[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
// Copyright 2017 the V8 project authors. All rights reserved.
|
|
|
|
|
// Use of this source code is governed by a BSD-style license that can be
|
|
|
|
|
// found in the LICENSE file.
|
|
|
|
|
|
|
|
|
|
#include <stddef.h>
|
|
|
|
|
#include <stdint.h>
|
|
|
|
|
#include <stdlib.h>
|
|
|
|
|
|
|
|
|
|
#include <algorithm>
|
|
|
|
|
|
|
|
|
|
#include "include/v8.h"
|
2019-05-22 07:55:37 +00:00
|
|
|
#include "src/execution/isolate.h"
|
2019-05-23 08:51:46 +00:00
|
|
|
#include "src/objects/objects-inl.h"
|
|
|
|
|
#include "src/objects/objects.h"
|
2019-05-29 13:30:15 +00:00
|
|
|
#include "src/utils/ostreams.h"
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
#include "src/wasm/wasm-module-builder.h"
|
|
|
|
|
#include "src/wasm/wasm-module.h"
|
2020-06-24 11:17:46 +00:00
|
|
|
#include "src/wasm/wasm-opcodes-inl.h"
|
2019-05-29 13:30:15 +00:00
|
|
|
#include "test/common/wasm/flag-utils.h"
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
#include "test/common/wasm/test-signatures.h"
|
|
|
|
|
#include "test/common/wasm/wasm-module-runner.h"
|
|
|
|
|
#include "test/fuzzer/fuzzer-support.h"
|
2017-05-08 09:22:54 +00:00
|
|
|
#include "test/fuzzer/wasm-fuzzer-common.h"
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
|
2017-09-01 13:20:46 +00:00
|
|
|
namespace v8 {
|
|
|
|
|
namespace internal {
|
|
|
|
|
namespace wasm {
|
|
|
|
|
namespace fuzzer {
|
2017-08-09 08:11:21 +00:00
|
|
|
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
namespace {
|
|
|
|
|
|
2017-12-21 14:31:42 +00:00
|
|
|
constexpr int kMaxFunctions = 4;
|
2018-01-29 16:45:59 +00:00
|
|
|
constexpr int kMaxGlobals = 64;
|
2020-05-18 08:31:21 +00:00
|
|
|
constexpr int kMaxParameters = 15;
|
|
|
|
|
constexpr int kMaxReturns = 15;
|
2017-12-21 14:31:42 +00:00
|
|
|
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
class DataRange {
|
2018-07-12 09:05:43 +00:00
|
|
|
Vector<const uint8_t> data_;
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
|
|
|
|
|
public:
|
2018-07-12 09:05:43 +00:00
|
|
|
explicit DataRange(Vector<const uint8_t> data) : data_(data) {}
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
|
2017-12-21 14:30:32 +00:00
|
|
|
// Don't accidentally pass DataRange by value. This will reuse bytes and might
|
|
|
|
|
// lead to OOM because the end might not be reached.
|
|
|
|
|
// Define move constructor and move assignment, disallow copy constructor and
|
|
|
|
|
// copy assignment (below).
|
2018-08-02 08:16:29 +00:00
|
|
|
DataRange(DataRange&& other) V8_NOEXCEPT : DataRange(other.data_) {
|
|
|
|
|
other.data_ = {};
|
|
|
|
|
}
|
|
|
|
|
DataRange& operator=(DataRange&& other) V8_NOEXCEPT {
|
2017-12-21 14:30:32 +00:00
|
|
|
data_ = other.data_;
|
2018-07-12 09:05:43 +00:00
|
|
|
other.data_ = {};
|
2017-12-21 14:30:32 +00:00
|
|
|
return *this;
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
}
|
|
|
|
|
|
2018-07-12 09:05:43 +00:00
|
|
|
size_t size() const { return data_.size(); }
|
2017-12-21 14:30:32 +00:00
|
|
|
|
|
|
|
|
DataRange split() {
|
2018-07-12 09:05:43 +00:00
|
|
|
uint16_t num_bytes = get<uint16_t>() % std::max(size_t{1}, data_.size());
|
|
|
|
|
DataRange split(data_.SubVector(0, num_bytes));
|
2017-12-21 14:30:32 +00:00
|
|
|
data_ += num_bytes;
|
|
|
|
|
return split;
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
}
|
|
|
|
|
|
2018-11-15 15:08:32 +00:00
|
|
|
template <typename T, size_t max_bytes = sizeof(T)>
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
T get() {
|
2020-05-15 08:23:21 +00:00
|
|
|
// DISABLE FOR BOOL
|
|
|
|
|
// The -O3 on release will break the result. This creates a different
|
|
|
|
|
// observable side effect when invoking get<bool> between debug and release
|
|
|
|
|
// version, which eventually makes the code output different as well as
|
|
|
|
|
// raising various unrecoverable errors on runtime. It is caused by
|
|
|
|
|
// undefined behavior of assigning boolean via memcpy from randomized bytes.
|
|
|
|
|
STATIC_ASSERT(!(std::is_same<T, bool>::value));
|
2018-11-15 15:08:32 +00:00
|
|
|
STATIC_ASSERT(max_bytes <= sizeof(T));
|
2017-12-21 14:31:42 +00:00
|
|
|
// We want to support the case where we have less than sizeof(T) bytes
|
|
|
|
|
// remaining in the slice. For example, if we emit an i32 constant, it's
|
|
|
|
|
// okay if we don't have a full four bytes available, we'll just use what
|
|
|
|
|
// we have. We aren't concerned about endianness because we are generating
|
|
|
|
|
// arbitrary expressions.
|
2018-11-15 15:08:32 +00:00
|
|
|
const size_t num_bytes = std::min(max_bytes, data_.size());
|
2017-12-21 14:31:42 +00:00
|
|
|
T result = T();
|
2019-04-29 11:06:49 +00:00
|
|
|
memcpy(&result, data_.begin(), num_bytes);
|
2017-12-21 14:31:42 +00:00
|
|
|
data_ += num_bytes;
|
|
|
|
|
return result;
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
}
|
2017-12-21 14:30:32 +00:00
|
|
|
|
|
|
|
|
DISALLOW_COPY_AND_ASSIGN(DataRange);
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
};
|
|
|
|
|
|
2019-07-08 11:44:42 +00:00
|
|
|
ValueType GetValueType(DataRange* data) {
|
2020-02-06 18:26:29 +00:00
|
|
|
// TODO(v8:8460): We do not add kWasmS128 here yet because this method is used
|
|
|
|
|
// to generate globals, and since we do not have v128.const yet, there is no
|
|
|
|
|
// way to specify an initial value a global of this type.
|
2019-07-08 11:44:42 +00:00
|
|
|
switch (data->get<uint8_t>() % 4) {
|
2017-12-21 13:53:15 +00:00
|
|
|
case 0:
|
|
|
|
|
return kWasmI32;
|
|
|
|
|
case 1:
|
|
|
|
|
return kWasmI64;
|
|
|
|
|
case 2:
|
|
|
|
|
return kWasmF32;
|
|
|
|
|
case 3:
|
|
|
|
|
return kWasmF64;
|
|
|
|
|
}
|
|
|
|
|
UNREACHABLE();
|
|
|
|
|
}
|
|
|
|
|
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
class WasmGenerator {
|
2020-03-12 14:29:51 +00:00
|
|
|
template <WasmOpcode Op, ValueType::Kind... Args>
|
2019-07-08 11:44:42 +00:00
|
|
|
void op(DataRange* data) {
|
2017-11-06 15:48:17 +00:00
|
|
|
Generate<Args...>(data);
|
|
|
|
|
builder_->Emit(Op);
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
}
|
|
|
|
|
|
2017-11-15 11:51:15 +00:00
|
|
|
class BlockScope {
|
|
|
|
|
public:
|
2020-05-18 08:31:21 +00:00
|
|
|
BlockScope(WasmGenerator* gen, WasmOpcode block_type,
|
2020-06-18 12:55:08 +00:00
|
|
|
Vector<const ValueType> param_types,
|
|
|
|
|
Vector<const ValueType> result_types,
|
|
|
|
|
Vector<const ValueType> br_types)
|
2017-11-15 11:51:15 +00:00
|
|
|
: gen_(gen) {
|
2020-06-18 12:55:08 +00:00
|
|
|
gen->blocks_.emplace_back(br_types.begin(), br_types.end());
|
2020-05-18 08:31:21 +00:00
|
|
|
if (param_types.size() == 0 && result_types.size() == 0) {
|
|
|
|
|
gen->builder_->EmitWithU8(block_type, kWasmStmt.value_type_code());
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
if (param_types.size() == 0 && result_types.size() == 1) {
|
|
|
|
|
gen->builder_->EmitWithU8(block_type,
|
|
|
|
|
result_types[0].value_type_code());
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
// Multi-value block.
|
|
|
|
|
Zone* zone = gen->builder_->builder()->zone();
|
|
|
|
|
FunctionSig::Builder builder(zone, result_types.size(),
|
|
|
|
|
param_types.size());
|
|
|
|
|
for (auto& type : param_types) {
|
|
|
|
|
DCHECK_NE(type, kWasmStmt);
|
|
|
|
|
builder.AddParam(type);
|
|
|
|
|
}
|
|
|
|
|
for (auto& type : result_types) {
|
|
|
|
|
DCHECK_NE(type, kWasmStmt);
|
|
|
|
|
builder.AddReturn(type);
|
|
|
|
|
}
|
|
|
|
|
FunctionSig* sig = builder.Build();
|
|
|
|
|
int sig_id = gen->builder_->builder()->AddSignature(sig);
|
|
|
|
|
gen->builder_->EmitWithI32V(block_type, sig_id);
|
2017-11-15 11:51:15 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
~BlockScope() {
|
|
|
|
|
gen_->builder_->Emit(kExprEnd);
|
|
|
|
|
gen_->blocks_.pop_back();
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
private:
|
|
|
|
|
WasmGenerator* const gen_;
|
|
|
|
|
};
|
|
|
|
|
|
2020-06-18 12:55:08 +00:00
|
|
|
void block(Vector<const ValueType> param_types,
|
|
|
|
|
Vector<const ValueType> return_types, DataRange* data) {
|
2020-05-18 08:31:21 +00:00
|
|
|
BlockScope block_scope(this, kExprBlock, param_types, return_types,
|
|
|
|
|
return_types);
|
|
|
|
|
ConsumeAndGenerate(param_types, return_types, data);
|
|
|
|
|
}
|
|
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
template <ValueType::Kind T>
|
2019-07-08 11:44:42 +00:00
|
|
|
void block(DataRange* data) {
|
2020-06-18 12:55:08 +00:00
|
|
|
block({}, VectorOf({ValueType::Primitive(T)}), data);
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
}
|
|
|
|
|
|
2020-06-18 12:55:08 +00:00
|
|
|
void loop(Vector<const ValueType> param_types,
|
|
|
|
|
Vector<const ValueType> return_types, DataRange* data) {
|
2020-05-20 09:21:27 +00:00
|
|
|
BlockScope block_scope(this, kExprLoop, param_types, return_types,
|
|
|
|
|
param_types);
|
|
|
|
|
ConsumeAndGenerate(param_types, return_types, data);
|
|
|
|
|
}
|
|
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
template <ValueType::Kind T>
|
2019-07-08 11:44:42 +00:00
|
|
|
void loop(DataRange* data) {
|
2020-06-18 12:55:08 +00:00
|
|
|
loop({}, VectorOf({ValueType::Primitive(T)}), data);
|
2017-11-15 11:51:15 +00:00
|
|
|
}
|
2017-11-06 15:48:17 +00:00
|
|
|
|
2018-01-30 11:38:01 +00:00
|
|
|
enum IfType { kIf, kIfElse };
|
|
|
|
|
|
2020-06-18 12:55:08 +00:00
|
|
|
void if_(Vector<const ValueType> param_types,
|
|
|
|
|
Vector<const ValueType> return_types, IfType type, DataRange* data) {
|
2020-05-20 09:21:27 +00:00
|
|
|
// One-armed "if" are only valid if the input and output types are the same.
|
|
|
|
|
DCHECK_IMPLIES(type == kIf, param_types == return_types);
|
|
|
|
|
Generate(kWasmI32, data);
|
|
|
|
|
BlockScope block_scope(this, kExprIf, param_types, return_types,
|
|
|
|
|
return_types);
|
|
|
|
|
ConsumeAndGenerate(param_types, return_types, data);
|
|
|
|
|
if (type == kIfElse) {
|
|
|
|
|
builder_->Emit(kExprElse);
|
|
|
|
|
ConsumeAndGenerate(param_types, return_types, data);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
template <ValueType::Kind T, IfType type>
|
2019-07-08 11:44:42 +00:00
|
|
|
void if_(DataRange* data) {
|
2020-03-12 14:29:51 +00:00
|
|
|
static_assert(T == ValueType::kStmt || type == kIfElse,
|
2018-01-30 11:38:01 +00:00
|
|
|
"if without else cannot produce a value");
|
2020-06-18 12:55:08 +00:00
|
|
|
if_({},
|
|
|
|
|
T == ValueType::kStmt ? Vector<ValueType>{}
|
|
|
|
|
: VectorOf({ValueType::Primitive(T)}),
|
|
|
|
|
type, data);
|
2020-05-20 09:21:27 +00:00
|
|
|
}
|
|
|
|
|
|
2020-06-18 12:55:08 +00:00
|
|
|
void any_block(Vector<const ValueType> param_types,
|
|
|
|
|
Vector<const ValueType> return_types, DataRange* data) {
|
2020-05-20 09:21:27 +00:00
|
|
|
uint8_t block_type = data->get<uint8_t>() % 4;
|
|
|
|
|
switch (block_type) {
|
|
|
|
|
case 0:
|
|
|
|
|
block(param_types, return_types, data);
|
|
|
|
|
return;
|
|
|
|
|
case 1:
|
|
|
|
|
loop(param_types, return_types, data);
|
|
|
|
|
return;
|
|
|
|
|
case 2:
|
|
|
|
|
if (param_types == return_types) {
|
|
|
|
|
if_({}, {}, kIf, data);
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
V8_FALLTHROUGH;
|
|
|
|
|
case 3:
|
|
|
|
|
if_(param_types, return_types, kIfElse, data);
|
|
|
|
|
return;
|
2018-01-30 11:38:01 +00:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2019-07-08 11:44:42 +00:00
|
|
|
void br(DataRange* data) {
|
2017-11-15 11:51:15 +00:00
|
|
|
// There is always at least the block representing the function body.
|
|
|
|
|
DCHECK(!blocks_.empty());
|
2019-07-08 11:44:42 +00:00
|
|
|
const uint32_t target_block = data->get<uint32_t>() % blocks_.size();
|
2020-05-12 14:39:52 +00:00
|
|
|
const auto break_types = blocks_[target_block];
|
2017-11-06 15:48:17 +00:00
|
|
|
|
2020-05-18 08:31:21 +00:00
|
|
|
Generate(VectorOf(break_types), data);
|
2017-11-08 21:03:20 +00:00
|
|
|
builder_->EmitWithI32V(
|
|
|
|
|
kExprBr, static_cast<uint32_t>(blocks_.size()) - 1 - target_block);
|
2017-11-06 15:48:17 +00:00
|
|
|
}
|
|
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
template <ValueType::Kind wanted_type>
|
2019-07-08 11:44:42 +00:00
|
|
|
void br_if(DataRange* data) {
|
2018-01-29 16:42:03 +00:00
|
|
|
// There is always at least the block representing the function body.
|
|
|
|
|
DCHECK(!blocks_.empty());
|
2019-07-08 11:44:42 +00:00
|
|
|
const uint32_t target_block = data->get<uint32_t>() % blocks_.size();
|
2020-06-18 12:55:08 +00:00
|
|
|
const auto break_types = VectorOf(blocks_[target_block]);
|
2020-05-12 14:39:52 +00:00
|
|
|
|
2020-06-18 12:55:08 +00:00
|
|
|
Generate(break_types, data);
|
2018-01-29 16:42:03 +00:00
|
|
|
Generate(kWasmI32, data);
|
|
|
|
|
builder_->EmitWithI32V(
|
|
|
|
|
kExprBrIf, static_cast<uint32_t>(blocks_.size()) - 1 - target_block);
|
2020-06-18 12:55:08 +00:00
|
|
|
ConsumeAndGenerate(break_types,
|
|
|
|
|
wanted_type == ValueType::kStmt
|
|
|
|
|
? Vector<ValueType>{}
|
|
|
|
|
: VectorOf({ValueType::Primitive(wanted_type)}),
|
|
|
|
|
data);
|
2018-01-29 16:42:03 +00:00
|
|
|
}
|
|
|
|
|
|
2017-11-08 21:03:20 +00:00
|
|
|
// TODO(eholk): make this function constexpr once gcc supports it
|
|
|
|
|
static uint8_t max_alignment(WasmOpcode memop) {
|
|
|
|
|
switch (memop) {
|
2020-03-12 23:53:38 +00:00
|
|
|
case kExprS128LoadMem:
|
|
|
|
|
case kExprS128StoreMem:
|
|
|
|
|
return 4;
|
2017-11-08 21:03:20 +00:00
|
|
|
case kExprI64LoadMem:
|
|
|
|
|
case kExprF64LoadMem:
|
|
|
|
|
case kExprI64StoreMem:
|
|
|
|
|
case kExprF64StoreMem:
|
2020-01-14 18:17:34 +00:00
|
|
|
case kExprI64AtomicStore:
|
|
|
|
|
case kExprI64AtomicLoad:
|
2020-01-21 14:43:23 +00:00
|
|
|
case kExprI64AtomicAdd:
|
|
|
|
|
case kExprI64AtomicSub:
|
|
|
|
|
case kExprI64AtomicAnd:
|
|
|
|
|
case kExprI64AtomicOr:
|
|
|
|
|
case kExprI64AtomicXor:
|
2020-01-30 14:34:16 +00:00
|
|
|
case kExprI64AtomicExchange:
|
|
|
|
|
case kExprI64AtomicCompareExchange:
|
2020-05-11 17:12:45 +00:00
|
|
|
case kExprI16x8Load8x8S:
|
|
|
|
|
case kExprI16x8Load8x8U:
|
|
|
|
|
case kExprI32x4Load16x4S:
|
|
|
|
|
case kExprI32x4Load16x4U:
|
|
|
|
|
case kExprI64x2Load32x2S:
|
|
|
|
|
case kExprI64x2Load32x2U:
|
|
|
|
|
case kExprS64x2LoadSplat:
|
2017-11-08 21:03:20 +00:00
|
|
|
return 3;
|
|
|
|
|
case kExprI32LoadMem:
|
|
|
|
|
case kExprI64LoadMem32S:
|
|
|
|
|
case kExprI64LoadMem32U:
|
|
|
|
|
case kExprF32LoadMem:
|
|
|
|
|
case kExprI32StoreMem:
|
|
|
|
|
case kExprI64StoreMem32:
|
|
|
|
|
case kExprF32StoreMem:
|
2020-01-14 18:17:34 +00:00
|
|
|
case kExprI32AtomicStore:
|
|
|
|
|
case kExprI64AtomicStore32U:
|
|
|
|
|
case kExprI32AtomicLoad:
|
|
|
|
|
case kExprI64AtomicLoad32U:
|
2020-01-21 14:43:23 +00:00
|
|
|
case kExprI32AtomicAdd:
|
|
|
|
|
case kExprI32AtomicSub:
|
|
|
|
|
case kExprI32AtomicAnd:
|
|
|
|
|
case kExprI32AtomicOr:
|
|
|
|
|
case kExprI32AtomicXor:
|
2020-01-30 14:34:16 +00:00
|
|
|
case kExprI32AtomicExchange:
|
|
|
|
|
case kExprI32AtomicCompareExchange:
|
2020-01-21 14:43:23 +00:00
|
|
|
case kExprI64AtomicAdd32U:
|
|
|
|
|
case kExprI64AtomicSub32U:
|
|
|
|
|
case kExprI64AtomicAnd32U:
|
|
|
|
|
case kExprI64AtomicOr32U:
|
|
|
|
|
case kExprI64AtomicXor32U:
|
2020-01-30 14:34:16 +00:00
|
|
|
case kExprI64AtomicExchange32U:
|
|
|
|
|
case kExprI64AtomicCompareExchange32U:
|
2020-05-11 17:12:45 +00:00
|
|
|
case kExprS32x4LoadSplat:
|
2017-11-08 21:03:20 +00:00
|
|
|
return 2;
|
|
|
|
|
case kExprI32LoadMem16S:
|
|
|
|
|
case kExprI32LoadMem16U:
|
|
|
|
|
case kExprI64LoadMem16S:
|
|
|
|
|
case kExprI64LoadMem16U:
|
|
|
|
|
case kExprI32StoreMem16:
|
|
|
|
|
case kExprI64StoreMem16:
|
2020-01-14 18:17:34 +00:00
|
|
|
case kExprI32AtomicStore16U:
|
|
|
|
|
case kExprI64AtomicStore16U:
|
|
|
|
|
case kExprI32AtomicLoad16U:
|
|
|
|
|
case kExprI64AtomicLoad16U:
|
2020-01-21 14:43:23 +00:00
|
|
|
case kExprI32AtomicAdd16U:
|
|
|
|
|
case kExprI32AtomicSub16U:
|
|
|
|
|
case kExprI32AtomicAnd16U:
|
|
|
|
|
case kExprI32AtomicOr16U:
|
|
|
|
|
case kExprI32AtomicXor16U:
|
2020-01-30 14:34:16 +00:00
|
|
|
case kExprI32AtomicExchange16U:
|
|
|
|
|
case kExprI32AtomicCompareExchange16U:
|
2020-01-21 14:43:23 +00:00
|
|
|
case kExprI64AtomicAdd16U:
|
|
|
|
|
case kExprI64AtomicSub16U:
|
|
|
|
|
case kExprI64AtomicAnd16U:
|
|
|
|
|
case kExprI64AtomicOr16U:
|
|
|
|
|
case kExprI64AtomicXor16U:
|
2020-01-30 14:34:16 +00:00
|
|
|
case kExprI64AtomicExchange16U:
|
|
|
|
|
case kExprI64AtomicCompareExchange16U:
|
2020-05-11 17:12:45 +00:00
|
|
|
case kExprS16x8LoadSplat:
|
2017-11-08 21:03:20 +00:00
|
|
|
return 1;
|
|
|
|
|
case kExprI32LoadMem8S:
|
|
|
|
|
case kExprI32LoadMem8U:
|
|
|
|
|
case kExprI64LoadMem8S:
|
|
|
|
|
case kExprI64LoadMem8U:
|
|
|
|
|
case kExprI32StoreMem8:
|
|
|
|
|
case kExprI64StoreMem8:
|
2020-01-14 18:17:34 +00:00
|
|
|
case kExprI32AtomicStore8U:
|
|
|
|
|
case kExprI64AtomicStore8U:
|
|
|
|
|
case kExprI32AtomicLoad8U:
|
|
|
|
|
case kExprI64AtomicLoad8U:
|
2020-01-21 14:43:23 +00:00
|
|
|
case kExprI32AtomicAdd8U:
|
|
|
|
|
case kExprI32AtomicSub8U:
|
|
|
|
|
case kExprI32AtomicAnd8U:
|
|
|
|
|
case kExprI32AtomicOr8U:
|
|
|
|
|
case kExprI32AtomicXor8U:
|
2020-01-30 14:34:16 +00:00
|
|
|
case kExprI32AtomicExchange8U:
|
|
|
|
|
case kExprI32AtomicCompareExchange8U:
|
2020-01-21 14:43:23 +00:00
|
|
|
case kExprI64AtomicAdd8U:
|
|
|
|
|
case kExprI64AtomicSub8U:
|
|
|
|
|
case kExprI64AtomicAnd8U:
|
|
|
|
|
case kExprI64AtomicOr8U:
|
|
|
|
|
case kExprI64AtomicXor8U:
|
2020-01-30 14:34:16 +00:00
|
|
|
case kExprI64AtomicExchange8U:
|
|
|
|
|
case kExprI64AtomicCompareExchange8U:
|
2020-05-11 17:12:45 +00:00
|
|
|
case kExprS8x16LoadSplat:
|
2017-11-08 21:03:20 +00:00
|
|
|
return 0;
|
|
|
|
|
default:
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
template <WasmOpcode memory_op, ValueType::Kind... arg_types>
|
2019-07-08 11:44:42 +00:00
|
|
|
void memop(DataRange* data) {
|
|
|
|
|
const uint8_t align = data->get<uint8_t>() % (max_alignment(memory_op) + 1);
|
|
|
|
|
const uint32_t offset = data->get<uint32_t>();
|
2017-11-06 15:48:17 +00:00
|
|
|
|
2017-11-08 10:25:14 +00:00
|
|
|
// Generate the index and the arguments, if any.
|
2020-03-12 14:29:51 +00:00
|
|
|
Generate<ValueType::kI32, arg_types...>(data);
|
2017-11-06 15:48:17 +00:00
|
|
|
|
2020-03-09 20:52:40 +00:00
|
|
|
if (WasmOpcodes::IsPrefixOpcode(static_cast<WasmOpcode>(memory_op >> 8))) {
|
2020-03-04 22:53:37 +00:00
|
|
|
DCHECK(memory_op >> 8 == kAtomicPrefix || memory_op >> 8 == kSimdPrefix);
|
2020-01-14 18:17:34 +00:00
|
|
|
builder_->EmitWithPrefix(memory_op);
|
|
|
|
|
} else {
|
|
|
|
|
builder_->Emit(memory_op);
|
|
|
|
|
}
|
2017-11-06 15:48:17 +00:00
|
|
|
builder_->EmitU32V(align);
|
|
|
|
|
builder_->EmitU32V(offset);
|
|
|
|
|
}
|
|
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
template <WasmOpcode Op, ValueType::Kind... Args>
|
2020-01-21 14:43:23 +00:00
|
|
|
void atomic_op(DataRange* data) {
|
|
|
|
|
const uint8_t align = data->get<uint8_t>() % (max_alignment(Op) + 1);
|
|
|
|
|
const uint32_t offset = data->get<uint32_t>();
|
|
|
|
|
|
|
|
|
|
Generate<Args...>(data);
|
|
|
|
|
builder_->EmitWithPrefix(Op);
|
|
|
|
|
|
|
|
|
|
builder_->EmitU32V(align);
|
|
|
|
|
builder_->EmitU32V(offset);
|
|
|
|
|
}
|
|
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
template <WasmOpcode Op, ValueType::Kind... Args>
|
2020-05-20 07:08:11 +00:00
|
|
|
void op_with_prefix(DataRange* data) {
|
2020-02-06 18:26:29 +00:00
|
|
|
Generate<Args...>(data);
|
|
|
|
|
builder_->EmitWithPrefix(Op);
|
|
|
|
|
}
|
|
|
|
|
|
2020-07-08 19:10:14 +00:00
|
|
|
void simd_const(DataRange* data) {
|
|
|
|
|
builder_->EmitWithPrefix(kExprS128Const);
|
|
|
|
|
for (int i = 0; i < kSimd128Size; i++) {
|
|
|
|
|
builder_->EmitByte(data->get<byte>());
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2020-05-28 00:06:31 +00:00
|
|
|
template <WasmOpcode Op, int lanes, ValueType::Kind... Args>
|
|
|
|
|
void simd_lane_op(DataRange* data) {
|
|
|
|
|
Generate<Args...>(data);
|
|
|
|
|
builder_->EmitWithPrefix(Op);
|
|
|
|
|
builder_->EmitByte(data->get<byte>() % lanes);
|
|
|
|
|
}
|
|
|
|
|
|
2020-04-03 23:56:53 +00:00
|
|
|
void simd_shuffle(DataRange* data) {
|
|
|
|
|
Generate<ValueType::kS128, ValueType::kS128>(data);
|
|
|
|
|
builder_->EmitWithPrefix(kExprS8x16Shuffle);
|
|
|
|
|
for (int i = 0; i < kSimd128Size; i++) {
|
|
|
|
|
builder_->EmitByte(static_cast<uint8_t>(data->get<byte>() % 32));
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2019-07-08 11:44:42 +00:00
|
|
|
void drop(DataRange* data) {
|
2017-12-21 13:53:15 +00:00
|
|
|
Generate(GetValueType(data), data);
|
|
|
|
|
builder_->Emit(kExprDrop);
|
|
|
|
|
}
|
|
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
template <ValueType::Kind wanted_type>
|
2019-07-08 11:44:42 +00:00
|
|
|
void call(DataRange* data) {
|
[wasm-gc] Change ValueType representation to account for new types
Motivation:
Changes to the typed function references and gc proposals solidified
the notion of heap type, clarified nullable vs. non-nullable reference
types, and introduced rtts, which contain an integer depth field in
addition to a heap type. This required us to overhaul our ValueType
representation, which results in extensive changes.
To keep this CL "small", we do not try to implement the binary encoding
as described in the proposals, but rather devise a simpler one of our
own (see below). Also, we do not try to implement additional
functionality for the new types.
Changes:
- Introduce HeapType. Move heap types from ValueType to HeapType.
- Introduce Nullability for reference types.
- Rework ValueType helper methods.
- Introduce rtts in ValueType with an integer depth field. Include depth
in the ValueType encoding.
- Make the constructor of ValueType private, instead expose static
functions which explicitly state what they create.
- Change every switch statement on ValueType::Kind. Sometimes, we need
nested switches.
- Introduce temporary constants in ValueTypeCode for nullable types,
use them for decoding.
- In WasmGlobalObject, split 'flags' into 'raw_type' and 'is_mutable'.
- Change IsSubtypeOfRef to IsSubtypeOfHeap and implement changes in
subtyping.
- kWasmFuncRef initializers are now non-nullable. Initializers are
only required to be subtypes of the declared global type.
- Change tests and fuzzers as needed.
Bug: v8:7748
Change-Id: If41f783bd4128443b07e94188cea7dd53ab0bfa5
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2247657
Commit-Queue: Manos Koukoutos <manoskouk@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68408}
2020-06-18 11:24:07 +00:00
|
|
|
call(data, ValueType::Primitive(wanted_type));
|
2017-12-21 14:31:42 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void Convert(ValueType src, ValueType dst) {
|
|
|
|
|
auto idx = [](ValueType t) -> int {
|
2020-03-12 14:29:51 +00:00
|
|
|
switch (t.kind()) {
|
|
|
|
|
case ValueType::kI32:
|
2017-12-21 14:31:42 +00:00
|
|
|
return 0;
|
2020-03-12 14:29:51 +00:00
|
|
|
case ValueType::kI64:
|
2017-12-21 14:31:42 +00:00
|
|
|
return 1;
|
2020-03-12 14:29:51 +00:00
|
|
|
case ValueType::kF32:
|
2017-12-21 14:31:42 +00:00
|
|
|
return 2;
|
2020-03-12 14:29:51 +00:00
|
|
|
case ValueType::kF64:
|
2017-12-21 14:31:42 +00:00
|
|
|
return 3;
|
|
|
|
|
default:
|
|
|
|
|
UNREACHABLE();
|
|
|
|
|
}
|
|
|
|
|
};
|
|
|
|
|
static constexpr WasmOpcode kConvertOpcodes[] = {
|
|
|
|
|
// {i32, i64, f32, f64} -> i32
|
|
|
|
|
kExprNop, kExprI32ConvertI64, kExprI32SConvertF32, kExprI32SConvertF64,
|
|
|
|
|
// {i32, i64, f32, f64} -> i64
|
|
|
|
|
kExprI64SConvertI32, kExprNop, kExprI64SConvertF32, kExprI64SConvertF64,
|
|
|
|
|
// {i32, i64, f32, f64} -> f32
|
|
|
|
|
kExprF32SConvertI32, kExprF32SConvertI64, kExprNop, kExprF32ConvertF64,
|
|
|
|
|
// {i32, i64, f32, f64} -> f64
|
|
|
|
|
kExprF64SConvertI32, kExprF64SConvertI64, kExprF64ConvertF32, kExprNop};
|
|
|
|
|
int arr_idx = idx(dst) << 2 | idx(src);
|
|
|
|
|
builder_->Emit(kConvertOpcodes[arr_idx]);
|
|
|
|
|
}
|
|
|
|
|
|
2019-07-08 11:44:42 +00:00
|
|
|
void ConvertOrGenerate(ValueType src, ValueType dst, DataRange* data) {
|
2018-01-29 16:42:03 +00:00
|
|
|
if (src == dst) return;
|
|
|
|
|
if (src == kWasmStmt && dst != kWasmStmt) {
|
|
|
|
|
Generate(dst, data);
|
|
|
|
|
} else if (dst == kWasmStmt && src != kWasmStmt) {
|
|
|
|
|
builder_->Emit(kExprDrop);
|
|
|
|
|
} else {
|
|
|
|
|
Convert(src, dst);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2019-07-08 11:44:42 +00:00
|
|
|
void call(DataRange* data, ValueType wanted_type) {
|
|
|
|
|
int func_index = data->get<uint8_t>() % functions_.size();
|
2017-12-21 14:31:42 +00:00
|
|
|
FunctionSig* sig = functions_[func_index];
|
|
|
|
|
// Generate arguments.
|
|
|
|
|
for (size_t i = 0; i < sig->parameter_count(); ++i) {
|
|
|
|
|
Generate(sig->GetParam(i), data);
|
|
|
|
|
}
|
|
|
|
|
// Emit call.
|
|
|
|
|
builder_->EmitWithU32V(kExprCallFunction, func_index);
|
2020-05-12 14:39:52 +00:00
|
|
|
if (sig->return_count() == 0 && wanted_type != kWasmStmt) {
|
2017-12-21 14:31:42 +00:00
|
|
|
// The call did not generate a value. Thus just generate it here.
|
|
|
|
|
Generate(wanted_type, data);
|
2020-05-12 14:39:52 +00:00
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
if (wanted_type == kWasmStmt) {
|
|
|
|
|
// The call did generate values, but we did not want one.
|
|
|
|
|
for (size_t i = 0; i < sig->return_count(); ++i) {
|
|
|
|
|
builder_->Emit(kExprDrop);
|
|
|
|
|
}
|
|
|
|
|
return;
|
|
|
|
|
}
|
2020-06-18 12:55:08 +00:00
|
|
|
auto return_types = VectorOf(sig->returns().begin(), sig->return_count());
|
|
|
|
|
auto wanted_types =
|
|
|
|
|
VectorOf(&wanted_type, wanted_type == kWasmStmt ? 0 : 1);
|
2020-05-18 08:31:21 +00:00
|
|
|
ConsumeAndGenerate(return_types, wanted_types, data);
|
2017-12-21 14:31:42 +00:00
|
|
|
}
|
|
|
|
|
|
2018-01-29 16:45:59 +00:00
|
|
|
struct Var {
|
2018-01-02 18:52:49 +00:00
|
|
|
uint32_t index;
|
|
|
|
|
ValueType type = kWasmStmt;
|
2018-01-29 16:45:59 +00:00
|
|
|
Var() = default;
|
|
|
|
|
Var(uint32_t index, ValueType type) : index(index), type(type) {}
|
2018-01-02 18:52:49 +00:00
|
|
|
bool is_valid() const { return type != kWasmStmt; }
|
|
|
|
|
};
|
|
|
|
|
|
2019-07-08 11:44:42 +00:00
|
|
|
Var GetRandomLocal(DataRange* data) {
|
2018-01-02 18:52:49 +00:00
|
|
|
uint32_t num_params =
|
|
|
|
|
static_cast<uint32_t>(builder_->signature()->parameter_count());
|
|
|
|
|
uint32_t num_locals = static_cast<uint32_t>(locals_.size());
|
|
|
|
|
if (num_params + num_locals == 0) return {};
|
2019-07-08 11:44:42 +00:00
|
|
|
uint32_t index = data->get<uint8_t>() % (num_params + num_locals);
|
2018-01-02 18:52:49 +00:00
|
|
|
ValueType type = index < num_params ? builder_->signature()->GetParam(index)
|
|
|
|
|
: locals_[index - num_params];
|
|
|
|
|
return {index, type};
|
|
|
|
|
}
|
|
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
template <ValueType::Kind wanted_type>
|
2019-07-08 11:44:42 +00:00
|
|
|
void local_op(DataRange* data, WasmOpcode opcode) {
|
2018-01-29 16:45:59 +00:00
|
|
|
Var local = GetRandomLocal(data);
|
2018-01-11 12:56:19 +00:00
|
|
|
// If there are no locals and no parameters, just generate any value (if a
|
|
|
|
|
// value is needed), or do nothing.
|
|
|
|
|
if (!local.is_valid()) {
|
2020-03-12 14:29:51 +00:00
|
|
|
if (wanted_type == ValueType::kStmt) return;
|
2018-01-11 12:56:19 +00:00
|
|
|
return Generate<wanted_type>(data);
|
|
|
|
|
}
|
2018-01-02 18:52:49 +00:00
|
|
|
|
2019-10-08 12:38:48 +00:00
|
|
|
if (opcode != kExprLocalGet) Generate(local.type, data);
|
2018-01-11 12:56:19 +00:00
|
|
|
builder_->EmitWithU32V(opcode, local.index);
|
2020-03-12 14:29:51 +00:00
|
|
|
if (wanted_type != ValueType::kStmt && local.type.kind() != wanted_type) {
|
[wasm-gc] Change ValueType representation to account for new types
Motivation:
Changes to the typed function references and gc proposals solidified
the notion of heap type, clarified nullable vs. non-nullable reference
types, and introduced rtts, which contain an integer depth field in
addition to a heap type. This required us to overhaul our ValueType
representation, which results in extensive changes.
To keep this CL "small", we do not try to implement the binary encoding
as described in the proposals, but rather devise a simpler one of our
own (see below). Also, we do not try to implement additional
functionality for the new types.
Changes:
- Introduce HeapType. Move heap types from ValueType to HeapType.
- Introduce Nullability for reference types.
- Rework ValueType helper methods.
- Introduce rtts in ValueType with an integer depth field. Include depth
in the ValueType encoding.
- Make the constructor of ValueType private, instead expose static
functions which explicitly state what they create.
- Change every switch statement on ValueType::Kind. Sometimes, we need
nested switches.
- Introduce temporary constants in ValueTypeCode for nullable types,
use them for decoding.
- In WasmGlobalObject, split 'flags' into 'raw_type' and 'is_mutable'.
- Change IsSubtypeOfRef to IsSubtypeOfHeap and implement changes in
subtyping.
- kWasmFuncRef initializers are now non-nullable. Initializers are
only required to be subtypes of the declared global type.
- Change tests and fuzzers as needed.
Bug: v8:7748
Change-Id: If41f783bd4128443b07e94188cea7dd53ab0bfa5
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2247657
Commit-Queue: Manos Koukoutos <manoskouk@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68408}
2020-06-18 11:24:07 +00:00
|
|
|
Convert(local.type, ValueType::Primitive(wanted_type));
|
2018-01-11 12:56:19 +00:00
|
|
|
}
|
2018-01-02 18:52:49 +00:00
|
|
|
}
|
|
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
template <ValueType::Kind wanted_type>
|
2019-07-08 11:44:42 +00:00
|
|
|
void get_local(DataRange* data) {
|
2020-03-12 14:29:51 +00:00
|
|
|
static_assert(wanted_type != ValueType::kStmt, "illegal type");
|
2019-10-08 12:38:48 +00:00
|
|
|
local_op<wanted_type>(data, kExprLocalGet);
|
2018-01-11 12:56:19 +00:00
|
|
|
}
|
|
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
void set_local(DataRange* data) {
|
|
|
|
|
local_op<ValueType::kStmt>(data, kExprLocalSet);
|
|
|
|
|
}
|
2018-01-02 18:52:49 +00:00
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
template <ValueType::Kind wanted_type>
|
2019-07-08 11:44:42 +00:00
|
|
|
void tee_local(DataRange* data) {
|
2019-10-08 12:38:48 +00:00
|
|
|
local_op<wanted_type>(data, kExprLocalTee);
|
2018-01-02 18:52:49 +00:00
|
|
|
}
|
|
|
|
|
|
2018-11-15 15:08:32 +00:00
|
|
|
template <size_t num_bytes>
|
2019-07-08 11:44:42 +00:00
|
|
|
void i32_const(DataRange* data) {
|
|
|
|
|
builder_->EmitI32Const(data->get<int32_t, num_bytes>());
|
2018-11-15 15:08:32 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
template <size_t num_bytes>
|
2019-07-08 11:44:42 +00:00
|
|
|
void i64_const(DataRange* data) {
|
|
|
|
|
builder_->EmitI64Const(data->get<int64_t, num_bytes>());
|
2018-11-15 15:08:32 +00:00
|
|
|
}
|
|
|
|
|
|
2019-07-08 11:44:42 +00:00
|
|
|
Var GetRandomGlobal(DataRange* data, bool ensure_mutable) {
|
2018-01-29 16:45:59 +00:00
|
|
|
uint32_t index;
|
|
|
|
|
if (ensure_mutable) {
|
|
|
|
|
if (mutable_globals_.empty()) return {};
|
2019-07-08 11:44:42 +00:00
|
|
|
index = mutable_globals_[data->get<uint8_t>() % mutable_globals_.size()];
|
2018-01-29 16:45:59 +00:00
|
|
|
} else {
|
|
|
|
|
if (globals_.empty()) return {};
|
2019-07-08 11:44:42 +00:00
|
|
|
index = data->get<uint8_t>() % globals_.size();
|
2018-01-29 16:45:59 +00:00
|
|
|
}
|
|
|
|
|
ValueType type = globals_[index];
|
|
|
|
|
return {index, type};
|
|
|
|
|
}
|
|
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
template <ValueType::Kind wanted_type>
|
2019-07-08 11:44:42 +00:00
|
|
|
void global_op(DataRange* data) {
|
2020-03-12 14:29:51 +00:00
|
|
|
constexpr bool is_set = wanted_type == ValueType::kStmt;
|
2018-01-29 16:45:59 +00:00
|
|
|
Var global = GetRandomGlobal(data, is_set);
|
|
|
|
|
// If there are no globals, just generate any value (if a value is needed),
|
|
|
|
|
// or do nothing.
|
|
|
|
|
if (!global.is_valid()) {
|
2020-03-12 14:29:51 +00:00
|
|
|
if (wanted_type == ValueType::kStmt) return;
|
2018-01-29 16:45:59 +00:00
|
|
|
return Generate<wanted_type>(data);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (is_set) Generate(global.type, data);
|
2019-10-08 13:16:30 +00:00
|
|
|
builder_->EmitWithU32V(is_set ? kExprGlobalSet : kExprGlobalGet,
|
2018-01-29 16:45:59 +00:00
|
|
|
global.index);
|
2020-03-12 14:29:51 +00:00
|
|
|
if (!is_set && global.type.kind() != wanted_type) {
|
[wasm-gc] Change ValueType representation to account for new types
Motivation:
Changes to the typed function references and gc proposals solidified
the notion of heap type, clarified nullable vs. non-nullable reference
types, and introduced rtts, which contain an integer depth field in
addition to a heap type. This required us to overhaul our ValueType
representation, which results in extensive changes.
To keep this CL "small", we do not try to implement the binary encoding
as described in the proposals, but rather devise a simpler one of our
own (see below). Also, we do not try to implement additional
functionality for the new types.
Changes:
- Introduce HeapType. Move heap types from ValueType to HeapType.
- Introduce Nullability for reference types.
- Rework ValueType helper methods.
- Introduce rtts in ValueType with an integer depth field. Include depth
in the ValueType encoding.
- Make the constructor of ValueType private, instead expose static
functions which explicitly state what they create.
- Change every switch statement on ValueType::Kind. Sometimes, we need
nested switches.
- Introduce temporary constants in ValueTypeCode for nullable types,
use them for decoding.
- In WasmGlobalObject, split 'flags' into 'raw_type' and 'is_mutable'.
- Change IsSubtypeOfRef to IsSubtypeOfHeap and implement changes in
subtyping.
- kWasmFuncRef initializers are now non-nullable. Initializers are
only required to be subtypes of the declared global type.
- Change tests and fuzzers as needed.
Bug: v8:7748
Change-Id: If41f783bd4128443b07e94188cea7dd53ab0bfa5
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2247657
Commit-Queue: Manos Koukoutos <manoskouk@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68408}
2020-06-18 11:24:07 +00:00
|
|
|
Convert(global.type, ValueType::Primitive(wanted_type));
|
2018-01-29 16:45:59 +00:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
template <ValueType::Kind wanted_type>
|
2019-07-08 11:44:42 +00:00
|
|
|
void get_global(DataRange* data) {
|
2020-03-12 14:29:51 +00:00
|
|
|
static_assert(wanted_type != ValueType::kStmt, "illegal type");
|
2018-01-29 16:45:59 +00:00
|
|
|
global_op<wanted_type>(data);
|
|
|
|
|
}
|
|
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
template <ValueType::Kind select_type>
|
2019-07-08 11:44:42 +00:00
|
|
|
void select_with_type(DataRange* data) {
|
2020-03-12 14:29:51 +00:00
|
|
|
static_assert(select_type != ValueType::kStmt, "illegal type for select");
|
|
|
|
|
Generate<select_type, select_type, ValueType::kI32>(data);
|
2019-05-28 10:06:42 +00:00
|
|
|
// num_types is always 1.
|
|
|
|
|
uint8_t num_types = 1;
|
|
|
|
|
builder_->EmitWithU8U8(kExprSelectWithType, num_types,
|
[wasm-gc] Change ValueType representation to account for new types
Motivation:
Changes to the typed function references and gc proposals solidified
the notion of heap type, clarified nullable vs. non-nullable reference
types, and introduced rtts, which contain an integer depth field in
addition to a heap type. This required us to overhaul our ValueType
representation, which results in extensive changes.
To keep this CL "small", we do not try to implement the binary encoding
as described in the proposals, but rather devise a simpler one of our
own (see below). Also, we do not try to implement additional
functionality for the new types.
Changes:
- Introduce HeapType. Move heap types from ValueType to HeapType.
- Introduce Nullability for reference types.
- Rework ValueType helper methods.
- Introduce rtts in ValueType with an integer depth field. Include depth
in the ValueType encoding.
- Make the constructor of ValueType private, instead expose static
functions which explicitly state what they create.
- Change every switch statement on ValueType::Kind. Sometimes, we need
nested switches.
- Introduce temporary constants in ValueTypeCode for nullable types,
use them for decoding.
- In WasmGlobalObject, split 'flags' into 'raw_type' and 'is_mutable'.
- Change IsSubtypeOfRef to IsSubtypeOfHeap and implement changes in
subtyping.
- kWasmFuncRef initializers are now non-nullable. Initializers are
only required to be subtypes of the declared global type.
- Change tests and fuzzers as needed.
Bug: v8:7748
Change-Id: If41f783bd4128443b07e94188cea7dd53ab0bfa5
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2247657
Commit-Queue: Manos Koukoutos <manoskouk@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68408}
2020-06-18 11:24:07 +00:00
|
|
|
ValueType::Primitive(select_type).value_type_code());
|
2019-05-28 10:06:42 +00:00
|
|
|
}
|
|
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
void set_global(DataRange* data) { global_op<ValueType::kStmt>(data); }
|
2018-01-29 16:45:59 +00:00
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
template <ValueType::Kind... Types>
|
2019-07-08 11:44:42 +00:00
|
|
|
void sequence(DataRange* data) {
|
2018-08-01 15:14:06 +00:00
|
|
|
Generate<Types...>(data);
|
2017-11-06 15:48:17 +00:00
|
|
|
}
|
|
|
|
|
|
2019-07-08 11:44:42 +00:00
|
|
|
void current_memory(DataRange* data) {
|
2017-11-06 15:48:17 +00:00
|
|
|
builder_->EmitWithU8(kExprMemorySize, 0);
|
|
|
|
|
}
|
|
|
|
|
|
2019-07-08 11:44:42 +00:00
|
|
|
void grow_memory(DataRange* data);
|
2017-11-07 17:38:57 +00:00
|
|
|
|
2019-10-17 08:46:01 +00:00
|
|
|
using GenerateFn = void (WasmGenerator::*const)(DataRange*);
|
2017-11-06 15:48:17 +00:00
|
|
|
|
|
|
|
|
template <size_t N>
|
2019-10-17 08:46:01 +00:00
|
|
|
void GenerateOneOf(GenerateFn (&alternatives)[N], DataRange* data) {
|
2017-11-06 15:48:17 +00:00
|
|
|
static_assert(N < std::numeric_limits<uint8_t>::max(),
|
2019-10-17 08:46:01 +00:00
|
|
|
"Too many alternatives. Use a bigger type if needed.");
|
2019-07-08 11:44:42 +00:00
|
|
|
const auto which = data->get<uint8_t>();
|
2017-11-06 15:48:17 +00:00
|
|
|
|
2019-10-17 08:46:01 +00:00
|
|
|
GenerateFn alternate = alternatives[which % N];
|
2017-11-06 15:48:17 +00:00
|
|
|
(this->*alternate)(data);
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
}
|
|
|
|
|
|
2017-11-03 16:43:29 +00:00
|
|
|
struct GeneratorRecursionScope {
|
|
|
|
|
explicit GeneratorRecursionScope(WasmGenerator* gen) : gen(gen) {
|
|
|
|
|
++gen->recursion_depth;
|
2017-11-06 15:48:17 +00:00
|
|
|
DCHECK_LE(gen->recursion_depth, kMaxRecursionDepth);
|
2017-11-03 16:43:29 +00:00
|
|
|
}
|
|
|
|
|
~GeneratorRecursionScope() {
|
|
|
|
|
DCHECK_GT(gen->recursion_depth, 0);
|
|
|
|
|
--gen->recursion_depth;
|
|
|
|
|
}
|
|
|
|
|
WasmGenerator* gen;
|
|
|
|
|
};
|
|
|
|
|
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
public:
|
2018-01-02 18:52:49 +00:00
|
|
|
WasmGenerator(WasmFunctionBuilder* fn,
|
2018-01-29 16:45:59 +00:00
|
|
|
const std::vector<FunctionSig*>& functions,
|
|
|
|
|
const std::vector<ValueType>& globals,
|
2019-07-08 11:44:42 +00:00
|
|
|
const std::vector<uint8_t>& mutable_globals, DataRange* data)
|
2018-01-29 16:45:59 +00:00
|
|
|
: builder_(fn),
|
|
|
|
|
functions_(functions),
|
|
|
|
|
globals_(globals),
|
|
|
|
|
mutable_globals_(mutable_globals) {
|
2017-12-21 14:31:42 +00:00
|
|
|
FunctionSig* sig = fn->signature();
|
2020-05-20 09:21:27 +00:00
|
|
|
blocks_.emplace_back();
|
|
|
|
|
for (size_t i = 0; i < sig->return_count(); ++i) {
|
|
|
|
|
blocks_.back().push_back(sig->GetReturn(i));
|
2020-05-12 14:39:52 +00:00
|
|
|
}
|
2018-01-02 18:52:49 +00:00
|
|
|
|
|
|
|
|
constexpr uint32_t kMaxLocals = 32;
|
2019-07-08 11:44:42 +00:00
|
|
|
locals_.resize(data->get<uint8_t>() % kMaxLocals);
|
2018-01-02 18:52:49 +00:00
|
|
|
for (ValueType& local : locals_) {
|
|
|
|
|
local = GetValueType(data);
|
|
|
|
|
fn->AddLocal(local);
|
|
|
|
|
}
|
2017-11-15 11:51:15 +00:00
|
|
|
}
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
|
2019-07-08 11:44:42 +00:00
|
|
|
void Generate(ValueType type, DataRange* data);
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
template <ValueType::Kind T>
|
2019-07-08 11:44:42 +00:00
|
|
|
void Generate(DataRange* data);
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
template <ValueType::Kind T1, ValueType::Kind T2, ValueType::Kind... Ts>
|
2019-07-08 11:44:42 +00:00
|
|
|
void Generate(DataRange* data) {
|
2019-09-26 08:16:50 +00:00
|
|
|
// TODO(clemensb): Implement a more even split.
|
2019-07-08 11:44:42 +00:00
|
|
|
auto first_data = data->split();
|
|
|
|
|
Generate<T1>(&first_data);
|
2017-12-21 14:30:32 +00:00
|
|
|
Generate<T2, Ts...>(data);
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
}
|
|
|
|
|
|
2020-05-18 08:31:21 +00:00
|
|
|
std::vector<ValueType> GenerateTypes(DataRange* data);
|
|
|
|
|
void Generate(Vector<const ValueType> types, DataRange* data);
|
2020-06-18 12:55:08 +00:00
|
|
|
void ConsumeAndGenerate(Vector<const ValueType> parameter_types,
|
|
|
|
|
Vector<const ValueType> return_types,
|
2020-05-18 08:31:21 +00:00
|
|
|
DataRange* data);
|
|
|
|
|
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
private:
|
2017-09-01 13:20:46 +00:00
|
|
|
WasmFunctionBuilder* builder_;
|
2020-05-12 14:39:52 +00:00
|
|
|
std::vector<std::vector<ValueType>> blocks_;
|
2017-12-21 14:31:42 +00:00
|
|
|
const std::vector<FunctionSig*>& functions_;
|
2018-01-02 18:52:49 +00:00
|
|
|
std::vector<ValueType> locals_;
|
2018-01-29 16:45:59 +00:00
|
|
|
std::vector<ValueType> globals_;
|
|
|
|
|
std::vector<uint8_t> mutable_globals_; // indexes into {globals_}.
|
2017-11-03 16:43:29 +00:00
|
|
|
uint32_t recursion_depth = 0;
|
|
|
|
|
|
|
|
|
|
static constexpr uint32_t kMaxRecursionDepth = 64;
|
|
|
|
|
|
|
|
|
|
bool recursion_limit_reached() {
|
|
|
|
|
return recursion_depth >= kMaxRecursionDepth;
|
|
|
|
|
}
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
};
|
|
|
|
|
|
2020-05-18 08:31:21 +00:00
|
|
|
template <>
|
|
|
|
|
void WasmGenerator::block<ValueType::kStmt>(DataRange* data) {
|
|
|
|
|
block({}, {}, data);
|
|
|
|
|
}
|
|
|
|
|
|
2020-05-20 09:21:27 +00:00
|
|
|
template <>
|
|
|
|
|
void WasmGenerator::loop<ValueType::kStmt>(DataRange* data) {
|
|
|
|
|
loop({}, {}, data);
|
|
|
|
|
}
|
|
|
|
|
|
2017-11-06 15:07:14 +00:00
|
|
|
template <>
|
2020-03-12 14:29:51 +00:00
|
|
|
void WasmGenerator::Generate<ValueType::kStmt>(DataRange* data) {
|
2017-11-06 15:07:14 +00:00
|
|
|
GeneratorRecursionScope rec_scope(this);
|
2019-07-08 11:44:42 +00:00
|
|
|
if (recursion_limit_reached() || data->size() == 0) return;
|
2017-11-06 15:48:17 +00:00
|
|
|
|
2019-10-17 08:46:01 +00:00
|
|
|
constexpr GenerateFn alternatives[] = {
|
2020-03-12 14:29:51 +00:00
|
|
|
&WasmGenerator::sequence<ValueType::kStmt, ValueType::kStmt>,
|
|
|
|
|
&WasmGenerator::sequence<ValueType::kStmt, ValueType::kStmt,
|
|
|
|
|
ValueType::kStmt, ValueType::kStmt>,
|
|
|
|
|
&WasmGenerator::sequence<ValueType::kStmt, ValueType::kStmt,
|
|
|
|
|
ValueType::kStmt, ValueType::kStmt,
|
|
|
|
|
ValueType::kStmt, ValueType::kStmt,
|
|
|
|
|
ValueType::kStmt, ValueType::kStmt>,
|
|
|
|
|
&WasmGenerator::block<ValueType::kStmt>,
|
|
|
|
|
&WasmGenerator::loop<ValueType::kStmt>,
|
|
|
|
|
&WasmGenerator::if_<ValueType::kStmt, kIf>,
|
|
|
|
|
&WasmGenerator::if_<ValueType::kStmt, kIfElse>,
|
2017-11-15 11:51:15 +00:00
|
|
|
&WasmGenerator::br,
|
2020-03-12 14:29:51 +00:00
|
|
|
&WasmGenerator::br_if<ValueType::kStmt>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::memop<kExprI32StoreMem, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::memop<kExprI32StoreMem8, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::memop<kExprI32StoreMem16, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::memop<kExprI64StoreMem, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::memop<kExprI64StoreMem8, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::memop<kExprI64StoreMem16, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::memop<kExprI64StoreMem32, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::memop<kExprF32StoreMem, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::memop<kExprF64StoreMem, ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::memop<kExprI32AtomicStore, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::memop<kExprI32AtomicStore8U, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::memop<kExprI32AtomicStore16U, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::memop<kExprI64AtomicStore, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::memop<kExprI64AtomicStore8U, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::memop<kExprI64AtomicStore16U, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::memop<kExprI64AtomicStore32U, ValueType::kI64>,
|
2020-03-12 23:53:38 +00:00
|
|
|
&WasmGenerator::memop<kExprS128StoreMem, ValueType::kS128>,
|
2017-12-21 13:53:15 +00:00
|
|
|
|
2017-12-21 14:31:42 +00:00
|
|
|
&WasmGenerator::drop,
|
|
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
&WasmGenerator::call<ValueType::kStmt>,
|
2018-01-11 12:56:19 +00:00
|
|
|
|
2018-01-29 16:45:59 +00:00
|
|
|
&WasmGenerator::set_local,
|
|
|
|
|
&WasmGenerator::set_global};
|
2017-11-06 15:07:14 +00:00
|
|
|
|
2019-10-17 08:46:01 +00:00
|
|
|
GenerateOneOf(alternatives, data);
|
2017-11-06 15:07:14 +00:00
|
|
|
}
|
|
|
|
|
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
template <>
|
2020-03-12 14:29:51 +00:00
|
|
|
void WasmGenerator::Generate<ValueType::kI32>(DataRange* data) {
|
2017-11-03 16:43:29 +00:00
|
|
|
GeneratorRecursionScope rec_scope(this);
|
2019-07-08 11:44:42 +00:00
|
|
|
if (recursion_limit_reached() || data->size() <= 1) {
|
|
|
|
|
builder_->EmitI32Const(data->get<uint32_t>());
|
2017-11-06 15:48:17 +00:00
|
|
|
return;
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
}
|
2017-11-06 15:48:17 +00:00
|
|
|
|
2019-10-17 08:46:01 +00:00
|
|
|
constexpr GenerateFn alternatives[] = {
|
2018-11-15 15:08:32 +00:00
|
|
|
&WasmGenerator::i32_const<1>,
|
|
|
|
|
&WasmGenerator::i32_const<2>,
|
|
|
|
|
&WasmGenerator::i32_const<3>,
|
|
|
|
|
&WasmGenerator::i32_const<4>,
|
|
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
&WasmGenerator::sequence<ValueType::kI32, ValueType::kStmt>,
|
|
|
|
|
&WasmGenerator::sequence<ValueType::kStmt, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::sequence<ValueType::kStmt, ValueType::kI32,
|
|
|
|
|
ValueType::kStmt>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::op<kExprI32Eqz, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op<kExprI32Eq, ValueType::kI32, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op<kExprI32Ne, ValueType::kI32, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op<kExprI32LtS, ValueType::kI32, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op<kExprI32LtU, ValueType::kI32, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op<kExprI32GeS, ValueType::kI32, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op<kExprI32GeU, ValueType::kI32, ValueType::kI32>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::op<kExprI64Eqz, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprI64Eq, ValueType::kI64, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprI64Ne, ValueType::kI64, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprI64LtS, ValueType::kI64, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprI64LtU, ValueType::kI64, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprI64GeS, ValueType::kI64, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprI64GeU, ValueType::kI64, ValueType::kI64>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::op<kExprF32Eq, ValueType::kF32, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::op<kExprF32Ne, ValueType::kF32, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::op<kExprF32Lt, ValueType::kF32, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::op<kExprF32Ge, ValueType::kF32, ValueType::kF32>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::op<kExprF64Eq, ValueType::kF64, ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::op<kExprF64Ne, ValueType::kF64, ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::op<kExprF64Lt, ValueType::kF64, ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::op<kExprF64Ge, ValueType::kF64, ValueType::kF64>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::op<kExprI32Add, ValueType::kI32, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op<kExprI32Sub, ValueType::kI32, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op<kExprI32Mul, ValueType::kI32, ValueType::kI32>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::op<kExprI32DivS, ValueType::kI32, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op<kExprI32DivU, ValueType::kI32, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op<kExprI32RemS, ValueType::kI32, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op<kExprI32RemU, ValueType::kI32, ValueType::kI32>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::op<kExprI32And, ValueType::kI32, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op<kExprI32Ior, ValueType::kI32, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op<kExprI32Xor, ValueType::kI32, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op<kExprI32Shl, ValueType::kI32, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op<kExprI32ShrU, ValueType::kI32, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op<kExprI32ShrS, ValueType::kI32, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op<kExprI32Ror, ValueType::kI32, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op<kExprI32Rol, ValueType::kI32, ValueType::kI32>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::op<kExprI32Clz, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op<kExprI32Ctz, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op<kExprI32Popcnt, ValueType::kI32>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::op<kExprI32ConvertI64, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprI32SConvertF32, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::op<kExprI32UConvertF32, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::op<kExprI32SConvertF64, ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::op<kExprI32UConvertF64, ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::op<kExprI32ReinterpretF32, ValueType::kF32>,
|
|
|
|
|
|
2020-05-20 07:08:11 +00:00
|
|
|
&WasmGenerator::op_with_prefix<kExprI32SConvertSatF32, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32UConvertSatF32, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32SConvertSatF64, ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32UConvertSatF64, ValueType::kF64>,
|
2020-05-14 15:05:44 +00:00
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
&WasmGenerator::block<ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::loop<ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::if_<ValueType::kI32, kIfElse>,
|
|
|
|
|
&WasmGenerator::br_if<ValueType::kI32>,
|
2017-11-06 15:48:17 +00:00
|
|
|
|
|
|
|
|
&WasmGenerator::memop<kExprI32LoadMem>,
|
|
|
|
|
&WasmGenerator::memop<kExprI32LoadMem8S>,
|
|
|
|
|
&WasmGenerator::memop<kExprI32LoadMem8U>,
|
|
|
|
|
&WasmGenerator::memop<kExprI32LoadMem16S>,
|
|
|
|
|
&WasmGenerator::memop<kExprI32LoadMem16U>,
|
2020-01-14 18:17:34 +00:00
|
|
|
&WasmGenerator::memop<kExprI32AtomicLoad>,
|
|
|
|
|
&WasmGenerator::memop<kExprI32AtomicLoad8U>,
|
|
|
|
|
&WasmGenerator::memop<kExprI32AtomicLoad16U>,
|
2017-11-06 15:48:17 +00:00
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
&WasmGenerator::atomic_op<kExprI32AtomicAdd, ValueType::kI32,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI32AtomicSub, ValueType::kI32,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI32AtomicAnd, ValueType::kI32,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI32AtomicOr, ValueType::kI32,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI32AtomicXor, ValueType::kI32,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI32AtomicExchange, ValueType::kI32,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI32AtomicCompareExchange, ValueType::kI32,
|
|
|
|
|
ValueType::kI32, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI32AtomicAdd8U, ValueType::kI32,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI32AtomicSub8U, ValueType::kI32,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI32AtomicAnd8U, ValueType::kI32,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI32AtomicOr8U, ValueType::kI32,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI32AtomicXor8U, ValueType::kI32,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI32AtomicExchange8U, ValueType::kI32,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI32AtomicCompareExchange8U,
|
|
|
|
|
ValueType::kI32, ValueType::kI32,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI32AtomicAdd16U, ValueType::kI32,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI32AtomicSub16U, ValueType::kI32,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI32AtomicAnd16U, ValueType::kI32,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI32AtomicOr16U, ValueType::kI32,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI32AtomicXor16U, ValueType::kI32,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI32AtomicExchange16U, ValueType::kI32,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI32AtomicCompareExchange16U,
|
|
|
|
|
ValueType::kI32, ValueType::kI32,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
|
2020-05-21 18:27:47 +00:00
|
|
|
&WasmGenerator::op_with_prefix<kExprV8x16AnyTrue, ValueType::kS128>,
|
2020-06-12 22:51:09 +00:00
|
|
|
&WasmGenerator::op_with_prefix<kExprV8x16AllTrue, ValueType::kS128>,
|
2020-06-16 01:45:27 +00:00
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16BitMask, ValueType::kS128>,
|
2020-05-21 18:27:47 +00:00
|
|
|
&WasmGenerator::op_with_prefix<kExprV16x8AnyTrue, ValueType::kS128>,
|
2020-06-12 22:51:09 +00:00
|
|
|
&WasmGenerator::op_with_prefix<kExprV16x8AllTrue, ValueType::kS128>,
|
2020-06-16 01:45:27 +00:00
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8BitMask, ValueType::kS128>,
|
2020-05-21 18:27:47 +00:00
|
|
|
&WasmGenerator::op_with_prefix<kExprV32x4AnyTrue, ValueType::kS128>,
|
2020-06-12 22:51:09 +00:00
|
|
|
&WasmGenerator::op_with_prefix<kExprV32x4AllTrue, ValueType::kS128>,
|
2020-06-16 01:45:27 +00:00
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4BitMask, ValueType::kS128>,
|
2020-05-28 00:06:31 +00:00
|
|
|
&WasmGenerator::simd_lane_op<kExprI8x16ExtractLaneS, 16,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::simd_lane_op<kExprI8x16ExtractLaneU, 16,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::simd_lane_op<kExprI16x8ExtractLaneS, 8, ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::simd_lane_op<kExprI16x8ExtractLaneU, 8, ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::simd_lane_op<kExprI32x4ExtractLane, 4, ValueType::kS128>,
|
2020-02-06 18:26:29 +00:00
|
|
|
|
2017-11-06 15:48:17 +00:00
|
|
|
&WasmGenerator::current_memory,
|
2017-12-21 14:31:42 +00:00
|
|
|
&WasmGenerator::grow_memory,
|
|
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
&WasmGenerator::get_local<ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::tee_local<ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::get_global<ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op<kExprSelect, ValueType::kI32, ValueType::kI32,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::select_with_type<ValueType::kI32>,
|
2018-01-02 18:52:49 +00:00
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
&WasmGenerator::call<ValueType::kI32>};
|
2017-11-06 15:48:17 +00:00
|
|
|
|
2019-10-17 08:46:01 +00:00
|
|
|
GenerateOneOf(alternatives, data);
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
template <>
|
2020-03-12 14:29:51 +00:00
|
|
|
void WasmGenerator::Generate<ValueType::kI64>(DataRange* data) {
|
2017-11-03 16:43:29 +00:00
|
|
|
GeneratorRecursionScope rec_scope(this);
|
2019-07-08 11:44:42 +00:00
|
|
|
if (recursion_limit_reached() || data->size() <= 1) {
|
|
|
|
|
builder_->EmitI64Const(data->get<int64_t>());
|
2017-11-06 15:48:17 +00:00
|
|
|
return;
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
}
|
2017-11-06 15:48:17 +00:00
|
|
|
|
2019-10-17 08:46:01 +00:00
|
|
|
constexpr GenerateFn alternatives[] = {
|
2018-11-15 15:08:32 +00:00
|
|
|
&WasmGenerator::i64_const<1>,
|
|
|
|
|
&WasmGenerator::i64_const<2>,
|
|
|
|
|
&WasmGenerator::i64_const<3>,
|
|
|
|
|
&WasmGenerator::i64_const<4>,
|
|
|
|
|
&WasmGenerator::i64_const<5>,
|
|
|
|
|
&WasmGenerator::i64_const<6>,
|
|
|
|
|
&WasmGenerator::i64_const<7>,
|
|
|
|
|
&WasmGenerator::i64_const<8>,
|
|
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
&WasmGenerator::sequence<ValueType::kI64, ValueType::kStmt>,
|
|
|
|
|
&WasmGenerator::sequence<ValueType::kStmt, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::sequence<ValueType::kStmt, ValueType::kI64,
|
|
|
|
|
ValueType::kStmt>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::op<kExprI64Add, ValueType::kI64, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprI64Sub, ValueType::kI64, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprI64Mul, ValueType::kI64, ValueType::kI64>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::op<kExprI64DivS, ValueType::kI64, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprI64DivU, ValueType::kI64, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprI64RemS, ValueType::kI64, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprI64RemU, ValueType::kI64, ValueType::kI64>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::op<kExprI64And, ValueType::kI64, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprI64Ior, ValueType::kI64, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprI64Xor, ValueType::kI64, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprI64Shl, ValueType::kI64, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprI64ShrU, ValueType::kI64, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprI64ShrS, ValueType::kI64, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprI64Ror, ValueType::kI64, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprI64Rol, ValueType::kI64, ValueType::kI64>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::op<kExprI64Clz, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprI64Ctz, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprI64Popcnt, ValueType::kI64>,
|
|
|
|
|
|
2020-05-20 12:33:26 +00:00
|
|
|
&WasmGenerator::op_with_prefix<kExprI64SConvertSatF32, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI64UConvertSatF32, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI64SConvertSatF64, ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI64UConvertSatF64, ValueType::kF64>,
|
|
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
&WasmGenerator::block<ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::loop<ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::if_<ValueType::kI64, kIfElse>,
|
|
|
|
|
&WasmGenerator::br_if<ValueType::kI64>,
|
2017-11-06 15:48:17 +00:00
|
|
|
|
|
|
|
|
&WasmGenerator::memop<kExprI64LoadMem>,
|
|
|
|
|
&WasmGenerator::memop<kExprI64LoadMem8S>,
|
|
|
|
|
&WasmGenerator::memop<kExprI64LoadMem8U>,
|
|
|
|
|
&WasmGenerator::memop<kExprI64LoadMem16S>,
|
|
|
|
|
&WasmGenerator::memop<kExprI64LoadMem16U>,
|
|
|
|
|
&WasmGenerator::memop<kExprI64LoadMem32S>,
|
2017-12-21 14:31:42 +00:00
|
|
|
&WasmGenerator::memop<kExprI64LoadMem32U>,
|
2020-01-14 18:17:34 +00:00
|
|
|
&WasmGenerator::memop<kExprI64AtomicLoad>,
|
|
|
|
|
&WasmGenerator::memop<kExprI64AtomicLoad8U>,
|
|
|
|
|
&WasmGenerator::memop<kExprI64AtomicLoad16U>,
|
|
|
|
|
&WasmGenerator::memop<kExprI64AtomicLoad32U>,
|
2017-12-21 14:31:42 +00:00
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicAdd, ValueType::kI32,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicSub, ValueType::kI32,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicAnd, ValueType::kI32,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicOr, ValueType::kI32,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicXor, ValueType::kI32,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicExchange, ValueType::kI32,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicCompareExchange, ValueType::kI32,
|
|
|
|
|
ValueType::kI64, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicAdd8U, ValueType::kI32,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicSub8U, ValueType::kI32,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicAnd8U, ValueType::kI32,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicOr8U, ValueType::kI32,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicXor8U, ValueType::kI32,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicExchange8U, ValueType::kI32,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicCompareExchange8U,
|
|
|
|
|
ValueType::kI32, ValueType::kI64,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicAdd16U, ValueType::kI32,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicSub16U, ValueType::kI32,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicAnd16U, ValueType::kI32,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicOr16U, ValueType::kI32,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicXor16U, ValueType::kI32,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicExchange16U, ValueType::kI32,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicCompareExchange16U,
|
|
|
|
|
ValueType::kI32, ValueType::kI64,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicAdd32U, ValueType::kI32,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicSub32U, ValueType::kI32,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicAnd32U, ValueType::kI32,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicOr32U, ValueType::kI32,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicXor32U, ValueType::kI32,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicExchange32U, ValueType::kI32,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::atomic_op<kExprI64AtomicCompareExchange32U,
|
|
|
|
|
ValueType::kI32, ValueType::kI64,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
|
2020-05-28 00:06:31 +00:00
|
|
|
&WasmGenerator::simd_lane_op<kExprI64x2ExtractLane, 2, ValueType::kS128>,
|
|
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
&WasmGenerator::get_local<ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::tee_local<ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::get_global<ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprSelect, ValueType::kI64, ValueType::kI64,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::select_with_type<ValueType::kI64>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::call<ValueType::kI64>};
|
2017-11-06 15:48:17 +00:00
|
|
|
|
2019-10-17 08:46:01 +00:00
|
|
|
GenerateOneOf(alternatives, data);
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
template <>
|
2020-03-12 14:29:51 +00:00
|
|
|
void WasmGenerator::Generate<ValueType::kF32>(DataRange* data) {
|
2017-11-03 16:43:29 +00:00
|
|
|
GeneratorRecursionScope rec_scope(this);
|
2019-07-08 11:44:42 +00:00
|
|
|
if (recursion_limit_reached() || data->size() <= sizeof(float)) {
|
|
|
|
|
builder_->EmitF32Const(data->get<float>());
|
2017-11-06 15:48:17 +00:00
|
|
|
return;
|
|
|
|
|
}
|
2017-11-06 15:07:14 +00:00
|
|
|
|
2019-10-17 08:46:01 +00:00
|
|
|
constexpr GenerateFn alternatives[] = {
|
2020-03-12 14:29:51 +00:00
|
|
|
&WasmGenerator::sequence<ValueType::kF32, ValueType::kStmt>,
|
|
|
|
|
&WasmGenerator::sequence<ValueType::kStmt, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::sequence<ValueType::kStmt, ValueType::kF32,
|
|
|
|
|
ValueType::kStmt>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::op<kExprF32Abs, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::op<kExprF32Neg, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::op<kExprF32Ceil, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::op<kExprF32Floor, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::op<kExprF32Trunc, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::op<kExprF32NearestInt, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::op<kExprF32Sqrt, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::op<kExprF32Add, ValueType::kF32, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::op<kExprF32Sub, ValueType::kF32, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::op<kExprF32Mul, ValueType::kF32, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::op<kExprF32Div, ValueType::kF32, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::op<kExprF32Min, ValueType::kF32, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::op<kExprF32Max, ValueType::kF32, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::op<kExprF32CopySign, ValueType::kF32, ValueType::kF32>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::op<kExprF32SConvertI32, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op<kExprF32UConvertI32, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op<kExprF32SConvertI64, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprF32UConvertI64, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprF32ConvertF64, ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::op<kExprF32ReinterpretI32, ValueType::kI32>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::block<ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::loop<ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::if_<ValueType::kF32, kIfElse>,
|
|
|
|
|
&WasmGenerator::br_if<ValueType::kF32>,
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
|
2017-12-21 14:31:42 +00:00
|
|
|
&WasmGenerator::memop<kExprF32LoadMem>,
|
|
|
|
|
|
2020-05-28 00:06:31 +00:00
|
|
|
&WasmGenerator::simd_lane_op<kExprF32x4ExtractLane, 4, ValueType::kS128>,
|
|
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
&WasmGenerator::get_local<ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::tee_local<ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::get_global<ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::op<kExprSelect, ValueType::kF32, ValueType::kF32,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::select_with_type<ValueType::kF32>,
|
2018-01-02 18:52:49 +00:00
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
&WasmGenerator::call<ValueType::kF32>};
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
|
2019-10-17 08:46:01 +00:00
|
|
|
GenerateOneOf(alternatives, data);
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
template <>
|
2020-03-12 14:29:51 +00:00
|
|
|
void WasmGenerator::Generate<ValueType::kF64>(DataRange* data) {
|
2017-11-03 16:43:29 +00:00
|
|
|
GeneratorRecursionScope rec_scope(this);
|
2019-07-08 11:44:42 +00:00
|
|
|
if (recursion_limit_reached() || data->size() <= sizeof(double)) {
|
|
|
|
|
builder_->EmitF64Const(data->get<double>());
|
2017-11-06 15:48:17 +00:00
|
|
|
return;
|
|
|
|
|
}
|
2017-11-06 15:07:14 +00:00
|
|
|
|
2019-10-17 08:46:01 +00:00
|
|
|
constexpr GenerateFn alternatives[] = {
|
2020-03-12 14:29:51 +00:00
|
|
|
&WasmGenerator::sequence<ValueType::kF64, ValueType::kStmt>,
|
|
|
|
|
&WasmGenerator::sequence<ValueType::kStmt, ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::sequence<ValueType::kStmt, ValueType::kF64,
|
|
|
|
|
ValueType::kStmt>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::op<kExprF64Abs, ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::op<kExprF64Neg, ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::op<kExprF64Ceil, ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::op<kExprF64Floor, ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::op<kExprF64Trunc, ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::op<kExprF64NearestInt, ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::op<kExprF64Sqrt, ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::op<kExprF64Add, ValueType::kF64, ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::op<kExprF64Sub, ValueType::kF64, ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::op<kExprF64Mul, ValueType::kF64, ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::op<kExprF64Div, ValueType::kF64, ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::op<kExprF64Min, ValueType::kF64, ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::op<kExprF64Max, ValueType::kF64, ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::op<kExprF64CopySign, ValueType::kF64, ValueType::kF64>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::op<kExprF64SConvertI32, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op<kExprF64UConvertI32, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op<kExprF64SConvertI64, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprF64UConvertI64, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op<kExprF64ConvertF32, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::op<kExprF64ReinterpretI64, ValueType::kI64>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::block<ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::loop<ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::if_<ValueType::kF64, kIfElse>,
|
|
|
|
|
&WasmGenerator::br_if<ValueType::kF64>,
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
|
2017-12-21 14:31:42 +00:00
|
|
|
&WasmGenerator::memop<kExprF64LoadMem>,
|
|
|
|
|
|
2020-05-28 00:06:31 +00:00
|
|
|
&WasmGenerator::simd_lane_op<kExprF64x2ExtractLane, 2, ValueType::kS128>,
|
|
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
&WasmGenerator::get_local<ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::tee_local<ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::get_global<ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::op<kExprSelect, ValueType::kF64, ValueType::kF64,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::select_with_type<ValueType::kF64>,
|
2018-01-02 18:52:49 +00:00
|
|
|
|
2020-03-12 14:29:51 +00:00
|
|
|
&WasmGenerator::call<ValueType::kF64>};
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
|
2019-10-17 08:46:01 +00:00
|
|
|
GenerateOneOf(alternatives, data);
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
}
|
|
|
|
|
|
2020-02-06 18:26:29 +00:00
|
|
|
template <>
|
2020-03-12 14:29:51 +00:00
|
|
|
void WasmGenerator::Generate<ValueType::kS128>(DataRange* data) {
|
2020-02-06 18:26:29 +00:00
|
|
|
GeneratorRecursionScope rec_scope(this);
|
|
|
|
|
if (recursion_limit_reached() || data->size() <= sizeof(int32_t)) {
|
|
|
|
|
// TODO(v8:8460): v128.const is not implemented yet, and we need a way to
|
|
|
|
|
// "bottom-out", so use a splat to generate this.
|
|
|
|
|
builder_->EmitI32Const(data->get<int32_t>());
|
|
|
|
|
builder_->EmitWithPrefix(kExprI8x16Splat);
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
constexpr GenerateFn alternatives[] = {
|
2020-07-08 19:10:14 +00:00
|
|
|
&WasmGenerator::simd_const,
|
2020-06-01 22:02:19 +00:00
|
|
|
&WasmGenerator::simd_lane_op<kExprI8x16ReplaceLane, 16, ValueType::kS128,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::simd_lane_op<kExprI16x8ReplaceLane, 8, ValueType::kS128,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::simd_lane_op<kExprI32x4ReplaceLane, 4, ValueType::kS128,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::simd_lane_op<kExprI64x2ReplaceLane, 2, ValueType::kS128,
|
|
|
|
|
ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::simd_lane_op<kExprF32x4ReplaceLane, 4, ValueType::kS128,
|
|
|
|
|
ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::simd_lane_op<kExprF64x2ReplaceLane, 2, ValueType::kS128,
|
|
|
|
|
ValueType::kF64>,
|
|
|
|
|
|
2020-05-20 07:08:11 +00:00
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16Splat, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16Eq, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16Ne, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16LtS, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16LtU, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16GtS, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16GtU, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16LeS, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16LeU, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16GeS, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16GeU, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
2020-06-01 22:50:40 +00:00
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16Abs, ValueType::kS128>,
|
2020-05-20 07:08:11 +00:00
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16Neg, ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16Shl, ValueType::kS128,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16ShrS, ValueType::kS128,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16ShrU, ValueType::kS128,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16Add, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16AddSaturateS, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16AddSaturateU, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16Sub, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16SubSaturateS, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16SubSaturateU, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16MinS, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16MinU, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
2020-03-17 22:17:40 +00:00
|
|
|
// I8x16Mul is prototyped but not in the proposal, thus omitted here.
|
2020-05-20 07:08:11 +00:00
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16MaxS, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16MaxU, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
2020-05-21 00:06:23 +00:00
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16RoundingAverageU,
|
|
|
|
|
ValueType::kS128, ValueType::kS128>,
|
2020-05-20 07:08:11 +00:00
|
|
|
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8Splat, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8Eq, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8Ne, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8LtS, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8LtU, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8GtS, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8GtU, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8LeS, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8LeU, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8GeS, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8GeU, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
2020-06-01 22:50:40 +00:00
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8Abs, ValueType::kS128>,
|
2020-05-20 07:08:11 +00:00
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8Neg, ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8Shl, ValueType::kS128,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8ShrS, ValueType::kS128,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8ShrU, ValueType::kS128,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8Add, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8AddSaturateS, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8AddSaturateU, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8Sub, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8SubSaturateS, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8SubSaturateU, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8Mul, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8MinS, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8MinU, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8MaxS, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8MaxU, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
2020-05-21 00:06:23 +00:00
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8RoundingAverageU,
|
|
|
|
|
ValueType::kS128, ValueType::kS128>,
|
2020-05-20 07:08:11 +00:00
|
|
|
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4Splat, ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4Eq, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4Ne, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4LtS, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4LtU, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4GtS, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4GtU, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4LeS, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4LeU, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4GeS, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4GeU, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
2020-06-01 22:50:40 +00:00
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4Abs, ValueType::kS128>,
|
2020-05-20 07:08:11 +00:00
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4Neg, ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4Shl, ValueType::kS128,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4ShrS, ValueType::kS128,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4ShrU, ValueType::kS128,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4Add, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4Sub, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4Mul, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4MinS, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4MinU, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4MaxS, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4MaxU, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI64x2Splat, ValueType::kI64>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI64x2Neg, ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI64x2Shl, ValueType::kS128,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI64x2ShrS, ValueType::kS128,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI64x2ShrU, ValueType::kS128,
|
|
|
|
|
ValueType::kI32>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI64x2Add, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI64x2Sub, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI64x2Mul, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF32x4Splat, ValueType::kF32>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF32x4Eq, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF32x4Ne, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF32x4Lt, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF32x4Gt, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF32x4Le, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF32x4Ge, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF32x4Abs, ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF32x4Neg, ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF32x4Sqrt, ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF32x4Add, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF32x4Sub, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF32x4Mul, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF32x4Div, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF32x4Min, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF32x4Max, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF64x2Splat, ValueType::kF64>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF64x2Eq, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF64x2Ne, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF64x2Lt, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF64x2Gt, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF64x2Le, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF64x2Ge, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF64x2Abs, ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF64x2Neg, ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF64x2Sqrt, ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF64x2Add, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF64x2Sub, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF64x2Mul, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF64x2Div, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF64x2Min, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF64x2Max, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4SConvertF32x4, ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4UConvertF32x4, ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF32x4SConvertI32x4, ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprF32x4UConvertI32x4, ValueType::kS128>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16SConvertI16x8, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI8x16UConvertI16x8, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8SConvertI32x4, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8UConvertI32x4, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8SConvertI8x16Low,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8SConvertI8x16High,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8UConvertI8x16Low,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI16x8UConvertI8x16High,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4SConvertI16x8Low,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4SConvertI16x8High,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4UConvertI16x8Low,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprI32x4UConvertI16x8High,
|
|
|
|
|
ValueType::kS128>,
|
2020-05-08 21:08:31 +00:00
|
|
|
|
2020-06-01 22:37:03 +00:00
|
|
|
&WasmGenerator::op_with_prefix<kExprS128Not, ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprS128And, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprS128AndNot, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprS128Or, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprS128Xor, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
|
|
|
|
&WasmGenerator::op_with_prefix<kExprS128Select, ValueType::kS128,
|
|
|
|
|
ValueType::kS128, ValueType::kS128>,
|
|
|
|
|
|
2020-04-03 23:56:53 +00:00
|
|
|
&WasmGenerator::simd_shuffle,
|
2020-05-20 07:08:11 +00:00
|
|
|
&WasmGenerator::op_with_prefix<kExprS8x16Swizzle, ValueType::kS128,
|
|
|
|
|
ValueType::kS128>,
|
2020-04-03 23:56:53 +00:00
|
|
|
|
2020-05-11 17:12:45 +00:00
|
|
|
&WasmGenerator::memop<kExprS128LoadMem>,
|
|
|
|
|
&WasmGenerator::memop<kExprI16x8Load8x8S>,
|
|
|
|
|
&WasmGenerator::memop<kExprI16x8Load8x8U>,
|
|
|
|
|
&WasmGenerator::memop<kExprI32x4Load16x4S>,
|
|
|
|
|
&WasmGenerator::memop<kExprI32x4Load16x4U>,
|
|
|
|
|
&WasmGenerator::memop<kExprI64x2Load32x2S>,
|
|
|
|
|
&WasmGenerator::memop<kExprI64x2Load32x2U>,
|
|
|
|
|
&WasmGenerator::memop<kExprS8x16LoadSplat>,
|
|
|
|
|
&WasmGenerator::memop<kExprS16x8LoadSplat>,
|
|
|
|
|
&WasmGenerator::memop<kExprS32x4LoadSplat>,
|
|
|
|
|
&WasmGenerator::memop<kExprS64x2LoadSplat>};
|
2020-02-06 18:26:29 +00:00
|
|
|
|
|
|
|
|
GenerateOneOf(alternatives, data);
|
|
|
|
|
}
|
|
|
|
|
|
2019-07-08 11:44:42 +00:00
|
|
|
void WasmGenerator::grow_memory(DataRange* data) {
|
2020-03-12 14:29:51 +00:00
|
|
|
Generate<ValueType::kI32>(data);
|
2018-10-26 17:28:37 +00:00
|
|
|
builder_->EmitWithU8(kExprMemoryGrow, 0);
|
2017-11-07 17:38:57 +00:00
|
|
|
}
|
|
|
|
|
|
2019-07-08 11:44:42 +00:00
|
|
|
void WasmGenerator::Generate(ValueType type, DataRange* data) {
|
2020-03-12 14:29:51 +00:00
|
|
|
switch (type.kind()) {
|
|
|
|
|
case ValueType::kStmt:
|
|
|
|
|
return Generate<ValueType::kStmt>(data);
|
|
|
|
|
case ValueType::kI32:
|
|
|
|
|
return Generate<ValueType::kI32>(data);
|
|
|
|
|
case ValueType::kI64:
|
|
|
|
|
return Generate<ValueType::kI64>(data);
|
|
|
|
|
case ValueType::kF32:
|
|
|
|
|
return Generate<ValueType::kF32>(data);
|
|
|
|
|
case ValueType::kF64:
|
|
|
|
|
return Generate<ValueType::kF64>(data);
|
|
|
|
|
case ValueType::kS128:
|
|
|
|
|
return Generate<ValueType::kS128>(data);
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
default:
|
|
|
|
|
UNREACHABLE();
|
|
|
|
|
}
|
|
|
|
|
}
|
2017-12-21 14:31:42 +00:00
|
|
|
|
2020-05-18 08:31:21 +00:00
|
|
|
std::vector<ValueType> WasmGenerator::GenerateTypes(DataRange* data) {
|
|
|
|
|
std::vector<ValueType> types;
|
|
|
|
|
int num_params = int{data->get<uint8_t>()} % (kMaxParameters + 1);
|
|
|
|
|
for (int i = 0; i < num_params; ++i) {
|
|
|
|
|
types.push_back(GetValueType(data));
|
|
|
|
|
}
|
|
|
|
|
return types;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void WasmGenerator::Generate(Vector<const ValueType> types, DataRange* data) {
|
|
|
|
|
// Maybe emit a multi-value block with the expected return type. Use a
|
|
|
|
|
// non-default value to indicate block generation to avoid recursion when we
|
|
|
|
|
// reach the end of the data.
|
|
|
|
|
bool generate_block = data->get<uint8_t>() % 32 == 1;
|
|
|
|
|
if (generate_block) {
|
|
|
|
|
GeneratorRecursionScope rec_scope(this);
|
|
|
|
|
if (!recursion_limit_reached()) {
|
|
|
|
|
const auto param_types = GenerateTypes(data);
|
|
|
|
|
Generate(VectorOf(param_types), data);
|
2020-06-18 12:55:08 +00:00
|
|
|
any_block(VectorOf(param_types), types, data);
|
2020-05-18 08:31:21 +00:00
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (types.size() == 0) {
|
|
|
|
|
Generate(kWasmStmt, data);
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
if (types.size() == 1) {
|
|
|
|
|
Generate(types[0], data);
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Split the types in two halves and recursively generate each half.
|
|
|
|
|
// Each half is non empty to ensure termination.
|
|
|
|
|
size_t split_index = data->get<uint8_t>() % (types.size() - 1) + 1;
|
2020-06-18 12:55:08 +00:00
|
|
|
Vector<const ValueType> lower_half = types.SubVector(0, split_index);
|
|
|
|
|
Vector<const ValueType> upper_half =
|
2020-06-22 07:24:35 +00:00
|
|
|
types.SubVector(split_index, types.size());
|
2020-05-18 08:31:21 +00:00
|
|
|
DataRange first_range = data->split();
|
|
|
|
|
Generate(lower_half, &first_range);
|
|
|
|
|
Generate(upper_half, data);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Emit code to match an arbitrary signature.
|
2020-06-18 12:55:08 +00:00
|
|
|
void WasmGenerator::ConsumeAndGenerate(Vector<const ValueType> param_types,
|
|
|
|
|
Vector<const ValueType> return_types,
|
|
|
|
|
DataRange* data) {
|
2020-05-18 08:31:21 +00:00
|
|
|
if (param_types.size() == 0) {
|
2020-06-18 12:55:08 +00:00
|
|
|
Generate(return_types, data);
|
2020-05-18 08:31:21 +00:00
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
// Keep exactly one of the parameters on the stack with a combination of drops
|
|
|
|
|
// and selects, convert this value to the first return type, and generate the
|
|
|
|
|
// remaining types.
|
|
|
|
|
// TODO(thibaudm): Improve this strategy to potentially generate any sequence
|
|
|
|
|
// of instructions matching the given signature.
|
|
|
|
|
size_t return_index = data->get<uint8_t>() % param_types.size();
|
|
|
|
|
for (size_t i = param_types.size() - 1; i > return_index; --i) {
|
|
|
|
|
builder_->Emit(kExprDrop);
|
|
|
|
|
}
|
|
|
|
|
for (size_t i = return_index; i > 0; --i) {
|
|
|
|
|
Convert(param_types[i], param_types[i - 1]);
|
|
|
|
|
builder_->EmitI32Const(0);
|
|
|
|
|
builder_->Emit(kExprSelect);
|
|
|
|
|
}
|
|
|
|
|
if (return_types.empty()) {
|
|
|
|
|
builder_->Emit(kExprDrop);
|
|
|
|
|
} else {
|
|
|
|
|
Convert(param_types[0], return_types[0]);
|
2020-06-18 12:55:08 +00:00
|
|
|
Generate(return_types + 1, data);
|
2020-05-18 08:31:21 +00:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2019-07-08 11:44:42 +00:00
|
|
|
FunctionSig* GenerateSig(Zone* zone, DataRange* data) {
|
2017-12-21 14:31:42 +00:00
|
|
|
// Generate enough parameters to spill some to the stack.
|
2019-07-08 11:44:42 +00:00
|
|
|
int num_params = int{data->get<uint8_t>()} % (kMaxParameters + 1);
|
2020-05-12 14:39:52 +00:00
|
|
|
int num_returns = int{data->get<uint8_t>()} % (kMaxReturns + 1);
|
2017-12-21 14:31:42 +00:00
|
|
|
|
2020-05-12 14:39:52 +00:00
|
|
|
FunctionSig::Builder builder(zone, num_returns, num_params);
|
|
|
|
|
for (int i = 0; i < num_returns; ++i) builder.AddReturn(GetValueType(data));
|
2017-12-21 14:31:42 +00:00
|
|
|
for (int i = 0; i < num_params; ++i) builder.AddParam(GetValueType(data));
|
|
|
|
|
return builder.Build();
|
|
|
|
|
}
|
|
|
|
|
|
2017-09-04 10:05:10 +00:00
|
|
|
} // namespace
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
|
2017-05-08 09:22:54 +00:00
|
|
|
class WasmCompileFuzzer : public WasmExecutionFuzzer {
|
2017-09-04 10:05:10 +00:00
|
|
|
bool GenerateModule(
|
2018-07-12 09:05:43 +00:00
|
|
|
Isolate* isolate, Zone* zone, Vector<const uint8_t> data,
|
2019-07-08 11:44:42 +00:00
|
|
|
ZoneBuffer* buffer, int32_t* num_args,
|
|
|
|
|
std::unique_ptr<WasmValue[]>* interpreter_args,
|
|
|
|
|
std::unique_ptr<Handle<Object>[]>* compiler_args) override {
|
2017-05-08 09:22:54 +00:00
|
|
|
TestSignatures sigs;
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
|
2017-05-08 09:22:54 +00:00
|
|
|
WasmModuleBuilder builder(zone);
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
|
2018-07-12 09:05:43 +00:00
|
|
|
DataRange range(data);
|
2017-12-21 14:31:42 +00:00
|
|
|
std::vector<FunctionSig*> function_signatures;
|
|
|
|
|
function_signatures.push_back(sigs.i_iii());
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
|
2017-12-21 14:31:42 +00:00
|
|
|
static_assert(kMaxFunctions >= 1, "need min. 1 function");
|
|
|
|
|
int num_functions = 1 + (range.get<uint8_t>() % kMaxFunctions);
|
|
|
|
|
|
|
|
|
|
for (int i = 1; i < num_functions; ++i) {
|
2019-07-08 11:44:42 +00:00
|
|
|
function_signatures.push_back(GenerateSig(zone, &range));
|
2017-12-21 14:31:42 +00:00
|
|
|
}
|
|
|
|
|
|
2018-01-29 16:45:59 +00:00
|
|
|
int num_globals = range.get<uint8_t>() % (kMaxGlobals + 1);
|
|
|
|
|
std::vector<ValueType> globals;
|
|
|
|
|
std::vector<uint8_t> mutable_globals;
|
|
|
|
|
globals.reserve(num_globals);
|
|
|
|
|
mutable_globals.reserve(num_globals);
|
|
|
|
|
|
|
|
|
|
for (int i = 0; i < num_globals; ++i) {
|
2019-07-08 11:44:42 +00:00
|
|
|
ValueType type = GetValueType(&range);
|
2018-01-29 16:45:59 +00:00
|
|
|
// 1/8 of globals are immutable.
|
|
|
|
|
const bool mutability = (range.get<uint8_t>() % 8) != 0;
|
2019-07-11 15:02:44 +00:00
|
|
|
builder.AddGlobal(type, mutability, WasmInitExpr());
|
2018-01-29 16:45:59 +00:00
|
|
|
globals.push_back(type);
|
|
|
|
|
if (mutability) mutable_globals.push_back(static_cast<uint8_t>(i));
|
|
|
|
|
}
|
|
|
|
|
|
2017-12-21 14:31:42 +00:00
|
|
|
for (int i = 0; i < num_functions; ++i) {
|
|
|
|
|
DataRange function_range =
|
|
|
|
|
i == num_functions - 1 ? std::move(range) : range.split();
|
|
|
|
|
|
|
|
|
|
FunctionSig* sig = function_signatures[i];
|
|
|
|
|
WasmFunctionBuilder* f = builder.AddFunction(sig);
|
|
|
|
|
|
2018-01-29 16:45:59 +00:00
|
|
|
WasmGenerator gen(f, function_signatures, globals, mutable_globals,
|
2019-07-08 11:44:42 +00:00
|
|
|
&function_range);
|
2020-05-18 08:31:21 +00:00
|
|
|
Vector<const ValueType> return_types(sig->returns().begin(),
|
|
|
|
|
sig->return_count());
|
|
|
|
|
gen.Generate(return_types, &function_range);
|
2017-12-21 14:31:42 +00:00
|
|
|
|
|
|
|
|
f->Emit(kExprEnd);
|
|
|
|
|
if (i == 0) builder.AddExport(CStrVector("main"), f);
|
|
|
|
|
}
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
|
2017-08-29 08:47:47 +00:00
|
|
|
builder.SetMaxMemorySize(32);
|
2020-01-14 18:17:34 +00:00
|
|
|
// We enable shared memory to be able to test atomics.
|
|
|
|
|
builder.SetHasSharedMemory();
|
2019-07-08 11:44:42 +00:00
|
|
|
builder.WriteTo(buffer);
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
|
2019-07-08 11:44:42 +00:00
|
|
|
*num_args = 3;
|
|
|
|
|
interpreter_args->reset(
|
2017-07-14 13:49:01 +00:00
|
|
|
new WasmValue[3]{WasmValue(1), WasmValue(2), WasmValue(3)});
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
|
2019-07-08 11:44:42 +00:00
|
|
|
compiler_args->reset(new Handle<Object>[3] {
|
|
|
|
|
handle(Smi::FromInt(1), isolate), handle(Smi::FromInt(2), isolate),
|
|
|
|
|
handle(Smi::FromInt(3), isolate)
|
|
|
|
|
});
|
2017-05-08 09:22:54 +00:00
|
|
|
return true;
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
}
|
2017-05-08 09:22:54 +00:00
|
|
|
};
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
|
2017-05-08 09:22:54 +00:00
|
|
|
extern "C" int LLVMFuzzerTestOneInput(const uint8_t* data, size_t size) {
|
2017-11-08 21:03:20 +00:00
|
|
|
constexpr bool require_valid = true;
|
2020-06-09 15:54:14 +00:00
|
|
|
EXPERIMENTAL_FLAG_SCOPE(reftypes);
|
2018-11-20 14:41:36 +00:00
|
|
|
WasmCompileFuzzer().FuzzWasmModule({data, size}, require_valid);
|
|
|
|
|
return 0;
|
[wasm] Syntax- and Type-aware Fuzzer
This is the beginning of a new fuzzer that generates
correct-by-construction Wasm modules. This should allow us to better
exercise the compiler and correctness aspects of fuzzing. It is based off
of ahaas' original Wasm fuzzer.
At the moment, it can generate expressions made up of most binops, and
also nested blocks with unconditional breaks. Future CLs will add
additional constructs, such as br_if, loops, memory access, etc.
The way the fuzzer works is that it starts with an array of arbitrary
data provided by libfuzzer. It uses the data to generate an expression.
Care is taken to make use of the entire string. Basically, the
generator has a bunch of grammar-like rules for how to construct an
expression of a given type. For example, an i32 can be made by adding
two other i32s, or by wrapping an i64. The process then continues
recursively until all the data is consumed.
We generate an expression from a slice of data as follows:
* If the slice is less than or equal to the size of the type (e.g. 4
bytes for i32), then it will emit the entire slice as a constant.
* Otherwise, it will consume the first 4 bytes of the slice and use
this to select which rule to apply. Each rule then consumes the
remainder of the slice in an appropriate way. For example:
* Unary ops use the remainder of the slice to generate the argument.
* Binary ops consume another four bytes and mod this with the length
of the remaining slice to split the slice into two parts. Each of
these subslices are then used to generate one of the arguments to
the binop.
* Blocks are basically like a unary op, but a stack of block types is
maintained to facilitate branches. For blocks that end in a break,
the first four bytes of a slice are used to select the break depth
and the stack determines what type of expression to generate.
The goal is that once this generator is complete, it will provide a one
to one mapping between binary strings and valid Wasm modules.
Review-Url: https://codereview.chromium.org/2658723006
Cr-Commit-Position: refs/heads/master@{#43289}
2017-02-17 17:06:29 +00:00
|
|
|
}
|
2017-09-01 13:20:46 +00:00
|
|
|
|
|
|
|
|
} // namespace fuzzer
|
|
|
|
|
} // namespace wasm
|
|
|
|
|
} // namespace internal
|
|
|
|
|
} // namespace v8
|
