jsutil.h

/* -*- Mode: C++; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 2 -*-
 * vim: set ts=8 sts=2 et sw=2 tw=80:
 * This Source Code Form is subject to the terms of the Mozilla Public
 * License, v. 2.0. If a copy of the MPL was not distributed with this
 * file, You can obtain one at http://mozilla.org/MPL/2.0/. */

/*
 * PR assertion checker.
 */

#ifndef jsutil_h
#define jsutil_h

#include "mozilla/Assertions.h"
#include "mozilla/HashFunctions.h"
#include "mozilla/MathAlgorithms.h"
#include "mozilla/MemoryChecking.h"
#include "mozilla/PodOperations.h"

#include <limits.h>

#include "js/Initialization.h"
#include "js/Utility.h"
#include "js/Value.h"

/* Crash diagnostics by default in debug and on nightly channel. */
#if defined(DEBUG) || defined(NIGHTLY_BUILD)
#  define JS_CRASH_DIAGNOSTICS 1
#endif

/* Enable poisoning in crash-diagnostics and zeal builds. */
#if defined(JS_CRASH_DIAGNOSTICS) || defined(JS_GC_ZEAL)
#  define JS_GC_POISONING 1
#endif

#if defined(JS_DEBUG)
#  define JS_DIAGNOSTICS_ASSERT(expr) MOZ_ASSERT(expr)
#elif defined(JS_CRASH_DIAGNOSTICS)
#  define JS_DIAGNOSTICS_ASSERT(expr)         \
    do {                                      \
      if (MOZ_UNLIKELY(!(expr))) MOZ_CRASH(); \
    } while (0)
#else
#  define JS_DIAGNOSTICS_ASSERT(expr) ((void)0)
#endif

static MOZ_ALWAYS_INLINE void* js_memcpy(void* dst_, const void* src_,
                                         size_t len) {
  char* dst = (char*)dst_;
  const char* src = (const char*)src_;
  MOZ_ASSERT_IF(dst >= src, (size_t)(dst - src) >= len);
  MOZ_ASSERT_IF(src >= dst, (size_t)(src - dst) >= len);

  return memcpy(dst, src, len);
}

namespace js {

// An internal version of JS_IsInitialized() that returns whether SpiderMonkey
// is currently initialized or is in the process of being initialized.
inline bool IsInitialized() {
  using namespace JS::detail;
  return libraryInitState == InitState::Initializing ||
         libraryInitState == InitState::Running;
}

template <class T>
static inline void Reverse(T* beg, T* end) {
  while (beg != end) {
    if (--end == beg) {
      return;
    }
    T tmp = *beg;
    *beg = *end;
    *end = tmp;
    ++beg;
  }
}

template <class T, class Pred>
static inline T* RemoveIf(T* begin, T* end, Pred pred) {
  T* result = begin;
  for (T* p = begin; p != end; p++) {
    if (!pred(*p)) {
      *result++ = *p;
    }
  }
  return result;
}

template <class Container, class Pred>
static inline size_t EraseIf(Container& c, Pred pred) {
  auto newEnd = RemoveIf(c.begin(), c.end(), pred);
  size_t removed = c.end() - newEnd;
  c.shrinkBy(removed);
  return removed;
}

template <class T>
static inline T* Find(T* beg, T* end, const T& v) {
  for (T* p = beg; p != end; ++p) {
    if (*p == v) {
      return p;
    }
  }
  return end;
}

template <class Container>
static inline typename Container::ElementType* Find(
    Container& c, const typename Container::ElementType& v) {
  return Find(c.begin(), c.end(), v);
}

template <typename InputIterT, typename CallableT>
void ForEach(InputIterT begin, InputIterT end, CallableT f) {
  for (; begin != end; ++begin) {
    f(*begin);
  }
}

template <class Container1, class Container2>
static inline bool EqualContainers(const Container1& lhs,
                                   const Container2& rhs) {
  if (lhs.length() != rhs.length()) {
    return false;
  }
  for (size_t i = 0, n = lhs.length(); i < n; i++) {
    if (lhs[i] != rhs[i]) {
      return false;
    }
  }
  return true;
}

template <class Container>
static inline HashNumber AddContainerToHash(const Container& c,
                                            HashNumber hn = 0) {
  for (size_t i = 0; i < c.length(); i++) {
    hn = mozilla::AddToHash(hn, HashNumber(c[i]));
  }
  return hn;
}

template <class T>
static constexpr inline T Min(T t1, T t2) {
  return t1 < t2 ? t1 : t2;
}

template <class T>
static constexpr inline T Max(T t1, T t2) {
  return t1 > t2 ? t1 : t2;
}

template <typename T, typename U>
static constexpr U ComputeByteAlignment(T bytes, U alignment) {
  static_assert(mozilla::IsUnsigned<U>::value,
                "alignment amount must be unsigned");

  return (alignment - (bytes % alignment)) % alignment;
}

template <typename T, typename U>
static constexpr T AlignBytes(T bytes, U alignment) {
  static_assert(mozilla::IsUnsigned<U>::value,
                "alignment amount must be unsigned");

  return bytes + ComputeByteAlignment(bytes, alignment);
}

/*****************************************************************************/

/* A bit array is an array of bits represented by an array of words (size_t). */

static const size_t BitArrayElementBits = sizeof(size_t) * CHAR_BIT;

static inline unsigned NumWordsForBitArrayOfLength(size_t length) {
  return (length + (BitArrayElementBits - 1)) / BitArrayElementBits;
}

static inline unsigned BitArrayIndexToWordIndex(size_t length,
                                                size_t bitIndex) {
  unsigned wordIndex = bitIndex / BitArrayElementBits;
  MOZ_ASSERT(wordIndex < length);
  return wordIndex;
}

static inline size_t BitArrayIndexToWordMask(size_t i) {
  return size_t(1) << (i % BitArrayElementBits);
}

static inline bool IsBitArrayElementSet(const size_t* array, size_t length,
                                        size_t i) {
  return array[BitArrayIndexToWordIndex(length, i)] &
         BitArrayIndexToWordMask(i);
}

static inline bool IsAnyBitArrayElementSet(const size_t* array, size_t length) {
  unsigned numWords = NumWordsForBitArrayOfLength(length);
  for (unsigned i = 0; i < numWords; ++i) {
    if (array[i]) {
      return true;
    }
  }
  return false;
}

static inline void SetBitArrayElement(size_t* array, size_t length, size_t i) {
  array[BitArrayIndexToWordIndex(length, i)] |= BitArrayIndexToWordMask(i);
}

static inline void ClearBitArrayElement(size_t* array, size_t length,
                                        size_t i) {
  array[BitArrayIndexToWordIndex(length, i)] &= ~BitArrayIndexToWordMask(i);
}

static inline void ClearAllBitArrayElements(size_t* array, size_t length) {
  for (unsigned i = 0; i < length; ++i) {
    array[i] = 0;
  }
}

// Placement-new elements of an array. This should optimize away for types with
// trivial default initiation.
template <typename T>
static void DefaultInitializeElements(void* arrayPtr, size_t length) {
  uintptr_t elem = reinterpret_cast<uintptr_t>(arrayPtr);
  MOZ_ASSERT(elem % alignof(T) == 0);

  for (size_t i = 0; i < length; ++i) {
    new (reinterpret_cast<void*>(elem)) T;
    elem += sizeof(T);
  }
}

} /* namespace js */

namespace mozilla {

/**
 * Set the first |aNElem| T elements in |aDst| to |aSrc|.
 */
template <typename T>
static MOZ_ALWAYS_INLINE void PodSet(T* aDst, const T& aSrc, size_t aNElem) {
  for (const T* dstend = aDst + aNElem; aDst < dstend; ++aDst) {
    *aDst = aSrc;
  }
}

} /* namespace mozilla */

/*
 * Patterns used by SpiderMonkey to overwrite unused memory. If you are
 * accessing an object with one of these pattern, you probably have a dangling
 * pointer. These values should be odd, see the comment in IsThingPoisoned.
 *
 * Note: new patterns should also be added to the array in IsThingPoisoned!
 *
 * We try to keep our IRC bot, mrgiggles, up to date with these and other
 * patterns:
 * https://bitbucket.org/sfink/mrgiggles/src/default/plugins/knowledge/__init__.py
 */
const uint8_t JS_FRESH_NURSERY_PATTERN = 0x2F;
const uint8_t JS_SWEPT_NURSERY_PATTERN = 0x2B;
const uint8_t JS_ALLOCATED_NURSERY_PATTERN = 0x2D;
const uint8_t JS_FRESH_TENURED_PATTERN = 0x4F;
const uint8_t JS_MOVED_TENURED_PATTERN = 0x49;
const uint8_t JS_SWEPT_TENURED_PATTERN = 0x4B;
const uint8_t JS_ALLOCATED_TENURED_PATTERN = 0x4D;
const uint8_t JS_FREED_HEAP_PTR_PATTERN = 0x6B;
const uint8_t JS_FREED_CHUNK_PATTERN = 0x8B;
const uint8_t JS_FREED_ARENA_PATTERN = 0x9B;
const uint8_t JS_SWEPT_TI_PATTERN = 0x6F;
const uint8_t JS_FRESH_MARK_STACK_PATTERN = 0x9F;
const uint8_t JS_RESET_VALUE_PATTERN = 0xBB;
const uint8_t JS_POISONED_JSSCRIPT_DATA_PATTERN = 0xDB;
const uint8_t JS_OOB_PARSE_NODE_PATTERN = 0xFF;
const uint8_t JS_LIFO_UNDEFINED_PATTERN = 0xcd;
const uint8_t JS_LIFO_UNINITIALIZED_PATTERN = 0xce;

// Even ones
const uint8_t JS_NEW_NATIVE_ITERATOR_PATTERN = 0xCC;
const uint8_t JS_SCOPE_DATA_TRAILING_NAMES_PATTERN = 0xCC;

/*
 * Ensure JS_SWEPT_CODE_PATTERN is a byte pattern that will crash immediately
 * when executed, so either an undefined instruction or an instruction that's
 * illegal in user mode.
 */
#if defined(JS_CODEGEN_X86) || defined(JS_CODEGEN_X64) || \
    defined(JS_CODEGEN_NONE)
#  define JS_SWEPT_CODE_PATTERN 0xED  // IN instruction, crashes in user mode.
#elif defined(JS_CODEGEN_ARM) || defined(JS_CODEGEN_ARM64)
#  define JS_SWEPT_CODE_PATTERN 0xA3  // undefined instruction
#elif defined(JS_CODEGEN_MIPS32) || defined(JS_CODEGEN_MIPS64)
#  define JS_SWEPT_CODE_PATTERN 0x01  // undefined instruction
#else
#  error "JS_SWEPT_CODE_PATTERN not defined for this platform"
#endif

enum class MemCheckKind : uint8_t {
  // Marks a region as poisoned. Memory sanitizers like ASan will crash when
  // accessing it (both reads and writes).
  MakeNoAccess,

  // Marks a region as having undefined contents. In ASan builds this just
  // unpoisons the memory. MSan and Valgrind can also use this to find
  // reads of uninitialized memory.
  MakeUndefined,
};

static MOZ_ALWAYS_INLINE void SetMemCheckKind(void* ptr, size_t bytes,
                                              MemCheckKind kind) {
  switch (kind) {
    case MemCheckKind::MakeUndefined:
      MOZ_MAKE_MEM_UNDEFINED(ptr, bytes);
      return;
    case MemCheckKind::MakeNoAccess:
      MOZ_MAKE_MEM_NOACCESS(ptr, bytes);
      return;
  }
  MOZ_CRASH("Invalid kind");
}

namespace js {

static inline void PoisonImpl(void* ptr, uint8_t value, size_t num) {
  // Without a valid Value tag, a poisoned Value may look like a valid
  // floating point number. To ensure that we crash more readily when
  // observing a poisoned Value, we make the poison an invalid ObjectValue.
  // Unfortunately, this adds about 2% more overhead, so we can only enable
  // it in debug.
#if defined(DEBUG)
  uintptr_t poison;
  memset(&poison, value, sizeof(poison));
#  if defined(JS_PUNBOX64)
  poison = poison & ((uintptr_t(1) << JSVAL_TAG_SHIFT) - 1);
#  endif
  JS::Value v = js::PoisonedObjectValue(poison);

  size_t value_count = num / sizeof(v);
  size_t byte_count = num % sizeof(v);
  mozilla::PodSet(reinterpret_cast<JS::Value*>(ptr), v, value_count);
  if (byte_count) {
    uint8_t* bytes = static_cast<uint8_t*>(ptr);
    uint8_t* end = bytes + num;
    mozilla::PodSet(end - byte_count, value, byte_count);
  }
#else   // !DEBUG
  memset(ptr, value, num);
#endif  // !DEBUG
}

// Unconditionally poison a region on memory.
static inline void AlwaysPoison(void* ptr, uint8_t value, size_t num,
                                MemCheckKind kind) {
  PoisonImpl(ptr, value, num);
  SetMemCheckKind(ptr, num, kind);
}

// JSGC_DISABLE_POISONING environment variable
extern bool gDisablePoisoning;

// Poison a region of memory in debug and nightly builds (plus builds where GC
// zeal is configured). Can be disabled by setting the JSGC_DISABLE_POISONING
// environment variable.
static inline void Poison(void* ptr, uint8_t value, size_t num,
                          MemCheckKind kind) {
#if defined(JS_GC_POISONING)
  if (!js::gDisablePoisoning) {
    PoisonImpl(ptr, value, num);
  }
#endif
  SetMemCheckKind(ptr, num, kind);
}

// Poison a region of memory in debug builds. Can be disabled by setting the
// JSGC_DISABLE_POISONING environment variable.
static inline void DebugOnlyPoison(void* ptr, uint8_t value, size_t num,
                                   MemCheckKind kind) {
#if defined(DEBUG)
  Poison(ptr, value, num, kind);
#else
  SetMemCheckKind(ptr, num, kind);
#endif
}

}  // namespace js

#endif /* jsutil_h */