2017-04-15 14:20:00 +00:00
|
|
|
/* Copyright (C) 2017 the mpv developers
|
|
|
|
*
|
|
|
|
* Permission to use, copy, modify, and/or distribute this software for any
|
Replace talloc
There are multiple reasons to do this. One big reason is the license:
talloc is LGPLv3+, which forces mpv to be licensed as GPLv3+.
Another one is that our talloc copy contains modifications, which makes
it essentially incompatible with upstream talloc (in particular, our
version aborts on out of memory conditions - well, it wasn't my idea).
Updating from upstream is also a bit involved - the talloc source is
not really organized in a way to allow copying it into projects (and
this isn't an intended use-case).
Finally, talloc is kind of big and bloated. The replacement halves the
amount of code - mainly because we didn't use all talloc features. It's
even more extreme if you compare upstream talloc (~4700 lines) and the
new allocator without talloc compat (~900 lines).
The replacement provides all features we need. It also doesn't clash
with talloc. (The talloc compatibility wrapper uses macros to avoid
introducing linker-level symbols which could clash with libtalloc.)
It also tries to lower the overhead (only 4 words opposed to 10 words
in talloc for leaf nodes in release mode). Debugging features like leak
reporting can be enabled at compile time and add somewhat more overhead.
Though I'm not sure whether the overhead reduction was actually
successful: allocations with children need an "extra" header, which adds
plenty of overhead, and it turns out that almost half of all allocations
have children. Maybe the implementation could be simplified and the
extra header removed - even then, overhead would be lower than talloc's.
Currently, debugging features can be entirely deactivated by defining
NDEBUG - I'm not sure if anything defines this directly yet, though.
Unlike in talloc, the leak reporting stuff is thread-safe. (That's also
why it's far less elegant, and requires extra list pointers.)
Comes with a compatibility layer, so no changes to mpv source code
are needed. The idea is that we will pretend to be using talloc for
a while, so that we can revert to our old talloc implementation at
any time for debugging purposes.
Some inspiration was taken from Mesa's ralloc:
http://cgit.freedesktop.org/mesa/mesa/tree/src/glsl/ralloc.h
This is another talloc replacement, but lacks some features we need
(getting size of an allocation, debugging features, being able to
access children in the dtor).
There's some information in ta/README what will happen next and how the
transition is expected to progress.
2013-10-12 23:17:45 +00:00
|
|
|
* purpose with or without fee is hereby granted, provided that the above
|
|
|
|
* copyright notice and this permission notice appear in all copies.
|
|
|
|
*
|
|
|
|
* THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
|
|
|
|
* WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
|
|
|
|
* MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
|
|
|
|
* ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
|
|
|
|
* WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
|
|
|
|
* ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
|
|
|
|
* OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <stdlib.h>
|
|
|
|
#include <string.h>
|
|
|
|
#include <stdio.h>
|
|
|
|
#include <assert.h>
|
2016-02-07 12:27:46 +00:00
|
|
|
#include "osdep/strnlen.h"
|
Replace talloc
There are multiple reasons to do this. One big reason is the license:
talloc is LGPLv3+, which forces mpv to be licensed as GPLv3+.
Another one is that our talloc copy contains modifications, which makes
it essentially incompatible with upstream talloc (in particular, our
version aborts on out of memory conditions - well, it wasn't my idea).
Updating from upstream is also a bit involved - the talloc source is
not really organized in a way to allow copying it into projects (and
this isn't an intended use-case).
Finally, talloc is kind of big and bloated. The replacement halves the
amount of code - mainly because we didn't use all talloc features. It's
even more extreme if you compare upstream talloc (~4700 lines) and the
new allocator without talloc compat (~900 lines).
The replacement provides all features we need. It also doesn't clash
with talloc. (The talloc compatibility wrapper uses macros to avoid
introducing linker-level symbols which could clash with libtalloc.)
It also tries to lower the overhead (only 4 words opposed to 10 words
in talloc for leaf nodes in release mode). Debugging features like leak
reporting can be enabled at compile time and add somewhat more overhead.
Though I'm not sure whether the overhead reduction was actually
successful: allocations with children need an "extra" header, which adds
plenty of overhead, and it turns out that almost half of all allocations
have children. Maybe the implementation could be simplified and the
extra header removed - even then, overhead would be lower than talloc's.
Currently, debugging features can be entirely deactivated by defining
NDEBUG - I'm not sure if anything defines this directly yet, though.
Unlike in talloc, the leak reporting stuff is thread-safe. (That's also
why it's far less elegant, and requires extra list pointers.)
Comes with a compatibility layer, so no changes to mpv source code
are needed. The idea is that we will pretend to be using talloc for
a while, so that we can revert to our old talloc implementation at
any time for debugging purposes.
Some inspiration was taken from Mesa's ralloc:
http://cgit.freedesktop.org/mesa/mesa/tree/src/glsl/ralloc.h
This is another talloc replacement, but lacks some features we need
(getting size of an allocation, debugging features, being able to
access children in the dtor).
There's some information in ta/README what will happen next and how the
transition is expected to progress.
2013-10-12 23:17:45 +00:00
|
|
|
|
|
|
|
#define TA_NO_WRAPPERS
|
|
|
|
#include "ta.h"
|
|
|
|
|
|
|
|
// Return element_size * count. If it overflows, return (size_t)-1 (SIZE_MAX).
|
|
|
|
// I.e. this returns the equivalent of: MIN(element_size * count, SIZE_MAX).
|
|
|
|
// The idea is that every real memory allocator will reject (size_t)-1, thus
|
|
|
|
// this is a valid way to handle too large array allocation requests.
|
|
|
|
size_t ta_calc_array_size(size_t element_size, size_t count)
|
|
|
|
{
|
|
|
|
if (count > (((size_t)-1) / element_size))
|
|
|
|
return (size_t)-1;
|
|
|
|
return element_size * count;
|
|
|
|
}
|
|
|
|
|
2014-01-02 22:54:59 +00:00
|
|
|
// This is used when an array has to be enlarged for appending new elements.
|
|
|
|
// Return a "good" size for the new array (in number of elements). This returns
|
2014-02-23 15:51:00 +00:00
|
|
|
// a value > nextidx, unless the calculation overflows, in which case SIZE_MAX
|
2014-01-02 22:54:59 +00:00
|
|
|
// is returned.
|
|
|
|
size_t ta_calc_prealloc_elems(size_t nextidx)
|
|
|
|
{
|
|
|
|
if (nextidx >= ((size_t)-1) / 2 - 1)
|
|
|
|
return (size_t)-1;
|
|
|
|
return (nextidx + 1) * 2;
|
|
|
|
}
|
|
|
|
|
2020-02-23 16:45:05 +00:00
|
|
|
/* Create an empty (size 0) TA allocation.
|
Replace talloc
There are multiple reasons to do this. One big reason is the license:
talloc is LGPLv3+, which forces mpv to be licensed as GPLv3+.
Another one is that our talloc copy contains modifications, which makes
it essentially incompatible with upstream talloc (in particular, our
version aborts on out of memory conditions - well, it wasn't my idea).
Updating from upstream is also a bit involved - the talloc source is
not really organized in a way to allow copying it into projects (and
this isn't an intended use-case).
Finally, talloc is kind of big and bloated. The replacement halves the
amount of code - mainly because we didn't use all talloc features. It's
even more extreme if you compare upstream talloc (~4700 lines) and the
new allocator without talloc compat (~900 lines).
The replacement provides all features we need. It also doesn't clash
with talloc. (The talloc compatibility wrapper uses macros to avoid
introducing linker-level symbols which could clash with libtalloc.)
It also tries to lower the overhead (only 4 words opposed to 10 words
in talloc for leaf nodes in release mode). Debugging features like leak
reporting can be enabled at compile time and add somewhat more overhead.
Though I'm not sure whether the overhead reduction was actually
successful: allocations with children need an "extra" header, which adds
plenty of overhead, and it turns out that almost half of all allocations
have children. Maybe the implementation could be simplified and the
extra header removed - even then, overhead would be lower than talloc's.
Currently, debugging features can be entirely deactivated by defining
NDEBUG - I'm not sure if anything defines this directly yet, though.
Unlike in talloc, the leak reporting stuff is thread-safe. (That's also
why it's far less elegant, and requires extra list pointers.)
Comes with a compatibility layer, so no changes to mpv source code
are needed. The idea is that we will pretend to be using talloc for
a while, so that we can revert to our old talloc implementation at
any time for debugging purposes.
Some inspiration was taken from Mesa's ralloc:
http://cgit.freedesktop.org/mesa/mesa/tree/src/glsl/ralloc.h
This is another talloc replacement, but lacks some features we need
(getting size of an allocation, debugging features, being able to
access children in the dtor).
There's some information in ta/README what will happen next and how the
transition is expected to progress.
2013-10-12 23:17:45 +00:00
|
|
|
*/
|
|
|
|
void *ta_new_context(void *ta_parent)
|
|
|
|
{
|
2020-02-23 16:45:05 +00:00
|
|
|
return ta_alloc_size(ta_parent, 0);
|
Replace talloc
There are multiple reasons to do this. One big reason is the license:
talloc is LGPLv3+, which forces mpv to be licensed as GPLv3+.
Another one is that our talloc copy contains modifications, which makes
it essentially incompatible with upstream talloc (in particular, our
version aborts on out of memory conditions - well, it wasn't my idea).
Updating from upstream is also a bit involved - the talloc source is
not really organized in a way to allow copying it into projects (and
this isn't an intended use-case).
Finally, talloc is kind of big and bloated. The replacement halves the
amount of code - mainly because we didn't use all talloc features. It's
even more extreme if you compare upstream talloc (~4700 lines) and the
new allocator without talloc compat (~900 lines).
The replacement provides all features we need. It also doesn't clash
with talloc. (The talloc compatibility wrapper uses macros to avoid
introducing linker-level symbols which could clash with libtalloc.)
It also tries to lower the overhead (only 4 words opposed to 10 words
in talloc for leaf nodes in release mode). Debugging features like leak
reporting can be enabled at compile time and add somewhat more overhead.
Though I'm not sure whether the overhead reduction was actually
successful: allocations with children need an "extra" header, which adds
plenty of overhead, and it turns out that almost half of all allocations
have children. Maybe the implementation could be simplified and the
extra header removed - even then, overhead would be lower than talloc's.
Currently, debugging features can be entirely deactivated by defining
NDEBUG - I'm not sure if anything defines this directly yet, though.
Unlike in talloc, the leak reporting stuff is thread-safe. (That's also
why it's far less elegant, and requires extra list pointers.)
Comes with a compatibility layer, so no changes to mpv source code
are needed. The idea is that we will pretend to be using talloc for
a while, so that we can revert to our old talloc implementation at
any time for debugging purposes.
Some inspiration was taken from Mesa's ralloc:
http://cgit.freedesktop.org/mesa/mesa/tree/src/glsl/ralloc.h
This is another talloc replacement, but lacks some features we need
(getting size of an allocation, debugging features, being able to
access children in the dtor).
There's some information in ta/README what will happen next and how the
transition is expected to progress.
2013-10-12 23:17:45 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/* Set parent of ptr to ta_parent, return the ptr.
|
|
|
|
* Note that ta_parent==NULL will simply unset the current parent of ptr.
|
|
|
|
*/
|
|
|
|
void *ta_steal_(void *ta_parent, void *ptr)
|
|
|
|
{
|
2020-02-23 16:45:05 +00:00
|
|
|
ta_set_parent(ptr, ta_parent);
|
Replace talloc
There are multiple reasons to do this. One big reason is the license:
talloc is LGPLv3+, which forces mpv to be licensed as GPLv3+.
Another one is that our talloc copy contains modifications, which makes
it essentially incompatible with upstream talloc (in particular, our
version aborts on out of memory conditions - well, it wasn't my idea).
Updating from upstream is also a bit involved - the talloc source is
not really organized in a way to allow copying it into projects (and
this isn't an intended use-case).
Finally, talloc is kind of big and bloated. The replacement halves the
amount of code - mainly because we didn't use all talloc features. It's
even more extreme if you compare upstream talloc (~4700 lines) and the
new allocator without talloc compat (~900 lines).
The replacement provides all features we need. It also doesn't clash
with talloc. (The talloc compatibility wrapper uses macros to avoid
introducing linker-level symbols which could clash with libtalloc.)
It also tries to lower the overhead (only 4 words opposed to 10 words
in talloc for leaf nodes in release mode). Debugging features like leak
reporting can be enabled at compile time and add somewhat more overhead.
Though I'm not sure whether the overhead reduction was actually
successful: allocations with children need an "extra" header, which adds
plenty of overhead, and it turns out that almost half of all allocations
have children. Maybe the implementation could be simplified and the
extra header removed - even then, overhead would be lower than talloc's.
Currently, debugging features can be entirely deactivated by defining
NDEBUG - I'm not sure if anything defines this directly yet, though.
Unlike in talloc, the leak reporting stuff is thread-safe. (That's also
why it's far less elegant, and requires extra list pointers.)
Comes with a compatibility layer, so no changes to mpv source code
are needed. The idea is that we will pretend to be using talloc for
a while, so that we can revert to our old talloc implementation at
any time for debugging purposes.
Some inspiration was taken from Mesa's ralloc:
http://cgit.freedesktop.org/mesa/mesa/tree/src/glsl/ralloc.h
This is another talloc replacement, but lacks some features we need
(getting size of an allocation, debugging features, being able to
access children in the dtor).
There's some information in ta/README what will happen next and how the
transition is expected to progress.
2013-10-12 23:17:45 +00:00
|
|
|
return ptr;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Duplicate the memory at ptr with the given size.
|
|
|
|
*/
|
|
|
|
void *ta_memdup(void *ta_parent, void *ptr, size_t size)
|
|
|
|
{
|
|
|
|
if (!ptr) {
|
|
|
|
assert(!size);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
void *res = ta_alloc_size(ta_parent, size);
|
|
|
|
if (!res)
|
|
|
|
return NULL;
|
|
|
|
memcpy(res, ptr, size);
|
|
|
|
return res;
|
|
|
|
}
|
|
|
|
|
|
|
|
// *str = *str[0..at] + append[0..append_len]
|
|
|
|
// (append_len being a maximum length; shorter if embedded \0s are encountered)
|
|
|
|
static bool strndup_append_at(char **str, size_t at, const char *append,
|
|
|
|
size_t append_len)
|
|
|
|
{
|
|
|
|
assert(ta_get_size(*str) >= at);
|
|
|
|
|
|
|
|
if (!*str && !append)
|
|
|
|
return true; // stays NULL, but not an OOM condition
|
|
|
|
|
|
|
|
size_t real_len = append ? strnlen(append, append_len) : 0;
|
|
|
|
if (append_len > real_len)
|
|
|
|
append_len = real_len;
|
|
|
|
|
|
|
|
if (ta_get_size(*str) < at + append_len + 1) {
|
|
|
|
char *t = ta_realloc_size(NULL, *str, at + append_len + 1);
|
|
|
|
if (!t)
|
|
|
|
return false;
|
|
|
|
*str = t;
|
|
|
|
}
|
|
|
|
|
2015-03-23 17:03:45 +00:00
|
|
|
if (append_len)
|
|
|
|
memcpy(*str + at, append, append_len);
|
|
|
|
|
Replace talloc
There are multiple reasons to do this. One big reason is the license:
talloc is LGPLv3+, which forces mpv to be licensed as GPLv3+.
Another one is that our talloc copy contains modifications, which makes
it essentially incompatible with upstream talloc (in particular, our
version aborts on out of memory conditions - well, it wasn't my idea).
Updating from upstream is also a bit involved - the talloc source is
not really organized in a way to allow copying it into projects (and
this isn't an intended use-case).
Finally, talloc is kind of big and bloated. The replacement halves the
amount of code - mainly because we didn't use all talloc features. It's
even more extreme if you compare upstream talloc (~4700 lines) and the
new allocator without talloc compat (~900 lines).
The replacement provides all features we need. It also doesn't clash
with talloc. (The talloc compatibility wrapper uses macros to avoid
introducing linker-level symbols which could clash with libtalloc.)
It also tries to lower the overhead (only 4 words opposed to 10 words
in talloc for leaf nodes in release mode). Debugging features like leak
reporting can be enabled at compile time and add somewhat more overhead.
Though I'm not sure whether the overhead reduction was actually
successful: allocations with children need an "extra" header, which adds
plenty of overhead, and it turns out that almost half of all allocations
have children. Maybe the implementation could be simplified and the
extra header removed - even then, overhead would be lower than talloc's.
Currently, debugging features can be entirely deactivated by defining
NDEBUG - I'm not sure if anything defines this directly yet, though.
Unlike in talloc, the leak reporting stuff is thread-safe. (That's also
why it's far less elegant, and requires extra list pointers.)
Comes with a compatibility layer, so no changes to mpv source code
are needed. The idea is that we will pretend to be using talloc for
a while, so that we can revert to our old talloc implementation at
any time for debugging purposes.
Some inspiration was taken from Mesa's ralloc:
http://cgit.freedesktop.org/mesa/mesa/tree/src/glsl/ralloc.h
This is another talloc replacement, but lacks some features we need
(getting size of an allocation, debugging features, being able to
access children in the dtor).
There's some information in ta/README what will happen next and how the
transition is expected to progress.
2013-10-12 23:17:45 +00:00
|
|
|
(*str)[at + append_len] = '\0';
|
|
|
|
|
|
|
|
ta_dbg_mark_as_string(*str);
|
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Return a copy of str.
|
|
|
|
* Returns NULL on OOM.
|
|
|
|
*/
|
|
|
|
char *ta_strdup(void *ta_parent, const char *str)
|
|
|
|
{
|
|
|
|
return ta_strndup(ta_parent, str, str ? strlen(str) : 0);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Return a copy of str. If the string is longer than n, copy only n characters
|
|
|
|
* (the returned allocation will be n+1 bytes and contain a terminating '\0').
|
|
|
|
* The returned string will have the length MIN(strlen(str), n)
|
|
|
|
* If str==NULL, return NULL. Returns NULL on OOM as well.
|
|
|
|
*/
|
|
|
|
char *ta_strndup(void *ta_parent, const char *str, size_t n)
|
|
|
|
{
|
|
|
|
if (!str)
|
|
|
|
return NULL;
|
|
|
|
char *new = NULL;
|
|
|
|
strndup_append_at(&new, 0, str, n);
|
2020-02-23 16:45:05 +00:00
|
|
|
ta_set_parent(new, ta_parent);
|
Replace talloc
There are multiple reasons to do this. One big reason is the license:
talloc is LGPLv3+, which forces mpv to be licensed as GPLv3+.
Another one is that our talloc copy contains modifications, which makes
it essentially incompatible with upstream talloc (in particular, our
version aborts on out of memory conditions - well, it wasn't my idea).
Updating from upstream is also a bit involved - the talloc source is
not really organized in a way to allow copying it into projects (and
this isn't an intended use-case).
Finally, talloc is kind of big and bloated. The replacement halves the
amount of code - mainly because we didn't use all talloc features. It's
even more extreme if you compare upstream talloc (~4700 lines) and the
new allocator without talloc compat (~900 lines).
The replacement provides all features we need. It also doesn't clash
with talloc. (The talloc compatibility wrapper uses macros to avoid
introducing linker-level symbols which could clash with libtalloc.)
It also tries to lower the overhead (only 4 words opposed to 10 words
in talloc for leaf nodes in release mode). Debugging features like leak
reporting can be enabled at compile time and add somewhat more overhead.
Though I'm not sure whether the overhead reduction was actually
successful: allocations with children need an "extra" header, which adds
plenty of overhead, and it turns out that almost half of all allocations
have children. Maybe the implementation could be simplified and the
extra header removed - even then, overhead would be lower than talloc's.
Currently, debugging features can be entirely deactivated by defining
NDEBUG - I'm not sure if anything defines this directly yet, though.
Unlike in talloc, the leak reporting stuff is thread-safe. (That's also
why it's far less elegant, and requires extra list pointers.)
Comes with a compatibility layer, so no changes to mpv source code
are needed. The idea is that we will pretend to be using talloc for
a while, so that we can revert to our old talloc implementation at
any time for debugging purposes.
Some inspiration was taken from Mesa's ralloc:
http://cgit.freedesktop.org/mesa/mesa/tree/src/glsl/ralloc.h
This is another talloc replacement, but lacks some features we need
(getting size of an allocation, debugging features, being able to
access children in the dtor).
There's some information in ta/README what will happen next and how the
transition is expected to progress.
2013-10-12 23:17:45 +00:00
|
|
|
return new;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Append a to *str. If *str is NULL, the string is newly allocated, otherwise
|
|
|
|
* ta_realloc() is used on *str as needed.
|
|
|
|
* Return success or failure (it can fail due to OOM only).
|
|
|
|
*/
|
|
|
|
bool ta_strdup_append(char **str, const char *a)
|
|
|
|
{
|
|
|
|
return strndup_append_at(str, *str ? strlen(*str) : 0, a, (size_t)-1);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Like ta_strdup_append(), but use ta_get_size(*str)-1 instead of strlen(*str).
|
|
|
|
* (See also: ta_asprintf_append_buffer())
|
|
|
|
*/
|
|
|
|
bool ta_strdup_append_buffer(char **str, const char *a)
|
|
|
|
{
|
|
|
|
size_t size = ta_get_size(*str);
|
|
|
|
if (size > 0)
|
|
|
|
size -= 1;
|
|
|
|
return strndup_append_at(str, size, a, (size_t)-1);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Like ta_strdup_append(), but limit the length of a with n.
|
|
|
|
* (See also: ta_strndup())
|
|
|
|
*/
|
|
|
|
bool ta_strndup_append(char **str, const char *a, size_t n)
|
|
|
|
{
|
|
|
|
return strndup_append_at(str, *str ? strlen(*str) : 0, a, n);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Like ta_strdup_append_buffer(), but limit the length of a with n.
|
|
|
|
* (See also: ta_strndup())
|
|
|
|
*/
|
|
|
|
bool ta_strndup_append_buffer(char **str, const char *a, size_t n)
|
|
|
|
{
|
|
|
|
size_t size = ta_get_size(*str);
|
|
|
|
if (size > 0)
|
|
|
|
size -= 1;
|
|
|
|
return strndup_append_at(str, size, a, n);
|
|
|
|
}
|
|
|
|
|
2024-06-25 03:03:12 +00:00
|
|
|
TA_PRF(3, 0)
|
Replace talloc
There are multiple reasons to do this. One big reason is the license:
talloc is LGPLv3+, which forces mpv to be licensed as GPLv3+.
Another one is that our talloc copy contains modifications, which makes
it essentially incompatible with upstream talloc (in particular, our
version aborts on out of memory conditions - well, it wasn't my idea).
Updating from upstream is also a bit involved - the talloc source is
not really organized in a way to allow copying it into projects (and
this isn't an intended use-case).
Finally, talloc is kind of big and bloated. The replacement halves the
amount of code - mainly because we didn't use all talloc features. It's
even more extreme if you compare upstream talloc (~4700 lines) and the
new allocator without talloc compat (~900 lines).
The replacement provides all features we need. It also doesn't clash
with talloc. (The talloc compatibility wrapper uses macros to avoid
introducing linker-level symbols which could clash with libtalloc.)
It also tries to lower the overhead (only 4 words opposed to 10 words
in talloc for leaf nodes in release mode). Debugging features like leak
reporting can be enabled at compile time and add somewhat more overhead.
Though I'm not sure whether the overhead reduction was actually
successful: allocations with children need an "extra" header, which adds
plenty of overhead, and it turns out that almost half of all allocations
have children. Maybe the implementation could be simplified and the
extra header removed - even then, overhead would be lower than talloc's.
Currently, debugging features can be entirely deactivated by defining
NDEBUG - I'm not sure if anything defines this directly yet, though.
Unlike in talloc, the leak reporting stuff is thread-safe. (That's also
why it's far less elegant, and requires extra list pointers.)
Comes with a compatibility layer, so no changes to mpv source code
are needed. The idea is that we will pretend to be using talloc for
a while, so that we can revert to our old talloc implementation at
any time for debugging purposes.
Some inspiration was taken from Mesa's ralloc:
http://cgit.freedesktop.org/mesa/mesa/tree/src/glsl/ralloc.h
This is another talloc replacement, but lacks some features we need
(getting size of an allocation, debugging features, being able to
access children in the dtor).
There's some information in ta/README what will happen next and how the
transition is expected to progress.
2013-10-12 23:17:45 +00:00
|
|
|
static bool ta_vasprintf_append_at(char **str, size_t at, const char *fmt,
|
|
|
|
va_list ap)
|
|
|
|
{
|
|
|
|
assert(ta_get_size(*str) >= at);
|
|
|
|
|
|
|
|
int size;
|
|
|
|
va_list copy;
|
|
|
|
va_copy(copy, ap);
|
|
|
|
char c;
|
|
|
|
size = vsnprintf(&c, 1, fmt, copy);
|
|
|
|
va_end(copy);
|
|
|
|
|
|
|
|
if (size < 0)
|
|
|
|
return false;
|
|
|
|
|
|
|
|
if (ta_get_size(*str) < at + size + 1) {
|
|
|
|
char *t = ta_realloc_size(NULL, *str, at + size + 1);
|
|
|
|
if (!t)
|
|
|
|
return false;
|
|
|
|
*str = t;
|
|
|
|
}
|
|
|
|
vsnprintf(*str + at, size + 1, fmt, ap);
|
|
|
|
|
|
|
|
ta_dbg_mark_as_string(*str);
|
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Like snprintf(); returns the formatted string as allocation (or NULL on OOM
|
|
|
|
* or snprintf() errors).
|
|
|
|
*/
|
|
|
|
char *ta_asprintf(void *ta_parent, const char *fmt, ...)
|
|
|
|
{
|
|
|
|
char *res;
|
|
|
|
va_list ap;
|
|
|
|
va_start(ap, fmt);
|
|
|
|
res = ta_vasprintf(ta_parent, fmt, ap);
|
|
|
|
va_end(ap);
|
|
|
|
return res;
|
|
|
|
}
|
|
|
|
|
|
|
|
char *ta_vasprintf(void *ta_parent, const char *fmt, va_list ap)
|
|
|
|
{
|
|
|
|
char *res = NULL;
|
|
|
|
ta_vasprintf_append_at(&res, 0, fmt, ap);
|
2020-02-23 16:45:05 +00:00
|
|
|
ta_set_parent(res, ta_parent);
|
|
|
|
if (!res) {
|
Replace talloc
There are multiple reasons to do this. One big reason is the license:
talloc is LGPLv3+, which forces mpv to be licensed as GPLv3+.
Another one is that our talloc copy contains modifications, which makes
it essentially incompatible with upstream talloc (in particular, our
version aborts on out of memory conditions - well, it wasn't my idea).
Updating from upstream is also a bit involved - the talloc source is
not really organized in a way to allow copying it into projects (and
this isn't an intended use-case).
Finally, talloc is kind of big and bloated. The replacement halves the
amount of code - mainly because we didn't use all talloc features. It's
even more extreme if you compare upstream talloc (~4700 lines) and the
new allocator without talloc compat (~900 lines).
The replacement provides all features we need. It also doesn't clash
with talloc. (The talloc compatibility wrapper uses macros to avoid
introducing linker-level symbols which could clash with libtalloc.)
It also tries to lower the overhead (only 4 words opposed to 10 words
in talloc for leaf nodes in release mode). Debugging features like leak
reporting can be enabled at compile time and add somewhat more overhead.
Though I'm not sure whether the overhead reduction was actually
successful: allocations with children need an "extra" header, which adds
plenty of overhead, and it turns out that almost half of all allocations
have children. Maybe the implementation could be simplified and the
extra header removed - even then, overhead would be lower than talloc's.
Currently, debugging features can be entirely deactivated by defining
NDEBUG - I'm not sure if anything defines this directly yet, though.
Unlike in talloc, the leak reporting stuff is thread-safe. (That's also
why it's far less elegant, and requires extra list pointers.)
Comes with a compatibility layer, so no changes to mpv source code
are needed. The idea is that we will pretend to be using talloc for
a while, so that we can revert to our old talloc implementation at
any time for debugging purposes.
Some inspiration was taken from Mesa's ralloc:
http://cgit.freedesktop.org/mesa/mesa/tree/src/glsl/ralloc.h
This is another talloc replacement, but lacks some features we need
(getting size of an allocation, debugging features, being able to
access children in the dtor).
There's some information in ta/README what will happen next and how the
transition is expected to progress.
2013-10-12 23:17:45 +00:00
|
|
|
ta_free(res);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
return res;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Append the formatted string to *str (after strlen(*str)). The allocation is
|
|
|
|
* ta_realloced if needed.
|
|
|
|
* Returns false on OOM or snprintf() errors, with *str left untouched.
|
|
|
|
*/
|
|
|
|
bool ta_asprintf_append(char **str, const char *fmt, ...)
|
|
|
|
{
|
|
|
|
bool res;
|
|
|
|
va_list ap;
|
|
|
|
va_start(ap, fmt);
|
|
|
|
res = ta_vasprintf_append(str, fmt, ap);
|
|
|
|
va_end(ap);
|
|
|
|
return res;
|
|
|
|
}
|
|
|
|
|
|
|
|
bool ta_vasprintf_append(char **str, const char *fmt, va_list ap)
|
|
|
|
{
|
2014-11-21 04:12:55 +00:00
|
|
|
return ta_vasprintf_append_at(str, *str ? strlen(*str) : 0, fmt, ap);
|
Replace talloc
There are multiple reasons to do this. One big reason is the license:
talloc is LGPLv3+, which forces mpv to be licensed as GPLv3+.
Another one is that our talloc copy contains modifications, which makes
it essentially incompatible with upstream talloc (in particular, our
version aborts on out of memory conditions - well, it wasn't my idea).
Updating from upstream is also a bit involved - the talloc source is
not really organized in a way to allow copying it into projects (and
this isn't an intended use-case).
Finally, talloc is kind of big and bloated. The replacement halves the
amount of code - mainly because we didn't use all talloc features. It's
even more extreme if you compare upstream talloc (~4700 lines) and the
new allocator without talloc compat (~900 lines).
The replacement provides all features we need. It also doesn't clash
with talloc. (The talloc compatibility wrapper uses macros to avoid
introducing linker-level symbols which could clash with libtalloc.)
It also tries to lower the overhead (only 4 words opposed to 10 words
in talloc for leaf nodes in release mode). Debugging features like leak
reporting can be enabled at compile time and add somewhat more overhead.
Though I'm not sure whether the overhead reduction was actually
successful: allocations with children need an "extra" header, which adds
plenty of overhead, and it turns out that almost half of all allocations
have children. Maybe the implementation could be simplified and the
extra header removed - even then, overhead would be lower than talloc's.
Currently, debugging features can be entirely deactivated by defining
NDEBUG - I'm not sure if anything defines this directly yet, though.
Unlike in talloc, the leak reporting stuff is thread-safe. (That's also
why it's far less elegant, and requires extra list pointers.)
Comes with a compatibility layer, so no changes to mpv source code
are needed. The idea is that we will pretend to be using talloc for
a while, so that we can revert to our old talloc implementation at
any time for debugging purposes.
Some inspiration was taken from Mesa's ralloc:
http://cgit.freedesktop.org/mesa/mesa/tree/src/glsl/ralloc.h
This is another talloc replacement, but lacks some features we need
(getting size of an allocation, debugging features, being able to
access children in the dtor).
There's some information in ta/README what will happen next and how the
transition is expected to progress.
2013-10-12 23:17:45 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/* Append the formatted string at the end of the allocation of *str. It
|
|
|
|
* overwrites the last byte of the allocation too (which is assumed to be the
|
2013-11-20 17:13:05 +00:00
|
|
|
* '\0' terminating the string). Compared to ta_asprintf_append(), this is
|
Replace talloc
There are multiple reasons to do this. One big reason is the license:
talloc is LGPLv3+, which forces mpv to be licensed as GPLv3+.
Another one is that our talloc copy contains modifications, which makes
it essentially incompatible with upstream talloc (in particular, our
version aborts on out of memory conditions - well, it wasn't my idea).
Updating from upstream is also a bit involved - the talloc source is
not really organized in a way to allow copying it into projects (and
this isn't an intended use-case).
Finally, talloc is kind of big and bloated. The replacement halves the
amount of code - mainly because we didn't use all talloc features. It's
even more extreme if you compare upstream talloc (~4700 lines) and the
new allocator without talloc compat (~900 lines).
The replacement provides all features we need. It also doesn't clash
with talloc. (The talloc compatibility wrapper uses macros to avoid
introducing linker-level symbols which could clash with libtalloc.)
It also tries to lower the overhead (only 4 words opposed to 10 words
in talloc for leaf nodes in release mode). Debugging features like leak
reporting can be enabled at compile time and add somewhat more overhead.
Though I'm not sure whether the overhead reduction was actually
successful: allocations with children need an "extra" header, which adds
plenty of overhead, and it turns out that almost half of all allocations
have children. Maybe the implementation could be simplified and the
extra header removed - even then, overhead would be lower than talloc's.
Currently, debugging features can be entirely deactivated by defining
NDEBUG - I'm not sure if anything defines this directly yet, though.
Unlike in talloc, the leak reporting stuff is thread-safe. (That's also
why it's far less elegant, and requires extra list pointers.)
Comes with a compatibility layer, so no changes to mpv source code
are needed. The idea is that we will pretend to be using talloc for
a while, so that we can revert to our old talloc implementation at
any time for debugging purposes.
Some inspiration was taken from Mesa's ralloc:
http://cgit.freedesktop.org/mesa/mesa/tree/src/glsl/ralloc.h
This is another talloc replacement, but lacks some features we need
(getting size of an allocation, debugging features, being able to
access children in the dtor).
There's some information in ta/README what will happen next and how the
transition is expected to progress.
2013-10-12 23:17:45 +00:00
|
|
|
* useful if you know that the string ends with the allocation, so that the
|
|
|
|
* extra strlen() can be avoided for better performance.
|
|
|
|
* Returns false on OOM or snprintf() errors, with *str left untouched.
|
|
|
|
*/
|
|
|
|
bool ta_asprintf_append_buffer(char **str, const char *fmt, ...)
|
|
|
|
{
|
|
|
|
bool res;
|
|
|
|
va_list ap;
|
|
|
|
va_start(ap, fmt);
|
|
|
|
res = ta_vasprintf_append_buffer(str, fmt, ap);
|
|
|
|
va_end(ap);
|
|
|
|
return res;
|
|
|
|
}
|
|
|
|
|
|
|
|
bool ta_vasprintf_append_buffer(char **str, const char *fmt, va_list ap)
|
|
|
|
{
|
|
|
|
size_t size = ta_get_size(*str);
|
|
|
|
if (size > 0)
|
|
|
|
size -= 1;
|
|
|
|
return ta_vasprintf_append_at(str, size, fmt, ap);
|
|
|
|
}
|
|
|
|
|
|
|
|
void *ta_xmemdup(void *ta_parent, void *ptr, size_t size)
|
|
|
|
{
|
|
|
|
void *new = ta_memdup(ta_parent, ptr, size);
|
|
|
|
ta_oom_b(new || !ptr);
|
|
|
|
return new;
|
|
|
|
}
|
|
|
|
|
|
|
|
void *ta_xrealloc_size(void *ta_parent, void *ptr, size_t size)
|
|
|
|
{
|
|
|
|
ptr = ta_realloc_size(ta_parent, ptr, size);
|
|
|
|
ta_oom_b(ptr || !size);
|
|
|
|
return ptr;
|
|
|
|
}
|
|
|
|
|
|
|
|
char *ta_xstrdup(void *ta_parent, const char *str)
|
|
|
|
{
|
|
|
|
char *res = ta_strdup(ta_parent, str);
|
|
|
|
ta_oom_b(res || !str);
|
|
|
|
return res;
|
|
|
|
}
|
|
|
|
|
|
|
|
char *ta_xstrndup(void *ta_parent, const char *str, size_t n)
|
|
|
|
{
|
|
|
|
char *res = ta_strndup(ta_parent, str, n);
|
|
|
|
ta_oom_b(res || !str);
|
|
|
|
return res;
|
|
|
|
}
|