { range, .. } => Some((COMPONENT_MODULE_SECTION, range.clone())), + InstanceSection(s) => Some((COMPONENT_CORE_INSTANCE_SECTION, s.range())), + CoreTypeSection(s) => Some((COMPONENT_CORE_TYPE_SECTION, s.range())), + ComponentSection { range, .. } => Some((COMPONENT_SECTION, range.clone())), + ComponentInstanceSection(s) => Some((COMPONENT_INSTANCE_SECTION, s.range())), + ComponentAliasSection(s) => Some((COMPONENT_ALIAS_SECTION, s.range())), + ComponentTypeSection(s) => Some((COMPONENT_TYPE_SECTION, s.range())), + ComponentCanonicalSection(s) => Some((COMPONENT_CANONICAL_SECTION, s.range())), + ComponentStartSection { range, .. } => Some((COMPONENT_START_SECTION, range.clone())), + ComponentImportSection(s) => Some((COMPONENT_IMPORT_SECTION, s.range())), + ComponentExportSection(s) => Some((COMPONENT_EXPORT_SECTION, s.range())), + + CustomSection(c) => Some((CUSTOM_SECTION, c.range())), + + UnknownSection { id, range, .. } => Some((*id, range.clone())), + + End(_) => None, + } + } +} + +impl fmt::Debug for Payload<'_> { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + use Payload::*; + match self { + Version { + num, + encoding, + range, + } => f + .debug_struct("Version") + .field("num", num) + .field("encoding", encoding) + .field("range", range) + .finish(), + + // Module sections + TypeSection(_) => f.debug_tuple("TypeSection").field(&"...").finish(), + ImportSection(_) => f.debug_tuple("ImportSection").field(&"...").finish(), + FunctionSection(_) => f.debug_tuple("FunctionSection").field(&"...").finish(), + TableSection(_) => f.debug_tuple("TableSection").field(&"...").finish(), + MemorySection(_) => f.debug_tuple("MemorySection").field(&"...").finish(), + TagSection(_) => f.debug_tuple("TagSection").field(&"...").finish(), + GlobalSection(_) => f.debug_tuple("GlobalSection").field(&"...").finish(), + ExportSection(_) => f.debug_tuple("ExportSection").field(&"...").finish(), + ElementSection(_) => f.debug_tuple("ElementSection").field(&"...").finish(), + DataSection(_) => f.debug_tuple("DataSection").field(&"...").finish(), + StartSection { func, range } => f + .debug_struct("StartSection") + .field("func", func) + .field("range", range) + .finish(), + DataCountSection { count, range } => f + .debug_struct("DataCountSection") + .field("count", count) + .field("range", range) + .finish(), + CodeSectionStart { count, range, size } => f + .debug_struct("CodeSectionStart") + .field("count", count) + .field("range", range) + .field("size", size) + .finish(), + CodeSectionEntry(_) => f.debug_tuple("CodeSectionEntry").field(&"...").finish(), + + // Component sections + ModuleSection { parser: _, range } => f + .debug_struct("ModuleSection") + .field("range", range) + .finish(), + InstanceSection(_) => f.debug_tuple("InstanceSection").field(&"...").finish(), + CoreTypeSection(_) => f.debug_tuple("CoreTypeSection").field(&"...").finish(), + ComponentSection { parser: _, range } => f + .debug_struct("ComponentSection") + .field("range", range) + .finish(), + ComponentInstanceSection(_) => f + .debug_tuple("ComponentInstanceSection") + .field(&"...") + .finish(), + ComponentAliasSection(_) => f + .debug_tuple("ComponentAliasSection") + .field(&"...") + .finish(), + ComponentTypeSection(_) => f.debug_tuple("ComponentTypeSection").field(&"...").finish(), + ComponentCanonicalSection(_) => f + .debug_tuple("ComponentCanonicalSection") + .field(&"...") + .finish(), + ComponentStartSection { .. } => f + .debug_tuple("ComponentStartSection") + .field(&"...") + .finish(), + ComponentImportSection(_) => f + .debug_tuple("ComponentImportSection") + .field(&"...") + .finish(), + ComponentExportSection(_) => f + .debug_tuple("ComponentExportSection") + .field(&"...") + .finish(), + + CustomSection(c) => f.debug_tuple("CustomSection").field(c).finish(), + + UnknownSection { id, range, .. } => f + .debug_struct("UnknownSection") + .field("id", id) + .field("range", range) + .finish(), + + End(offset) => f.debug_tuple("End").field(offset).finish(), + } + } +} + +fn clear_hint(mut err: BinaryReaderError) -> BinaryReaderError { + err.inner.needed_hint = None; + err +} + +#[cfg(test)] +mod tests { + use super::*; + + macro_rules! assert_matches { + ($a:expr, $b:pat $(,)?) => { + match $a { + $b => {} + a => panic!("`{:?}` doesn't match `{}`", a, stringify!($b)), + } + }; + } + + #[test] + fn header() { + assert!(Parser::default().parse(&[], true).is_err()); + assert_matches!( + Parser::default().parse(&[], false), + Ok(Chunk::NeedMoreData(4)), + ); + assert_matches!( + Parser::default().parse(b"\0", false), + Ok(Chunk::NeedMoreData(3)), + ); + assert_matches!( + Parser::default().parse(b"\0asm", false), + Ok(Chunk::NeedMoreData(4)), + ); + assert_matches!( + Parser::default().parse(b"\0asm\x01\0\0\0", false), + Ok(Chunk::Parsed { + consumed: 8, + payload: Payload::Version { num: 1, .. }, + }), + ); + } + + #[test] + fn header_iter() { + for _ in Parser::default().parse_all(&[]) {} + for _ in Parser::default().parse_all(b"\0") {} + for _ in Parser::default().parse_all(b"\0asm") {} + for _ in Parser::default().parse_all(b"\0asm\x01\x01\x01\x01") {} + } + + fn parser_after_header() -> Parser { + let mut p = Parser::default(); + assert_matches!( + p.parse(b"\0asm\x01\0\0\0", false), + Ok(Chunk::Parsed { + consumed: 8, + payload: Payload::Version { + num: WASM_MODULE_VERSION, + encoding: Encoding::Module, + .. + }, + }), + ); + p + } + + fn parser_after_component_header() -> Parser { + let mut p = Parser::default(); + assert_matches!( + p.parse(b"\0asm\x0c\0\x01\0", false), + Ok(Chunk::Parsed { + consumed: 8, + payload: Payload::Version { + num: WASM_COMPONENT_VERSION, + encoding: Encoding::Component, + .. + }, + }), + ); + p + } + + #[test] + fn start_section() { + assert_matches!( + parser_after_header().parse(&[], false), + Ok(Chunk::NeedMoreData(1)), + ); + assert!(parser_after_header().parse(&[8], true).is_err()); + assert!(parser_after_header().parse(&[8, 1], true).is_err()); + assert!(parser_after_header().parse(&[8, 2], true).is_err()); + assert_matches!( + parser_after_header().parse(&[8], false), + Ok(Chunk::NeedMoreData(1)), + ); + assert_matches!( + parser_after_header().parse(&[8, 1], false), + Ok(Chunk::NeedMoreData(1)), + ); + assert_matches!( + parser_after_header().parse(&[8, 2], false), + Ok(Chunk::NeedMoreData(2)), + ); + assert_matches!( + parser_after_header().parse(&[8, 1, 1], false), + Ok(Chunk::Parsed { + consumed: 3, + payload: Payload::StartSection { func: 1, .. }, + }), + ); + assert!(parser_after_header().parse(&[8, 2, 1, 1], false).is_err()); + assert!(parser_after_header().parse(&[8, 0], false).is_err()); + } + + #[test] + fn end_works() { + assert_matches!( + parser_after_header().parse(&[], true), + Ok(Chunk::Parsed { + consumed: 0, + payload: Payload::End(8), + }), + ); + } + + #[test] + fn type_section() { + assert!(parser_after_header().parse(&[1], true).is_err()); + assert!(parser_after_header().parse(&[1, 0], false).is_err()); + assert!(parser_after_header().parse(&[8, 2], true).is_err()); + assert_matches!( + parser_after_header().parse(&[1], false), + Ok(Chunk::NeedMoreData(1)), + ); + assert_matches!( + parser_after_header().parse(&[1, 1], false), + Ok(Chunk::NeedMoreData(1)), + ); + assert_matches!( + parser_after_header().parse(&[1, 1, 1], false), + Ok(Chunk::Parsed { + consumed: 3, + payload: Payload::TypeSection(_), + }), + ); + assert_matches!( + parser_after_header().parse(&[1, 1, 1, 2, 3, 4], false), + Ok(Chunk::Parsed { + consumed: 3, + payload: Payload::TypeSection(_), + }), + ); + } + + #[test] + fn custom_section() { + assert!(parser_after_header().parse(&[0], true).is_err()); + assert!(parser_after_header().parse(&[0, 0], false).is_err()); + assert!(parser_after_header().parse(&[0, 1, 1], false).is_err()); + assert_matches!( + parser_after_header().parse(&[0, 2, 1], false), + Ok(Chunk::NeedMoreData(1)), + ); + assert_matches!( + parser_after_header().parse(&[0, 1, 0], false), + Ok(Chunk::Parsed { + consumed: 3, + payload: Payload::CustomSection(CustomSectionReader { + name: "", + data_offset: 11, + data: b"", + range: Range { start: 10, end: 11 }, + }), + }), + ); + assert_matches!( + parser_after_header().parse(&[0, 2, 1, b'a'], false), + Ok(Chunk::Parsed { + consumed: 4, + payload: Payload::CustomSection(CustomSectionReader { + name: "a", + data_offset: 12, + data: b"", + range: Range { start: 10, end: 12 }, + }), + }), + ); + assert_matches!( + parser_after_header().parse(&[0, 2, 0, b'a'], false), + Ok(Chunk::Parsed { + consumed: 4, + payload: Payload::CustomSection(CustomSectionReader { + name: "", + data_offset: 11, + data: b"a", + range: Range { start: 10, end: 12 }, + }), + }), + ); + } + + #[test] + fn function_section() { + assert!(parser_after_header().parse(&[10], true).is_err()); + assert!(parser_after_header().parse(&[10, 0], true).is_err()); + assert!(parser_after_header().parse(&[10, 1], true).is_err()); + assert_matches!( + parser_after_header().parse(&[10], false), + Ok(Chunk::NeedMoreData(1)) + ); + assert_matches!( + parser_after_header().parse(&[10, 1], false), + Ok(Chunk::NeedMoreData(1)) + ); + let mut p = parser_after_header(); + assert_matches!( + p.parse(&[10, 1, 0], false), + Ok(Chunk::Parsed { + consumed: 3, + payload: Payload::CodeSectionStart { count: 0, .. }, + }), + ); + assert_matches!( + p.parse(&[], true), + Ok(Chunk::Parsed { + consumed: 0, + payload: Payload::End(11), + }), + ); + let mut p = parser_after_header(); + assert_matches!( + p.parse(&[10, 2, 1, 0], false), + Ok(Chunk::Parsed { + consumed: 3, + payload: Payload::CodeSectionStart { count: 1, .. }, + }), + ); + assert_matches!( + p.parse(&[0], false), + Ok(Chunk::Parsed { + consumed: 1, + payload: Payload::CodeSectionEntry(_), + }), + ); + assert_matches!( + p.parse(&[], true), + Ok(Chunk::Parsed { + consumed: 0, + payload: Payload::End(12), + }), + ); + + // 1 byte section with 1 function can't read the function body because + // the section is too small + let mut p = parser_after_header(); + assert_matches!( + p.parse(&[10, 1, 1], false), + Ok(Chunk::Parsed { + consumed: 3, + payload: Payload::CodeSectionStart { count: 1, .. }, + }), + ); + assert_eq!( + p.parse(&[0], false).unwrap_err().message(), + "unexpected end-of-file" + ); + + // section with 2 functions but section is cut off + let mut p = parser_after_header(); + assert_matches!( + p.parse(&[10, 2, 2], false), + Ok(Chunk::Parsed { + consumed: 3, + payload: Payload::CodeSectionStart { count: 2, .. }, + }), + ); + assert_matches!( + p.parse(&[0], false), + Ok(Chunk::Parsed { + consumed: 1, + payload: Payload::CodeSectionEntry(_), + }), + ); + assert_matches!(p.parse(&[], false), Ok(Chunk::NeedMoreData(1))); + assert_eq!( + p.parse(&[0], false).unwrap_err().message(), + "unexpected end-of-file", + ); + + // trailing data is bad + let mut p = parser_after_header(); + assert_matches!( + p.parse(&[10, 3, 1], false), + Ok(Chunk::Parsed { + consumed: 3, + payload: Payload::CodeSectionStart { count: 1, .. }, + }), + ); + assert_matches!( + p.parse(&[0], false), + Ok(Chunk::Parsed { + consumed: 1, + payload: Payload::CodeSectionEntry(_), + }), + ); + assert_eq!( + p.parse(&[0], false).unwrap_err().message(), + "trailing bytes at end of section", + ); + } + + #[test] + fn single_module() { + let mut p = parser_after_component_header(); + assert_matches!(p.parse(&[4], false), Ok(Chunk::NeedMoreData(1))); + + // A module that's 8 bytes in length + let mut sub = match p.parse(&[1, 8], false) { + Ok(Chunk::Parsed { + consumed: 2, + payload: Payload::ModuleSection { parser, .. }, + }) => parser, + other => panic!("bad parse {:?}", other), + }; + + // Parse the header of the submodule with the sub-parser. + assert_matches!(sub.parse(&[], false), Ok(Chunk::NeedMoreData(4))); + assert_matches!(sub.parse(b"\0asm", false), Ok(Chunk::NeedMoreData(4))); + assert_matches!( + sub.parse(b"\0asm\x01\0\0\0", false), + Ok(Chunk::Parsed { + consumed: 8, + payload: Payload::Version { + num: 1, + encoding: Encoding::Module, + .. + }, + }), + ); + + // The sub-parser should be byte-limited so the next byte shouldn't get + // consumed, it's intended for the parent parser. + assert_matches!( + sub.parse(&[10], false), + Ok(Chunk::Parsed { + consumed: 0, + payload: Payload::End(18), + }), + ); + + // The parent parser should now be back to resuming, and we simulate it + // being done with bytes to ensure that it's safely at the end, + // completing the module code section. + assert_matches!(p.parse(&[], false), Ok(Chunk::NeedMoreData(1))); + assert_matches!( + p.parse(&[], true), + Ok(Chunk::Parsed { + consumed: 0, + payload: Payload::End(18), + }), + ); + } + + #[test] + fn nested_section_too_big() { + let mut p = parser_after_component_header(); + + // A module that's 10 bytes in length + let mut sub = match p.parse(&[1, 10], false) { + Ok(Chunk::Parsed { + consumed: 2, + payload: Payload::ModuleSection { parser, .. }, + }) => parser, + other => panic!("bad parse {:?}", other), + }; + + // use 8 bytes to parse the header, leaving 2 remaining bytes in our + // module. + assert_matches!( + sub.parse(b"\0asm\x01\0\0\0", false), + Ok(Chunk::Parsed { + consumed: 8, + payload: Payload::Version { num: 1, .. }, + }), + ); + + // We can't parse a section which declares its bigger than the outer + // module. This is a custom section, one byte big, with one content byte. The + // content byte, however, lives outside of the parent's module code + // section. + assert_eq!( + sub.parse(&[0, 1, 0], false).unwrap_err().message(), + "section too large", + ); + } +} diff --git a/third_party/rust/wasmparser/src/readers.rs b/third_party/rust/wasmparser/src/readers.rs new file mode 100644 index 0000000000..e2b25da7cf --- /dev/null +++ b/third_party/rust/wasmparser/src/readers.rs @@ -0,0 +1,316 @@ +/* Copyright 2018 Mozilla Foundation + * + * 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. + */ + +use crate::{BinaryReader, BinaryReaderError, Result}; +use std::fmt; +use std::marker; +use std::ops::Range; + +mod component; +mod core; + +pub use self::component::*; +pub use self::core::*; + +/// A trait implemented for items that can be decoded directly from a +/// `BinaryReader`, or that which can be parsed from the WebAssembly binary +/// format. +/// +/// Note that this is also accessible as a [`BinaryReader::read`] method. +pub trait FromReader<'a>: Sized { + /// Attempts to read `Self` from the provided binary reader, returning an + /// error if it is unable to do so. + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self>; +} + +impl<'a> FromReader<'a> for u32 { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + reader.read_var_u32() + } +} + +impl<'a> FromReader<'a> for &'a str { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + reader.read_string() + } +} + +impl<'a, T, U> FromReader<'a> for (T, U) +where + T: FromReader<'a>, + U: FromReader<'a>, +{ + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + Ok((reader.read()?, reader.read()?)) + } +} + +/// A generic structure for reading a section of a WebAssembly binary which has +/// a limited number of items within it. +/// +/// Many WebAssembly sections are a count of items followed by that many items. +/// This helper structure can be used to parse these sections and provides +/// an iteration-based API for reading the contents. +/// +/// Note that this always implements the [`Clone`] trait to represent the +/// ability to parse the section multiple times. +pub struct SectionLimited<'a, T> { + reader: BinaryReader<'a>, + count: u32, + _marker: marker::PhantomData<T>, +} + +impl<'a, T> SectionLimited<'a, T> { + /// Creates a new section reader from the provided contents. + /// + /// The `data` provided here is the data of the section itself that will be + /// parsed. The `offset` argument is the byte offset, in the original wasm + /// binary, that the section was found. The `offset` argument is used + /// for error reporting. + /// + /// # Errors + /// + /// Returns an error if a 32-bit count couldn't be read from the `data`. + pub fn new(data: &'a [u8], offset: usize) -> Result<Self> { + let mut reader = BinaryReader::new_with_offset(data, offset); + let count = reader.read_var_u32()?; + Ok(SectionLimited { + reader, + count, + _marker: marker::PhantomData, + }) + } + + /// Returns the count of total items within this section. + pub fn count(&self) -> u32 { + self.count + } + + /// Returns whether the original byte offset of this section. + pub fn original_position(&self) -> usize { + self.reader.original_position() + } + + /// Returns the range, as byte offsets, of this section within the original + /// wasm binary. + pub fn range(&self) -> Range<usize> { + self.reader.range() + } + + /// Returns an iterator which yields not only each item in this section but + /// additionally the offset of each item within the section. + pub fn into_iter_with_offsets(self) -> SectionLimitedIntoIterWithOffsets<'a, T> + where + T: FromReader<'a>, + { + SectionLimitedIntoIterWithOffsets { + iter: self.into_iter(), + } + } +} + +impl<T> Clone for SectionLimited<'_, T> { + fn clone(&self) -> Self { + SectionLimited { + reader: self.reader.clone(), + count: self.count, + _marker: self._marker, + } + } +} + +impl<T> fmt::Debug for SectionLimited<'_, T> { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.debug_struct("SectionLimited") + .field("count", &self.count) + .field("range", &self.range()) + .finish() + } +} + +impl<'a, T> IntoIterator for SectionLimited<'a, T> +where + T: FromReader<'a>, +{ + type Item = Result<T>; + type IntoIter = SectionLimitedIntoIter<'a, T>; + + fn into_iter(self) -> Self::IntoIter { + SectionLimitedIntoIter { + remaining: self.count, + section: self, + end: false, + } + } +} + +/// A consuming iterator of a [`SectionLimited`]. +/// +/// This is created via the [`IntoIterator`] `impl` for the [`SectionLimited`] +/// type. +pub struct SectionLimitedIntoIter<'a, T> { + section: SectionLimited<'a, T>, + remaining: u32, + end: bool, +} + +impl<T> SectionLimitedIntoIter<'_, T> { + /// Returns the current byte offset of the section within this iterator. + pub fn original_position(&self) -> usize { + self.section.reader.original_position() + } +} + +impl<'a, T> Iterator for SectionLimitedIntoIter<'a, T> +where + T: FromReader<'a>, +{ + type Item = Result<T>; + + fn next(&mut self) -> Option<Result<T>> { + if self.end { + return None; + } + if self.remaining == 0 { + self.end = true; + if self.section.reader.eof() { + return None; + } + return Some(Err(BinaryReaderError::new( + "section size mismatch: unexpected data at the end of the section", + self.section.reader.original_position(), + ))); + } + let result = self.section.reader.read(); + self.end = result.is_err(); + self.remaining -= 1; + Some(result) + } + + fn size_hint(&self) -> (usize, Option<usize>) { + let remaining = self.remaining as usize; + (remaining, Some(remaining)) + } +} + +impl<'a, T> ExactSizeIterator for SectionLimitedIntoIter<'a, T> where T: FromReader<'a> {} + +/// An iterator over a limited section iterator. +pub struct SectionLimitedIntoIterWithOffsets<'a, T> { + iter: SectionLimitedIntoIter<'a, T>, +} + +impl<'a, T> Iterator for SectionLimitedIntoIterWithOffsets<'a, T> +where + T: FromReader<'a>, +{ + type Item = Result<(usize, T)>; + + fn next(&mut self) -> Option<Self::Item> { + let pos = self.iter.section.reader.original_position(); + Some(self.iter.next()?.map(|item| (pos, item))) + } + + fn size_hint(&self) -> (usize, Option<usize>) { + self.iter.size_hint() + } +} + +impl<'a, T> ExactSizeIterator for SectionLimitedIntoIterWithOffsets<'a, T> where T: FromReader<'a> {} + +/// A trait implemented for subsections of another outer section. +/// +/// This is currently only used for subsections within custom sections, such as +/// the `name` section of core wasm. +/// +/// This is used in conjunction with [`Subsections`]. +pub trait Subsection<'a>: Sized { + /// Converts the section identifier provided with the section contents into + /// a typed section + fn from_reader(id: u8, reader: BinaryReader<'a>) -> Result<Self>; +} + +/// Iterator/reader over the contents of a section which is composed of +/// subsections. +/// +/// This reader is used for the core `name` section, for example. This type +/// primarily implements [`Iterator`] for advancing through the sections. +pub struct Subsections<'a, T> { + reader: BinaryReader<'a>, + _marker: marker::PhantomData<T>, +} + +impl<'a, T> Subsections<'a, T> { + /// Creates a new reader for the specified section contents starting at + /// `offset` within the original wasm file. + pub fn new(data: &'a [u8], offset: usize) -> Self { + Subsections { + reader: BinaryReader::new_with_offset(data, offset), + _marker: marker::PhantomData, + } + } + + /// Returns whether the original byte offset of this section. + pub fn original_position(&self) -> usize { + self.reader.original_position() + } + + /// Returns the range, as byte offsets, of this section within the original + /// wasm binary. + pub fn range(&self) -> Range<usize> { + self.reader.range() + } + + fn read(&mut self) -> Result<T> + where + T: Subsection<'a>, + { + let subsection_id = self.reader.read_u7()?; + let reader = self.reader.read_reader("unexpected end of section")?; + T::from_reader(subsection_id, reader) + } +} + +impl<T> Clone for Subsections<'_, T> { + fn clone(&self) -> Self { + Subsections { + reader: self.reader.clone(), + _marker: self._marker, + } + } +} + +impl<T> fmt::Debug for Subsections<'_, T> { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.debug_struct("Subsections") + .field("range", &self.range()) + .finish() + } +} + +impl<'a, T> Iterator for Subsections<'a, T> +where + T: Subsection<'a>, +{ + type Item = Result<T>; + + fn next(&mut self) -> Option<Result<T>> { + if self.reader.eof() { + None + } else { + Some(self.read()) + } + } +} diff --git a/third_party/rust/wasmparser/src/readers/component.rs b/third_party/rust/wasmparser/src/readers/component.rs new file mode 100644 index 0000000000..24b490d0c3 --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/component.rs @@ -0,0 +1,17 @@ +mod aliases; +mod canonicals; +mod exports; +mod imports; +mod instances; +mod names; +mod start; +mod types; + +pub use self::aliases::*; +pub use self::canonicals::*; +pub use self::exports::*; +pub use self::imports::*; +pub use self::instances::*; +pub use self::names::*; +pub use self::start::*; +pub use self::types::*; diff --git a/third_party/rust/wasmparser/src/readers/component/aliases.rs b/third_party/rust/wasmparser/src/readers/component/aliases.rs new file mode 100644 index 0000000000..fb71d579b4 --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/component/aliases.rs @@ -0,0 +1,119 @@ +use crate::{BinaryReader, ComponentExternalKind, ExternalKind, FromReader, Result}; + +/// Represents the kind of an outer alias in a WebAssembly component. +#[derive(Clone, Copy, Debug, Eq, PartialEq)] +pub enum ComponentOuterAliasKind { + /// The alias is to a core module. + CoreModule, + /// The alias is to a core type. + CoreType, + /// The alias is to a component type. + Type, + /// The alias is to a component. + Component, +} + +/// Represents an alias in a WebAssembly component. +#[derive(Debug, Clone)] +pub enum ComponentAlias<'a> { + /// The alias is to an export of a component instance. + InstanceExport { + /// The alias kind. + kind: ComponentExternalKind, + /// The instance index. + instance_index: u32, + /// The export name. + name: &'a str, + }, + /// The alias is to an export of a module instance. + CoreInstanceExport { + /// The alias kind. + kind: ExternalKind, + /// The instance index. + instance_index: u32, + /// The export name. + name: &'a str, + }, + /// The alias is to an outer item. + Outer { + /// The alias kind. + kind: ComponentOuterAliasKind, + /// The outward count, starting at zero for the current component. + count: u32, + /// The index of the item within the outer component. + index: u32, + }, +} + +/// Section reader for the component alias section +pub type ComponentAliasSectionReader<'a> = crate::SectionLimited<'a, ComponentAlias<'a>>; + +impl<'a> FromReader<'a> for ComponentAlias<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + // We don't know what type of alias it is yet, so just read the sort bytes + let offset = reader.original_position(); + let byte1 = reader.read_u8()?; + let byte2 = if byte1 == 0x00 { + Some(reader.read_u8()?) + } else { + None + }; + + Ok(match reader.read_u8()? { + 0x00 => ComponentAlias::InstanceExport { + kind: ComponentExternalKind::from_bytes(byte1, byte2, offset)?, + instance_index: reader.read_var_u32()?, + name: reader.read_string()?, + }, + 0x01 => ComponentAlias::CoreInstanceExport { + kind: BinaryReader::external_kind_from_byte( + byte2.ok_or_else(|| { + BinaryReader::invalid_leading_byte_error( + byte1, + "core instance export kind", + offset, + ) + })?, + offset, + )?, + instance_index: reader.read_var_u32()?, + name: reader.read_string()?, + }, + 0x02 => ComponentAlias::Outer { + kind: component_outer_alias_kind_from_bytes(byte1, byte2, offset)?, + count: reader.read_var_u32()?, + index: reader.read_var_u32()?, + }, + x => reader.invalid_leading_byte(x, "alias")?, + }) + } +} + +fn component_outer_alias_kind_from_bytes( + byte1: u8, + byte2: Option<u8>, + offset: usize, +) -> Result<ComponentOuterAliasKind> { + Ok(match byte1 { + 0x00 => match byte2.unwrap() { + 0x10 => ComponentOuterAliasKind::CoreType, + 0x11 => ComponentOuterAliasKind::CoreModule, + x => { + return Err(BinaryReader::invalid_leading_byte_error( + x, + "component outer alias kind", + offset + 1, + )) + } + }, + 0x03 => ComponentOuterAliasKind::Type, + 0x04 => ComponentOuterAliasKind::Component, + x => { + return Err(BinaryReader::invalid_leading_byte_error( + x, + "component outer alias kind", + offset, + )) + } + }) +} diff --git a/third_party/rust/wasmparser/src/readers/component/canonicals.rs b/third_party/rust/wasmparser/src/readers/component/canonicals.rs new file mode 100644 index 0000000000..e360d029c4 --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/component/canonicals.rs @@ -0,0 +1,95 @@ +use crate::limits::MAX_WASM_CANONICAL_OPTIONS; +use crate::{BinaryReader, FromReader, Result, SectionLimited}; + +/// Represents options for component functions. +#[derive(Debug, Clone, Copy, PartialEq, Eq)] +pub enum CanonicalOption { + /// The string types in the function signature are UTF-8 encoded. + UTF8, + /// The string types in the function signature are UTF-16 encoded. + UTF16, + /// The string types in the function signature are compact UTF-16 encoded. + CompactUTF16, + /// The memory to use if the lifting or lowering of a function requires memory access. + /// + /// The value is an index to a core memory. + Memory(u32), + /// The realloc function to use if the lifting or lowering of a function requires memory + /// allocation. + /// + /// The value is an index to a core function of type `(func (param i32 i32 i32 i32) (result i32))`. + Realloc(u32), + /// The post-return function to use if the lifting of a function requires + /// cleanup after the function returns. + PostReturn(u32), +} + +/// Represents a canonical function in a WebAssembly component. +#[derive(Debug, Clone)] +pub enum CanonicalFunction { + /// The function lifts a core WebAssembly function to the canonical ABI. + Lift { + /// The index of the core WebAssembly function to lift. + core_func_index: u32, + /// The index of the lifted function's type. + type_index: u32, + /// The canonical options for the function. + options: Box<[CanonicalOption]>, + }, + /// The function lowers a canonical ABI function to a core WebAssembly function. + Lower { + /// The index of the function to lower. + func_index: u32, + /// The canonical options for the function. + options: Box<[CanonicalOption]>, + }, +} + +/// A reader for the canonical section of a WebAssembly component. +pub type ComponentCanonicalSectionReader<'a> = SectionLimited<'a, CanonicalFunction>; + +impl<'a> FromReader<'a> for CanonicalFunction { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<CanonicalFunction> { + Ok(match reader.read_u8()? { + 0x00 => match reader.read_u8()? { + 0x00 => { + let core_func_index = reader.read_var_u32()?; + let options = reader + .read_iter(MAX_WASM_CANONICAL_OPTIONS, "canonical options")? + .collect::<Result<_>>()?; + let type_index = reader.read_var_u32()?; + CanonicalFunction::Lift { + core_func_index, + options, + type_index, + } + } + x => return reader.invalid_leading_byte(x, "canonical function lift"), + }, + 0x01 => match reader.read_u8()? { + 0x00 => CanonicalFunction::Lower { + func_index: reader.read_var_u32()?, + options: reader + .read_iter(MAX_WASM_CANONICAL_OPTIONS, "canonical options")? + .collect::<Result<_>>()?, + }, + x => return reader.invalid_leading_byte(x, "canonical function lower"), + }, + x => return reader.invalid_leading_byte(x, "canonical function"), + }) + } +} + +impl<'a> FromReader<'a> for CanonicalOption { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + Ok(match reader.read_u8()? { + 0x00 => CanonicalOption::UTF8, + 0x01 => CanonicalOption::UTF16, + 0x02 => CanonicalOption::CompactUTF16, + 0x03 => CanonicalOption::Memory(reader.read_var_u32()?), + 0x04 => CanonicalOption::Realloc(reader.read_var_u32()?), + 0x05 => CanonicalOption::PostReturn(reader.read_var_u32()?), + x => return reader.invalid_leading_byte(x, "canonical option"), + }) + } +} diff --git a/third_party/rust/wasmparser/src/readers/component/exports.rs b/third_party/rust/wasmparser/src/readers/component/exports.rs new file mode 100644 index 0000000000..8ce5f43a00 --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/component/exports.rs @@ -0,0 +1,105 @@ +use crate::{BinaryReader, ComponentTypeRef, FromReader, Result, SectionLimited}; + +/// Represents the kind of an external items of a WebAssembly component. +#[derive(Clone, Copy, Debug, Eq, PartialEq)] +pub enum ComponentExternalKind { + /// The external kind is a core module. + Module, + /// The external kind is a function. + Func, + /// The external kind is a value. + Value, + /// The external kind is a type. + Type, + /// The external kind is an instance. + Instance, + /// The external kind is a component. + Component, +} + +impl ComponentExternalKind { + pub(crate) fn from_bytes( + byte1: u8, + byte2: Option<u8>, + offset: usize, + ) -> Result<ComponentExternalKind> { + Ok(match byte1 { + 0x00 => match byte2.unwrap() { + 0x11 => ComponentExternalKind::Module, + x => { + return Err(BinaryReader::invalid_leading_byte_error( + x, + "component external kind", + offset + 1, + )) + } + }, + 0x01 => ComponentExternalKind::Func, + 0x02 => ComponentExternalKind::Value, + 0x03 => ComponentExternalKind::Type, + 0x04 => ComponentExternalKind::Component, + 0x05 => ComponentExternalKind::Instance, + x => { + return Err(BinaryReader::invalid_leading_byte_error( + x, + "component external kind", + offset, + )) + } + }) + } +} + +/// Represents an export in a WebAssembly component. +#[derive(Debug, Clone)] +pub struct ComponentExport<'a> { + /// The name of the exported item. + pub name: &'a str, + /// The optional URL of the exported item. + pub url: &'a str, + /// The kind of the export. + pub kind: ComponentExternalKind, + /// The index of the exported item. + pub index: u32, + /// An optionally specified type ascribed to this export. + pub ty: Option<ComponentTypeRef>, +} + +/// A reader for the export section of a WebAssembly component. +pub type ComponentExportSectionReader<'a> = SectionLimited<'a, ComponentExport<'a>>; + +impl<'a> FromReader<'a> for ComponentExport<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + Ok(ComponentExport { + name: reader.read()?, + url: reader.read()?, + kind: reader.read()?, + index: reader.read()?, + ty: match reader.read_u8()? { + 0x00 => None, + 0x01 => Some(reader.read()?), + other => { + return Err(BinaryReader::invalid_leading_byte_error( + other, + "optional component export type", + reader.original_position() - 1, + )) + } + }, + }) + } +} + +impl<'a> FromReader<'a> for ComponentExternalKind { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + let offset = reader.original_position(); + let byte1 = reader.read_u8()?; + let byte2 = if byte1 == 0x00 { + Some(reader.read_u8()?) + } else { + None + }; + + ComponentExternalKind::from_bytes(byte1, byte2, offset) + } +} diff --git a/third_party/rust/wasmparser/src/readers/component/imports.rs b/third_party/rust/wasmparser/src/readers/component/imports.rs new file mode 100644 index 0000000000..c1313c11e2 --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/component/imports.rs @@ -0,0 +1,109 @@ +use crate::{ + BinaryReader, ComponentExternalKind, ComponentValType, FromReader, Result, SectionLimited, +}; + +/// Represents the type bounds for imports and exports. +#[derive(Clone, Copy, Debug, Eq, PartialEq)] +pub enum TypeBounds { + /// The type is bounded by equality. + Eq, +} + +impl<'a> FromReader<'a> for TypeBounds { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + Ok(match reader.read_u8()? { + 0x00 => TypeBounds::Eq, + x => return reader.invalid_leading_byte(x, "type bound"), + }) + } +} + +/// Represents a reference to a component type. +#[derive(Clone, Copy, Debug, Eq, PartialEq)] +pub enum ComponentTypeRef { + /// The reference is to a core module type. + /// + /// The index is expected to be core type index to a core module type. + Module(u32), + /// The reference is to a function type. + /// + /// The index is expected to be a type index to a function type. + Func(u32), + /// The reference is to a value type. + Value(ComponentValType), + /// The reference is to a bounded type. + /// + /// The index is expected to be a type index. + Type(TypeBounds, u32), + /// The reference is to an instance type. + /// + /// The index is a type index to an instance type. + Instance(u32), + /// The reference is to a component type. + /// + /// The index is a type index to a component type. + Component(u32), +} + +impl ComponentTypeRef { + /// Returns the corresponding [`ComponentExternalKind`] for this reference. + pub fn kind(&self) -> ComponentExternalKind { + match self { + ComponentTypeRef::Module(_) => ComponentExternalKind::Module, + ComponentTypeRef::Func(_) => ComponentExternalKind::Func, + ComponentTypeRef::Value(_) => ComponentExternalKind::Value, + ComponentTypeRef::Type(..) => ComponentExternalKind::Type, + ComponentTypeRef::Instance(_) => ComponentExternalKind::Instance, + ComponentTypeRef::Component(_) => ComponentExternalKind::Component, + } + } +} + +impl<'a> FromReader<'a> for ComponentTypeRef { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + Ok(match reader.read()? { + ComponentExternalKind::Module => ComponentTypeRef::Module(reader.read()?), + ComponentExternalKind::Func => ComponentTypeRef::Func(reader.read()?), + ComponentExternalKind::Value => ComponentTypeRef::Value(reader.read()?), + ComponentExternalKind::Type => ComponentTypeRef::Type(reader.read()?, reader.read()?), + ComponentExternalKind::Instance => ComponentTypeRef::Instance(reader.read()?), + ComponentExternalKind::Component => ComponentTypeRef::Component(reader.read()?), + }) + } +} + +/// Represents an import in a WebAssembly component +#[derive(Debug, Copy, Clone)] +pub struct ComponentImport<'a> { + /// The name of the imported item. + pub name: &'a str, + /// The optional URL of the imported item. + pub url: &'a str, + /// The type reference for the import. + pub ty: ComponentTypeRef, +} + +impl<'a> FromReader<'a> for ComponentImport<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + Ok(ComponentImport { + name: reader.read()?, + url: reader.read()?, + ty: reader.read()?, + }) + } +} + +/// A reader for the import section of a WebAssembly component. +/// +/// # Examples +/// +/// ``` +/// use wasmparser::ComponentImportSectionReader; +/// let data: &[u8] = &[0x01, 0x01, 0x41, 0x00, 0x01, 0x66]; +/// let reader = ComponentImportSectionReader::new(data, 0).unwrap(); +/// for import in reader { +/// let import = import.expect("import"); +/// println!("Import: {:?}", import); +/// } +/// ``` +pub type ComponentImportSectionReader<'a> = SectionLimited<'a, ComponentImport<'a>>; diff --git a/third_party/rust/wasmparser/src/readers/component/instances.rs b/third_party/rust/wasmparser/src/readers/component/instances.rs new file mode 100644 index 0000000000..8166395edc --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/component/instances.rs @@ -0,0 +1,164 @@ +use crate::limits::{MAX_WASM_INSTANTIATION_ARGS, MAX_WASM_INSTANTIATION_EXPORTS}; +use crate::{ + BinaryReader, ComponentExport, ComponentExternalKind, Export, FromReader, Result, + SectionLimited, +}; + +/// Represents the kind of an instantiation argument for a core instance. +#[derive(Debug, Copy, Clone, PartialEq, Eq)] +pub enum InstantiationArgKind { + /// The instantiation argument is a core instance. + Instance, +} + +/// Represents an argument to instantiating a WebAssembly module. +#[derive(Debug, Clone)] +pub struct InstantiationArg<'a> { + /// The name of the module argument. + pub name: &'a str, + /// The kind of the module argument. + pub kind: InstantiationArgKind, + /// The index of the argument item. + pub index: u32, +} + +/// Represents an instance of a WebAssembly module. +#[derive(Debug, Clone)] +pub enum Instance<'a> { + /// The instance is from instantiating a WebAssembly module. + Instantiate { + /// The module index. + module_index: u32, + /// The module's instantiation arguments. + args: Box<[InstantiationArg<'a>]>, + }, + /// The instance is a from exporting local items. + FromExports(Box<[Export<'a>]>), +} + +/// A reader for the core instance section of a WebAssembly component. +/// +/// # Examples +/// +/// ``` +/// use wasmparser::InstanceSectionReader; +/// # let data: &[u8] = &[0x01, 0x00, 0x00, 0x01, 0x03, b'f', b'o', b'o', 0x12, 0x00]; +/// let mut reader = InstanceSectionReader::new(data, 0).unwrap(); +/// for inst in reader { +/// println!("Instance {:?}", inst.expect("instance")); +/// } +/// ``` +pub type InstanceSectionReader<'a> = SectionLimited<'a, Instance<'a>>; + +impl<'a> FromReader<'a> for Instance<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + Ok(match reader.read_u8()? { + 0x00 => Instance::Instantiate { + module_index: reader.read_var_u32()?, + args: reader + .read_iter(MAX_WASM_INSTANTIATION_ARGS, "core instantiation arguments")? + .collect::<Result<_>>()?, + }, + 0x01 => Instance::FromExports( + reader + .read_iter(MAX_WASM_INSTANTIATION_ARGS, "core instantiation arguments")? + .collect::<Result<_>>()?, + ), + x => return reader.invalid_leading_byte(x, "core instance"), + }) + } +} + +impl<'a> FromReader<'a> for InstantiationArg<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + Ok(InstantiationArg { + name: reader.read()?, + kind: reader.read()?, + index: reader.read()?, + }) + } +} + +impl<'a> FromReader<'a> for InstantiationArgKind { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + Ok(match reader.read_u8()? { + 0x12 => InstantiationArgKind::Instance, + x => return reader.invalid_leading_byte(x, "instantiation arg kind"), + }) + } +} + +/// Represents an argument to instantiating a WebAssembly component. +#[derive(Debug, Clone)] +pub struct ComponentInstantiationArg<'a> { + /// The name of the component argument. + pub name: &'a str, + /// The kind of the component argument. + pub kind: ComponentExternalKind, + /// The index of the argument item. + pub index: u32, +} + +/// Represents an instance in a WebAssembly component. +#[derive(Debug, Clone)] +pub enum ComponentInstance<'a> { + /// The instance is from instantiating a WebAssembly component. + Instantiate { + /// The component index. + component_index: u32, + /// The component's instantiation arguments. + args: Box<[ComponentInstantiationArg<'a>]>, + }, + /// The instance is a from exporting local items. + FromExports(Box<[ComponentExport<'a>]>), +} + +/// A reader for the component instance section of a WebAssembly component. +/// +/// # Examples +/// +/// ``` +/// use wasmparser::ComponentInstanceSectionReader; +/// # let data: &[u8] = &[0x01, 0x00, 0x00, 0x01, 0x03, b'f', b'o', b'o', 0x01, 0x00]; +/// let mut reader = ComponentInstanceSectionReader::new(data, 0).unwrap(); +/// for inst in reader { +/// println!("Instance {:?}", inst.expect("instance")); +/// } +/// ``` +pub type ComponentInstanceSectionReader<'a> = SectionLimited<'a, ComponentInstance<'a>>; + +impl<'a> FromReader<'a> for ComponentInstance<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + Ok(match reader.read_u8()? { + 0x00 => ComponentInstance::Instantiate { + component_index: reader.read_var_u32()?, + args: reader + .read_iter(MAX_WASM_INSTANTIATION_ARGS, "instantiation arguments")? + .collect::<Result<_>>()?, + }, + 0x01 => ComponentInstance::FromExports( + (0..reader.read_size(MAX_WASM_INSTANTIATION_EXPORTS, "instantiation exports")?) + .map(|_| { + Ok(ComponentExport { + name: reader.read()?, + url: "", + kind: reader.read()?, + index: reader.read()?, + ty: None, + }) + }) + .collect::<Result<_>>()?, + ), + x => return reader.invalid_leading_byte(x, "instance"), + }) + } +} +impl<'a> FromReader<'a> for ComponentInstantiationArg<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + Ok(ComponentInstantiationArg { + name: reader.read()?, + kind: reader.read()?, + index: reader.read()?, + }) + } +} diff --git a/third_party/rust/wasmparser/src/readers/component/names.rs b/third_party/rust/wasmparser/src/readers/component/names.rs new file mode 100644 index 0000000000..19de2752d0 --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/component/names.rs @@ -0,0 +1,102 @@ +use crate::{BinaryReader, BinaryReaderError, NameMap, Result, Subsection, Subsections}; +use std::ops::Range; + +/// Type used to iterate and parse the contents of the `component-name` custom +/// section in compnents, similar to the `name` section of core modules. +pub type ComponentNameSectionReader<'a> = Subsections<'a, ComponentName<'a>>; + +/// Represents a name read from the names custom section. +#[derive(Clone)] +#[allow(missing_docs)] +pub enum ComponentName<'a> { + Component { + name: &'a str, + name_range: Range<usize>, + }, + CoreFuncs(NameMap<'a>), + CoreGlobals(NameMap<'a>), + CoreMemories(NameMap<'a>), + CoreTables(NameMap<'a>), + CoreModules(NameMap<'a>), + CoreInstances(NameMap<'a>), + CoreTypes(NameMap<'a>), + Types(NameMap<'a>), + Instances(NameMap<'a>), + Components(NameMap<'a>), + Funcs(NameMap<'a>), + Values(NameMap<'a>), + + /// An unknown [name subsection](https://webassembly.github.io/spec/core/appendix/custom.html#subsections). + Unknown { + /// The identifier for this subsection. + ty: u8, + /// The contents of this subsection. + data: &'a [u8], + /// The range of bytes, relative to the start of the original data + /// stream, that the contents of this subsection reside in. + range: Range<usize>, + }, +} + +impl<'a> Subsection<'a> for ComponentName<'a> { + fn from_reader(id: u8, mut reader: BinaryReader<'a>) -> Result<Self> { + let data = reader.remaining_buffer(); + let offset = reader.original_position(); + Ok(match id { + 0 => { + let name = reader.read_string()?; + if !reader.eof() { + return Err(BinaryReaderError::new( + "trailing data at the end of a name", + reader.original_position(), + )); + } + ComponentName::Component { + name, + name_range: offset..offset + reader.position, + } + } + 1 => { + let ctor: fn(NameMap<'a>) -> ComponentName<'a> = match reader.read_u8()? { + 0x00 => match reader.read_u8()? { + 0x00 => ComponentName::CoreFuncs, + 0x01 => ComponentName::CoreTables, + 0x02 => ComponentName::CoreMemories, + 0x03 => ComponentName::CoreGlobals, + 0x10 => ComponentName::CoreTypes, + 0x11 => ComponentName::CoreModules, + 0x12 => ComponentName::CoreInstances, + _ => { + return Ok(ComponentName::Unknown { + ty: 1, + data, + range: offset..offset + data.len(), + }); + } + }, + 0x01 => ComponentName::Funcs, + 0x02 => ComponentName::Values, + 0x03 => ComponentName::Types, + 0x04 => ComponentName::Components, + 0x05 => ComponentName::Instances, + _ => { + return Ok(ComponentName::Unknown { + ty: 1, + data, + range: offset..offset + data.len(), + }); + } + }; + ctor(NameMap::new( + reader.remaining_buffer(), + reader.original_position(), + )?) + } + ty => ComponentName::Unknown { + ty, + data, + range: offset..offset + data.len(), + }, + }) + } +} diff --git a/third_party/rust/wasmparser/src/readers/component/start.rs b/third_party/rust/wasmparser/src/readers/component/start.rs new file mode 100644 index 0000000000..dc01fa4340 --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/component/start.rs @@ -0,0 +1,30 @@ +use crate::limits::{MAX_WASM_FUNCTION_RETURNS, MAX_WASM_START_ARGS}; +use crate::{BinaryReader, FromReader, Result}; + +/// Represents the start function in a WebAssembly component. +#[derive(Debug, Clone)] +pub struct ComponentStartFunction { + /// The index to the start function. + pub func_index: u32, + /// The start function arguments. + /// + /// The arguments are specified by value index. + pub arguments: Box<[u32]>, + /// The number of expected results for the start function. + pub results: u32, +} + +impl<'a> FromReader<'a> for ComponentStartFunction { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + let func_index = reader.read_var_u32()?; + let arguments = reader + .read_iter(MAX_WASM_START_ARGS, "start function arguments")? + .collect::<Result<_>>()?; + let results = reader.read_size(MAX_WASM_FUNCTION_RETURNS, "start function results")? as u32; + Ok(ComponentStartFunction { + func_index, + arguments, + results, + }) + } +} diff --git a/third_party/rust/wasmparser/src/readers/component/types.rs b/third_party/rust/wasmparser/src/readers/component/types.rs new file mode 100644 index 0000000000..b0e9687a4d --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/component/types.rs @@ -0,0 +1,508 @@ +use crate::limits::*; +use crate::{ + BinaryReader, ComponentAlias, ComponentImport, ComponentTypeRef, FromReader, FuncType, Import, + Result, SectionLimited, Type, TypeRef, +}; + +/// Represents the kind of an outer core alias in a WebAssembly component. +#[derive(Clone, Copy, Debug, Eq, PartialEq)] +pub enum OuterAliasKind { + /// The alias is to a core type. + Type, +} + +/// Represents a core type in a WebAssembly component. +#[derive(Debug, Clone)] +pub enum CoreType<'a> { + /// The type is for a core function. + Func(FuncType), + /// The type is for a core module. + Module(Box<[ModuleTypeDeclaration<'a>]>), +} + +impl<'a> FromReader<'a> for CoreType<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + Ok(match reader.read_u8()? { + 0x60 => CoreType::Func(reader.read()?), + 0x50 => CoreType::Module( + reader + .read_iter(MAX_WASM_MODULE_TYPE_DECLS, "module type declaration")? + .collect::<Result<_>>()?, + ), + x => return reader.invalid_leading_byte(x, "core type"), + }) + } +} + +/// Represents a module type declaration in a WebAssembly component. +#[derive(Debug, Clone)] +pub enum ModuleTypeDeclaration<'a> { + /// The module type definition is for a type. + Type(Type), + /// The module type definition is for an export. + Export { + /// The name of the exported item. + name: &'a str, + /// The type reference of the export. + ty: TypeRef, + }, + /// The module type declaration is for an outer alias. + OuterAlias { + /// The alias kind. + kind: OuterAliasKind, + /// The outward count, starting at zero for the current type. + count: u32, + /// The index of the item within the outer type. + index: u32, + }, + /// The module type definition is for an import. + Import(Import<'a>), +} + +impl<'a> FromReader<'a> for ModuleTypeDeclaration<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + Ok(match reader.read_u8()? { + 0x00 => ModuleTypeDeclaration::Import(reader.read()?), + 0x01 => ModuleTypeDeclaration::Type(reader.read()?), + 0x02 => { + let kind = match reader.read_u8()? { + 0x10 => OuterAliasKind::Type, + x => { + return reader.invalid_leading_byte(x, "outer alias kind"); + } + }; + match reader.read_u8()? { + 0x01 => ModuleTypeDeclaration::OuterAlias { + kind, + count: reader.read()?, + index: reader.read()?, + }, + x => { + return reader.invalid_leading_byte(x, "outer alias target"); + } + } + } + 0x03 => ModuleTypeDeclaration::Export { + name: reader.read()?, + ty: reader.read()?, + }, + x => return reader.invalid_leading_byte(x, "type definition"), + }) + } +} + +/// A reader for the core type section of a WebAssembly component. +/// +/// # Examples +/// ``` +/// use wasmparser::CoreTypeSectionReader; +/// # let data: &[u8] = &[0x01, 0x60, 0x00, 0x00]; +/// let mut reader = CoreTypeSectionReader::new(data, 0).unwrap(); +/// for ty in reader { +/// println!("Type {:?}", ty.expect("type")); +/// } +/// ``` +pub type CoreTypeSectionReader<'a> = SectionLimited<'a, CoreType<'a>>; + +/// Represents a value type in a WebAssembly component. +#[derive(Debug, Clone, Copy, PartialEq, Eq)] +pub enum ComponentValType { + /// The value type is a primitive type. + Primitive(PrimitiveValType), + /// The value type is a reference to a defined type. + Type(u32), +} + +impl<'a> FromReader<'a> for ComponentValType { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + if let Some(ty) = PrimitiveValType::from_byte(reader.peek()?) { + reader.position += 1; + return Ok(ComponentValType::Primitive(ty)); + } + + Ok(ComponentValType::Type(reader.read_var_s33()? as u32)) + } +} + +impl<'a> FromReader<'a> for Option<ComponentValType> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + match reader.read_u8()? { + 0x0 => Ok(None), + 0x1 => Ok(Some(reader.read()?)), + x => reader.invalid_leading_byte(x, "optional component value type"), + } + } +} + +/// Represents a primitive value type. +#[derive(Debug, Clone, Copy, PartialEq, Eq)] +pub enum PrimitiveValType { + /// The type is a boolean. + Bool, + /// The type is a signed 8-bit integer. + S8, + /// The type is an unsigned 8-bit integer. + U8, + /// The type is a signed 16-bit integer. + S16, + /// The type is an unsigned 16-bit integer. + U16, + /// The type is a signed 32-bit integer. + S32, + /// The type is an unsigned 32-bit integer. + U32, + /// The type is a signed 64-bit integer. + S64, + /// The type is an unsigned 64-bit integer. + U64, + /// The type is a 32-bit floating point number. + Float32, + /// The type is a 64-bit floating point number. + Float64, + /// The type is a Unicode character. + Char, + /// The type is a string. + String, +} + +impl PrimitiveValType { + fn from_byte(byte: u8) -> Option<PrimitiveValType> { + Some(match byte { + 0x7f => PrimitiveValType::Bool, + 0x7e => PrimitiveValType::S8, + 0x7d => PrimitiveValType::U8, + 0x7c => PrimitiveValType::S16, + 0x7b => PrimitiveValType::U16, + 0x7a => PrimitiveValType::S32, + 0x79 => PrimitiveValType::U32, + 0x78 => PrimitiveValType::S64, + 0x77 => PrimitiveValType::U64, + 0x76 => PrimitiveValType::Float32, + 0x75 => PrimitiveValType::Float64, + 0x74 => PrimitiveValType::Char, + 0x73 => PrimitiveValType::String, + _ => return None, + }) + } + + pub(crate) fn requires_realloc(&self) -> bool { + matches!(self, Self::String) + } + + /// Determines if primitive value type `a` is a subtype of `b`. + pub fn is_subtype_of(a: Self, b: Self) -> bool { + // Note that this intentionally diverges from the upstream specification + // at this time and only considers exact equality for subtyping + // relationships. + // + // More information can be found in the subtyping implementation for + // component functions. + a == b + } +} + +/// Represents a type in a WebAssembly component. +#[derive(Debug, Clone)] +pub enum ComponentType<'a> { + /// The type is a component defined type. + Defined(ComponentDefinedType<'a>), + /// The type is a function type. + Func(ComponentFuncType<'a>), + /// The type is a component type. + Component(Box<[ComponentTypeDeclaration<'a>]>), + /// The type is an instance type. + Instance(Box<[InstanceTypeDeclaration<'a>]>), +} + +impl<'a> FromReader<'a> for ComponentType<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + Ok(match reader.read_u8()? { + 0x40 => { + let params = reader + .read_iter(MAX_WASM_FUNCTION_PARAMS, "component function parameters")? + .collect::<Result<_>>()?; + let results = reader.read()?; + ComponentType::Func(ComponentFuncType { params, results }) + } + 0x41 => ComponentType::Component( + reader + .read_iter(MAX_WASM_COMPONENT_TYPE_DECLS, "component type declaration")? + .collect::<Result<_>>()?, + ), + 0x42 => ComponentType::Instance( + reader + .read_iter(MAX_WASM_INSTANCE_TYPE_DECLS, "instance type declaration")? + .collect::<Result<_>>()?, + ), + x => { + if let Some(ty) = PrimitiveValType::from_byte(x) { + ComponentType::Defined(ComponentDefinedType::Primitive(ty)) + } else { + ComponentType::Defined(ComponentDefinedType::read(reader, x)?) + } + } + }) + } +} + +/// Represents part of a component type declaration in a WebAssembly component. +#[derive(Debug, Clone)] +pub enum ComponentTypeDeclaration<'a> { + /// The component type declaration is for a core type. + CoreType(CoreType<'a>), + /// The component type declaration is for a type. + Type(ComponentType<'a>), + /// The component type declaration is for an alias. + Alias(ComponentAlias<'a>), + /// The component type declaration is for an export. + Export { + /// The name of the export. + name: &'a str, + /// The optional URL of the export. + url: &'a str, + /// The type reference for the export. + ty: ComponentTypeRef, + }, + /// The component type declaration is for an import. + Import(ComponentImport<'a>), +} + +impl<'a> FromReader<'a> for ComponentTypeDeclaration<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + // Component types are effectively instance types with the additional + // variant of imports; check for imports here or delegate to + // `InstanceTypeDeclaration` with the appropriate conversions. + if reader.peek()? == 0x03 { + reader.position += 1; + return Ok(ComponentTypeDeclaration::Import(reader.read()?)); + } + + Ok(match reader.read()? { + InstanceTypeDeclaration::CoreType(t) => ComponentTypeDeclaration::CoreType(t), + InstanceTypeDeclaration::Type(t) => ComponentTypeDeclaration::Type(t), + InstanceTypeDeclaration::Alias(a) => ComponentTypeDeclaration::Alias(a), + InstanceTypeDeclaration::Export { name, url, ty } => { + ComponentTypeDeclaration::Export { name, url, ty } + } + }) + } +} + +/// Represents an instance type declaration in a WebAssembly component. +#[derive(Debug, Clone)] +pub enum InstanceTypeDeclaration<'a> { + /// The component type declaration is for a core type. + CoreType(CoreType<'a>), + /// The instance type declaration is for a type. + Type(ComponentType<'a>), + /// The instance type declaration is for an alias. + Alias(ComponentAlias<'a>), + /// The instance type declaration is for an export. + Export { + /// The name of the export. + name: &'a str, + /// The URL for the export. + url: &'a str, + /// The type reference for the export. + ty: ComponentTypeRef, + }, +} + +impl<'a> FromReader<'a> for InstanceTypeDeclaration<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + Ok(match reader.read_u8()? { + 0x00 => InstanceTypeDeclaration::CoreType(reader.read()?), + 0x01 => InstanceTypeDeclaration::Type(reader.read()?), + 0x02 => InstanceTypeDeclaration::Alias(reader.read()?), + 0x04 => InstanceTypeDeclaration::Export { + name: reader.read()?, + url: reader.read()?, + ty: reader.read()?, + }, + x => return reader.invalid_leading_byte(x, "component or instance type declaration"), + }) + } +} + +/// Represents the result type of a component function. +#[derive(Debug, Clone)] +pub enum ComponentFuncResult<'a> { + /// The function returns a singular, unnamed type. + Unnamed(ComponentValType), + /// The function returns zero or more named types. + Named(Box<[(&'a str, ComponentValType)]>), +} + +impl<'a> FromReader<'a> for ComponentFuncResult<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + Ok(match reader.read_u8()? { + 0x00 => ComponentFuncResult::Unnamed(reader.read()?), + 0x01 => ComponentFuncResult::Named( + reader + .read_iter(MAX_WASM_FUNCTION_RETURNS, "component function results")? + .collect::<Result<_>>()?, + ), + x => return reader.invalid_leading_byte(x, "component function results"), + }) + } +} + +impl ComponentFuncResult<'_> { + /// Gets the count of types returned by the function. + pub fn type_count(&self) -> usize { + match self { + Self::Unnamed(_) => 1, + Self::Named(vec) => vec.len(), + } + } + + /// Iterates over the types returned by the function. + pub fn iter(&self) -> impl Iterator<Item = (Option<&str>, &ComponentValType)> { + enum Either<L, R> { + Left(L), + Right(R), + } + + impl<L, R> Iterator for Either<L, R> + where + L: Iterator, + R: Iterator<Item = L::Item>, + { + type Item = L::Item; + + fn next(&mut self) -> Option<Self::Item> { + match self { + Either::Left(l) => l.next(), + Either::Right(r) => r.next(), + } + } + } + + match self { + Self::Unnamed(ty) => Either::Left(std::iter::once(ty).map(|ty| (None, ty))), + Self::Named(vec) => Either::Right(vec.iter().map(|(n, ty)| (Some(*n), ty))), + } + } +} + +/// Represents a type of a function in a WebAssembly component. +#[derive(Debug, Clone)] +pub struct ComponentFuncType<'a> { + /// The function parameters. + pub params: Box<[(&'a str, ComponentValType)]>, + /// The function result. + pub results: ComponentFuncResult<'a>, +} + +/// Represents a case in a variant type. +#[derive(Debug, Clone, PartialEq, Eq)] +pub struct VariantCase<'a> { + /// The name of the variant case. + pub name: &'a str, + /// The value type of the variant case. + pub ty: Option<ComponentValType>, + /// The index of the variant case that is refined by this one. + pub refines: Option<u32>, +} + +impl<'a> FromReader<'a> for VariantCase<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + Ok(VariantCase { + name: reader.read()?, + ty: reader.read()?, + refines: match reader.read_u8()? { + 0x0 => None, + 0x1 => Some(reader.read_var_u32()?), + x => return reader.invalid_leading_byte(x, "variant case refines"), + }, + }) + } +} + +/// Represents a defined type in a WebAssembly component. +#[derive(Debug, Clone, PartialEq, Eq)] +pub enum ComponentDefinedType<'a> { + /// The type is one of the primitive value types. + Primitive(PrimitiveValType), + /// The type is a record with the given fields. + Record(Box<[(&'a str, ComponentValType)]>), + /// The type is a variant with the given cases. + Variant(Box<[VariantCase<'a>]>), + /// The type is a list of the given value type. + List(ComponentValType), + /// The type is a tuple of the given value types. + Tuple(Box<[ComponentValType]>), + /// The type is flags with the given names. + Flags(Box<[&'a str]>), + /// The type is an enum with the given tags. + Enum(Box<[&'a str]>), + /// The type is a union of the given value types. + Union(Box<[ComponentValType]>), + /// The type is an option of the given value type. + Option(ComponentValType), + /// The type is a result type. + Result { + /// The type returned for success. + ok: Option<ComponentValType>, + /// The type returned for failure. + err: Option<ComponentValType>, + }, +} + +impl<'a> ComponentDefinedType<'a> { + fn read(reader: &mut BinaryReader<'a>, byte: u8) -> Result<ComponentDefinedType<'a>> { + Ok(match byte { + 0x72 => ComponentDefinedType::Record( + reader + .read_iter(MAX_WASM_RECORD_FIELDS, "record field")? + .collect::<Result<_>>()?, + ), + 0x71 => ComponentDefinedType::Variant( + reader + .read_iter(MAX_WASM_VARIANT_CASES, "variant cases")? + .collect::<Result<_>>()?, + ), + 0x70 => ComponentDefinedType::List(reader.read()?), + 0x6f => ComponentDefinedType::Tuple( + reader + .read_iter(MAX_WASM_TUPLE_TYPES, "tuple types")? + .collect::<Result<_>>()?, + ), + 0x6e => ComponentDefinedType::Flags( + reader + .read_iter(MAX_WASM_FLAG_NAMES, "flag names")? + .collect::<Result<_>>()?, + ), + 0x6d => ComponentDefinedType::Enum( + reader + .read_iter(MAX_WASM_ENUM_CASES, "enum cases")? + .collect::<Result<_>>()?, + ), + 0x6c => ComponentDefinedType::Union( + reader + .read_iter(MAX_WASM_UNION_TYPES, "union types")? + .collect::<Result<_>>()?, + ), + 0x6b => ComponentDefinedType::Option(reader.read()?), + 0x6a => ComponentDefinedType::Result { + ok: reader.read()?, + err: reader.read()?, + }, + x => return reader.invalid_leading_byte(x, "component defined type"), + }) + } +} + +/// A reader for the type section of a WebAssembly component. +/// +/// # Examples +/// +/// ``` +/// use wasmparser::ComponentTypeSectionReader; +/// let data: &[u8] = &[0x01, 0x40, 0x01, 0x03, b'f', b'o', b'o', 0x73, 0x00, 0x73]; +/// let mut reader = ComponentTypeSectionReader::new(data, 0).unwrap(); +/// for ty in reader { +/// println!("Type {:?}", ty.expect("type")); +/// } +/// ``` +pub type ComponentTypeSectionReader<'a> = SectionLimited<'a, ComponentType<'a>>; diff --git a/third_party/rust/wasmparser/src/readers/core.rs b/third_party/rust/wasmparser/src/readers/core.rs new file mode 100644 index 0000000000..c42bbf8d9e --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/core.rs @@ -0,0 +1,33 @@ +mod code; +mod custom; +mod data; +mod elements; +mod exports; +mod functions; +mod globals; +mod imports; +mod init; +mod memories; +mod names; +mod operators; +mod producers; +mod tables; +mod tags; +mod types; + +pub use self::code::*; +pub use self::custom::*; +pub use self::data::*; +pub use self::elements::*; +pub use self::exports::*; +pub use self::functions::*; +pub use self::globals::*; +pub use self::imports::*; +pub use self::init::*; +pub use self::memories::*; +pub use self::names::*; +pub use self::operators::*; +pub use self::producers::*; +pub use self::tables::*; +pub use self::tags::*; +pub use self::types::*; diff --git a/third_party/rust/wasmparser/src/readers/core/code.rs b/third_party/rust/wasmparser/src/readers/core/code.rs new file mode 100644 index 0000000000..2a463727e8 --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/core/code.rs @@ -0,0 +1,146 @@ +/* Copyright 2018 Mozilla Foundation + * + * 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. + */ + +use crate::{BinaryReader, FromReader, OperatorsReader, Result, SectionLimited, ValType}; +use std::ops::Range; + +/// A reader for the code section of a WebAssembly module. +pub type CodeSectionReader<'a> = SectionLimited<'a, FunctionBody<'a>>; + +/// Represents a WebAssembly function body. +#[derive(Debug, Clone)] +pub struct FunctionBody<'a> { + reader: BinaryReader<'a>, +} + +impl<'a> FunctionBody<'a> { + /// Constructs a new `FunctionBody` for the given data and offset. + pub fn new(offset: usize, data: &'a [u8]) -> Self { + Self { + reader: BinaryReader::new_with_offset(data, offset), + } + } + + /// Whether or not to allow 64-bit memory arguments in the + /// function body. + /// + /// This is intended to be `true` when support for the memory64 + /// WebAssembly proposal is also enabled. + pub fn allow_memarg64(&mut self, allow: bool) { + self.reader.allow_memarg64(allow); + } + + /// Gets a binary reader for this function body. + pub fn get_binary_reader(&self) -> BinaryReader<'a> { + self.reader.clone() + } + + fn skip_locals(reader: &mut BinaryReader) -> Result<()> { + let count = reader.read_var_u32()?; + for _ in 0..count { + reader.read_var_u32()?; + reader.read::<ValType>()?; + } + Ok(()) + } + + /// Gets the locals reader for this function body. + pub fn get_locals_reader(&self) -> Result<LocalsReader<'a>> { + let mut reader = self.reader.clone(); + let count = reader.read_var_u32()?; + Ok(LocalsReader { reader, count }) + } + + /// Gets the operators reader for this function body. + pub fn get_operators_reader(&self) -> Result<OperatorsReader<'a>> { + let mut reader = self.reader.clone(); + Self::skip_locals(&mut reader)?; + Ok(OperatorsReader::new(reader)) + } + + /// Gets the range of the function body. + pub fn range(&self) -> Range<usize> { + self.reader.range() + } +} + +impl<'a> FromReader<'a> for FunctionBody<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + let reader = reader.read_reader("function body extends past end of the code section")?; + Ok(FunctionBody { reader }) + } +} + +/// A reader for a function body's locals. +pub struct LocalsReader<'a> { + reader: BinaryReader<'a>, + count: u32, +} + +impl<'a> LocalsReader<'a> { + /// Gets the count of locals in the function body. + pub fn get_count(&self) -> u32 { + self.count + } + + /// Gets the original position of the reader. + pub fn original_position(&self) -> usize { + self.reader.original_position() + } + + /// Reads an item from the reader. + pub fn read(&mut self) -> Result<(u32, ValType)> { + let count = self.reader.read()?; + let value_type = self.reader.read()?; + Ok((count, value_type)) + } +} + +impl<'a> IntoIterator for LocalsReader<'a> { + type Item = Result<(u32, ValType)>; + type IntoIter = LocalsIterator<'a>; + fn into_iter(self) -> Self::IntoIter { + let count = self.count; + LocalsIterator { + reader: self, + left: count, + err: false, + } + } +} + +/// An iterator over locals in a function body. +pub struct LocalsIterator<'a> { + reader: LocalsReader<'a>, + left: u32, + err: bool, +} + +impl<'a> Iterator for LocalsIterator<'a> { + type Item = Result<(u32, ValType)>; + fn next(&mut self) -> Option<Self::Item> { + if self.err || self.left == 0 { + return None; + } + let result = self.reader.read(); + self.err = result.is_err(); + self.left -= 1; + Some(result) + } + fn size_hint(&self) -> (usize, Option<usize>) { + let count = self.reader.get_count() as usize; + (count, Some(count)) + } +} diff --git a/third_party/rust/wasmparser/src/readers/core/custom.rs b/third_party/rust/wasmparser/src/readers/core/custom.rs new file mode 100644 index 0000000000..a04fe5a1ac --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/core/custom.rs @@ -0,0 +1,63 @@ +use crate::{BinaryReader, Result}; +use std::ops::Range; + +/// A reader for custom sections of a WebAssembly module. +#[derive(Clone)] +pub struct CustomSectionReader<'a> { + // NB: these fields are public to the crate to make testing easier. + pub(crate) name: &'a str, + pub(crate) data_offset: usize, + pub(crate) data: &'a [u8], + pub(crate) range: Range<usize>, +} + +impl<'a> CustomSectionReader<'a> { + /// Constructs a new `CustomSectionReader` for the given data and offset. + pub fn new(data: &'a [u8], offset: usize) -> Result<CustomSectionReader<'a>> { + let mut reader = BinaryReader::new_with_offset(data, offset); + let name = reader.read_string()?; + let data_offset = reader.original_position(); + let data = reader.remaining_buffer(); + let range = reader.range(); + Ok(CustomSectionReader { + name, + data_offset, + data, + range, + }) + } + + /// The name of the custom section. + pub fn name(&self) -> &'a str { + self.name + } + + /// The offset, relative to the start of the original module or component, + /// that the `data` payload for this custom section starts at. + pub fn data_offset(&self) -> usize { + self.data_offset + } + + /// The actual contents of the custom section. + pub fn data(&self) -> &'a [u8] { + self.data + } + + /// The range of bytes that specify this whole custom section (including + /// both the name of this custom section and its data) specified in + /// offsets relative to the start of the byte stream. + pub fn range(&self) -> Range<usize> { + self.range.clone() + } +} + +impl<'a> std::fmt::Debug for CustomSectionReader<'a> { + fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { + f.debug_struct("CustomSectionReader") + .field("name", &self.name) + .field("data_offset", &self.data_offset) + .field("data", &"...") + .field("range", &self.range) + .finish() + } +} diff --git a/third_party/rust/wasmparser/src/readers/core/data.rs b/third_party/rust/wasmparser/src/readers/core/data.rs new file mode 100644 index 0000000000..5ea5f99457 --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/core/data.rs @@ -0,0 +1,96 @@ +/* Copyright 2018 Mozilla Foundation + * + * 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. + */ + +use crate::{BinaryReader, BinaryReaderError, ConstExpr, FromReader, Result, SectionLimited}; +use std::ops::Range; + +/// Represents a data segment in a core WebAssembly module. +#[derive(Debug, Clone)] +pub struct Data<'a> { + /// The kind of data segment. + pub kind: DataKind<'a>, + /// The data of the data segment. + pub data: &'a [u8], + /// The range of the data segment. + pub range: Range<usize>, +} + +/// The kind of data segment. +#[derive(Debug, Copy, Clone)] +pub enum DataKind<'a> { + /// The data segment is passive. + Passive, + /// The data segment is active. + Active { + /// The memory index for the data segment. + memory_index: u32, + /// The initialization expression for the data segment. + offset_expr: ConstExpr<'a>, + }, +} + +/// A reader for the data section of a WebAssembly module. +pub type DataSectionReader<'a> = SectionLimited<'a, Data<'a>>; + +impl<'a> FromReader<'a> for Data<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + let segment_start = reader.original_position(); + + // The current handling of the flags is largely specified in the `bulk-memory` proposal, + // which at the time this comment is written has been merged to the main specification + // draft. + // + // Notably, this proposal allows multiple different encodings of the memory index 0. `00` + // and `02 00` are both valid ways to specify the 0-th memory. However it also makes + // another encoding of the 0-th memory `80 00` no longer valid. + // + // We, however maintain this by parsing `flags` as a LEB128 integer. In that case, `80 00` + // encoding is parsed out as `0` and is therefore assigned a `memidx` 0, even though the + // current specification draft does not allow for this. + // + // See also https://github.com/WebAssembly/spec/issues/1439 + let flags = reader.read_var_u32()?; + let kind = match flags { + 1 => DataKind::Passive, + 0 | 2 => { + let memory_index = if flags == 0 { + 0 + } else { + reader.read_var_u32()? + }; + let offset_expr = reader.read()?; + DataKind::Active { + memory_index, + offset_expr, + } + } + _ => { + return Err(BinaryReaderError::new( + "invalid flags byte in data segment", + segment_start, + )); + } + }; + + let data = reader.read_reader( + "unexpected end of section or function: data segment extends past end of the section", + )?; + Ok(Data { + kind, + data: data.remaining_buffer(), + range: segment_start..data.range().end, + }) + } +} diff --git a/third_party/rust/wasmparser/src/readers/core/elements.rs b/third_party/rust/wasmparser/src/readers/core/elements.rs new file mode 100644 index 0000000000..7e37e7d7b6 --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/core/elements.rs @@ -0,0 +1,158 @@ +/* Copyright 2018 Mozilla Foundation + * + * 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. + */ + +use crate::{ + BinaryReader, BinaryReaderError, ConstExpr, ExternalKind, FromReader, RefType, Result, + SectionLimited, +}; +use std::ops::Range; + +/// Represents a core WebAssembly element segment. +#[derive(Clone)] +pub struct Element<'a> { + /// The kind of the element segment. + pub kind: ElementKind<'a>, + /// The initial elements of the element segment. + pub items: ElementItems<'a>, + /// The type of the elements. + pub ty: RefType, + /// The range of the the element segment. + pub range: Range<usize>, +} + +/// The kind of element segment. +#[derive(Clone)] +pub enum ElementKind<'a> { + /// The element segment is passive. + Passive, + /// The element segment is active. + Active { + /// The index of the table being initialized. + table_index: u32, + /// The initial expression of the element segment. + offset_expr: ConstExpr<'a>, + }, + /// The element segment is declared. + Declared, +} + +/// Represents the items of an element segment. +#[derive(Clone)] +pub enum ElementItems<'a> { + /// This element contains function indices. + Functions(SectionLimited<'a, u32>), + /// This element contains constant expressions used to initialize the table. + Expressions(SectionLimited<'a, ConstExpr<'a>>), +} + +/// A reader for the element section of a WebAssembly module. +pub type ElementSectionReader<'a> = SectionLimited<'a, Element<'a>>; + +impl<'a> FromReader<'a> for Element<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + let elem_start = reader.original_position(); + // The current handling of the flags is largely specified in the `bulk-memory` proposal, + // which at the time this commend is written has been merged to the main specification + // draft. + // + // Notably, this proposal allows multiple different encodings of the table index 0. `00` + // and `02 00` are both valid ways to specify the 0-th table. However it also makes + // another encoding of the 0-th memory `80 00` no longer valid. + // + // We, however maintain this support by parsing `flags` as a LEB128 integer. In that case, + // `80 00` encoding is parsed out as `0` and is therefore assigned a `tableidx` 0, even + // though the current specification draft does not allow for this. + // + // See also https://github.com/WebAssembly/spec/issues/1439 + let flags = reader.read_var_u32()?; + if (flags & !0b111) != 0 { + return Err(BinaryReaderError::new( + "invalid flags byte in element segment", + reader.original_position() - 1, + )); + } + let kind = if flags & 0b001 != 0 { + if flags & 0b010 != 0 { + ElementKind::Declared + } else { + ElementKind::Passive + } + } else { + let table_index = if flags & 0b010 == 0 { + 0 + } else { + reader.read_var_u32()? + }; + let offset_expr = reader.read()?; + ElementKind::Active { + table_index, + offset_expr, + } + }; + let exprs = flags & 0b100 != 0; + let ty = if flags & 0b011 != 0 { + if exprs { + reader.read()? + } else { + match reader.read()? { + ExternalKind::Func => RefType::FUNCREF, + _ => { + return Err(BinaryReaderError::new( + "only the function external type is supported in elem segment", + reader.original_position() - 1, + )); + } + } + } + } else { + RefType::FUNCREF + }; + // FIXME(#188) ideally wouldn't have to do skips here + let data = reader.skip(|reader| { + let items_count = reader.read_var_u32()?; + if exprs { + for _ in 0..items_count { + reader.skip_const_expr()?; + } + } else { + for _ in 0..items_count { + reader.read_var_u32()?; + } + } + Ok(()) + })?; + let items = if exprs { + ElementItems::Expressions(SectionLimited::new( + data.remaining_buffer(), + data.original_position(), + )?) + } else { + ElementItems::Functions(SectionLimited::new( + data.remaining_buffer(), + data.original_position(), + )?) + }; + + let elem_end = reader.original_position(); + let range = elem_start..elem_end; + + Ok(Element { + kind, + items, + ty, + range, + }) + } +} diff --git a/third_party/rust/wasmparser/src/readers/core/exports.rs b/third_party/rust/wasmparser/src/readers/core/exports.rs new file mode 100644 index 0000000000..c1bd62626b --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/core/exports.rs @@ -0,0 +1,65 @@ +/* Copyright 2018 Mozilla Foundation + * + * 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. + */ + +use crate::{BinaryReader, FromReader, Result, SectionLimited}; + +/// A reader for the export section of a WebAssembly module. +pub type ExportSectionReader<'a> = SectionLimited<'a, Export<'a>>; + +/// External types as defined [here]. +/// +/// [here]: https://webassembly.github.io/spec/core/syntax/types.html#external-types +#[derive(Debug, Copy, Clone, PartialEq, Eq)] +pub enum ExternalKind { + /// The external kind is a function. + Func, + /// The external kind if a table. + Table, + /// The external kind is a memory. + Memory, + /// The external kind is a global. + Global, + /// The external kind is a tag. + Tag, +} + +/// Represents an export in a WebAssembly module. +#[derive(Debug, Copy, Clone)] +pub struct Export<'a> { + /// The name of the exported item. + pub name: &'a str, + /// The kind of the export. + pub kind: ExternalKind, + /// The index of the exported item. + pub index: u32, +} + +impl<'a> FromReader<'a> for Export<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + Ok(Export { + name: reader.read_string()?, + kind: reader.read()?, + index: reader.read_var_u32()?, + }) + } +} + +impl<'a> FromReader<'a> for ExternalKind { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + let offset = reader.original_position(); + let byte = reader.read_u8()?; + BinaryReader::external_kind_from_byte(byte, offset) + } +} diff --git a/third_party/rust/wasmparser/src/readers/core/functions.rs b/third_party/rust/wasmparser/src/readers/core/functions.rs new file mode 100644 index 0000000000..ebddce05a3 --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/core/functions.rs @@ -0,0 +1,17 @@ +/* Copyright 2018 Mozilla Foundation + * + * 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. + */ + +/// A reader for the function section of a WebAssembly module. +pub type FunctionSectionReader<'a> = crate::SectionLimited<'a, u32>; diff --git a/third_party/rust/wasmparser/src/readers/core/globals.rs b/third_party/rust/wasmparser/src/readers/core/globals.rs new file mode 100644 index 0000000000..6fd99bc0b8 --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/core/globals.rs @@ -0,0 +1,49 @@ +/* Copyright 2018 Mozilla Foundation + * + * 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. + */ + +use crate::{BinaryReader, ConstExpr, FromReader, GlobalType, Result, SectionLimited}; + +/// Represents a core WebAssembly global. +#[derive(Debug, Copy, Clone)] +pub struct Global<'a> { + /// The global's type. + pub ty: GlobalType, + /// The global's initialization expression. + pub init_expr: ConstExpr<'a>, +} + +/// A reader for the global section of a WebAssembly module. +pub type GlobalSectionReader<'a> = SectionLimited<'a, Global<'a>>; + +impl<'a> FromReader<'a> for Global<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + let ty = reader.read()?; + let init_expr = reader.read()?; + Ok(Global { ty, init_expr }) + } +} + +impl<'a> FromReader<'a> for GlobalType { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + Ok(GlobalType { + content_type: reader.read()?, + mutable: match reader.read_u8()? { + 0x00 => false, + 0x01 => true, + _ => bail!(reader.original_position() - 1, "malformed mutability",), + }, + }) + } +} diff --git a/third_party/rust/wasmparser/src/readers/core/imports.rs b/third_party/rust/wasmparser/src/readers/core/imports.rs new file mode 100644 index 0000000000..d2a33c89e3 --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/core/imports.rs @@ -0,0 +1,76 @@ +/* Copyright 2018 Mozilla Foundation + * + * 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. + */ + +use crate::{ + BinaryReader, ExternalKind, FromReader, GlobalType, MemoryType, Result, SectionLimited, + TableType, TagType, +}; + +/// Represents a reference to a type definition in a WebAssembly module. +#[derive(Debug, Clone, Copy)] +pub enum TypeRef { + /// The type is a function. + /// + /// The value is an index into the type section. + Func(u32), + /// The type is a table. + Table(TableType), + /// The type is a memory. + Memory(MemoryType), + /// The type is a global. + Global(GlobalType), + /// The type is a tag. + /// + /// This variant is only used for the exception handling proposal. + /// + /// The value is an index in the types index space. + Tag(TagType), +} + +/// Represents an import in a WebAssembly module. +#[derive(Debug, Copy, Clone)] +pub struct Import<'a> { + /// The module being imported from. + pub module: &'a str, + /// The name of the imported item. + pub name: &'a str, + /// The type of the imported item. + pub ty: TypeRef, +} + +/// A reader for the import section of a WebAssembly module. +pub type ImportSectionReader<'a> = SectionLimited<'a, Import<'a>>; + +impl<'a> FromReader<'a> for Import<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + Ok(Import { + module: reader.read()?, + name: reader.read()?, + ty: reader.read()?, + }) + } +} + +impl<'a> FromReader<'a> for TypeRef { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + Ok(match reader.read()? { + ExternalKind::Func => TypeRef::Func(reader.read_var_u32()?), + ExternalKind::Table => TypeRef::Table(reader.read()?), + ExternalKind::Memory => TypeRef::Memory(reader.read()?), + ExternalKind::Global => TypeRef::Global(reader.read()?), + ExternalKind::Tag => TypeRef::Tag(reader.read()?), + }) + } +} diff --git a/third_party/rust/wasmparser/src/readers/core/init.rs b/third_party/rust/wasmparser/src/readers/core/init.rs new file mode 100644 index 0000000000..fcd3bd73c9 --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/core/init.rs @@ -0,0 +1,51 @@ +/* Copyright 2018 Mozilla Foundation + * + * 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. + */ + +use crate::{BinaryReader, FromReader, OperatorsReader, Result}; + +/// Represents an initialization expression. +#[derive(Debug, Copy, Clone)] +pub struct ConstExpr<'a> { + offset: usize, + data: &'a [u8], +} + +impl<'a> ConstExpr<'a> { + /// Constructs a new `ConstExpr` from the given data and offset. + pub fn new(data: &[u8], offset: usize) -> ConstExpr { + ConstExpr { offset, data } + } + + /// Gets a binary reader for the initialization expression. + pub fn get_binary_reader(&self) -> BinaryReader<'a> { + BinaryReader::new_with_offset(self.data, self.offset) + } + + /// Gets an operators reader for the initialization expression. + pub fn get_operators_reader(&self) -> OperatorsReader<'a> { + OperatorsReader::new(self.get_binary_reader()) + } +} + +impl<'a> FromReader<'a> for ConstExpr<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + // FIXME(#188) ideally shouldn't need to skip here + let reader = reader.skip(|r| r.skip_const_expr())?; + Ok(ConstExpr::new( + reader.remaining_buffer(), + reader.original_position(), + )) + } +} diff --git a/third_party/rust/wasmparser/src/readers/core/memories.rs b/third_party/rust/wasmparser/src/readers/core/memories.rs new file mode 100644 index 0000000000..d1941b1cdc --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/core/memories.rs @@ -0,0 +1,56 @@ +/* Copyright 2018 Mozilla Foundation + * + * 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. + */ + +use crate::{BinaryReader, FromReader, MemoryType, Result, SectionLimited}; + +/// A reader for the memory section of a WebAssembly module. +pub type MemorySectionReader<'a> = SectionLimited<'a, MemoryType>; + +impl<'a> FromReader<'a> for MemoryType { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + let pos = reader.original_position(); + let flags = reader.read_u8()?; + if (flags & !0b111) != 0 { + bail!(pos, "invalid memory limits flags"); + } + + let memory64 = flags & 0b100 != 0; + let shared = flags & 0b010 != 0; + let has_max = flags & 0b001 != 0; + Ok(MemoryType { + memory64, + shared, + // FIXME(WebAssembly/memory64#21) as currently specified if the + // `shared` flag is set we should be reading a 32-bit limits field + // here. That seems a bit odd to me at the time of this writing so + // I've taken the liberty of reading a 64-bit limits field in those + // situations. I suspect that this is a typo in the spec, but if not + // we'll need to update this to read a 32-bit limits field when the + // shared flag is set. + initial: if memory64 { + reader.read_var_u64()? + } else { + reader.read_var_u32()?.into() + }, + maximum: if !has_max { + None + } else if memory64 { + Some(reader.read_var_u64()?) + } else { + Some(reader.read_var_u32()?.into()) + }, + }) + } +} diff --git a/third_party/rust/wasmparser/src/readers/core/names.rs b/third_party/rust/wasmparser/src/readers/core/names.rs new file mode 100644 index 0000000000..aa8a11dde2 --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/core/names.rs @@ -0,0 +1,153 @@ +/* Copyright 2018 Mozilla Foundation + * + * 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. + */ + +use crate::{ + BinaryReader, BinaryReaderError, FromReader, Result, SectionLimited, Subsection, Subsections, +}; +use std::ops::Range; + +/// Represents a name map from the names custom section. +pub type NameMap<'a> = SectionLimited<'a, Naming<'a>>; + +/// Represents a name for an index from the names section. +#[derive(Debug, Copy, Clone)] +pub struct Naming<'a> { + /// The index being named. + pub index: u32, + /// The name for the index. + pub name: &'a str, +} + +impl<'a> FromReader<'a> for Naming<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + let index = reader.read_var_u32()?; + let name = reader.read_string()?; + Ok(Naming { index, name }) + } +} + +/// Represents a reader for indirect names from the names custom section. +pub type IndirectNameMap<'a> = SectionLimited<'a, IndirectNaming<'a>>; + +/// Represents an indirect name in the names custom section. +#[derive(Debug, Clone)] +pub struct IndirectNaming<'a> { + /// The indirect index of the name. + pub index: u32, + /// The map of names within the `index` prior. + pub names: NameMap<'a>, +} + +impl<'a> FromReader<'a> for IndirectNaming<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + let index = reader.read_var_u32()?; + + // Skip the `NameMap` manually here. + // + // FIXME(#188) shouldn't need to skip here + let names = reader.skip(|reader| { + let count = reader.read_var_u32()?; + for _ in 0..count { + reader.read_var_u32()?; + reader.skip_string()?; + } + Ok(()) + })?; + + Ok(IndirectNaming { + index, + names: NameMap::new(names.remaining_buffer(), names.original_position())?, + }) + } +} + +/// Represents a name read from the names custom section. +#[derive(Clone)] +pub enum Name<'a> { + /// The name is for the module. + Module { + /// The specified name. + name: &'a str, + /// The byte range that `name` occupies in the original binary. + name_range: Range<usize>, + }, + /// The name is for the functions. + Function(NameMap<'a>), + /// The name is for the function locals. + Local(IndirectNameMap<'a>), + /// The name is for the function labels. + Label(IndirectNameMap<'a>), + /// The name is for the types. + Type(NameMap<'a>), + /// The name is for the tables. + Table(NameMap<'a>), + /// The name is for the memories. + Memory(NameMap<'a>), + /// The name is for the globals. + Global(NameMap<'a>), + /// The name is for the element segments. + Element(NameMap<'a>), + /// The name is for the data segments. + Data(NameMap<'a>), + /// An unknown [name subsection](https://webassembly.github.io/spec/core/appendix/custom.html#subsections). + Unknown { + /// The identifier for this subsection. + ty: u8, + /// The contents of this subsection. + data: &'a [u8], + /// The range of bytes, relative to the start of the original data + /// stream, that the contents of this subsection reside in. + range: Range<usize>, + }, +} + +/// A reader for the name custom section of a WebAssembly module. +pub type NameSectionReader<'a> = Subsections<'a, Name<'a>>; + +impl<'a> Subsection<'a> for Name<'a> { + fn from_reader(id: u8, mut reader: BinaryReader<'a>) -> Result<Self> { + let data = reader.remaining_buffer(); + let offset = reader.original_position(); + Ok(match id { + 0 => { + let name = reader.read_string()?; + if !reader.eof() { + return Err(BinaryReaderError::new( + "trailing data at the end of a name", + reader.original_position(), + )); + } + Name::Module { + name, + name_range: offset..offset + reader.position, + } + } + 1 => Name::Function(NameMap::new(data, offset)?), + 2 => Name::Local(IndirectNameMap::new(data, offset)?), + 3 => Name::Label(IndirectNameMap::new(data, offset)?), + 4 => Name::Type(NameMap::new(data, offset)?), + 5 => Name::Table(NameMap::new(data, offset)?), + 6 => Name::Memory(NameMap::new(data, offset)?), + 7 => Name::Global(NameMap::new(data, offset)?), + 8 => Name::Element(NameMap::new(data, offset)?), + 9 => Name::Data(NameMap::new(data, offset)?), + ty => Name::Unknown { + ty, + data, + range: offset..offset + data.len(), + }, + }) + } +} diff --git a/third_party/rust/wasmparser/src/readers/core/operators.rs b/third_party/rust/wasmparser/src/readers/core/operators.rs new file mode 100644 index 0000000000..d1312c259f --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/core/operators.rs @@ -0,0 +1,354 @@ +/* Copyright 2018 Mozilla Foundation + * + * 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. + */ + +use crate::{BinaryReader, BinaryReaderError, Result, ValType}; + +/// Represents a block type. +#[derive(Debug, Copy, Clone, PartialEq, Eq)] +pub enum BlockType { + /// The block produces consumes nor produces any values. + Empty, + /// The block produces a singular value of the given type ([] -> \[t]). + Type(ValType), + /// The block is described by a function type. + /// + /// The index is to a function type in the types section. + FuncType(u32), +} + +/// Represents a memory immediate in a WebAssembly memory instruction. +#[derive(Debug, Copy, Clone)] +pub struct MemArg { + /// Alignment, stored as `n` where the actual alignment is `2^n` + pub align: u8, + /// Maximum alignment, stored as `n` where the actual alignment is `2^n`. + /// + /// Note that this field is not actually read from the binary format, it + /// will be a constant depending on which instruction this `MemArg` is a + /// payload for. + pub max_align: u8, + /// A fixed byte-offset that this memory immediate specifies. + /// + /// Note that the memory64 proposal can specify a full 64-bit byte offset + /// while otherwise only 32-bit offsets are allowed. Once validated + /// memory immediates for 32-bit memories are guaranteed to be at most + /// `u32::MAX` whereas 64-bit memories can use the full 64-bits. + pub offset: u64, + /// The index of the memory this immediate points to. + /// + /// Note that this points within the module's own memory index space, and + /// is always zero unless the multi-memory proposal of WebAssembly is + /// enabled. + pub memory: u32, +} + +/// A br_table entries representation. +#[derive(Clone)] +pub struct BrTable<'a> { + pub(crate) reader: crate::BinaryReader<'a>, + pub(crate) cnt: u32, + pub(crate) default: u32, +} + +/// An IEEE binary32 immediate floating point value, represented as a u32 +/// containing the bit pattern. +/// +/// All bit patterns are allowed. +#[derive(Copy, Clone, Debug, Eq, PartialEq, Hash)] +pub struct Ieee32(pub(crate) u32); + +impl Ieee32 { + /// Gets the underlying bits of the 32-bit float. + pub fn bits(self) -> u32 { + self.0 + } +} + +/// An IEEE binary64 immediate floating point value, represented as a u64 +/// containing the bit pattern. +/// +/// All bit patterns are allowed. +#[derive(Copy, Clone, Debug, Eq, PartialEq, Hash)] +pub struct Ieee64(pub(crate) u64); + +impl Ieee64 { + /// Gets the underlying bits of the 64-bit float. + pub fn bits(self) -> u64 { + self.0 + } +} + +/// Represents a 128-bit vector value. +#[derive(Copy, Clone, Debug, Eq, PartialEq, Hash)] +pub struct V128(pub(crate) [u8; 16]); + +impl V128 { + /// Gets the bytes of the vector value. + pub fn bytes(&self) -> &[u8; 16] { + &self.0 + } + + /// Gets a signed 128-bit integer value from the vector's bytes. + pub fn i128(&self) -> i128 { + i128::from_le_bytes(self.0) + } +} + +macro_rules! define_operator { + ($(@$proposal:ident $op:ident $({ $($payload:tt)* })? => $visit:ident)*) => { + /// Instructions as defined [here]. + /// + /// [here]: https://webassembly.github.io/spec/core/binary/instructions.html + #[derive(Debug, Clone)] + #[allow(missing_docs)] + pub enum Operator<'a> { + $( + $op $({ $($payload)* })?, + )* + } + } +} +for_each_operator!(define_operator); + +/// A reader for a core WebAssembly function's operators. +#[derive(Clone)] +pub struct OperatorsReader<'a> { + pub(crate) reader: BinaryReader<'a>, +} + +impl<'a> OperatorsReader<'a> { + pub(crate) fn new(reader: BinaryReader<'a>) -> OperatorsReader<'a> { + OperatorsReader { reader } + } + + /// Determines if the reader is at the end of the operators. + pub fn eof(&self) -> bool { + self.reader.eof() + } + + /// Gets the original position of the reader. + pub fn original_position(&self) -> usize { + self.reader.original_position() + } + + /// Whether or not to allow 64-bit memory arguments in the + /// the operators being read. + /// + /// This is intended to be `true` when support for the memory64 + /// WebAssembly proposal is also enabled. + pub fn allow_memarg64(&mut self, allow: bool) { + self.reader.allow_memarg64(allow); + } + + /// Ensures the reader is at the end. + /// + /// This function returns an error if there is extra data after the operators. + pub fn ensure_end(&self) -> Result<()> { + if self.eof() { + return Ok(()); + } + Err(BinaryReaderError::new( + "unexpected data at the end of operators", + self.reader.original_position(), + )) + } + + /// Reads an operator from the reader. + pub fn read(&mut self) -> Result<Operator<'a>> { + self.reader.read_operator() + } + + /// Converts to an iterator of operators paired with offsets. + pub fn into_iter_with_offsets(self) -> OperatorsIteratorWithOffsets<'a> { + OperatorsIteratorWithOffsets { + reader: self, + err: false, + } + } + + /// Reads an operator with its offset. + pub fn read_with_offset(&mut self) -> Result<(Operator<'a>, usize)> { + let pos = self.reader.original_position(); + Ok((self.read()?, pos)) + } + + /// Visit a single operator with the specified [`VisitOperator`] instance. + /// + /// See [`BinaryReader::visit_operator`] for more information. + pub fn visit_operator<T>(&mut self, visitor: &mut T) -> Result<<T as VisitOperator<'a>>::Output> + where + T: VisitOperator<'a>, + { + self.reader.visit_operator(visitor) + } + + /// Gets a binary reader from this operators reader. + pub fn get_binary_reader(&self) -> BinaryReader<'a> { + self.reader.clone() + } +} + +impl<'a> IntoIterator for OperatorsReader<'a> { + type Item = Result<Operator<'a>>; + type IntoIter = OperatorsIterator<'a>; + + /// Reads content of the code section. + /// + /// # Examples + /// ``` + /// use wasmparser::{Operator, CodeSectionReader, Result}; + /// # let data: &[u8] = &[ + /// # 0x01, 0x03, 0x00, 0x01, 0x0b]; + /// let code_reader = CodeSectionReader::new(data, 0).unwrap(); + /// for body in code_reader { + /// let body = body.expect("function body"); + /// let mut op_reader = body.get_operators_reader().expect("op reader"); + /// let ops = op_reader.into_iter().collect::<Result<Vec<Operator>>>().expect("ops"); + /// assert!( + /// if let [Operator::Nop, Operator::End] = ops.as_slice() { true } else { false }, + /// "found {:?}", + /// ops + /// ); + /// } + /// ``` + fn into_iter(self) -> Self::IntoIter { + OperatorsIterator { + reader: self, + err: false, + } + } +} + +/// An iterator over a function's operators. +pub struct OperatorsIterator<'a> { + reader: OperatorsReader<'a>, + err: bool, +} + +impl<'a> Iterator for OperatorsIterator<'a> { + type Item = Result<Operator<'a>>; + + fn next(&mut self) -> Option<Self::Item> { + if self.err || self.reader.eof() { + return None; + } + let result = self.reader.read(); + self.err = result.is_err(); + Some(result) + } +} + +/// An iterator over a function's operators with offsets. +pub struct OperatorsIteratorWithOffsets<'a> { + reader: OperatorsReader<'a>, + err: bool, +} + +impl<'a> Iterator for OperatorsIteratorWithOffsets<'a> { + type Item = Result<(Operator<'a>, usize)>; + + /// Reads content of the code section with offsets. + /// + /// # Examples + /// ``` + /// use wasmparser::{Operator, CodeSectionReader, Result}; + /// # let data: &[u8] = &[ + /// # 0x01, 0x03, 0x00, /* offset = 23 */ 0x01, 0x0b]; + /// let code_reader = CodeSectionReader::new(data, 20).unwrap(); + /// for body in code_reader { + /// let body = body.expect("function body"); + /// let mut op_reader = body.get_operators_reader().expect("op reader"); + /// let ops = op_reader.into_iter_with_offsets().collect::<Result<Vec<(Operator, usize)>>>().expect("ops"); + /// assert!( + /// if let [(Operator::Nop, 23), (Operator::End, 24)] = ops.as_slice() { true } else { false }, + /// "found {:?}", + /// ops + /// ); + /// } + /// ``` + fn next(&mut self) -> Option<Self::Item> { + if self.err || self.reader.eof() { + return None; + } + let result = self.reader.read_with_offset(); + self.err = result.is_err(); + Some(result) + } +} + +macro_rules! define_visit_operator { + ($(@$proposal:ident $op:ident $({ $($arg:ident: $argty:ty),* })? => $visit:ident)*) => { + $( + fn $visit(&mut self $($(,$arg: $argty)*)?) -> Self::Output; + )* + } +} + +/// Trait implemented by types that can visit all [`Operator`] variants. +#[allow(missing_docs)] +pub trait VisitOperator<'a> { + /// The result type of the visitor. + type Output: 'a; + + /// Visits the [`Operator`] `op` using the given `offset`. + /// + /// # Note + /// + /// This is a convenience method that is intended for non-performance + /// critical use cases. For performance critical implementations users + /// are recommended to directly use the respective `visit` methods or + /// implement [`VisitOperator`] on their own. + fn visit_operator(&mut self, op: &Operator<'a>) -> Self::Output { + macro_rules! visit_operator { + ($(@$proposal:ident $op:ident $({ $($arg:ident: $argty:ty),* })? => $visit:ident)*) => { + match op { + $( + Operator::$op $({ $($arg),* })? => self.$visit($($($arg.clone()),*)?), + )* + } + } + + } + for_each_operator!(visit_operator) + } + + for_each_operator!(define_visit_operator); +} + +macro_rules! define_visit_operator_delegate { + ($(@$proposal:ident $op:ident $({ $($arg:ident: $argty:ty),* })? => $visit:ident)*) => { + $( + fn $visit(&mut self $($(,$arg: $argty)*)?) -> Self::Output { + V::$visit(&mut *self, $($($arg),*)?) + } + )* + } +} + +impl<'a, 'b, V: VisitOperator<'a> + ?Sized> VisitOperator<'a> for &'b mut V { + type Output = V::Output; + fn visit_operator(&mut self, op: &Operator<'a>) -> Self::Output { + V::visit_operator(*self, op) + } + for_each_operator!(define_visit_operator_delegate); +} + +impl<'a, V: VisitOperator<'a> + ?Sized> VisitOperator<'a> for Box<V> { + type Output = V::Output; + fn visit_operator(&mut self, op: &Operator<'a>) -> Self::Output { + V::visit_operator(&mut *self, op) + } + for_each_operator!(define_visit_operator_delegate); +} diff --git a/third_party/rust/wasmparser/src/readers/core/producers.rs b/third_party/rust/wasmparser/src/readers/core/producers.rs new file mode 100644 index 0000000000..07785ed75a --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/core/producers.rs @@ -0,0 +1,78 @@ +/* Copyright 2019 Mozilla Foundation + * + * 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. + */ + +use crate::{BinaryReader, FromReader, Result, SectionLimited}; + +/// A reader for the producers custom section of a WebAssembly module. +/// +/// # Examples +/// +/// ``` +/// # let data: &[u8] = &[0x01, 0x08, 0x6c, 0x61, 0x6e, 0x67, 0x75, 0x61, 0x67, 0x65, +/// # 0x02, 0x03, 0x77, 0x61, 0x74, 0x01, 0x31, 0x01, 0x43, 0x03, 0x39, 0x2e, 0x30]; +/// use wasmparser::{ProducersSectionReader, ProducersFieldValue, Result}; +/// let reader = ProducersSectionReader::new(data, 0).expect("producers reader"); +/// let field = reader.into_iter().next().unwrap().expect("producers field"); +/// assert!(field.name == "language"); +/// let value = field.values.into_iter().collect::<Result<Vec<_>>>().expect("values"); +/// assert!(value.len() == 2); +/// assert!(value[0].name == "wat" && value[0].version == "1"); +/// assert!(value[1].name == "C" && value[1].version == "9.0"); +/// ``` +pub type ProducersSectionReader<'a> = SectionLimited<'a, ProducersField<'a>>; + +/// A field from the producers custom section. +#[derive(Debug, Clone)] +pub struct ProducersField<'a> { + /// The name of the field. + pub name: &'a str, + /// The values specified for this field + pub values: SectionLimited<'a, ProducersFieldValue<'a>>, +} + +impl<'a> FromReader<'a> for ProducersField<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + let name = reader.read_string()?; + let values = reader.skip(|reader| { + // FIXME(#188) ideally shouldn't need to skip here + for _ in 0..reader.read_var_u32()? { + reader.skip_string()?; + reader.skip_string()?; + } + Ok(()) + })?; + Ok(ProducersField { + name, + values: SectionLimited::new(values.remaining_buffer(), values.original_position())?, + }) + } +} + +/// Represents a field value in the producers custom section. +#[derive(Debug, Copy, Clone)] +pub struct ProducersFieldValue<'a> { + /// The field name. + pub name: &'a str, + /// The field version. + pub version: &'a str, +} + +impl<'a> FromReader<'a> for ProducersFieldValue<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + let name = reader.read_string()?; + let version = reader.read_string()?; + Ok(ProducersFieldValue { name, version }) + } +} diff --git a/third_party/rust/wasmparser/src/readers/core/tables.rs b/third_party/rust/wasmparser/src/readers/core/tables.rs new file mode 100644 index 0000000000..211e415efd --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/core/tables.rs @@ -0,0 +1,87 @@ +/* Copyright 2018 Mozilla Foundation + * + * 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. + */ + +use crate::{BinaryReader, ConstExpr, FromReader, Result, SectionLimited, TableType}; + +/// A reader for the table section of a WebAssembly module. +pub type TableSectionReader<'a> = SectionLimited<'a, Table<'a>>; + +/// Type information about a table defined in the table section of a WebAssembly +/// module. +#[derive(Debug)] +pub struct Table<'a> { + /// The type of this table, including its element type and its limits. + pub ty: TableType, + /// The initialization expression for the table. + pub init: TableInit<'a>, +} + +/// Different modes of initializing a table. +#[derive(Debug)] +pub enum TableInit<'a> { + /// The table is initialized to all null elements. + RefNull, + /// Each element in the table is initialized with the specified constant + /// expression. + Expr(ConstExpr<'a>), +} + +impl<'a> FromReader<'a> for Table<'a> { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + let has_init_expr = if reader.peek()? == 0x40 { + reader.read_u8()?; + true + } else { + false + }; + + if has_init_expr { + if reader.read_u8()? != 0x00 { + bail!(reader.original_position() - 1, "invalid table encoding"); + } + } + + let ty = reader.read::<TableType>()?; + let init = if has_init_expr { + TableInit::Expr(reader.read()?) + } else { + TableInit::RefNull + }; + Ok(Table { ty, init }) + } +} + +impl<'a> FromReader<'a> for TableType { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + let element_type = reader.read()?; + let has_max = match reader.read_u8()? { + 0x00 => false, + 0x01 => true, + _ => { + bail!( + reader.original_position() - 1, + "invalid table resizable limits flags", + ) + } + }; + let initial = reader.read()?; + let maximum = if has_max { Some(reader.read()?) } else { None }; + Ok(TableType { + element_type, + initial, + maximum, + }) + } +} diff --git a/third_party/rust/wasmparser/src/readers/core/tags.rs b/third_party/rust/wasmparser/src/readers/core/tags.rs new file mode 100644 index 0000000000..746b3ea7ac --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/core/tags.rs @@ -0,0 +1,32 @@ +/* Copyright 2020 Mozilla Foundation + * + * 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. + */ + +use crate::{BinaryReader, FromReader, Result, SectionLimited, TagKind, TagType}; + +/// A reader for the tags section of a WebAssembly module. +pub type TagSectionReader<'a> = SectionLimited<'a, TagType>; + +impl<'a> FromReader<'a> for TagType { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + let attribute = reader.read_u8()?; + if attribute != 0 { + bail!(reader.original_position() - 1, "invalid tag attributes"); + } + Ok(TagType { + kind: TagKind::Exception, + func_type_idx: reader.read_var_u32()?, + }) + } +} diff --git a/third_party/rust/wasmparser/src/readers/core/types.rs b/third_party/rust/wasmparser/src/readers/core/types.rs new file mode 100644 index 0000000000..4358e2670c --- /dev/null +++ b/third_party/rust/wasmparser/src/readers/core/types.rs @@ -0,0 +1,380 @@ +/* Copyright 2018 Mozilla Foundation + * + * 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. + */ + +use crate::limits::{MAX_WASM_FUNCTION_PARAMS, MAX_WASM_FUNCTION_RETURNS}; +use crate::{BinaryReader, FromReader, Result, SectionLimited}; +use std::fmt::Debug; + +/// Represents the types of values in a WebAssembly module. +#[derive(Debug, Copy, Clone, PartialEq, Eq, Hash)] +pub enum ValType { + /// The value type is i32. + I32, + /// The value type is i64. + I64, + /// The value type is f32. + F32, + /// The value type is f64. + F64, + /// The value type is v128. + V128, + /// The value type is a reference. Which type of reference is decided by + /// RefType. This is a change in syntax from the function references proposal, + /// which now provides FuncRef and ExternRef as sugar for the generic ref + /// construct. + Ref(RefType), +} + +/// A reference type. When the function references feature is disabled, this +/// only represents funcref and externref, using the following format: +/// RefType { nullable: true, heap_type: Func | Extern }) +#[derive(Debug, Copy, Clone, PartialEq, Eq, Hash)] +#[repr(packed)] +pub struct RefType { + /// Whether it's nullable + pub nullable: bool, + /// The relevant heap type + pub heap_type: HeapType, +} + +impl RefType { + /// Alias for the wasm `funcref` type. + pub const FUNCREF: RefType = RefType { + nullable: true, + heap_type: HeapType::Func, + }; + /// Alias for the wasm `externref` type. + pub const EXTERNREF: RefType = RefType { + nullable: true, + heap_type: HeapType::Extern, + }; +} + +impl From<RefType> for ValType { + fn from(ty: RefType) -> ValType { + ValType::Ref(ty) + } +} + +/// Used as a performance optimization in HeapType. Call `.into()` to get the u32 +// A u16 forces 2-byte alignment, which forces HeapType to be 4 bytes, +// which forces ValType to 5 bytes. This newtype is annotated as unaligned to +// store the necessary bits compactly +#[derive(Debug, Copy, Clone, PartialEq, Eq, Hash)] +#[repr(packed)] +pub struct PackedIndex(u16); + +impl TryFrom<u32> for PackedIndex { + type Error = (); + + fn try_from(idx: u32) -> Result<PackedIndex, ()> { + idx.try_into().map(PackedIndex).map_err(|_| ()) + } +} + +impl From<PackedIndex> for u32 { + fn from(x: PackedIndex) -> u32 { + x.0 as u32 + } +} + +/// A heap type from function references. When the proposal is disabled, Index +/// is an invalid type. +#[derive(Debug, Copy, Clone, PartialEq, Eq, Hash)] +pub enum HeapType { + /// Function type index + /// Note: [PackedIndex] may need to be unpacked + TypedFunc(PackedIndex), + /// From reference types + Func, + /// From reference types + Extern, +} + +impl ValType { + /// Alias for the wasm `funcref` type. + pub const FUNCREF: ValType = ValType::Ref(RefType::FUNCREF); + /// Alias for the wasm `externref` type. + pub const EXTERNREF: ValType = ValType::Ref(RefType::EXTERNREF); + + /// Returns whether this value type is a "reference type". + /// + /// Only reference types are allowed in tables, for example, and with some + /// instructions. Current reference types include `funcref` and `externref`. + pub fn is_reference_type(&self) -> bool { + matches!(self, ValType::Ref(_)) + } + /// Whether the type is defaultable according to function references + /// spec. This amounts to whether it's a non-nullable ref + pub fn is_defaultable(&self) -> bool { + !matches!( + self, + ValType::Ref(RefType { + nullable: false, + .. + }) + ) + } + + pub(crate) fn is_valtype_byte(byte: u8) -> bool { + match byte { + 0x7F | 0x7E | 0x7D | 0x7C | 0x7B | 0x70 | 0x6F | 0x6B | 0x6C => true, + _ => false, + } + } +} + +impl<'a> FromReader<'a> for ValType { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + match reader.peek()? { + 0x7F => { + reader.position += 1; + Ok(ValType::I32) + } + 0x7E => { + reader.position += 1; + Ok(ValType::I64) + } + 0x7D => { + reader.position += 1; + Ok(ValType::F32) + } + 0x7C => { + reader.position += 1; + Ok(ValType::F64) + } + 0x7B => { + reader.position += 1; + Ok(ValType::V128) + } + 0x70 | 0x6F | 0x6B | 0x6C => Ok(ValType::Ref(reader.read()?)), + _ => bail!(reader.original_position(), "invalid value type"), + } + } +} + +impl<'a> FromReader<'a> for RefType { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + match reader.read()? { + 0x70 => Ok(RefType::FUNCREF), + 0x6F => Ok(RefType::EXTERNREF), + byte @ (0x6B | 0x6C) => Ok(RefType { + nullable: byte == 0x6C, + heap_type: reader.read()?, + }), + _ => bail!(reader.original_position(), "malformed reference type"), + } + } +} + +impl<'a> FromReader<'a> for HeapType { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + match reader.peek()? { + 0x70 => { + reader.position += 1; + Ok(HeapType::Func) + } + 0x6F => { + reader.position += 1; + Ok(HeapType::Extern) + } + _ => { + let idx = match u32::try_from(reader.read_var_s33()?) { + Ok(idx) => idx, + Err(_) => { + bail!(reader.original_position(), "invalid function heap type",); + } + }; + match idx.try_into() { + Ok(packed) => Ok(HeapType::TypedFunc(packed)), + Err(_) => { + bail!(reader.original_position(), "function index too large"); + } + } + } + } + } +} + +/// Represents a type in a WebAssembly module. +#[derive(Debug, Clone)] +pub enum Type { + /// The type is for a function. + Func(FuncType), +} + +/// Represents a type of a function in a WebAssembly module. +#[derive(Clone, Eq, PartialEq, Hash)] +pub struct FuncType { + /// The combined parameters and result types. + params_results: Box<[ValType]>, + /// The number of parameter types. + len_params: usize, +} + +impl Debug for FuncType { + fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { + f.debug_struct("FuncType") + .field("params", &self.params()) + .field("returns", &self.results()) + .finish() + } +} + +impl FuncType { + /// Creates a new [`FuncType`] from the given `params` and `results`. + pub fn new<P, R>(params: P, results: R) -> Self + where + P: IntoIterator<Item = ValType>, + R: IntoIterator<Item = ValType>, + { + let mut buffer = params.into_iter().collect::<Vec<_>>(); + let len_params = buffer.len(); + buffer.extend(results); + Self { + params_results: buffer.into(), + len_params, + } + } + + /// Creates a new [`FuncType`] fom its raw parts. + /// + /// # Panics + /// + /// If `len_params` is greater than the length of `params_results` combined. + pub(crate) fn from_raw_parts(params_results: Box<[ValType]>, len_params: usize) -> Self { + assert!(len_params <= params_results.len()); + Self { + params_results, + len_params, + } + } + + /// Returns a shared slice to the parameter types of the [`FuncType`]. + #[inline] + pub fn params(&self) -> &[ValType] { + &self.params_results[..self.len_params] + } + + /// Returns a shared slice to the result types of the [`FuncType`]. + #[inline] + pub fn results(&self) -> &[ValType] { + &self.params_results[self.len_params..] + } +} + +/// Represents a table's type. +#[derive(Debug, Copy, Clone, PartialEq, Eq, Hash)] +pub struct TableType { + /// The table's element type. + pub element_type: RefType, + /// Initial size of this table, in elements. + pub initial: u32, + /// Optional maximum size of the table, in elements. + pub maximum: Option<u32>, +} + +/// Represents a memory's type. +#[derive(Debug, Copy, Clone, PartialEq, Eq, Hash)] +pub struct MemoryType { + /// Whether or not this is a 64-bit memory, using i64 as an index. If this + /// is false it's a 32-bit memory using i32 as an index. + /// + /// This is part of the memory64 proposal in WebAssembly. + pub memory64: bool, + + /// Whether or not this is a "shared" memory, indicating that it should be + /// send-able across threads and the `maximum` field is always present for + /// valid types. + /// + /// This is part of the threads proposal in WebAssembly. + pub shared: bool, + + /// Initial size of this memory, in wasm pages. + /// + /// For 32-bit memories (when `memory64` is `false`) this is guaranteed to + /// be at most `u32::MAX` for valid types. + pub initial: u64, + + /// Optional maximum size of this memory, in wasm pages. + /// + /// For 32-bit memories (when `memory64` is `false`) this is guaranteed to + /// be at most `u32::MAX` for valid types. This field is always present for + /// valid wasm memories when `shared` is `true`. + pub maximum: Option<u64>, +} + +impl MemoryType { + /// Gets the index type for the memory. + pub fn index_type(&self) -> ValType { + if self.memory64 { + ValType::I64 + } else { + ValType::I32 + } + } +} + +/// Represents a global's type. +#[derive(Debug, Copy, Clone, PartialEq, Eq, Hash)] +pub struct GlobalType { + /// The global's type. + pub content_type: ValType, + /// Whether or not the global is mutable. + pub mutable: bool, +} + +/// Represents a tag kind. +#[derive(Clone, Copy, Debug)] +pub enum TagKind { + /// The tag is an exception type. + Exception, +} + +/// A tag's type. +#[derive(Clone, Copy, Debug)] +pub struct TagType { + /// The kind of tag + pub kind: TagKind, + /// The function type this tag uses. + pub func_type_idx: u32, +} + +/// A reader for the type section of a WebAssembly module. +pub type TypeSectionReader<'a> = SectionLimited<'a, Type>; + +impl<'a> FromReader<'a> for Type { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + Ok(match reader.read_u8()? { + 0x60 => Type::Func(reader.read()?), + x => return reader.invalid_leading_byte(x, "type"), + }) + } +} + +impl<'a> FromReader<'a> for FuncType { + fn from_reader(reader: &mut BinaryReader<'a>) -> Result<Self> { + let mut params_results = reader + .read_iter(MAX_WASM_FUNCTION_PARAMS, "function params")? + .collect::<Result<Vec<_>>>()?; + let len_params = params_results.len(); + let results = reader.read_iter(MAX_WASM_FUNCTION_RETURNS, "function returns")?; + params_results.reserve(results.size_hint().0); + for result in results { + params_results.push(result?); + } + Ok(FuncType::from_raw_parts(params_results.into(), len_params)) + } +} diff --git a/third_party/rust/wasmparser/src/resources.rs b/third_party/rust/wasmparser/src/resources.rs new file mode 100644 index 0000000000..2f0cb5309d --- /dev/null +++ b/third_party/rust/wasmparser/src/resources.rs @@ -0,0 +1,395 @@ +/* Copyright 2019 Mozilla Foundation + * + * 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. + */ + +use crate::{ + BinaryReaderError, FuncType, GlobalType, HeapType, MemoryType, RefType, TableType, ValType, + WasmFeatures, +}; +use std::ops::Range; + +/// Types that qualify as Wasm function types for validation purposes. +pub trait WasmFuncType { + /// Returns the number of input types. + fn len_inputs(&self) -> usize; + /// Returns the number of output types. + fn len_outputs(&self) -> usize; + /// Returns the type at given index if any. + /// + /// # Note + /// + /// The returned type may be wrapped by the user crate and thus + /// the actually returned type only has to be comparable to a Wasm type. + fn input_at(&self, at: u32) -> Option<ValType>; + /// Returns the type at given index if any. + /// + /// # Note + /// + /// The returned type may be wrapped by the user crate and thus + /// the actually returned type only has to be comparable to a Wasm type. + fn output_at(&self, at: u32) -> Option<ValType>; + + /// Returns the list of inputs as an iterator. + fn inputs(&self) -> WasmFuncTypeInputs<'_, Self> + where + Self: Sized, + { + WasmFuncTypeInputs { + func_type: self, + range: 0..self.len_inputs() as u32, + } + } + + /// Returns the list of outputs as an iterator. + fn outputs(&self) -> WasmFuncTypeOutputs<'_, Self> + where + Self: Sized, + { + WasmFuncTypeOutputs { + func_type: self, + range: 0..self.len_outputs() as u32, + } + } +} + +impl<T> WasmFuncType for &'_ T +where + T: ?Sized + WasmFuncType, +{ + fn len_inputs(&self) -> usize { + T::len_inputs(self) + } + fn len_outputs(&self) -> usize { + T::len_outputs(self) + } + fn input_at(&self, at: u32) -> Option<ValType> { + T::input_at(self, at) + } + fn output_at(&self, at: u32) -> Option<ValType> { + T::output_at(self, at) + } +} + +/// Iterator over the inputs of a Wasm function type. +pub struct WasmFuncTypeInputs<'a, T> { + /// The iterated-over function type. + func_type: &'a T, + /// The range we're iterating over. + range: Range<u32>, +} + +impl<T> Iterator for WasmFuncTypeInputs<'_, T> +where + T: WasmFuncType, +{ + type Item = crate::ValType; + + fn next(&mut self) -> Option<Self::Item> { + self.range + .next() + .map(|i| self.func_type.input_at(i).unwrap()) + } + + fn size_hint(&self) -> (usize, Option<usize>) { + self.range.size_hint() + } +} + +impl<T> DoubleEndedIterator for WasmFuncTypeInputs<'_, T> +where + T: WasmFuncType, +{ + fn next_back(&mut self) -> Option<Self::Item> { + self.range + .next_back() + .map(|i| self.func_type.input_at(i).unwrap()) + } +} + +impl<T> ExactSizeIterator for WasmFuncTypeInputs<'_, T> +where + T: WasmFuncType, +{ + fn len(&self) -> usize { + self.range.len() + } +} + +impl<'a, T> Clone for WasmFuncTypeInputs<'a, T> { + fn clone(&self) -> WasmFuncTypeInputs<'a, T> { + WasmFuncTypeInputs { + func_type: self.func_type, + range: self.range.clone(), + } + } +} + +/// Iterator over the outputs of a Wasm function type. +pub struct WasmFuncTypeOutputs<'a, T> { + /// The iterated-over function type. + func_type: &'a T, + /// The range we're iterating over. + range: Range<u32>, +} + +impl<T> Iterator for WasmFuncTypeOutputs<'_, T> +where + T: WasmFuncType, +{ + type Item = crate::ValType; + + fn next(&mut self) -> Option<Self::Item> { + self.range + .next() + .map(|i| self.func_type.output_at(i).unwrap()) + } + + fn size_hint(&self) -> (usize, Option<usize>) { + self.range.size_hint() + } +} + +impl<T> DoubleEndedIterator for WasmFuncTypeOutputs<'_, T> +where + T: WasmFuncType, +{ + fn next_back(&mut self) -> Option<Self::Item> { + self.range + .next_back() + .map(|i| self.func_type.output_at(i).unwrap()) + } +} + +impl<T> ExactSizeIterator for WasmFuncTypeOutputs<'_, T> +where + T: WasmFuncType, +{ + fn len(&self) -> usize { + self.range.len() + } +} + +impl<'a, T> Clone for WasmFuncTypeOutputs<'a, T> { + fn clone(&self) -> WasmFuncTypeOutputs<'a, T> { + WasmFuncTypeOutputs { + func_type: self.func_type, + range: self.range.clone(), + } + } +} + +/// Types that qualify as Wasm validation database. +/// +/// # Note +/// +/// The `wasmparser` crate provides a builtin validation framework but allows +/// users of this crate to also feed the parsed Wasm into their own data +/// structure while parsing and also validate at the same time without +/// the need of an additional parsing or validation step or copying data around. +pub trait WasmModuleResources { + /// The function type used for validation. + type FuncType: WasmFuncType; + + /// Returns the table at given index if any. + fn table_at(&self, at: u32) -> Option<TableType>; + /// Returns the linear memory at given index. + fn memory_at(&self, at: u32) -> Option<MemoryType>; + /// Returns the tag at given index. + fn tag_at(&self, at: u32) -> Option<&Self::FuncType>; + /// Returns the global variable at given index. + fn global_at(&self, at: u32) -> Option<GlobalType>; + /// Returns the `FuncType` associated with the given type index. + fn func_type_at(&self, type_idx: u32) -> Option<&Self::FuncType>; + /// Returns the type index associated with the given function + /// index. type_of_function = func_type_at(type_index_of_function) + fn type_index_of_function(&self, func_idx: u32) -> Option<u32>; + /// Returns the `FuncType` associated with the given function index. + fn type_of_function(&self, func_idx: u32) -> Option<&Self::FuncType>; + /// Returns the element type at the given index. + fn element_type_at(&self, at: u32) -> Option<RefType>; + /// Under the function references proposal, returns whether t1 <= + /// t2. Otherwise, returns whether t1 == t2 + fn matches(&self, t1: ValType, t2: ValType) -> bool; + /// Check a value type. This requires using func_type_at to check references + fn check_value_type( + &self, + t: ValType, + features: &WasmFeatures, + offset: usize, + ) -> Result<(), BinaryReaderError>; + + /// Checks that a `HeapType` is valid, notably its function index if one is + /// used. + fn check_heap_type( + &self, + heap_type: HeapType, + features: &WasmFeatures, + offset: usize, + ) -> Result<(), BinaryReaderError> { + // Delegate to the generic value type validation which will have the + // same validity checks. + self.check_value_type( + RefType { + nullable: true, + heap_type, + } + .into(), + features, + offset, + ) + } + + /// Returns the number of elements. + fn element_count(&self) -> u32; + /// Returns the number of bytes in the Wasm data section. + fn data_count(&self) -> Option<u32>; + /// Returns whether the function index is referenced in the module anywhere + /// outside of the start/function sections. + fn is_function_referenced(&self, idx: u32) -> bool; +} + +impl<T> WasmModuleResources for &'_ T +where + T: ?Sized + WasmModuleResources, +{ + type FuncType = T::FuncType; + + fn table_at(&self, at: u32) -> Option<TableType> { + T::table_at(self, at) + } + fn memory_at(&self, at: u32) -> Option<MemoryType> { + T::memory_at(self, at) + } + fn tag_at(&self, at: u32) -> Option<&Self::FuncType> { + T::tag_at(self, at) + } + fn global_at(&self, at: u32) -> Option<GlobalType> { + T::global_at(self, at) + } + fn func_type_at(&self, at: u32) -> Option<&Self::FuncType> { + T::func_type_at(self, at) + } + fn type_index_of_function(&self, func_idx: u32) -> Option<u32> { + T::type_index_of_function(self, func_idx) + } + fn type_of_function(&self, func_idx: u32) -> Option<&Self::FuncType> { + T::type_of_function(self, func_idx) + } + fn check_value_type( + &self, + t: ValType, + features: &WasmFeatures, + offset: usize, + ) -> Result<(), BinaryReaderError> { + T::check_value_type(self, t, features, offset) + } + fn element_type_at(&self, at: u32) -> Option<RefType> { + T::element_type_at(self, at) + } + fn matches(&self, t1: ValType, t2: ValType) -> bool { + T::matches(self, t1, t2) + } + + fn element_count(&self) -> u32 { + T::element_count(self) + } + fn data_count(&self) -> Option<u32> { + T::data_count(self) + } + fn is_function_referenced(&self, idx: u32) -> bool { + T::is_function_referenced(self, idx) + } +} + +impl<T> WasmModuleResources for std::sync::Arc<T> +where + T: WasmModuleResources, +{ + type FuncType = T::FuncType; + + fn table_at(&self, at: u32) -> Option<TableType> { + T::table_at(self, at) + } + + fn memory_at(&self, at: u32) -> Option<MemoryType> { + T::memory_at(self, at) + } + + fn tag_at(&self, at: u32) -> Option<&Self::FuncType> { + T::tag_at(self, at) + } + + fn global_at(&self, at: u32) -> Option<GlobalType> { + T::global_at(self, at) + } + + fn func_type_at(&self, type_idx: u32) -> Option<&Self::FuncType> { + T::func_type_at(self, type_idx) + } + + fn type_index_of_function(&self, func_idx: u32) -> Option<u32> { + T::type_index_of_function(self, func_idx) + } + + fn type_of_function(&self, func_idx: u32) -> Option<&Self::FuncType> { + T::type_of_function(self, func_idx) + } + + fn check_value_type( + &self, + t: ValType, + features: &WasmFeatures, + offset: usize, + ) -> Result<(), BinaryReaderError> { + T::check_value_type(self, t, features, offset) + } + + fn element_type_at(&self, at: u32) -> Option<RefType> { + T::element_type_at(self, at) + } + + fn matches(&self, t1: ValType, t2: ValType) -> bool { + T::matches(self, t1, t2) + } + + fn element_count(&self) -> u32 { + T::element_count(self) + } + + fn data_count(&self) -> Option<u32> { + T::data_count(self) + } + + fn is_function_referenced(&self, idx: u32) -> bool { + T::is_function_referenced(self, idx) + } +} + +impl WasmFuncType for FuncType { + fn len_inputs(&self) -> usize { + self.params().len() + } + + fn len_outputs(&self) -> usize { + self.results().len() + } + + fn input_at(&self, at: u32) -> Option<ValType> { + self.params().get(at as usize).copied() + } + + fn output_at(&self, at: u32) -> Option<ValType> { + self.results().get(at as usize).copied() + } +} diff --git a/third_party/rust/wasmparser/src/validator.rs b/third_party/rust/wasmparser/src/validator.rs new file mode 100644 index 0000000000..04c207130b --- /dev/null +++ b/third_party/rust/wasmparser/src/validator.rs @@ -0,0 +1,1514 @@ +/* Copyright 2018 Mozilla Foundation + * + * 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. + */ + +use crate::{ + limits::*, BinaryReaderError, Encoding, FromReader, FunctionBody, HeapType, Parser, Payload, + Result, SectionLimited, ValType, WASM_COMPONENT_VERSION, WASM_MODULE_VERSION, +}; +use std::mem; +use std::ops::Range; +use std::sync::Arc; + +/// Test whether the given buffer contains a valid WebAssembly module or component, +/// analogous to [`WebAssembly.validate`][js] in the JS API. +/// +/// This functions requires the bytes to validate are entirely resident in memory. +/// Additionally this validates the given bytes with the default set of WebAssembly +/// features implemented by `wasmparser`. +/// +/// For more fine-tuned control over validation it's recommended to review the +/// documentation of [`Validator`]. +/// +/// Upon success, the type information for the top-level module or component will +/// be returned. +/// +/// [js]: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/WebAssembly/validate +pub fn validate(bytes: &[u8]) -> Result<Types> { + Validator::new().validate_all(bytes) +} + +#[test] +fn test_validate() { + assert!(validate(&[0x0, 0x61, 0x73, 0x6d, 0x1, 0x0, 0x0, 0x0]).is_ok()); + assert!(validate(&[0x0, 0x61, 0x73, 0x6d, 0x2, 0x0, 0x0, 0x0]).is_err()); +} + +mod component; +mod core; +mod func; +mod operators; +pub mod types; + +use self::component::*; +pub use self::core::ValidatorResources; +use self::core::*; +use self::types::{TypeAlloc, Types, TypesRef}; +pub use func::{FuncToValidate, FuncValidator, FuncValidatorAllocations}; +pub use operators::{Frame, FrameKind}; + +fn check_max(cur_len: usize, amt_added: u32, max: usize, desc: &str, offset: usize) -> Result<()> { + if max + .checked_sub(cur_len) + .and_then(|amt| amt.checked_sub(amt_added as usize)) + .is_none() + { + if max == 1 { + bail!(offset, "multiple {desc}"); + } + + bail!(offset, "{desc} count exceeds limit of {max}"); + } + + Ok(()) +} + +fn combine_type_sizes(a: u32, b: u32, offset: usize) -> Result<u32> { + match a.checked_add(b) { + Some(sum) if sum < MAX_WASM_TYPE_SIZE => Ok(sum), + _ => Err(format_err!( + offset, + "effective type size exceeds the limit of {MAX_WASM_TYPE_SIZE}", + )), + } +} + +/// Validator for a WebAssembly binary module or component. +/// +/// This structure encapsulates state necessary to validate a WebAssembly +/// binary. This implements validation as defined by the [core +/// specification][core]. A `Validator` is designed, like +/// [`Parser`], to accept incremental input over time. +/// Additionally a `Validator` is also designed for parallel validation of +/// functions as they are received. +/// +/// It's expected that you'll be using a [`Parser`] in tandem with a +/// `Validator`. As each [`Payload`](crate::Payload) is received from a +/// [`Parser`] you'll pass it into a `Validator` to test the validity of the +/// payload. Note that all payloads received from a [`Parser`] are expected to +/// be passed to a [`Validator`]. For example if you receive +/// [`Payload::TypeSection`](crate::Payload) you'll call +/// [`Validator::type_section`] to validate this. +/// +/// The design of [`Validator`] is intended that you'll interleave, in your own +/// application's processing, calls to validation. Each variant, after it's +/// received, will be validated and then your application would proceed as +/// usual. At all times, however, you'll have access to the [`Validator`] and +/// the validation context up to that point. This enables applications to check +/// the types of functions and learn how many globals there are, for example. +/// +/// [core]: https://webassembly.github.io/spec/core/valid/index.html +#[derive(Default)] +pub struct Validator { + /// The current state of the validator. + state: State, + + /// The global type space used by the validator and any sub-validators. + types: TypeAlloc, + + /// The module state when parsing a WebAssembly module. + module: Option<ModuleState>, + + /// With the component model enabled, this stores the pushed component states. + /// The top of the stack is the current component state. + components: Vec<ComponentState>, + + /// Enabled WebAssembly feature flags, dictating what's valid and what + /// isn't. + features: WasmFeatures, +} + +#[derive(Debug, Clone, Copy, Eq, PartialEq)] +enum State { + /// A header has not yet been parsed. + /// + /// The value is the expected encoding for the header. + Unparsed(Option<Encoding>), + /// A module header has been parsed. + /// + /// The associated module state is available via [`Validator::module`]. + Module, + /// A component header has been parsed. + /// + /// The associated component state exists at the top of the + /// validator's [`Validator::components`] stack. + Component, + /// The parse has completed and no more data is expected. + End, +} + +impl State { + fn ensure_parsable(&self, offset: usize) -> Result<()> { + match self { + Self::Module | Self::Component => Ok(()), + Self::Unparsed(_) => Err(BinaryReaderError::new( + "unexpected section before header was parsed", + offset, + )), + Self::End => Err(BinaryReaderError::new( + "unexpected section after parsing has completed", + offset, + )), + } + } + + fn ensure_module(&self, section: &str, offset: usize) -> Result<()> { + self.ensure_parsable(offset)?; + + match self { + Self::Module => Ok(()), + Self::Component => Err(format_err!( + offset, + "unexpected module {section} section while parsing a component", + )), + _ => unreachable!(), + } + } + + fn ensure_component(&self, section: &str, offset: usize) -> Result<()> { + self.ensure_parsable(offset)?; + + match self { + Self::Component => Ok(()), + Self::Module => Err(format_err!( + offset, + "unexpected component {section} section while parsing a module", + )), + _ => unreachable!(), + } + } +} + +impl Default for State { + fn default() -> Self { + Self::Unparsed(None) + } +} + +/// Flags for features that are enabled for validation. +#[derive(Hash, Debug, Copy, Clone)] +pub struct WasmFeatures { + /// The WebAssembly `mutable-global` proposal (enabled by default) + pub mutable_global: bool, + /// The WebAssembly `nontrapping-float-to-int-conversions` proposal (enabled by default) + pub saturating_float_to_int: bool, + /// The WebAssembly `sign-extension-ops` proposal (enabled by default) + pub sign_extension: bool, + /// The WebAssembly reference types proposal (enabled by default) + pub reference_types: bool, + /// The WebAssembly multi-value proposal (enabled by default) + pub multi_value: bool, + /// The WebAssembly bulk memory operations proposal (enabled by default) + pub bulk_memory: bool, + /// The WebAssembly SIMD proposal (enabled by default) + pub simd: bool, + /// The WebAssembly Relaxed SIMD proposal + pub relaxed_simd: bool, + /// The WebAssembly threads proposal + pub threads: bool, + /// The WebAssembly tail-call proposal + pub tail_call: bool, + /// Whether or not floating-point instructions are enabled. + /// + /// This is enabled by default can be used to disallow floating-point + /// operators and types. + /// + /// This does not correspond to a WebAssembly proposal but is instead + /// intended for embeddings which have stricter-than-usual requirements + /// about execution. Floats in WebAssembly can have different NaN patterns + /// across hosts which can lead to host-dependent execution which some + /// runtimes may not desire. + pub floats: bool, + /// The WebAssembly multi memory proposal + pub multi_memory: bool, + /// The WebAssembly exception handling proposal + pub exceptions: bool, + /// The WebAssembly memory64 proposal + pub memory64: bool, + /// The WebAssembly extended_const proposal + pub extended_const: bool, + /// The WebAssembly component model proposal. + pub component_model: bool, + /// The WebAssembly typed function references proposal + pub function_references: bool, + /// The WebAssembly memory control proposal + pub memory_control: bool, +} + +impl WasmFeatures { + /// NOTE: This only checks that the value type corresponds to the feature set!! + /// + /// To check that reference types are valid, we need access to the module + /// types. Use module.check_value_type. + pub(crate) fn check_value_type(&self, ty: ValType) -> Result<(), &'static str> { + match ty { + ValType::I32 | ValType::I64 => Ok(()), + ValType::F32 | ValType::F64 => { + if self.floats { + Ok(()) + } else { + Err("floating-point support is disabled") + } + } + ValType::Ref(r) => { + if self.reference_types { + if !self.function_references { + match (r.heap_type, r.nullable) { + (_, false) => { + Err("function references required for non-nullable types") + } + (HeapType::TypedFunc(_), _) => { + Err("function references required for index reference types") + } + _ => Ok(()), + } + } else { + Ok(()) + } + } else { + Err("reference types support is not enabled") + } + } + ValType::V128 => { + if self.simd { + Ok(()) + } else { + Err("SIMD support is not enabled") + } + } + } + } +} + +impl Default for WasmFeatures { + fn default() -> WasmFeatures { + WasmFeatures { + // Off-by-default features. + relaxed_simd: false, + threads: false, + multi_memory: false, + exceptions: false, + memory64: false, + extended_const: false, + component_model: false, + function_references: false, + memory_control: false, + + // On-by-default features (phase 4 or greater). + mutable_global: true, + saturating_float_to_int: true, + sign_extension: true, + bulk_memory: true, + multi_value: true, + reference_types: true, + tail_call: true, + simd: true, + floats: true, + } + } +} + +/// Possible return values from [`Validator::payload`]. +#[allow(clippy::large_enum_variant)] +pub enum ValidPayload<'a> { + /// The payload validated, no further action need be taken. + Ok, + /// The payload validated, but it started a nested module or component. + /// + /// This result indicates that the specified parser should be used instead + /// of the currently-used parser until this returned one ends. + Parser(Parser), + /// A function was found to be validate. + Func(FuncToValidate<ValidatorResources>, FunctionBody<'a>), + /// The end payload was validated and the types known to the validator + /// are provided. + End(Types), +} + +impl Validator { + /// Creates a new [`Validator`] ready to validate a WebAssembly module + /// or component. + /// + /// The new validator will receive payloads parsed from + /// [`Parser`], and expects the first payload received to be + /// the version header from the parser. + pub fn new() -> Validator { + Validator::default() + } + + /// Creates a new [`Validator`] which has the specified set of wasm + /// features activated for validation. + /// + /// This function is the same as [`Validator::new`] except it also allows + /// you to customize the active wasm features in use for validation. This + /// can allow enabling experimental proposals or also turning off + /// on-by-default wasm proposals. + pub fn new_with_features(features: WasmFeatures) -> Validator { + let mut ret = Validator::new(); + ret.features = features; + ret + } + + /// Returns the wasm features used for this validator. + pub fn features(&self) -> &WasmFeatures { + &self.features + } + + /// Validates an entire in-memory module or component with this validator. + /// + /// This function will internally create a [`Parser`] to parse the `bytes` + /// provided. The entire module or component specified by `bytes` will be + /// parsed and validated. + /// + /// Upon success, the type information for the top-level module or component + /// will be returned. + pub fn validate_all(&mut self, bytes: &[u8]) -> Result<Types> { + let mut functions_to_validate = Vec::new(); + let mut last_types = None; + for payload in Parser::new(0).parse_all(bytes) { + match self.payload(&payload?)? { + ValidPayload::Func(a, b) => { + functions_to_validate.push((a, b)); + } + ValidPayload::End(types) => { + // Only the last (top-level) type information will be returned + last_types = Some(types); + } + _ => {} + } + } + + let mut allocs = FuncValidatorAllocations::default(); + for (func, body) in functions_to_validate { + let mut validator = func.into_validator(allocs); + validator.validate(&body)?; + allocs = validator.into_allocations(); + } + + Ok(last_types.unwrap()) + } + + /// Gets the types known by the validator so far within the + /// module/component `level` modules/components up from the + /// module/component currently being parsed. + /// + /// For instance, calling `validator.types(0)` will get the types of the + /// module/component currently being parsed, and `validator.types(1)` will + /// get the types of the component containing that module/component. + /// + /// Returns `None` if there is no module/component that many levels up. + pub fn types(&self, mut level: usize) -> Option<TypesRef> { + if let Some(module) = &self.module { + if level == 0 { + return Some(TypesRef::from_module(&self.types, &module.module)); + } else { + level -= 1; + } + } + + self.components + .iter() + .nth_back(level) + .map(|component| TypesRef::from_component(&self.types, component)) + } + + /// Convenience function to validate a single [`Payload`]. + /// + /// This function is intended to be used as a convenience. It will + /// internally perform any validation necessary to validate the [`Payload`] + /// provided. The convenience part is that you're likely already going to + /// be matching on [`Payload`] in your application, at which point it's more + /// appropriate to call the individual methods on [`Validator`] per-variant + /// in [`Payload`], such as [`Validator::type_section`]. + /// + /// This function returns a [`ValidPayload`] variant on success, indicating + /// one of a few possible actions that need to be taken after a payload is + /// validated. For example function contents are not validated here, they're + /// returned through [`ValidPayload`] for validation by the caller. + pub fn payload<'a>(&mut self, payload: &Payload<'a>) -> Result<ValidPayload<'a>> { + use crate::Payload::*; + match payload { + Version { + num, + encoding, + range, + } => self.version(*num, *encoding, range)?, + + // Module sections + TypeSection(s) => self.type_section(s)?, + ImportSection(s) => self.import_section(s)?, + FunctionSection(s) => self.function_section(s)?, + TableSection(s) => self.table_section(s)?, + MemorySection(s) => self.memory_section(s)?, + TagSection(s) => self.tag_section(s)?, + GlobalSection(s) => self.global_section(s)?, + ExportSection(s) => self.export_section(s)?, + StartSection { func, range } => self.start_section(*func, range)?, + ElementSection(s) => self.element_section(s)?, + DataCountSection { count, range } => self.data_count_section(*count, range)?, + CodeSectionStart { + count, + range, + size: _, + } => self.code_section_start(*count, range)?, + CodeSectionEntry(body) => { + let func_validator = self.code_section_entry(body)?; + return Ok(ValidPayload::Func(func_validator, body.clone())); + } + DataSection(s) => self.data_section(s)?, + + // Component sections + ModuleSection { parser, range, .. } => { + self.module_section(range)?; + return Ok(ValidPayload::Parser(parser.clone())); + } + InstanceSection(s) => self.instance_section(s)?, + CoreTypeSection(s) => self.core_type_section(s)?, + ComponentSection { parser, range, .. } => { + self.component_section(range)?; + return Ok(ValidPayload::Parser(parser.clone())); + } + ComponentInstanceSection(s) => self.component_instance_section(s)?, + ComponentAliasSection(s) => self.component_alias_section(s)?, + ComponentTypeSection(s) => self.component_type_section(s)?, + ComponentCanonicalSection(s) => self.component_canonical_section(s)?, + ComponentStartSection { start, range } => self.component_start_section(start, range)?, + ComponentImportSection(s) => self.component_import_section(s)?, + ComponentExportSection(s) => self.component_export_section(s)?, + + End(offset) => return Ok(ValidPayload::End(self.end(*offset)?)), + + CustomSection { .. } => {} // no validation for custom sections + UnknownSection { id, range, .. } => self.unknown_section(*id, range)?, + } + Ok(ValidPayload::Ok) + } + + /// Validates [`Payload::Version`](crate::Payload). + pub fn version(&mut self, num: u16, encoding: Encoding, range: &Range<usize>) -> Result<()> { + match &self.state { + State::Unparsed(expected) => { + if let Some(expected) = expected { + if *expected != encoding { + bail!( + range.start, + "expected a version header for a {}", + match expected { + Encoding::Module => "module", + Encoding::Component => "component", + } + ); + } + } + } + _ => { + return Err(BinaryReaderError::new( + "wasm version header out of order", + range.start, + )) + } + } + + self.state = match encoding { + Encoding::Module => { + if num == WASM_MODULE_VERSION { + assert!(self.module.is_none()); + self.module = Some(ModuleState::default()); + State::Module + } else { + bail!(range.start, "unknown binary version: {num:#x}"); + } + } + Encoding::Component => { + if !self.features.component_model { + bail!( + range.start, + "unknown binary version and encoding combination: {num:#x} and 0x1, \ + note: encoded as a component but the WebAssembly component model feature \ + is not enabled - enable the feature to allow component validation", + ); + } + if num == WASM_COMPONENT_VERSION { + self.components.push(ComponentState::default()); + State::Component + } else if num < WASM_COMPONENT_VERSION { + bail!(range.start, "unsupported component version: {num:#x}"); + } else { + bail!(range.start, "unknown component version: {num:#x}"); + } + } + }; + + Ok(()) + } + + /// Validates [`Payload::TypeSection`](crate::Payload). + pub fn type_section(&mut self, section: &crate::TypeSectionReader<'_>) -> Result<()> { + self.process_module_section( + Order::Type, + section, + "type", + |state, _, types, count, offset| { + check_max( + state.module.types.len(), + count, + MAX_WASM_TYPES, + "types", + offset, + )?; + types.reserve(count as usize); + state.module.assert_mut().types.reserve(count as usize); + Ok(()) + }, + |state, features, types, def, offset| { + state + .module + .assert_mut() + .add_type(def, features, types, offset, false /* checked above */) + }, + ) + } + + /// Validates [`Payload::ImportSection`](crate::Payload). + /// + /// This method should only be called when parsing a module. + pub fn import_section(&mut self, section: &crate::ImportSectionReader<'_>) -> Result<()> { + self.process_module_section( + Order::Import, + section, + "import", + |_, _, _, _, _| Ok(()), // add_import will check limits + |state, features, types, import, offset| { + state + .module + .assert_mut() + .add_import(import, features, types, offset) + }, + ) + } + + /// Validates [`Payload::FunctionSection`](crate::Payload). + /// + /// This method should only be called when parsing a module. + pub fn function_section(&mut self, section: &crate::FunctionSectionReader<'_>) -> Result<()> { + self.process_module_section( + Order::Function, + section, + "function", + |state, _, _, count, offset| { + check_max( + state.module.functions.len(), + count, + MAX_WASM_FUNCTIONS, + "functions", + offset, + )?; + state.module.assert_mut().functions.reserve(count as usize); + debug_assert!(state.expected_code_bodies.is_none()); + state.expected_code_bodies = Some(count); + Ok(()) + }, + |state, _, types, ty, offset| state.module.assert_mut().add_function(ty, types, offset), + ) + } + + /// Validates [`Payload::TableSection`](crate::Payload). + /// + /// This method should only be called when parsing a module. + pub fn table_section(&mut self, section: &crate::TableSectionReader<'_>) -> Result<()> { + let features = self.features; + self.process_module_section( + Order::Table, + section, + "table", + |state, _, _, count, offset| { + check_max( + state.module.tables.len(), + count, + state.module.max_tables(&features), + "tables", + offset, + )?; + state.module.assert_mut().tables.reserve(count as usize); + Ok(()) + }, + |state, features, types, table, offset| state.add_table(table, features, types, offset), + ) + } + + /// Validates [`Payload::MemorySection`](crate::Payload). + /// + /// This method should only be called when parsing a module. + pub fn memory_section(&mut self, section: &crate::MemorySectionReader<'_>) -> Result<()> { + self.process_module_section( + Order::Memory, + section, + "memory", + |state, features, _, count, offset| { + check_max( + state.module.memories.len(), + count, + state.module.max_memories(features), + "memories", + offset, + )?; + state.module.assert_mut().memories.reserve(count as usize); + Ok(()) + }, + |state, features, _, ty, offset| { + state.module.assert_mut().add_memory(ty, features, offset) + }, + ) + } + + /// Validates [`Payload::TagSection`](crate::Payload). + /// + /// This method should only be called when parsing a module. + pub fn tag_section(&mut self, section: &crate::TagSectionReader<'_>) -> Result<()> { + if !self.features.exceptions { + return Err(BinaryReaderError::new( + "exceptions proposal not enabled", + section.range().start, + )); + } + + self.process_module_section( + Order::Tag, + section, + "tag", + |state, _, _, count, offset| { + check_max( + state.module.tags.len(), + count, + MAX_WASM_TAGS, + "tags", + offset, + )?; + state.module.assert_mut().tags.reserve(count as usize); + Ok(()) + }, + |state, features, types, ty, offset| { + state + .module + .assert_mut() + .add_tag(ty, features, types, offset) + }, + ) + } + + /// Validates [`Payload::GlobalSection`](crate::Payload). + /// + /// This method should only be called when parsing a module. + pub fn global_section(&mut self, section: &crate::GlobalSectionReader<'_>) -> Result<()> { + self.process_module_section( + Order::Global, + section, + "global", + |state, _, _, count, offset| { + check_max( + state.module.globals.len(), + count, + MAX_WASM_GLOBALS, + "globals", + offset, + )?; + state.module.assert_mut().globals.reserve(count as usize); + Ok(()) + }, + |state, features, types, global, offset| { + state.add_global(global, features, types, offset) + }, + ) + } + + /// Validates [`Payload::ExportSection`](crate::Payload). + /// + /// This method should only be called when parsing a module. + pub fn export_section(&mut self, section: &crate::ExportSectionReader<'_>) -> Result<()> { + self.process_module_section( + Order::Export, + section, + "export", + |state, _, _, count, offset| { + check_max( + state.module.exports.len(), + count, + MAX_WASM_EXPORTS, + "exports", + offset, + )?; + state.module.assert_mut().exports.reserve(count as usize); + Ok(()) + }, + |state, features, _, e, offset| { + let state = state.module.assert_mut(); + let ty = state.export_to_entity_type(&e, offset)?; + state.add_export(e.name, ty, features, offset, false /* checked above */) + }, + ) + } + + /// Validates [`Payload::StartSection`](crate::Payload). + /// + /// This method should only be called when parsing a module. + pub fn start_section(&mut self, func: u32, range: &Range<usize>) -> Result<()> { + let offset = range.start; + self.state.ensure_module("start", offset)?; + let state = self.module.as_mut().unwrap(); + state.update_order(Order::Start, offset)?; + + let ty = state.module.get_func_type(func, &self.types, offset)?; + if !ty.params().is_empty() || !ty.results().is_empty() { + return Err(BinaryReaderError::new( + "invalid start function type", + offset, + )); + } + + Ok(()) + } + + /// Validates [`Payload::ElementSection`](crate::Payload). + /// + /// This method should only be called when parsing a module. + pub fn element_section(&mut self, section: &crate::ElementSectionReader<'_>) -> Result<()> { + self.process_module_section( + Order::Element, + section, + "element", + |state, _, _, count, offset| { + check_max( + state.module.element_types.len(), + count, + MAX_WASM_ELEMENT_SEGMENTS, + "element segments", + offset, + )?; + state + .module + .assert_mut() + .element_types + .reserve(count as usize); + Ok(()) + }, + |state, features, types, e, offset| { + state.add_element_segment(e, features, types, offset) + }, + ) + } + + /// Validates [`Payload::DataCountSection`](crate::Payload). + /// + /// This method should only be called when parsing a module. + pub fn data_count_section(&mut self, count: u32, range: &Range<usize>) -> Result<()> { + let offset = range.start; + self.state.ensure_module("data count", offset)?; + + let state = self.module.as_mut().unwrap(); + state.update_order(Order::DataCount, offset)?; + + if count > MAX_WASM_DATA_SEGMENTS as u32 { + return Err(BinaryReaderError::new( + "data count section specifies too many data segments", + offset, + )); + } + + state.module.assert_mut().data_count = Some(count); + Ok(()) + } + + /// Validates [`Payload::CodeSectionStart`](crate::Payload). + /// + /// This method should only be called when parsing a module. + pub fn code_section_start(&mut self, count: u32, range: &Range<usize>) -> Result<()> { + let offset = range.start; + self.state.ensure_module("code", offset)?; + + let state = self.module.as_mut().unwrap(); + state.update_order(Order::Code, offset)?; + + match state.expected_code_bodies.take() { + Some(n) if n == count => {} + Some(_) => { + return Err(BinaryReaderError::new( + "function and code section have inconsistent lengths", + offset, + )); + } + // empty code sections are allowed even if the function section is + // missing + None if count == 0 => {} + None => { + return Err(BinaryReaderError::new( + "code section without function section", + offset, + )) + } + } + + // Take a snapshot of the types when we start the code section. + state.module.assert_mut().snapshot = Some(Arc::new(self.types.commit())); + + Ok(()) + } + + /// Validates [`Payload::CodeSectionEntry`](crate::Payload). + /// + /// This function will prepare a [`FuncToValidate`] which can be used to + /// create a [`FuncValidator`] to validate the function. The function body + /// provided will not be parsed or validated by this function. + /// + /// Note that the returned [`FuncToValidate`] is "connected" to this + /// [`Validator`] in that it uses the internal context of this validator for + /// validating the function. The [`FuncToValidate`] can be sent to another + /// thread, for example, to offload actual processing of functions + /// elsewhere. + /// + /// This method should only be called when parsing a module. + pub fn code_section_entry( + &mut self, + body: &crate::FunctionBody, + ) -> Result<FuncToValidate<ValidatorResources>> { + let offset = body.range().start; + self.state.ensure_module("code", offset)?; + + let state = self.module.as_mut().unwrap(); + + let (index, ty) = state.next_code_index_and_type(offset)?; + Ok(FuncToValidate::new( + index, + ty, + ValidatorResources(state.module.arc().clone()), + &self.features, + )) + } + + /// Validates [`Payload::DataSection`](crate::Payload). + /// + /// This method should only be called when parsing a module. + pub fn data_section(&mut self, section: &crate::DataSectionReader<'_>) -> Result<()> { + self.process_module_section( + Order::Data, + section, + "data", + |state, _, _, count, offset| { + state.data_segment_count = count; + check_max(0, count, MAX_WASM_DATA_SEGMENTS, "data segments", offset) + }, + |state, features, types, d, offset| state.add_data_segment(d, features, types, offset), + ) + } + + /// Validates [`Payload::ModuleSection`](crate::Payload). + /// + /// This method should only be called when parsing a component. + pub fn module_section(&mut self, range: &Range<usize>) -> Result<()> { + self.state.ensure_component("module", range.start)?; + + let current = self.components.last_mut().unwrap(); + check_max( + current.core_modules.len(), + 1, + MAX_WASM_MODULES, + "modules", + range.start, + )?; + + match mem::replace(&mut self.state, State::Unparsed(Some(Encoding::Module))) { + State::Component => {} + _ => unreachable!(), + } + + Ok(()) + } + + /// Validates [`Payload::InstanceSection`](crate::Payload). + /// + /// This method should only be called when parsing a component. + pub fn instance_section(&mut self, section: &crate::InstanceSectionReader) -> Result<()> { + self.process_component_section( + section, + "core instance", + |components, _, count, offset| { + let current = components.last_mut().unwrap(); + check_max( + current.instance_count(), + count, + MAX_WASM_INSTANCES, + "instances", + offset, + )?; + current.core_instances.reserve(count as usize); + Ok(()) + }, + |components, types, _, instance, offset| { + components + .last_mut() + .unwrap() + .add_core_instance(instance, types, offset) + }, + ) + } + + /// Validates [`Payload::CoreTypeSection`](crate::Payload). + /// + /// This method should only be called when parsing a component. + pub fn core_type_section(&mut self, section: &crate::CoreTypeSectionReader<'_>) -> Result<()> { + self.process_component_section( + section, + "core type", + |components, types, count, offset| { + let current = components.last_mut().unwrap(); + check_max(current.type_count(), count, MAX_WASM_TYPES, "types", offset)?; + types.reserve(count as usize); + current.core_types.reserve(count as usize); + Ok(()) + }, + |components, types, features, ty, offset| { + ComponentState::add_core_type( + components, ty, features, types, offset, false, /* checked above */ + ) + }, + ) + } + + /// Validates [`Payload::ComponentSection`](crate::Payload). + /// + /// This method should only be called when parsing a component. + pub fn component_section(&mut self, range: &Range<usize>) -> Result<()> { + self.state.ensure_component("component", range.start)?; + + let current = self.components.last_mut().unwrap(); + check_max( + current.components.len(), + 1, + MAX_WASM_COMPONENTS, + "components", + range.start, + )?; + + match mem::replace(&mut self.state, State::Unparsed(Some(Encoding::Component))) { + State::Component => {} + _ => unreachable!(), + } + + Ok(()) + } + + /// Validates [`Payload::ComponentInstanceSection`](crate::Payload). + /// + /// This method should only be called when parsing a component. + pub fn component_instance_section( + &mut self, + section: &crate::ComponentInstanceSectionReader, + ) -> Result<()> { + self.process_component_section( + section, + "instance", + |components, _, count, offset| { + let current = components.last_mut().unwrap(); + check_max( + current.instance_count(), + count, + MAX_WASM_INSTANCES, + "instances", + offset, + )?; + current.instances.reserve(count as usize); + Ok(()) + }, + |components, types, _, instance, offset| { + components + .last_mut() + .unwrap() + .add_instance(instance, types, offset) + }, + ) + } + + /// Validates [`Payload::ComponentAliasSection`](crate::Payload). + /// + /// This method should only be called when parsing a component. + pub fn component_alias_section( + &mut self, + section: &crate::ComponentAliasSectionReader<'_>, + ) -> Result<()> { + self.process_component_section( + section, + "alias", + |_, _, _, _| Ok(()), // maximums checked via `add_alias` + |components, types, _, alias, offset| -> Result<(), BinaryReaderError> { + ComponentState::add_alias(components, alias, types, offset) + }, + ) + } + + /// Validates [`Payload::ComponentTypeSection`](crate::Payload). + /// + /// This method should only be called when parsing a component. + pub fn component_type_section( + &mut self, + section: &crate::ComponentTypeSectionReader, + ) -> Result<()> { + self.process_component_section( + section, + "type", + |components, types, count, offset| { + let current = components.last_mut().unwrap(); + check_max(current.type_count(), count, MAX_WASM_TYPES, "types", offset)?; + types.reserve(count as usize); + current.types.reserve(count as usize); + Ok(()) + }, + |components, types, features, ty, offset| { + ComponentState::add_type( + components, ty, features, types, offset, false, /* checked above */ + ) + }, + ) + } + + /// Validates [`Payload::ComponentCanonicalSection`](crate::Payload). + /// + /// This method should only be called when parsing a component. + pub fn component_canonical_section( + &mut self, + section: &crate::ComponentCanonicalSectionReader, + ) -> Result<()> { + self.process_component_section( + section, + "function", + |components, _, count, offset| { + let current = components.last_mut().unwrap(); + check_max( + current.function_count(), + count, + MAX_WASM_FUNCTIONS, + "functions", + offset, + )?; + current.funcs.reserve(count as usize); + Ok(()) + }, + |components, types, _, func, offset| { + let current = components.last_mut().unwrap(); + match func { + crate::CanonicalFunction::Lift { + core_func_index, + type_index, + options, + } => current.lift_function( + core_func_index, + type_index, + options.into_vec(), + types, + offset, + ), + crate::CanonicalFunction::Lower { + func_index, + options, + } => current.lower_function(func_index, options.into_vec(), types, offset), + } + }, + ) + } + + /// Validates [`Payload::ComponentStartSection`](crate::Payload). + /// + /// This method should only be called when parsing a component. + pub fn component_start_section( + &mut self, + f: &crate::ComponentStartFunction, + range: &Range<usize>, + ) -> Result<()> { + self.state.ensure_component("start", range.start)?; + + // let mut section = section.clone(); + // let f = section.read()?; + + // if !section.eof() { + // return Err(BinaryReaderError::new( + // "trailing data at the end of the start section", + // section.original_position(), + // )); + // } + + self.components.last_mut().unwrap().add_start( + f.func_index, + &f.arguments, + f.results, + &self.types, + range.start, + ) + } + + /// Validates [`Payload::ComponentImportSection`](crate::Payload). + /// + /// This method should only be called when parsing a component. + pub fn component_import_section( + &mut self, + section: &crate::ComponentImportSectionReader, + ) -> Result<()> { + self.process_component_section( + section, + "import", + |_, _, _, _| Ok(()), // add_import will check limits + |components, types, _, import, offset| { + components + .last_mut() + .unwrap() + .add_import(import, types, offset) + }, + ) + } + + /// Validates [`Payload::ComponentExportSection`](crate::Payload). + /// + /// This method should only be called when parsing a component. + pub fn component_export_section( + &mut self, + section: &crate::ComponentExportSectionReader, + ) -> Result<()> { + self.process_component_section( + section, + "export", + |components, _, count, offset| { + let current = components.last_mut().unwrap(); + check_max( + current.externs.len(), + count, + MAX_WASM_EXPORTS, + "imports and exports", + offset, + )?; + current.externs.reserve(count as usize); + Ok(()) + }, + |components, types, _, export, offset| { + let current = components.last_mut().unwrap(); + let ty = current.export_to_entity_type(&export, types, offset)?; + current.add_export( + export.name, + export.url, + ty, + offset, + false, /* checked above */ + ) + }, + ) + } + + /// Validates [`Payload::UnknownSection`](crate::Payload). + /// + /// Currently always returns an error. + pub fn unknown_section(&mut self, id: u8, range: &Range<usize>) -> Result<()> { + Err(format_err!(range.start, "malformed section id: {id}")) + } + + /// Validates [`Payload::End`](crate::Payload). + /// + /// Returns the types known to the validator for the module or component. + pub fn end(&mut self, offset: usize) -> Result<Types> { + match std::mem::replace(&mut self.state, State::End) { + State::Unparsed(_) => Err(BinaryReaderError::new( + "cannot call `end` before a header has been parsed", + offset, + )), + State::End => Err(BinaryReaderError::new( + "cannot call `end` after parsing has completed", + offset, + )), + State::Module => { + let mut state = self.module.take().unwrap(); + state.validate_end(offset)?; + + // If there's a parent component, we'll add a module to the parent state + // and continue to validate the component + if let Some(parent) = self.components.last_mut() { + parent.add_core_module(&state.module, &mut self.types, offset)?; + self.state = State::Component; + } + + Ok(Types::from_module( + self.types.commit(), + state.module.arc().clone(), + )) + } + State::Component => { + let mut component = self.components.pop().unwrap(); + + // Validate that all values were used for the component + if let Some(index) = component.values.iter().position(|(_, used)| !*used) { + return Err( + format_err!(offset,"value index {index} was not used as part of an instantiation, start function, or export" + ) + ); + } + + // If there's a parent component, pop the stack, add it to the parent, + // and continue to validate the component + if let Some(parent) = self.components.last_mut() { + parent.add_component(&mut component, &mut self.types); + self.state = State::Component; + } + + Ok(Types::from_component(self.types.commit(), component)) + } + } + } + + fn process_module_section<'a, T>( + &mut self, + order: Order, + section: &SectionLimited<'a, T>, + name: &str, + validate_section: impl FnOnce( + &mut ModuleState, + &WasmFeatures, + &mut TypeAlloc, + u32, + usize, + ) -> Result<()>, + mut validate_item: impl FnMut( + &mut ModuleState, + &WasmFeatures, + &mut TypeAlloc, + T, + usize, + ) -> Result<()>, + ) -> Result<()> + where + T: FromReader<'a>, + { + let offset = section.range().start; + self.state.ensure_module(name, offset)?; + + let state = self.module.as_mut().unwrap(); + state.update_order(order, offset)?; + + validate_section( + state, + &self.features, + &mut self.types, + section.count(), + offset, + )?; + + for item in section.clone().into_iter_with_offsets() { + let (offset, item) = item?; + validate_item(state, &self.features, &mut self.types, item, offset)?; + } + + Ok(()) + } + + fn process_component_section<'a, T>( + &mut self, + section: &SectionLimited<'a, T>, + name: &str, + validate_section: impl FnOnce( + &mut Vec<ComponentState>, + &mut TypeAlloc, + u32, + usize, + ) -> Result<()>, + mut validate_item: impl FnMut( + &mut Vec<ComponentState>, + &mut TypeAlloc, + &WasmFeatures, + T, + usize, + ) -> Result<()>, + ) -> Result<()> + where + T: FromReader<'a>, + { + let offset = section.range().start; + + if !self.features.component_model { + return Err(BinaryReaderError::new( + "component model feature is not enabled", + offset, + )); + } + + self.state.ensure_component(name, offset)?; + validate_section( + &mut self.components, + &mut self.types, + section.count(), + offset, + )?; + + for item in section.clone().into_iter_with_offsets() { + let (offset, item) = item?; + validate_item( + &mut self.components, + &mut self.types, + &self.features, + item, + offset, + )?; + } + + Ok(()) + } +} + +#[cfg(test)] +mod tests { + use crate::{GlobalType, MemoryType, RefType, TableType, ValType, Validator, WasmFeatures}; + use anyhow::Result; + + #[test] + fn test_module_type_information() -> Result<()> { + let bytes = wat::parse_str( + r#" + (module + (type (func (param i32 i64) (result i32))) + (memory 1 5) + (table 10 funcref) + (global (mut i32) (i32.const 0)) + (func (type 0) (i32.const 0)) + (tag (param i64 i32)) + (elem funcref (ref.func 0)) + ) + "#, + )?; + + let mut validator = Validator::new_with_features(WasmFeatures { + exceptions: true, + ..Default::default() + }); + + let types = validator.validate_all(&bytes)?; + + assert_eq!(types.type_count(), 2); + assert_eq!(types.memory_count(), 1); + assert_eq!(types.table_count(), 1); + assert_eq!(types.global_count(), 1); + assert_eq!(types.function_count(), 1); + assert_eq!(types.tag_count(), 1); + assert_eq!(types.element_count(), 1); + assert_eq!(types.module_count(), 0); + assert_eq!(types.component_count(), 0); + assert_eq!(types.instance_count(), 0); + assert_eq!(types.value_count(), 0); + + match types.func_type_at(0) { + Some(ty) => { + assert_eq!(ty.params(), [ValType::I32, ValType::I64]); + assert_eq!(ty.results(), [ValType::I32]); + } + _ => unreachable!(), + } + + match types.func_type_at(1) { + Some(ty) => { + assert_eq!(ty.params(), [ValType::I64, ValType::I32]); + assert_eq!(ty.results(), []); + } + _ => unreachable!(), + } + + assert_eq!( + types.memory_at(0), + Some(MemoryType { + memory64: false, + shared: false, + initial: 1, + maximum: Some(5) + }) + ); + + assert_eq!( + types.table_at(0), + Some(TableType { + initial: 10, + maximum: None, + element_type: RefType::FUNCREF, + }) + ); + + assert_eq!( + types.global_at(0), + Some(GlobalType { + content_type: ValType::I32, + mutable: true + }) + ); + + match types.function_at(0) { + Some(ty) => { + assert_eq!(ty.params(), [ValType::I32, ValType::I64]); + assert_eq!(ty.results(), [ValType::I32]); + } + _ => unreachable!(), + } + + match types.tag_at(0) { + Some(ty) => { + assert_eq!(ty.params(), [ValType::I64, ValType::I32]); + assert_eq!(ty.results(), []); + } + _ => unreachable!(), + } + + assert_eq!(types.element_at(0), Some(RefType::FUNCREF)); + + Ok(()) + } + + #[test] + fn test_type_id_aliasing() -> Result<()> { + let bytes = wat::parse_str( + r#" + (component + (type $T (list string)) + (alias outer 0 $T (type $A1)) + (alias outer 0 $T (type $A2)) + ) + "#, + )?; + + let mut validator = Validator::new_with_features(WasmFeatures { + component_model: true, + ..Default::default() + }); + + let types = validator.validate_all(&bytes)?; + + let t_id = types.id_from_type_index(0, false).unwrap(); + let a1_id = types.id_from_type_index(1, false).unwrap(); + let a2_id = types.id_from_type_index(2, false).unwrap(); + + // The ids should all be different + assert!(t_id != a1_id); + assert!(t_id != a2_id); + assert!(a1_id != a2_id); + + // However, they should all point to the same type + assert!(std::ptr::eq( + types.type_from_id(t_id).unwrap(), + types.type_from_id(a1_id).unwrap() + )); + assert!(std::ptr::eq( + types.type_from_id(t_id).unwrap(), + types.type_from_id(a2_id).unwrap() + )); + + Ok(()) + } +} diff --git a/third_party/rust/wasmparser/src/validator/component.rs b/third_party/rust/wasmparser/src/validator/component.rs new file mode 100644 index 0000000000..641b18a2cc --- /dev/null +++ b/third_party/rust/wasmparser/src/validator/component.rs @@ -0,0 +1,2101 @@ +//! State relating to validating a WebAssembly component. + +use super::{ + check_max, combine_type_sizes, + core::Module, + types::{ + ComponentFuncType, ComponentInstanceType, ComponentInstanceTypeKind, ComponentType, + ComponentValType, EntityType, InstanceType, KebabString, ModuleType, RecordType, Type, + TypeAlloc, TypeId, TypeList, VariantCase, + }, +}; +use crate::{ + limits::*, + types::{ + ComponentDefinedType, ComponentEntityType, InstanceTypeKind, KebabStr, LoweringInfo, + TupleType, UnionType, VariantType, + }, + BinaryReaderError, CanonicalOption, ComponentExternalKind, ComponentOuterAliasKind, + ComponentTypeRef, ExternalKind, FuncType, GlobalType, InstantiationArgKind, MemoryType, Result, + TableType, TypeBounds, ValType, WasmFeatures, +}; +use indexmap::{map::Entry, IndexMap, IndexSet}; +use std::{collections::HashSet, mem}; +use url::Url; + +fn to_kebab_str<'a>(s: &'a str, desc: &str, offset: usize) -> Result<&'a KebabStr> { + match KebabStr::new(s) { + Some(s) => Ok(s), + None => { + if s.is_empty() { + bail!(offset, "{desc} name cannot be empty"); + } + + bail!(offset, "{desc} name `{s}` is not in kebab case"); + } + } +} + +fn parse_url(url: &str, offset: usize) -> Result<Option<Url>> { + if url.is_empty() { + return Ok(None); + } + + Url::parse(url) + .map(Some) + .map_err(|e| BinaryReaderError::new(e.to_string(), offset)) +} + +pub(crate) struct ComponentState { + // Core index spaces + pub core_types: Vec<TypeId>, + pub core_modules: Vec<TypeId>, + pub core_instances: Vec<TypeId>, + pub core_funcs: Vec<TypeId>, + pub core_memories: Vec<MemoryType>, + pub core_tables: Vec<TableType>, + pub core_globals: Vec<GlobalType>, + pub core_tags: Vec<TypeId>, + + // Component index spaces + pub types: Vec<TypeId>, + pub funcs: Vec<TypeId>, + pub values: Vec<(ComponentValType, bool)>, + pub instances: Vec<TypeId>, + pub components: Vec<TypeId>, + + /// A set of all imports and exports since they share the same namespace. + pub externs: IndexMap<KebabString, (Option<Url>, ComponentEntityType, ExternKind)>, + + // Note: URL validation requires unique URLs by byte comparison, so + // strings are used here and the URLs are not normalized. + import_urls: HashSet<String>, + export_urls: HashSet<String>, + + has_start: bool, + type_size: u32, +} + +pub enum ExternKind { + Import, + Export, +} + +impl ExternKind { + fn desc(&self) -> &'static str { + match self { + ExternKind::Import => "import", + ExternKind::Export => "export", + } + } +} + +impl ComponentState { + pub fn type_count(&self) -> usize { + self.core_types.len() + self.types.len() + } + + pub fn instance_count(&self) -> usize { + self.core_instances.len() + self.instances.len() + } + + pub fn function_count(&self) -> usize { + self.core_funcs.len() + self.funcs.len() + } + + pub fn add_core_type( + components: &mut [Self], + ty: crate::CoreType, + features: &WasmFeatures, + types: &mut TypeAlloc, + offset: usize, + check_limit: bool, + ) -> Result<()> { + let ty = match ty { + crate::CoreType::Func(ty) => Type::Func(ty), + crate::CoreType::Module(decls) => Type::Module(Self::create_module_type( + components, + decls.into_vec(), + features, + types, + offset, + )?), + }; + + let current = components.last_mut().unwrap(); + + if check_limit { + check_max(current.type_count(), 1, MAX_WASM_TYPES, "types", offset)?; + } + + let id = types.push_defined(ty); + current.core_types.push(id); + + Ok(()) + } + + pub fn add_core_module( + &mut self, + module: &Module, + types: &mut TypeAlloc, + offset: usize, + ) -> Result<()> { + let imports = module.imports_for_module_type(offset)?; + + // We have to clone the module's imports and exports here + // because we cannot take the data out of the `MaybeOwned` + // as it might be shared with a function validator. + let ty = Type::Module(ModuleType { + type_size: module.type_size, + imports, + exports: module.exports.clone(), + }); + + let id = types.push_anon(ty); + self.core_modules.push(id); + + Ok(()) + } + + pub fn add_core_instance( + &mut self, + instance: crate::Instance, + types: &mut TypeAlloc, + offset: usize, + ) -> Result<()> { + let instance = match instance { + crate::Instance::Instantiate { module_index, args } => { + self.instantiate_module(module_index, args.into_vec(), types, offset)? + } + crate::Instance::FromExports(exports) => { + self.instantiate_core_exports(exports.into_vec(), types, offset)? + } + }; + + self.core_instances.push(instance); + + Ok(()) + } + + pub fn add_type( + components: &mut Vec<Self>, + ty: crate::ComponentType, + features: &WasmFeatures, + types: &mut TypeAlloc, + offset: usize, + check_limit: bool, + ) -> Result<()> { + assert!(!components.is_empty()); + let ty = match ty { + crate::ComponentType::Defined(ty) => Type::Defined( + components + .last_mut() + .unwrap() + .create_defined_type(ty, types, offset)?, + ), + crate::ComponentType::Func(ty) => Type::ComponentFunc( + components + .last_mut() + .unwrap() + .create_function_type(ty, types, offset)?, + ), + crate::ComponentType::Component(decls) => Type::Component(Self::create_component_type( + components, + decls.into_vec(), + features, + types, + offset, + )?), + crate::ComponentType::Instance(decls) => Type::ComponentInstance( + Self::create_instance_type(components, decls.into_vec(), features, types, offset)?, + ), + }; + + let current = components.last_mut().unwrap(); + if check_limit { + check_max(current.type_count(), 1, MAX_WASM_TYPES, "types", offset)?; + } + + let id = types.push_defined(ty); + current.types.push(id); + + Ok(()) + } + + pub fn add_import( + &mut self, + import: crate::ComponentImport, + types: &mut TypeAlloc, + offset: usize, + ) -> Result<()> { + let entity = self.check_type_ref(&import.ty, types, offset)?; + self.add_entity(entity, false, offset)?; + let name = to_kebab_str(import.name, "import", offset)?; + + match self.externs.entry(name.to_owned()) { + Entry::Occupied(e) => { + bail!( + offset, + "import name `{name}` conflicts with previous {desc} name `{prev}`", + name = import.name, + prev = e.key(), + desc = e.get().2.desc(), + ); + } + Entry::Vacant(e) => { + let url = parse_url(import.url, offset)?; + if let Some(url) = url.as_ref() { + if !self.import_urls.insert(url.to_string()) { + bail!(offset, "duplicate import URL `{url}`"); + } + } + + self.type_size = combine_type_sizes(self.type_size, entity.type_size(), offset)?; + e.insert((url, entity, ExternKind::Import)); + } + } + + Ok(()) + } + + fn add_entity( + &mut self, + ty: ComponentEntityType, + value_used: bool, + offset: usize, + ) -> Result<()> { + let (len, max, desc) = match ty { + ComponentEntityType::Module(id) => { + self.core_modules.push(id); + (self.core_modules.len(), MAX_WASM_MODULES, "modules") + } + ComponentEntityType::Component(id) => { + self.components.push(id); + (self.components.len(), MAX_WASM_COMPONENTS, "components") + } + ComponentEntityType::Instance(id) => { + self.instances.push(id); + (self.instance_count(), MAX_WASM_INSTANCES, "instances") + } + ComponentEntityType::Func(id) => { + self.funcs.push(id); + (self.function_count(), MAX_WASM_FUNCTIONS, "functions") + } + ComponentEntityType::Value(ty) => { + self.values.push((ty, value_used)); + (self.values.len(), MAX_WASM_VALUES, "values") + } + ComponentEntityType::Type { created, .. } => { + self.types.push(created); + (self.types.len(), MAX_WASM_TYPES, "types") + } + }; + + check_max(len, 0, max, desc, offset)?; + Ok(()) + } + + pub fn add_export( + &mut self, + name: &str, + url: &str, + ty: ComponentEntityType, + offset: usize, + check_limit: bool, + ) -> Result<()> { + if check_limit { + check_max( + self.externs.len(), + 1, + MAX_WASM_EXPORTS, + "imports and exports", + offset, + )?; + } + self.add_entity(ty, true, offset)?; + + let name = to_kebab_str(name, "export", offset)?; + + match self.externs.entry(name.to_owned()) { + Entry::Occupied(e) => { + bail!( + offset, + "export name `{name}` conflicts with previous {desc} name `{prev}`", + prev = e.key(), + desc = e.get().2.desc(), + ); + } + Entry::Vacant(e) => { + let url = parse_url(url, offset)?; + if let Some(url) = url.as_ref() { + if !self.export_urls.insert(url.to_string()) { + bail!(offset, "duplicate export URL `{url}`"); + } + } + + self.type_size = combine_type_sizes(self.type_size, ty.type_size(), offset)?; + e.insert((url, ty, ExternKind::Export)); + } + } + + Ok(()) + } + + pub fn lift_function( + &mut self, + core_func_index: u32, + type_index: u32, + options: Vec<CanonicalOption>, + types: &TypeList, + offset: usize, + ) -> Result<()> { + let ty = self.function_type_at(type_index, types, offset)?; + let core_ty = types[self.core_function_at(core_func_index, offset)?] + .as_func_type() + .unwrap(); + + // Lifting a function is for an export, so match the expected canonical ABI + // export signature + let info = ty.lower(types, false); + self.check_options(Some(core_ty), &info, &options, types, offset)?; + + if core_ty.params() != info.params.as_slice() { + bail!( + offset, + "lowered parameter types `{:?}` do not match parameter types \ + `{:?}` of core function {core_func_index}", + info.params.as_slice(), + core_ty.params(), + ); + } + + if core_ty.results() != info.results.as_slice() { + bail!( + offset, + "lowered result types `{:?}` do not match result types \ + `{:?}` of core function {core_func_index}", + info.results.as_slice(), + core_ty.results() + ); + } + + self.funcs.push(self.types[type_index as usize]); + + Ok(()) + } + + pub fn lower_function( + &mut self, + func_index: u32, + options: Vec<CanonicalOption>, + types: &mut TypeAlloc, + offset: usize, + ) -> Result<()> { + let ty = types[self.function_at(func_index, offset)?] + .as_component_func_type() + .unwrap(); + + // Lowering a function is for an import, so use a function type that matches + // the expected canonical ABI import signature. + let info = ty.lower(types, true); + + self.check_options(None, &info, &options, types, offset)?; + + let lowered_ty = Type::Func(info.into_func_type()); + + let id = types.push_anon(lowered_ty); + self.core_funcs.push(id); + + Ok(()) + } + + pub fn add_component(&mut self, component: &mut Self, types: &mut TypeAlloc) { + let ty = Type::Component(component.take_component_type()); + let id = types.push_anon(ty); + self.components.push(id); + } + + pub fn add_instance( + &mut self, + instance: crate::ComponentInstance, + types: &mut TypeAlloc, + offset: usize, + ) -> Result<()> { + let instance = match instance { + crate::ComponentInstance::Instantiate { + component_index, + args, + } => self.instantiate_component(component_index, args.into_vec(), types, offset)?, + crate::ComponentInstance::FromExports(exports) => { + self.instantiate_exports(exports.into_vec(), types, offset)? + } + }; + + self.instances.push(instance); + + Ok(()) + } + + pub fn add_alias( + components: &mut [Self], + alias: crate::ComponentAlias, + types: &mut TypeAlloc, + offset: usize, + ) -> Result<()> { + match alias { + crate::ComponentAlias::InstanceExport { + instance_index, + kind, + name, + } => components.last_mut().unwrap().alias_instance_export( + instance_index, + kind, + name, + types, + offset, + ), + crate::ComponentAlias::CoreInstanceExport { + instance_index, + kind, + name, + } => components.last_mut().unwrap().alias_core_instance_export( + instance_index, + kind, + name, + types, + offset, + ), + crate::ComponentAlias::Outer { kind, count, index } => match kind { + ComponentOuterAliasKind::CoreModule => { + Self::alias_module(components, count, index, offset) + } + ComponentOuterAliasKind::CoreType => { + Self::alias_core_type(components, count, index, types, offset) + } + ComponentOuterAliasKind::Type => { + Self::alias_type(components, count, index, types, offset) + } + ComponentOuterAliasKind::Component => { + Self::alias_component(components, count, index, offset) + } + }, + } + } + + pub fn add_start( + &mut self, + func_index: u32, + args: &[u32], + results: u32, + types: &TypeList, + offset: usize, + ) -> Result<()> { + if self.has_start { + return Err(BinaryReaderError::new( + "component cannot have more than one start function", + offset, + )); + } + + let ft = types[self.function_at(func_index, offset)?] + .as_component_func_type() + .unwrap(); + + if ft.params.len() != args.len() { + bail!( + offset, + "component start function requires {} arguments but was given {}", + ft.params.len(), + args.len() + ); + } + + if ft.results.len() as u32 != results { + bail!( + offset, + "component start function has a result count of {results} \ + but the function type has a result count of {type_results}", + type_results = ft.results.len(), + ); + } + + for (i, ((_, ty), arg)) in ft.params.iter().zip(args).enumerate() { + // Ensure the value's type is a subtype of the parameter type + if !ComponentValType::internal_is_subtype_of( + self.value_at(*arg, offset)?, + types, + ty, + types, + ) { + bail!( + offset, + "value type mismatch for component start function argument {i}" + ); + } + } + + for (_, ty) in ft.results.iter() { + self.values.push((*ty, false)); + } + + self.has_start = true; + + Ok(()) + } + + fn check_options( + &self, + core_ty: Option<&FuncType>, + info: &LoweringInfo, + options: &[CanonicalOption], + types: &TypeList, + offset: usize, + ) -> Result<()> { + fn display(option: CanonicalOption) -> &'static str { + match option { + CanonicalOption::UTF8 => "utf8", + CanonicalOption::UTF16 => "utf16", + CanonicalOption::CompactUTF16 => "latin1-utf16", + CanonicalOption::Memory(_) => "memory", + CanonicalOption::Realloc(_) => "realloc", + CanonicalOption::PostReturn(_) => "post-return", + } + } + + let mut encoding = None; + let mut memory = None; + let mut realloc = None; + let mut post_return = None; + + for option in options { + match option { + CanonicalOption::UTF8 | CanonicalOption::UTF16 | CanonicalOption::CompactUTF16 => { + match encoding { + Some(existing) => { + bail!( + offset, + "canonical encoding option `{}` conflicts with option `{}`", + display(existing), + display(*option), + ) + } + None => encoding = Some(*option), + } + } + CanonicalOption::Memory(idx) => { + memory = match memory { + None => { + self.memory_at(*idx, offset)?; + Some(*idx) + } + Some(_) => { + return Err(BinaryReaderError::new( + "canonical option `memory` is specified more than once", + offset, + )) + } + } + } + CanonicalOption::Realloc(idx) => { + realloc = match realloc { + None => { + let ty = types[self.core_function_at(*idx, offset)?] + .as_func_type() + .unwrap(); + if ty.params() + != [ValType::I32, ValType::I32, ValType::I32, ValType::I32] + || ty.results() != [ValType::I32] + { + return Err(BinaryReaderError::new( + "canonical option `realloc` uses a core function with an incorrect signature", + offset, + )); + } + Some(*idx) + } + Some(_) => { + return Err(BinaryReaderError::new( + "canonical option `realloc` is specified more than once", + offset, + )) + } + } + } + CanonicalOption::PostReturn(idx) => { + post_return = match post_return { + None => { + let core_ty = core_ty.ok_or_else(|| { + BinaryReaderError::new( + "canonical option `post-return` cannot be specified for lowerings", + offset, + ) + })?; + + let ty = types[self.core_function_at(*idx, offset)?] + .as_func_type() + .unwrap(); + + if ty.params() != core_ty.results() || !ty.results().is_empty() { + return Err(BinaryReaderError::new( + "canonical option `post-return` uses a core function with an incorrect signature", + offset, + )); + } + Some(*idx) + } + Some(_) => { + return Err(BinaryReaderError::new( + "canonical option `post-return` is specified more than once", + offset, + )) + } + } + } + } + } + + if info.requires_memory && memory.is_none() { + return Err(BinaryReaderError::new( + "canonical option `memory` is required", + offset, + )); + } + + if info.requires_realloc && realloc.is_none() { + return Err(BinaryReaderError::new( + "canonical option `realloc` is required", + offset, + )); + } + + Ok(()) + } + + fn check_type_ref( + &self, + ty: &ComponentTypeRef, + types: &mut TypeAlloc, + offset: usize, + ) -> Result<ComponentEntityType> { + Ok(match ty { + ComponentTypeRef::Module(index) => { + let id = self.type_at(*index, true, offset)?; + types[id].as_module_type().ok_or_else(|| { + format_err!(offset, "core type index {index} is not a module type") + })?; + ComponentEntityType::Module(id) + } + ComponentTypeRef::Func(index) => { + let id = self.type_at(*index, false, offset)?; + types[id].as_component_func_type().ok_or_else(|| { + format_err!(offset, "type index {index} is not a function type") + })?; + ComponentEntityType::Func(id) + } + ComponentTypeRef::Value(ty) => { + let ty = match ty { + crate::ComponentValType::Primitive(ty) => ComponentValType::Primitive(*ty), + crate::ComponentValType::Type(index) => { + ComponentValType::Type(self.defined_type_at(*index, types, offset)?) + } + }; + ComponentEntityType::Value(ty) + } + ComponentTypeRef::Type(TypeBounds::Eq, index) => { + let referenced = self.type_at(*index, false, offset)?; + let created = types.with_unique(referenced); + ComponentEntityType::Type { + referenced, + created, + } + } + ComponentTypeRef::Instance(index) => { + let id = self.type_at(*index, false, offset)?; + types[id].as_component_instance_type().ok_or_else(|| { + format_err!(offset, "type index {index} is not an instance type") + })?; + ComponentEntityType::Instance(id) + } + ComponentTypeRef::Component(index) => { + let id = self.type_at(*index, false, offset)?; + types[id].as_component_type().ok_or_else(|| { + format_err!(offset, "type index {index} is not a component type") + })?; + ComponentEntityType::Component(id) + } + }) + } + + pub fn export_to_entity_type( + &mut self, + export: &crate::ComponentExport, + types: &mut TypeAlloc, + offset: usize, + ) -> Result<ComponentEntityType> { + let actual = match export.kind { + ComponentExternalKind::Module => { + ComponentEntityType::Module(self.module_at(export.index, offset)?) + } + ComponentExternalKind::Func => { + ComponentEntityType::Func(self.function_at(export.index, offset)?) + } + ComponentExternalKind::Value => { + ComponentEntityType::Value(*self.value_at(export.index, offset)?) + } + ComponentExternalKind::Type => { + let referenced = self.type_at(export.index, false, offset)?; + let created = types.with_unique(referenced); + ComponentEntityType::Type { + referenced, + created, + } + } + ComponentExternalKind::Instance => { + ComponentEntityType::Instance(self.instance_at(export.index, offset)?) + } + ComponentExternalKind::Component => { + ComponentEntityType::Component(self.component_at(export.index, offset)?) + } + }; + + let ascribed = match &export.ty { + Some(ty) => self.check_type_ref(ty, types, offset)?, + None => return Ok(actual), + }; + + if !ComponentEntityType::internal_is_subtype_of(&actual, types, &ascribed, types) { + bail!( + offset, + "ascribed type of export is not compatible with item's type" + ); + } + + Ok(ascribed) + } + + fn create_module_type( + components: &[Self], + decls: Vec<crate::ModuleTypeDeclaration>, + features: &WasmFeatures, + types: &mut TypeAlloc, + offset: usize, + ) -> Result<ModuleType> { + let mut state = Module::default(); + + for decl in decls { + match decl { + crate::ModuleTypeDeclaration::Type(ty) => { + state.add_type(ty, features, types, offset, true)?; + } + crate::ModuleTypeDeclaration::Export { name, ty } => { + let ty = state.check_type_ref(&ty, features, types, offset)?; + state.add_export(name, ty, features, offset, true)?; + } + crate::ModuleTypeDeclaration::OuterAlias { kind, count, index } => { + if count > 1 { + return Err(BinaryReaderError::new( + "outer type aliases in module type declarations are limited to a maximum count of 1", + offset, + )); + } + match kind { + crate::OuterAliasKind::Type => { + let ty = if count == 0 { + // Local alias, check the local module state + state.type_at(index, offset)? + } else { + // Otherwise, check the enclosing component state + let component = + Self::check_alias_count(components, count - 1, offset)?; + component.type_at(index, true, offset)? + }; + + check_max(state.types.len(), 1, MAX_WASM_TYPES, "types", offset)?; + + state.types.push(ty); + } + } + } + crate::ModuleTypeDeclaration::Import(import) => { + state.add_import(import, features, types, offset)?; + } + } + } + + let imports = state.imports_for_module_type(offset)?; + + Ok(ModuleType { + type_size: state.type_size, + imports, + exports: state.exports, + }) + } + + fn create_component_type( + components: &mut Vec<Self>, + decls: Vec<crate::ComponentTypeDeclaration>, + features: &WasmFeatures, + types: &mut TypeAlloc, + offset: usize, + ) -> Result<ComponentType> { + components.push(ComponentState::default()); + + for decl in decls { + match decl { + crate::ComponentTypeDeclaration::CoreType(ty) => { + Self::add_core_type(components, ty, features, types, offset, true)?; + } + crate::ComponentTypeDeclaration::Type(ty) => { + Self::add_type(components, ty, features, types, offset, true)?; + } + crate::ComponentTypeDeclaration::Export { name, url, ty } => { + let current = components.last_mut().unwrap(); + let ty = current.check_type_ref(&ty, types, offset)?; + current.add_export(name, url, ty, offset, true)?; + } + crate::ComponentTypeDeclaration::Import(import) => { + components + .last_mut() + .unwrap() + .add_import(import, types, offset)?; + } + crate::ComponentTypeDeclaration::Alias(alias) => { + Self::add_alias(components, alias, types, offset)?; + } + }; + } + + let mut state = components.pop().unwrap(); + + Ok(state.take_component_type()) + } + + fn create_instance_type( + components: &mut Vec<Self>, + decls: Vec<crate::InstanceTypeDeclaration>, + features: &WasmFeatures, + types: &mut TypeAlloc, + offset: usize, + ) -> Result<ComponentInstanceType> { + components.push(ComponentState::default()); + + for decl in decls { + match decl { + crate::InstanceTypeDeclaration::CoreType(ty) => { + Self::add_core_type(components, ty, features, types, offset, true)?; + } + crate::InstanceTypeDeclaration::Type(ty) => { + Self::add_type(components, ty, features, types, offset, true)?; + } + crate::InstanceTypeDeclaration::Export { name, url, ty } => { + let current = components.last_mut().unwrap(); + let ty = current.check_type_ref(&ty, types, offset)?; + current.add_export(name, url, ty, offset, true)?; + } + crate::InstanceTypeDeclaration::Alias(alias) => { + Self::add_alias(components, alias, types, offset)?; + } + }; + } + + let state = components.pop().unwrap(); + + Ok(ComponentInstanceType { + type_size: state.type_size, + kind: ComponentInstanceTypeKind::Defined( + state + .externs + .into_iter() + .filter_map(|(name, (url, ty, kind))| match kind { + ExternKind::Export => Some((name, (url, ty))), + ExternKind::Import => None, + }) + .collect(), + ), + }) + } + + fn create_function_type( + &self, + ty: crate::ComponentFuncType, + types: &TypeList, + offset: usize, + ) -> Result<ComponentFuncType> { + let mut type_size = 1; + + let mut set = + HashSet::with_capacity(std::cmp::max(ty.params.len(), ty.results.type_count())); + + let params = ty + .params + .iter() + .map(|(name, ty)| { + let name = to_kebab_str(name, "function parameter", offset)?; + if !set.insert(name) { + bail!( + offset, + "function parameter name `{name}` conflicts with previous parameter name `{prev}`", + prev = set.get(&name).unwrap(), + ); + } + + let ty = self.create_component_val_type(*ty, types, offset)?; + type_size = combine_type_sizes(type_size, ty.type_size(), offset)?; + Ok((name.to_owned(), ty)) + }) + .collect::<Result<_>>()?; + + set.clear(); + + let results = ty + .results + .iter() + .map(|(name, ty)| { + let name = name + .map(|name| { + let name = to_kebab_str(name, "function result", offset)?; + if !set.insert(name) { + bail!( + offset, + "function result name `{name}` conflicts with previous result name `{prev}`", + prev = set.get(name).unwrap(), + ); + } + + Ok(name.to_owned()) + }) + .transpose()?; + + let ty = self.create_component_val_type(*ty, types, offset)?; + type_size = combine_type_sizes(type_size, ty.type_size(), offset)?; + Ok((name, ty)) + }) + .collect::<Result<_>>()?; + + Ok(ComponentFuncType { + type_size, + params, + results, + }) + } + + fn instantiate_module( + &self, + module_index: u32, + module_args: Vec<crate::InstantiationArg>, + types: &mut TypeAlloc, + offset: usize, + ) -> Result<TypeId> { + fn insert_arg<'a>( + name: &'a str, + arg: &'a InstanceType, + args: &mut IndexMap<&'a str, &'a InstanceType>, + offset: usize, + ) -> Result<()> { + if args.insert(name, arg).is_some() { + bail!( + offset, + "duplicate module instantiation argument named `{name}`" + ); + } + + Ok(()) + } + + let module_type_id = self.module_at(module_index, offset)?; + let mut args = IndexMap::new(); + + // Populate the arguments + for module_arg in module_args { + match module_arg.kind { + InstantiationArgKind::Instance => { + let instance_type = types[self.core_instance_at(module_arg.index, offset)?] + .as_instance_type() + .unwrap(); + insert_arg(module_arg.name, instance_type, &mut args, offset)?; + } + } + } + + // Validate the arguments + let module_type = types[module_type_id].as_module_type().unwrap(); + for ((module, name), expected) in module_type.imports.iter() { + let instance = args.get(module.as_str()).ok_or_else(|| { + format_err!( + offset, + "missing module instantiation argument named `{module}`" + ) + })?; + + let arg = instance + .internal_exports(types) + .get(name.as_str()) + .ok_or_else(|| { + format_err!( + offset, + "module instantiation argument `{module}` does not \ + export an item named `{name}`", + ) + })?; + + match (arg, expected) { + (EntityType::Func(_), EntityType::Func(_)) + | (EntityType::Table(_), EntityType::Table(_)) + | (EntityType::Memory(_), EntityType::Memory(_)) + | (EntityType::Global(_), EntityType::Global(_)) + | (EntityType::Tag(_), EntityType::Tag(_)) => {} + _ => { + bail!( + offset, + "module instantiation argument `{module}` exports \ + an item named `{name}` but it is not a {}", + expected.desc() + ) + } + } + + if !EntityType::internal_is_subtype_of(arg, types, expected, types) { + bail!( + offset, + "{} type mismatch for export `{name}` of module \ + instantiation argument `{module}`", + expected.desc(), + ); + } + } + + let ty = Type::Instance(InstanceType { + type_size: module_type + .exports + .iter() + .fold(1, |acc, (_, ty)| acc + ty.type_size()), + kind: InstanceTypeKind::Instantiated(module_type_id), + }); + + Ok(types.push_anon(ty)) + } + + fn instantiate_component( + &mut self, + component_index: u32, + component_args: Vec<crate::ComponentInstantiationArg>, + types: &mut TypeAlloc, + offset: usize, + ) -> Result<TypeId> { + fn insert_arg<'a>( + name: &'a str, + arg: ComponentEntityType, + args: &mut IndexMap<&'a KebabStr, ComponentEntityType>, + offset: usize, + ) -> Result<()> { + let name = to_kebab_str(name, "instantiation argument", offset)?; + match args.entry(name) { + Entry::Occupied(e) => { + bail!( + offset, + "instantiation argument `{name}` conflicts with previous argument `{prev}`", + prev = e.key() + ); + } + Entry::Vacant(e) => { + e.insert(arg); + } + } + + Ok(()) + } + + let component_type_id = self.component_at(component_index, offset)?; + let mut args = IndexMap::new(); + + // Populate the arguments + for component_arg in component_args { + match component_arg.kind { + ComponentExternalKind::Module => { + insert_arg( + component_arg.name, + ComponentEntityType::Module(self.module_at(component_arg.index, offset)?), + &mut args, + offset, + )?; + } + ComponentExternalKind::Component => { + insert_arg( + component_arg.name, + ComponentEntityType::Component( + self.component_at(component_arg.index, offset)?, + ), + &mut args, + offset, + )?; + } + ComponentExternalKind::Instance => { + insert_arg( + component_arg.name, + ComponentEntityType::Instance( + self.instance_at(component_arg.index, offset)?, + ), + &mut args, + offset, + )?; + } + ComponentExternalKind::Func => { + insert_arg( + component_arg.name, + ComponentEntityType::Func(self.function_at(component_arg.index, offset)?), + &mut args, + offset, + )?; + } + ComponentExternalKind::Value => { + insert_arg( + component_arg.name, + ComponentEntityType::Value(*self.value_at(component_arg.index, offset)?), + &mut args, + offset, + )?; + } + ComponentExternalKind::Type => { + let ty = self.type_at(component_arg.index, false, offset)?; + insert_arg( + component_arg.name, + ComponentEntityType::Type { + referenced: ty, + created: ty, + }, + &mut args, + offset, + )?; + } + } + } + + // Validate the arguments + let component_type = types[component_type_id].as_component_type().unwrap(); + for (name, (_, expected)) in component_type.imports.iter() { + match args.get(&name.as_kebab_str()) { + Some(arg) => { + match (arg, expected) { + (ComponentEntityType::Module(_), ComponentEntityType::Module(_)) + | (ComponentEntityType::Component(_), ComponentEntityType::Component(_)) + | (ComponentEntityType::Instance(_), ComponentEntityType::Instance(_)) + | (ComponentEntityType::Func(_), ComponentEntityType::Func(_)) + | (ComponentEntityType::Value(_), ComponentEntityType::Value(_)) + | (ComponentEntityType::Type { .. }, ComponentEntityType::Type { .. }) => {} + _ => { + bail!( + offset, + "expected component instantiation argument `{name}` to be a {desc}", + desc = expected.desc() + ) + } + }; + + if !ComponentEntityType::internal_is_subtype_of(arg, types, expected, types) { + bail!( + offset, + "type mismatch for component instantiation argument `{name}`" + ); + } + } + None => { + bail!( + offset, + "missing component instantiation argument named `{name}`" + ); + } + } + } + + let ty = Type::ComponentInstance(ComponentInstanceType { + type_size: component_type + .exports + .iter() + .fold(1, |acc, (_, (_, ty))| acc + ty.type_size()), + kind: ComponentInstanceTypeKind::Instantiated(component_type_id), + }); + + Ok(types.push_anon(ty)) + } + + fn instantiate_exports( + &mut self, + exports: Vec<crate::ComponentExport>, + types: &mut TypeAlloc, + offset: usize, + ) -> Result<TypeId> { + fn insert_export( + name: &str, + export: ComponentEntityType, + exports: &mut IndexMap<KebabString, (Option<Url>, ComponentEntityType)>, + type_size: &mut u32, + offset: usize, + ) -> Result<()> { + let name = to_kebab_str(name, "instance export", offset)?; + match exports.entry(name.to_owned()) { + Entry::Occupied(e) => bail!( + offset, + "instance export name `{name}` conflicts with previous export name `{prev}`", + prev = e.key() + ), + Entry::Vacant(e) => { + *type_size = combine_type_sizes(*type_size, export.type_size(), offset)?; + e.insert((None, export)); + } + } + + Ok(()) + } + + let mut type_size = 1; + let mut inst_exports = IndexMap::new(); + for export in exports { + assert!(export.ty.is_none()); + match export.kind { + ComponentExternalKind::Module => { + insert_export( + export.name, + ComponentEntityType::Module(self.module_at(export.index, offset)?), + &mut inst_exports, + &mut type_size, + offset, + )?; + } + ComponentExternalKind::Component => { + insert_export( + export.name, + ComponentEntityType::Component(self.component_at(export.index, offset)?), + &mut inst_exports, + &mut type_size, + offset, + )?; + } + ComponentExternalKind::Instance => { + insert_export( + export.name, + ComponentEntityType::Instance(self.instance_at(export.index, offset)?), + &mut inst_exports, + &mut type_size, + offset, + )?; + } + ComponentExternalKind::Func => { + insert_export( + export.name, + ComponentEntityType::Func(self.function_at(export.index, offset)?), + &mut inst_exports, + &mut type_size, + offset, + )?; + } + ComponentExternalKind::Value => { + insert_export( + export.name, + ComponentEntityType::Value(*self.value_at(export.index, offset)?), + &mut inst_exports, + &mut type_size, + offset, + )?; + } + ComponentExternalKind::Type => { + let ty = self.type_at(export.index, false, offset)?; + insert_export( + export.name, + ComponentEntityType::Type { + referenced: ty, + // The created type index here isn't used anywhere + // in index spaces because a "bag of exports" + // doesn't build up its own index spaces. Just fill + // in the same index here in this case as what's + // referenced. + created: ty, + }, + &mut inst_exports, + &mut type_size, + offset, + )?; + } + } + } + + let ty = Type::ComponentInstance(ComponentInstanceType { + type_size, + kind: ComponentInstanceTypeKind::Exports(inst_exports), + }); + + Ok(types.push_anon(ty)) + } + + fn instantiate_core_exports( + &mut self, + exports: Vec<crate::Export>, + types: &mut TypeAlloc, + offset: usize, + ) -> Result<TypeId> { + fn insert_export( + name: &str, + export: EntityType, + exports: &mut IndexMap<String, EntityType>, + type_size: &mut u32, + offset: usize, + ) -> Result<()> { + *type_size = combine_type_sizes(*type_size, export.type_size(), offset)?; + + if exports.insert(name.to_string(), export).is_some() { + bail!( + offset, + "duplicate instantiation export name `{name}` already defined", + ) + } + + Ok(()) + } + + let mut type_size = 1; + let mut inst_exports = IndexMap::new(); + for export in exports { + match export.kind { + ExternalKind::Func => { + insert_export( + export.name, + EntityType::Func(self.core_function_at(export.index, offset)?), + &mut inst_exports, + &mut type_size, + offset, + )?; + } + ExternalKind::Table => insert_export( + export.name, + EntityType::Table(*self.table_at(export.index, offset)?), + &mut inst_exports, + &mut type_size, + offset, + )?, + ExternalKind::Memory => insert_export( + export.name, + EntityType::Memory(*self.memory_at(export.index, offset)?), + &mut inst_exports, + &mut type_size, + offset, + )?, + ExternalKind::Global => { + insert_export( + export.name, + EntityType::Global(*self.global_at(export.index, offset)?), + &mut inst_exports, + &mut type_size, + offset, + )?; + } + ExternalKind::Tag => insert_export( + export.name, + EntityType::Tag(self.core_function_at(export.index, offset)?), + &mut inst_exports, + &mut type_size, + offset, + )?, + } + } + + let ty = Type::Instance(InstanceType { + type_size, + kind: InstanceTypeKind::Exports(inst_exports), + }); + + Ok(types.push_anon(ty)) + } + + fn alias_core_instance_export( + &mut self, + instance_index: u32, + kind: ExternalKind, + name: &str, + types: &TypeList, + offset: usize, + ) -> Result<()> { + macro_rules! push_module_export { + ($expected:path, $collection:ident, $ty:literal) => {{ + match self.core_instance_export(instance_index, name, types, offset)? { + $expected(ty) => { + self.$collection.push(*ty); + Ok(()) + } + _ => { + bail!( + offset, + "export `{name}` for core instance {instance_index} is not a {}", + $ty + ) + } + } + }}; + } + + match kind { + ExternalKind::Func => { + check_max( + self.function_count(), + 1, + MAX_WASM_FUNCTIONS, + "functions", + offset, + )?; + push_module_export!(EntityType::Func, core_funcs, "function") + } + ExternalKind::Table => { + check_max(self.core_tables.len(), 1, MAX_WASM_TABLES, "tables", offset)?; + push_module_export!(EntityType::Table, core_tables, "table") + } + ExternalKind::Memory => { + check_max( + self.core_memories.len(), + 1, + MAX_WASM_MEMORIES, + "memories", + offset, + )?; + push_module_export!(EntityType::Memory, core_memories, "memory") + } + ExternalKind::Global => { + check_max( + self.core_globals.len(), + 1, + MAX_WASM_GLOBALS, + "globals", + offset, + )?; + push_module_export!(EntityType::Global, core_globals, "global") + } + ExternalKind::Tag => { + check_max(self.core_tags.len(), 1, MAX_WASM_TAGS, "tags", offset)?; + push_module_export!(EntityType::Tag, core_tags, "tag") + } + } + } + + fn alias_instance_export( + &mut self, + instance_index: u32, + kind: ComponentExternalKind, + name: &str, + types: &mut TypeAlloc, + offset: usize, + ) -> Result<()> { + let name = to_kebab_str(name, "alias export", offset)?; + + macro_rules! push_component_export { + ($expected:path, $collection:ident, $ty:literal) => {{ + match self.instance_export(instance_index, name, types, offset)? { + $expected(ty) => { + self.$collection.push(*ty); + Ok(()) + } + _ => { + bail!( + offset, + "export `{name}` for instance {instance_index} is not a {}", + $ty + ) + } + } + }}; + } + + match kind { + ComponentExternalKind::Module => { + check_max( + self.core_modules.len(), + 1, + MAX_WASM_MODULES, + "modules", + offset, + )?; + push_component_export!(ComponentEntityType::Module, core_modules, "module") + } + ComponentExternalKind::Component => { + check_max( + self.components.len(), + 1, + MAX_WASM_COMPONENTS, + "components", + offset, + )?; + push_component_export!(ComponentEntityType::Component, components, "component") + } + ComponentExternalKind::Instance => { + check_max( + self.instance_count(), + 1, + MAX_WASM_INSTANCES, + "instances", + offset, + )?; + push_component_export!(ComponentEntityType::Instance, instances, "instance") + } + ComponentExternalKind::Func => { + check_max( + self.function_count(), + 1, + MAX_WASM_FUNCTIONS, + "functions", + offset, + )?; + push_component_export!(ComponentEntityType::Func, funcs, "function") + } + ComponentExternalKind::Value => { + check_max(self.values.len(), 1, MAX_WASM_VALUES, "values", offset)?; + match self.instance_export(instance_index, name, types, offset)? { + ComponentEntityType::Value(ty) => { + self.values.push((*ty, false)); + Ok(()) + } + _ => bail!( + offset, + "export `{name}` for instance {instance_index} is not a value", + ), + } + } + ComponentExternalKind::Type => { + check_max(self.type_count(), 1, MAX_WASM_TYPES, "types", offset)?; + match *self.instance_export(instance_index, name, types, offset)? { + ComponentEntityType::Type { created, .. } => { + let id = types.with_unique(created); + self.types.push(id); + Ok(()) + } + _ => { + bail!( + offset, + "export `{name}` for instance {instance_index} is not a type", + ) + } + } + } + } + } + + fn alias_module(components: &mut [Self], count: u32, index: u32, offset: usize) -> Result<()> { + let component = Self::check_alias_count(components, count, offset)?; + let ty = component.module_at(index, offset)?; + + let current = components.last_mut().unwrap(); + check_max( + current.core_modules.len(), + 1, + MAX_WASM_MODULES, + "modules", + offset, + )?; + + current.core_modules.push(ty); + Ok(()) + } + + fn alias_component( + components: &mut [Self], + count: u32, + index: u32, + offset: usize, + ) -> Result<()> { + let component = Self::check_alias_count(components, count, offset)?; + let ty = component.component_at(index, offset)?; + + let current = components.last_mut().unwrap(); + check_max( + current.components.len(), + 1, + MAX_WASM_COMPONENTS, + "components", + offset, + )?; + + current.components.push(ty); + Ok(()) + } + + fn alias_core_type( + components: &mut [Self], + count: u32, + index: u32, + types: &mut TypeAlloc, + offset: usize, + ) -> Result<()> { + let component = Self::check_alias_count(components, count, offset)?; + let ty = component.type_at(index, true, offset)?; + + let current = components.last_mut().unwrap(); + check_max(current.type_count(), 1, MAX_WASM_TYPES, "types", offset)?; + + let id = types.with_unique(ty); + current.core_types.push(id); + + Ok(()) + } + + fn alias_type( + components: &mut [Self], + count: u32, + index: u32, + types: &mut TypeAlloc, + offset: usize, + ) -> Result<()> { + let component = Self::check_alias_count(components, count, offset)?; + let ty = component.type_at(index, false, offset)?; + + let current = components.last_mut().unwrap(); + check_max(current.type_count(), 1, MAX_WASM_TYPES, "types", offset)?; + + let id = types.with_unique(ty); + current.types.push(id); + + Ok(()) + } + + fn check_alias_count(components: &[Self], count: u32, offset: usize) -> Result<&Self> { + let count = count as usize; + if count >= components.len() { + bail!(offset, "invalid outer alias count of {count}"); + } + + Ok(&components[components.len() - count - 1]) + } + + fn create_defined_type( + &self, + ty: crate::ComponentDefinedType, + types: &TypeList, + offset: usize, + ) -> Result<ComponentDefinedType> { + match ty { + crate::ComponentDefinedType::Primitive(ty) => Ok(ComponentDefinedType::Primitive(ty)), + crate::ComponentDefinedType::Record(fields) => { + self.create_record_type(fields.as_ref(), types, offset) + } + crate::ComponentDefinedType::Variant(cases) => { + self.create_variant_type(cases.as_ref(), types, offset) + } + crate::ComponentDefinedType::List(ty) => Ok(ComponentDefinedType::List( + self.create_component_val_type(ty, types, offset)?, + )), + crate::ComponentDefinedType::Tuple(tys) => { + self.create_tuple_type(tys.as_ref(), types, offset) + } + crate::ComponentDefinedType::Flags(names) => { + self.create_flags_type(names.as_ref(), offset) + } + crate::ComponentDefinedType::Enum(cases) => { + self.create_enum_type(cases.as_ref(), offset) + } + crate::ComponentDefinedType::Union(tys) => { + self.create_union_type(tys.as_ref(), types, offset) + } + crate::ComponentDefinedType::Option(ty) => Ok(ComponentDefinedType::Option( + self.create_component_val_type(ty, types, offset)?, + )), + crate::ComponentDefinedType::Result { ok, err } => Ok(ComponentDefinedType::Result { + ok: ok + .map(|ty| self.create_component_val_type(ty, types, offset)) + .transpose()?, + err: err + .map(|ty| self.create_component_val_type(ty, types, offset)) + .transpose()?, + }), + } + } + + fn create_record_type( + &self, + fields: &[(&str, crate::ComponentValType)], + types: &TypeList, + offset: usize, + ) -> Result<ComponentDefinedType> { + let mut type_size = 1; + let mut field_map = IndexMap::with_capacity(fields.len()); + + for (name, ty) in fields { + let name = to_kebab_str(name, "record field", offset)?; + let ty = self.create_component_val_type(*ty, types, offset)?; + + match field_map.entry(name.to_owned()) { + Entry::Occupied(e) => bail!( + offset, + "record field name `{name}` conflicts with previous field name `{prev}`", + prev = e.key() + ), + Entry::Vacant(e) => { + type_size = combine_type_sizes(type_size, ty.type_size(), offset)?; + e.insert(ty); + } + } + } + + Ok(ComponentDefinedType::Record(RecordType { + type_size, + fields: field_map, + })) + } + + fn create_variant_type( + &self, + cases: &[crate::VariantCase], + types: &TypeList, + offset: usize, + ) -> Result<ComponentDefinedType> { + let mut type_size = 1; + let mut case_map: IndexMap<KebabString, VariantCase> = IndexMap::with_capacity(cases.len()); + + if cases.is_empty() { + return Err(BinaryReaderError::new( + "variant type must have at least one case", + offset, + )); + } + + if cases.len() > u32::MAX as usize { + return Err(BinaryReaderError::new( + "variant type cannot be represented with a 32-bit discriminant value", + offset, + )); + } + + for (i, case) in cases.iter().enumerate() { + if let Some(refines) = case.refines { + if refines >= i as u32 { + return Err(BinaryReaderError::new( + "variant case can only refine a previously defined case", + offset, + )); + } + } + + let name = to_kebab_str(case.name, "variant case", offset)?; + + let ty = case + .ty + .map(|ty| self.create_component_val_type(ty, types, offset)) + .transpose()?; + + match case_map.entry(name.to_owned()) { + Entry::Occupied(e) => bail!( + offset, + "variant case name `{name}` conflicts with previous case name `{prev}`", + name = case.name, + prev = e.key() + ), + Entry::Vacant(e) => { + type_size = combine_type_sizes( + type_size, + ty.map(|ty| ty.type_size()).unwrap_or(1), + offset, + )?; + + // Safety: the use of `KebabStr::new_unchecked` here is safe because the string + // was already verified to be kebab case. + e.insert(VariantCase { + ty, + refines: case + .refines + .map(|i| KebabStr::new_unchecked(cases[i as usize].name).to_owned()), + }); + } + } + } + + Ok(ComponentDefinedType::Variant(VariantType { + type_size, + cases: case_map, + })) + } + + fn create_tuple_type( + &self, + tys: &[crate::ComponentValType], + types: &TypeList, + offset: usize, + ) -> Result<ComponentDefinedType> { + let mut type_size = 1; + let types = tys + .iter() + .map(|ty| { + let ty = self.create_component_val_type(*ty, types, offset)?; + type_size = combine_type_sizes(type_size, ty.type_size(), offset)?; + Ok(ty) + }) + .collect::<Result<_>>()?; + + Ok(ComponentDefinedType::Tuple(TupleType { type_size, types })) + } + + fn create_flags_type(&self, names: &[&str], offset: usize) -> Result<ComponentDefinedType> { + let mut names_set = IndexSet::with_capacity(names.len()); + + for name in names { + let name = to_kebab_str(name, "flag", offset)?; + if !names_set.insert(name.to_owned()) { + bail!( + offset, + "flag name `{name}` conflicts with previous flag name `{prev}`", + prev = names_set.get(name).unwrap() + ); + } + } + + Ok(ComponentDefinedType::Flags(names_set)) + } + + fn create_enum_type(&self, cases: &[&str], offset: usize) -> Result<ComponentDefinedType> { + if cases.len() > u32::MAX as usize { + return Err(BinaryReaderError::new( + "enumeration type cannot be represented with a 32-bit discriminant value", + offset, + )); + } + + let mut tags = IndexSet::with_capacity(cases.len()); + + for tag in cases { + let tag = to_kebab_str(tag, "enum tag", offset)?; + if !tags.insert(tag.to_owned()) { + bail!( + offset, + "enum tag name `{tag}` conflicts with previous tag name `{prev}`", + prev = tags.get(tag).unwrap() + ); + } + } + + Ok(ComponentDefinedType::Enum(tags)) + } + + fn create_union_type( + &self, + tys: &[crate::ComponentValType], + types: &TypeList, + offset: usize, + ) -> Result<ComponentDefinedType> { + let mut type_size = 1; + let types = tys + .iter() + .map(|ty| { + let ty = self.create_component_val_type(*ty, types, offset)?; + type_size = combine_type_sizes(type_size, ty.type_size(), offset)?; + Ok(ty) + }) + .collect::<Result<_>>()?; + + Ok(ComponentDefinedType::Union(UnionType { type_size, types })) + } + + fn create_component_val_type( + &self, + ty: crate::ComponentValType, + types: &TypeList, + offset: usize, + ) -> Result<ComponentValType> { + Ok(match ty { + crate::ComponentValType::Primitive(pt) => ComponentValType::Primitive(pt), + crate::ComponentValType::Type(idx) => { + ComponentValType::Type(self.defined_type_at(idx, types, offset)?) + } + }) + } + + pub fn type_at(&self, idx: u32, core: bool, offset: usize) -> Result<TypeId> { + let types = if core { &self.core_types } else { &self.types }; + types + .get(idx as usize) + .copied() + .ok_or_else(|| format_err!(offset, "unknown type {idx}: type index out of bounds")) + } + + fn function_type_at<'a>( + &self, + idx: u32, + types: &'a TypeList, + offset: usize, + ) -> Result<&'a ComponentFuncType> { + types[self.type_at(idx, false, offset)?] + .as_component_func_type() + .ok_or_else(|| format_err!(offset, "type index {idx} is not a function type")) + } + + fn function_at(&self, idx: u32, offset: usize) -> Result<TypeId> { + self.funcs.get(idx as usize).copied().ok_or_else(|| { + format_err!( + offset, + "unknown function {idx}: function index out of bounds" + ) + }) + } + + fn component_at(&self, idx: u32, offset: usize) -> Result<TypeId> { + self.components.get(idx as usize).copied().ok_or_else(|| { + format_err!( + offset, + "unknown component {idx}: component index out of bounds" + ) + }) + } + + fn instance_at(&self, idx: u32, offset: usize) -> Result<TypeId> { + self.instances.get(idx as usize).copied().ok_or_else(|| { + format_err!( + offset, + "unknown instance {idx}: instance index out of bounds" + ) + }) + } + + fn instance_export<'a>( + &self, + instance_index: u32, + name: &KebabStr, + types: &'a TypeList, + offset: usize, + ) -> Result<&'a ComponentEntityType> { + match types[self.instance_at(instance_index, offset)?] + .as_component_instance_type() + .unwrap() + .internal_exports(types) + .get(name) + { + Some((_, ty)) => Ok(ty), + None => bail!( + offset, + "instance {instance_index} has no export named `{name}`" + ), + } + } + + fn value_at(&mut self, idx: u32, offset: usize) -> Result<&ComponentValType> { + match self.values.get_mut(idx as usize) { + Some((ty, used)) if !*used => { + *used = true; + Ok(ty) + } + Some(_) => bail!(offset, "value {idx} cannot be used more than once"), + None => bail!(offset, "unknown value {idx}: value index out of bounds"), + } + } + + fn defined_type_at(&self, idx: u32, types: &TypeList, offset: usize) -> Result<TypeId> { + let id = self.type_at(idx, false, offset)?; + match &types[id] { + Type::Defined(_) => Ok(id), + _ => bail!(offset, "type index {} is not a defined type", idx), + } + } + + fn core_function_at(&self, idx: u32, offset: usize) -> Result<TypeId> { + match self.core_funcs.get(idx as usize) { + Some(id) => Ok(*id), + None => bail!( + offset, + "unknown core function {idx}: function index out of bounds" + ), + } + } + + fn module_at(&self, idx: u32, offset: usize) -> Result<TypeId> { + match self.core_modules.get(idx as usize) { + Some(id) => Ok(*id), + None => bail!(offset, "unknown module {idx}: module index out of bounds"), + } + } + + fn core_instance_at(&self, idx: u32, offset: usize) -> Result<TypeId> { + match self.core_instances.get(idx as usize) { + Some(id) => Ok(*id), + None => bail!( + offset, + "unknown core instance {idx}: instance index out of bounds" + ), + } + } + + fn core_instance_export<'a>( + &self, + instance_index: u32, + name: &str, + types: &'a TypeList, + offset: usize, + ) -> Result<&'a EntityType> { + match types[self.core_instance_at(instance_index, offset)?] + .as_instance_type() + .unwrap() + .internal_exports(types) + .get(name) + { + Some(export) => Ok(export), + None => bail!( + offset, + "core instance {instance_index} has no export named `{name}`" + ), + } + } + + fn global_at(&self, idx: u32, offset: usize) -> Result<&GlobalType> { + match self.core_globals.get(idx as usize) { + Some(t) => Ok(t), + None => bail!(offset, "unknown global {idx}: global index out of bounds"), + } + } + + fn table_at(&self, idx: u32, offset: usize) -> Result<&TableType> { + match self.core_tables.get(idx as usize) { + Some(t) => Ok(t), + None => bail!(offset, "unknown table {idx}: table index out of bounds"), + } + } + + fn memory_at(&self, idx: u32, offset: usize) -> Result<&MemoryType> { + match self.core_memories.get(idx as usize) { + Some(t) => Ok(t), + None => bail!(offset, "unknown memory {idx}: memory index out of bounds"), + } + } + + fn take_component_type(&mut self) -> ComponentType { + let mut ty = ComponentType { + type_size: self.type_size, + imports: Default::default(), + exports: Default::default(), + }; + + for (name, (url, t, kind)) in mem::take(&mut self.externs) { + let map = match kind { + ExternKind::Import => &mut ty.imports, + ExternKind::Export => &mut ty.exports, + }; + let prev = map.insert(name, (url, t)); + assert!(prev.is_none()); + } + + ty + } +} + +impl Default for ComponentState { + fn default() -> Self { + Self { + core_types: Default::default(), + core_modules: Default::default(), + core_instances: Default::default(), + core_funcs: Default::default(), + core_memories: Default::default(), + core_tables: Default::default(), + core_globals: Default::default(), + core_tags: Default::default(), + types: Default::default(), + funcs: Default::default(), + values: Default::default(), + instances: Default::default(), + components: Default::default(), + externs: Default::default(), + export_urls: Default::default(), + import_urls: Default::default(), + has_start: Default::default(), + type_size: 1, + } + } +} diff --git a/third_party/rust/wasmparser/src/validator/core.rs b/third_party/rust/wasmparser/src/validator/core.rs new file mode 100644 index 0000000000..5707e1e73b --- /dev/null +++ b/third_party/rust/wasmparser/src/validator/core.rs @@ -0,0 +1,1278 @@ +//! State relating to validating a WebAssembly module. +//! +use super::{ + check_max, combine_type_sizes, + operators::{ty_to_str, OperatorValidator, OperatorValidatorAllocations}, + types::{EntityType, Type, TypeAlloc, TypeId, TypeList}, +}; +use crate::limits::*; +use crate::validator::core::arc::MaybeOwned; +use crate::{ + BinaryReaderError, ConstExpr, Data, DataKind, Element, ElementKind, ExternalKind, FuncType, + Global, GlobalType, HeapType, MemoryType, RefType, Result, Table, TableInit, TableType, + TagType, TypeRef, ValType, VisitOperator, WasmFeatures, WasmFuncType, WasmModuleResources, +}; +use indexmap::IndexMap; +use std::mem; +use std::{collections::HashSet, sync::Arc}; + +// Section order for WebAssembly modules. +// +// Component sections are unordered and allow for duplicates, +// so this isn't used for components. +#[derive(Copy, Clone, PartialOrd, Ord, PartialEq, Eq, Debug)] +pub enum Order { + Initial, + Type, + Import, + Function, + Table, + Memory, + Tag, + Global, + Export, + Start, + Element, + DataCount, + Code, + Data, +} + +impl Default for Order { + fn default() -> Order { + Order::Initial + } +} + +#[derive(Default)] +pub(crate) struct ModuleState { + /// Internal state that is incrementally built-up for the module being + /// validated. This houses type information for all wasm items, like + /// functions. Note that this starts out as a solely owned `Arc<T>` so we can + /// get mutable access, but after we get to the code section this is never + /// mutated to we can clone it cheaply and hand it to sub-validators. + pub module: arc::MaybeOwned<Module>, + + /// Where we are, order-wise, in the wasm binary. + order: Order, + + /// The number of data segments in the data section (if present). + pub data_segment_count: u32, + + /// The number of functions we expect to be defined in the code section, or + /// basically the length of the function section if it was found. The next + /// index is where we are, in the code section index space, for the next + /// entry in the code section (used to figure out what type is next for the + /// function being validated). + pub expected_code_bodies: Option<u32>, + + const_expr_allocs: OperatorValidatorAllocations, + + /// When parsing the code section, represents the current index in the section. + code_section_index: Option<usize>, +} + +impl ModuleState { + pub fn update_order(&mut self, order: Order, offset: usize) -> Result<()> { + if self.order >= order { + return Err(BinaryReaderError::new("section out of order", offset)); + } + + self.order = order; + + Ok(()) + } + + pub fn validate_end(&self, offset: usize) -> Result<()> { + // Ensure that the data count section, if any, was correct. + if let Some(data_count) = self.module.data_count { + if data_count != self.data_segment_count { + return Err(BinaryReaderError::new( + "data count and data section have inconsistent lengths", + offset, + )); + } + } + // Ensure that the function section, if nonzero, was paired with a code + // section with the appropriate length. + if let Some(n) = self.expected_code_bodies { + if n > 0 { + return Err(BinaryReaderError::new( + "function and code section have inconsistent lengths", + offset, + )); + } + } + + Ok(()) + } + + pub fn next_code_index_and_type(&mut self, offset: usize) -> Result<(u32, u32)> { + let index = self + .code_section_index + .get_or_insert(self.module.num_imported_functions as usize); + + if *index >= self.module.functions.len() { + return Err(BinaryReaderError::new( + "code section entry exceeds number of functions", + offset, + )); + } + + let ty = self.module.functions[*index]; + *index += 1; + + Ok(((*index - 1) as u32, ty)) + } + + pub fn add_global( + &mut self, + global: Global, + features: &WasmFeatures, + types: &TypeList, + offset: usize, + ) -> Result<()> { + self.module + .check_global_type(&global.ty, features, types, offset)?; + self.check_const_expr(&global.init_expr, global.ty.content_type, features, types)?; + self.module.assert_mut().globals.push(global.ty); + Ok(()) + } + + pub fn add_table( + &mut self, + table: Table<'_>, + features: &WasmFeatures, + types: &TypeList, + offset: usize, + ) -> Result<()> { + self.module + .check_table_type(&table.ty, features, types, offset)?; + + match &table.init { + TableInit::RefNull => { + if !table.ty.element_type.nullable { + bail!(offset, "type mismatch: non-defaultable element type"); + } + } + TableInit::Expr(expr) => { + if !features.function_references { + bail!( + offset, + "tables with expression initializers require \ + the function-references proposal" + ); + } + self.check_const_expr(expr, table.ty.element_type.into(), features, types)?; + } + } + self.module.assert_mut().tables.push(table.ty); + Ok(()) + } + + pub fn add_data_segment( + &mut self, + data: Data, + features: &WasmFeatures, + types: &TypeList, + offset: usize, + ) -> Result<()> { + match data.kind { + DataKind::Passive => Ok(()), + DataKind::Active { + memory_index, + offset_expr, + } => { + let ty = self.module.memory_at(memory_index, offset)?.index_type(); + self.check_const_expr(&offset_expr, ty, features, types) + } + } + } + + pub fn add_element_segment( + &mut self, + e: Element, + features: &WasmFeatures, + types: &TypeList, + offset: usize, + ) -> Result<()> { + // the `funcref` value type is allowed all the way back to the MVP, so + // don't check it here + if e.ty != RefType::FUNCREF { + self.module + .check_value_type(ValType::Ref(e.ty), features, types, offset)?; + } + match e.kind { + ElementKind::Active { + table_index, + offset_expr, + } => { + let table = self.module.table_at(table_index, offset)?; + if !self + .module + .matches(ValType::Ref(e.ty), ValType::Ref(table.element_type), types) + { + return Err(BinaryReaderError::new( + format!( + "type mismatch: invalid element type `{}` for table type `{}`", + ty_to_str(e.ty.into()), + ty_to_str(table.element_type.into()), + ), + offset, + )); + } + + self.check_const_expr(&offset_expr, ValType::I32, features, types)?; + } + ElementKind::Passive | ElementKind::Declared => { + if !features.bulk_memory { + return Err(BinaryReaderError::new( + "bulk memory must be enabled", + offset, + )); + } + } + } + + let validate_count = |count: u32| -> Result<(), BinaryReaderError> { + if count > MAX_WASM_TABLE_ENTRIES as u32 { + Err(BinaryReaderError::new( + "number of elements is out of bounds", + offset, + )) + } else { + Ok(()) + } + }; + match e.items { + crate::ElementItems::Functions(reader) => { + let count = reader.count(); + if !e.ty.nullable && count <= 0 { + return Err(BinaryReaderError::new( + "a non-nullable element must come with an initialization expression", + offset, + )); + } + validate_count(count)?; + for f in reader.into_iter_with_offsets() { + let (offset, f) = f?; + self.module.get_func_type(f, types, offset)?; + self.module.assert_mut().function_references.insert(f); + } + } + crate::ElementItems::Expressions(reader) => { + validate_count(reader.count())?; + for expr in reader { + self.check_const_expr(&expr?, ValType::Ref(e.ty), features, types)?; + } + } + } + self.module.assert_mut().element_types.push(e.ty); + Ok(()) + } + + fn check_const_expr( + &mut self, + expr: &ConstExpr<'_>, + expected_ty: ValType, + features: &WasmFeatures, + types: &TypeList, + ) -> Result<()> { + let mut validator = VisitConstOperator { + offset: 0, + order: self.order, + uninserted_funcref: false, + ops: OperatorValidator::new_const_expr( + features, + expected_ty, + mem::take(&mut self.const_expr_allocs), + ), + resources: OperatorValidatorResources { + types, + module: &mut self.module, + }, + }; + + let mut ops = expr.get_operators_reader(); + while !ops.eof() { + validator.offset = ops.original_position(); + ops.visit_operator(&mut validator)??; + } + validator.ops.finish(ops.original_position())?; + + // See comment in `RefFunc` below for why this is an assert. + assert!(!validator.uninserted_funcref); + + self.const_expr_allocs = validator.ops.into_allocations(); + + return Ok(()); + + struct VisitConstOperator<'a> { + offset: usize, + uninserted_funcref: bool, + ops: OperatorValidator, + resources: OperatorValidatorResources<'a>, + order: Order, + } + + impl VisitConstOperator<'_> { + fn validator(&mut self) -> impl VisitOperator<'_, Output = Result<()>> { + self.ops.with_resources(&self.resources, self.offset) + } + + fn validate_extended_const(&mut self) -> Result<()> { + if self.ops.features.extended_const { + Ok(()) + } else { + Err(BinaryReaderError::new( + "constant expression required: non-constant operator", + self.offset, + )) + } + } + + fn validate_global(&mut self, index: u32) -> Result<()> { + let module = &self.resources.module; + let global = module.global_at(index, self.offset)?; + if index >= module.num_imported_globals { + return Err(BinaryReaderError::new( + "constant expression required: global.get of locally defined global", + self.offset, + )); + } + if global.mutable { + return Err(BinaryReaderError::new( + "constant expression required: global.get of mutable global", + self.offset, + )); + } + Ok(()) + } + + // Functions in initialization expressions are only valid in + // element segment initialization expressions and globals. In + // these contexts we want to record all function references. + // + // Initialization expressions can also be found in the data + // section, however. A `RefFunc` instruction in those situations + // is always invalid and needs to produce a validation error. In + // this situation, though, we can no longer modify + // the state since it's been "snapshot" already for + // parallel validation of functions. + // + // If we cannot modify the function references then this function + // *should* result in a validation error, but we defer that + // validation error to happen later. The `uninserted_funcref` + // boolean here is used to track this and will cause a panic + // (aka a fuzz bug) if we somehow forget to emit an error somewhere + // else. + fn insert_ref_func(&mut self, index: u32) { + if self.order == Order::Data { + self.uninserted_funcref = true; + } else { + self.resources + .module + .assert_mut() + .function_references + .insert(index); + } + } + } + + macro_rules! define_visit_operator { + ($(@$proposal:ident $op:ident $({ $($arg:ident: $argty:ty),* })? => $visit:ident)*) => { + $( + #[allow(unused_variables)] + fn $visit(&mut self $($(,$arg: $argty)*)?) -> Self::Output { + define_visit_operator!(@visit self $visit $($($arg)*)?) + } + )* + }; + + // These are always valid in const expressions + (@visit $self:ident visit_i32_const $val:ident) => {{ + $self.validator().visit_i32_const($val) + }}; + (@visit $self:ident visit_i64_const $val:ident) => {{ + $self.validator().visit_i64_const($val) + }}; + (@visit $self:ident visit_f32_const $val:ident) => {{ + $self.validator().visit_f32_const($val) + }}; + (@visit $self:ident visit_f64_const $val:ident) => {{ + $self.validator().visit_f64_const($val) + }}; + (@visit $self:ident visit_v128_const $val:ident) => {{ + $self.validator().visit_v128_const($val) + }}; + (@visit $self:ident visit_ref_null $val:ident) => {{ + $self.validator().visit_ref_null($val) + }}; + (@visit $self:ident visit_end) => {{ + $self.validator().visit_end() + }}; + + + // These are valid const expressions when the extended-const proposal is enabled. + (@visit $self:ident visit_i32_add) => {{ + $self.validate_extended_const()?; + $self.validator().visit_i32_add() + }}; + (@visit $self:ident visit_i32_sub) => {{ + $self.validate_extended_const()?; + $self.validator().visit_i32_sub() + }}; + (@visit $self:ident visit_i32_mul) => {{ + $self.validate_extended_const()?; + $self.validator().visit_i32_mul() + }}; + (@visit $self:ident visit_i64_add) => {{ + $self.validate_extended_const()?; + $self.validator().visit_i64_add() + }}; + (@visit $self:ident visit_i64_sub) => {{ + $self.validate_extended_const()?; + $self.validator().visit_i64_sub() + }}; + (@visit $self:ident visit_i64_mul) => {{ + $self.validate_extended_const()?; + $self.validator().visit_i64_mul() + }}; + + // `global.get` is a valid const expression for imported, immutable + // globals. + (@visit $self:ident visit_global_get $idx:ident) => {{ + $self.validate_global($idx)?; + $self.validator().visit_global_get($idx) + }}; + // `ref.func`, if it's in a `global` initializer, will insert into + // the set of referenced functions so it's processed here. + (@visit $self:ident visit_ref_func $idx:ident) => {{ + $self.insert_ref_func($idx); + $self.validator().visit_ref_func($idx) + }}; + + (@visit $self:ident $op:ident $($args:tt)*) => {{ + Err(BinaryReaderError::new( + "constant expression required: non-constant operator", + $self.offset, + )) + }} + } + + impl<'a> VisitOperator<'a> for VisitConstOperator<'a> { + type Output = Result<()>; + + for_each_operator!(define_visit_operator); + } + } +} + +pub(crate) struct Module { + // This is set once the code section starts. + // `WasmModuleResources` implementations use the snapshot to + // enable parallel validation of functions. + pub snapshot: Option<Arc<TypeList>>, + // Stores indexes into the validator's types list. + pub types: Vec<TypeId>, + pub tables: Vec<TableType>, + pub memories: Vec<MemoryType>, + pub globals: Vec<GlobalType>, + pub element_types: Vec<RefType>, + pub data_count: Option<u32>, + // Stores indexes into `types`. + pub functions: Vec<u32>, + pub tags: Vec<TypeId>, + pub function_references: HashSet<u32>, + pub imports: IndexMap<(String, String), Vec<EntityType>>, + pub exports: IndexMap<String, EntityType>, + pub type_size: u32, + num_imported_globals: u32, + num_imported_functions: u32, +} + +impl Module { + pub fn add_type( + &mut self, + ty: crate::Type, + features: &WasmFeatures, + types: &mut TypeAlloc, + offset: usize, + check_limit: bool, + ) -> Result<()> { + let ty = match ty { + crate::Type::Func(t) => { + for ty in t.params().iter().chain(t.results()) { + self.check_value_type(*ty, features, types, offset)?; + } + if t.results().len() > 1 && !features.multi_value { + return Err(BinaryReaderError::new( + "func type returns multiple values but the multi-value feature is not enabled", + offset, + )); + } + Type::Func(t) + } + }; + + if check_limit { + check_max(self.types.len(), 1, MAX_WASM_TYPES, "types", offset)?; + } + + let id = types.push_defined(ty); + self.types.push(id); + Ok(()) + } + + pub fn add_import( + &mut self, + import: crate::Import, + features: &WasmFeatures, + types: &TypeList, + offset: usize, + ) -> Result<()> { + let entity = self.check_type_ref(&import.ty, features, types, offset)?; + + let (len, max, desc) = match import.ty { + TypeRef::Func(type_index) => { + self.functions.push(type_index); + self.num_imported_functions += 1; + (self.functions.len(), MAX_WASM_FUNCTIONS, "functions") + } + TypeRef::Table(ty) => { + self.tables.push(ty); + (self.tables.len(), self.max_tables(features), "tables") + } + TypeRef::Memory(ty) => { + self.memories.push(ty); + (self.memories.len(), self.max_memories(features), "memories") + } + TypeRef::Tag(ty) => { + self.tags.push(self.types[ty.func_type_idx as usize]); + (self.tags.len(), MAX_WASM_TAGS, "tags") + } + TypeRef::Global(ty) => { + if !features.mutable_global && ty.mutable { + return Err(BinaryReaderError::new( + "mutable global support is not enabled", + offset, + )); + } + self.globals.push(ty); + self.num_imported_globals += 1; + (self.globals.len(), MAX_WASM_GLOBALS, "globals") + } + }; + + check_max(len, 0, max, desc, offset)?; + + self.type_size = combine_type_sizes(self.type_size, entity.type_size(), offset)?; + + self.imports + .entry((import.module.to_string(), import.name.to_string())) + .or_default() + .push(entity); + + Ok(()) + } + + pub fn add_export( + &mut self, + name: &str, + ty: EntityType, + features: &WasmFeatures, + offset: usize, + check_limit: bool, + ) -> Result<()> { + if !features.mutable_global { + if let EntityType::Global(global_type) = ty { + if global_type.mutable { + return Err(BinaryReaderError::new( + "mutable global support is not enabled", + offset, + )); + } + } + } + + if check_limit { + check_max(self.exports.len(), 1, MAX_WASM_EXPORTS, "exports", offset)?; + } + + self.type_size = combine_type_sizes(self.type_size, ty.type_size(), offset)?; + + match self.exports.insert(name.to_string(), ty) { + Some(_) => Err(format_err!( + offset, + "duplicate export name `{name}` already defined" + )), + None => Ok(()), + } + } + + pub fn add_function(&mut self, type_index: u32, types: &TypeList, offset: usize) -> Result<()> { + self.func_type_at(type_index, types, offset)?; + self.functions.push(type_index); + Ok(()) + } + + pub fn add_memory( + &mut self, + ty: MemoryType, + features: &WasmFeatures, + offset: usize, + ) -> Result<()> { + self.check_memory_type(&ty, features, offset)?; + self.memories.push(ty); + Ok(()) + } + + pub fn add_tag( + &mut self, + ty: TagType, + features: &WasmFeatures, + types: &TypeList, + offset: usize, + ) -> Result<()> { + self.check_tag_type(&ty, features, types, offset)?; + self.tags.push(self.types[ty.func_type_idx as usize]); + Ok(()) + } + + pub fn type_at(&self, idx: u32, offset: usize) -> Result<TypeId> { + self.types + .get(idx as usize) + .copied() + .ok_or_else(|| format_err!(offset, "unknown type {idx}: type index out of bounds")) + } + + fn func_type_at<'a>( + &self, + type_index: u32, + types: &'a TypeList, + offset: usize, + ) -> Result<&'a FuncType> { + types[self.type_at(type_index, offset)?] + .as_func_type() + .ok_or_else(|| format_err!(offset, "type index {type_index} is not a function type")) + } + + pub fn check_type_ref( + &self, + type_ref: &TypeRef, + features: &WasmFeatures, + types: &TypeList, + offset: usize, + ) -> Result<EntityType> { + Ok(match type_ref { + TypeRef::Func(type_index) => { + self.func_type_at(*type_index, types, offset)?; + EntityType::Func(self.types[*type_index as usize]) + } + TypeRef::Table(t) => { + self.check_table_type(t, features, types, offset)?; + EntityType::Table(*t) + } + TypeRef::Memory(t) => { + self.check_memory_type(t, features, offset)?; + EntityType::Memory(*t) + } + TypeRef::Tag(t) => { + self.check_tag_type(t, features, types, offset)?; + EntityType::Tag(self.types[t.func_type_idx as usize]) + } + TypeRef::Global(t) => { + self.check_global_type(t, features, types, offset)?; + EntityType::Global(*t) + } + }) + } + + fn check_table_type( + &self, + ty: &TableType, + features: &WasmFeatures, + types: &TypeList, + offset: usize, + ) -> Result<()> { + // the `funcref` value type is allowed all the way back to the MVP, so + // don't check it here + if ty.element_type != RefType::FUNCREF { + self.check_value_type(ValType::Ref(ty.element_type), features, types, offset)? + } + + self.check_limits(ty.initial, ty.maximum, offset)?; + if ty.initial > MAX_WASM_TABLE_ENTRIES as u32 { + return Err(BinaryReaderError::new( + "minimum table size is out of bounds", + offset, + )); + } + Ok(()) + } + + fn check_memory_type( + &self, + ty: &MemoryType, + features: &WasmFeatures, + offset: usize, + ) -> Result<()> { + self.check_limits(ty.initial, ty.maximum, offset)?; + let (true_maximum, err) = if ty.memory64 { + if !features.memory64 { + return Err(BinaryReaderError::new( + "memory64 must be enabled for 64-bit memories", + offset, + )); + } + ( + MAX_WASM_MEMORY64_PAGES, + "memory size must be at most 2**48 pages", + ) + } else { + ( + MAX_WASM_MEMORY32_PAGES, + "memory size must be at most 65536 pages (4GiB)", + ) + }; + if ty.initial > true_maximum { + return Err(BinaryReaderError::new(err, offset)); + } + if let Some(maximum) = ty.maximum { + if maximum > true_maximum { + return Err(BinaryReaderError::new(err, offset)); + } + } + if ty.shared { + if !features.threads { + return Err(BinaryReaderError::new( + "threads must be enabled for shared memories", + offset, + )); + } + if ty.maximum.is_none() { + return Err(BinaryReaderError::new( + "shared memory must have maximum size", + offset, + )); + } + } + Ok(()) + } + + pub(crate) fn imports_for_module_type( + &self, + offset: usize, + ) -> Result<IndexMap<(String, String), EntityType>> { + // Ensure imports are unique, which is a requirement of the component model + self.imports + .iter() + .map(|((module, name), types)| { + if types.len() != 1 { + bail!( + offset, + "module has a duplicate import name `{module}:{name}` \ + that is not allowed in components", + ); + } + Ok(((module.clone(), name.clone()), types[0])) + }) + .collect::<Result<_>>() + } + + fn check_value_type( + &self, + ty: ValType, + features: &WasmFeatures, + types: &TypeList, + offset: usize, + ) -> Result<()> { + match features.check_value_type(ty) { + Ok(()) => Ok(()), + Err(e) => Err(BinaryReaderError::new(e, offset)), + }?; + // The above only checks the value type for features. + // We must check it if it's a reference. + match ty { + ValType::Ref(rt) => { + self.check_ref_type(rt, types, offset)?; + } + _ => (), + } + Ok(()) + } + + fn check_ref_type(&self, ty: RefType, types: &TypeList, offset: usize) -> Result<()> { + // Check that the heap type is valid + match ty.heap_type { + HeapType::Func | HeapType::Extern => (), + HeapType::TypedFunc(type_index) => { + // Just check that the index is valid + self.func_type_at(type_index.into(), types, offset)?; + } + } + Ok(()) + } + + fn eq_valtypes(&self, ty1: ValType, ty2: ValType, types: &TypeList) -> bool { + match (ty1, ty2) { + (ValType::Ref(rt1), ValType::Ref(rt2)) => { + rt1.nullable == rt2.nullable + && match (rt1.heap_type, rt2.heap_type) { + (HeapType::Func, HeapType::Func) => true, + (HeapType::Extern, HeapType::Extern) => true, + (HeapType::TypedFunc(n1), HeapType::TypedFunc(n2)) => { + let n1 = self.func_type_at(n1.into(), types, 0).unwrap(); + let n2 = self.func_type_at(n2.into(), types, 0).unwrap(); + self.eq_fns(n1, n2, types) + } + (_, _) => false, + } + } + _ => ty1 == ty2, + } + } + fn eq_fns(&self, f1: &impl WasmFuncType, f2: &impl WasmFuncType, types: &TypeList) -> bool { + f1.len_inputs() == f2.len_inputs() + && f2.len_outputs() == f2.len_outputs() + && f1 + .inputs() + .zip(f2.inputs()) + .all(|(t1, t2)| self.eq_valtypes(t1, t2, types)) + && f1 + .outputs() + .zip(f2.outputs()) + .all(|(t1, t2)| self.eq_valtypes(t1, t2, types)) + } + + pub(crate) fn matches(&self, ty1: ValType, ty2: ValType, types: &TypeList) -> bool { + fn matches_null(null1: bool, null2: bool) -> bool { + (null1 == null2) || null2 + } + + let matches_heap = |ty1: HeapType, ty2: HeapType, types: &TypeList| -> bool { + match (ty1, ty2) { + (HeapType::TypedFunc(n1), HeapType::TypedFunc(n2)) => { + // Check whether the defined types are (structurally) equivalent. + let n1 = self.func_type_at(n1.into(), types, 0).unwrap(); + let n2 = self.func_type_at(n2.into(), types, 0).unwrap(); + self.eq_fns(n1, n2, types) + } + (HeapType::TypedFunc(_), HeapType::Func) => true, + (_, _) => ty1 == ty2, + } + }; + + let matches_ref = |ty1: RefType, ty2: RefType, types: &TypeList| -> bool { + matches_heap(ty1.heap_type, ty2.heap_type, types) + && matches_null(ty1.nullable, ty2.nullable) + }; + + match (ty1, ty2) { + (ValType::Ref(rt1), ValType::Ref(rt2)) => matches_ref(rt1, rt2, types), + (_, _) => ty1 == ty2, + } + } + + fn check_tag_type( + &self, + ty: &TagType, + features: &WasmFeatures, + types: &TypeList, + offset: usize, + ) -> Result<()> { + if !features.exceptions { + return Err(BinaryReaderError::new( + "exceptions proposal not enabled", + offset, + )); + } + let ty = self.func_type_at(ty.func_type_idx, types, offset)?; + if !ty.results().is_empty() { + return Err(BinaryReaderError::new( + "invalid exception type: non-empty tag result type", + offset, + )); + } + Ok(()) + } + + fn check_global_type( + &self, + ty: &GlobalType, + features: &WasmFeatures, + types: &TypeList, + offset: usize, + ) -> Result<()> { + self.check_value_type(ty.content_type, features, types, offset) + } + + fn check_limits<T>(&self, initial: T, maximum: Option<T>, offset: usize) -> Result<()> + where + T: Into<u64>, + { + if let Some(max) = maximum { + if initial.into() > max.into() { + return Err(BinaryReaderError::new( + "size minimum must not be greater than maximum", + offset, + )); + } + } + Ok(()) + } + + pub fn max_tables(&self, features: &WasmFeatures) -> usize { + if features.reference_types { + MAX_WASM_TABLES + } else { + 1 + } + } + + pub fn max_memories(&self, features: &WasmFeatures) -> usize { + if features.multi_memory { + MAX_WASM_MEMORIES + } else { + 1 + } + } + + pub fn export_to_entity_type( + &mut self, + export: &crate::Export, + offset: usize, + ) -> Result<EntityType> { + let check = |ty: &str, index: u32, total: usize| { + if index as usize >= total { + Err(format_err!( + offset, + "unknown {ty} {index}: exported {ty} index out of bounds", + )) + } else { + Ok(()) + } + }; + + Ok(match export.kind { + ExternalKind::Func => { + check("function", export.index, self.functions.len())?; + self.function_references.insert(export.index); + EntityType::Func(self.types[self.functions[export.index as usize] as usize]) + } + ExternalKind::Table => { + check("table", export.index, self.tables.len())?; + EntityType::Table(self.tables[export.index as usize]) + } + ExternalKind::Memory => { + check("memory", export.index, self.memories.len())?; + EntityType::Memory(self.memories[export.index as usize]) + } + ExternalKind::Global => { + check("global", export.index, self.globals.len())?; + EntityType::Global(self.globals[export.index as usize]) + } + ExternalKind::Tag => { + check("tag", export.index, self.tags.len())?; + EntityType::Tag(self.tags[export.index as usize]) + } + }) + } + + pub fn get_func_type<'a>( + &self, + func_idx: u32, + types: &'a TypeList, + offset: usize, + ) -> Result<&'a FuncType> { + match self.functions.get(func_idx as usize) { + Some(idx) => self.func_type_at(*idx, types, offset), + None => Err(format_err!( + offset, + "unknown function {func_idx}: func index out of bounds", + )), + } + } + + fn global_at(&self, idx: u32, offset: usize) -> Result<&GlobalType> { + match self.globals.get(idx as usize) { + Some(t) => Ok(t), + None => Err(format_err!( + offset, + "unknown global {idx}: global index out of bounds" + )), + } + } + + fn table_at(&self, idx: u32, offset: usize) -> Result<&TableType> { + match self.tables.get(idx as usize) { + Some(t) => Ok(t), + None => Err(format_err!( + offset, + "unknown table {idx}: table index out of bounds" + )), + } + } + + fn memory_at(&self, idx: u32, offset: usize) -> Result<&MemoryType> { + match self.memories.get(idx as usize) { + Some(t) => Ok(t), + None => Err(format_err!( + offset, + "unknown memory {idx}: memory index out of bounds" + )), + } + } +} + +impl Default for Module { + fn default() -> Self { + Self { + snapshot: Default::default(), + types: Default::default(), + tables: Default::default(), + memories: Default::default(), + globals: Default::default(), + element_types: Default::default(), + data_count: Default::default(), + functions: Default::default(), + tags: Default::default(), + function_references: Default::default(), + imports: Default::default(), + exports: Default::default(), + type_size: 1, + num_imported_globals: Default::default(), + num_imported_functions: Default::default(), + } + } +} + +struct OperatorValidatorResources<'a> { + module: &'a mut MaybeOwned<Module>, + types: &'a TypeList, +} + +impl WasmModuleResources for OperatorValidatorResources<'_> { + type FuncType = crate::FuncType; + + fn table_at(&self, at: u32) -> Option<TableType> { + self.module.tables.get(at as usize).cloned() + } + + fn memory_at(&self, at: u32) -> Option<MemoryType> { + self.module.memories.get(at as usize).cloned() + } + + fn tag_at(&self, at: u32) -> Option<&Self::FuncType> { + Some( + self.types[*self.module.tags.get(at as usize)?] + .as_func_type() + .unwrap(), + ) + } + + fn global_at(&self, at: u32) -> Option<GlobalType> { + self.module.globals.get(at as usize).cloned() + } + + fn func_type_at(&self, at: u32) -> Option<&Self::FuncType> { + Some( + self.types[*self.module.types.get(at as usize)?] + .as_func_type() + .unwrap(), + ) + } + + fn type_index_of_function(&self, at: u32) -> Option<u32> { + self.module.functions.get(at as usize).cloned() + } + + fn type_of_function(&self, at: u32) -> Option<&Self::FuncType> { + self.func_type_at(self.type_index_of_function(at)?) + } + + fn check_value_type(&self, t: ValType, features: &WasmFeatures, offset: usize) -> Result<()> { + self.module + .check_value_type(t, features, self.types, offset) + } + + fn element_type_at(&self, at: u32) -> Option<RefType> { + self.module.element_types.get(at as usize).cloned() + } + + fn matches(&self, t1: ValType, t2: ValType) -> bool { + self.module.matches(t1, t2, self.types) + } + + fn element_count(&self) -> u32 { + self.module.element_types.len() as u32 + } + + fn data_count(&self) -> Option<u32> { + self.module.data_count + } + + fn is_function_referenced(&self, idx: u32) -> bool { + self.module.function_references.contains(&idx) + } +} + +/// The implementation of [`WasmModuleResources`] used by +/// [`Validator`](crate::Validator). +pub struct ValidatorResources(pub(crate) Arc<Module>); + +impl WasmModuleResources for ValidatorResources { + type FuncType = crate::FuncType; + + fn table_at(&self, at: u32) -> Option<TableType> { + self.0.tables.get(at as usize).cloned() + } + + fn memory_at(&self, at: u32) -> Option<MemoryType> { + self.0.memories.get(at as usize).cloned() + } + + fn tag_at(&self, at: u32) -> Option<&Self::FuncType> { + Some( + self.0.snapshot.as_ref().unwrap()[*self.0.tags.get(at as usize)?] + .as_func_type() + .unwrap(), + ) + } + + fn global_at(&self, at: u32) -> Option<GlobalType> { + self.0.globals.get(at as usize).cloned() + } + + fn func_type_at(&self, at: u32) -> Option<&Self::FuncType> { + Some( + self.0.snapshot.as_ref().unwrap()[*self.0.types.get(at as usize)?] + .as_func_type() + .unwrap(), + ) + } + + fn type_index_of_function(&self, at: u32) -> Option<u32> { + self.0.functions.get(at as usize).cloned() + } + + fn type_of_function(&self, at: u32) -> Option<&Self::FuncType> { + self.func_type_at(self.type_index_of_function(at)?) + } + + fn check_value_type(&self, t: ValType, features: &WasmFeatures, offset: usize) -> Result<()> { + self.0 + .check_value_type(t, features, self.0.snapshot.as_ref().unwrap(), offset) + } + + fn element_type_at(&self, at: u32) -> Option<RefType> { + self.0.element_types.get(at as usize).cloned() + } + + fn matches(&self, t1: ValType, t2: ValType) -> bool { + self.0.matches(t1, t2, self.0.snapshot.as_ref().unwrap()) + } + + fn element_count(&self) -> u32 { + self.0.element_types.len() as u32 + } + + fn data_count(&self) -> Option<u32> { + self.0.data_count + } + + fn is_function_referenced(&self, idx: u32) -> bool { + self.0.function_references.contains(&idx) + } +} + +const _: () = { + fn assert_send<T: Send>() {} + + // Assert that `ValidatorResources` is Send so function validation + // can be parallelizable + fn assert() { + assert_send::<ValidatorResources>(); + } +}; + +mod arc { + use std::ops::Deref; + use std::sync::Arc; + + enum Inner<T> { + Owned(T), + Shared(Arc<T>), + + Empty, // Only used for swapping from owned to shared. + } + + pub struct MaybeOwned<T> { + inner: Inner<T>, + } + + impl<T> MaybeOwned<T> { + #[inline] + fn as_mut(&mut self) -> Option<&mut T> { + match &mut self.inner { + Inner::Owned(x) => Some(x), + Inner::Shared(_) => None, + Inner::Empty => Self::unreachable(), + } + } + + #[inline] + pub fn assert_mut(&mut self) -> &mut T { + self.as_mut().unwrap() + } + + pub fn arc(&mut self) -> &Arc<T> { + self.make_shared(); + match &self.inner { + Inner::Shared(x) => x, + _ => Self::unreachable(), + } + } + + #[inline] + fn make_shared(&mut self) { + if let Inner::Shared(_) = self.inner { + return; + } + + let inner = std::mem::replace(&mut self.inner, Inner::Empty); + let x = match inner { + Inner::Owned(x) => x, + _ => Self::unreachable(), + }; + let x = Arc::new(x); + self.inner = Inner::Shared(x); + } + + #[cold] + #[inline(never)] + fn unreachable() -> ! { + unreachable!() + } + } + + impl<T: Default> Default for MaybeOwned<T> { + fn default() -> MaybeOwned<T> { + MaybeOwned { + inner: Inner::Owned(T::default()), + } + } + } + + impl<T> Deref for MaybeOwned<T> { + type Target = T; + + fn deref(&self) -> &T { + match &self.inner { + Inner::Owned(x) => x, + Inner::Shared(x) => x, + Inner::Empty => Self::unreachable(), + } + } + } +} diff --git a/third_party/rust/wasmparser/src/validator/func.rs b/third_party/rust/wasmparser/src/validator/func.rs new file mode 100644 index 0000000000..4d405f9615 --- /dev/null +++ b/third_party/rust/wasmparser/src/validator/func.rs @@ -0,0 +1,348 @@ +use super::operators::{Frame, OperatorValidator, OperatorValidatorAllocations}; +use crate::{BinaryReader, Result, ValType, VisitOperator}; +use crate::{FunctionBody, Operator, WasmFeatures, WasmModuleResources}; + +/// Resources necessary to perform validation of a function. +/// +/// This structure is created by +/// [`Validator::code_section_entry`](crate::Validator::code_section_entry) and +/// is created per-function in a WebAssembly module. This structure is suitable +/// for sending to other threads while the original +/// [`Validator`](crate::Validator) continues processing other functions. +pub struct FuncToValidate<T> { + resources: T, + index: u32, + ty: u32, + features: WasmFeatures, +} + +impl<T: WasmModuleResources> FuncToValidate<T> { + /// Creates a new function to validate which will have the specified + /// configuration parameters: + /// + /// * `index` - the core wasm function index being validated + /// * `ty` - the core wasm type index of the function being validated, + /// defining the results and parameters to the function. + /// * `resources` - metadata and type information about the module that + /// this function is validated within. + /// * `features` - enabled WebAssembly features. + pub fn new(index: u32, ty: u32, resources: T, features: &WasmFeatures) -> FuncToValidate<T> { + FuncToValidate { + resources, + index, + ty, + features: *features, + } + } + + /// Converts this [`FuncToValidate`] into a [`FuncValidator`] using the + /// `allocs` provided. + /// + /// This method, in conjunction with [`FuncValidator::into_allocations`], + /// provides a means to reuse allocations across validation of each + /// individual function. Note that it is also sufficient to call this + /// method with `Default::default()` if no prior allocations are + /// available. + /// + /// # Panics + /// + /// If a `FuncToValidate` was created with an invalid `ty` index then this + /// function will panic. + pub fn into_validator(self, allocs: FuncValidatorAllocations) -> FuncValidator<T> { + let FuncToValidate { + resources, + index, + ty, + features, + } = self; + let validator = + OperatorValidator::new_func(ty, 0, &features, &resources, allocs.0).unwrap(); + FuncValidator { + validator, + resources, + index, + } + } +} + +/// Validation context for a WebAssembly function. +/// +/// This is a finalized validator which is ready to process a [`FunctionBody`]. +/// This is created from the [`FuncToValidate::into_validator`] method. +pub struct FuncValidator<T> { + validator: OperatorValidator, + resources: T, + index: u32, +} + +/// External handle to the internal allocations used during function validation. +/// +/// This is created with either the `Default` implementation or with +/// [`FuncValidator::into_allocations`]. It is then passed as an argument to +/// [`FuncToValidate::into_validator`] to provide a means of reusing allocations +/// between each function. +#[derive(Default)] +pub struct FuncValidatorAllocations(OperatorValidatorAllocations); + +impl<T: WasmModuleResources> FuncValidator<T> { + /// Convenience function to validate an entire function's body. + /// + /// You may not end up using this in final implementations because you'll + /// often want to interleave validation with parsing. + pub fn validate(&mut self, body: &FunctionBody<'_>) -> Result<()> { + let mut reader = body.get_binary_reader(); + self.read_locals(&mut reader)?; + reader.allow_memarg64(self.validator.features.memory64); + while !reader.eof() { + reader.visit_operator(&mut self.visitor(reader.original_position()))??; + } + self.finish(reader.original_position()) + } + + /// Reads the local definitions from the given `BinaryReader`, often sourced + /// from a `FunctionBody`. + /// + /// This function will automatically advance the `BinaryReader` forward, + /// leaving reading operators up to the caller afterwards. + pub fn read_locals(&mut self, reader: &mut BinaryReader<'_>) -> Result<()> { + for _ in 0..reader.read_var_u32()? { + let offset = reader.original_position(); + let cnt = reader.read()?; + let ty = reader.read()?; + self.define_locals(offset, cnt, ty)?; + } + Ok(()) + } + + /// Defines locals into this validator. + /// + /// This should be used if the application is already reading local + /// definitions and there's no need to re-parse the function again. + pub fn define_locals(&mut self, offset: usize, count: u32, ty: ValType) -> Result<()> { + self.validator + .define_locals(offset, count, ty, &self.resources) + } + + /// Validates the next operator in a function. + /// + /// This functions is expected to be called once-per-operator in a + /// WebAssembly function. Each operator's offset in the original binary and + /// the operator itself are passed to this function to provide more useful + /// error messages. + pub fn op(&mut self, offset: usize, operator: &Operator<'_>) -> Result<()> { + self.visitor(offset).visit_operator(operator) + } + + /// Get the operator visitor for the next operator in the function. + /// + /// The returned visitor is intended to visit just one instruction at the `offset`. + /// + /// # Example + /// + /// ``` + /// # use wasmparser::{WasmModuleResources, FuncValidator, FunctionBody, Result}; + /// pub fn validate<R>(validator: &mut FuncValidator<R>, body: &FunctionBody<'_>) -> Result<()> + /// where R: WasmModuleResources + /// { + /// let mut operator_reader = body.get_binary_reader(); + /// while !operator_reader.eof() { + /// let mut visitor = validator.visitor(operator_reader.original_position()); + /// operator_reader.visit_operator(&mut visitor)??; + /// } + /// validator.finish(operator_reader.original_position()) + /// } + /// ``` + pub fn visitor<'this, 'a: 'this>( + &'this mut self, + offset: usize, + ) -> impl VisitOperator<'a, Output = Result<()>> + 'this { + self.validator.with_resources(&self.resources, offset) + } + + /// Function that must be called after the last opcode has been processed. + /// + /// This will validate that the function was properly terminated with the + /// `end` opcode. If this function is not called then the function will not + /// be properly validated. + /// + /// The `offset` provided to this function will be used as a position for an + /// error if validation fails. + pub fn finish(&mut self, offset: usize) -> Result<()> { + self.validator.finish(offset) + } + + /// Returns the underlying module resources that this validator is using. + pub fn resources(&self) -> &T { + &self.resources + } + + /// The index of the function within the module's function index space that + /// is being validated. + pub fn index(&self) -> u32 { + self.index + } + + /// Returns the number of defined local variables in the function. + pub fn len_locals(&self) -> u32 { + self.validator.locals.len_locals() + } + + /// Returns the type of the local variable at the given `index` if any. + pub fn get_local_type(&self, index: u32) -> Option<ValType> { + self.validator.locals.get(index) + } + + /// Get the current height of the operand stack. + /// + /// This returns the height of the whole operand stack for this function, + /// not just for the current control frame. + pub fn operand_stack_height(&self) -> u32 { + self.validator.operand_stack_height() as u32 + } + + /// Returns the optional value type of the value operand at the given + /// `depth` from the top of the operand stack. + /// + /// - Returns `None` if the `depth` is out of bounds. + /// - Returns `Some(None)` if there is a value with unknown type + /// at the given `depth`. + /// + /// # Note + /// + /// A `depth` of 0 will refer to the last operand on the stack. + pub fn get_operand_type(&self, depth: usize) -> Option<Option<ValType>> { + self.validator.peek_operand_at(depth) + } + + /// Returns the number of frames on the control flow stack. + /// + /// This returns the height of the whole control stack for this function, + /// not just for the current control frame. + pub fn control_stack_height(&self) -> u32 { + self.validator.control_stack_height() as u32 + } + + /// Returns a shared reference to the control flow [`Frame`] of the + /// control flow stack at the given `depth` if any. + /// + /// Returns `None` if the `depth` is out of bounds. + /// + /// # Note + /// + /// A `depth` of 0 will refer to the last frame on the stack. + pub fn get_control_frame(&self, depth: usize) -> Option<&Frame> { + self.validator.get_frame(depth) + } + + /// Consumes this validator and returns the underlying allocations that + /// were used during the validation process. + /// + /// The returned value here can be paired with + /// [`FuncToValidate::into_validator`] to reuse the allocations already + /// created by this validator. + pub fn into_allocations(self) -> FuncValidatorAllocations { + FuncValidatorAllocations(self.validator.into_allocations()) + } +} + +#[cfg(test)] +mod tests { + use super::*; + use crate::WasmFuncType; + + struct EmptyResources; + + impl WasmModuleResources for EmptyResources { + type FuncType = EmptyFuncType; + + fn table_at(&self, _at: u32) -> Option<crate::TableType> { + todo!() + } + fn memory_at(&self, _at: u32) -> Option<crate::MemoryType> { + todo!() + } + fn tag_at(&self, _at: u32) -> Option<&Self::FuncType> { + todo!() + } + fn global_at(&self, _at: u32) -> Option<crate::GlobalType> { + todo!() + } + fn func_type_at(&self, _type_idx: u32) -> Option<&Self::FuncType> { + Some(&EmptyFuncType) + } + fn type_index_of_function(&self, _at: u32) -> Option<u32> { + todo!() + } + fn type_of_function(&self, _func_idx: u32) -> Option<&Self::FuncType> { + todo!() + } + fn check_value_type( + &self, + _t: ValType, + _features: &WasmFeatures, + _offset: usize, + ) -> Result<()> { + Ok(()) + } + fn element_type_at(&self, _at: u32) -> Option<crate::RefType> { + todo!() + } + fn matches(&self, _t1: ValType, _t2: ValType) -> bool { + todo!() + } + fn element_count(&self) -> u32 { + todo!() + } + fn data_count(&self) -> Option<u32> { + todo!() + } + fn is_function_referenced(&self, _idx: u32) -> bool { + todo!() + } + } + + struct EmptyFuncType; + + impl WasmFuncType for EmptyFuncType { + fn len_inputs(&self) -> usize { + 0 + } + fn len_outputs(&self) -> usize { + 0 + } + fn input_at(&self, _at: u32) -> Option<ValType> { + todo!() + } + fn output_at(&self, _at: u32) -> Option<ValType> { + todo!() + } + } + + #[test] + fn operand_stack_height() { + let mut v = FuncToValidate::new(0, 0, EmptyResources, &Default::default()) + .into_validator(Default::default()); + + // Initially zero values on the stack. + assert_eq!(v.operand_stack_height(), 0); + + // Pushing a constant value makes use have one value on the stack. + assert!(v.op(0, &Operator::I32Const { value: 0 }).is_ok()); + assert_eq!(v.operand_stack_height(), 1); + + // Entering a new control block does not affect the stack height. + assert!(v + .op( + 1, + &Operator::Block { + blockty: crate::BlockType::Empty + } + ) + .is_ok()); + assert_eq!(v.operand_stack_height(), 1); + + // Pushing another constant value makes use have two values on the stack. + assert!(v.op(2, &Operator::I32Const { value: 99 }).is_ok()); + assert_eq!(v.operand_stack_height(), 2); + } +} diff --git a/third_party/rust/wasmparser/src/validator/operators.rs b/third_party/rust/wasmparser/src/validator/operators.rs new file mode 100644 index 0000000000..54fee8acc6 --- /dev/null +++ b/third_party/rust/wasmparser/src/validator/operators.rs @@ -0,0 +1,3474 @@ +/* Copyright 2019 Mozilla Foundation + * + * 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. + */ + +// The basic validation algorithm here is copied from the "Validation +// Algorithm" section of the WebAssembly specification - +// https://webassembly.github.io/spec/core/appendix/algorithm.html. +// +// That algorithm is followed pretty closely here, namely `push_operand`, +// `pop_operand`, `push_ctrl`, and `pop_ctrl`. If anything here is a bit +// confusing it's recommended to read over that section to see how it maps to +// the various methods here. + +use crate::{ + limits::MAX_WASM_FUNCTION_LOCALS, BinaryReaderError, BlockType, BrTable, HeapType, Ieee32, + Ieee64, MemArg, RefType, Result, ValType, VisitOperator, WasmFeatures, WasmFuncType, + WasmModuleResources, V128, +}; +use std::ops::{Deref, DerefMut}; + +pub(crate) struct OperatorValidator { + pub(super) locals: Locals, + pub(super) local_inits: Vec<bool>, + + // This is a list of flags for wasm features which are used to gate various + // instructions. + pub(crate) features: WasmFeatures, + + // Temporary storage used during the validation of `br_table`. + br_table_tmp: Vec<MaybeType>, + + /// The `control` list is the list of blocks that we're currently in. + control: Vec<Frame>, + /// The `operands` is the current type stack. + operands: Vec<MaybeType>, + /// When local_inits is modified, the relevant index is recorded here to be + /// undone when control pops + inits: Vec<u32>, + + /// Offset of the `end` instruction which emptied the `control` stack, which + /// must be the end of the function. + end_which_emptied_control: Option<usize>, +} + +// No science was performed in the creation of this number, feel free to change +// it if you so like. +const MAX_LOCALS_TO_TRACK: usize = 50; + +pub(super) struct Locals { + // Total number of locals in the function. + num_locals: u32, + + // The first MAX_LOCALS_TO_TRACK locals in a function. This is used to + // optimize the theoretically common case where most functions don't have + // many locals and don't need a full binary search in the entire local space + // below. + first: Vec<ValType>, + + // This is a "compressed" list of locals for this function. The list of + // locals are represented as a list of tuples. The second element is the + // type of the local, and the first element is monotonically increasing as + // you visit elements of this list. The first element is the maximum index + // of the local, after the previous index, of the type specified. + // + // This allows us to do a binary search on the list for a local's index for + // `local.{get,set,tee}`. We do a binary search for the index desired, and + // it either lies in a "hole" where the maximum index is specified later, + // or it's at the end of the list meaning it's out of bounds. + all: Vec<(u32, ValType)>, +} + +/// A Wasm control flow block on the control flow stack during Wasm validation. +// +// # Dev. Note +// +// This structure corresponds to `ctrl_frame` as specified at in the validation +// appendix of the wasm spec +#[derive(Debug, Copy, Clone)] +pub struct Frame { + /// Indicator for what kind of instruction pushed this frame. + pub kind: FrameKind, + /// The type signature of this frame, represented as a singular return type + /// or a type index pointing into the module's types. + pub block_type: BlockType, + /// The index, below which, this frame cannot modify the operand stack. + pub height: usize, + /// Whether this frame is unreachable so far. + pub unreachable: bool, + /// The number of initializations in the stack at the time of its creation + pub init_height: usize, +} + +/// The kind of a control flow [`Frame`]. +#[derive(Copy, Clone, Debug, PartialEq, Eq)] +pub enum FrameKind { + /// A Wasm `block` control block. + Block, + /// A Wasm `if` control block. + If, + /// A Wasm `else` control block. + Else, + /// A Wasm `loop` control block. + Loop, + /// A Wasm `try` control block. + /// + /// # Note + /// + /// This belongs to the Wasm exception handling proposal. + Try, + /// A Wasm `catch` control block. + /// + /// # Note + /// + /// This belongs to the Wasm exception handling proposal. + Catch, + /// A Wasm `catch_all` control block. + /// + /// # Note + /// + /// This belongs to the Wasm exception handling proposal. + CatchAll, +} + +struct OperatorValidatorTemp<'validator, 'resources, T> { + offset: usize, + inner: &'validator mut OperatorValidator, + resources: &'resources T, +} + +#[derive(Default)] +pub struct OperatorValidatorAllocations { + br_table_tmp: Vec<MaybeType>, + control: Vec<Frame>, + operands: Vec<MaybeType>, + local_inits: Vec<bool>, + inits: Vec<u32>, + locals_first: Vec<ValType>, + locals_all: Vec<(u32, ValType)>, +} + +/// Type storage within the validator. +/// +/// This is used to manage the operand stack and notably isn't just `ValType` to +/// handle unreachable code and the "bottom" type. +#[derive(Debug, Copy, Clone)] +enum MaybeType { + Bot, + HeapBot, + Type(ValType), +} + +// The validator is pretty performance-sensitive and `MaybeType` is the main +// unit of storage, so assert that it doesn't exceed 4 bytes which is the +// current expected size. +const _: () = { + assert!(std::mem::size_of::<MaybeType>() == 4); +}; + +impl From<ValType> for MaybeType { + fn from(ty: ValType) -> MaybeType { + MaybeType::Type(ty) + } +} + +impl OperatorValidator { + fn new(features: &WasmFeatures, allocs: OperatorValidatorAllocations) -> Self { + let OperatorValidatorAllocations { + br_table_tmp, + control, + operands, + local_inits, + inits, + locals_first, + locals_all, + } = allocs; + debug_assert!(br_table_tmp.is_empty()); + debug_assert!(control.is_empty()); + debug_assert!(operands.is_empty()); + debug_assert!(local_inits.is_empty()); + debug_assert!(inits.is_empty()); + debug_assert!(locals_first.is_empty()); + debug_assert!(locals_all.is_empty()); + OperatorValidator { + locals: Locals { + num_locals: 0, + first: locals_first, + all: locals_all, + }, + local_inits, + inits, + features: *features, + br_table_tmp, + operands, + control, + end_which_emptied_control: None, + } + } + + /// Creates a new operator validator which will be used to validate a + /// function whose type is the `ty` index specified. + /// + /// The `resources` are used to learn about the function type underlying + /// `ty`. + pub fn new_func<T>( + ty: u32, + offset: usize, + features: &WasmFeatures, + resources: &T, + allocs: OperatorValidatorAllocations, + ) -> Result<Self> + where + T: WasmModuleResources, + { + let mut ret = OperatorValidator::new(features, allocs); + ret.control.push(Frame { + kind: FrameKind::Block, + block_type: BlockType::FuncType(ty), + height: 0, + unreachable: false, + init_height: 0, + }); + let params = OperatorValidatorTemp { + // This offset is used by the `func_type_at` and `inputs`. + offset, + inner: &mut ret, + resources, + } + .func_type_at(ty)? + .inputs(); + for ty in params { + ret.locals.define(1, ty); + ret.local_inits.push(true); + } + Ok(ret) + } + + /// Creates a new operator validator which will be used to validate an + /// `init_expr` constant expression which should result in the `ty` + /// specified. + pub fn new_const_expr( + features: &WasmFeatures, + ty: ValType, + allocs: OperatorValidatorAllocations, + ) -> Self { + let mut ret = OperatorValidator::new(features, allocs); + ret.control.push(Frame { + kind: FrameKind::Block, + block_type: BlockType::Type(ty), + height: 0, + unreachable: false, + init_height: 0, + }); + ret + } + + pub fn define_locals( + &mut self, + offset: usize, + count: u32, + ty: ValType, + resources: &impl WasmModuleResources, + ) -> Result<()> { + resources.check_value_type(ty, &self.features, offset)?; + if count == 0 { + return Ok(()); + } + if !self.locals.define(count, ty) { + return Err(BinaryReaderError::new( + "too many locals: locals exceed maximum", + offset, + )); + } + self.local_inits + .resize(self.local_inits.len() + count as usize, ty.is_defaultable()); + Ok(()) + } + + /// Returns the current operands stack height. + pub fn operand_stack_height(&self) -> usize { + self.operands.len() + } + + /// Returns the optional value type of the value operand at the given + /// `depth` from the top of the operand stack. + /// + /// - Returns `None` if the `depth` is out of bounds. + /// - Returns `Some(None)` if there is a value with unknown type + /// at the given `depth`. + /// + /// # Note + /// + /// A `depth` of 0 will refer to the last operand on the stack. + pub fn peek_operand_at(&self, depth: usize) -> Option<Option<ValType>> { + Some(match self.operands.iter().rev().nth(depth)? { + MaybeType::Type(t) => Some(*t), + MaybeType::Bot | MaybeType::HeapBot => None, + }) + } + + /// Returns the number of frames on the control flow stack. + pub fn control_stack_height(&self) -> usize { + self.control.len() + } + + pub fn get_frame(&self, depth: usize) -> Option<&Frame> { + self.control.iter().rev().nth(depth) + } + + /// Create a temporary [`OperatorValidatorTemp`] for validation. + pub fn with_resources<'a, 'validator, 'resources, T>( + &'validator mut self, + resources: &'resources T, + offset: usize, + ) -> impl VisitOperator<'a, Output = Result<()>> + 'validator + where + T: WasmModuleResources, + 'resources: 'validator, + { + WasmProposalValidator(OperatorValidatorTemp { + offset, + inner: self, + resources, + }) + } + + pub fn finish(&mut self, offset: usize) -> Result<()> { + if self.control.last().is_some() { + bail!( + offset, + "control frames remain at end of function: END opcode expected" + ); + } + + // The `end` opcode is one byte which means that the `offset` here + // should point just beyond the `end` opcode which emptied the control + // stack. If not that means more instructions were present after the + // control stack was emptied. + if offset != self.end_which_emptied_control.unwrap() + 1 { + return Err(self.err_beyond_end(offset)); + } + Ok(()) + } + + fn err_beyond_end(&self, offset: usize) -> BinaryReaderError { + format_err!(offset, "operators remaining after end of function") + } + + pub fn into_allocations(self) -> OperatorValidatorAllocations { + fn truncate<T>(mut tmp: Vec<T>) -> Vec<T> { + tmp.truncate(0); + tmp + } + OperatorValidatorAllocations { + br_table_tmp: truncate(self.br_table_tmp), + control: truncate(self.control), + operands: truncate(self.operands), + local_inits: truncate(self.local_inits), + inits: truncate(self.inits), + locals_first: truncate(self.locals.first), + locals_all: truncate(self.locals.all), + } + } +} + +impl<R> Deref for OperatorValidatorTemp<'_, '_, R> { + type Target = OperatorValidator; + fn deref(&self) -> &OperatorValidator { + self.inner + } +} + +impl<R> DerefMut for OperatorValidatorTemp<'_, '_, R> { + fn deref_mut(&mut self) -> &mut OperatorValidator { + self.inner + } +} + +impl<'resources, R: WasmModuleResources> OperatorValidatorTemp<'_, 'resources, R> { + /// Pushes a type onto the operand stack. + /// + /// This is used by instructions to represent a value that is pushed to the + /// operand stack. This can fail, but only if `Type` is feature gated. + /// Otherwise the push operation always succeeds. + fn push_operand<T>(&mut self, ty: T) -> Result<()> + where + T: Into<MaybeType>, + { + let maybe_ty = ty.into(); + self.operands.push(maybe_ty); + Ok(()) + } + + /// Attempts to pop a type from the operand stack. + /// + /// This function is used to remove types from the operand stack. The + /// `expected` argument can be used to indicate that a type is required, or + /// simply that something is needed to be popped. + /// + /// If `expected` is `Some(T)` then this will be guaranteed to return + /// `T`, and it will only return success if the current block is + /// unreachable or if `T` was found at the top of the operand stack. + /// + /// If `expected` is `None` then it indicates that something must be on the + /// operand stack, but it doesn't matter what's on the operand stack. This + /// is useful for polymorphic instructions like `select`. + /// + /// If `Some(T)` is returned then `T` was popped from the operand stack and + /// matches `expected`. If `None` is returned then it means that `None` was + /// expected and a type was successfully popped, but its exact type is + /// indeterminate because the current block is unreachable. + fn pop_operand(&mut self, expected: Option<ValType>) -> Result<MaybeType> { + // This method is one of the hottest methods in the validator so to + // improve codegen this method contains a fast-path success case where + // if the top operand on the stack is as expected it's returned + // immediately. This is the most common case where the stack will indeed + // have the expected type and all we need to do is pop it off. + // + // Note that this still has to be careful to be correct, though. For + // efficiency an operand is unconditionally popped and on success it is + // matched against the state of the world to see if we could actually + // pop it. If we shouldn't have popped it then it's passed to the slow + // path to get pushed back onto the stack. + let popped = match self.operands.pop() { + Some(MaybeType::Type(actual_ty)) => { + if Some(actual_ty) == expected { + if let Some(control) = self.control.last() { + if self.operands.len() >= control.height { + return Ok(MaybeType::Type(actual_ty)); + } + } + } + Some(MaybeType::Type(actual_ty)) + } + other => other, + }; + + self._pop_operand(expected, popped) + } + + // This is the "real" implementation of `pop_operand` which is 100% + // spec-compliant with little attention paid to efficiency since this is the + // slow-path from the actual `pop_operand` function above. + #[cold] + fn _pop_operand( + &mut self, + expected: Option<ValType>, + popped: Option<MaybeType>, + ) -> Result<MaybeType> { + self.operands.extend(popped); + let control = match self.control.last() { + Some(c) => c, + None => return Err(self.err_beyond_end(self.offset)), + }; + let actual = if self.operands.len() == control.height && control.unreachable { + MaybeType::Bot + } else { + if self.operands.len() == control.height { + let desc = match expected { + Some(ty) => ty_to_str(ty), + None => "a type", + }; + bail!( + self.offset, + "type mismatch: expected {desc} but nothing on stack" + ) + } else { + self.operands.pop().unwrap() + } + }; + if let Some(expected) = expected { + match (actual, expected) { + // The bottom type matches all expectations + (MaybeType::Bot, _) + // The "heap bottom" type only matches other references types, + // but not any integer types. + | (MaybeType::HeapBot, ValType::Ref(_)) => {} + + // Use the `matches` predicate to test if a found type matches + // the expectation. + (MaybeType::Type(actual), expected) => { + if !self.resources.matches(actual, expected) { + bail!( + self.offset, + "type mismatch: expected {}, found {}", + ty_to_str(expected), + ty_to_str(actual) + ); + } + } + + // A "heap bottom" type cannot match any numeric types. + ( + MaybeType::HeapBot, + ValType::I32 | ValType::I64 | ValType::F32 | ValType::F64 | ValType::V128, + ) => { + bail!( + self.offset, + "type mismatche: expected {}, found heap type", + ty_to_str(expected) + ) + } + } + } + Ok(actual) + } + + fn pop_ref(&mut self) -> Result<Option<RefType>> { + match self.pop_operand(None)? { + MaybeType::Bot | MaybeType::HeapBot => Ok(None), + MaybeType::Type(ValType::Ref(rt)) => Ok(Some(rt)), + MaybeType::Type(ty) => bail!( + self.offset, + "type mismatch: expected ref but found {}", + ty_to_str(ty) + ), + } + } + + /// Fetches the type for the local at `idx`, returning an error if it's out + /// of bounds. + fn local(&self, idx: u32) -> Result<ValType> { + match self.locals.get(idx) { + Some(ty) => Ok(ty), + None => bail!( + self.offset, + "unknown local {}: local index out of bounds", + idx + ), + } + } + + /// Flags the current control frame as unreachable, additionally truncating + /// the currently active operand stack. + fn unreachable(&mut self) -> Result<()> { + let control = match self.control.last_mut() { + Some(frame) => frame, + None => return Err(self.err_beyond_end(self.offset)), + }; + control.unreachable = true; + let new_height = control.height; + self.operands.truncate(new_height); + Ok(()) + } + + /// Pushes a new frame onto the control stack. + /// + /// This operation is used when entering a new block such as an if, loop, + /// or block itself. The `kind` of block is specified which indicates how + /// breaks interact with this block's type. Additionally the type signature + /// of the block is specified by `ty`. + fn push_ctrl(&mut self, kind: FrameKind, ty: BlockType) -> Result<()> { + // Push a new frame which has a snapshot of the height of the current + // operand stack. + let height = self.operands.len(); + let init_height = self.inits.len(); + self.control.push(Frame { + kind, + block_type: ty, + height, + unreachable: false, + init_height, + }); + // All of the parameters are now also available in this control frame, + // so we push them here in order. + for ty in self.params(ty)? { + self.push_operand(ty)?; + } + Ok(()) + } + + /// Pops a frame from the control stack. + /// + /// This function is used when exiting a block and leaves a block scope. + /// Internally this will validate that blocks have the correct result type. + fn pop_ctrl(&mut self) -> Result<Frame> { + // Read the expected type and expected height of the operand stack the + // end of the frame. + let frame = match self.control.last() { + Some(f) => f, + None => return Err(self.err_beyond_end(self.offset)), + }; + let ty = frame.block_type; + let height = frame.height; + let init_height = frame.init_height; + + // reset_locals in the spec + for init in self.inits.split_off(init_height) { + self.local_inits[init as usize] = false; + } + + // Pop all the result types, in reverse order, from the operand stack. + // These types will, possibly, be transferred to the next frame. + for ty in self.results(ty)?.rev() { + self.pop_operand(Some(ty))?; + } + + // Make sure that the operand stack has returned to is original + // height... + if self.operands.len() != height { + bail!( + self.offset, + "type mismatch: values remaining on stack at end of block" + ); + } + + // And then we can remove it! + Ok(self.control.pop().unwrap()) + } + + /// Validates a relative jump to the `depth` specified. + /// + /// Returns the type signature of the block that we're jumping to as well + /// as the kind of block if the jump is valid. Otherwise returns an error. + fn jump(&self, depth: u32) -> Result<(BlockType, FrameKind)> { + if self.control.is_empty() { + return Err(self.err_beyond_end(self.offset)); + } + match (self.control.len() - 1).checked_sub(depth as usize) { + Some(i) => { + let frame = &self.control[i]; + Ok((frame.block_type, frame.kind)) + } + None => bail!(self.offset, "unknown label: branch depth too large"), + } + } + + /// Validates that `memory_index` is valid in this module, and returns the + /// type of address used to index the memory specified. + fn check_memory_index(&self, memory_index: u32) -> Result<ValType> { + match self.resources.memory_at(memory_index) { + Some(mem) => Ok(mem.index_type()), + None => bail!(self.offset, "unknown memory {}", memory_index), + } + } + + /// Validates a `memarg for alignment and such (also the memory it + /// references), and returns the type of index used to address the memory. + fn check_memarg(&self, memarg: MemArg) -> Result<ValType> { + let index_ty = self.check_memory_index(memarg.memory)?; + if memarg.align > memarg.max_align { + bail!(self.offset, "alignment must not be larger than natural"); + } + if index_ty == ValType::I32 && memarg.offset > u64::from(u32::MAX) { + bail!(self.offset, "offset out of range: must be <= 2**32"); + } + Ok(index_ty) + } + + fn check_floats_enabled(&self) -> Result<()> { + if !self.features.floats { + bail!(self.offset, "floating-point instruction disallowed"); + } + Ok(()) + } + + fn check_shared_memarg(&self, memarg: MemArg) -> Result<ValType> { + if memarg.align != memarg.max_align { + bail!( + self.offset, + "atomic instructions must always specify maximum alignment" + ); + } + self.check_memory_index(memarg.memory) + } + + fn check_simd_lane_index(&self, index: u8, max: u8) -> Result<()> { + if index >= max { + bail!(self.offset, "SIMD index out of bounds"); + } + Ok(()) + } + + /// Validates a block type, primarily with various in-flight proposals. + fn check_block_type(&self, ty: BlockType) -> Result<()> { + match ty { + BlockType::Empty => Ok(()), + BlockType::Type(t) => self + .resources + .check_value_type(t, &self.features, self.offset), + BlockType::FuncType(idx) => { + if !self.features.multi_value { + bail!( + self.offset, + "blocks, loops, and ifs may only produce a resulttype \ + when multi-value is not enabled", + ); + } + self.func_type_at(idx)?; + Ok(()) + } + } + } + + /// Validates a `call` instruction, ensuring that the function index is + /// in-bounds and the right types are on the stack to call the function. + fn check_call(&mut self, function_index: u32) -> Result<()> { + let ty = match self.resources.type_index_of_function(function_index) { + Some(i) => i, + None => { + bail!( + self.offset, + "unknown function {function_index}: function index out of bounds", + ); + } + }; + self.check_call_ty(ty) + } + + fn check_call_ty(&mut self, type_index: u32) -> Result<()> { + let ty = match self.resources.func_type_at(type_index) { + Some(i) => i, + None => { + bail!( + self.offset, + "unknown type {type_index}: type index out of bounds", + ); + } + }; + for ty in ty.inputs().rev() { + self.pop_operand(Some(ty))?; + } + for ty in ty.outputs() { + self.push_operand(ty)?; + } + Ok(()) + } + + /// Validates a call to an indirect function, very similar to `check_call`. + fn check_call_indirect(&mut self, index: u32, table_index: u32) -> Result<()> { + match self.resources.table_at(table_index) { + None => { + bail!(self.offset, "unknown table: table index out of bounds"); + } + Some(tab) => { + if !self + .resources + .matches(ValType::Ref(tab.element_type), ValType::FUNCREF) + { + bail!( + self.offset, + "indirect calls must go through a table with type <= funcref", + ); + } + } + } + let ty = self.func_type_at(index)?; + self.pop_operand(Some(ValType::I32))?; + for ty in ty.inputs().rev() { + self.pop_operand(Some(ty))?; + } + for ty in ty.outputs() { + self.push_operand(ty)?; + } + Ok(()) + } + + /// Validates a `return` instruction, popping types from the operand + /// stack that the function needs. + fn check_return(&mut self) -> Result<()> { + if self.control.is_empty() { + return Err(self.err_beyond_end(self.offset)); + } + for ty in self.results(self.control[0].block_type)?.rev() { + self.pop_operand(Some(ty))?; + } + self.unreachable()?; + Ok(()) + } + + /// Checks the validity of a common comparison operator. + fn check_cmp_op(&mut self, ty: ValType) -> Result<()> { + self.pop_operand(Some(ty))?; + self.pop_operand(Some(ty))?; + self.push_operand(ValType::I32)?; + Ok(()) + } + + /// Checks the validity of a common float comparison operator. + fn check_fcmp_op(&mut self, ty: ValType) -> Result<()> { + debug_assert!(matches!(ty, ValType::F32 | ValType::F64)); + self.check_floats_enabled()?; + self.check_cmp_op(ty) + } + + /// Checks the validity of a common unary operator. + fn check_unary_op(&mut self, ty: ValType) -> Result<()> { + self.pop_operand(Some(ty))?; + self.push_operand(ty)?; + Ok(()) + } + + /// Checks the validity of a common unary float operator. + fn check_funary_op(&mut self, ty: ValType) -> Result<()> { + debug_assert!(matches!(ty, ValType::F32 | ValType::F64)); + self.check_floats_enabled()?; + self.check_unary_op(ty) + } + + /// Checks the validity of a common conversion operator. + fn check_conversion_op(&mut self, into: ValType, from: ValType) -> Result<()> { + self.pop_operand(Some(from))?; + self.push_operand(into)?; + Ok(()) + } + + /// Checks the validity of a common conversion operator. + fn check_fconversion_op(&mut self, into: ValType, from: ValType) -> Result<()> { + debug_assert!(matches!(into, ValType::F32 | ValType::F64)); + self.check_floats_enabled()?; + self.check_conversion_op(into, from) + } + + /// Checks the validity of a common binary operator. + fn check_binary_op(&mut self, ty: ValType) -> Result<()> { + self.pop_operand(Some(ty))?; + self.pop_operand(Some(ty))?; + self.push_operand(ty)?; + Ok(()) + } + + /// Checks the validity of a common binary float operator. + fn check_fbinary_op(&mut self, ty: ValType) -> Result<()> { + debug_assert!(matches!(ty, ValType::F32 | ValType::F64)); + self.check_floats_enabled()?; + self.check_binary_op(ty) + } + + /// Checks the validity of an atomic load operator. + fn check_atomic_load(&mut self, memarg: MemArg, load_ty: ValType) -> Result<()> { + let ty = self.check_shared_memarg(memarg)?; + self.pop_operand(Some(ty))?; + self.push_operand(load_ty)?; + Ok(()) + } + + /// Checks the validity of an atomic store operator. + fn check_atomic_store(&mut self, memarg: MemArg, store_ty: ValType) -> Result<()> { + let ty = self.check_shared_memarg(memarg)?; + self.pop_operand(Some(store_ty))?; + self.pop_operand(Some(ty))?; + Ok(()) + } + + /// Checks the validity of a common atomic binary operator. + fn check_atomic_binary_op(&mut self, memarg: MemArg, op_ty: ValType) -> Result<()> { + let ty = self.check_shared_memarg(memarg)?; + self.pop_operand(Some(op_ty))?; + self.pop_operand(Some(ty))?; + self.push_operand(op_ty)?; + Ok(()) + } + + /// Checks the validity of an atomic compare exchange operator. + fn check_atomic_binary_cmpxchg(&mut self, memarg: MemArg, op_ty: ValType) -> Result<()> { + let ty = self.check_shared_memarg(memarg)?; + self.pop_operand(Some(op_ty))?; + self.pop_operand(Some(op_ty))?; + self.pop_operand(Some(ty))?; + self.push_operand(op_ty)?; + Ok(()) + } + + /// Checks a [`V128`] splat operator. + fn check_v128_splat(&mut self, src_ty: ValType) -> Result<()> { + self.pop_operand(Some(src_ty))?; + self.push_operand(ValType::V128)?; + Ok(()) + } + + /// Checks a [`V128`] binary operator. + fn check_v128_binary_op(&mut self) -> Result<()> { + self.pop_operand(Some(ValType::V128))?; + self.pop_operand(Some(ValType::V128))?; + self.push_operand(ValType::V128)?; + Ok(()) + } + + /// Checks a [`V128`] binary float operator. + fn check_v128_fbinary_op(&mut self) -> Result<()> { + self.check_floats_enabled()?; + self.check_v128_binary_op() + } + + /// Checks a [`V128`] binary operator. + fn check_v128_unary_op(&mut self) -> Result<()> { + self.pop_operand(Some(ValType::V128))?; + self.push_operand(ValType::V128)?; + Ok(()) + } + + /// Checks a [`V128`] binary operator. + fn check_v128_funary_op(&mut self) -> Result<()> { + self.check_floats_enabled()?; + self.check_v128_unary_op() + } + + /// Checks a [`V128`] relaxed ternary operator. + fn check_v128_ternary_op(&mut self) -> Result<()> { + self.pop_operand(Some(ValType::V128))?; + self.pop_operand(Some(ValType::V128))?; + self.pop_operand(Some(ValType::V128))?; + self.push_operand(ValType::V128)?; + Ok(()) + } + + /// Checks a [`V128`] relaxed ternary operator. + fn check_v128_bitmask_op(&mut self) -> Result<()> { + self.pop_operand(Some(ValType::V128))?; + self.push_operand(ValType::I32)?; + Ok(()) + } + + /// Checks a [`V128`] relaxed ternary operator. + fn check_v128_shift_op(&mut self) -> Result<()> { + self.pop_operand(Some(ValType::I32))?; + self.pop_operand(Some(ValType::V128))?; + self.push_operand(ValType::V128)?; + Ok(()) + } + + /// Checks a [`V128`] common load operator. + fn check_v128_load_op(&mut self, memarg: MemArg) -> Result<()> { + let idx = self.check_memarg(memarg)?; + self.pop_operand(Some(idx))?; + self.push_operand(ValType::V128)?; + Ok(()) + } + + fn func_type_at(&self, at: u32) -> Result<&'resources R::FuncType> { + self.resources + .func_type_at(at) + .ok_or_else(|| format_err!(self.offset, "unknown type: type index out of bounds")) + } + + fn tag_at(&self, at: u32) -> Result<&'resources R::FuncType> { + self.resources + .tag_at(at) + .ok_or_else(|| format_err!(self.offset, "unknown tag {}: tag index out of bounds", at)) + } + + fn params(&self, ty: BlockType) -> Result<impl PreciseIterator<Item = ValType> + 'resources> { + Ok(match ty { + BlockType::Empty | BlockType::Type(_) => Either::B(None.into_iter()), + BlockType::FuncType(t) => Either::A(self.func_type_at(t)?.inputs()), + }) + } + + fn results(&self, ty: BlockType) -> Result<impl PreciseIterator<Item = ValType> + 'resources> { + Ok(match ty { + BlockType::Empty => Either::B(None.into_iter()), + BlockType::Type(t) => Either::B(Some(t).into_iter()), + BlockType::FuncType(t) => Either::A(self.func_type_at(t)?.outputs()), + }) + } + + fn label_types( + &self, + ty: BlockType, + kind: FrameKind, + ) -> Result<impl PreciseIterator<Item = ValType> + 'resources> { + Ok(match kind { + FrameKind::Loop => Either::A(self.params(ty)?), + _ => Either::B(self.results(ty)?), + }) + } +} + +pub fn ty_to_str(ty: ValType) -> &'static str { + match ty { + ValType::I32 => "i32", + ValType::I64 => "i64", + ValType::F32 => "f32", + ValType::F64 => "f64", + ValType::V128 => "v128", + ValType::FUNCREF => "funcref", + ValType::EXTERNREF => "externref", + ValType::Ref(RefType { + nullable: false, + heap_type: HeapType::Func, + }) => "(ref func)", + ValType::Ref(RefType { + nullable: false, + heap_type: HeapType::Extern, + }) => "(ref extern)", + ValType::Ref(RefType { + nullable: false, + heap_type: HeapType::TypedFunc(_), + }) => "(ref $type)", + ValType::Ref(RefType { + nullable: true, + heap_type: HeapType::TypedFunc(_), + }) => "(ref null $type)", + } +} + +/// A wrapper "visitor" around the real operator validator internally which +/// exists to check that the required wasm feature is enabled to proceed with +/// validation. +/// +/// This validator is macro-generated to ensure that the proposal listed in this +/// crate's macro matches the one that's validated here. Each instruction's +/// visit method validates the specified proposal is enabled and then delegates +/// to `OperatorValidatorTemp` to perform the actual opcode validation. +struct WasmProposalValidator<'validator, 'resources, T>( + OperatorValidatorTemp<'validator, 'resources, T>, +); + +impl<T> WasmProposalValidator<'_, '_, T> { + fn check_enabled(&self, flag: bool, desc: &str) -> Result<()> { + if flag { + return Ok(()); + } + bail!(self.0.offset, "{desc} support is not enabled"); + } +} + +macro_rules! validate_proposal { + ($( @$proposal:ident $op:ident $({ $($arg:ident: $argty:ty),* })? => $visit:ident)*) => { + $( + fn $visit(&mut self $($(,$arg: $argty)*)?) -> Result<()> { + validate_proposal!(validate self $proposal); + self.0.$visit($( $($arg),* )?) + } + )* + }; + + (validate self mvp) => {}; + (validate $self:ident $proposal:ident) => { + $self.check_enabled($self.0.features.$proposal, validate_proposal!(desc $proposal))? + }; + + (desc simd) => ("SIMD"); + (desc relaxed_simd) => ("relaxed SIMD"); + (desc threads) => ("threads"); + (desc saturating_float_to_int) => ("saturating float to int conversions"); + (desc reference_types) => ("reference types"); + (desc bulk_memory) => ("bulk memory"); + (desc sign_extension) => ("sign extension operations"); + (desc exceptions) => ("exceptions"); + (desc tail_call) => ("tail calls"); + (desc function_references) => ("function references"); + (desc memory_control) => ("memory control"); +} + +impl<'a, T> VisitOperator<'a> for WasmProposalValidator<'_, '_, T> +where + T: WasmModuleResources, +{ + type Output = Result<()>; + + for_each_operator!(validate_proposal); +} + +impl<'a, T> VisitOperator<'a> for OperatorValidatorTemp<'_, '_, T> +where + T: WasmModuleResources, +{ + type Output = Result<()>; + + fn visit_nop(&mut self) -> Self::Output { + Ok(()) + } + fn visit_unreachable(&mut self) -> Self::Output { + self.unreachable()?; + Ok(()) + } + fn visit_block(&mut self, ty: BlockType) -> Self::Output { + self.check_block_type(ty)?; + for ty in self.params(ty)?.rev() { + self.pop_operand(Some(ty))?; + } + self.push_ctrl(FrameKind::Block, ty)?; + Ok(()) + } + fn visit_loop(&mut self, ty: BlockType) -> Self::Output { + self.check_block_type(ty)?; + for ty in self.params(ty)?.rev() { + self.pop_operand(Some(ty))?; + } + self.push_ctrl(FrameKind::Loop, ty)?; + Ok(()) + } + fn visit_if(&mut self, ty: BlockType) -> Self::Output { + self.check_block_type(ty)?; + self.pop_operand(Some(ValType::I32))?; + for ty in self.params(ty)?.rev() { + self.pop_operand(Some(ty))?; + } + self.push_ctrl(FrameKind::If, ty)?; + Ok(()) + } + fn visit_else(&mut self) -> Self::Output { + let frame = self.pop_ctrl()?; + if frame.kind != FrameKind::If { + bail!(self.offset, "else found outside of an `if` block"); + } + self.push_ctrl(FrameKind::Else, frame.block_type)?; + Ok(()) + } + fn visit_try(&mut self, ty: BlockType) -> Self::Output { + self.check_block_type(ty)?; + for ty in self.params(ty)?.rev() { + self.pop_operand(Some(ty))?; + } + self.push_ctrl(FrameKind::Try, ty)?; + Ok(()) + } + fn visit_catch(&mut self, index: u32) -> Self::Output { + let frame = self.pop_ctrl()?; + if frame.kind != FrameKind::Try && frame.kind != FrameKind::Catch { + bail!(self.offset, "catch found outside of an `try` block"); + } + // Start a new frame and push `exnref` value. + let height = self.operands.len(); + let init_height = self.inits.len(); + self.control.push(Frame { + kind: FrameKind::Catch, + block_type: frame.block_type, + height, + unreachable: false, + init_height, + }); + // Push exception argument types. + let ty = self.tag_at(index)?; + for ty in ty.inputs() { + self.push_operand(ty)?; + } + Ok(()) + } + fn visit_throw(&mut self, index: u32) -> Self::Output { + // Check values associated with the exception. + let ty = self.tag_at(index)?; + for ty in ty.inputs().rev() { + self.pop_operand(Some(ty))?; + } + if ty.outputs().len() > 0 { + bail!( + self.offset, + "result type expected to be empty for exception" + ); + } + self.unreachable()?; + Ok(()) + } + fn visit_rethrow(&mut self, relative_depth: u32) -> Self::Output { + // This is not a jump, but we need to check that the `rethrow` + // targets an actual `catch` to get the exception. + let (_, kind) = self.jump(relative_depth)?; + if kind != FrameKind::Catch && kind != FrameKind::CatchAll { + bail!( + self.offset, + "invalid rethrow label: target was not a `catch` block" + ); + } + self.unreachable()?; + Ok(()) + } + fn visit_delegate(&mut self, relative_depth: u32) -> Self::Output { + let frame = self.pop_ctrl()?; + if frame.kind != FrameKind::Try { + bail!(self.offset, "delegate found outside of an `try` block"); + } + // This operation is not a jump, but we need to check the + // depth for validity + let _ = self.jump(relative_depth)?; + for ty in self.results(frame.block_type)? { + self.push_operand(ty)?; + } + Ok(()) + } + fn visit_catch_all(&mut self) -> Self::Output { + let frame = self.pop_ctrl()?; + if frame.kind == FrameKind::CatchAll { + bail!(self.offset, "only one catch_all allowed per `try` block"); + } else if frame.kind != FrameKind::Try && frame.kind != FrameKind::Catch { + bail!(self.offset, "catch_all found outside of a `try` block"); + } + let height = self.operands.len(); + let init_height = self.inits.len(); + self.control.push(Frame { + kind: FrameKind::CatchAll, + block_type: frame.block_type, + height, + unreachable: false, + init_height, + }); + Ok(()) + } + fn visit_end(&mut self) -> Self::Output { + let mut frame = self.pop_ctrl()?; + + // Note that this `if` isn't included in the appendix right + // now, but it's used to allow for `if` statements that are + // missing an `else` block which have the same parameter/return + // types on the block (since that's valid). + if frame.kind == FrameKind::If { + self.push_ctrl(FrameKind::Else, frame.block_type)?; + frame = self.pop_ctrl()?; + } + for ty in self.results(frame.block_type)? { + self.push_operand(ty)?; + } + + if self.control.is_empty() && self.end_which_emptied_control.is_none() { + assert_ne!(self.offset, 0); + self.end_which_emptied_control = Some(self.offset); + } + Ok(()) + } + fn visit_br(&mut self, relative_depth: u32) -> Self::Output { + let (ty, kind) = self.jump(relative_depth)?; + for ty in self.label_types(ty, kind)?.rev() { + self.pop_operand(Some(ty))?; + } + self.unreachable()?; + Ok(()) + } + fn visit_br_if(&mut self, relative_depth: u32) -> Self::Output { + self.pop_operand(Some(ValType::I32))?; + let (ty, kind) = self.jump(relative_depth)?; + let types = self.label_types(ty, kind)?; + for ty in types.clone().rev() { + self.pop_operand(Some(ty))?; + } + for ty in types { + self.push_operand(ty)?; + } + Ok(()) + } + fn visit_br_table(&mut self, table: BrTable) -> Self::Output { + self.pop_operand(Some(ValType::I32))?; + let default = self.jump(table.default())?; + let default_types = self.label_types(default.0, default.1)?; + for element in table.targets() { + let relative_depth = element?; + let block = self.jump(relative_depth)?; + let tys = self.label_types(block.0, block.1)?; + if tys.len() != default_types.len() { + bail!( + self.offset, + "type mismatch: br_table target labels have different number of types" + ); + } + debug_assert!(self.br_table_tmp.is_empty()); + for ty in tys.rev() { + let ty = self.pop_operand(Some(ty))?; + self.br_table_tmp.push(ty); + } + for ty in self.inner.br_table_tmp.drain(..).rev() { + self.inner.operands.push(ty); + } + } + for ty in default_types.rev() { + self.pop_operand(Some(ty))?; + } + self.unreachable()?; + Ok(()) + } + fn visit_return(&mut self) -> Self::Output { + self.check_return()?; + Ok(()) + } + fn visit_call(&mut self, function_index: u32) -> Self::Output { + self.check_call(function_index)?; + Ok(()) + } + fn visit_return_call(&mut self, function_index: u32) -> Self::Output { + self.check_call(function_index)?; + self.check_return()?; + Ok(()) + } + fn visit_call_ref(&mut self, hty: HeapType) -> Self::Output { + self.resources + .check_heap_type(hty, &self.features, self.offset)?; + // If `None` is popped then that means a "bottom" type was popped which + // is always considered equivalent to the `hty` tag. + if let Some(rt) = self.pop_ref()? { + let expected = RefType { + nullable: true, + heap_type: hty, + }; + if !self + .resources + .matches(ValType::Ref(rt), ValType::Ref(expected)) + { + bail!( + self.offset, + "type mismatch: funcref on stack does not match specified type", + ); + } + } + match hty { + HeapType::TypedFunc(type_index) => self.check_call_ty(type_index.into())?, + _ => bail!( + self.offset, + "type mismatch: instruction requires function reference type", + ), + } + Ok(()) + } + fn visit_return_call_ref(&mut self, hty: HeapType) -> Self::Output { + self.visit_call_ref(hty)?; + self.check_return() + } + fn visit_call_indirect( + &mut self, + index: u32, + table_index: u32, + table_byte: u8, + ) -> Self::Output { + if table_byte != 0 && !self.features.reference_types { + bail!( + self.offset, + "reference-types not enabled: zero byte expected" + ); + } + self.check_call_indirect(index, table_index)?; + Ok(()) + } + fn visit_return_call_indirect(&mut self, index: u32, table_index: u32) -> Self::Output { + self.check_call_indirect(index, table_index)?; + self.check_return()?; + Ok(()) + } + fn visit_drop(&mut self) -> Self::Output { + self.pop_operand(None)?; + Ok(()) + } + fn visit_select(&mut self) -> Self::Output { + self.pop_operand(Some(ValType::I32))?; + let ty1 = self.pop_operand(None)?; + let ty2 = self.pop_operand(None)?; + + let ty = match (ty1, ty2) { + // All heap-related types aren't allowed with the `select` + // instruction + (MaybeType::HeapBot, _) + | (_, MaybeType::HeapBot) + | (MaybeType::Type(ValType::Ref(_)), _) + | (_, MaybeType::Type(ValType::Ref(_))) => { + bail!( + self.offset, + "type mismatch: select only takes integral types" + ) + } + + // If one operand is the "bottom" type then whatever the other + // operand is is the result of the `select` + (MaybeType::Bot, t) | (t, MaybeType::Bot) => t, + + // Otherwise these are two integral types and they must match for + // `select` to typecheck. + (t @ MaybeType::Type(t1), MaybeType::Type(t2)) => { + if t1 != t2 { + bail!( + self.offset, + "type mismatch: select operands have different types" + ); + } + t + } + }; + self.push_operand(ty)?; + Ok(()) + } + fn visit_typed_select(&mut self, ty: ValType) -> Self::Output { + self.resources + .check_value_type(ty, &self.features, self.offset)?; + self.pop_operand(Some(ValType::I32))?; + self.pop_operand(Some(ty))?; + self.pop_operand(Some(ty))?; + self.push_operand(ty)?; + Ok(()) + } + fn visit_local_get(&mut self, local_index: u32) -> Self::Output { + let ty = self.local(local_index)?; + if !self.local_inits[local_index as usize] { + bail!(self.offset, "uninitialized local: {}", local_index); + } + self.push_operand(ty)?; + Ok(()) + } + fn visit_local_set(&mut self, local_index: u32) -> Self::Output { + let ty = self.local(local_index)?; + self.pop_operand(Some(ty))?; + if !self.local_inits[local_index as usize] { + self.local_inits[local_index as usize] = true; + self.inits.push(local_index); + } + Ok(()) + } + fn visit_local_tee(&mut self, local_index: u32) -> Self::Output { + let ty = self.local(local_index)?; + self.pop_operand(Some(ty))?; + if !self.local_inits[local_index as usize] { + self.local_inits[local_index as usize] = true; + self.inits.push(local_index); + } + + self.push_operand(ty)?; + Ok(()) + } + fn visit_global_get(&mut self, global_index: u32) -> Self::Output { + if let Some(ty) = self.resources.global_at(global_index) { + self.push_operand(ty.content_type)?; + } else { + bail!(self.offset, "unknown global: global index out of bounds"); + }; + Ok(()) + } + fn visit_global_set(&mut self, global_index: u32) -> Self::Output { + if let Some(ty) = self.resources.global_at(global_index) { + if !ty.mutable { + bail!( + self.offset, + "global is immutable: cannot modify it with `global.set`" + ); + } + self.pop_operand(Some(ty.content_type))?; + } else { + bail!(self.offset, "unknown global: global index out of bounds"); + }; + Ok(()) + } + fn visit_i32_load(&mut self, memarg: MemArg) -> Self::Output { + let ty = self.check_memarg(memarg)?; + self.pop_operand(Some(ty))?; + self.push_operand(ValType::I32)?; + Ok(()) + } + fn visit_i64_load(&mut self, memarg: MemArg) -> Self::Output { + let ty = self.check_memarg(memarg)?; + self.pop_operand(Some(ty))?; + self.push_operand(ValType::I64)?; + Ok(()) + } + fn visit_f32_load(&mut self, memarg: MemArg) -> Self::Output { + self.check_floats_enabled()?; + let ty = self.check_memarg(memarg)?; + self.pop_operand(Some(ty))?; + self.push_operand(ValType::F32)?; + Ok(()) + } + fn visit_f64_load(&mut self, memarg: MemArg) -> Self::Output { + self.check_floats_enabled()?; + let ty = self.check_memarg(memarg)?; + self.pop_operand(Some(ty))?; + self.push_operand(ValType::F64)?; + Ok(()) + } + fn visit_i32_load8_s(&mut self, memarg: MemArg) -> Self::Output { + let ty = self.check_memarg(memarg)?; + self.pop_operand(Some(ty))?; + self.push_operand(ValType::I32)?; + Ok(()) + } + fn visit_i32_load8_u(&mut self, memarg: MemArg) -> Self::Output { + self.visit_i32_load8_s(memarg) + } + fn visit_i32_load16_s(&mut self, memarg: MemArg) -> Self::Output { + let ty = self.check_memarg(memarg)?; + self.pop_operand(Some(ty))?; + self.push_operand(ValType::I32)?; + Ok(()) + } + fn visit_i32_load16_u(&mut self, memarg: MemArg) -> Self::Output { + self.visit_i32_load16_s(memarg) + } + fn visit_i64_load8_s(&mut self, memarg: MemArg) -> Self::Output { + let ty = self.check_memarg(memarg)?; + self.pop_operand(Some(ty))?; + self.push_operand(ValType::I64)?; + Ok(()) + } + fn visit_i64_load8_u(&mut self, memarg: MemArg) -> Self::Output { + self.visit_i64_load8_s(memarg) + } + fn visit_i64_load16_s(&mut self, memarg: MemArg) -> Self::Output { + let ty = self.check_memarg(memarg)?; + self.pop_operand(Some(ty))?; + self.push_operand(ValType::I64)?; + Ok(()) + } + fn visit_i64_load16_u(&mut self, memarg: MemArg) -> Self::Output { + self.visit_i64_load16_s(memarg) + } + fn visit_i64_load32_s(&mut self, memarg: MemArg) -> Self::Output { + let ty = self.check_memarg(memarg)?; + self.pop_operand(Some(ty))?; + self.push_operand(ValType::I64)?; + Ok(()) + } + fn visit_i64_load32_u(&mut self, memarg: MemArg) -> Self::Output { + self.visit_i64_load32_s(memarg) + } + fn visit_i32_store(&mut self, memarg: MemArg) -> Self::Output { + let ty = self.check_memarg(memarg)?; + self.pop_operand(Some(ValType::I32))?; + self.pop_operand(Some(ty))?; + Ok(()) + } + fn visit_i64_store(&mut self, memarg: MemArg) -> Self::Output { + let ty = self.check_memarg(memarg)?; + self.pop_operand(Some(ValType::I64))?; + self.pop_operand(Some(ty))?; + Ok(()) + } + fn visit_f32_store(&mut self, memarg: MemArg) -> Self::Output { + self.check_floats_enabled()?; + let ty = self.check_memarg(memarg)?; + self.pop_operand(Some(ValType::F32))?; + self.pop_operand(Some(ty))?; + Ok(()) + } + fn visit_f64_store(&mut self, memarg: MemArg) -> Self::Output { + self.check_floats_enabled()?; + let ty = self.check_memarg(memarg)?; + self.pop_operand(Some(ValType::F64))?; + self.pop_operand(Some(ty))?; + Ok(()) + } + fn visit_i32_store8(&mut self, memarg: MemArg) -> Self::Output { + let ty = self.check_memarg(memarg)?; + self.pop_operand(Some(ValType::I32))?; + self.pop_operand(Some(ty))?; + Ok(()) + } + fn visit_i32_store16(&mut self, memarg: MemArg) -> Self::Output { + let ty = self.check_memarg(memarg)?; + self.pop_operand(Some(ValType::I32))?; + self.pop_operand(Some(ty))?; + Ok(()) + } + fn visit_i64_store8(&mut self, memarg: MemArg) -> Self::Output { + let ty = self.check_memarg(memarg)?; + self.pop_operand(Some(ValType::I64))?; + self.pop_operand(Some(ty))?; + Ok(()) + } + fn visit_i64_store16(&mut self, memarg: MemArg) -> Self::Output { + let ty = self.check_memarg(memarg)?; + self.pop_operand(Some(ValType::I64))?; + self.pop_operand(Some(ty))?; + Ok(()) + } + fn visit_i64_store32(&mut self, memarg: MemArg) -> Self::Output { + let ty = self.check_memarg(memarg)?; + self.pop_operand(Some(ValType::I64))?; + self.pop_operand(Some(ty))?; + Ok(()) + } + fn visit_memory_size(&mut self, mem: u32, mem_byte: u8) -> Self::Output { + if mem_byte != 0 && !self.features.multi_memory { + bail!(self.offset, "multi-memory not enabled: zero byte expected"); + } + let index_ty = self.check_memory_index(mem)?; + self.push_operand(index_ty)?; + Ok(()) + } + fn visit_memory_grow(&mut self, mem: u32, mem_byte: u8) -> Self::Output { + if mem_byte != 0 && !self.features.multi_memory { + bail!(self.offset, "multi-memory not enabled: zero byte expected"); + } + let index_ty = self.check_memory_index(mem)?; + self.pop_operand(Some(index_ty))?; + self.push_operand(index_ty)?; + Ok(()) + } + fn visit_i32_const(&mut self, _value: i32) -> Self::Output { + self.push_operand(ValType::I32)?; + Ok(()) + } + fn visit_i64_const(&mut self, _value: i64) -> Self::Output { + self.push_operand(ValType::I64)?; + Ok(()) + } + fn visit_f32_const(&mut self, _value: Ieee32) -> Self::Output { + self.check_floats_enabled()?; + self.push_operand(ValType::F32)?; + Ok(()) + } + fn visit_f64_const(&mut self, _value: Ieee64) -> Self::Output { + self.check_floats_enabled()?; + self.push_operand(ValType::F64)?; + Ok(()) + } + fn visit_i32_eqz(&mut self) -> Self::Output { + self.pop_operand(Some(ValType::I32))?; + self.push_operand(ValType::I32)?; + Ok(()) + } + fn visit_i32_eq(&mut self) -> Self::Output { + self.check_cmp_op(ValType::I32) + } + fn visit_i32_ne(&mut self) -> Self::Output { + self.check_cmp_op(ValType::I32) + } + fn visit_i32_lt_s(&mut self) -> Self::Output { + self.check_cmp_op(ValType::I32) + } + fn visit_i32_lt_u(&mut self) -> Self::Output { + self.check_cmp_op(ValType::I32) + } + fn visit_i32_gt_s(&mut self) -> Self::Output { + self.check_cmp_op(ValType::I32) + } + fn visit_i32_gt_u(&mut self) -> Self::Output { + self.check_cmp_op(ValType::I32) + } + fn visit_i32_le_s(&mut self) -> Self::Output { + self.check_cmp_op(ValType::I32) + } + fn visit_i32_le_u(&mut self) -> Self::Output { + self.check_cmp_op(ValType::I32) + } + fn visit_i32_ge_s(&mut self) -> Self::Output { + self.check_cmp_op(ValType::I32) + } + fn visit_i32_ge_u(&mut self) -> Self::Output { + self.check_cmp_op(ValType::I32) + } + fn visit_i64_eqz(&mut self) -> Self::Output { + self.pop_operand(Some(ValType::I64))?; + self.push_operand(ValType::I32)?; + Ok(()) + } + fn visit_i64_eq(&mut self) -> Self::Output { + self.check_cmp_op(ValType::I64) + } + fn visit_i64_ne(&mut self) -> Self::Output { + self.check_cmp_op(ValType::I64) + } + fn visit_i64_lt_s(&mut self) -> Self::Output { + self.check_cmp_op(ValType::I64) + } + fn visit_i64_lt_u(&mut self) -> Self::Output { + self.check_cmp_op(ValType::I64) + } + fn visit_i64_gt_s(&mut self) -> Self::Output { + self.check_cmp_op(ValType::I64) + } + fn visit_i64_gt_u(&mut self) -> Self::Output { + self.check_cmp_op(ValType::I64) + } + fn visit_i64_le_s(&mut self) -> Self::Output { + self.check_cmp_op(ValType::I64) + } + fn visit_i64_le_u(&mut self) -> Self::Output { + self.check_cmp_op(ValType::I64) + } + fn visit_i64_ge_s(&mut self) -> Self::Output { + self.check_cmp_op(ValType::I64) + } + fn visit_i64_ge_u(&mut self) -> Self::Output { + self.check_cmp_op(ValType::I64) + } + fn visit_f32_eq(&mut self) -> Self::Output { + self.check_fcmp_op(ValType::F32) + } + fn visit_f32_ne(&mut self) -> Self::Output { + self.check_fcmp_op(ValType::F32) + } + fn visit_f32_lt(&mut self) -> Self::Output { + self.check_fcmp_op(ValType::F32) + } + fn visit_f32_gt(&mut self) -> Self::Output { + self.check_fcmp_op(ValType::F32) + } + fn visit_f32_le(&mut self) -> Self::Output { + self.check_fcmp_op(ValType::F32) + } + fn visit_f32_ge(&mut self) -> Self::Output { + self.check_fcmp_op(ValType::F32) + } + fn visit_f64_eq(&mut self) -> Self::Output { + self.check_fcmp_op(ValType::F64) + } + fn visit_f64_ne(&mut self) -> Self::Output { + self.check_fcmp_op(ValType::F64) + } + fn visit_f64_lt(&mut self) -> Self::Output { + self.check_fcmp_op(ValType::F64) + } + fn visit_f64_gt(&mut self) -> Self::Output { + self.check_fcmp_op(ValType::F64) + } + fn visit_f64_le(&mut self) -> Self::Output { + self.check_fcmp_op(ValType::F64) + } + fn visit_f64_ge(&mut self) -> Self::Output { + self.check_fcmp_op(ValType::F64) + } + fn visit_i32_clz(&mut self) -> Self::Output { + self.check_unary_op(ValType::I32) + } + fn visit_i32_ctz(&mut self) -> Self::Output { + self.check_unary_op(ValType::I32) + } + fn visit_i32_popcnt(&mut self) -> Self::Output { + self.check_unary_op(ValType::I32) + } + fn visit_i32_add(&mut self) -> Self::Output { + self.check_binary_op(ValType::I32) + } + fn visit_i32_sub(&mut self) -> Self::Output { + self.check_binary_op(ValType::I32) + } + fn visit_i32_mul(&mut self) -> Self::Output { + self.check_binary_op(ValType::I32) + } + fn visit_i32_div_s(&mut self) -> Self::Output { + self.check_binary_op(ValType::I32) + } + fn visit_i32_div_u(&mut self) -> Self::Output { + self.check_binary_op(ValType::I32) + } + fn visit_i32_rem_s(&mut self) -> Self::Output { + self.check_binary_op(ValType::I32) + } + fn visit_i32_rem_u(&mut self) -> Self::Output { + self.check_binary_op(ValType::I32) + } + fn visit_i32_and(&mut self) -> Self::Output { + self.check_binary_op(ValType::I32) + } + fn visit_i32_or(&mut self) -> Self::Output { + self.check_binary_op(ValType::I32) + } + fn visit_i32_xor(&mut self) -> Self::Output { + self.check_binary_op(ValType::I32) + } + fn visit_i32_shl(&mut self) -> Self::Output { + self.check_binary_op(ValType::I32) + } + fn visit_i32_shr_s(&mut self) -> Self::Output { + self.check_binary_op(ValType::I32) + } + fn visit_i32_shr_u(&mut self) -> Self::Output { + self.check_binary_op(ValType::I32) + } + fn visit_i32_rotl(&mut self) -> Self::Output { + self.check_binary_op(ValType::I32) + } + fn visit_i32_rotr(&mut self) -> Self::Output { + self.check_binary_op(ValType::I32) + } + fn visit_i64_clz(&mut self) -> Self::Output { + self.check_unary_op(ValType::I64) + } + fn visit_i64_ctz(&mut self) -> Self::Output { + self.check_unary_op(ValType::I64) + } + fn visit_i64_popcnt(&mut self) -> Self::Output { + self.check_unary_op(ValType::I64) + } + fn visit_i64_add(&mut self) -> Self::Output { + self.check_binary_op(ValType::I64) + } + fn visit_i64_sub(&mut self) -> Self::Output { + self.check_binary_op(ValType::I64) + } + fn visit_i64_mul(&mut self) -> Self::Output { + self.check_binary_op(ValType::I64) + } + fn visit_i64_div_s(&mut self) -> Self::Output { + self.check_binary_op(ValType::I64) + } + fn visit_i64_div_u(&mut self) -> Self::Output { + self.check_binary_op(ValType::I64) + } + fn visit_i64_rem_s(&mut self) -> Self::Output { + self.check_binary_op(ValType::I64) + } + fn visit_i64_rem_u(&mut self) -> Self::Output { + self.check_binary_op(ValType::I64) + } + fn visit_i64_and(&mut self) -> Self::Output { + self.check_binary_op(ValType::I64) + } + fn visit_i64_or(&mut self) -> Self::Output { + self.check_binary_op(ValType::I64) + } + fn visit_i64_xor(&mut self) -> Self::Output { + self.check_binary_op(ValType::I64) + } + fn visit_i64_shl(&mut self) -> Self::Output { + self.check_binary_op(ValType::I64) + } + fn visit_i64_shr_s(&mut self) -> Self::Output { + self.check_binary_op(ValType::I64) + } + fn visit_i64_shr_u(&mut self) -> Self::Output { + self.check_binary_op(ValType::I64) + } + fn visit_i64_rotl(&mut self) -> Self::Output { + self.check_binary_op(ValType::I64) + } + fn visit_i64_rotr(&mut self) -> Self::Output { + self.check_binary_op(ValType::I64) + } + fn visit_f32_abs(&mut self) -> Self::Output { + self.check_funary_op(ValType::F32) + } + fn visit_f32_neg(&mut self) -> Self::Output { + self.check_funary_op(ValType::F32) + } + fn visit_f32_ceil(&mut self) -> Self::Output { + self.check_funary_op(ValType::F32) + } + fn visit_f32_floor(&mut self) -> Self::Output { + self.check_funary_op(ValType::F32) + } + fn visit_f32_trunc(&mut self) -> Self::Output { + self.check_funary_op(ValType::F32) + } + fn visit_f32_nearest(&mut self) -> Self::Output { + self.check_funary_op(ValType::F32) + } + fn visit_f32_sqrt(&mut self) -> Self::Output { + self.check_funary_op(ValType::F32) + } + fn visit_f32_add(&mut self) -> Self::Output { + self.check_fbinary_op(ValType::F32) + } + fn visit_f32_sub(&mut self) -> Self::Output { + self.check_fbinary_op(ValType::F32) + } + fn visit_f32_mul(&mut self) -> Self::Output { + self.check_fbinary_op(ValType::F32) + } + fn visit_f32_div(&mut self) -> Self::Output { + self.check_fbinary_op(ValType::F32) + } + fn visit_f32_min(&mut self) -> Self::Output { + self.check_fbinary_op(ValType::F32) + } + fn visit_f32_max(&mut self) -> Self::Output { + self.check_fbinary_op(ValType::F32) + } + fn visit_f32_copysign(&mut self) -> Self::Output { + self.check_fbinary_op(ValType::F32) + } + fn visit_f64_abs(&mut self) -> Self::Output { + self.check_funary_op(ValType::F64) + } + fn visit_f64_neg(&mut self) -> Self::Output { + self.check_funary_op(ValType::F64) + } + fn visit_f64_ceil(&mut self) -> Self::Output { + self.check_funary_op(ValType::F64) + } + fn visit_f64_floor(&mut self) -> Self::Output { + self.check_funary_op(ValType::F64) + } + fn visit_f64_trunc(&mut self) -> Self::Output { + self.check_funary_op(ValType::F64) + } + fn visit_f64_nearest(&mut self) -> Self::Output { + self.check_funary_op(ValType::F64) + } + fn visit_f64_sqrt(&mut self) -> Self::Output { + self.check_funary_op(ValType::F64) + } + fn visit_f64_add(&mut self) -> Self::Output { + self.check_fbinary_op(ValType::F64) + } + fn visit_f64_sub(&mut self) -> Self::Output { + self.check_fbinary_op(ValType::F64) + } + fn visit_f64_mul(&mut self) -> Self::Output { + self.check_fbinary_op(ValType::F64) + } + fn visit_f64_div(&mut self) -> Self::Output { + self.check_fbinary_op(ValType::F64) + } + fn visit_f64_min(&mut self) -> Self::Output { + self.check_fbinary_op(ValType::F64) + } + fn visit_f64_max(&mut self) -> Self::Output { + self.check_fbinary_op(ValType::F64) + } + fn visit_f64_copysign(&mut self) -> Self::Output { + self.check_fbinary_op(ValType::F64) + } + fn visit_i32_wrap_i64(&mut self) -> Self::Output { + self.check_conversion_op(ValType::I32, ValType::I64) + } + fn visit_i32_trunc_f32_s(&mut self) -> Self::Output { + self.check_conversion_op(ValType::I32, ValType::F32) + } + fn visit_i32_trunc_f32_u(&mut self) -> Self::Output { + self.check_conversion_op(ValType::I32, ValType::F32) + } + fn visit_i32_trunc_f64_s(&mut self) -> Self::Output { + self.check_conversion_op(ValType::I32, ValType::F64) + } + fn visit_i32_trunc_f64_u(&mut self) -> Self::Output { + self.check_conversion_op(ValType::I32, ValType::F64) + } + fn visit_i64_extend_i32_s(&mut self) -> Self::Output { + self.check_conversion_op(ValType::I64, ValType::I32) + } + fn visit_i64_extend_i32_u(&mut self) -> Self::Output { + self.check_conversion_op(ValType::I64, ValType::I32) + } + fn visit_i64_trunc_f32_s(&mut self) -> Self::Output { + self.check_conversion_op(ValType::I64, ValType::F32) + } + fn visit_i64_trunc_f32_u(&mut self) -> Self::Output { + self.check_conversion_op(ValType::I64, ValType::F32) + } + fn visit_i64_trunc_f64_s(&mut self) -> Self::Output { + self.check_conversion_op(ValType::I64, ValType::F64) + } + fn visit_i64_trunc_f64_u(&mut self) -> Self::Output { + self.check_conversion_op(ValType::I64, ValType::F64) + } + fn visit_f32_convert_i32_s(&mut self) -> Self::Output { + self.check_fconversion_op(ValType::F32, ValType::I32) + } + fn visit_f32_convert_i32_u(&mut self) -> Self::Output { + self.check_fconversion_op(ValType::F32, ValType::I32) + } + fn visit_f32_convert_i64_s(&mut self) -> Self::Output { + self.check_fconversion_op(ValType::F32, ValType::I64) + } + fn visit_f32_convert_i64_u(&mut self) -> Self::Output { + self.check_fconversion_op(ValType::F32, ValType::I64) + } + fn visit_f32_demote_f64(&mut self) -> Self::Output { + self.check_fconversion_op(ValType::F32, ValType::F64) + } + fn visit_f64_convert_i32_s(&mut self) -> Self::Output { + self.check_fconversion_op(ValType::F64, ValType::I32) + } + fn visit_f64_convert_i32_u(&mut self) -> Self::Output { + self.check_fconversion_op(ValType::F64, ValType::I32) + } + fn visit_f64_convert_i64_s(&mut self) -> Self::Output { + self.check_fconversion_op(ValType::F64, ValType::I64) + } + fn visit_f64_convert_i64_u(&mut self) -> Self::Output { + self.check_fconversion_op(ValType::F64, ValType::I64) + } + fn visit_f64_promote_f32(&mut self) -> Self::Output { + self.check_fconversion_op(ValType::F64, ValType::F32) + } + fn visit_i32_reinterpret_f32(&mut self) -> Self::Output { + self.check_conversion_op(ValType::I32, ValType::F32) + } + fn visit_i64_reinterpret_f64(&mut self) -> Self::Output { + self.check_conversion_op(ValType::I64, ValType::F64) + } + fn visit_f32_reinterpret_i32(&mut self) -> Self::Output { + self.check_fconversion_op(ValType::F32, ValType::I32) + } + fn visit_f64_reinterpret_i64(&mut self) -> Self::Output { + self.check_fconversion_op(ValType::F64, ValType::I64) + } + fn visit_i32_trunc_sat_f32_s(&mut self) -> Self::Output { + self.check_conversion_op(ValType::I32, ValType::F32) + } + fn visit_i32_trunc_sat_f32_u(&mut self) -> Self::Output { + self.check_conversion_op(ValType::I32, ValType::F32) + } + fn visit_i32_trunc_sat_f64_s(&mut self) -> Self::Output { + self.check_conversion_op(ValType::I32, ValType::F64) + } + fn visit_i32_trunc_sat_f64_u(&mut self) -> Self::Output { + self.check_conversion_op(ValType::I32, ValType::F64) + } + fn visit_i64_trunc_sat_f32_s(&mut self) -> Self::Output { + self.check_conversion_op(ValType::I64, ValType::F32) + } + fn visit_i64_trunc_sat_f32_u(&mut self) -> Self::Output { + self.check_conversion_op(ValType::I64, ValType::F32) + } + fn visit_i64_trunc_sat_f64_s(&mut self) -> Self::Output { + self.check_conversion_op(ValType::I64, ValType::F64) + } + fn visit_i64_trunc_sat_f64_u(&mut self) -> Self::Output { + self.check_conversion_op(ValType::I64, ValType::F64) + } + fn visit_i32_extend8_s(&mut self) -> Self::Output { + self.check_unary_op(ValType::I32) + } + fn visit_i32_extend16_s(&mut self) -> Self::Output { + self.check_unary_op(ValType::I32) + } + fn visit_i64_extend8_s(&mut self) -> Self::Output { + self.check_unary_op(ValType::I64) + } + fn visit_i64_extend16_s(&mut self) -> Self::Output { + self.check_unary_op(ValType::I64) + } + fn visit_i64_extend32_s(&mut self) -> Self::Output { + self.check_unary_op(ValType::I64) + } + fn visit_i32_atomic_load(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_load(memarg, ValType::I32) + } + fn visit_i32_atomic_load16_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_load(memarg, ValType::I32) + } + fn visit_i32_atomic_load8_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_load(memarg, ValType::I32) + } + fn visit_i64_atomic_load(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_load(memarg, ValType::I64) + } + fn visit_i64_atomic_load32_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_load(memarg, ValType::I64) + } + fn visit_i64_atomic_load16_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_load(memarg, ValType::I64) + } + fn visit_i64_atomic_load8_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_load(memarg, ValType::I64) + } + fn visit_i32_atomic_store(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_store(memarg, ValType::I32) + } + fn visit_i32_atomic_store16(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_store(memarg, ValType::I32) + } + fn visit_i32_atomic_store8(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_store(memarg, ValType::I32) + } + fn visit_i64_atomic_store(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_store(memarg, ValType::I64) + } + fn visit_i64_atomic_store32(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_store(memarg, ValType::I64) + } + fn visit_i64_atomic_store16(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_store(memarg, ValType::I64) + } + fn visit_i64_atomic_store8(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_store(memarg, ValType::I64) + } + fn visit_i32_atomic_rmw_add(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I32) + } + fn visit_i32_atomic_rmw_sub(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I32) + } + fn visit_i32_atomic_rmw_and(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I32) + } + fn visit_i32_atomic_rmw_or(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I32) + } + fn visit_i32_atomic_rmw_xor(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I32) + } + fn visit_i32_atomic_rmw16_add_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I32) + } + fn visit_i32_atomic_rmw16_sub_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I32) + } + fn visit_i32_atomic_rmw16_and_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I32) + } + fn visit_i32_atomic_rmw16_or_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I32) + } + fn visit_i32_atomic_rmw16_xor_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I32) + } + fn visit_i32_atomic_rmw8_add_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I32) + } + fn visit_i32_atomic_rmw8_sub_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I32) + } + fn visit_i32_atomic_rmw8_and_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I32) + } + fn visit_i32_atomic_rmw8_or_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I32) + } + fn visit_i32_atomic_rmw8_xor_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I32) + } + fn visit_i64_atomic_rmw_add(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw_sub(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw_and(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw_or(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw_xor(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw32_add_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw32_sub_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw32_and_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw32_or_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw32_xor_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw16_add_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw16_sub_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw16_and_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw16_or_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw16_xor_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw8_add_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw8_sub_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw8_and_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw8_or_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw8_xor_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I64) + } + fn visit_i32_atomic_rmw_xchg(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I32) + } + fn visit_i32_atomic_rmw16_xchg_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I32) + } + fn visit_i32_atomic_rmw8_xchg_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I32) + } + fn visit_i32_atomic_rmw_cmpxchg(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_cmpxchg(memarg, ValType::I32) + } + fn visit_i32_atomic_rmw16_cmpxchg_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_cmpxchg(memarg, ValType::I32) + } + fn visit_i32_atomic_rmw8_cmpxchg_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_cmpxchg(memarg, ValType::I32) + } + fn visit_i64_atomic_rmw_xchg(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw32_xchg_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw16_xchg_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw8_xchg_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw_cmpxchg(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_cmpxchg(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw32_cmpxchg_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_cmpxchg(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw16_cmpxchg_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_cmpxchg(memarg, ValType::I64) + } + fn visit_i64_atomic_rmw8_cmpxchg_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_cmpxchg(memarg, ValType::I64) + } + fn visit_memory_atomic_notify(&mut self, memarg: MemArg) -> Self::Output { + self.check_atomic_binary_op(memarg, ValType::I32) + } + fn visit_memory_atomic_wait32(&mut self, memarg: MemArg) -> Self::Output { + let ty = self.check_shared_memarg(memarg)?; + self.pop_operand(Some(ValType::I64))?; + self.pop_operand(Some(ValType::I32))?; + self.pop_operand(Some(ty))?; + self.push_operand(ValType::I32)?; + Ok(()) + } + fn visit_memory_atomic_wait64(&mut self, memarg: MemArg) -> Self::Output { + let ty = self.check_shared_memarg(memarg)?; + self.pop_operand(Some(ValType::I64))?; + self.pop_operand(Some(ValType::I64))?; + self.pop_operand(Some(ty))?; + self.push_operand(ValType::I32)?; + Ok(()) + } + fn visit_atomic_fence(&mut self) -> Self::Output { + Ok(()) + } + fn visit_ref_null(&mut self, heap_type: HeapType) -> Self::Output { + self.resources + .check_heap_type(heap_type, &self.features, self.offset)?; + self.push_operand(ValType::Ref(RefType { + nullable: true, + heap_type, + }))?; + Ok(()) + } + + fn visit_ref_as_non_null(&mut self) -> Self::Output { + let ty = match self.pop_ref()? { + Some(ty) => MaybeType::Type(ValType::Ref(RefType { + nullable: false, + heap_type: ty.heap_type, + })), + None => MaybeType::HeapBot, + }; + self.push_operand(ty)?; + Ok(()) + } + fn visit_br_on_null(&mut self, relative_depth: u32) -> Self::Output { + let ty = match self.pop_ref()? { + None => MaybeType::HeapBot, + Some(ty) => MaybeType::Type(ValType::Ref(RefType { + nullable: false, + heap_type: ty.heap_type, + })), + }; + let (ft, kind) = self.jump(relative_depth)?; + for ty in self.label_types(ft, kind)?.rev() { + self.pop_operand(Some(ty))?; + } + for ty in self.label_types(ft, kind)? { + self.push_operand(ty)?; + } + self.push_operand(ty)?; + Ok(()) + } + fn visit_br_on_non_null(&mut self, relative_depth: u32) -> Self::Output { + let ty = self.pop_ref()?; + let (ft, kind) = self.jump(relative_depth)?; + let mut lts = self.label_types(ft, kind)?; + match (lts.next_back(), ty) { + (None, _) => bail!( + self.offset, + "type mismatch: br_on_non_null target has no label types", + ), + (Some(ValType::Ref(_)), None) => {} + (Some(rt1 @ ValType::Ref(_)), Some(rt0)) => { + // Switch rt0, our popped type, to a non-nullable type and + // perform the match because if the branch is taken it's a + // non-null value. + let ty = RefType { + nullable: false, + heap_type: rt0.heap_type, + }; + if !self.resources.matches(ty.into(), rt1) { + bail!( + self.offset, + "type mismatch: expected {} but found {}", + ty_to_str(rt0.into()), + ty_to_str(rt1) + ) + } + } + (Some(_), _) => bail!( + self.offset, + "type mismatch: br_on_non_null target does not end with heap type", + ), + } + for ty in self.label_types(ft, kind)?.rev().skip(1) { + self.pop_operand(Some(ty))?; + } + for ty in lts { + self.push_operand(ty)?; + } + Ok(()) + } + fn visit_ref_is_null(&mut self) -> Self::Output { + self.pop_ref()?; + self.push_operand(ValType::I32)?; + Ok(()) + } + fn visit_ref_func(&mut self, function_index: u32) -> Self::Output { + let type_index = match self.resources.type_index_of_function(function_index) { + Some(idx) => idx, + None => bail!( + self.offset, + "unknown function {}: function index out of bounds", + function_index, + ), + }; + if !self.resources.is_function_referenced(function_index) { + bail!(self.offset, "undeclared function reference"); + } + + // FIXME(#924) this should not be conditional based on enabled + // proposals. + if self.features.function_references { + let heap_type = HeapType::TypedFunc(match type_index.try_into() { + Ok(packed) => packed, + Err(_) => { + bail!(self.offset, "type index of `ref.func` target too large") + } + }); + self.push_operand(ValType::Ref(RefType { + nullable: false, + heap_type, + }))?; + } else { + self.push_operand(ValType::FUNCREF)?; + } + Ok(()) + } + fn visit_v128_load(&mut self, memarg: MemArg) -> Self::Output { + let ty = self.check_memarg(memarg)?; + self.pop_operand(Some(ty))?; + self.push_operand(ValType::V128)?; + Ok(()) + } + fn visit_v128_store(&mut self, memarg: MemArg) -> Self::Output { + let ty = self.check_memarg(memarg)?; + self.pop_operand(Some(ValType::V128))?; + self.pop_operand(Some(ty))?; + Ok(()) + } + fn visit_v128_const(&mut self, _value: V128) -> Self::Output { + self.push_operand(ValType::V128)?; + Ok(()) + } + fn visit_i8x16_splat(&mut self) -> Self::Output { + self.check_v128_splat(ValType::I32) + } + fn visit_i16x8_splat(&mut self) -> Self::Output { + self.check_v128_splat(ValType::I32) + } + fn visit_i32x4_splat(&mut self) -> Self::Output { + self.check_v128_splat(ValType::I32) + } + fn visit_i64x2_splat(&mut self) -> Self::Output { + self.check_v128_splat(ValType::I64) + } + fn visit_f32x4_splat(&mut self) -> Self::Output { + self.check_floats_enabled()?; + self.check_v128_splat(ValType::F32) + } + fn visit_f64x2_splat(&mut self) -> Self::Output { + self.check_floats_enabled()?; + self.check_v128_splat(ValType::F64) + } + fn visit_i8x16_extract_lane_s(&mut self, lane: u8) -> Self::Output { + self.check_simd_lane_index(lane, 16)?; + self.pop_operand(Some(ValType::V128))?; + self.push_operand(ValType::I32)?; + Ok(()) + } + fn visit_i8x16_extract_lane_u(&mut self, lane: u8) -> Self::Output { + self.visit_i8x16_extract_lane_s(lane) + } + fn visit_i16x8_extract_lane_s(&mut self, lane: u8) -> Self::Output { + self.check_simd_lane_index(lane, 8)?; + self.pop_operand(Some(ValType::V128))?; + self.push_operand(ValType::I32)?; + Ok(()) + } + fn visit_i16x8_extract_lane_u(&mut self, lane: u8) -> Self::Output { + self.visit_i16x8_extract_lane_s(lane) + } + fn visit_i32x4_extract_lane(&mut self, lane: u8) -> Self::Output { + self.check_simd_lane_index(lane, 4)?; + self.pop_operand(Some(ValType::V128))?; + self.push_operand(ValType::I32)?; + Ok(()) + } + fn visit_i8x16_replace_lane(&mut self, lane: u8) -> Self::Output { + self.check_simd_lane_index(lane, 16)?; + self.pop_operand(Some(ValType::I32))?; + self.pop_operand(Some(ValType::V128))?; + self.push_operand(ValType::V128)?; + Ok(()) + } + fn visit_i16x8_replace_lane(&mut self, lane: u8) -> Self::Output { + self.check_simd_lane_index(lane, 8)?; + self.pop_operand(Some(ValType::I32))?; + self.pop_operand(Some(ValType::V128))?; + self.push_operand(ValType::V128)?; + Ok(()) + } + fn visit_i32x4_replace_lane(&mut self, lane: u8) -> Self::Output { + self.check_simd_lane_index(lane, 4)?; + self.pop_operand(Some(ValType::I32))?; + self.pop_operand(Some(ValType::V128))?; + self.push_operand(ValType::V128)?; + Ok(()) + } + fn visit_i64x2_extract_lane(&mut self, lane: u8) -> Self::Output { + self.check_simd_lane_index(lane, 2)?; + self.pop_operand(Some(ValType::V128))?; + self.push_operand(ValType::I64)?; + Ok(()) + } + fn visit_i64x2_replace_lane(&mut self, lane: u8) -> Self::Output { + self.check_simd_lane_index(lane, 2)?; + self.pop_operand(Some(ValType::I64))?; + self.pop_operand(Some(ValType::V128))?; + self.push_operand(ValType::V128)?; + Ok(()) + } + fn visit_f32x4_extract_lane(&mut self, lane: u8) -> Self::Output { + self.check_floats_enabled()?; + self.check_simd_lane_index(lane, 4)?; + self.pop_operand(Some(ValType::V128))?; + self.push_operand(ValType::F32)?; + Ok(()) + } + fn visit_f32x4_replace_lane(&mut self, lane: u8) -> Self::Output { + self.check_floats_enabled()?; + self.check_simd_lane_index(lane, 4)?; + self.pop_operand(Some(ValType::F32))?; + self.pop_operand(Some(ValType::V128))?; + self.push_operand(ValType::V128)?; + Ok(()) + } + fn visit_f64x2_extract_lane(&mut self, lane: u8) -> Self::Output { + self.check_floats_enabled()?; + self.check_simd_lane_index(lane, 2)?; + self.pop_operand(Some(ValType::V128))?; + self.push_operand(ValType::F64)?; + Ok(()) + } + fn visit_f64x2_replace_lane(&mut self, lane: u8) -> Self::Output { + self.check_floats_enabled()?; + self.check_simd_lane_index(lane, 2)?; + self.pop_operand(Some(ValType::F64))?; + self.pop_operand(Some(ValType::V128))?; + self.push_operand(ValType::V128)?; + Ok(()) + } + fn visit_f32x4_eq(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f32x4_ne(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f32x4_lt(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f32x4_gt(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f32x4_le(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f32x4_ge(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f64x2_eq(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f64x2_ne(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f64x2_lt(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f64x2_gt(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f64x2_le(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f64x2_ge(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f32x4_add(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f32x4_sub(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f32x4_mul(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f32x4_div(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f32x4_min(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f32x4_max(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f32x4_pmin(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f32x4_pmax(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f64x2_add(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f64x2_sub(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f64x2_mul(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f64x2_div(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f64x2_min(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f64x2_max(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f64x2_pmin(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_f64x2_pmax(&mut self) -> Self::Output { + self.check_v128_fbinary_op() + } + fn visit_i8x16_eq(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i8x16_ne(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i8x16_lt_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i8x16_lt_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i8x16_gt_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i8x16_gt_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i8x16_le_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i8x16_le_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i8x16_ge_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i8x16_ge_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_eq(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_ne(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_lt_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_lt_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_gt_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_gt_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_le_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_le_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_ge_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_ge_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i32x4_eq(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i32x4_ne(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i32x4_lt_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i32x4_lt_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i32x4_gt_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i32x4_gt_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i32x4_le_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i32x4_le_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i32x4_ge_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i32x4_ge_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i64x2_eq(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i64x2_ne(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i64x2_lt_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i64x2_gt_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i64x2_le_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i64x2_ge_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_v128_and(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_v128_andnot(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_v128_or(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_v128_xor(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i8x16_add(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i8x16_add_sat_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i8x16_add_sat_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i8x16_sub(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i8x16_sub_sat_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i8x16_sub_sat_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i8x16_min_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i8x16_min_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i8x16_max_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i8x16_max_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_add(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_add_sat_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_add_sat_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_sub(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_sub_sat_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_sub_sat_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_mul(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_min_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_min_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_max_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_max_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i32x4_add(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i32x4_sub(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i32x4_mul(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i32x4_min_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i32x4_min_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i32x4_max_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i32x4_max_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i32x4_dot_i16x8_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i64x2_add(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i64x2_sub(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i64x2_mul(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i8x16_avgr_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_avgr_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i8x16_narrow_i16x8_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i8x16_narrow_i16x8_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_narrow_i32x4_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_narrow_i32x4_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_extmul_low_i8x16_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_extmul_high_i8x16_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_extmul_low_i8x16_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_extmul_high_i8x16_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i32x4_extmul_low_i16x8_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i32x4_extmul_high_i16x8_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i32x4_extmul_low_i16x8_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i32x4_extmul_high_i16x8_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i64x2_extmul_low_i32x4_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i64x2_extmul_high_i32x4_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i64x2_extmul_low_i32x4_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i64x2_extmul_high_i32x4_u(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_q15mulr_sat_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_f32x4_ceil(&mut self) -> Self::Output { + self.check_v128_funary_op() + } + fn visit_f32x4_floor(&mut self) -> Self::Output { + self.check_v128_funary_op() + } + fn visit_f32x4_trunc(&mut self) -> Self::Output { + self.check_v128_funary_op() + } + fn visit_f32x4_nearest(&mut self) -> Self::Output { + self.check_v128_funary_op() + } + fn visit_f64x2_ceil(&mut self) -> Self::Output { + self.check_v128_funary_op() + } + fn visit_f64x2_floor(&mut self) -> Self::Output { + self.check_v128_funary_op() + } + fn visit_f64x2_trunc(&mut self) -> Self::Output { + self.check_v128_funary_op() + } + fn visit_f64x2_nearest(&mut self) -> Self::Output { + self.check_v128_funary_op() + } + fn visit_f32x4_abs(&mut self) -> Self::Output { + self.check_v128_funary_op() + } + fn visit_f32x4_neg(&mut self) -> Self::Output { + self.check_v128_funary_op() + } + fn visit_f32x4_sqrt(&mut self) -> Self::Output { + self.check_v128_funary_op() + } + fn visit_f64x2_abs(&mut self) -> Self::Output { + self.check_v128_funary_op() + } + fn visit_f64x2_neg(&mut self) -> Self::Output { + self.check_v128_funary_op() + } + fn visit_f64x2_sqrt(&mut self) -> Self::Output { + self.check_v128_funary_op() + } + fn visit_f32x4_demote_f64x2_zero(&mut self) -> Self::Output { + self.check_v128_funary_op() + } + fn visit_f64x2_promote_low_f32x4(&mut self) -> Self::Output { + self.check_v128_funary_op() + } + fn visit_f64x2_convert_low_i32x4_s(&mut self) -> Self::Output { + self.check_v128_funary_op() + } + fn visit_f64x2_convert_low_i32x4_u(&mut self) -> Self::Output { + self.check_v128_funary_op() + } + fn visit_i32x4_trunc_sat_f32x4_s(&mut self) -> Self::Output { + self.check_v128_funary_op() + } + fn visit_i32x4_trunc_sat_f32x4_u(&mut self) -> Self::Output { + self.check_v128_funary_op() + } + fn visit_i32x4_trunc_sat_f64x2_s_zero(&mut self) -> Self::Output { + self.check_v128_funary_op() + } + fn visit_i32x4_trunc_sat_f64x2_u_zero(&mut self) -> Self::Output { + self.check_v128_funary_op() + } + fn visit_f32x4_convert_i32x4_s(&mut self) -> Self::Output { + self.check_v128_funary_op() + } + fn visit_f32x4_convert_i32x4_u(&mut self) -> Self::Output { + self.check_v128_funary_op() + } + fn visit_v128_not(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i8x16_abs(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i8x16_neg(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i8x16_popcnt(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i16x8_abs(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i16x8_neg(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i32x4_abs(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i32x4_neg(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i64x2_abs(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i64x2_neg(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i16x8_extend_low_i8x16_s(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i16x8_extend_high_i8x16_s(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i16x8_extend_low_i8x16_u(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i16x8_extend_high_i8x16_u(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i32x4_extend_low_i16x8_s(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i32x4_extend_high_i16x8_s(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i32x4_extend_low_i16x8_u(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i32x4_extend_high_i16x8_u(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i64x2_extend_low_i32x4_s(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i64x2_extend_high_i32x4_s(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i64x2_extend_low_i32x4_u(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i64x2_extend_high_i32x4_u(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i16x8_extadd_pairwise_i8x16_s(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i16x8_extadd_pairwise_i8x16_u(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i32x4_extadd_pairwise_i16x8_s(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i32x4_extadd_pairwise_i16x8_u(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_v128_bitselect(&mut self) -> Self::Output { + self.pop_operand(Some(ValType::V128))?; + self.pop_operand(Some(ValType::V128))?; + self.pop_operand(Some(ValType::V128))?; + self.push_operand(ValType::V128)?; + Ok(()) + } + fn visit_i8x16_relaxed_swizzle(&mut self) -> Self::Output { + self.pop_operand(Some(ValType::V128))?; + self.pop_operand(Some(ValType::V128))?; + self.push_operand(ValType::V128)?; + Ok(()) + } + fn visit_i32x4_relaxed_trunc_f32x4_s(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i32x4_relaxed_trunc_f32x4_u(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i32x4_relaxed_trunc_f64x2_s_zero(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_i32x4_relaxed_trunc_f64x2_u_zero(&mut self) -> Self::Output { + self.check_v128_unary_op() + } + fn visit_f32x4_relaxed_madd(&mut self) -> Self::Output { + self.check_v128_ternary_op() + } + fn visit_f32x4_relaxed_nmadd(&mut self) -> Self::Output { + self.check_v128_ternary_op() + } + fn visit_f64x2_relaxed_madd(&mut self) -> Self::Output { + self.check_v128_ternary_op() + } + fn visit_f64x2_relaxed_nmadd(&mut self) -> Self::Output { + self.check_v128_ternary_op() + } + fn visit_i8x16_relaxed_laneselect(&mut self) -> Self::Output { + self.check_v128_ternary_op() + } + fn visit_i16x8_relaxed_laneselect(&mut self) -> Self::Output { + self.check_v128_ternary_op() + } + fn visit_i32x4_relaxed_laneselect(&mut self) -> Self::Output { + self.check_v128_ternary_op() + } + fn visit_i64x2_relaxed_laneselect(&mut self) -> Self::Output { + self.check_v128_ternary_op() + } + fn visit_f32x4_relaxed_min(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_f32x4_relaxed_max(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_f64x2_relaxed_min(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_f64x2_relaxed_max(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_relaxed_q15mulr_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i16x8_relaxed_dot_i8x16_i7x16_s(&mut self) -> Self::Output { + self.check_v128_binary_op() + } + fn visit_i32x4_relaxed_dot_i8x16_i7x16_add_s(&mut self) -> Self::Output { + self.check_v128_ternary_op() + } + fn visit_v128_any_true(&mut self) -> Self::Output { + self.check_v128_bitmask_op() + } + fn visit_i8x16_all_true(&mut self) -> Self::Output { + self.check_v128_bitmask_op() + } + fn visit_i8x16_bitmask(&mut self) -> Self::Output { + self.check_v128_bitmask_op() + } + fn visit_i16x8_all_true(&mut self) -> Self::Output { + self.check_v128_bitmask_op() + } + fn visit_i16x8_bitmask(&mut self) -> Self::Output { + self.check_v128_bitmask_op() + } + fn visit_i32x4_all_true(&mut self) -> Self::Output { + self.check_v128_bitmask_op() + } + fn visit_i32x4_bitmask(&mut self) -> Self::Output { + self.check_v128_bitmask_op() + } + fn visit_i64x2_all_true(&mut self) -> Self::Output { + self.check_v128_bitmask_op() + } + fn visit_i64x2_bitmask(&mut self) -> Self::Output { + self.check_v128_bitmask_op() + } + fn visit_i8x16_shl(&mut self) -> Self::Output { + self.check_v128_shift_op() + } + fn visit_i8x16_shr_s(&mut self) -> Self::Output { + self.check_v128_shift_op() + } + fn visit_i8x16_shr_u(&mut self) -> Self::Output { + self.check_v128_shift_op() + } + fn visit_i16x8_shl(&mut self) -> Self::Output { + self.check_v128_shift_op() + } + fn visit_i16x8_shr_s(&mut self) -> Self::Output { + self.check_v128_shift_op() + } + fn visit_i16x8_shr_u(&mut self) -> Self::Output { + self.check_v128_shift_op() + } + fn visit_i32x4_shl(&mut self) -> Self::Output { + self.check_v128_shift_op() + } + fn visit_i32x4_shr_s(&mut self) -> Self::Output { + self.check_v128_shift_op() + } + fn visit_i32x4_shr_u(&mut self) -> Self::Output { + self.check_v128_shift_op() + } + fn visit_i64x2_shl(&mut self) -> Self::Output { + self.check_v128_shift_op() + } + fn visit_i64x2_shr_s(&mut self) -> Self::Output { + self.check_v128_shift_op() + } + fn visit_i64x2_shr_u(&mut self) -> Self::Output { + self.check_v128_shift_op() + } + fn visit_i8x16_swizzle(&mut self) -> Self::Output { + self.pop_operand(Some(ValType::V128))?; + self.pop_operand(Some(ValType::V128))?; + self.push_operand(ValType::V128)?; + Ok(()) + } + fn visit_i8x16_shuffle(&mut self, lanes: [u8; 16]) -> Self::Output { + self.pop_operand(Some(ValType::V128))?; + self.pop_operand(Some(ValType::V128))?; + for i in lanes { + self.check_simd_lane_index(i, 32)?; + } + self.push_operand(ValType::V128)?; + Ok(()) + } + fn visit_v128_load8_splat(&mut self, memarg: MemArg) -> Self::Output { + let ty = self.check_memarg(memarg)?; + self.pop_operand(Some(ty))?; + self.push_operand(ValType::V128)?; + Ok(()) + } + fn visit_v128_load16_splat(&mut self, memarg: MemArg) -> Self::Output { + let ty = self.check_memarg(memarg)?; + self.pop_operand(Some(ty))?; + self.push_operand(ValType::V128)?; + Ok(()) + } + fn visit_v128_load32_splat(&mut self, memarg: MemArg) -> Self::Output { + let ty = self.check_memarg(memarg)?; + self.pop_operand(Some(ty))?; + self.push_operand(ValType::V128)?; + Ok(()) + } + fn visit_v128_load32_zero(&mut self, memarg: MemArg) -> Self::Output { + self.visit_v128_load32_splat(memarg) + } + fn visit_v128_load64_splat(&mut self, memarg: MemArg) -> Self::Output { + self.check_v128_load_op(memarg) + } + fn visit_v128_load64_zero(&mut self, memarg: MemArg) -> Self::Output { + self.check_v128_load_op(memarg) + } + fn visit_v128_load8x8_s(&mut self, memarg: MemArg) -> Self::Output { + self.check_v128_load_op(memarg) + } + fn visit_v128_load8x8_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_v128_load_op(memarg) + } + fn visit_v128_load16x4_s(&mut self, memarg: MemArg) -> Self::Output { + self.check_v128_load_op(memarg) + } + fn visit_v128_load16x4_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_v128_load_op(memarg) + } + fn visit_v128_load32x2_s(&mut self, memarg: MemArg) -> Self::Output { + self.check_v128_load_op(memarg) + } + fn visit_v128_load32x2_u(&mut self, memarg: MemArg) -> Self::Output { + self.check_v128_load_op(memarg) + } + fn visit_v128_load8_lane(&mut self, memarg: MemArg, lane: u8) -> Self::Output { + let idx = self.check_memarg(memarg)?; + self.check_simd_lane_index(lane, 16)?; + self.pop_operand(Some(ValType::V128))?; + self.pop_operand(Some(idx))?; + self.push_operand(ValType::V128)?; + Ok(()) + } + fn visit_v128_load16_lane(&mut self, memarg: MemArg, lane: u8) -> Self::Output { + let idx = self.check_memarg(memarg)?; + self.check_simd_lane_index(lane, 8)?; + self.pop_operand(Some(ValType::V128))?; + self.pop_operand(Some(idx))?; + self.push_operand(ValType::V128)?; + Ok(()) + } + fn visit_v128_load32_lane(&mut self, memarg: MemArg, lane: u8) -> Self::Output { + let idx = self.check_memarg(memarg)?; + self.check_simd_lane_index(lane, 4)?; + self.pop_operand(Some(ValType::V128))?; + self.pop_operand(Some(idx))?; + self.push_operand(ValType::V128)?; + Ok(()) + } + fn visit_v128_load64_lane(&mut self, memarg: MemArg, lane: u8) -> Self::Output { + let idx = self.check_memarg(memarg)?; + self.check_simd_lane_index(lane, 2)?; + self.pop_operand(Some(ValType::V128))?; + self.pop_operand(Some(idx))?; + self.push_operand(ValType::V128)?; + Ok(()) + } + fn visit_v128_store8_lane(&mut self, memarg: MemArg, lane: u8) -> Self::Output { + let idx = self.check_memarg(memarg)?; + self.check_simd_lane_index(lane, 16)?; + self.pop_operand(Some(ValType::V128))?; + self.pop_operand(Some(idx))?; + Ok(()) + } + fn visit_v128_store16_lane(&mut self, memarg: MemArg, lane: u8) -> Self::Output { + let idx = self.check_memarg(memarg)?; + self.check_simd_lane_index(lane, 8)?; + self.pop_operand(Some(ValType::V128))?; + self.pop_operand(Some(idx))?; + Ok(()) + } + fn visit_v128_store32_lane(&mut self, memarg: MemArg, lane: u8) -> Self::Output { + let idx = self.check_memarg(memarg)?; + self.check_simd_lane_index(lane, 4)?; + self.pop_operand(Some(ValType::V128))?; + self.pop_operand(Some(idx))?; + Ok(()) + } + fn visit_v128_store64_lane(&mut self, memarg: MemArg, lane: u8) -> Self::Output { + let idx = self.check_memarg(memarg)?; + self.check_simd_lane_index(lane, 2)?; + self.pop_operand(Some(ValType::V128))?; + self.pop_operand(Some(idx))?; + Ok(()) + } + fn visit_memory_init(&mut self, segment: u32, mem: u32) -> Self::Output { + let ty = self.check_memory_index(mem)?; + match self.resources.data_count() { + None => bail!(self.offset, "data count section required"), + Some(count) if segment < count => {} + Some(_) => bail!(self.offset, "unknown data segment {}", segment), + } + self.pop_operand(Some(ValType::I32))?; + self.pop_operand(Some(ValType::I32))?; + self.pop_operand(Some(ty))?; + Ok(()) + } + fn visit_data_drop(&mut self, segment: u32) -> Self::Output { + match self.resources.data_count() { + None => bail!(self.offset, "data count section required"), + Some(count) if segment < count => {} + Some(_) => bail!(self.offset, "unknown data segment {}", segment), + } + Ok(()) + } + fn visit_memory_copy(&mut self, dst: u32, src: u32) -> Self::Output { + let dst_ty = self.check_memory_index(dst)?; + let src_ty = self.check_memory_index(src)?; + + // The length operand here is the smaller of src/dst, which is + // i32 if one is i32 + self.pop_operand(Some(match src_ty { + ValType::I32 => ValType::I32, + _ => dst_ty, + }))?; + + // ... and the offset into each memory is required to be + // whatever the indexing type is for that memory + self.pop_operand(Some(src_ty))?; + self.pop_operand(Some(dst_ty))?; + Ok(()) + } + fn visit_memory_fill(&mut self, mem: u32) -> Self::Output { + let ty = self.check_memory_index(mem)?; + self.pop_operand(Some(ty))?; + self.pop_operand(Some(ValType::I32))?; + self.pop_operand(Some(ty))?; + Ok(()) + } + fn visit_memory_discard(&mut self, mem: u32) -> Self::Output { + let ty = self.check_memory_index(mem)?; + self.pop_operand(Some(ty))?; + self.pop_operand(Some(ty))?; + Ok(()) + } + fn visit_table_init(&mut self, segment: u32, table: u32) -> Self::Output { + if table > 0 {} + let table = match self.resources.table_at(table) { + Some(table) => table, + None => bail!( + self.offset, + "unknown table {}: table index out of bounds", + table + ), + }; + let segment_ty = match self.resources.element_type_at(segment) { + Some(ty) => ty, + None => bail!( + self.offset, + "unknown elem segment {}: segment index out of bounds", + segment + ), + }; + if segment_ty != table.element_type { + bail!(self.offset, "type mismatch"); + } + self.pop_operand(Some(ValType::I32))?; + self.pop_operand(Some(ValType::I32))?; + self.pop_operand(Some(ValType::I32))?; + Ok(()) + } + fn visit_elem_drop(&mut self, segment: u32) -> Self::Output { + if segment >= self.resources.element_count() { + bail!( + self.offset, + "unknown elem segment {}: segment index out of bounds", + segment + ); + } + Ok(()) + } + fn visit_table_copy(&mut self, dst_table: u32, src_table: u32) -> Self::Output { + if src_table > 0 || dst_table > 0 {} + let (src, dst) = match ( + self.resources.table_at(src_table), + self.resources.table_at(dst_table), + ) { + (Some(a), Some(b)) => (a, b), + _ => bail!(self.offset, "table index out of bounds"), + }; + if !self.resources.matches( + ValType::Ref(src.element_type), + ValType::Ref(dst.element_type), + ) { + bail!(self.offset, "type mismatch"); + } + self.pop_operand(Some(ValType::I32))?; + self.pop_operand(Some(ValType::I32))?; + self.pop_operand(Some(ValType::I32))?; + Ok(()) + } + fn visit_table_get(&mut self, table: u32) -> Self::Output { + let ty = match self.resources.table_at(table) { + Some(ty) => ty.element_type, + None => bail!(self.offset, "table index out of bounds"), + }; + self.pop_operand(Some(ValType::I32))?; + self.push_operand(ValType::Ref(ty))?; + Ok(()) + } + fn visit_table_set(&mut self, table: u32) -> Self::Output { + let ty = match self.resources.table_at(table) { + Some(ty) => ty.element_type, + None => bail!(self.offset, "table index out of bounds"), + }; + self.pop_operand(Some(ValType::Ref(ty)))?; + self.pop_operand(Some(ValType::I32))?; + Ok(()) + } + fn visit_table_grow(&mut self, table: u32) -> Self::Output { + let ty = match self.resources.table_at(table) { + Some(ty) => ty.element_type, + None => bail!(self.offset, "table index out of bounds"), + }; + self.pop_operand(Some(ValType::I32))?; + self.pop_operand(Some(ValType::Ref(ty)))?; + self.push_operand(ValType::I32)?; + Ok(()) + } + fn visit_table_size(&mut self, table: u32) -> Self::Output { + if self.resources.table_at(table).is_none() { + bail!(self.offset, "table index out of bounds"); + } + self.push_operand(ValType::I32)?; + Ok(()) + } + fn visit_table_fill(&mut self, table: u32) -> Self::Output { + let ty = match self.resources.table_at(table) { + Some(ty) => ty.element_type, + None => bail!(self.offset, "table index out of bounds"), + }; + self.pop_operand(Some(ValType::I32))?; + self.pop_operand(Some(ValType::Ref(ty)))?; + self.pop_operand(Some(ValType::I32))?; + Ok(()) + } +} + +#[derive(Clone)] +enum Either<A, B> { + A(A), + B(B), +} + +impl<A, B> Iterator for Either<A, B> +where + A: Iterator, + B: Iterator<Item = A::Item>, +{ + type Item = A::Item; + fn next(&mut self) -> Option<A::Item> { + match self { + Either::A(a) => a.next(), + Either::B(b) => b.next(), + } + } +} + +impl<A, B> DoubleEndedIterator for Either<A, B> +where + A: DoubleEndedIterator, + B: DoubleEndedIterator<Item = A::Item>, +{ + fn next_back(&mut self) -> Option<A::Item> { + match self { + Either::A(a) => a.next_back(), + Either::B(b) => b.next_back(), + } + } +} + +impl<A, B> ExactSizeIterator for Either<A, B> +where + A: ExactSizeIterator, + B: ExactSizeIterator<Item = A::Item>, +{ + fn len(&self) -> usize { + match self { + Either::A(a) => a.len(), + Either::B(b) => b.len(), + } + } +} + +trait PreciseIterator: ExactSizeIterator + DoubleEndedIterator + Clone {} +impl<T: ExactSizeIterator + DoubleEndedIterator + Clone> PreciseIterator for T {} + +impl Locals { + /// Defines another group of `count` local variables of type `ty`. + /// + /// Returns `true` if the definition was successful. Local variable + /// definition is unsuccessful in case the amount of total variables + /// after definition exceeds the allowed maximum number. + fn define(&mut self, count: u32, ty: ValType) -> bool { + match self.num_locals.checked_add(count) { + Some(n) => self.num_locals = n, + None => return false, + } + if self.num_locals > (MAX_WASM_FUNCTION_LOCALS as u32) { + return false; + } + for _ in 0..count { + if self.first.len() >= MAX_LOCALS_TO_TRACK { + break; + } + self.first.push(ty); + } + self.all.push((self.num_locals - 1, ty)); + true + } + + /// Returns the number of defined local variables. + pub(super) fn len_locals(&self) -> u32 { + self.num_locals + } + + /// Returns the type of the local variable at the given index if any. + #[inline] + pub(super) fn get(&self, idx: u32) -> Option<ValType> { + match self.first.get(idx as usize) { + Some(ty) => Some(*ty), + None => self.get_bsearch(idx), + } + } + + fn get_bsearch(&self, idx: u32) -> Option<ValType> { + match self.all.binary_search_by_key(&idx, |(idx, _)| *idx) { + // If this index would be inserted at the end of the list, then the + // index is out of bounds and we return an error. + Err(i) if i == self.all.len() => None, + + // If `Ok` is returned we found the index exactly, or if `Err` is + // returned the position is the one which is the least index + // greater that `idx`, which is still the type of `idx` according + // to our "compressed" representation. In both cases we access the + // list at index `i`. + Ok(i) | Err(i) => Some(self.all[i].1), + } + } +} diff --git a/third_party/rust/wasmparser/src/validator/types.rs b/third_party/rust/wasmparser/src/validator/types.rs new file mode 100644 index 0000000000..ce0559d34c --- /dev/null +++ b/third_party/rust/wasmparser/src/validator/types.rs @@ -0,0 +1,2166 @@ +//! Types relating to type information provided by validation. + +use super::{component::ComponentState, core::Module}; +use crate::{ + Export, ExternalKind, FuncType, GlobalType, Import, MemoryType, PrimitiveValType, RefType, + TableType, TypeRef, ValType, +}; +use indexmap::{IndexMap, IndexSet}; +use std::collections::HashMap; +use std::{ + borrow::Borrow, + fmt, + hash::{Hash, Hasher}, + mem, + ops::{Deref, DerefMut}, + sync::Arc, +}; +use url::Url; + +/// The maximum number of parameters in the canonical ABI that can be passed by value. +/// +/// Functions that exceed this limit will instead pass parameters indirectly from +/// linear memory via a single pointer parameter. +const MAX_FLAT_FUNC_PARAMS: usize = 16; +/// The maximum number of results in the canonical ABI that can be returned by a function. +/// +/// Functions that exceed this limit have their results written to linear memory via an +/// additional pointer parameter (imports) or return a single pointer value (exports). +const MAX_FLAT_FUNC_RESULTS: usize = 1; + +/// The maximum lowered types, including a possible type for a return pointer parameter. +const MAX_LOWERED_TYPES: usize = MAX_FLAT_FUNC_PARAMS + 1; + +/// Represents a kebab string slice used in validation. +/// +/// This is a wrapper around `str` that ensures the slice is +/// a valid kebab case string according to the component model +/// specification. +/// +/// It also provides an equality and hashing implementation +/// that ignores ASCII case. +#[derive(Debug, Eq)] +#[repr(transparent)] +pub struct KebabStr(str); + +impl KebabStr { + /// Creates a new kebab string slice. + /// + /// Returns `None` if the given string is not a valid kebab string. + pub fn new<'a>(s: impl AsRef<str> + 'a) -> Option<&'a Self> { + let s = Self::new_unchecked(s); + if s.is_kebab_case() { + Some(s) + } else { + None + } + } + + pub(crate) fn new_unchecked<'a>(s: impl AsRef<str> + 'a) -> &'a Self { + // Safety: `KebabStr` is a transparent wrapper around `str` + // Therefore transmuting `&str` to `&KebabStr` is safe. + unsafe { std::mem::transmute::<_, &Self>(s.as_ref()) } + } + + /// Gets the underlying string slice. + pub fn as_str(&self) -> &str { + &self.0 + } + + /// Converts the slice to an owned string. + pub fn to_kebab_string(&self) -> KebabString { + KebabString(self.to_string()) + } + + fn is_kebab_case(&self) -> bool { + let mut lower = false; + let mut upper = false; + for c in self.chars() { + match c { + 'a'..='z' if !lower && !upper => lower = true, + 'A'..='Z' if !lower && !upper => upper = true, + 'a'..='z' if lower => {} + 'A'..='Z' if upper => {} + '0'..='9' if lower || upper => {} + '-' if lower || upper => { + lower = false; + upper = false; + } + _ => return false, + } + } + + !self.is_empty() && !self.ends_with('-') + } +} + +impl Deref for KebabStr { + type Target = str; + + fn deref(&self) -> &str { + self.as_str() + } +} + +impl PartialEq for KebabStr { + fn eq(&self, other: &Self) -> bool { + if self.len() != other.len() { + return false; + } + + self.chars() + .zip(other.chars()) + .all(|(a, b)| a.to_ascii_lowercase() == b.to_ascii_lowercase()) + } +} + +impl PartialEq<KebabString> for KebabStr { + fn eq(&self, other: &KebabString) -> bool { + self.eq(other.as_kebab_str()) + } +} + +impl Hash for KebabStr { + fn hash<H: Hasher>(&self, state: &mut H) { + self.len().hash(state); + + for b in self.chars() { + b.to_ascii_lowercase().hash(state); + } + } +} + +impl fmt::Display for KebabStr { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + (self as &str).fmt(f) + } +} + +impl ToOwned for KebabStr { + type Owned = KebabString; + + fn to_owned(&self) -> Self::Owned { + self.to_kebab_string() + } +} + +/// Represents an owned kebab string for validation. +/// +/// This is a wrapper around `String` that ensures the string is +/// a valid kebab case string according to the component model +/// specification. +/// +/// It also provides an equality and hashing implementation +/// that ignores ASCII case. +#[derive(Debug, Clone, Eq)] +pub struct KebabString(String); + +impl KebabString { + /// Creates a new kebab string. + /// + /// Returns `None` if the given string is not a valid kebab string. + pub fn new(s: impl Into<String>) -> Option<Self> { + let s = s.into(); + if KebabStr::new(&s).is_some() { + Some(Self(s)) + } else { + None + } + } + + /// Gets the underlying string. + pub fn as_str(&self) -> &str { + self.0.as_str() + } + + /// Converts the kebab string to a kebab string slice. + pub fn as_kebab_str(&self) -> &KebabStr { + // Safety: internal string is always valid kebab-case + KebabStr::new_unchecked(self.as_str()) + } +} + +impl Deref for KebabString { + type Target = KebabStr; + + fn deref(&self) -> &Self::Target { + self.as_kebab_str() + } +} + +impl Borrow<KebabStr> for KebabString { + fn borrow(&self) -> &KebabStr { + self.as_kebab_str() + } +} + +impl PartialEq for KebabString { + fn eq(&self, other: &Self) -> bool { + self.as_kebab_str().eq(other.as_kebab_str()) + } +} + +impl PartialEq<KebabStr> for KebabString { + fn eq(&self, other: &KebabStr) -> bool { + self.as_kebab_str().eq(other) + } +} + +impl Hash for KebabString { + fn hash<H: Hasher>(&self, state: &mut H) { + self.as_kebab_str().hash(state) + } +} + +impl fmt::Display for KebabString { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + self.as_kebab_str().fmt(f) + } +} + +impl From<KebabString> for String { + fn from(s: KebabString) -> String { + s.0 + } +} + +/// A simple alloc-free list of types used for calculating lowered function signatures. +pub(crate) struct LoweredTypes { + types: [ValType; MAX_LOWERED_TYPES], + len: usize, + max: usize, +} + +impl LoweredTypes { + fn new(max: usize) -> Self { + assert!(max <= MAX_LOWERED_TYPES); + Self { + types: [ValType::I32; MAX_LOWERED_TYPES], + len: 0, + max, + } + } + + fn len(&self) -> usize { + self.len + } + + fn maxed(&self) -> bool { + self.len == self.max + } + + fn get_mut(&mut self, index: usize) -> Option<&mut ValType> { + if index < self.len { + Some(&mut self.types[index]) + } else { + None + } + } + + fn push(&mut self, ty: ValType) -> bool { + if self.maxed() { + return false; + } + + self.types[self.len] = ty; + self.len += 1; + true + } + + fn clear(&mut self) { + self.len = 0; + } + + pub fn as_slice(&self) -> &[ValType] { + &self.types[..self.len] + } + + pub fn iter(&self) -> impl Iterator<Item = ValType> + '_ { + self.as_slice().iter().copied() + } +} + +/// Represents information about a component function type lowering. +pub(crate) struct LoweringInfo { + pub(crate) params: LoweredTypes, + pub(crate) results: LoweredTypes, + pub(crate) requires_memory: bool, + pub(crate) requires_realloc: bool, +} + +impl LoweringInfo { + pub(crate) fn into_func_type(self) -> FuncType { + FuncType::new( + self.params.as_slice().iter().copied(), + self.results.as_slice().iter().copied(), + ) + } +} + +impl Default for LoweringInfo { + fn default() -> Self { + Self { + params: LoweredTypes::new(MAX_FLAT_FUNC_PARAMS), + results: LoweredTypes::new(MAX_FLAT_FUNC_RESULTS), + requires_memory: false, + requires_realloc: false, + } + } +} + +fn push_primitive_wasm_types(ty: &PrimitiveValType, lowered_types: &mut LoweredTypes) -> bool { + match ty { + PrimitiveValType::Bool + | PrimitiveValType::S8 + | PrimitiveValType::U8 + | PrimitiveValType::S16 + | PrimitiveValType::U16 + | PrimitiveValType::S32 + | PrimitiveValType::U32 + | PrimitiveValType::Char => lowered_types.push(ValType::I32), + PrimitiveValType::S64 | PrimitiveValType::U64 => lowered_types.push(ValType::I64), + PrimitiveValType::Float32 => lowered_types.push(ValType::F32), + PrimitiveValType::Float64 => lowered_types.push(ValType::F64), + PrimitiveValType::String => { + lowered_types.push(ValType::I32) && lowered_types.push(ValType::I32) + } + } +} + +/// Represents a unique identifier for a type known to a [`crate::Validator`]. +#[derive(Debug, Copy, Clone, PartialEq, Eq, Hash)] +pub struct TypeId { + /// The index into the global list of types. + pub(crate) index: usize, + /// The effective type size for the type. + /// + /// This is stored as part of the ID to avoid having to recurse through + /// the global type list when calculating type sizes. + pub(crate) type_size: u32, + /// A unique integer assigned to this type. + /// + /// The purpose of this field is to ensure that two different `TypeId` + /// representations can be handed out for two different aliased types within + /// a component that actually point to the same underlying type (as pointed + /// to by the `index` field). + unique_id: u32, +} + +// The size of `TypeId` was seen to have a large-ish impact in #844, so this +// assert ensures that it stays relatively small. +const _: () = { + assert!(std::mem::size_of::<TypeId>() <= 16); +}; + +/// A unified type definition for validating WebAssembly modules and components. +#[derive(Debug)] +pub enum Type { + /// The definition is for a core function type. + Func(FuncType), + /// The definition is for a core module type. + /// + /// This variant is only supported when parsing a component. + Module(ModuleType), + /// The definition is for a core module instance type. + /// + /// This variant is only supported when parsing a component. + Instance(InstanceType), + /// The definition is for a component type. + /// + /// This variant is only supported when parsing a component. + Component(ComponentType), + /// The definition is for a component instance type. + /// + /// This variant is only supported when parsing a component. + ComponentInstance(ComponentInstanceType), + /// The definition is for a component function type. + /// + /// This variant is only supported when parsing a component. + ComponentFunc(ComponentFuncType), + /// The definition is for a component defined type. + /// + /// This variant is only supported when parsing a component. + Defined(ComponentDefinedType), +} + +impl Type { + /// Converts the type to a core function type. + pub fn as_func_type(&self) -> Option<&FuncType> { + match self { + Self::Func(ty) => Some(ty), + _ => None, + } + } + + /// Converts the type to a core module type. + pub fn as_module_type(&self) -> Option<&ModuleType> { + match self { + Self::Module(ty) => Some(ty), + _ => None, + } + } + + /// Converts the type to a core module instance type. + pub fn as_instance_type(&self) -> Option<&InstanceType> { + match self { + Self::Instance(ty) => Some(ty), + _ => None, + } + } + + /// Converts the type to a component type. + pub fn as_component_type(&self) -> Option<&ComponentType> { + match self { + Self::Component(ty) => Some(ty), + _ => None, + } + } + + /// Converts the type to a component instance type. + pub fn as_component_instance_type(&self) -> Option<&ComponentInstanceType> { + match self { + Self::ComponentInstance(ty) => Some(ty), + _ => None, + } + } + + /// Converts the type to a component function type. + pub fn as_component_func_type(&self) -> Option<&ComponentFuncType> { + match self { + Self::ComponentFunc(ty) => Some(ty), + _ => None, + } + } + + /// Converts the type to a component defined type. + pub fn as_defined_type(&self) -> Option<&ComponentDefinedType> { + match self { + Self::Defined(ty) => Some(ty), + _ => None, + } + } + + pub(crate) fn type_size(&self) -> u32 { + match self { + Self::Func(ty) => 1 + (ty.params().len() + ty.results().len()) as u32, + Self::Module(ty) => ty.type_size, + Self::Instance(ty) => ty.type_size, + Self::Component(ty) => ty.type_size, + Self::ComponentInstance(ty) => ty.type_size, + Self::ComponentFunc(ty) => ty.type_size, + Self::Defined(ty) => ty.type_size(), + } + } +} + +/// A component value type. +#[derive(Debug, Clone, Copy)] +pub enum ComponentValType { + /// The value type is one of the primitive types. + Primitive(PrimitiveValType), + /// The type is represented with the given type identifier. + Type(TypeId), +} + +impl ComponentValType { + pub(crate) fn requires_realloc(&self, types: &TypeList) -> bool { + match self { + ComponentValType::Primitive(ty) => ty.requires_realloc(), + ComponentValType::Type(ty) => types[*ty] + .as_defined_type() + .unwrap() + .requires_realloc(types), + } + } + + /// Determines if component value type `a` is a subtype of `b`. + pub fn is_subtype_of(a: &Self, at: TypesRef, b: &Self, bt: TypesRef) -> bool { + Self::internal_is_subtype_of(a, at.list, b, bt.list) + } + + pub(crate) fn internal_is_subtype_of(a: &Self, at: &TypeList, b: &Self, bt: &TypeList) -> bool { + match (a, b) { + (ComponentValType::Primitive(a), ComponentValType::Primitive(b)) => { + PrimitiveValType::is_subtype_of(*a, *b) + } + (ComponentValType::Type(a), ComponentValType::Type(b)) => { + ComponentDefinedType::internal_is_subtype_of( + at[*a].as_defined_type().unwrap(), + at, + bt[*b].as_defined_type().unwrap(), + bt, + ) + } + (ComponentValType::Primitive(a), ComponentValType::Type(b)) => { + match bt[*b].as_defined_type().unwrap() { + ComponentDefinedType::Primitive(b) => PrimitiveValType::is_subtype_of(*a, *b), + _ => false, + } + } + (ComponentValType::Type(a), ComponentValType::Primitive(b)) => { + match at[*a].as_defined_type().unwrap() { + ComponentDefinedType::Primitive(a) => PrimitiveValType::is_subtype_of(*a, *b), + _ => false, + } + } + } + } + + fn push_wasm_types(&self, types: &TypeList, lowered_types: &mut LoweredTypes) -> bool { + match self { + Self::Primitive(ty) => push_primitive_wasm_types(ty, lowered_types), + Self::Type(id) => types[*id] + .as_defined_type() + .unwrap() + .push_wasm_types(types, lowered_types), + } + } + + pub(crate) fn type_size(&self) -> u32 { + match self { + Self::Primitive(_) => 1, + Self::Type(id) => id.type_size, + } + } +} + +/// The entity type for imports and exports of a module. +#[derive(Debug, Clone, Copy)] +pub enum EntityType { + /// The entity is a function. + Func(TypeId), + /// The entity is a table. + Table(TableType), + /// The entity is a memory. + Memory(MemoryType), + /// The entity is a global. + Global(GlobalType), + /// The entity is a tag. + Tag(TypeId), +} + +impl EntityType { + /// Determines if entity type `a` is a subtype of `b`. + pub fn is_subtype_of(a: &Self, at: TypesRef, b: &Self, bt: TypesRef) -> bool { + Self::internal_is_subtype_of(a, at.list, b, bt.list) + } + + pub(crate) fn internal_is_subtype_of(a: &Self, at: &TypeList, b: &Self, bt: &TypeList) -> bool { + macro_rules! limits_match { + ($a:expr, $b:expr) => {{ + let a = $a; + let b = $b; + a.initial >= b.initial + && match b.maximum { + Some(b_max) => match a.maximum { + Some(a_max) => a_max <= b_max, + None => false, + }, + None => true, + } + }}; + } + + match (a, b) { + (EntityType::Func(a), EntityType::Func(b)) => { + at[*a].as_func_type().unwrap() == bt[*b].as_func_type().unwrap() + } + (EntityType::Table(a), EntityType::Table(b)) => { + a.element_type == b.element_type && limits_match!(a, b) + } + (EntityType::Memory(a), EntityType::Memory(b)) => { + a.shared == b.shared && a.memory64 == b.memory64 && limits_match!(a, b) + } + (EntityType::Global(a), EntityType::Global(b)) => a == b, + (EntityType::Tag(a), EntityType::Tag(b)) => { + at[*a].as_func_type().unwrap() == bt[*b].as_func_type().unwrap() + } + _ => false, + } + } + + pub(crate) fn desc(&self) -> &'static str { + match self { + Self::Func(_) => "function", + Self::Table(_) => "table", + Self::Memory(_) => "memory", + Self::Global(_) => "global", + Self::Tag(_) => "tag", + } + } + + pub(crate) fn type_size(&self) -> u32 { + match self { + Self::Func(id) | Self::Tag(id) => id.type_size, + Self::Table(_) | Self::Memory(_) | Self::Global(_) => 1, + } + } +} + +trait ModuleImportKey { + fn module(&self) -> &str; + fn name(&self) -> &str; +} + +impl<'a> Borrow<dyn ModuleImportKey + 'a> for (String, String) { + fn borrow(&self) -> &(dyn ModuleImportKey + 'a) { + self + } +} + +impl Hash for (dyn ModuleImportKey + '_) { + fn hash<H: Hasher>(&self, state: &mut H) { + self.module().hash(state); + self.name().hash(state); + } +} + +impl PartialEq for (dyn ModuleImportKey + '_) { + fn eq(&self, other: &Self) -> bool { + self.module() == other.module() && self.name() == other.name() + } +} + +impl Eq for (dyn ModuleImportKey + '_) {} + +impl ModuleImportKey for (String, String) { + fn module(&self) -> &str { + &self.0 + } + + fn name(&self) -> &str { + &self.1 + } +} + +impl ModuleImportKey for (&str, &str) { + fn module(&self) -> &str { + self.0 + } + + fn name(&self) -> &str { + self.1 + } +} + +/// Represents a core module type. +#[derive(Debug, Clone)] +pub struct ModuleType { + /// The effective type size for the module type. + pub(crate) type_size: u32, + /// The imports of the module type. + pub imports: IndexMap<(String, String), EntityType>, + /// The exports of the module type. + pub exports: IndexMap<String, EntityType>, +} + +impl ModuleType { + /// Looks up an import by its module and name. + /// + /// Returns `None` if the import was not found. + pub fn lookup_import(&self, module: &str, name: &str) -> Option<&EntityType> { + self.imports.get(&(module, name) as &dyn ModuleImportKey) + } + + /// Determines if module type `a` is a subtype of `b`. + pub fn is_subtype_of(a: &Self, at: TypesRef, b: &Self, bt: TypesRef) -> bool { + Self::internal_is_subtype_of(a, at.list, b, bt.list) + } + + pub(crate) fn internal_is_subtype_of(a: &Self, at: &TypeList, b: &Self, bt: &TypeList) -> bool { + // For module type subtyping, all exports in the other module type + // must be present in this module type's exports (i.e. it can export + // *more* than what this module type needs). + // However, for imports, the check is reversed (i.e. it is okay + // to import *less* than what this module type needs). + a.imports.iter().all(|(k, a)| match b.imports.get(k) { + Some(b) => EntityType::internal_is_subtype_of(b, bt, a, at), + None => false, + }) && b.exports.iter().all(|(k, b)| match a.exports.get(k) { + Some(a) => EntityType::internal_is_subtype_of(a, at, b, bt), + None => false, + }) + } +} + +/// Represents the kind of module instance type. +#[derive(Debug, Clone)] +pub enum InstanceTypeKind { + /// The instance type is the result of instantiating a module type. + Instantiated(TypeId), + /// The instance type is the result of instantiating from exported items. + Exports(IndexMap<String, EntityType>), +} + +/// Represents a module instance type. +#[derive(Debug, Clone)] +pub struct InstanceType { + /// The effective type size for the module instance type. + pub(crate) type_size: u32, + /// The kind of module instance type. + pub kind: InstanceTypeKind, +} + +impl InstanceType { + /// Gets the exports of the instance type. + pub fn exports<'a>(&'a self, types: TypesRef<'a>) -> &'a IndexMap<String, EntityType> { + self.internal_exports(types.list) + } + + pub(crate) fn internal_exports<'a>( + &'a self, + types: &'a TypeList, + ) -> &'a IndexMap<String, EntityType> { + match &self.kind { + InstanceTypeKind::Instantiated(id) => &types[*id].as_module_type().unwrap().exports, + InstanceTypeKind::Exports(exports) => exports, + } + } +} + +/// The entity type for imports and exports of a component. +#[derive(Debug, Clone, Copy)] +pub enum ComponentEntityType { + /// The entity is a core module. + Module(TypeId), + /// The entity is a function. + Func(TypeId), + /// The entity is a value. + Value(ComponentValType), + /// The entity is a type. + Type { + /// This is the identifier of the type that was referenced when this + /// entity was created. + referenced: TypeId, + /// This is the identifier of the type that was created when this type + /// was imported or exported from the component. + /// + /// Note that the underlying type information for the `referenced` + /// field and for this `created` field is the same, but these two types + /// will hash to different values. + created: TypeId, + }, + /// The entity is a component instance. + Instance(TypeId), + /// The entity is a component. + Component(TypeId), +} + +impl ComponentEntityType { + /// Determines if component entity type `a` is a subtype of `b`. + pub fn is_subtype_of(a: &Self, at: TypesRef, b: &Self, bt: TypesRef) -> bool { + Self::internal_is_subtype_of(a, at.list, b, bt.list) + } + + pub(crate) fn internal_is_subtype_of(a: &Self, at: &TypeList, b: &Self, bt: &TypeList) -> bool { + match (a, b) { + (Self::Module(a), Self::Module(b)) => ModuleType::internal_is_subtype_of( + at[*a].as_module_type().unwrap(), + at, + bt[*b].as_module_type().unwrap(), + bt, + ), + (Self::Func(a), Self::Func(b)) => ComponentFuncType::internal_is_subtype_of( + at[*a].as_component_func_type().unwrap(), + at, + bt[*b].as_component_func_type().unwrap(), + bt, + ), + (Self::Value(a), Self::Value(b)) => { + ComponentValType::internal_is_subtype_of(a, at, b, bt) + } + (Self::Type { referenced: a, .. }, Self::Type { referenced: b, .. }) => { + ComponentDefinedType::internal_is_subtype_of( + at[*a].as_defined_type().unwrap(), + at, + bt[*b].as_defined_type().unwrap(), + bt, + ) + } + (Self::Instance(a), Self::Instance(b)) => { + ComponentInstanceType::internal_is_subtype_of( + at[*a].as_component_instance_type().unwrap(), + at, + bt[*b].as_component_instance_type().unwrap(), + bt, + ) + } + (Self::Component(a), Self::Component(b)) => ComponentType::internal_is_subtype_of( + at[*a].as_component_type().unwrap(), + at, + bt[*b].as_component_type().unwrap(), + bt, + ), + _ => false, + } + } + + pub(crate) fn desc(&self) -> &'static str { + match self { + Self::Module(_) => "module", + Self::Func(_) => "function", + Self::Value(_) => "value", + Self::Type { .. } => "type", + Self::Instance(_) => "instance", + Self::Component(_) => "component", + } + } + + pub(crate) fn type_size(&self) -> u32 { + match self { + Self::Module(ty) + | Self::Func(ty) + | Self::Type { referenced: ty, .. } + | Self::Instance(ty) + | Self::Component(ty) => ty.type_size, + Self::Value(ty) => ty.type_size(), + } + } +} + +/// Represents a type of a component. +#[derive(Debug, Clone)] +pub struct ComponentType { + /// The effective type size for the component type. + pub(crate) type_size: u32, + /// The imports of the component type. + pub imports: IndexMap<KebabString, (Option<Url>, ComponentEntityType)>, + /// The exports of the component type. + pub exports: IndexMap<KebabString, (Option<Url>, ComponentEntityType)>, +} + +impl ComponentType { + /// Determines if component type `a` is a subtype of `b`. + pub fn is_subtype_of(a: &Self, at: TypesRef, b: &Self, bt: TypesRef) -> bool { + Self::internal_is_subtype_of(a, at.list, b, bt.list) + } + + pub(crate) fn internal_is_subtype_of(a: &Self, at: &TypeList, b: &Self, bt: &TypeList) -> bool { + // For component type subtyping, all exports in the other component type + // must be present in this component type's exports (i.e. it can export + // *more* than what this component type needs). + // However, for imports, the check is reversed (i.e. it is okay + // to import *less* than what this component type needs). + a.imports.iter().all(|(k, (_, a))| match b.imports.get(k) { + Some((_, b)) => ComponentEntityType::internal_is_subtype_of(b, bt, a, at), + None => false, + }) && b.exports.iter().all(|(k, (_, b))| match a.exports.get(k) { + Some((_, a)) => ComponentEntityType::internal_is_subtype_of(a, at, b, bt), + None => false, + }) + } +} + +/// Represents the kind of a component instance. +#[derive(Debug, Clone)] +pub enum ComponentInstanceTypeKind { + /// The instance type is from a definition. + Defined(IndexMap<KebabString, (Option<Url>, ComponentEntityType)>), + /// The instance type is the result of instantiating a component type. + Instantiated(TypeId), + /// The instance type is the result of instantiating from exported items. + Exports(IndexMap<KebabString, (Option<Url>, ComponentEntityType)>), +} + +/// Represents a type of a component instance. +#[derive(Debug, Clone)] +pub struct ComponentInstanceType { + /// The effective type size for the instance type. + pub(crate) type_size: u32, + /// The kind of instance type. + pub kind: ComponentInstanceTypeKind, +} + +impl ComponentInstanceType { + /// Gets the exports of the instance type. + pub fn exports<'a>( + &'a self, + types: TypesRef<'a>, + ) -> impl ExactSizeIterator<Item = (&'a KebabStr, &'a Option<Url>, ComponentEntityType)> + Clone + { + self.internal_exports(types.list) + .iter() + .map(|(n, (u, t))| (n.as_kebab_str(), u, *t)) + } + + pub(crate) fn internal_exports<'a>( + &'a self, + types: &'a TypeList, + ) -> &'a IndexMap<KebabString, (Option<Url>, ComponentEntityType)> { + match &self.kind { + ComponentInstanceTypeKind::Defined(exports) + | ComponentInstanceTypeKind::Exports(exports) => exports, + ComponentInstanceTypeKind::Instantiated(id) => { + &types[*id].as_component_type().unwrap().exports + } + } + } + + /// Determines if component instance type `a` is a subtype of `b`. + pub fn is_subtype_of(a: &Self, at: TypesRef, b: &Self, bt: TypesRef) -> bool { + Self::internal_is_subtype_of(a, at.list, b, bt.list) + } + + pub(crate) fn internal_is_subtype_of(a: &Self, at: &TypeList, b: &Self, bt: &TypeList) -> bool { + let exports = a.internal_exports(at); + + // For instance type subtyping, all exports in the other instance type + // must be present in this instance type's exports (i.e. it can export + // *more* than what this instance type needs). + b.internal_exports(bt) + .iter() + .all(|(k, (_, b))| match exports.get(k) { + Some((_, a)) => ComponentEntityType::internal_is_subtype_of(a, at, b, bt), + None => false, + }) + } +} + +/// Represents a type of a component function. +#[derive(Debug, Clone)] +pub struct ComponentFuncType { + /// The effective type size for the component function type. + pub(crate) type_size: u32, + /// The function parameters. + pub params: Box<[(KebabString, ComponentValType)]>, + /// The function's results. + pub results: Box<[(Option<KebabString>, ComponentValType)]>, +} + +impl ComponentFuncType { + /// Determines if component function type `a` is a subtype of `b`. + pub fn is_subtype_of(a: &Self, at: TypesRef, b: &Self, bt: TypesRef) -> bool { + Self::internal_is_subtype_of(a, at.list, b, bt.list) + } + + pub(crate) fn internal_is_subtype_of(a: &Self, at: &TypeList, b: &Self, bt: &TypeList) -> bool { + // Note that this intentionally diverges from the upstream specification + // in terms of subtyping. This is a full type-equality check which + // ensures that the structure of `a` exactly matches the structure of + // `b`. The rationale for this is: + // + // * Primarily in Wasmtime subtyping based on function types is not + // implemented. This includes both subtyping a host import and + // additionally handling subtyping as functions cross component + // boundaries. The host import subtyping (or component export + // subtyping) is not clear how to handle at all at this time. The + // subtyping of functions between components can more easily be + // handled by extending the `fact` compiler, but that hasn't been done + // yet. + // + // * The upstream specification is currently pretty intentionally vague + // precisely what subtyping is allowed. Implementing a strict check + // here is intended to be a conservative starting point for the + // component model which can be extended in the future if necessary. + // + // * The interaction with subtyping on bindings generation, for example, + // is a tricky problem that doesn't have a clear answer at this time. + // Effectively this is more rationale for being conservative in the + // first pass of the component model. + // + // So, in conclusion, the test here (and other places that reference + // this comment) is for exact type equality with no differences. + a.params.len() == b.params.len() + && a.results.len() == b.results.len() + && a.params + .iter() + .zip(b.params.iter()) + .all(|((an, a), (bn, b))| { + an == bn && ComponentValType::internal_is_subtype_of(a, at, b, bt) + }) + && a.results + .iter() + .zip(b.results.iter()) + .all(|((an, a), (bn, b))| { + an == bn && ComponentValType::internal_is_subtype_of(a, at, b, bt) + }) + } + + /// Lowers the component function type to core parameter and result types for the + /// canonical ABI. + pub(crate) fn lower(&self, types: &TypeList, import: bool) -> LoweringInfo { + let mut info = LoweringInfo::default(); + + for (_, ty) in self.params.iter() { + // When `import` is false, it means we're lifting a core function, + // check if the parameters needs realloc + if !import && !info.requires_realloc { + info.requires_realloc = ty.requires_realloc(types); + } + + if !ty.push_wasm_types(types, &mut info.params) { + // Too many parameters to pass directly + // Function will have a single pointer parameter to pass the arguments + // via linear memory + info.params.clear(); + assert!(info.params.push(ValType::I32)); + info.requires_memory = true; + + // We need realloc as well when lifting a function + if !import { + info.requires_realloc = true; + } + break; + } + } + + for (_, ty) in self.results.iter() { + // When `import` is true, it means we're lowering a component function, + // check if the result needs realloc + if import && !info.requires_realloc { + info.requires_realloc = ty.requires_realloc(types); + } + + if !ty.push_wasm_types(types, &mut info.results) { + // Too many results to return directly, either a retptr parameter will be used (import) + // or a single pointer will be returned (export) + info.results.clear(); + if import { + info.params.max = MAX_LOWERED_TYPES; + assert!(info.params.push(ValType::I32)); + } else { + assert!(info.results.push(ValType::I32)); + } + info.requires_memory = true; + break; + } + } + + // Memory is always required when realloc is required + info.requires_memory |= info.requires_realloc; + + info + } +} + +/// Represents a variant case. +#[derive(Debug, Clone)] +pub struct VariantCase { + /// The variant case type. + pub ty: Option<ComponentValType>, + /// The name of the variant case refined by this one. + pub refines: Option<KebabString>, +} + +/// Represents a record type. +#[derive(Debug, Clone)] +pub struct RecordType { + /// The effective type size for the record type. + pub(crate) type_size: u32, + /// The map of record fields. + pub fields: IndexMap<KebabString, ComponentValType>, +} + +/// Represents a variant type. +#[derive(Debug, Clone)] +pub struct VariantType { + /// The effective type size for the variant type. + pub(crate) type_size: u32, + /// The map of variant cases. + pub cases: IndexMap<KebabString, VariantCase>, +} + +/// Represents a tuple type. +#[derive(Debug, Clone)] +pub struct TupleType { + /// The effective type size for the tuple type. + pub(crate) type_size: u32, + /// The types of the tuple. + pub types: Box<[ComponentValType]>, +} + +/// Represents a union type. +#[derive(Debug, Clone)] +pub struct UnionType { + /// The inclusive type count for the union type. + pub(crate) type_size: u32, + /// The types of the union. + pub types: Box<[ComponentValType]>, +} + +/// Represents a component defined type. +#[derive(Debug, Clone)] +pub enum ComponentDefinedType { + /// The type is a primitive value type. + Primitive(PrimitiveValType), + /// The type is a record. + Record(RecordType), + /// The type is a variant. + Variant(VariantType), + /// The type is a list. + List(ComponentValType), + /// The type is a tuple. + Tuple(TupleType), + /// The type is a set of flags. + Flags(IndexSet<KebabString>), + /// The type is an enumeration. + Enum(IndexSet<KebabString>), + /// The type is a union. + Union(UnionType), + /// The type is an `option`. + Option(ComponentValType), + /// The type is a `result`. + Result { + /// The `ok` type. + ok: Option<ComponentValType>, + /// The `error` type. + err: Option<ComponentValType>, + }, +} + +impl ComponentDefinedType { + pub(crate) fn requires_realloc(&self, types: &TypeList) -> bool { + match self { + Self::Primitive(ty) => ty.requires_realloc(), + Self::Record(r) => r.fields.values().any(|ty| ty.requires_realloc(types)), + Self::Variant(v) => v.cases.values().any(|case| { + case.ty + .map(|ty| ty.requires_realloc(types)) + .unwrap_or(false) + }), + Self::List(_) => true, + Self::Tuple(t) => t.types.iter().any(|ty| ty.requires_realloc(types)), + Self::Union(u) => u.types.iter().any(|ty| ty.requires_realloc(types)), + Self::Flags(_) | Self::Enum(_) => false, + Self::Option(ty) => ty.requires_realloc(types), + Self::Result { ok, err } => { + ok.map(|ty| ty.requires_realloc(types)).unwrap_or(false) + || err.map(|ty| ty.requires_realloc(types)).unwrap_or(false) + } + } + } + + /// Determines if component defined type `a` is a subtype of `b`. + pub fn is_subtype_of(a: &Self, at: TypesRef, b: &Self, bt: TypesRef) -> bool { + Self::internal_is_subtype_of(a, at.list, b, bt.list) + } + + pub(crate) fn internal_is_subtype_of(a: &Self, at: &TypeList, b: &Self, bt: &TypeList) -> bool { + // Note that the implementation of subtyping here diverges from the + // upstream specification intentionally, see the documentation on + // function subtyping for more information. + match (a, b) { + (Self::Primitive(a), Self::Primitive(b)) => PrimitiveValType::is_subtype_of(*a, *b), + (Self::Record(a), Self::Record(b)) => { + a.fields.len() == b.fields.len() + && a.fields + .iter() + .zip(b.fields.iter()) + .all(|((aname, a), (bname, b))| { + aname == bname && ComponentValType::internal_is_subtype_of(a, at, b, bt) + }) + } + (Self::Variant(a), Self::Variant(b)) => { + a.cases.len() == b.cases.len() + && a.cases + .iter() + .zip(b.cases.iter()) + .all(|((aname, a), (bname, b))| { + aname == bname + && match (&a.ty, &b.ty) { + (Some(a), Some(b)) => { + ComponentValType::internal_is_subtype_of(a, at, b, bt) + } + (None, None) => true, + _ => false, + } + }) + } + (Self::List(a), Self::List(b)) | (Self::Option(a), Self::Option(b)) => { + ComponentValType::internal_is_subtype_of(a, at, b, bt) + } + (Self::Tuple(a), Self::Tuple(b)) => { + if a.types.len() != b.types.len() { + return false; + } + a.types + .iter() + .zip(b.types.iter()) + .all(|(a, b)| ComponentValType::internal_is_subtype_of(a, at, b, bt)) + } + (Self::Union(a), Self::Union(b)) => { + if a.types.len() != b.types.len() { + return false; + } + a.types + .iter() + .zip(b.types.iter()) + .all(|(a, b)| ComponentValType::internal_is_subtype_of(a, at, b, bt)) + } + (Self::Flags(a), Self::Flags(b)) | (Self::Enum(a), Self::Enum(b)) => { + a.len() == b.len() && a.iter().eq(b.iter()) + } + (Self::Result { ok: ao, err: ae }, Self::Result { ok: bo, err: be }) => { + Self::is_optional_subtype_of(*ao, at, *bo, bt) + && Self::is_optional_subtype_of(*ae, at, *be, bt) + } + _ => false, + } + } + + pub(crate) fn type_size(&self) -> u32 { + match self { + Self::Primitive(_) => 1, + Self::Flags(_) | Self::Enum(_) => 1, + Self::Record(r) => r.type_size, + Self::Variant(v) => v.type_size, + Self::Tuple(t) => t.type_size, + Self::Union(u) => u.type_size, + Self::List(ty) | Self::Option(ty) => ty.type_size(), + Self::Result { ok, err } => { + ok.map(|ty| ty.type_size()).unwrap_or(1) + err.map(|ty| ty.type_size()).unwrap_or(1) + } + } + } + + fn is_optional_subtype_of( + a: Option<ComponentValType>, + at: &TypeList, + b: Option<ComponentValType>, + bt: &TypeList, + ) -> bool { + match (a, b) { + (None, None) => true, + (Some(a), Some(b)) => ComponentValType::internal_is_subtype_of(&a, at, &b, bt), + _ => false, + } + } + fn push_wasm_types(&self, types: &TypeList, lowered_types: &mut LoweredTypes) -> bool { + match self { + Self::Primitive(ty) => push_primitive_wasm_types(ty, lowered_types), + Self::Record(r) => r + .fields + .iter() + .all(|(_, ty)| ty.push_wasm_types(types, lowered_types)), + Self::Variant(v) => Self::push_variant_wasm_types( + v.cases.iter().filter_map(|(_, case)| case.ty.as_ref()), + types, + lowered_types, + ), + Self::List(_) => lowered_types.push(ValType::I32) && lowered_types.push(ValType::I32), + Self::Tuple(t) => t + .types + .iter() + .all(|ty| ty.push_wasm_types(types, lowered_types)), + Self::Flags(names) => { + (0..(names.len() + 31) / 32).all(|_| lowered_types.push(ValType::I32)) + } + Self::Enum(_) => lowered_types.push(ValType::I32), + Self::Union(u) => Self::push_variant_wasm_types(u.types.iter(), types, lowered_types), + Self::Option(ty) => { + Self::push_variant_wasm_types([ty].into_iter(), types, lowered_types) + } + Self::Result { ok, err } => { + Self::push_variant_wasm_types(ok.iter().chain(err.iter()), types, lowered_types) + } + } + } + + fn push_variant_wasm_types<'a>( + cases: impl Iterator<Item = &'a ComponentValType>, + types: &TypeList, + lowered_types: &mut LoweredTypes, + ) -> bool { + // Push the discriminant + if !lowered_types.push(ValType::I32) { + return false; + } + + let start = lowered_types.len(); + + for ty in cases { + let mut temp = LoweredTypes::new(lowered_types.max); + + if !ty.push_wasm_types(types, &mut temp) { + return false; + } + + for (i, ty) in temp.iter().enumerate() { + match lowered_types.get_mut(start + i) { + Some(prev) => *prev = Self::join_types(*prev, ty), + None => { + if !lowered_types.push(ty) { + return false; + } + } + } + } + } + + true + } + + fn join_types(a: ValType, b: ValType) -> ValType { + use ValType::*; + + match (a, b) { + (I32, I32) | (I64, I64) | (F32, F32) | (F64, F64) => a, + (I32, F32) | (F32, I32) => I32, + (_, I64 | F64) | (I64 | F64, _) => I64, + _ => panic!("unexpected wasm type for canonical ABI"), + } + } +} + +#[allow(clippy::large_enum_variant)] +enum TypesKind { + Module(Arc<Module>), + Component(ComponentState), +} + +/// Represents the types known to a [`crate::Validator`] once validation has completed. +/// +/// The type information is returned via the [`crate::Validator::end`] method. +pub struct Types { + list: TypeList, + kind: TypesKind, +} + +#[derive(Clone, Copy)] +enum TypesRefKind<'a> { + Module(&'a Module), + Component(&'a ComponentState), +} + +/// Represents the types known to a [`crate::Validator`] during validation. +/// +/// Retrieved via the [`crate::Validator::types`] method. +#[derive(Clone, Copy)] +pub struct TypesRef<'a> { + list: &'a TypeList, + kind: TypesRefKind<'a>, +} + +impl<'a> TypesRef<'a> { + pub(crate) fn from_module(types: &'a TypeList, module: &'a Module) -> Self { + Self { + list: types, + kind: TypesRefKind::Module(module), + } + } + + pub(crate) fn from_component(types: &'a TypeList, component: &'a ComponentState) -> Self { + Self { + list: types, + kind: TypesRefKind::Component(component), + } + } + + fn types(&self, core: bool) -> Option<&'a [TypeId]> { + Some(match &self.kind { + TypesRefKind::Module(module) => { + if core { + &module.types + } else { + return None; + } + } + TypesRefKind::Component(component) => { + if core { + &component.core_types + } else { + &component.types + } + } + }) + } + + /// Gets a type based on its type id. + /// + /// Returns `None` if the type id is unknown. + pub fn type_from_id(&self, id: TypeId) -> Option<&'a Type> { + self.list.get(id.index) + } + + /// Gets a type id from a type index. + /// + /// Returns `None` if the type index is out of bounds or the type has not + /// been parsed yet. + pub fn id_from_type_index(&self, index: u32, core: bool) -> Option<TypeId> { + self.types(core)?.get(index as usize).copied() + } + + /// Gets a type at the given type index. + /// + /// Returns `None` if the type index is out of bounds or the type has not + /// been parsed yet. + pub fn type_at(&self, index: u32, core: bool) -> Option<&'a Type> { + self.type_from_id(*self.types(core)?.get(index as usize)?) + } + + /// Gets a defined core function type at the given type index. + /// + /// Returns `None` if the type index is out of bounds or the type has not + /// been parsed yet. + pub fn func_type_at(&self, index: u32) -> Option<&'a FuncType> { + match self.type_at(index, true)? { + Type::Func(ty) => Some(ty), + _ => None, + } + } + + /// Gets the type of a table at the given table index. + /// + /// Returns `None` if the type index is out of bounds or the type has not + /// been parsed yet. + pub fn table_at(&self, index: u32) -> Option<TableType> { + let tables = match &self.kind { + TypesRefKind::Module(module) => &module.tables, + TypesRefKind::Component(component) => &component.core_tables, + }; + + tables.get(index as usize).copied() + } + + /// Gets the type of a memory at the given memory index. + /// + /// Returns `None` if the type index is out of bounds or the type has not + /// been parsed yet. + pub fn memory_at(&self, index: u32) -> Option<MemoryType> { + let memories = match &self.kind { + TypesRefKind::Module(module) => &module.memories, + TypesRefKind::Component(component) => &component.core_memories, + }; + + memories.get(index as usize).copied() + } + + /// Gets the type of a global at the given global index. + /// + /// Returns `None` if the type index is out of bounds or the type has not + /// been parsed yet. + pub fn global_at(&self, index: u32) -> Option<GlobalType> { + let globals = match &self.kind { + TypesRefKind::Module(module) => &module.globals, + TypesRefKind::Component(component) => &component.core_globals, + }; + + globals.get(index as usize).copied() + } + + /// Gets the type of a tag at the given tag index. + /// + /// Returns `None` if the type index is out of bounds or the type has not + /// been parsed yet. + pub fn tag_at(&self, index: u32) -> Option<&'a FuncType> { + let tags = match &self.kind { + TypesRefKind::Module(module) => &module.tags, + TypesRefKind::Component(component) => &component.core_tags, + }; + + Some( + self.list[*tags.get(index as usize)?] + .as_func_type() + .unwrap(), + ) + } + + /// Gets the type of a core function at the given function index. + /// + /// Returns `None` if the type index is out of bounds or the type has not + /// been parsed yet. + pub fn function_at(&self, index: u32) -> Option<&'a FuncType> { + let id = match &self.kind { + TypesRefKind::Module(module) => { + &module.types[*module.functions.get(index as usize)? as usize] + } + TypesRefKind::Component(component) => component.core_funcs.get(index as usize)?, + }; + + match &self.list[*id] { + Type::Func(ty) => Some(ty), + _ => None, + } + } + + /// Gets the type of an element segment at the given element segment index. + /// + /// Returns `None` if the type index is out of bounds or the type has not + /// been parsed yet. + pub fn element_at(&self, index: u32) -> Option<RefType> { + match &self.kind { + TypesRefKind::Module(module) => module.element_types.get(index as usize).copied(), + TypesRefKind::Component(_) => None, + } + } + + /// Gets the type of a component function at the given function index. + /// + /// Returns `None` if the type index is out of bounds or the type has not + /// been parsed yet. + pub fn component_function_at(&self, index: u32) -> Option<&'a ComponentFuncType> { + match &self.kind { + TypesRefKind::Module(_) => None, + TypesRefKind::Component(component) => Some( + self.list[*component.funcs.get(index as usize)?] + .as_component_func_type() + .unwrap(), + ), + } + } + + /// Gets the type of a module at the given module index. + /// + /// Returns `None` if the type index is out of bounds or the type has not + /// been parsed yet. + pub fn module_at(&self, index: u32) -> Option<&'a ModuleType> { + match &self.kind { + TypesRefKind::Module(_) => None, + TypesRefKind::Component(component) => Some( + self.list[*component.core_modules.get(index as usize)?] + .as_module_type() + .unwrap(), + ), + } + } + + /// Gets the type of a module instance at the given module instance index. + /// + /// Returns `None` if the type index is out of bounds or the type has not + /// been parsed yet. + pub fn instance_at(&self, index: u32) -> Option<&'a InstanceType> { + match &self.kind { + TypesRefKind::Module(_) => None, + TypesRefKind::Component(component) => { + let id = component.core_instances.get(index as usize)?; + match &self.list[*id] { + Type::Instance(ty) => Some(ty), + _ => None, + } + } + } + } + + /// Gets the type of a component at the given component index. + /// + /// Returns `None` if the type index is out of bounds or the type has not + /// been parsed yet. + pub fn component_at(&self, index: u32) -> Option<&'a ComponentType> { + match &self.kind { + TypesRefKind::Module(_) => None, + TypesRefKind::Component(component) => Some( + self.list[*component.components.get(index as usize)?] + .as_component_type() + .unwrap(), + ), + } + } + + /// Gets the type of an component instance at the given component instance index. + /// + /// Returns `None` if the type index is out of bounds or the type has not + /// been parsed yet. + pub fn component_instance_at(&self, index: u32) -> Option<&'a ComponentInstanceType> { + match &self.kind { + TypesRefKind::Module(_) => None, + TypesRefKind::Component(component) => { + let id = component.instances.get(index as usize)?; + match &self.list[*id] { + Type::ComponentInstance(ty) => Some(ty), + _ => None, + } + } + } + } + + /// Gets the type of a value at the given value index. + /// + /// Returns `None` if the type index is out of bounds or the type has not + /// been parsed yet. + pub fn value_at(&self, index: u32) -> Option<ComponentValType> { + match &self.kind { + TypesRefKind::Module(_) => None, + TypesRefKind::Component(component) => { + component.values.get(index as usize).map(|(r, _)| *r) + } + } + } + + /// Gets the entity type for the given import. + pub fn entity_type_from_import(&self, import: &Import) -> Option<EntityType> { + match &self.kind { + TypesRefKind::Module(module) => Some(match import.ty { + TypeRef::Func(idx) => EntityType::Func(*module.types.get(idx as usize)?), + TypeRef::Table(ty) => EntityType::Table(ty), + TypeRef::Memory(ty) => EntityType::Memory(ty), + TypeRef::Global(ty) => EntityType::Global(ty), + TypeRef::Tag(ty) => EntityType::Tag(*module.types.get(ty.func_type_idx as usize)?), + }), + TypesRefKind::Component(_) => None, + } + } + + /// Gets the entity type from the given export. + pub fn entity_type_from_export(&self, export: &Export) -> Option<EntityType> { + match &self.kind { + TypesRefKind::Module(module) => Some(match export.kind { + ExternalKind::Func => EntityType::Func( + module.types[*module.functions.get(export.index as usize)? as usize], + ), + ExternalKind::Table => { + EntityType::Table(*module.tables.get(export.index as usize)?) + } + ExternalKind::Memory => { + EntityType::Memory(*module.memories.get(export.index as usize)?) + } + ExternalKind::Global => { + EntityType::Global(*module.globals.get(export.index as usize)?) + } + ExternalKind::Tag => EntityType::Tag( + module.types[*module.functions.get(export.index as usize)? as usize], + ), + }), + TypesRefKind::Component(_) => None, + } + } + + /// Gets the component entity type for the given component import. + pub fn component_entity_type_of_extern(&self, name: &str) -> Option<ComponentEntityType> { + match &self.kind { + TypesRefKind::Module(_) => None, + TypesRefKind::Component(component) => { + let key = KebabStr::new(name)?; + Some(component.externs.get(key)?.1) + } + } + } +} + +impl Types { + pub(crate) fn from_module(types: TypeList, module: Arc<Module>) -> Self { + Self { + list: types, + kind: TypesKind::Module(module), + } + } + + pub(crate) fn from_component(types: TypeList, component: ComponentState) -> Self { + Self { + list: types, + kind: TypesKind::Component(component), + } + } + + /// Gets a reference to this validation type information. + pub fn as_ref(&self) -> TypesRef { + TypesRef { + list: &self.list, + kind: match &self.kind { + TypesKind::Module(module) => TypesRefKind::Module(module), + TypesKind::Component(component) => TypesRefKind::Component(component), + }, + } + } + + /// Gets a type based on its type id. + /// + /// Returns `None` if the type id is unknown. + pub fn type_from_id(&self, id: TypeId) -> Option<&Type> { + self.as_ref().type_from_id(id) + } + + /// Gets a type id from a type index. + /// + /// Returns `None` if the type index is out of bounds. + pub fn id_from_type_index(&self, index: u32, core: bool) -> Option<TypeId> { + self.as_ref().id_from_type_index(index, core) + } + + /// Gets a type at the given type index. + /// + /// Returns `None` if the index is out of bounds. + pub fn type_at(&self, index: u32, core: bool) -> Option<&Type> { + self.as_ref().type_at(index, core) + } + + /// Gets a defined core function type at the given type index. + /// + /// Returns `None` if the index is out of bounds. + pub fn func_type_at(&self, index: u32) -> Option<&FuncType> { + self.as_ref().func_type_at(index) + } + + /// Gets the count of core types. + pub fn type_count(&self) -> usize { + match &self.kind { + TypesKind::Module(module) => module.types.len(), + TypesKind::Component(component) => component.core_types.len(), + } + } + + /// Gets the type of a table at the given table index. + /// + /// Returns `None` if the index is out of bounds. + pub fn table_at(&self, index: u32) -> Option<TableType> { + self.as_ref().table_at(index) + } + + /// Gets the count of imported and defined tables. + pub fn table_count(&self) -> usize { + match &self.kind { + TypesKind::Module(module) => module.tables.len(), + TypesKind::Component(component) => component.core_tables.len(), + } + } + + /// Gets the type of a memory at the given memory index. + /// + /// Returns `None` if the index is out of bounds. + pub fn memory_at(&self, index: u32) -> Option<MemoryType> { + self.as_ref().memory_at(index) + } + + /// Gets the count of imported and defined memories. + pub fn memory_count(&self) -> usize { + match &self.kind { + TypesKind::Module(module) => module.memories.len(), + TypesKind::Component(component) => component.core_memories.len(), + } + } + + /// Gets the type of a global at the given global index. + /// + /// Returns `None` if the index is out of bounds. + pub fn global_at(&self, index: u32) -> Option<GlobalType> { + self.as_ref().global_at(index) + } + + /// Gets the count of imported and defined globals. + pub fn global_count(&self) -> usize { + match &self.kind { + TypesKind::Module(module) => module.globals.len(), + TypesKind::Component(component) => component.core_globals.len(), + } + } + + /// Gets the type of a tag at the given tag index. + /// + /// Returns `None` if the index is out of bounds. + pub fn tag_at(&self, index: u32) -> Option<&FuncType> { + self.as_ref().tag_at(index) + } + + /// Gets the count of imported and defined tags. + pub fn tag_count(&self) -> usize { + match &self.kind { + TypesKind::Module(module) => module.tags.len(), + TypesKind::Component(component) => component.core_tags.len(), + } + } + + /// Gets the type of a core function at the given function index. + /// + /// Returns `None` if the index is out of bounds. + pub fn function_at(&self, index: u32) -> Option<&FuncType> { + self.as_ref().function_at(index) + } + + /// Gets the count of imported and defined core functions. + /// + /// The count also includes aliased core functions in components. + pub fn function_count(&self) -> usize { + match &self.kind { + TypesKind::Module(module) => module.functions.len(), + TypesKind::Component(component) => component.core_funcs.len(), + } + } + + /// Gets the type of an element segment at the given element segment index. + /// + /// Returns `None` if the index is out of bounds. + pub fn element_at(&self, index: u32) -> Option<RefType> { + match &self.kind { + TypesKind::Module(module) => module.element_types.get(index as usize).copied(), + TypesKind::Component(_) => None, + } + } + + /// Gets the count of element segments. + pub fn element_count(&self) -> usize { + match &self.kind { + TypesKind::Module(module) => module.element_types.len(), + TypesKind::Component(_) => 0, + } + } + + /// Gets the type of a component function at the given function index. + /// + /// Returns `None` if the index is out of bounds. + pub fn component_function_at(&self, index: u32) -> Option<&ComponentFuncType> { + self.as_ref().component_function_at(index) + } + + /// Gets the count of imported, exported, or aliased component functions. + pub fn component_function_count(&self) -> usize { + match &self.kind { + TypesKind::Module(_) => 0, + TypesKind::Component(component) => component.funcs.len(), + } + } + + /// Gets the type of a module at the given module index. + /// + /// Returns `None` if the index is out of bounds. + pub fn module_at(&self, index: u32) -> Option<&ModuleType> { + self.as_ref().module_at(index) + } + + /// Gets the count of imported, exported, or aliased modules. + pub fn module_count(&self) -> usize { + match &self.kind { + TypesKind::Module(_) => 0, + TypesKind::Component(component) => component.core_modules.len(), + } + } + + /// Gets the type of a module instance at the given module instance index. + /// + /// Returns `None` if the index is out of bounds. + pub fn instance_at(&self, index: u32) -> Option<&InstanceType> { + self.as_ref().instance_at(index) + } + + /// Gets the count of imported, exported, or aliased core module instances. + pub fn instance_count(&self) -> usize { + match &self.kind { + TypesKind::Module(_) => 0, + TypesKind::Component(component) => component.core_instances.len(), + } + } + + /// Gets the type of a component at the given component index. + /// + /// Returns `None` if the index is out of bounds. + pub fn component_at(&self, index: u32) -> Option<&ComponentType> { + self.as_ref().component_at(index) + } + + /// Gets the count of imported, exported, or aliased components. + pub fn component_count(&self) -> usize { + match &self.kind { + TypesKind::Module(_) => 0, + TypesKind::Component(component) => component.components.len(), + } + } + + /// Gets the type of an component instance at the given component instance index. + /// + /// Returns `None` if the index is out of bounds. + pub fn component_instance_at(&self, index: u32) -> Option<&ComponentInstanceType> { + self.as_ref().component_instance_at(index) + } + + /// Gets the count of imported, exported, or aliased component instances. + pub fn component_instance_count(&self) -> usize { + match &self.kind { + TypesKind::Module(_) => 0, + TypesKind::Component(component) => component.instances.len(), + } + } + + /// Gets the type of a value at the given value index. + /// + /// Returns `None` if the index is out of bounds. + pub fn value_at(&self, index: u32) -> Option<ComponentValType> { + self.as_ref().value_at(index) + } + + /// Gets the count of imported, exported, or aliased values. + pub fn value_count(&self) -> usize { + match &self.kind { + TypesKind::Module(_) => 0, + TypesKind::Component(component) => component.values.len(), + } + } + + /// Gets the entity type from the given import. + pub fn entity_type_from_import(&self, import: &Import) -> Option<EntityType> { + self.as_ref().entity_type_from_import(import) + } + + /// Gets the entity type from the given export. + pub fn entity_type_from_export(&self, export: &Export) -> Option<EntityType> { + self.as_ref().entity_type_from_export(export) + } + + /// Gets the component entity type for the given component import or export + /// name. + pub fn component_entity_type_of_extern(&self, name: &str) -> Option<ComponentEntityType> { + self.as_ref().component_entity_type_of_extern(name) + } + + /// Attempts to lookup the type id that `ty` is an alias of. + /// + /// Returns `None` if `ty` wasn't listed as aliasing a prior type. + pub fn peel_alias(&self, ty: TypeId) -> Option<TypeId> { + self.list.peel_alias(ty) + } +} + +/// This is a type which mirrors a subset of the `Vec<T>` API, but is intended +/// to be able to be cheaply snapshotted and cloned. +/// +/// When each module's code sections start we "commit" the current list of types +/// in the global list of types. This means that the temporary `cur` vec here is +/// pushed onto `snapshots` and wrapped up in an `Arc`. At that point we clone +/// this entire list (which is then O(modules), not O(types in all modules)) and +/// pass out as a context to each function validator. +/// +/// Otherwise, though, this type behaves as if it were a large `Vec<T>`, but +/// it's represented by lists of contiguous chunks. +pub(crate) struct SnapshotList<T> { + // All previous snapshots, the "head" of the list that this type represents. + // The first entry in this pair is the starting index for all elements + // contained in the list, and the second element is the list itself. Note + // the `Arc` wrapper around sub-lists, which makes cloning time for this + // `SnapshotList` O(snapshots) rather than O(snapshots_total), which for + // us in this context means the number of modules, not types. + // + // Note that this list is sorted least-to-greatest in order of the index for + // binary searching. + snapshots: Vec<Arc<Snapshot<T>>>, + + // This is the total length of all lists in the `snapshots` array. + snapshots_total: usize, + + // The current list of types for the current snapshot that are being built. + cur: Vec<T>, + + unique_mappings: HashMap<u32, u32>, + unique_counter: u32, +} + +struct Snapshot<T> { + prior_types: usize, + unique_counter: u32, + unique_mappings: HashMap<u32, u32>, + items: Vec<T>, +} + +impl<T> SnapshotList<T> { + /// Same as `<&[T]>::get` + pub(crate) fn get(&self, index: usize) -> Option<&T> { + // Check to see if this index falls on our local list + if index >= self.snapshots_total { + return self.cur.get(index - self.snapshots_total); + } + // ... and failing that we do a binary search to figure out which bucket + // it's in. Note the `i-1` in the `Err` case because if we don't find an + // exact match the type is located in the previous bucket. + let i = match self + .snapshots + .binary_search_by_key(&index, |snapshot| snapshot.prior_types) + { + Ok(i) => i, + Err(i) => i - 1, + }; + let snapshot = &self.snapshots[i]; + Some(&snapshot.items[index - snapshot.prior_types]) + } + + /// Same as `<&mut [T]>::get_mut`, except only works for indexes into the + /// current snapshot being built. + /// + /// # Panics + /// + /// Panics if an index is passed in which falls within the + /// previously-snapshotted list of types. This should never happen in our + /// context and the panic is intended to weed out possible bugs in + /// wasmparser. + pub(crate) fn get_mut(&mut self, index: usize) -> Option<&mut T> { + if index >= self.snapshots_total { + return self.cur.get_mut(index - self.snapshots_total); + } + panic!("cannot get a mutable reference in snapshotted part of list") + } + + /// Same as `Vec::push` + pub(crate) fn push(&mut self, val: T) { + self.cur.push(val); + } + + /// Same as `<[T]>::len` + pub(crate) fn len(&self) -> usize { + self.cur.len() + self.snapshots_total + } + + /// Reserve space for an additional count of items. + pub(crate) fn reserve(&mut self, additional: usize) { + self.cur.reserve(additional); + } + + /// Commits previously pushed types into this snapshot vector, and returns a + /// clone of this list. + /// + /// The returned `SnapshotList` can be used to access all the same types as + /// this list itself. This list also is not changed (from an external + /// perspective) and can continue to access all the same types. + pub(crate) fn commit(&mut self) -> SnapshotList<T> { + // If the current chunk has new elements, commit them in to an + // `Arc`-wrapped vector in the snapshots list. Note the `shrink_to_fit` + // ahead of time to hopefully keep memory usage lower than it would + // otherwise be. Additionally note that the `unique_counter` is bumped + // here to ensure that the previous value of the unique counter is + // never used for an actual type so it's suitable for lookup via a + // binary search. + let len = self.cur.len(); + if len > 0 { + self.unique_counter += 1; + self.cur.shrink_to_fit(); + self.snapshots.push(Arc::new(Snapshot { + prior_types: self.snapshots_total, + unique_counter: self.unique_counter - 1, + unique_mappings: mem::take(&mut self.unique_mappings), + items: mem::take(&mut self.cur), + })); + self.snapshots_total += len; + } + SnapshotList { + snapshots: self.snapshots.clone(), + snapshots_total: self.snapshots_total, + unique_mappings: HashMap::new(), + unique_counter: self.unique_counter, + cur: Vec::new(), + } + } + + /// Modifies a `TypeId` to have the same contents but a fresh new unique id. + /// + /// This is used during aliasing with components to assign types a unique + /// identifier that isn't equivalent to anything else but still + /// points to the same underlying type. + pub fn with_unique(&mut self, mut ty: TypeId) -> TypeId { + self.unique_mappings + .insert(self.unique_counter, ty.unique_id); + ty.unique_id = self.unique_counter; + self.unique_counter += 1; + ty + } + + /// Attempts to lookup the type id that `ty` is an alias of. + /// + /// Returns `None` if `ty` wasn't listed as aliasing a prior type. + pub fn peel_alias(&self, ty: TypeId) -> Option<TypeId> { + // The unique counter in each snapshot is the unique counter at the + // time of the snapshot so it's guaranteed to never be used, meaning + // that `Ok` should never show up here. With an `Err` it's where the + // index would be placed meaning that the index in question is the + // smallest value over the unique id's value, meaning that slot has the + // mapping we're interested in. + let i = match self + .snapshots + .binary_search_by_key(&ty.unique_id, |snapshot| snapshot.unique_counter) + { + Ok(_) => unreachable!(), + Err(i) => i, + }; + + // If the `i` index is beyond the snapshot array then lookup in the + // current mappings instead since it may refer to a type not snapshot + // yet. + let unique_id = match self.snapshots.get(i) { + Some(snapshot) => *snapshot.unique_mappings.get(&ty.unique_id)?, + None => *self.unique_mappings.get(&ty.unique_id)?, + }; + Some(TypeId { unique_id, ..ty }) + } +} + +impl<T> std::ops::Index<usize> for SnapshotList<T> { + type Output = T; + + #[inline] + fn index(&self, index: usize) -> &T { + self.get(index).unwrap() + } +} + +impl<T> std::ops::IndexMut<usize> for SnapshotList<T> { + #[inline] + fn index_mut(&mut self, index: usize) -> &mut T { + self.get_mut(index).unwrap() + } +} + +impl<T> std::ops::Index<TypeId> for SnapshotList<T> { + type Output = T; + + #[inline] + fn index(&self, id: TypeId) -> &T { + self.get(id.index).unwrap() + } +} + +impl<T> std::ops::IndexMut<TypeId> for SnapshotList<T> { + #[inline] + fn index_mut(&mut self, id: TypeId) -> &mut T { + self.get_mut(id.index).unwrap() + } +} + +impl<T> Default for SnapshotList<T> { + fn default() -> SnapshotList<T> { + SnapshotList { + snapshots: Vec::new(), + snapshots_total: 0, + cur: Vec::new(), + unique_counter: 1, + unique_mappings: HashMap::new(), + } + } +} + +/// A snapshot list of types. +pub(crate) type TypeList = SnapshotList<Type>; + +/// Thin wrapper around `TypeList` which provides an allocator of unique ids for +/// types contained within this list. +pub(crate) struct TypeAlloc { + list: TypeList, +} + +impl Deref for TypeAlloc { + type Target = TypeList; + fn deref(&self) -> &TypeList { + &self.list + } +} + +impl DerefMut for TypeAlloc { + fn deref_mut(&mut self) -> &mut TypeList { + &mut self.list + } +} + +impl TypeAlloc { + /// Pushes a new anonymous type into this list which will have its + /// `unique_id` field cleared. + pub fn push_anon(&mut self, ty: Type) -> TypeId { + let index = self.list.len(); + let type_size = ty.type_size(); + self.list.push(ty); + TypeId { + index, + type_size, + unique_id: 0, + } + } + + /// Pushes a new defined type which has an index in core wasm onto this + /// list. + /// + /// The returned `TypeId` is guaranteed to be unique and not hash-equivalent + /// to any other prior ID in this list. + pub fn push_defined(&mut self, ty: Type) -> TypeId { + let id = self.push_anon(ty); + self.with_unique(id) + } +} + +impl Default for TypeAlloc { + fn default() -> TypeAlloc { + TypeAlloc { + list: Default::default(), + } + } +} |