/*
* Copyright © 2009, 2010 Codethink Limited
* Copyright © 2011 Collabora Ltd.
*
* SPDX-License-Identifier: LGPL-2.1-or-later
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library; if not, see .
*
* Author: Ryan Lortie
* Stef Walter
*/
#include "config.h"
#include "gbytes.h"
#include
#include
#include
#include
#include
#include
#include
#include
#if GLIB_SIZEOF_VOID_P == 8
# define G_BYTES_MAX_INLINE (128 - sizeof(GBytesInline))
#else
# define G_BYTES_MAX_INLINE (64 - sizeof(GBytesInline))
#endif
/**
* GBytes: (copy-func g_bytes_ref) (free-func g_bytes_unref)
*
* A simple reference counted data type representing an immutable sequence of
* zero or more bytes from an unspecified origin.
*
* The purpose of a `GBytes` is to keep the memory region that it holds
* alive for as long as anyone holds a reference to the bytes. When
* the last reference count is dropped, the memory is released. Multiple
* unrelated callers can use byte data in the `GBytes` without coordinating
* their activities, resting assured that the byte data will not change or
* move while they hold a reference.
*
* A `GBytes` can come from many different origins that may have
* different procedures for freeing the memory region. Examples are
* memory from [func@GLib.malloc], from memory slices, from a
* [struct@GLib.MappedFile] or memory from other allocators.
*
* `GBytes` work well as keys in [struct@GLib.HashTable]. Use
* [method@GLib.Bytes.equal] and [method@GLib.Bytes.hash] as parameters to
* [func@GLib.HashTable.new] or [func@GLib.HashTable.new_full].
* `GBytes` can also be used as keys in a [struct@GLib.Tree] by passing the
* [method@GLib.Bytes.compare] function to [ctor@GLib.Tree.new].
*
* The data pointed to by this bytes must not be modified. For a mutable
* array of bytes see [struct@GLib.ByteArray]. Use
* [method@GLib.Bytes.unref_to_array] to create a mutable array for a `GBytes`
* sequence. To create an immutable `GBytes` from a mutable
* [struct@GLib.ByteArray], use the [func@GLib.ByteArray.free_to_bytes]
* function.
*
* Since: 2.32
**/
/* Keep in sync with glib/tests/bytes.c */
struct _GBytes
{
gconstpointer data; /* may be NULL iff (size == 0) */
gsize size; /* may be 0 */
gatomicrefcount ref_count;
GDestroyNotify free_func;
gpointer user_data;
};
typedef struct
{
GBytes bytes;
/* Despite no guarantee about alignment in GBytes, it is nice to
* provide that to ensure that any code which predates support
* for inline data continues to work without disruption. malloc()
* on glibc systems would guarantee 2*sizeof(void*) aligned
* allocations and this matches that.
*/
gsize padding;
guint8 inline_data[];
} GBytesInline;
G_STATIC_ASSERT (G_STRUCT_OFFSET (GBytesInline, inline_data) == (6 * GLIB_SIZEOF_VOID_P));
/**
* g_bytes_new:
* @data: (transfer none) (array length=size) (element-type guint8) (nullable):
* the data to be used for the bytes
* @size: the size of @data
*
* Creates a new [struct@GLib.Bytes] from @data.
*
* @data is copied. If @size is 0, @data may be `NULL`.
*
* As an optimization, [ctor@GLib.Bytes.new] may avoid an extra allocation by
* copying the data within the resulting bytes structure if sufficiently small
* (since GLib 2.84).
*
* Returns: (transfer full): a new [struct@GLib.Bytes]
* Since: 2.32
*/
GBytes *
g_bytes_new (gconstpointer data,
gsize size)
{
g_return_val_if_fail (data != NULL || size == 0, NULL);
if (size <= G_BYTES_MAX_INLINE)
{
GBytesInline *bytes;
bytes = g_malloc (sizeof *bytes + size);
bytes->bytes.data = bytes->inline_data;
bytes->bytes.size = size;
bytes->bytes.free_func = NULL;
bytes->bytes.user_data = NULL;
g_atomic_ref_count_init (&bytes->bytes.ref_count);
memcpy (bytes->inline_data, data, size);
return (GBytes *)bytes;
}
return g_bytes_new_take (g_memdup2 (data, size), size);
}
/**
* g_bytes_new_take:
* @data: (transfer full) (array length=size) (element-type guint8) (nullable):
* the data to be used for the bytes
* @size: the size of @data
*
* Creates a new [struct@GLib.Bytes] from @data.
*
* After this call, @data belongs to the `GBytes` and may no longer be
* modified by the caller. The memory of @data has to be dynamically
* allocated and will eventually be freed with [func@GLib.free].
*
* For creating `GBytes` with memory from other allocators, see
* [ctor@GLib.Bytes.new_with_free_func].
*
* @data may be `NULL` if @size is 0.
*
* Returns: (transfer full): a new [struct@GLib.Bytes]
* Since: 2.32
*/
GBytes *
g_bytes_new_take (gpointer data,
gsize size)
{
return g_bytes_new_with_free_func (data, size, g_free, data);
}
/**
* g_bytes_new_static: (skip)
* @data: (transfer full) (array length=size) (element-type guint8) (nullable):
* the data to be used for the bytes
* @size: the size of @data
*
* Creates a new [struct@GLib.Bytes] from static data.
*
* @data must be static (ie: never modified or freed). It may be `NULL` if @size
* is 0.
*
* Returns: (transfer full): a new [struct@GLib.Bytes]
* Since: 2.32
*/
GBytes *
g_bytes_new_static (gconstpointer data,
gsize size)
{
return g_bytes_new_with_free_func (data, size, NULL, NULL);
}
/**
* g_bytes_new_with_free_func: (skip)
* @data: (array length=size) (element-type guint8) (nullable):
* the data to be used for the bytes
* @size: the size of @data
* @free_func: the function to call to release the data
* @user_data: data to pass to @free_func
*
* Creates a [struct@GLib.Bytes] from @data.
*
* When the last reference is dropped, @free_func will be called with the
* @user_data argument.
*
* @data must not be modified after this call is made until @free_func has
* been called to indicate that the bytes is no longer in use.
*
* @data may be `NULL` if @size is 0.
*
* Returns: (transfer full): a new [struct@GLib.Bytes]
* Since: 2.32
*/
GBytes *
g_bytes_new_with_free_func (gconstpointer data,
gsize size,
GDestroyNotify free_func,
gpointer user_data)
{
GBytes *bytes;
g_return_val_if_fail (data != NULL || size == 0, NULL);
bytes = g_new (GBytes, 1);
bytes->data = data;
bytes->size = size;
bytes->free_func = free_func;
bytes->user_data = user_data;
g_atomic_ref_count_init (&bytes->ref_count);
return (GBytes *)bytes;
}
/**
* g_bytes_new_from_bytes:
* @bytes: a [struct@GLib.Bytes]
* @offset: offset which subsection starts at
* @length: length of subsection
*
* Creates a [struct@GLib.Bytes] which is a subsection of another `GBytes`.
*
* The @offset + @length may not be longer than the size of @bytes.
*
* A reference to @bytes will be held by the newly created `GBytes` until
* the byte data is no longer needed.
*
* Since 2.56, if @offset is 0 and @length matches the size of @bytes, then
* @bytes will be returned with the reference count incremented by 1. If @bytes
* is a slice of another `GBytes`, then the resulting `GBytes` will reference
* the same `GBytes` instead of @bytes. This allows consumers to simplify the
* usage of `GBytes` when asynchronously writing to streams.
*
* Returns: (transfer full): a new [struct@GLib.Bytes]
* Since: 2.32
*/
GBytes *
g_bytes_new_from_bytes (GBytes *bytes,
gsize offset,
gsize length)
{
gchar *base;
/* Note that length may be 0. */
g_return_val_if_fail (bytes != NULL, NULL);
g_return_val_if_fail (offset <= bytes->size, NULL);
g_return_val_if_fail (offset + length <= bytes->size, NULL);
/* Avoid an extra GBytes if all bytes were requested */
if (offset == 0 && length == bytes->size)
return g_bytes_ref (bytes);
base = (gchar *)bytes->data + offset;
/* Avoid referencing intermediate GBytes. In practice, this should
* only loop once.
*/
while (bytes->free_func == (gpointer)g_bytes_unref)
bytes = bytes->user_data;
g_return_val_if_fail (bytes != NULL, NULL);
g_return_val_if_fail (base >= (gchar *)bytes->data, NULL);
g_return_val_if_fail (base <= (gchar *)bytes->data + bytes->size, NULL);
g_return_val_if_fail (base + length <= (gchar *)bytes->data + bytes->size, NULL);
return g_bytes_new_with_free_func (base, length,
(GDestroyNotify)g_bytes_unref, g_bytes_ref (bytes));
}
/**
* g_bytes_get_data:
* @bytes: a [struct@GLib.Bytes]
* @size: (out) (optional): location to return size of byte data
*
* Get the byte data in the [struct@GLib.Bytes].
*
* This data should not be modified.
*
* This function will always return the same pointer for a given `GBytes`.
*
* `NULL` may be returned if @size is 0. This is not guaranteed, as the `GBytes`
* may represent an empty string with @data non-`NULL` and @size as 0. `NULL`
* will not be returned if @size is non-zero.
*
* Returns: (transfer none) (array length=size) (element-type guint8) (nullable):
* a pointer to the byte data
* Since: 2.32
*/
gconstpointer
g_bytes_get_data (GBytes *bytes,
gsize *size)
{
g_return_val_if_fail (bytes != NULL, NULL);
if (size)
*size = bytes->size;
return bytes->data;
}
/**
* g_bytes_get_size:
* @bytes: a [struct@GLib.Bytes]
*
* Get the size of the byte data in the [struct@GLib.Bytes].
*
* This function will always return the same value for a given `GBytes`.
*
* Returns: the size
* Since: 2.32
*/
gsize
g_bytes_get_size (GBytes *bytes)
{
g_return_val_if_fail (bytes != NULL, 0);
return bytes->size;
}
/**
* g_bytes_ref:
* @bytes: a [struct@GLib.Bytes]
*
* Increase the reference count on @bytes.
*
* Returns: the [struct@GLib.Bytes]
* Since: 2.32
*/
GBytes *
g_bytes_ref (GBytes *bytes)
{
g_return_val_if_fail (bytes != NULL, NULL);
g_atomic_ref_count_inc (&bytes->ref_count);
return bytes;
}
/**
* g_bytes_unref:
* @bytes: (nullable): a [struct@GLib.Bytes]
*
* Releases a reference on @bytes.
*
* This may result in the bytes being freed. If @bytes is `NULL`, it will
* return immediately.
*
* Since: 2.32
*/
void
g_bytes_unref (GBytes *bytes)
{
if (bytes == NULL)
return;
if (g_atomic_ref_count_dec (&bytes->ref_count))
{
if (bytes->free_func != NULL)
bytes->free_func (bytes->user_data);
g_free (bytes);
}
}
/**
* g_bytes_equal:
* @bytes1: (type GLib.Bytes): a pointer to a [struct@GLib.Bytes]
* @bytes2: (type GLib.Bytes): a pointer to a [struct@GLib.Bytes] to compare with @bytes1
*
* Compares the two [struct@GLib.Bytes] values being pointed to and returns
* `TRUE` if they are equal.
*
* This function can be passed to [func@GLib.HashTable.new] as the
* @key_equal_func parameter, when using non-`NULL` `GBytes` pointers as keys in
* a [struct@GLib.HashTable].
*
* Returns: `TRUE` if the two keys match.
* Since: 2.32
*/
gboolean
g_bytes_equal (gconstpointer bytes1,
gconstpointer bytes2)
{
const GBytes *b1 = bytes1;
const GBytes *b2 = bytes2;
g_return_val_if_fail (bytes1 != NULL, FALSE);
g_return_val_if_fail (bytes2 != NULL, FALSE);
return b1->size == b2->size &&
(b1->size == 0 || memcmp (b1->data, b2->data, b1->size) == 0);
}
/**
* g_bytes_hash:
* @bytes: (type GLib.Bytes): a pointer to a [struct@GLib.Bytes] key
*
* Creates an integer hash code for the byte data in the [struct@GLib.Bytes].
*
* This function can be passed to [func@GLib.HashTable.new] as the
* @key_hash_func parameter, when using non-`NULL` `GBytes` pointers as keys in
* a [struct@GLib.HashTable].
*
* Returns: a hash value corresponding to the key.
* Since: 2.32
*/
guint
g_bytes_hash (gconstpointer bytes)
{
const GBytes *a = bytes;
const signed char *p, *e;
guint32 h = 5381;
g_return_val_if_fail (bytes != NULL, 0);
for (p = (signed char *)a->data, e = (signed char *)a->data + a->size; p != e; p++)
h = (h << 5) + h + *p;
return h;
}
/**
* g_bytes_compare:
* @bytes1: (type GLib.Bytes): a pointer to a [struct@GLib.Bytes]
* @bytes2: (type GLib.Bytes): a pointer to a [struct@GLib.Bytes] to compare with @bytes1
*
* Compares the two [struct@GLib.Bytes] values.
*
* This function can be used to sort `GBytes` instances in lexicographical
* order.
*
* If @bytes1 and @bytes2 have different length but the shorter one is a
* prefix of the longer one then the shorter one is considered to be less than
* the longer one. Otherwise the first byte where both differ is used for
* comparison. If @bytes1 has a smaller value at that position it is
* considered less, otherwise greater than @bytes2.
*
* Returns: a negative value if @bytes1 is less than @bytes2, a positive value
* if @bytes1 is greater than @bytes2, and zero if @bytes1 is equal to @bytes2
* Since: 2.32
*/
gint
g_bytes_compare (gconstpointer bytes1,
gconstpointer bytes2)
{
const GBytes *b1 = bytes1;
const GBytes *b2 = bytes2;
gint ret;
g_return_val_if_fail (bytes1 != NULL, 0);
g_return_val_if_fail (bytes2 != NULL, 0);
ret = memcmp (b1->data, b2->data, MIN (b1->size, b2->size));
if (ret == 0 && b1->size != b2->size)
ret = b1->size < b2->size ? -1 : 1;
return ret;
}
static gpointer
try_steal_and_unref (GBytes *bytes,
GDestroyNotify free_func,
gsize *size)
{
gpointer result;
if (bytes->free_func != free_func || bytes->data == NULL ||
bytes->user_data != bytes->data)
return NULL;
/* Are we the only reference? */
if (g_atomic_ref_count_compare (&bytes->ref_count, 1))
{
*size = bytes->size;
result = (gpointer)bytes->data;
g_assert (result != NULL); /* otherwise the case of @bytes being freed can’t be distinguished */
g_free (bytes);
return result;
}
return NULL;
}
/**
* g_bytes_unref_to_data:
* @bytes: (transfer full): a [struct@GLib.Bytes]
* @size: (out): location to place the length of the returned data
*
* Unreferences the bytes, and returns a pointer the same byte data
* contents.
*
* As an optimization, the byte data is returned without copying if this was
* the last reference to @bytes and @bytes was created with
* [ctor@GLib.Bytes.new], [ctor@GLib.Bytes.new_take] or
* [func@GLib.ByteArray.free_to_bytes] and the buffer was larger than the size
* [struct@GLib.Bytes] may internalize within its allocation. In all other cases
* the data is copied.
*
* Returns: (transfer full) (array length=size) (element-type guint8) (not nullable):
* a pointer to the same byte data, which should be freed with [func@GLib.free]
* Since: 2.32
*/
gpointer
g_bytes_unref_to_data (GBytes *bytes,
gsize *size)
{
gpointer result;
g_return_val_if_fail (bytes != NULL, NULL);
g_return_val_if_fail (size != NULL, NULL);
/*
* Optimal path: if this is was the last reference, then we can return
* the data from this GBytes without copying.
*/
result = try_steal_and_unref (bytes, g_free, size);
if (result == NULL)
{
/*
* Copy: Non g_malloc (or compatible) allocator, or static memory,
* so we have to copy, and then unref.
*/
result = g_memdup2 (bytes->data, bytes->size);
*size = bytes->size;
g_bytes_unref (bytes);
}
return result;
}
/**
* g_bytes_unref_to_array:
* @bytes: (transfer full): a [struct@GLib.Bytes]
*
* Unreferences the bytes, and returns a new mutable [struct@GLib.ByteArray]
* containing the same byte data.
*
* As an optimization, the byte data is transferred to the array without copying
* if this was the last reference to @bytes and @bytes was created with
* [ctor@GLib.Bytes.new], [ctor@GLib.Bytes.new_take] or
* [func@GLib.ByteArray.free_to_bytes] and the buffer was larger than the size
* [struct@GLib.Bytes] may internalize within its allocation. In all other cases
* the data is copied.
*
* Do not use it if @bytes contains more than %G_MAXUINT
* bytes. [struct@GLib.ByteArray] stores the length of its data in `guint`,
* which may be shorter than `gsize`, that @bytes is using.
*
* Returns: (transfer full): a new mutable [struct@GLib.ByteArray] containing
* the same byte data
* Since: 2.32
*/
GByteArray *
g_bytes_unref_to_array (GBytes *bytes)
{
gpointer data;
gsize size;
g_return_val_if_fail (bytes != NULL, NULL);
data = g_bytes_unref_to_data (bytes, &size);
return g_byte_array_new_take (data, size);
}
/**
* g_bytes_get_region:
* @bytes: a [struct@GLib.Bytes]
* @element_size: a non-zero element size
* @offset: an offset to the start of the region within the @bytes
* @n_elements: the number of elements in the region
*
* Gets a pointer to a region in @bytes.
*
* The region starts at @offset many bytes from the start of the data
* and contains @n_elements many elements of @element_size size.
*
* @n_elements may be zero, but @element_size must always be non-zero.
* Ideally, @element_size is a static constant (eg: `sizeof` a struct).
*
* This function does careful bounds checking (including checking for
* arithmetic overflows) and returns a non-`NULL` pointer if the
* specified region lies entirely within the @bytes. If the region is
* in some way out of range, or if an overflow has occurred, then `NULL`
* is returned.
*
* Note: it is possible to have a valid zero-size region. In this case,
* the returned pointer will be equal to the base pointer of the data of
* @bytes, plus @offset. This will be non-`NULL` except for the case
* where @bytes itself was a zero-sized region. Since it is unlikely
* that you will be using this function to check for a zero-sized region
* in a zero-sized @bytes, `NULL` effectively always means ‘error’.
*
* Returns: (nullable): the requested region, or `NULL` in case of an error
* Since: 2.70
*/
gconstpointer
g_bytes_get_region (GBytes *bytes,
gsize element_size,
gsize offset,
gsize n_elements)
{
gsize total_size;
gsize end_offset;
g_return_val_if_fail (element_size > 0, NULL);
/* No other assertion checks here. If something is wrong then we will
* simply crash (via NULL dereference or divide-by-zero).
*/
if (!g_size_checked_mul (&total_size, element_size, n_elements))
return NULL;
if (!g_size_checked_add (&end_offset, offset, total_size))
return NULL;
/* We now have:
*
* 0 <= offset <= end_offset
*
* So we need only check that end_offset is within the range of the
* size of @bytes and we're good to go.
*/
if (end_offset > bytes->size)
return NULL;
/* We now have:
*
* 0 <= offset <= end_offset <= bytes->size
*/
return ((guchar *) bytes->data) + offset;
}