1 | /* DistriNet malloc (dnmalloc): a more secure memory allocator.
|
---|
2 | Copyright (C) 2005, Yves Younan, Wouter Joosen, Frank Piessens
|
---|
3 | and Rainer Wichmann
|
---|
4 |
|
---|
5 | The authors can be contacted by:
|
---|
6 | Email: dnmalloc@fort-knox.org
|
---|
7 | Address:
|
---|
8 | #if defined(WITH_TPT)
|
---|
9 | Yves Younan
|
---|
10 | Celestijnenlaan 200A
|
---|
11 | B-3001 Heverlee
|
---|
12 | Belgium
|
---|
13 |
|
---|
14 | This library is free software; you can redistribute it and/or
|
---|
15 | modify it under the terms of the GNU Lesser General Public
|
---|
16 | License as published by the Free Software Foundation; either
|
---|
17 | version 2.1 of the License, or (at your option) any later version.
|
---|
18 |
|
---|
19 | This library is distributed in the hope that it will be useful,
|
---|
20 | but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
21 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
---|
22 | Lesser General Public License for more details.
|
---|
23 |
|
---|
24 | You should have received a copy of the GNU Lesser General Public
|
---|
25 | License along with this library; if not, write to the Free Software
|
---|
26 | Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
|
---|
27 |
|
---|
28 | */
|
---|
29 |
|
---|
30 | /* Current version: dnmalloc 1.0 */
|
---|
31 | /* Includes arc4random from OpenBSD, which is under the BDS license */
|
---|
32 |
|
---|
33 | /* Versions:
|
---|
34 | 0.1-0.5:
|
---|
35 | Proof of concept implementation by Hans Van den Eynden and Yves Younan
|
---|
36 | 0.6-0.7:
|
---|
37 | Bug fixes by Yves Younan
|
---|
38 | 0.8-1.0.beta4:
|
---|
39 | Reimplementation from scratch by Yves Younan
|
---|
40 | 1.0.beta4:
|
---|
41 | Public release
|
---|
42 | 1.0.beta5:
|
---|
43 | Prev_chunkinfo speeded up, was really slow because of the way we did lookups
|
---|
44 | A freechunkinfo region is now freed when it is completely empty and
|
---|
45 | not the current one
|
---|
46 |
|
---|
47 | 1.0 (Rainer Wichmann [support at la dash samhna dot org]):
|
---|
48 | ---------------------
|
---|
49 |
|
---|
50 | Compiler warnings fixed
|
---|
51 | Define REALLOC_ZERO_BYTES_FREES because it's what GNU libc does
|
---|
52 | (and what the standard says)
|
---|
53 | Removed unused code
|
---|
54 | Fix assert(aligned_OK(chunk(newp)));
|
---|
55 | -> assert(aligned_OK(chunk(oldp)));
|
---|
56 | Fix statistics in sYSMALLOc
|
---|
57 | Fix overwrite of av->top in sYSMALLOc
|
---|
58 | Provide own assert(), glibc assert() doesn't work (calls malloc)
|
---|
59 | Fix bug in mEMALIGn(), put remainder in hashtable before calling fREe
|
---|
60 | Remove cfree, independent_cmalloc, independent_comalloc (untested
|
---|
61 | public functions not covered by any standard)
|
---|
62 | Provide posix_memalign (that one is in the standard)
|
---|
63 | Move the malloc_state struct to mmapped memory protected by guard pages
|
---|
64 | Add arc4random function to initialize random canary on startup
|
---|
65 | Implement random canary at end of (re|m)alloced/memaligned buffer,
|
---|
66 | check at free/realloc
|
---|
67 | Remove code conditional on !HAVE_MMAP, since mmap is required anyway.
|
---|
68 | Use standard HAVE_foo macros (as generated by autoconf) instead of LACKS_foo
|
---|
69 |
|
---|
70 | Profiling: Reorder branches in hashtable_add, next_chunkinfo,
|
---|
71 | prev_chunkinfo, hashtable_insert, mALLOc, fREe, request2size,
|
---|
72 | checked_request2size (gcc predicts if{} branch to be taken).
|
---|
73 | Use UNLIKELY macro (gcc __builtin_expect()) where branch
|
---|
74 | reordering would make the code awkward.
|
---|
75 |
|
---|
76 | Portability: Hashtable always covers full 32bit address space to
|
---|
77 | avoid assumptions about memory layout.
|
---|
78 | Portability: Try hard to enforce mapping of mmapped memory into
|
---|
79 | 32bit address space, even on 64bit systems.
|
---|
80 | Portability: Provide a dnmalloc_pthread_init() function, since
|
---|
81 | pthread locking on HP-UX only works if initialized
|
---|
82 | after the application has entered main().
|
---|
83 | Portability: On *BSD, pthread_mutex_lock is unusable since it
|
---|
84 | calls malloc, use spinlocks instead.
|
---|
85 | Portability: Dynamically detect whether the heap is within
|
---|
86 | 32bit address range (e.g. on Linux x86_64, it isn't).
|
---|
87 | Don't use sbrk() if the heap is mapped to an address
|
---|
88 | outside the 32bit range, since this doesn't work with
|
---|
89 | the hashtable. New macro morecore32bit.
|
---|
90 |
|
---|
91 | Success on: HP-UX 11.11/pthread, Linux/pthread (32/64 bit),
|
---|
92 | FreeBSD/pthread, and Solaris 10 i386/pthread.
|
---|
93 | Fail on: OpenBSD/pthread (in _thread_machdep_save_float_state),
|
---|
94 | might be related to OpenBSD pthread internals (??).
|
---|
95 | Non-treaded version (#undef USE_MALLOC_LOCK)
|
---|
96 | works on OpenBSD.
|
---|
97 |
|
---|
98 | further to 1.0:
|
---|
99 | Valgrind client requests inserted (#define USE_VALGRIND)
|
---|
100 | Fix: malloc_consolidate (nextchunk->fd, nextchunk->bck may be NULL)
|
---|
101 | Portability: minsize = 32 bit on 64bit architecture
|
---|
102 | Minor cleanups
|
---|
103 | Fix: eliminate prototypes for memset, memcpy (they're in string.h)
|
---|
104 | Fix: one more malloc_consolidate segfault
|
---|
105 |
|
---|
106 | There may be some bugs left in this version. please use with caution.
|
---|
107 | */
|
---|
108 |
|
---|
109 |
|
---|
110 |
|
---|
111 | /* Please read the following papers for documentation:
|
---|
112 |
|
---|
113 | Yves Younan, Wouter Joosen, and Frank Piessens, A Methodology for Designing
|
---|
114 | Countermeasures against Current and Future Code Injection Attacks,
|
---|
115 | Proceedings of the Third IEEE International Information Assurance
|
---|
116 | Workshop 2005 (IWIA2005), College Park, Maryland, U.S.A., March 2005,
|
---|
117 | IEEE, IEEE Press.
|
---|
118 | http://www.fort-knox.org/younany_countermeasures.pdf
|
---|
119 |
|
---|
120 | Yves Younan, Wouter Joosen and Frank Piessens and Hans Van den
|
---|
121 | Eynden. Security of Memory Allocators for C and C++. Technical Report
|
---|
122 | CW419, Departement Computerwetenschappen, Katholieke Universiteit
|
---|
123 | Leuven, July 2005. http://www.fort-knox.org/CW419.pdf
|
---|
124 |
|
---|
125 | */
|
---|
126 |
|
---|
127 | /* Compile:
|
---|
128 | gcc -fPIC -rdynamic -c -Wall dnmalloc-portable.c
|
---|
129 | "Link":
|
---|
130 | Dynamic:
|
---|
131 | gcc -shared -Wl,-soname,libdnmalloc.so.0 -o libdnmalloc.so.0.0 dnmalloc-portable.o -lc
|
---|
132 | Static:
|
---|
133 | ar -rv libdnmalloc.a dnmalloc-portable.o
|
---|
134 |
|
---|
135 | */
|
---|
136 |
|
---|
137 | /*
|
---|
138 | dnmalloc is based on dlmalloc 2.7.2 (by Doug Lea (dl@cs.oswego.edu))
|
---|
139 | dlmalloc was released as public domain and contained the following license:
|
---|
140 |
|
---|
141 | "This is a version (aka dlmalloc) of malloc/free/realloc written by
|
---|
142 | Doug Lea and released to the public domain. Use, modify, and
|
---|
143 | redistribute this code without permission or acknowledgement in any
|
---|
144 | way you wish. Send questions, comments, complaints, performance
|
---|
145 | data, etc to dl@cs.oswego.edu
|
---|
146 |
|
---|
147 | * VERSION 2.7.2 Sat Aug 17 09:07:30 2002 Doug Lea (dl at gee)
|
---|
148 |
|
---|
149 | Note: There may be an updated version of this malloc obtainable at
|
---|
150 | ftp://gee.cs.oswego.edu/pub/misc/malloc.c
|
---|
151 | Check before installing!"
|
---|
152 |
|
---|
153 | */
|
---|
154 |
|
---|
155 | /* The following preprocessor macros are tested,
|
---|
156 | * and hence should have #define directives:
|
---|
157 | *
|
---|
158 | * HAVE_CONFIG_H Define to #include "config.h" (autoconf-generated)
|
---|
159 | *
|
---|
160 | * HAVE_UNISTD_H Define to #include <unistd.h>
|
---|
161 | *
|
---|
162 | * HAVE_SYS_UIO_H Define to #include <sys/uio.h> (for writev)
|
---|
163 | * HAVE_WRITEV Define if the 'writev' function is available
|
---|
164 | *
|
---|
165 | * HAVE_SYS_PARAM_H Define to #include <sys/param.h> (for pagesize)
|
---|
166 | *
|
---|
167 | * HAVE_MALLOC_H Define to #include <malloc.h> (for struct mallinfo)
|
---|
168 | *
|
---|
169 | * HAVE_FCNTL_H Define to #include <fcntl.h>
|
---|
170 | *
|
---|
171 | * HAVE_SYS_MMAN_H Define to #include <sys/mman.h>
|
---|
172 | * HAVE_MMAP Define if the 'mmap' function is available.
|
---|
173 | *
|
---|
174 | * HAVE_SCHED_H Define to #include <sched.h>
|
---|
175 | * HAVE_SCHED_YIELD Define id the 'sched_yield' function is available
|
---|
176 | */
|
---|
177 |
|
---|
178 |
|
---|
179 | /*
|
---|
180 | __STD_C should be nonzero if using ANSI-standard C compiler, a C++
|
---|
181 | compiler, or a C compiler sufficiently close to ANSI to get away
|
---|
182 | with it.
|
---|
183 | */
|
---|
184 |
|
---|
185 | #ifdef HAVE_CONFIG_H
|
---|
186 | #include "config.h"
|
---|
187 | #endif
|
---|
188 |
|
---|
189 | #ifdef USE_VALGRIND
|
---|
190 | #include <valgrind/memcheck.h>
|
---|
191 | #else
|
---|
192 | #define VALGRIND_FREELIKE_BLOCK(a,b) ((void)0)
|
---|
193 | #define VALGRIND_MALLOCLIKE_BLOCK(a,b,c,d) ((void)0)
|
---|
194 | #define VALGRIND_CREATE_MEMPOOL(a,b,c) ((void)0)
|
---|
195 | #define VALGRIND_MEMPOOL_ALLOC(a,b,c) ((void)0)
|
---|
196 | #define VALGRIND_MEMPOOL_FREE(a,b) ((void)0)
|
---|
197 | #define VALGRIND_DESTROY_MEMPOOL(a) ((void)0)
|
---|
198 | #define VALGRIND_MAKE_MEM_DEFINED(a,b) ((void)0)
|
---|
199 | #define VALGRIND_MAKE_MEM_UNDEFINED(a,b) ((void)0)
|
---|
200 | #define VALGRIND_MAKE_MEM_NOACCESS(a,b) ((void)0)
|
---|
201 | #endif
|
---|
202 |
|
---|
203 | #if defined (__GNUC__) && __GNUC__ > 2
|
---|
204 | # define LIKELY(expression) (__builtin_expect(!!(expression), 1))
|
---|
205 | # define UNLIKELY(expression) (__builtin_expect(!!(expression), 0))
|
---|
206 | # define __attribute_malloc__ __attribute__ ((__malloc__))
|
---|
207 | #else
|
---|
208 | # define LIKELY(x) (x)
|
---|
209 | # define UNLIKELY(x) (x)
|
---|
210 | # define __attribute_malloc__ /* Ignore */
|
---|
211 | #endif
|
---|
212 |
|
---|
213 | /*
|
---|
214 | Define HAVE_MREMAP to make realloc() use mremap() to re-allocate
|
---|
215 | large blocks. This is currently only possible on Linux with
|
---|
216 | kernel versions newer than 1.3.77.
|
---|
217 | */
|
---|
218 |
|
---|
219 | #ifndef HAVE_MREMAP
|
---|
220 | #ifdef linux
|
---|
221 | #define HAVE_MREMAP 1
|
---|
222 | #define _GNU_SOURCE
|
---|
223 | #else
|
---|
224 | #define HAVE_MREMAP 0
|
---|
225 | #endif
|
---|
226 | #endif /* HAVE_MREMAP */
|
---|
227 |
|
---|
228 |
|
---|
229 |
|
---|
230 | #ifndef __STD_C
|
---|
231 | #if defined(__STDC__) || defined(_cplusplus)
|
---|
232 | #define __STD_C 1
|
---|
233 | #else
|
---|
234 | #define __STD_C 0
|
---|
235 | #endif
|
---|
236 | #endif /*__STD_C*/
|
---|
237 |
|
---|
238 |
|
---|
239 | /*
|
---|
240 | Void_t* is the pointer type that malloc should say it returns
|
---|
241 | */
|
---|
242 |
|
---|
243 | #ifndef Void_t
|
---|
244 | #if (__STD_C || defined(WIN32))
|
---|
245 | #define Void_t void
|
---|
246 | #else
|
---|
247 | #define Void_t char
|
---|
248 | #endif
|
---|
249 | #endif /*Void_t*/
|
---|
250 |
|
---|
251 | #if __STD_C
|
---|
252 | #include <stddef.h> /* for size_t */
|
---|
253 | #else
|
---|
254 | #include <sys/types.h>
|
---|
255 | #endif
|
---|
256 |
|
---|
257 | #if !defined(USE_SYSTEM_MALLOC)
|
---|
258 |
|
---|
259 | #ifdef __cplusplus
|
---|
260 | extern "C" {
|
---|
261 | #endif
|
---|
262 |
|
---|
263 | /* define HAVE_UNISTD_H if your system has a <unistd.h>. */
|
---|
264 |
|
---|
265 | #ifdef HAVE_UNISTD_H
|
---|
266 | #include <unistd.h>
|
---|
267 | #endif
|
---|
268 |
|
---|
269 | #ifdef HAVE_SYS_UIO_H
|
---|
270 | #include <sys/uio.h>
|
---|
271 | #endif
|
---|
272 |
|
---|
273 | #include <stdio.h> /* needed for malloc_stats */
|
---|
274 | #include <errno.h> /* needed for optional MALLOC_FAILURE_ACTION */
|
---|
275 |
|
---|
276 | #include <string.h>
|
---|
277 | #include <stdlib.h>
|
---|
278 |
|
---|
279 | #include <sys/resource.h>
|
---|
280 |
|
---|
281 | extern int errno;
|
---|
282 |
|
---|
283 | /* 0: lazy,
|
---|
284 | * 1: medium (assertions compiled in),
|
---|
285 | * 2: high (guard pages at end of hash table and ciregions)
|
---|
286 | * 3: paranoid (guards at end of each allocated chunk, check at free)
|
---|
287 | */
|
---|
288 | #ifndef PARANOIA
|
---|
289 | #define PARANOIA 9
|
---|
290 | #endif
|
---|
291 |
|
---|
292 | /* Using assert() with multithreading will cause the code
|
---|
293 | * to deadlock since glibc __assert_fail will call malloc().
|
---|
294 | * We need our very own assert().
|
---|
295 | */
|
---|
296 | typedef void assert_handler_tp(const char * error, const char *file, int line);
|
---|
297 |
|
---|
298 | #if PARANOIA > 0
|
---|
299 |
|
---|
300 | #ifdef NDEBUG
|
---|
301 | #undef NDEBUG
|
---|
302 | #endif
|
---|
303 |
|
---|
304 | static void default_assert_handler(const char *error,
|
---|
305 | const char *file, int line)
|
---|
306 | {
|
---|
307 | #ifdef HAVE_WRITEV
|
---|
308 | struct iovec iov[5];
|
---|
309 | char * i1 = "assertion failed (";
|
---|
310 | char * i3 = "): ";
|
---|
311 | char * i5 = "\n";
|
---|
312 | int res = 0;
|
---|
313 |
|
---|
314 | iov[0].iov_base = i1; iov[0].iov_len = strlen(i1);
|
---|
315 | iov[1].iov_base = (char*) file; iov[1].iov_len = strlen(file);
|
---|
316 | iov[2].iov_base = i3; iov[2].iov_len = strlen(i3);
|
---|
317 | iov[3].iov_base = (char*) error; iov[3].iov_len = strlen(error);
|
---|
318 | iov[4].iov_base = i5; iov[4].iov_len = strlen(i5);
|
---|
319 | do {
|
---|
320 | res = writev(STDERR_FILENO, iov, 5);
|
---|
321 | } while (res < 0 && errno == EINTR);
|
---|
322 | #else
|
---|
323 | fputs("assertion failed (", stderr);
|
---|
324 | fputs(file, stderr);
|
---|
325 | fputs("): ", stderr);
|
---|
326 | fputs(error, stderr);
|
---|
327 | fputc('\n', stderr);
|
---|
328 | #endif
|
---|
329 | (void) line;
|
---|
330 | abort();
|
---|
331 | }
|
---|
332 | static assert_handler_tp *assert_handler = default_assert_handler;
|
---|
333 |
|
---|
334 |
|
---|
335 | #define assert(x) \
|
---|
336 | do { \
|
---|
337 | if (UNLIKELY(!(x))) { \
|
---|
338 | assert_handler(#x, __FILE__, __LINE__); \
|
---|
339 | } \
|
---|
340 | } while (0)
|
---|
341 |
|
---|
342 | #else
|
---|
343 |
|
---|
344 | static assert_handler_tp *assert_handler = NULL;
|
---|
345 | #define NDEBUG
|
---|
346 | #define assert(x) ((void)0)
|
---|
347 |
|
---|
348 | #endif
|
---|
349 |
|
---|
350 | assert_handler_tp *dnmalloc_set_handler(assert_handler_tp *new)
|
---|
351 | {
|
---|
352 | assert_handler_tp *old = assert_handler;
|
---|
353 | assert_handler = new;
|
---|
354 | return old;
|
---|
355 | }
|
---|
356 |
|
---|
357 |
|
---|
358 | #include <stdarg.h>
|
---|
359 |
|
---|
360 | /* define for debugging */
|
---|
361 | /* #define DNMALLOC_DEBUG */
|
---|
362 |
|
---|
363 | /* Do some extra checks? if not, covered by assrt()s */
|
---|
364 | /* #define DNMALLOC_CHECKS */
|
---|
365 |
|
---|
366 | /*
|
---|
367 | The unsigned integer type used for comparing any two chunk sizes.
|
---|
368 | This should be at least as wide as size_t, but should not be signed.
|
---|
369 | */
|
---|
370 |
|
---|
371 | #ifndef CHUNK_SIZE_T
|
---|
372 | #define CHUNK_SIZE_T unsigned long
|
---|
373 | #endif
|
---|
374 |
|
---|
375 | /*
|
---|
376 | The unsigned integer type used to hold addresses when they are are
|
---|
377 | manipulated as integers. Except that it is not defined on all
|
---|
378 | systems, intptr_t would suffice.
|
---|
379 | */
|
---|
380 | #ifndef PTR_UINT
|
---|
381 | #define PTR_UINT unsigned long
|
---|
382 | #endif
|
---|
383 |
|
---|
384 |
|
---|
385 | /*
|
---|
386 | INTERNAL_SIZE_T is the word-size used for internal bookkeeping
|
---|
387 | of chunk sizes.
|
---|
388 |
|
---|
389 | The default version is the same as size_t.
|
---|
390 |
|
---|
391 | While not strictly necessary, it is best to define this as an
|
---|
392 | unsigned type, even if size_t is a signed type. This may avoid some
|
---|
393 | artificial size limitations on some systems.
|
---|
394 |
|
---|
395 | On a 64-bit machine, you may be able to reduce malloc overhead by
|
---|
396 | defining INTERNAL_SIZE_T to be a 32 bit `unsigned int' at the
|
---|
397 | expense of not being able to handle more than 2^32 of malloced
|
---|
398 | space. If this limitation is acceptable, you are encouraged to set
|
---|
399 | this unless you are on a platform requiring 16byte alignments. In
|
---|
400 | this case the alignment requirements turn out to negate any
|
---|
401 | potential advantages of decreasing size_t word size.
|
---|
402 |
|
---|
403 | Implementors: Beware of the possible combinations of:
|
---|
404 | - INTERNAL_SIZE_T might be signed or unsigned, might be 32 or 64 bits,
|
---|
405 | and might be the same width as int or as long
|
---|
406 | - size_t might have different width and signedness as INTERNAL_SIZE_T
|
---|
407 | - int and long might be 32 or 64 bits, and might be the same width
|
---|
408 | To deal with this, most comparisons and difference computations
|
---|
409 | among INTERNAL_SIZE_Ts should cast them to CHUNK_SIZE_T, being
|
---|
410 | aware of the fact that casting an unsigned int to a wider long does
|
---|
411 | not sign-extend. (This also makes checking for negative numbers
|
---|
412 | awkward.) Some of these casts result in harmless compiler warnings
|
---|
413 | on some systems.
|
---|
414 | */
|
---|
415 |
|
---|
416 | #ifndef INTERNAL_SIZE_T
|
---|
417 | #define INTERNAL_SIZE_T size_t
|
---|
418 | #endif
|
---|
419 |
|
---|
420 | /* The corresponding word size */
|
---|
421 | #define SIZE_SZ (sizeof(INTERNAL_SIZE_T))
|
---|
422 |
|
---|
423 |
|
---|
424 |
|
---|
425 | /*
|
---|
426 | MALLOC_ALIGNMENT is the minimum alignment for malloc'ed chunks.
|
---|
427 | It must be a power of two at least 2 * SIZE_SZ, even on machines
|
---|
428 | for which smaller alignments would suffice. It may be defined as
|
---|
429 | larger than this though. Note however that code and data structures
|
---|
430 | are optimized for the case of 8-byte alignment.
|
---|
431 | */
|
---|
432 |
|
---|
433 |
|
---|
434 | #ifndef MALLOC_ALIGNMENT
|
---|
435 | #define MALLOC_ALIGNMENT (2 * SIZE_SZ)
|
---|
436 | #endif
|
---|
437 |
|
---|
438 | /* The corresponding bit mask value */
|
---|
439 | #define MALLOC_ALIGN_MASK (MALLOC_ALIGNMENT - 1)
|
---|
440 |
|
---|
441 |
|
---|
442 |
|
---|
443 | /*
|
---|
444 | REALLOC_ZERO_BYTES_FREES should be set if a call to
|
---|
445 | realloc with zero bytes should be the same as a call to free.
|
---|
446 | Some people think it should. Otherwise, since this malloc
|
---|
447 | returns a unique pointer for malloc(0), so does realloc(p, 0).
|
---|
448 | */
|
---|
449 |
|
---|
450 | #define REALLOC_ZERO_BYTES_FREES
|
---|
451 |
|
---|
452 | /*
|
---|
453 | TRIM_FASTBINS controls whether free() of a very small chunk can
|
---|
454 | immediately lead to trimming. Setting to true (1) can reduce memory
|
---|
455 | footprint, but will almost always slow down programs that use a lot
|
---|
456 | of small chunks.
|
---|
457 |
|
---|
458 | Define this only if you are willing to give up some speed to more
|
---|
459 | aggressively reduce system-level memory footprint when releasing
|
---|
460 | memory in programs that use many small chunks. You can get
|
---|
461 | essentially the same effect by setting MXFAST to 0, but this can
|
---|
462 | lead to even greater slowdowns in programs using many small chunks.
|
---|
463 | TRIM_FASTBINS is an in-between compile-time option, that disables
|
---|
464 | only those chunks bordering topmost memory from being placed in
|
---|
465 | fastbins.
|
---|
466 | */
|
---|
467 |
|
---|
468 | #ifndef TRIM_FASTBINS
|
---|
469 | #define TRIM_FASTBINS 0
|
---|
470 | #endif
|
---|
471 |
|
---|
472 |
|
---|
473 | /*
|
---|
474 | USE_DL_PREFIX will prefix all public routines with the string 'dl'.
|
---|
475 | This is necessary when you only want to use this malloc in one part
|
---|
476 | of a program, using your regular system malloc elsewhere.
|
---|
477 | */
|
---|
478 |
|
---|
479 | /* #define USE_DL_PREFIX */
|
---|
480 |
|
---|
481 |
|
---|
482 | /*
|
---|
483 | USE_MALLOC_LOCK causes wrapper functions to surround each
|
---|
484 | callable routine with pthread mutex lock/unlock.
|
---|
485 |
|
---|
486 | USE_MALLOC_LOCK forces USE_PUBLIC_MALLOC_WRAPPERS to be defined
|
---|
487 | */
|
---|
488 |
|
---|
489 | /* #define USE_MALLOC_LOCK */
|
---|
490 |
|
---|
491 |
|
---|
492 | /*
|
---|
493 | If USE_PUBLIC_MALLOC_WRAPPERS is defined, every public routine is
|
---|
494 | actually a wrapper function that first calls MALLOC_PREACTION, then
|
---|
495 | calls the internal routine, and follows it with
|
---|
496 | MALLOC_POSTACTION. This is needed for locking, but you can also use
|
---|
497 | this, without USE_MALLOC_LOCK, for purposes of interception,
|
---|
498 | instrumentation, etc. It is a sad fact that using wrappers often
|
---|
499 | noticeably degrades performance of malloc-intensive programs.
|
---|
500 | */
|
---|
501 |
|
---|
502 |
|
---|
503 | #ifdef USE_MALLOC_LOCK
|
---|
504 | #define USE_PUBLIC_MALLOC_WRAPPERS
|
---|
505 | #else
|
---|
506 | /* #define USE_PUBLIC_MALLOC_WRAPPERS */
|
---|
507 | #endif
|
---|
508 |
|
---|
509 |
|
---|
510 | /*
|
---|
511 | Two-phase name translation.
|
---|
512 | All of the actual routines are given mangled names.
|
---|
513 | When wrappers are used, they become the public callable versions.
|
---|
514 | When DL_PREFIX is used, the callable names are prefixed.
|
---|
515 | */
|
---|
516 |
|
---|
517 | #ifndef USE_PUBLIC_MALLOC_WRAPPERS
|
---|
518 | #define cALLOc public_cALLOc
|
---|
519 | #define fREe public_fREe
|
---|
520 | #define mALLOc public_mALLOc
|
---|
521 | #define mEMALIGn public_mEMALIGn
|
---|
522 | #define posix_mEMALIGn public_posix_mEMALIGn
|
---|
523 | #define rEALLOc public_rEALLOc
|
---|
524 | #define vALLOc public_vALLOc
|
---|
525 | #define pVALLOc public_pVALLOc
|
---|
526 | #define mALLINFo public_mALLINFo
|
---|
527 | #define mALLOPt public_mALLOPt
|
---|
528 | #define mTRIm public_mTRIm
|
---|
529 | #define mSTATs public_mSTATs
|
---|
530 | #define mUSABLe public_mUSABLe
|
---|
531 | #endif
|
---|
532 |
|
---|
533 | #ifdef USE_DL_PREFIX
|
---|
534 | #define public_cALLOc dlcalloc
|
---|
535 | #define public_fREe dlfree
|
---|
536 | #define public_mALLOc dlmalloc
|
---|
537 | #define public_mEMALIGn dlmemalign
|
---|
538 | #define public_posix_mEMALIGn dlposix_memalign
|
---|
539 | #define public_rEALLOc dlrealloc
|
---|
540 | #define public_vALLOc dlvalloc
|
---|
541 | #define public_pVALLOc dlpvalloc
|
---|
542 | #define public_mALLINFo dlmallinfo
|
---|
543 | #define public_mALLOPt dlmallopt
|
---|
544 | #define public_mTRIm dlmalloc_trim
|
---|
545 | #define public_mSTATs dlmalloc_stats
|
---|
546 | #define public_mUSABLe dlmalloc_usable_size
|
---|
547 | #else /* USE_DL_PREFIX */
|
---|
548 | #define public_cALLOc calloc
|
---|
549 | #define public_fREe free
|
---|
550 | #define public_mALLOc malloc
|
---|
551 | #define public_mEMALIGn memalign
|
---|
552 | #define public_posix_mEMALIGn posix_memalign
|
---|
553 | #define public_rEALLOc realloc
|
---|
554 | #define public_vALLOc valloc
|
---|
555 | #define public_pVALLOc pvalloc
|
---|
556 | #define public_mALLINFo mallinfo
|
---|
557 | #define public_mALLOPt mallopt
|
---|
558 | #define public_mTRIm malloc_trim
|
---|
559 | #define public_mSTATs malloc_stats
|
---|
560 | #define public_mUSABLe malloc_usable_size
|
---|
561 | #endif /* USE_DL_PREFIX */
|
---|
562 |
|
---|
563 |
|
---|
564 | /*
|
---|
565 | HAVE_MEMCPY should be defined if you are not otherwise using
|
---|
566 | ANSI STD C, but still have memcpy and memset in your C library
|
---|
567 | and want to use them in calloc and realloc. Otherwise simple
|
---|
568 | macro versions are defined below.
|
---|
569 |
|
---|
570 | USE_MEMCPY should be defined as 1 if you actually want to
|
---|
571 | have memset and memcpy called. People report that the macro
|
---|
572 | versions are faster than libc versions on some systems.
|
---|
573 |
|
---|
574 | Even if USE_MEMCPY is set to 1, loops to copy/clear small chunks
|
---|
575 | (of <= 36 bytes) are manually unrolled in realloc and calloc.
|
---|
576 | */
|
---|
577 |
|
---|
578 | #ifndef HAVE_MEMCPY
|
---|
579 | #define HAVE_MEMCPY
|
---|
580 | #endif
|
---|
581 |
|
---|
582 | #ifndef USE_MEMCPY
|
---|
583 | #ifdef HAVE_MEMCPY
|
---|
584 | #define USE_MEMCPY 1
|
---|
585 | #else
|
---|
586 | #define USE_MEMCPY 0
|
---|
587 | #endif
|
---|
588 | #endif
|
---|
589 |
|
---|
590 |
|
---|
591 | #if (__STD_C || defined(HAVE_MEMCPY))
|
---|
592 |
|
---|
593 | #ifdef WIN32
|
---|
594 | /* On Win32 memset and memcpy are already declared in windows.h */
|
---|
595 | #else
|
---|
596 | #if __STD_C
|
---|
597 | /* Defined in string.h */
|
---|
598 | #else
|
---|
599 | Void_t* memset();
|
---|
600 | Void_t* memcpy();
|
---|
601 | #endif
|
---|
602 | #endif
|
---|
603 | #endif
|
---|
604 |
|
---|
605 | /*
|
---|
606 | MALLOC_FAILURE_ACTION is the action to take before "return 0" when
|
---|
607 | malloc fails to be able to return memory, either because memory is
|
---|
608 | exhausted or because of illegal arguments.
|
---|
609 |
|
---|
610 | By default, sets errno if running on STD_C platform, else does nothing.
|
---|
611 | */
|
---|
612 |
|
---|
613 | #ifndef MALLOC_FAILURE_ACTION
|
---|
614 | #if __STD_C
|
---|
615 | #define MALLOC_FAILURE_ACTION \
|
---|
616 | errno = ENOMEM;
|
---|
617 |
|
---|
618 | #else
|
---|
619 | #define MALLOC_FAILURE_ACTION
|
---|
620 | #endif
|
---|
621 | #endif
|
---|
622 |
|
---|
623 | /*
|
---|
624 | MORECORE-related declarations. By default, rely on sbrk
|
---|
625 | */
|
---|
626 |
|
---|
627 |
|
---|
628 | #if !defined(HAVE_UNISTD_H)
|
---|
629 | #if !defined(__FreeBSD__) && !defined(__OpenBSD__) && !defined(__NetBSD__)
|
---|
630 | #if __STD_C
|
---|
631 | extern Void_t* sbrk(ptrdiff_t);
|
---|
632 | #else
|
---|
633 | extern Void_t* sbrk();
|
---|
634 | #endif
|
---|
635 | #endif
|
---|
636 | #endif
|
---|
637 |
|
---|
638 | /*
|
---|
639 | MORECORE_FAILURE is the value returned upon failure of MORECORE
|
---|
640 | as well as mmap. Since it cannot be an otherwise valid memory address,
|
---|
641 | and must reflect values of standard sys calls, you probably ought not
|
---|
642 | try to redefine it.
|
---|
643 | */
|
---|
644 |
|
---|
645 | #ifndef MORECORE_FAILURE
|
---|
646 | #define MORECORE_FAILURE ((void*)(-1UL))
|
---|
647 | #endif
|
---|
648 |
|
---|
649 | /*
|
---|
650 | MORECORE is the name of the routine to call to obtain more memory
|
---|
651 | from the system. See below for general guidance on writing
|
---|
652 | alternative MORECORE functions, as well as a version for WIN32 and a
|
---|
653 | sample version for pre-OSX macos.
|
---|
654 | */
|
---|
655 |
|
---|
656 | #ifndef MORECORE
|
---|
657 | #define MORECORE sbrk
|
---|
658 | #endif
|
---|
659 |
|
---|
660 |
|
---|
661 | /*
|
---|
662 | If MORECORE_CONTIGUOUS is true, take advantage of fact that
|
---|
663 | consecutive calls to MORECORE with positive arguments always return
|
---|
664 | contiguous increasing addresses. This is true of unix sbrk. Even
|
---|
665 | if not defined, when regions happen to be contiguous, malloc will
|
---|
666 | permit allocations spanning regions obtained from different
|
---|
667 | calls. But defining this when applicable enables some stronger
|
---|
668 | consistency checks and space efficiencies.
|
---|
669 | */
|
---|
670 |
|
---|
671 | #ifndef MORECORE_CONTIGUOUS
|
---|
672 | #define MORECORE_CONTIGUOUS 1
|
---|
673 | #endif
|
---|
674 |
|
---|
675 | /*
|
---|
676 | Define MORECORE_CANNOT_TRIM if your version of MORECORE
|
---|
677 | cannot release space back to the system when given negative
|
---|
678 | arguments. This is generally necessary only if you are using
|
---|
679 | a hand-crafted MORECORE function that cannot handle negative arguments.
|
---|
680 | */
|
---|
681 |
|
---|
682 | /* #define MORECORE_CANNOT_TRIM */
|
---|
683 |
|
---|
684 |
|
---|
685 | /*
|
---|
686 | This malloc requires mmap for heap management data. It is an error
|
---|
687 | if mmap is not available.
|
---|
688 |
|
---|
689 | Additionally, mmap will be used to satisfy large requests.
|
---|
690 | */
|
---|
691 |
|
---|
692 | #ifndef HAVE_MMAP
|
---|
693 | # error HAVE_MMAP not defined, has your operating system mmap?
|
---|
694 | #endif
|
---|
695 |
|
---|
696 | /*
|
---|
697 | Standard unix mmap using /dev/zero clears memory so calloc doesn't
|
---|
698 | need to.
|
---|
699 | */
|
---|
700 |
|
---|
701 | #ifndef MMAP_CLEARS
|
---|
702 | #define MMAP_CLEARS 1
|
---|
703 | #endif
|
---|
704 |
|
---|
705 |
|
---|
706 | /*
|
---|
707 | MMAP_AS_MORECORE_SIZE is the minimum mmap size argument to use if
|
---|
708 | sbrk fails, and mmap is used as a backup (which is done only if
|
---|
709 | HAVE_MMAP). The value must be a multiple of page size. This
|
---|
710 | backup strategy generally applies only when systems have "holes" in
|
---|
711 | address space, so sbrk cannot perform contiguous expansion, but
|
---|
712 | there is still space available on system. On systems for which
|
---|
713 | this is known to be useful (i.e. most linux kernels), this occurs
|
---|
714 | only when programs allocate huge amounts of memory. Between this,
|
---|
715 | and the fact that mmap regions tend to be limited, the size should
|
---|
716 | be large, to avoid too many mmap calls and thus avoid running out
|
---|
717 | of kernel resources.
|
---|
718 | */
|
---|
719 |
|
---|
720 | #ifndef MMAP_AS_MORECORE_SIZE
|
---|
721 | #define MMAP_AS_MORECORE_SIZE (1024 * 1024)
|
---|
722 | #endif
|
---|
723 |
|
---|
724 |
|
---|
725 | /*
|
---|
726 | The system page size. To the extent possible, this malloc manages
|
---|
727 | memory from the system in page-size units. Note that this value is
|
---|
728 | cached during initialization into a field of malloc_state. So even
|
---|
729 | if malloc_getpagesize is a function, it is only called once.
|
---|
730 |
|
---|
731 | The following mechanics for getpagesize were adapted from bsd/gnu
|
---|
732 | getpagesize.h. If none of the system-probes here apply, a value of
|
---|
733 | 4096 is used, which should be OK: If they don't apply, then using
|
---|
734 | the actual value probably doesn't impact performance.
|
---|
735 | */
|
---|
736 |
|
---|
737 |
|
---|
738 | #ifndef malloc_getpagesize
|
---|
739 |
|
---|
740 | # ifdef _SC_PAGESIZE /* some SVR4 systems omit an underscore */
|
---|
741 | # ifndef _SC_PAGE_SIZE
|
---|
742 | # define _SC_PAGE_SIZE _SC_PAGESIZE
|
---|
743 | # endif
|
---|
744 | # endif
|
---|
745 |
|
---|
746 | # ifdef _SC_PAGE_SIZE
|
---|
747 | # define malloc_getpagesize sysconf(_SC_PAGE_SIZE)
|
---|
748 | # else
|
---|
749 | # if defined(BSD) || defined(DGUX) || defined(HAVE_GETPAGESIZE)
|
---|
750 | extern size_t getpagesize();
|
---|
751 | # define malloc_getpagesize getpagesize()
|
---|
752 | # else
|
---|
753 | # ifdef WIN32 /* use supplied emulation of getpagesize */
|
---|
754 | # define malloc_getpagesize getpagesize()
|
---|
755 | # else
|
---|
756 | # if defined(HAVE_SYS_PARAM_H)
|
---|
757 | # include <sys/param.h>
|
---|
758 | # endif
|
---|
759 | # ifdef EXEC_PAGESIZE
|
---|
760 | # define malloc_getpagesize EXEC_PAGESIZE
|
---|
761 | # else
|
---|
762 | # ifdef NBPG
|
---|
763 | # ifndef CLSIZE
|
---|
764 | # define malloc_getpagesize NBPG
|
---|
765 | # else
|
---|
766 | # define malloc_getpagesize (NBPG * CLSIZE)
|
---|
767 | # endif
|
---|
768 | # else
|
---|
769 | # ifdef NBPC
|
---|
770 | # define malloc_getpagesize NBPC
|
---|
771 | # else
|
---|
772 | # ifdef PAGESIZE
|
---|
773 | # define malloc_getpagesize PAGESIZE
|
---|
774 | # else /* just guess */
|
---|
775 | # define malloc_getpagesize (4096)
|
---|
776 | # endif
|
---|
777 | # endif
|
---|
778 | # endif
|
---|
779 | # endif
|
---|
780 | # endif
|
---|
781 | # endif
|
---|
782 | # endif
|
---|
783 | #endif
|
---|
784 |
|
---|
785 | /*
|
---|
786 | This version of malloc supports the standard SVID/XPG mallinfo
|
---|
787 | routine that returns a struct containing usage properties and
|
---|
788 | statistics. It should work on any SVID/XPG compliant system that has
|
---|
789 | a /usr/include/malloc.h defining struct mallinfo. (If you'd like to
|
---|
790 | install such a thing yourself, cut out the preliminary declarations
|
---|
791 | as described above and below and save them in a malloc.h file. But
|
---|
792 | there's no compelling reason to bother to do this.)
|
---|
793 |
|
---|
794 | The main declaration needed is the mallinfo struct that is returned
|
---|
795 | (by-copy) by mallinfo(). The SVID/XPG malloinfo struct contains a
|
---|
796 | bunch of fields that are not even meaningful in this version of
|
---|
797 | malloc. These fields are are instead filled by mallinfo() with
|
---|
798 | other numbers that might be of interest.
|
---|
799 |
|
---|
800 | HAVE_MALLOC_H should be set if you have a
|
---|
801 | /usr/include/malloc.h file that includes a declaration of struct
|
---|
802 | mallinfo. If so, it is included; else an SVID2/XPG2 compliant
|
---|
803 | version is declared below. These must be precisely the same for
|
---|
804 | mallinfo() to work. The original SVID version of this struct,
|
---|
805 | defined on most systems with mallinfo, declares all fields as
|
---|
806 | ints. But some others define as unsigned long. If your system
|
---|
807 | defines the fields using a type of different width than listed here,
|
---|
808 | you must #include your system version and #define
|
---|
809 | HAVE_MALLOC_H.
|
---|
810 | */
|
---|
811 |
|
---|
812 | /* #define HAVE_MALLOC_H */
|
---|
813 |
|
---|
814 | /* On *BSD, malloc.h is deprecated, and on some *BSD including
|
---|
815 | * it may actually raise an error.
|
---|
816 | */
|
---|
817 | #if defined(HAVE_MALLOC_H) && !defined(__OpenBSD__) && !defined(__FreeBSD__) && !defined(__NetBSD__)
|
---|
818 | #include <malloc.h>
|
---|
819 | #else
|
---|
820 |
|
---|
821 | /* SVID2/XPG mallinfo structure */
|
---|
822 |
|
---|
823 | struct mallinfo {
|
---|
824 | int arena; /* non-mmapped space allocated from system */
|
---|
825 | int ordblks; /* number of free chunks */
|
---|
826 | int smblks; /* number of fastbin blocks */
|
---|
827 | int hblks; /* number of mmapped regions */
|
---|
828 | int hblkhd; /* space in mmapped regions */
|
---|
829 | int usmblks; /* maximum total allocated space */
|
---|
830 | int fsmblks; /* space available in freed fastbin blocks */
|
---|
831 | int uordblks; /* total allocated space */
|
---|
832 | int fordblks; /* total free space */
|
---|
833 | int keepcost; /* top-most, releasable (via malloc_trim) space */
|
---|
834 | };
|
---|
835 |
|
---|
836 | /*
|
---|
837 | SVID/XPG defines four standard parameter numbers for mallopt,
|
---|
838 | normally defined in malloc.h. Only one of these (M_MXFAST) is used
|
---|
839 | in this malloc. The others (M_NLBLKS, M_GRAIN, M_KEEP) don't apply,
|
---|
840 | so setting them has no effect. But this malloc also supports other
|
---|
841 | options in mallopt described below.
|
---|
842 | */
|
---|
843 | #endif
|
---|
844 |
|
---|
845 |
|
---|
846 | /* ---------- description of public routines ------------ */
|
---|
847 |
|
---|
848 | /*
|
---|
849 | malloc(size_t n)
|
---|
850 | Returns a pointer to a newly allocated chunk of at least n bytes, or null
|
---|
851 | if no space is available. Additionally, on failure, errno is
|
---|
852 | set to ENOMEM on ANSI C systems.
|
---|
853 |
|
---|
854 | If n is zero, malloc returns a minumum-sized chunk. (The minimum
|
---|
855 | size is 16 bytes on most 32bit systems, and 24 or 32 bytes on 64bit
|
---|
856 | systems.) On most systems, size_t is an unsigned type, so calls
|
---|
857 | with negative arguments are interpreted as requests for huge amounts
|
---|
858 | of space, which will often fail. The maximum supported value of n
|
---|
859 | differs across systems, but is in all cases less than the maximum
|
---|
860 | representable value of a size_t.
|
---|
861 | */
|
---|
862 | #if __STD_C
|
---|
863 | Void_t* public_mALLOc(size_t) __attribute_malloc__;
|
---|
864 | #else
|
---|
865 | Void_t* public_mALLOc();
|
---|
866 | #endif
|
---|
867 |
|
---|
868 | /*
|
---|
869 | free(Void_t* p)
|
---|
870 | Releases the chunk of memory pointed to by p, that had been previously
|
---|
871 | allocated using malloc or a related routine such as realloc.
|
---|
872 | It has no effect if p is null. It can have arbitrary (i.e., bad!)
|
---|
873 | effects if p has already been freed.
|
---|
874 |
|
---|
875 | Unless disabled (using mallopt), freeing very large spaces will
|
---|
876 | when possible, automatically trigger operations that give
|
---|
877 | back unused memory to the system, thus reducing program footprint.
|
---|
878 | */
|
---|
879 | #if __STD_C
|
---|
880 | void public_fREe(Void_t*);
|
---|
881 | #else
|
---|
882 | void public_fREe();
|
---|
883 | #endif
|
---|
884 |
|
---|
885 | /*
|
---|
886 | calloc(size_t n_elements, size_t element_size);
|
---|
887 | Returns a pointer to n_elements * element_size bytes, with all locations
|
---|
888 | set to zero.
|
---|
889 | */
|
---|
890 | #if __STD_C
|
---|
891 | Void_t* public_cALLOc(size_t, size_t) __attribute_malloc__;
|
---|
892 | #else
|
---|
893 | Void_t* public_cALLOc();
|
---|
894 | #endif
|
---|
895 |
|
---|
896 | /*
|
---|
897 | realloc(Void_t* p, size_t n)
|
---|
898 | Returns a pointer to a chunk of size n that contains the same data
|
---|
899 | as does chunk p up to the minimum of (n, p's size) bytes, or null
|
---|
900 | if no space is available.
|
---|
901 |
|
---|
902 | The returned pointer may or may not be the same as p. The algorithm
|
---|
903 | prefers extending p when possible, otherwise it employs the
|
---|
904 | equivalent of a malloc-copy-free sequence.
|
---|
905 |
|
---|
906 | If p is null, realloc is equivalent to malloc.
|
---|
907 |
|
---|
908 | If space is not available, realloc returns null, errno is set (if on
|
---|
909 | ANSI) and p is NOT freed.
|
---|
910 |
|
---|
911 | if n is for fewer bytes than already held by p, the newly unused
|
---|
912 | space is lopped off and freed if possible. Unless the #define
|
---|
913 | REALLOC_ZERO_BYTES_FREES is set, realloc with a size argument of
|
---|
914 | zero (re)allocates a minimum-sized chunk.
|
---|
915 |
|
---|
916 | Large chunks that were internally obtained via mmap will always
|
---|
917 | be reallocated using malloc-copy-free sequences unless
|
---|
918 | the system supports MREMAP (currently only linux).
|
---|
919 |
|
---|
920 | The old unix realloc convention of allowing the last-free'd chunk
|
---|
921 | to be used as an argument to realloc is not supported.
|
---|
922 | */
|
---|
923 | #if __STD_C
|
---|
924 | Void_t* public_rEALLOc(Void_t*, size_t) __attribute_malloc__;
|
---|
925 | #else
|
---|
926 | Void_t* public_rEALLOc();
|
---|
927 | #endif
|
---|
928 |
|
---|
929 | /*
|
---|
930 | memalign(size_t alignment, size_t n);
|
---|
931 | Returns a pointer to a newly allocated chunk of n bytes, aligned
|
---|
932 | in accord with the alignment argument.
|
---|
933 |
|
---|
934 | The alignment argument should be a power of two. If the argument is
|
---|
935 | not a power of two, the nearest greater power is used.
|
---|
936 | 8-byte alignment is guaranteed by normal malloc calls, so don't
|
---|
937 | bother calling memalign with an argument of 8 or less.
|
---|
938 |
|
---|
939 | Overreliance on memalign is a sure way to fragment space.
|
---|
940 | */
|
---|
941 | #if __STD_C
|
---|
942 | Void_t* public_mEMALIGn(size_t, size_t) __attribute_malloc__;
|
---|
943 | #else
|
---|
944 | Void_t* public_mEMALIGn();
|
---|
945 | #endif
|
---|
946 |
|
---|
947 | /*
|
---|
948 | posix_memalign(void** memptr, size_t alignment, size_t n);
|
---|
949 | Sets *memptr to the address of a newly allocated chunk of n bytes, aligned
|
---|
950 | in accord with the alignment argument. Returns 0 on success, otherwise
|
---|
951 | an error (EINVAL for incorrect alignment, ENOMEM for out of memory).
|
---|
952 |
|
---|
953 | The alignment must be a power of two, and a multiple of sizeof(void *).
|
---|
954 | */
|
---|
955 | #if __STD_C
|
---|
956 | int public_posix_mEMALIGn(Void_t**, size_t, size_t);
|
---|
957 | #else
|
---|
958 | int public_posix_mEMALIGn();
|
---|
959 | #endif
|
---|
960 |
|
---|
961 | /*
|
---|
962 | valloc(size_t n);
|
---|
963 | Equivalent to memalign(pagesize, n), where pagesize is the page
|
---|
964 | size of the system. If the pagesize is unknown, 4096 is used.
|
---|
965 | */
|
---|
966 | #if __STD_C
|
---|
967 | Void_t* public_vALLOc(size_t) __attribute_malloc__;
|
---|
968 | #else
|
---|
969 | Void_t* public_vALLOc();
|
---|
970 | #endif
|
---|
971 |
|
---|
972 |
|
---|
973 |
|
---|
974 | /*
|
---|
975 | mallopt(int parameter_number, int parameter_value)
|
---|
976 | Sets tunable parameters The format is to provide a
|
---|
977 | (parameter-number, parameter-value) pair. mallopt then sets the
|
---|
978 | corresponding parameter to the argument value if it can (i.e., so
|
---|
979 | long as the value is meaningful), and returns 1 if successful else
|
---|
980 | 0. SVID/XPG/ANSI defines four standard param numbers for mallopt,
|
---|
981 | normally defined in malloc.h. Only one of these (M_MXFAST) is used
|
---|
982 | in this malloc. The others (M_NLBLKS, M_GRAIN, M_KEEP) don't apply,
|
---|
983 | so setting them has no effect. But this malloc also supports four
|
---|
984 | other options in mallopt. See below for details. Briefly, supported
|
---|
985 | parameters are as follows (listed defaults are for "typical"
|
---|
986 | configurations).
|
---|
987 |
|
---|
988 | Symbol param # default allowed param values
|
---|
989 | M_MXFAST 1 64 0-80 (0 disables fastbins)
|
---|
990 | M_TRIM_THRESHOLD -1 256*1024 any (-1U disables trimming)
|
---|
991 | M_TOP_PAD -2 0 any
|
---|
992 | M_MMAP_THRESHOLD -3 256*1024 any (or 0 if no MMAP support)
|
---|
993 | M_MMAP_MAX -4 65536 any (0 disables use of mmap)
|
---|
994 | */
|
---|
995 | #if __STD_C
|
---|
996 | int public_mALLOPt(int, int);
|
---|
997 | #else
|
---|
998 | int public_mALLOPt();
|
---|
999 | #endif
|
---|
1000 |
|
---|
1001 |
|
---|
1002 | /*
|
---|
1003 | mallinfo()
|
---|
1004 | Returns (by copy) a struct containing various summary statistics:
|
---|
1005 |
|
---|
1006 | arena: current total non-mmapped bytes allocated from system
|
---|
1007 | ordblks: the number of free chunks
|
---|
1008 | smblks: the number of fastbin blocks (i.e., small chunks that
|
---|
1009 | have been freed but not use resused or consolidated)
|
---|
1010 | hblks: current number of mmapped regions
|
---|
1011 | hblkhd: total bytes held in mmapped regions
|
---|
1012 | usmblks: the maximum total allocated space. This will be greater
|
---|
1013 | than current total if trimming has occurred.
|
---|
1014 | fsmblks: total bytes held in fastbin blocks
|
---|
1015 | uordblks: current total allocated space (normal or mmapped)
|
---|
1016 | fordblks: total free space
|
---|
1017 | keepcost: the maximum number of bytes that could ideally be released
|
---|
1018 | back to system via malloc_trim. ("ideally" means that
|
---|
1019 | it ignores page restrictions etc.)
|
---|
1020 |
|
---|
1021 | Because these fields are ints, but internal bookkeeping may
|
---|
1022 | be kept as longs, the reported values may wrap around zero and
|
---|
1023 | thus be inaccurate.
|
---|
1024 | */
|
---|
1025 | #if __STD_C
|
---|
1026 | struct mallinfo public_mALLINFo(void);
|
---|
1027 | #else
|
---|
1028 | struct mallinfo public_mALLINFo();
|
---|
1029 | #endif
|
---|
1030 |
|
---|
1031 | /*
|
---|
1032 | pvalloc(size_t n);
|
---|
1033 | Equivalent to valloc(minimum-page-that-holds(n)), that is,
|
---|
1034 | round up n to nearest pagesize.
|
---|
1035 | */
|
---|
1036 | #if __STD_C
|
---|
1037 | Void_t* public_pVALLOc(size_t) __attribute_malloc__;
|
---|
1038 | #else
|
---|
1039 | Void_t* public_pVALLOc();
|
---|
1040 | #endif
|
---|
1041 |
|
---|
1042 | /*
|
---|
1043 | malloc_trim(size_t pad);
|
---|
1044 |
|
---|
1045 | If possible, gives memory back to the system (via negative
|
---|
1046 | arguments to sbrk) if there is unused memory at the `high' end of
|
---|
1047 | the malloc pool. You can call this after freeing large blocks of
|
---|
1048 | memory to potentially reduce the system-level memory requirements
|
---|
1049 | of a program. However, it cannot guarantee to reduce memory. Under
|
---|
1050 | some allocation patterns, some large free blocks of memory will be
|
---|
1051 | locked between two used chunks, so they cannot be given back to
|
---|
1052 | the system.
|
---|
1053 |
|
---|
1054 | The `pad' argument to malloc_trim represents the amount of free
|
---|
1055 | trailing space to leave untrimmed. If this argument is zero,
|
---|
1056 | only the minimum amount of memory to maintain internal data
|
---|
1057 | structures will be left (one page or less). Non-zero arguments
|
---|
1058 | can be supplied to maintain enough trailing space to service
|
---|
1059 | future expected allocations without having to re-obtain memory
|
---|
1060 | from the system.
|
---|
1061 |
|
---|
1062 | Malloc_trim returns 1 if it actually released any memory, else 0.
|
---|
1063 | On systems that do not support "negative sbrks", it will always
|
---|
1064 | rreturn 0.
|
---|
1065 | */
|
---|
1066 | #if __STD_C
|
---|
1067 | int public_mTRIm(size_t);
|
---|
1068 | #else
|
---|
1069 | int public_mTRIm();
|
---|
1070 | #endif
|
---|
1071 |
|
---|
1072 | /*
|
---|
1073 | malloc_usable_size(Void_t* p);
|
---|
1074 |
|
---|
1075 | Returns the number of bytes you can actually use in
|
---|
1076 | an allocated chunk, which may be more than you requested (although
|
---|
1077 | often not) due to alignment and minimum size constraints.
|
---|
1078 | You can use this many bytes without worrying about
|
---|
1079 | overwriting other allocated objects. This is not a particularly great
|
---|
1080 | programming practice. malloc_usable_size can be more useful in
|
---|
1081 | debugging and assertions, for example:
|
---|
1082 |
|
---|
1083 | p = malloc(n);
|
---|
1084 | assert(malloc_usable_size(p) >= 256);
|
---|
1085 |
|
---|
1086 | */
|
---|
1087 | #if __STD_C
|
---|
1088 | size_t public_mUSABLe(Void_t*);
|
---|
1089 | #else
|
---|
1090 | size_t public_mUSABLe();
|
---|
1091 | #endif
|
---|
1092 |
|
---|
1093 | /*
|
---|
1094 | malloc_stats();
|
---|
1095 | Prints on stderr the amount of space obtained from the system (both
|
---|
1096 | via sbrk and mmap), the maximum amount (which may be more than
|
---|
1097 | current if malloc_trim and/or munmap got called), and the current
|
---|
1098 | number of bytes allocated via malloc (or realloc, etc) but not yet
|
---|
1099 | freed. Note that this is the number of bytes allocated, not the
|
---|
1100 | number requested. It will be larger than the number requested
|
---|
1101 | because of alignment and bookkeeping overhead. Because it includes
|
---|
1102 | alignment wastage as being in use, this figure may be greater than
|
---|
1103 | zero even when no user-level chunks are allocated.
|
---|
1104 |
|
---|
1105 | The reported current and maximum system memory can be inaccurate if
|
---|
1106 | a program makes other calls to system memory allocation functions
|
---|
1107 | (normally sbrk) outside of malloc.
|
---|
1108 |
|
---|
1109 | malloc_stats prints only the most commonly interesting statistics.
|
---|
1110 | More information can be obtained by calling mallinfo.
|
---|
1111 |
|
---|
1112 | */
|
---|
1113 | #if __STD_C
|
---|
1114 | void public_mSTATs();
|
---|
1115 | #else
|
---|
1116 | void public_mSTATs();
|
---|
1117 | #endif
|
---|
1118 |
|
---|
1119 | /* mallopt tuning options */
|
---|
1120 |
|
---|
1121 | /*
|
---|
1122 | M_MXFAST is the maximum request size used for "fastbins", special bins
|
---|
1123 | that hold returned chunks without consolidating their spaces. This
|
---|
1124 | enables future requests for chunks of the same size to be handled
|
---|
1125 | very quickly, but can increase fragmentation, and thus increase the
|
---|
1126 | overall memory footprint of a program.
|
---|
1127 |
|
---|
1128 | This malloc manages fastbins very conservatively yet still
|
---|
1129 | efficiently, so fragmentation is rarely a problem for values less
|
---|
1130 | than or equal to the default. The maximum supported value of MXFAST
|
---|
1131 | is 80. You wouldn't want it any higher than this anyway. Fastbins
|
---|
1132 | are designed especially for use with many small structs, objects or
|
---|
1133 | strings -- the default handles structs/objects/arrays with sizes up
|
---|
1134 | to 16 4byte fields, or small strings representing words, tokens,
|
---|
1135 | etc. Using fastbins for larger objects normally worsens
|
---|
1136 | fragmentation without improving speed.
|
---|
1137 |
|
---|
1138 | M_MXFAST is set in REQUEST size units. It is internally used in
|
---|
1139 | chunksize units, which adds padding and alignment. You can reduce
|
---|
1140 | M_MXFAST to 0 to disable all use of fastbins. This causes the malloc
|
---|
1141 | algorithm to be a closer approximation of fifo-best-fit in all cases,
|
---|
1142 | not just for larger requests, but will generally cause it to be
|
---|
1143 | slower.
|
---|
1144 | */
|
---|
1145 |
|
---|
1146 |
|
---|
1147 | /* M_MXFAST is a standard SVID/XPG tuning option, usually listed in malloc.h */
|
---|
1148 | #ifndef M_MXFAST
|
---|
1149 | #define M_MXFAST 1
|
---|
1150 | #endif
|
---|
1151 |
|
---|
1152 | #ifndef DEFAULT_MXFAST
|
---|
1153 | #define DEFAULT_MXFAST 64
|
---|
1154 | #endif
|
---|
1155 |
|
---|
1156 |
|
---|
1157 | /*
|
---|
1158 | M_TRIM_THRESHOLD is the maximum amount of unused top-most memory
|
---|
1159 | to keep before releasing via malloc_trim in free().
|
---|
1160 |
|
---|
1161 | Automatic trimming is mainly useful in long-lived programs.
|
---|
1162 | Because trimming via sbrk can be slow on some systems, and can
|
---|
1163 | sometimes be wasteful (in cases where programs immediately
|
---|
1164 | afterward allocate more large chunks) the value should be high
|
---|
1165 | enough so that your overall system performance would improve by
|
---|
1166 | releasing this much memory.
|
---|
1167 |
|
---|
1168 | The trim threshold and the mmap control parameters (see below)
|
---|
1169 | can be traded off with one another. Trimming and mmapping are
|
---|
1170 | two different ways of releasing unused memory back to the
|
---|
1171 | system. Between these two, it is often possible to keep
|
---|
1172 | system-level demands of a long-lived program down to a bare
|
---|
1173 | minimum. For example, in one test suite of sessions measuring
|
---|
1174 | the XF86 X server on Linux, using a trim threshold of 128K and a
|
---|
1175 | mmap threshold of 192K led to near-minimal long term resource
|
---|
1176 | consumption.
|
---|
1177 |
|
---|
1178 | If you are using this malloc in a long-lived program, it should
|
---|
1179 | pay to experiment with these values. As a rough guide, you
|
---|
1180 | might set to a value close to the average size of a process
|
---|
1181 | (program) running on your system. Releasing this much memory
|
---|
1182 | would allow such a process to run in memory. Generally, it's
|
---|
1183 | worth it to tune for trimming rather tham memory mapping when a
|
---|
1184 | program undergoes phases where several large chunks are
|
---|
1185 | allocated and released in ways that can reuse each other's
|
---|
1186 | storage, perhaps mixed with phases where there are no such
|
---|
1187 | chunks at all. And in well-behaved long-lived programs,
|
---|
1188 | controlling release of large blocks via trimming versus mapping
|
---|
1189 | is usually faster.
|
---|
1190 |
|
---|
1191 | However, in most programs, these parameters serve mainly as
|
---|
1192 | protection against the system-level effects of carrying around
|
---|
1193 | massive amounts of unneeded memory. Since frequent calls to
|
---|
1194 | sbrk, mmap, and munmap otherwise degrade performance, the default
|
---|
1195 | parameters are set to relatively high values that serve only as
|
---|
1196 | safeguards.
|
---|
1197 |
|
---|
1198 | The trim value must be greater than page size to have any useful
|
---|
1199 | effect. To disable trimming completely, you can set to
|
---|
1200 | (unsigned long)(-1)
|
---|
1201 |
|
---|
1202 | Trim settings interact with fastbin (MXFAST) settings: Unless
|
---|
1203 | TRIM_FASTBINS is defined, automatic trimming never takes place upon
|
---|
1204 | freeing a chunk with size less than or equal to MXFAST. Trimming is
|
---|
1205 | instead delayed until subsequent freeing of larger chunks. However,
|
---|
1206 | you can still force an attempted trim by calling malloc_trim.
|
---|
1207 |
|
---|
1208 | Also, trimming is not generally possible in cases where
|
---|
1209 | the main arena is obtained via mmap.
|
---|
1210 |
|
---|
1211 | Note that the trick some people use of mallocing a huge space and
|
---|
1212 | then freeing it at program startup, in an attempt to reserve system
|
---|
1213 | memory, doesn't have the intended effect under automatic trimming,
|
---|
1214 | since that memory will immediately be returned to the system.
|
---|
1215 | */
|
---|
1216 |
|
---|
1217 | #define M_TRIM_THRESHOLD -1
|
---|
1218 |
|
---|
1219 | #ifndef DEFAULT_TRIM_THRESHOLD
|
---|
1220 | #define DEFAULT_TRIM_THRESHOLD (256 * 1024)
|
---|
1221 | #endif
|
---|
1222 |
|
---|
1223 | /*
|
---|
1224 | M_TOP_PAD is the amount of extra `padding' space to allocate or
|
---|
1225 | retain whenever sbrk is called. It is used in two ways internally:
|
---|
1226 |
|
---|
1227 | * When sbrk is called to extend the top of the arena to satisfy
|
---|
1228 | a new malloc request, this much padding is added to the sbrk
|
---|
1229 | request.
|
---|
1230 |
|
---|
1231 | * When malloc_trim is called automatically from free(),
|
---|
1232 | it is used as the `pad' argument.
|
---|
1233 |
|
---|
1234 | In both cases, the actual amount of padding is rounded
|
---|
1235 | so that the end of the arena is always a system page boundary.
|
---|
1236 |
|
---|
1237 | The main reason for using padding is to avoid calling sbrk so
|
---|
1238 | often. Having even a small pad greatly reduces the likelihood
|
---|
1239 | that nearly every malloc request during program start-up (or
|
---|
1240 | after trimming) will invoke sbrk, which needlessly wastes
|
---|
1241 | time.
|
---|
1242 |
|
---|
1243 | Automatic rounding-up to page-size units is normally sufficient
|
---|
1244 | to avoid measurable overhead, so the default is 0. However, in
|
---|
1245 | systems where sbrk is relatively slow, it can pay to increase
|
---|
1246 | this value, at the expense of carrying around more memory than
|
---|
1247 | the program needs.
|
---|
1248 | */
|
---|
1249 |
|
---|
1250 | #define M_TOP_PAD -2
|
---|
1251 |
|
---|
1252 | #ifndef DEFAULT_TOP_PAD
|
---|
1253 | #define DEFAULT_TOP_PAD (0)
|
---|
1254 | #endif
|
---|
1255 |
|
---|
1256 | /*
|
---|
1257 | M_MMAP_THRESHOLD is the request size threshold for using mmap()
|
---|
1258 | to service a request. Requests of at least this size that cannot
|
---|
1259 | be allocated using already-existing space will be serviced via mmap.
|
---|
1260 | (If enough normal freed space already exists it is used instead.)
|
---|
1261 |
|
---|
1262 | Using mmap segregates relatively large chunks of memory so that
|
---|
1263 | they can be individually obtained and released from the host
|
---|
1264 | system. A request serviced through mmap is never reused by any
|
---|
1265 | other request (at least not directly; the system may just so
|
---|
1266 | happen to remap successive requests to the same locations).
|
---|
1267 |
|
---|
1268 | Segregating space in this way has the benefits that:
|
---|
1269 |
|
---|
1270 | 1. Mmapped space can ALWAYS be individually released back
|
---|
1271 | to the system, which helps keep the system level memory
|
---|
1272 | demands of a long-lived program low.
|
---|
1273 | 2. Mapped memory can never become `locked' between
|
---|
1274 | other chunks, as can happen with normally allocated chunks, which
|
---|
1275 | means that even trimming via malloc_trim would not release them.
|
---|
1276 | 3. On some systems with "holes" in address spaces, mmap can obtain
|
---|
1277 | memory that sbrk cannot.
|
---|
1278 |
|
---|
1279 | However, it has the disadvantages that:
|
---|
1280 |
|
---|
1281 | 1. The space cannot be reclaimed, consolidated, and then
|
---|
1282 | used to service later requests, as happens with normal chunks.
|
---|
1283 | 2. It can lead to more wastage because of mmap page alignment
|
---|
1284 | requirements
|
---|
1285 | 3. It causes malloc performance to be more dependent on host
|
---|
1286 | system memory management support routines which may vary in
|
---|
1287 | implementation quality and may impose arbitrary
|
---|
1288 | limitations. Generally, servicing a request via normal
|
---|
1289 | malloc steps is faster than going through a system's mmap.
|
---|
1290 |
|
---|
1291 | The advantages of mmap nearly always outweigh disadvantages for
|
---|
1292 | "large" chunks, but the value of "large" varies across systems. The
|
---|
1293 | default is an empirically derived value that works well in most
|
---|
1294 | systems.
|
---|
1295 | */
|
---|
1296 |
|
---|
1297 | #define M_MMAP_THRESHOLD -3
|
---|
1298 |
|
---|
1299 | #ifndef DEFAULT_MMAP_THRESHOLD
|
---|
1300 | #define DEFAULT_MMAP_THRESHOLD (256 * 1024)
|
---|
1301 | #endif
|
---|
1302 |
|
---|
1303 | /*
|
---|
1304 | M_MMAP_MAX is the maximum number of requests to simultaneously
|
---|
1305 | service using mmap. This parameter exists because
|
---|
1306 | . Some systems have a limited number of internal tables for
|
---|
1307 | use by mmap, and using more than a few of them may degrade
|
---|
1308 | performance.
|
---|
1309 |
|
---|
1310 | The default is set to a value that serves only as a safeguard.
|
---|
1311 | Setting to 0 disables use of mmap for servicing large requests. If
|
---|
1312 | HAVE_MMAP is not set, the default value is 0, and attempts to set it
|
---|
1313 | to non-zero values in mallopt will fail.
|
---|
1314 | */
|
---|
1315 |
|
---|
1316 | #define M_MMAP_MAX -4
|
---|
1317 |
|
---|
1318 | #ifndef DEFAULT_MMAP_MAX
|
---|
1319 | #define DEFAULT_MMAP_MAX (65536)
|
---|
1320 | #endif
|
---|
1321 |
|
---|
1322 | #ifdef __cplusplus
|
---|
1323 | }; /* end of extern "C" */
|
---|
1324 | #endif
|
---|
1325 |
|
---|
1326 | /*
|
---|
1327 | ========================================================================
|
---|
1328 | To make a fully customizable malloc.h header file, cut everything
|
---|
1329 | above this line, put into file malloc.h, edit to suit, and #include it
|
---|
1330 | on the next line, as well as in programs that use this malloc.
|
---|
1331 | ========================================================================
|
---|
1332 | */
|
---|
1333 |
|
---|
1334 | /* #include "malloc.h" */
|
---|
1335 |
|
---|
1336 | /* --------------------- public wrappers ---------------------- */
|
---|
1337 |
|
---|
1338 | #ifdef USE_PUBLIC_MALLOC_WRAPPERS
|
---|
1339 |
|
---|
1340 | /* DL_STATIC used to make functions (deep down) consistent
|
---|
1341 | * with prototypes (otherwise the prototypes are static
|
---|
1342 | * with USE_PUBLIC_MALLOC_WRAPPERS, but the functions aren't).
|
---|
1343 | * The gcc compiler doesn't care, but the HP-UX compiler does.
|
---|
1344 | */
|
---|
1345 | #define DL_STATIC static
|
---|
1346 |
|
---|
1347 | /* Declare all routines as internal */
|
---|
1348 | #if __STD_C
|
---|
1349 | static Void_t* mALLOc(size_t) __attribute_malloc__;
|
---|
1350 | static void fREe(Void_t*);
|
---|
1351 | static Void_t* rEALLOc(Void_t*, size_t) __attribute_malloc__;
|
---|
1352 | static Void_t* mEMALIGn(size_t, size_t) __attribute_malloc__;
|
---|
1353 | static int posix_mEMALIGn(Void_t**, size_t, size_t);
|
---|
1354 | static Void_t* vALLOc(size_t) __attribute_malloc__;
|
---|
1355 | static Void_t* pVALLOc(size_t) __attribute_malloc__;
|
---|
1356 | static Void_t* cALLOc(size_t, size_t) __attribute_malloc__;
|
---|
1357 | static int mTRIm(size_t);
|
---|
1358 | static size_t mUSABLe(Void_t*);
|
---|
1359 | static void mSTATs();
|
---|
1360 | static int mALLOPt(int, int);
|
---|
1361 | static struct mallinfo mALLINFo(void);
|
---|
1362 | #else
|
---|
1363 | static Void_t* mALLOc();
|
---|
1364 | static void fREe();
|
---|
1365 | static Void_t* rEALLOc();
|
---|
1366 | static Void_t* mEMALIGn();
|
---|
1367 | static int posix_mEMALIGn();
|
---|
1368 | static Void_t* vALLOc();
|
---|
1369 | static Void_t* pVALLOc();
|
---|
1370 | static Void_t* cALLOc();
|
---|
1371 | static int mTRIm();
|
---|
1372 | static size_t mUSABLe();
|
---|
1373 | static void mSTATs();
|
---|
1374 | static int mALLOPt();
|
---|
1375 | static struct mallinfo mALLINFo();
|
---|
1376 | #endif
|
---|
1377 |
|
---|
1378 | /*
|
---|
1379 | MALLOC_PREACTION and MALLOC_POSTACTION should be
|
---|
1380 | defined to return 0 on success, and nonzero on failure.
|
---|
1381 | The return value of MALLOC_POSTACTION is currently ignored
|
---|
1382 | in wrapper functions since there is no reasonable default
|
---|
1383 | action to take on failure.
|
---|
1384 | */
|
---|
1385 |
|
---|
1386 |
|
---|
1387 | #ifdef USE_MALLOC_LOCK
|
---|
1388 |
|
---|
1389 | # ifdef WIN32
|
---|
1390 |
|
---|
1391 | static int mALLOC_MUTEx;
|
---|
1392 | #define MALLOC_PREACTION slwait(&mALLOC_MUTEx)
|
---|
1393 | #define MALLOC_POSTACTION slrelease(&mALLOC_MUTEx)
|
---|
1394 | int dnmalloc_pthread_init(void) { return 0; }
|
---|
1395 |
|
---|
1396 | # elif defined(__NetBSD__) || defined(__OpenBSD__) || defined(__FreeBSD__)
|
---|
1397 |
|
---|
1398 | # if defined(__NetBSD__)
|
---|
1399 | #include <reentrant.h>
|
---|
1400 | extern int __isthreaded;
|
---|
1401 | static mutex_t thread_lock = MUTEX_INITIALIZER;
|
---|
1402 | #define _MALLOC_LOCK() if (__isthreaded) mutex_lock(&thread_lock)
|
---|
1403 | #define _MALLOC_UNLOCK() if (__isthreaded) mutex_unlock(&thread_lock)
|
---|
1404 | void _malloc_prefork(void) { _MALLOC_LOCK(); }
|
---|
1405 | void _malloc_postfork(void) { _MALLOC_UNLOCK(); }
|
---|
1406 | # endif
|
---|
1407 |
|
---|
1408 | # if defined(__OpenBSD__)
|
---|
1409 | extern int __isthreaded;
|
---|
1410 | void _thread_malloc_lock(void);
|
---|
1411 | void _thread_malloc_unlock(void);
|
---|
1412 | #define _MALLOC_LOCK() if (__isthreaded) _thread_malloc_lock()
|
---|
1413 | #define _MALLOC_UNLOCK() if (__isthreaded) _thread_malloc_unlock()
|
---|
1414 | # endif
|
---|
1415 |
|
---|
1416 | # if defined(__FreeBSD__)
|
---|
1417 | extern int __isthreaded;
|
---|
1418 | struct _spinlock {
|
---|
1419 | volatile long access_lock;
|
---|
1420 | volatile long lock_owner;
|
---|
1421 | volatile char *fname;
|
---|
1422 | volatile int lineno;
|
---|
1423 | };
|
---|
1424 | typedef struct _spinlock spinlock_t;
|
---|
1425 | #define _SPINLOCK_INITIALIZER { 0, 0, 0, 0 }
|
---|
1426 | void _spinlock(spinlock_t *);
|
---|
1427 | void _spinunlock(spinlock_t *);
|
---|
1428 | /* # include "/usr/src/lib/libc/include/spinlock.h" */
|
---|
1429 | static spinlock_t thread_lock = _SPINLOCK_INITIALIZER;
|
---|
1430 | spinlock_t *__malloc_lock = &thread_lock;
|
---|
1431 | #define _MALLOC_LOCK() if (__isthreaded) _spinlock(&thread_lock)
|
---|
1432 | #define _MALLOC_UNLOCK() if (__isthreaded) _spinunlock(&thread_lock)
|
---|
1433 | # endif
|
---|
1434 |
|
---|
1435 | /* Common for all three *BSD
|
---|
1436 | */
|
---|
1437 | static int malloc_active = 0;
|
---|
1438 | static int dnmalloc_mutex_lock()
|
---|
1439 | {
|
---|
1440 | _MALLOC_LOCK();
|
---|
1441 | if (!malloc_active)
|
---|
1442 | {
|
---|
1443 | ++malloc_active;
|
---|
1444 | return 0;
|
---|
1445 | }
|
---|
1446 | assert(malloc_active == 0);
|
---|
1447 | _MALLOC_UNLOCK();
|
---|
1448 | errno = EDEADLK;
|
---|
1449 | return 1;
|
---|
1450 | }
|
---|
1451 | static int dnmalloc_mutex_unlock()
|
---|
1452 | {
|
---|
1453 | --malloc_active;
|
---|
1454 | _MALLOC_UNLOCK();
|
---|
1455 | return 0;
|
---|
1456 | }
|
---|
1457 | #define MALLOC_PREACTION dnmalloc_mutex_lock()
|
---|
1458 | #define MALLOC_POSTACTION dnmalloc_mutex_unlock()
|
---|
1459 | int dnmalloc_pthread_init(void) { return 0; }
|
---|
1460 |
|
---|
1461 | # else
|
---|
1462 |
|
---|
1463 | /* Wrapping malloc with pthread_mutex_lock/pthread_mutex_unlock
|
---|
1464 | *
|
---|
1465 | * Works fine on linux (no malloc in pthread_mutex_lock)
|
---|
1466 | * Works with on HP-UX if initialized after entering main()
|
---|
1467 | */
|
---|
1468 | #include <pthread.h>
|
---|
1469 | static int malloc_active = 0;
|
---|
1470 | void dnmalloc_fork_prepare(void);
|
---|
1471 | void dnmalloc_fork_parent(void);
|
---|
1472 | void dnmalloc_fork_child(void);
|
---|
1473 |
|
---|
1474 | #if !defined(__linux__)
|
---|
1475 |
|
---|
1476 | static pthread_mutex_t mALLOC_MUTEx;
|
---|
1477 | pthread_once_t dnmalloc_once_control = PTHREAD_ONCE_INIT;
|
---|
1478 | static int dnmalloc_use_mutex = 0;
|
---|
1479 | static void dnmalloc_pthread_init_int(void)
|
---|
1480 | {
|
---|
1481 | pthread_mutexattr_t mta;
|
---|
1482 | pthread_mutexattr_init(&mta);
|
---|
1483 | pthread_mutexattr_settype(&mta, PTHREAD_MUTEX_RECURSIVE);
|
---|
1484 | pthread_mutex_init(&(mALLOC_MUTEx), &mta);
|
---|
1485 | pthread_mutexattr_destroy(&mta);
|
---|
1486 | pthread_atfork(dnmalloc_fork_prepare,
|
---|
1487 | dnmalloc_fork_parent,
|
---|
1488 | dnmalloc_fork_child);
|
---|
1489 | dnmalloc_use_mutex = 1;
|
---|
1490 | }
|
---|
1491 | int dnmalloc_pthread_init(void)
|
---|
1492 | {
|
---|
1493 | return pthread_once(&dnmalloc_once_control, dnmalloc_pthread_init_int);
|
---|
1494 | }
|
---|
1495 |
|
---|
1496 | #else
|
---|
1497 |
|
---|
1498 | static pthread_mutex_t mALLOC_MUTEx = PTHREAD_MUTEX_INITIALIZER;
|
---|
1499 | static int dnmalloc_use_mutex = 1;
|
---|
1500 | int dnmalloc_pthread_init(void) {
|
---|
1501 | return pthread_atfork(dnmalloc_fork_prepare,
|
---|
1502 | dnmalloc_fork_parent,
|
---|
1503 | dnmalloc_fork_child);
|
---|
1504 | }
|
---|
1505 | #endif /* !defined(__linux__) */
|
---|
1506 |
|
---|
1507 | void dnmalloc_fork_prepare(void) {
|
---|
1508 | if (dnmalloc_use_mutex)
|
---|
1509 | pthread_mutex_lock(&mALLOC_MUTEx);
|
---|
1510 | }
|
---|
1511 | void dnmalloc_fork_parent(void) {
|
---|
1512 | if (dnmalloc_use_mutex)
|
---|
1513 | pthread_mutex_unlock(&mALLOC_MUTEx);
|
---|
1514 | }
|
---|
1515 | void dnmalloc_fork_child(void) {
|
---|
1516 | int rc = 0;
|
---|
1517 | #ifdef __GLIBC__
|
---|
1518 | if (dnmalloc_use_mutex)
|
---|
1519 | {
|
---|
1520 | pthread_mutex_unlock (&mALLOC_MUTEx);
|
---|
1521 | pthread_mutex_destroy(&mALLOC_MUTEx);
|
---|
1522 | rc = pthread_mutex_init(&mALLOC_MUTEx, NULL);
|
---|
1523 | }
|
---|
1524 | #else
|
---|
1525 | if (dnmalloc_use_mutex)
|
---|
1526 | rc = pthread_mutex_unlock(&mALLOC_MUTEx);
|
---|
1527 | #endif
|
---|
1528 | if (rc != 0)
|
---|
1529 | {
|
---|
1530 | fputs("fork_child failed", stderr);
|
---|
1531 | _exit(EXIT_FAILURE);
|
---|
1532 | }
|
---|
1533 | }
|
---|
1534 | static int dnmalloc_mutex_lock(pthread_mutex_t *mutex)
|
---|
1535 | {
|
---|
1536 | if (dnmalloc_use_mutex)
|
---|
1537 | {
|
---|
1538 | int rc = pthread_mutex_lock(mutex);
|
---|
1539 | if (rc == 0)
|
---|
1540 | {
|
---|
1541 | if (!malloc_active)
|
---|
1542 | {
|
---|
1543 | ++malloc_active;
|
---|
1544 | return 0;
|
---|
1545 | }
|
---|
1546 | assert(malloc_active == 0);
|
---|
1547 | (void) pthread_mutex_unlock(mutex);
|
---|
1548 | errno = EDEADLK;
|
---|
1549 | return 1;
|
---|
1550 | }
|
---|
1551 | return rc;
|
---|
1552 | }
|
---|
1553 | return 0;
|
---|
1554 | }
|
---|
1555 | static int dnmalloc_mutex_unlock(pthread_mutex_t *mutex)
|
---|
1556 | {
|
---|
1557 | if (dnmalloc_use_mutex)
|
---|
1558 | {
|
---|
1559 | --malloc_active;
|
---|
1560 | return pthread_mutex_unlock(mutex);
|
---|
1561 | }
|
---|
1562 | return 0;
|
---|
1563 | }
|
---|
1564 | # define MALLOC_PREACTION dnmalloc_mutex_lock(&mALLOC_MUTEx)
|
---|
1565 | # define MALLOC_POSTACTION dnmalloc_mutex_unlock(&mALLOC_MUTEx)
|
---|
1566 |
|
---|
1567 | # endif
|
---|
1568 |
|
---|
1569 | #else
|
---|
1570 |
|
---|
1571 | /* Substitute anything you like for these */
|
---|
1572 |
|
---|
1573 | # define MALLOC_PREACTION (0)
|
---|
1574 | # define MALLOC_POSTACTION (0)
|
---|
1575 | int dnmalloc_pthread_init(void) { return 0; }
|
---|
1576 |
|
---|
1577 | #endif /* USE_MALLOC_LOCK */
|
---|
1578 |
|
---|
1579 | Void_t* public_mALLOc(size_t bytes) {
|
---|
1580 | Void_t* m;
|
---|
1581 | if (MALLOC_PREACTION == 0) {
|
---|
1582 | m = mALLOc(bytes);
|
---|
1583 | (void) MALLOC_POSTACTION;
|
---|
1584 | return m;
|
---|
1585 | }
|
---|
1586 | return 0;
|
---|
1587 | }
|
---|
1588 |
|
---|
1589 | void public_fREe(Void_t* m) {
|
---|
1590 | if (MALLOC_PREACTION == 0) {
|
---|
1591 | fREe(m);
|
---|
1592 | (void) MALLOC_POSTACTION;
|
---|
1593 | }
|
---|
1594 | }
|
---|
1595 |
|
---|
1596 | Void_t* public_rEALLOc(Void_t* m, size_t bytes) {
|
---|
1597 | if (MALLOC_PREACTION == 0) {
|
---|
1598 | m = rEALLOc(m, bytes);
|
---|
1599 | (void) MALLOC_POSTACTION;
|
---|
1600 | return m;
|
---|
1601 | }
|
---|
1602 | return 0;
|
---|
1603 | }
|
---|
1604 |
|
---|
1605 | Void_t* public_mEMALIGn(size_t alignment, size_t bytes) {
|
---|
1606 | Void_t* m;
|
---|
1607 | if (MALLOC_PREACTION == 0) {
|
---|
1608 | m = mEMALIGn(alignment, bytes);
|
---|
1609 | (void) MALLOC_POSTACTION;
|
---|
1610 | return m;
|
---|
1611 | }
|
---|
1612 | return 0;
|
---|
1613 | }
|
---|
1614 |
|
---|
1615 | int public_posix_mEMALIGn(Void_t**memptr, size_t alignment, size_t bytes) {
|
---|
1616 | int m, ret;
|
---|
1617 | if ((ret = MALLOC_PREACTION) == 0) {
|
---|
1618 | m = posix_mEMALIGn(memptr, alignment, bytes);
|
---|
1619 | (void) MALLOC_POSTACTION;
|
---|
1620 | return m;
|
---|
1621 | }
|
---|
1622 | return ret;
|
---|
1623 | }
|
---|
1624 |
|
---|
1625 | Void_t* public_vALLOc(size_t bytes) {
|
---|
1626 | Void_t* m;
|
---|
1627 | if (MALLOC_PREACTION == 0) {
|
---|
1628 | m = vALLOc(bytes);
|
---|
1629 | (void) MALLOC_POSTACTION;
|
---|
1630 | return m;
|
---|
1631 | }
|
---|
1632 | return 0;
|
---|
1633 | }
|
---|
1634 |
|
---|
1635 | Void_t* public_pVALLOc(size_t bytes) {
|
---|
1636 | Void_t* m;
|
---|
1637 | if (MALLOC_PREACTION == 0) {
|
---|
1638 | m = pVALLOc(bytes);
|
---|
1639 | (void) MALLOC_POSTACTION;
|
---|
1640 | return m;
|
---|
1641 | }
|
---|
1642 | return 0;
|
---|
1643 | }
|
---|
1644 |
|
---|
1645 | Void_t* public_cALLOc(size_t n, size_t elem_size) {
|
---|
1646 | Void_t* m;
|
---|
1647 | if (MALLOC_PREACTION == 0) {
|
---|
1648 | m = cALLOc(n, elem_size);
|
---|
1649 | (void) MALLOC_POSTACTION;
|
---|
1650 | return m;
|
---|
1651 | }
|
---|
1652 | return 0;
|
---|
1653 | }
|
---|
1654 |
|
---|
1655 | int public_mTRIm(size_t s) {
|
---|
1656 | int result;
|
---|
1657 | if (MALLOC_PREACTION == 0) {
|
---|
1658 | result = mTRIm(s);
|
---|
1659 | (void) MALLOC_POSTACTION;
|
---|
1660 | return result;
|
---|
1661 | }
|
---|
1662 | return 0;
|
---|
1663 | }
|
---|
1664 |
|
---|
1665 | size_t public_mUSABLe(Void_t* m) {
|
---|
1666 | size_t result;
|
---|
1667 | if (MALLOC_PREACTION == 0) {
|
---|
1668 | result = mUSABLe(m);
|
---|
1669 | (void) MALLOC_POSTACTION;
|
---|
1670 | return result;
|
---|
1671 | }
|
---|
1672 | return 0;
|
---|
1673 | }
|
---|
1674 |
|
---|
1675 | void public_mSTATs() {
|
---|
1676 | if (MALLOC_PREACTION == 0) {
|
---|
1677 | mSTATs();
|
---|
1678 | (void) MALLOC_POSTACTION;
|
---|
1679 | }
|
---|
1680 | }
|
---|
1681 |
|
---|
1682 | struct mallinfo public_mALLINFo() {
|
---|
1683 | struct mallinfo m;
|
---|
1684 | if (MALLOC_PREACTION == 0) {
|
---|
1685 | m = mALLINFo();
|
---|
1686 | (void) MALLOC_POSTACTION;
|
---|
1687 | return m;
|
---|
1688 | } else {
|
---|
1689 | struct mallinfo nm = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 };
|
---|
1690 | return nm;
|
---|
1691 | }
|
---|
1692 | }
|
---|
1693 |
|
---|
1694 | int public_mALLOPt(int p, int v) {
|
---|
1695 | int result;
|
---|
1696 | if (MALLOC_PREACTION == 0) {
|
---|
1697 | result = mALLOPt(p, v);
|
---|
1698 | (void) MALLOC_POSTACTION;
|
---|
1699 | return result;
|
---|
1700 | }
|
---|
1701 | return 0;
|
---|
1702 | }
|
---|
1703 |
|
---|
1704 | #else
|
---|
1705 |
|
---|
1706 | int dnmalloc_pthread_init(void) { return 0; }
|
---|
1707 | #define DL_STATIC
|
---|
1708 |
|
---|
1709 | #endif /* USE_PUBLIC_MALLOC_WRAPPERS */
|
---|
1710 |
|
---|
1711 |
|
---|
1712 |
|
---|
1713 | /* ------------- Optional versions of memcopy ---------------- */
|
---|
1714 |
|
---|
1715 |
|
---|
1716 | #if USE_MEMCPY
|
---|
1717 |
|
---|
1718 | /*
|
---|
1719 | Note: memcpy is ONLY invoked with non-overlapping regions,
|
---|
1720 | so the (usually slower) memmove is not needed.
|
---|
1721 | */
|
---|
1722 |
|
---|
1723 | #define MALLOC_COPY(dest, src, nbytes) memcpy(dest, src, nbytes)
|
---|
1724 | #define MALLOC_ZERO(dest, nbytes) memset(dest, 0, nbytes)
|
---|
1725 |
|
---|
1726 | #else /* !USE_MEMCPY */
|
---|
1727 |
|
---|
1728 | /* Use Duff's device for good zeroing/copying performance. */
|
---|
1729 |
|
---|
1730 | #define MALLOC_ZERO(charp, nbytes) \
|
---|
1731 | do { \
|
---|
1732 | INTERNAL_SIZE_T* mzp = (INTERNAL_SIZE_T*)(charp); \
|
---|
1733 | CHUNK_SIZE_T mctmp = (nbytes)/sizeof(INTERNAL_SIZE_T); \
|
---|
1734 | long mcn; \
|
---|
1735 | if (mctmp < 8) mcn = 0; else { mcn = (mctmp-1)/8; mctmp %= 8; } \
|
---|
1736 | switch (mctmp) { \
|
---|
1737 | case 0: for(;;) { *mzp++ = 0; \
|
---|
1738 | case 7: *mzp++ = 0; \
|
---|
1739 | case 6: *mzp++ = 0; \
|
---|
1740 | case 5: *mzp++ = 0; \
|
---|
1741 | case 4: *mzp++ = 0; \
|
---|
1742 | case 3: *mzp++ = 0; \
|
---|
1743 | case 2: *mzp++ = 0; \
|
---|
1744 | case 1: *mzp++ = 0; if(mcn <= 0) break; mcn--; } \
|
---|
1745 | } \
|
---|
1746 | } while(0)
|
---|
1747 |
|
---|
1748 | #define MALLOC_COPY(dest,src,nbytes) \
|
---|
1749 | do { \
|
---|
1750 | INTERNAL_SIZE_T* mcsrc = (INTERNAL_SIZE_T*) src; \
|
---|
1751 | INTERNAL_SIZE_T* mcdst = (INTERNAL_SIZE_T*) dest; \
|
---|
1752 | CHUNK_SIZE_T mctmp = (nbytes)/sizeof(INTERNAL_SIZE_T); \
|
---|
1753 | long mcn; \
|
---|
1754 | if (mctmp < 8) mcn = 0; else { mcn = (mctmp-1)/8; mctmp %= 8; } \
|
---|
1755 | switch (mctmp) { \
|
---|
1756 | case 0: for(;;) { *mcdst++ = *mcsrc++; \
|
---|
1757 | case 7: *mcdst++ = *mcsrc++; \
|
---|
1758 | case 6: *mcdst++ = *mcsrc++; \
|
---|
1759 | case 5: *mcdst++ = *mcsrc++; \
|
---|
1760 | case 4: *mcdst++ = *mcsrc++; \
|
---|
1761 | case 3: *mcdst++ = *mcsrc++; \
|
---|
1762 | case 2: *mcdst++ = *mcsrc++; \
|
---|
1763 | case 1: *mcdst++ = *mcsrc++; if(mcn <= 0) break; mcn--; } \
|
---|
1764 | } \
|
---|
1765 | } while(0)
|
---|
1766 |
|
---|
1767 | #endif
|
---|
1768 |
|
---|
1769 | /* ------------------ MMAP support ------------------ */
|
---|
1770 |
|
---|
1771 |
|
---|
1772 | #if defined(HAVE_FCNTL_H)
|
---|
1773 | #include <fcntl.h>
|
---|
1774 | #endif
|
---|
1775 |
|
---|
1776 | #if defined(HAVE_SYS_MMAN_H)
|
---|
1777 | #include <sys/mman.h>
|
---|
1778 | #endif
|
---|
1779 |
|
---|
1780 | #if !defined(MAP_ANONYMOUS) && defined(MAP_ANON)
|
---|
1781 | #define MAP_ANONYMOUS MAP_ANON
|
---|
1782 | #endif
|
---|
1783 |
|
---|
1784 | /*
|
---|
1785 | Nearly all versions of mmap support MAP_ANONYMOUS,
|
---|
1786 | so the following is unlikely to be needed, but is
|
---|
1787 | supplied just in case.
|
---|
1788 | */
|
---|
1789 |
|
---|
1790 | #ifndef MAP_ANONYMOUS
|
---|
1791 |
|
---|
1792 | /* rw 19.05.2008 changed to avoid cached file descriptor, untested
|
---|
1793 | */
|
---|
1794 | void * anon_mmap (void *addr, size_t length, int prot, int flags)
|
---|
1795 | {
|
---|
1796 | void * retval = NULL;
|
---|
1797 | int dev_zero_fd = -1; /* File descriptor for /dev/zero. */
|
---|
1798 |
|
---|
1799 | dev_zero_fd = open("/dev/zero", O_RDWR);
|
---|
1800 | if (dev_zero_fd >= 0)
|
---|
1801 | {
|
---|
1802 | retval = mmap((addr), (size), (prot), (flags), dev_zero_fd, 0);
|
---|
1803 | /* closing the file descriptor does not unmap the region */
|
---|
1804 | close(dev_zero_fd);
|
---|
1805 | }
|
---|
1806 | return retval;
|
---|
1807 | }
|
---|
1808 |
|
---|
1809 | #define MMAP(addr, size, prot, flags) \
|
---|
1810 | (anon_mmap((addr), (size), (prot), (flags)))
|
---|
1811 |
|
---|
1812 |
|
---|
1813 | #else /* have MAP_ANONYMOUS */
|
---|
1814 |
|
---|
1815 | #if !defined(MAP_32BIT) && defined(MAP_ADDR32)
|
---|
1816 | #define MAP_32BIT MAP_ADDR32
|
---|
1817 | #endif
|
---|
1818 |
|
---|
1819 | #if defined(MAP_32BIT)
|
---|
1820 | #define MMAP(addr, size, prot, flags) \
|
---|
1821 | (mmap((addr), (size), (prot), (flags)|MAP_ANONYMOUS|MAP_32BIT, -1, 0))
|
---|
1822 | #elif defined(__sun)
|
---|
1823 | /*
|
---|
1824 | * Hint an address within 32bit address space
|
---|
1825 | */
|
---|
1826 | #define MMAP(addr, size, prot, flags) \
|
---|
1827 | (mmap((void*)0xC0000000, (size), (prot), (flags)|MAP_ANONYMOUS, -1, 0))
|
---|
1828 | #else
|
---|
1829 | /* *BSD */
|
---|
1830 | #define MMAP(addr, size, prot, flags) \
|
---|
1831 | (mmap((void*)0x80000000, (size), (prot), (flags)|MAP_ANONYMOUS, -1, 0))
|
---|
1832 | #endif
|
---|
1833 |
|
---|
1834 | #endif /* have MAP_ANONYMOUS */
|
---|
1835 |
|
---|
1836 |
|
---|
1837 | /*
|
---|
1838 | ----------------------- Chunk representations -----------------------
|
---|
1839 | */
|
---|
1840 |
|
---|
1841 | typedef void * mchunkptr;
|
---|
1842 |
|
---|
1843 | struct chunkinfo {
|
---|
1844 | INTERNAL_SIZE_T prev_size; /* Size of previous in bytes */
|
---|
1845 | INTERNAL_SIZE_T size; /* Size in bytes, including overhead. */
|
---|
1846 | INTERNAL_SIZE_T req; /* Original request size, for guard. */
|
---|
1847 | struct chunkinfo* hash_next; /* contains a pointer to the next chunk
|
---|
1848 | in the linked list if the hash
|
---|
1849 | value is the same as the chunk */
|
---|
1850 | struct chunkinfo* fd; /* double links -- used only if free. */
|
---|
1851 | struct chunkinfo* bk;
|
---|
1852 | mchunkptr chunk;
|
---|
1853 | };
|
---|
1854 |
|
---|
1855 | typedef struct chunkinfo* chunkinfoptr;
|
---|
1856 |
|
---|
1857 | struct cireginfo {
|
---|
1858 | unsigned long position;
|
---|
1859 | unsigned long *freebitmap;
|
---|
1860 | struct cireginfo* next;
|
---|
1861 | struct chunkinfo *freelist;
|
---|
1862 | struct chunkinfo *begin;
|
---|
1863 | unsigned long freecounter;
|
---|
1864 | };
|
---|
1865 |
|
---|
1866 | /*
|
---|
1867 | ---------- Size and alignment checks and conversions ----------
|
---|
1868 | */
|
---|
1869 |
|
---|
1870 | /* conversion from malloc headers to user pointers, and back */
|
---|
1871 | #define chunk(p) (p->chunk)
|
---|
1872 |
|
---|
1873 |
|
---|
1874 | #define chunk2mem(p) (chunk(p))
|
---|
1875 | #define mem2chunk(mem) (hashtable_lookup(mem))
|
---|
1876 |
|
---|
1877 | /* The smallest possible chunk */
|
---|
1878 | /* #define MIN_CHUNK_SIZE 16 */
|
---|
1879 | #if (SIZEOF_UNSIGNED_LONG == 8) || defined(__arch64__) || defined(__ia64__) || defined(__x86_64__) || defined(__LP64__) || defined(__64BIT__) || defined(_LP64) || defined(_M_IA64) || (defined(_MIPS_SZLONG) && (_MIPS_SZLONG == 64))
|
---|
1880 | # define MIN_CHUNK_SIZE 32
|
---|
1881 | #else
|
---|
1882 | # define MIN_CHUNK_SIZE 16
|
---|
1883 | #endif
|
---|
1884 |
|
---|
1885 | /* The smallest size we can malloc is an aligned minimal chunk */
|
---|
1886 |
|
---|
1887 | #define MINSIZE \
|
---|
1888 | (CHUNK_SIZE_T)(((MIN_CHUNK_SIZE+MALLOC_ALIGN_MASK) & ~MALLOC_ALIGN_MASK))
|
---|
1889 |
|
---|
1890 | /* Check if m has acceptable alignment */
|
---|
1891 |
|
---|
1892 | #define aligned_OK(m) (((PTR_UINT)((m)) & (MALLOC_ALIGN_MASK)) == 0)
|
---|
1893 |
|
---|
1894 | #define GUARD_SIZE 4
|
---|
1895 |
|
---|
1896 | /*
|
---|
1897 | Check if a request is so large that it would wrap around zero when
|
---|
1898 | padded and aligned. To simplify some other code, the bound is made
|
---|
1899 | low enough so that adding MINSIZE will also not wrap around zero.
|
---|
1900 |
|
---|
1901 | Make it 4*MINSIZE.
|
---|
1902 | */
|
---|
1903 |
|
---|
1904 | #define REQUEST_OUT_OF_RANGE(req) \
|
---|
1905 | ((CHUNK_SIZE_T)(req) >= \
|
---|
1906 | (CHUNK_SIZE_T)(INTERNAL_SIZE_T)(-4 * MINSIZE))
|
---|
1907 |
|
---|
1908 | /* pad request bytes into a usable size -- internal version */
|
---|
1909 |
|
---|
1910 | #define request2size(req) \
|
---|
1911 | (((req) + GUARD_SIZE + MALLOC_ALIGN_MASK >= MINSIZE) ? \
|
---|
1912 | ((req) + GUARD_SIZE + MALLOC_ALIGN_MASK) & ~MALLOC_ALIGN_MASK :\
|
---|
1913 | MINSIZE)
|
---|
1914 |
|
---|
1915 | /* Same, except also perform argument check */
|
---|
1916 |
|
---|
1917 | #define checked_request2size(req, sz) \
|
---|
1918 | if (!REQUEST_OUT_OF_RANGE(req)) { \
|
---|
1919 | (sz) = request2size(req); \
|
---|
1920 | assert((sz-req) >= GUARD_SIZE); \
|
---|
1921 | } else { \
|
---|
1922 | MALLOC_FAILURE_ACTION; \
|
---|
1923 | return 0; \
|
---|
1924 | }
|
---|
1925 |
|
---|
1926 | #if PARANOIA > 2
|
---|
1927 | static char * guard_set_p;
|
---|
1928 | static char * guard_set_q;
|
---|
1929 |
|
---|
1930 | #define guard_set(guard, P, request, sz) \
|
---|
1931 | assert((sz-request) >= GUARD_SIZE); \
|
---|
1932 | guard_set_p = (char*)(chunk(P)); \
|
---|
1933 | guard_set_p += request; \
|
---|
1934 | VALGRIND_MAKE_MEM_UNDEFINED(guard_set_p,GUARD_SIZE); \
|
---|
1935 | guard_set_q = (char*)(guard); \
|
---|
1936 | *guard_set_p = *guard_set_q; ++guard_set_p; ++guard_set_q; \
|
---|
1937 | *guard_set_p = *guard_set_q; ++guard_set_p; ++guard_set_q; \
|
---|
1938 | *guard_set_p = *guard_set_q; ++guard_set_p; ++guard_set_q; \
|
---|
1939 | *guard_set_p = *guard_set_q; \
|
---|
1940 | VALGRIND_MAKE_MEM_NOACCESS((((char*)chunk(P))+request),GUARD_SIZE); \
|
---|
1941 | (P)->req = request
|
---|
1942 |
|
---|
1943 | #define guard_check(guard, P) \
|
---|
1944 | VALGRIND_MAKE_MEM_DEFINED((((char *)chunk(P))+(P)->req), GUARD_SIZE); \
|
---|
1945 | assert(0 == memcmp((((char *)chunk(P))+(P)->req),(void*)(guard),GUARD_SIZE));\
|
---|
1946 | VALGRIND_MAKE_MEM_NOACCESS((((char *)chunk(P))+(P)->req), GUARD_SIZE);
|
---|
1947 |
|
---|
1948 | #else
|
---|
1949 | #define guard_set(guard, P, request, sz) ((void)0)
|
---|
1950 | #define guard_check(guard, P) ((void)0)
|
---|
1951 | #endif /* PARANOIA > 2 */
|
---|
1952 |
|
---|
1953 | /* dnmalloc forward declarations */
|
---|
1954 | static char * dnmalloc_arc4random(void);
|
---|
1955 | static void dnmalloc_init (void);
|
---|
1956 | static void malloc_mmap_state(void);
|
---|
1957 | static void cireg_extend (void);
|
---|
1958 | static chunkinfoptr cireg_getfree (void);
|
---|
1959 | static void hashtable_add (chunkinfoptr ci);
|
---|
1960 | static void hashtable_insert (chunkinfoptr ci_orig, chunkinfoptr ci_insert);
|
---|
1961 | static void hashtable_remove (mchunkptr p);
|
---|
1962 | static void hashtable_skiprm (chunkinfoptr ci_orig, chunkinfoptr ci_todelete);
|
---|
1963 | static chunkinfoptr hashtable_lookup (mchunkptr p);
|
---|
1964 | static chunkinfoptr next_chunkinfo (chunkinfoptr ci);
|
---|
1965 | static chunkinfoptr prev_chunkinfo (chunkinfoptr ci);
|
---|
1966 |
|
---|
1967 |
|
---|
1968 |
|
---|
1969 | /*
|
---|
1970 | --------------- Physical chunk operations ---------------
|
---|
1971 | */
|
---|
1972 |
|
---|
1973 |
|
---|
1974 | /* size field is or'ed with PREV_INUSE when previous adjacent chunk in use */
|
---|
1975 | #define PREV_INUSE 0x1
|
---|
1976 |
|
---|
1977 | /* extract inuse bit of previous chunk */
|
---|
1978 | #define prev_inuse(p) ((p)->size & PREV_INUSE)
|
---|
1979 |
|
---|
1980 | /* size field is or'ed with IS_MMAPPED if the chunk was obtained with mmap() */
|
---|
1981 | #define IS_MMAPPED 0x2
|
---|
1982 |
|
---|
1983 | /* check for mmap()'ed chunk */
|
---|
1984 | #define chunk_is_mmapped(p) ((p)->size & IS_MMAPPED)
|
---|
1985 |
|
---|
1986 |
|
---|
1987 | /* size field is or'ed when the chunk is in use */
|
---|
1988 | #define INUSE 0x4
|
---|
1989 |
|
---|
1990 | /* extract inuse bit of chunk */
|
---|
1991 | #define inuse(p) ((p)->size & INUSE)
|
---|
1992 |
|
---|
1993 | /*
|
---|
1994 | Bits to mask off when extracting size
|
---|
1995 |
|
---|
1996 | Note: IS_MMAPPED is intentionally not masked off from size field in
|
---|
1997 | macros for which mmapped chunks should never be seen. This should
|
---|
1998 | cause helpful core dumps to occur if it is tried by accident by
|
---|
1999 | people extending or adapting this malloc.
|
---|
2000 | */
|
---|
2001 | #define SIZE_BITS (PREV_INUSE|IS_MMAPPED|INUSE)
|
---|
2002 |
|
---|
2003 | /* Bits to mask off when extracting size of chunks for macros which do not use mmap */
|
---|
2004 | #define SIZE_NOMMAP (PREV_INUSE|INUSE)
|
---|
2005 |
|
---|
2006 | /* Get size, ignoring use bits */
|
---|
2007 | #define chunksize(p) ((p)->size & ~(SIZE_BITS))
|
---|
2008 |
|
---|
2009 | /* Ptr to chunkinfo of next physical malloc_chunk. */
|
---|
2010 | #define next_chunk(p) ((mchunkptr)( ((char*)(p)) + ((p)->size & SIZE_NOMMAP) ))
|
---|
2011 |
|
---|
2012 | /* Treat space at ptr + offset as a chunk */
|
---|
2013 | #define chunk_at_offset(p, s) ((mchunkptr)(((char*)(p)) + (s)))
|
---|
2014 |
|
---|
2015 | /* set/clear chunk as being inuse without otherwise disturbing */
|
---|
2016 | #define set_inuse(p) ((p)->size |= INUSE)
|
---|
2017 |
|
---|
2018 | #define clear_inuse(p) ((p)->size &= ~(INUSE))
|
---|
2019 |
|
---|
2020 | #define set_previnuse(p) ((p)->size |= PREV_INUSE)
|
---|
2021 |
|
---|
2022 | #define clear_previnuse(p) ((p)->size &= ~(PREV_INUSE))
|
---|
2023 |
|
---|
2024 | static void set_previnuse_next (chunkinfoptr p)
|
---|
2025 | {
|
---|
2026 | chunkinfoptr q;
|
---|
2027 | q = next_chunkinfo (p);
|
---|
2028 | if (q)
|
---|
2029 | set_previnuse (q);
|
---|
2030 | }
|
---|
2031 |
|
---|
2032 | #define set_all_inuse(p) \
|
---|
2033 | set_inuse(p); \
|
---|
2034 | set_previnuse_next(p);
|
---|
2035 |
|
---|
2036 |
|
---|
2037 | /* Set size at head, without disturbing its use bit */
|
---|
2038 | #define set_head_size(p, s) ((p)->size = (((p)->size & SIZE_NOMMAP) | (s)))
|
---|
2039 |
|
---|
2040 | /* Set size/use field */
|
---|
2041 | #define set_head(p, s) ((p)->size = (s))
|
---|
2042 |
|
---|
2043 | /*
|
---|
2044 | Bins
|
---|
2045 |
|
---|
2046 | An array of bin headers for free chunks. Each bin is doubly
|
---|
2047 | linked. The bins are approximately proportionally (log) spaced.
|
---|
2048 | There are a lot of these bins (128). This may look excessive, but
|
---|
2049 | works very well in practice. Most bins hold sizes that are
|
---|
2050 | unusual as malloc request sizes, but are more usual for fragments
|
---|
2051 | and consolidated sets of chunks, which is what these bins hold, so
|
---|
2052 | they can be found quickly. All procedures maintain the invariant
|
---|
2053 | that no consolidated chunk physically borders another one, so each
|
---|
2054 | chunk in a list is known to be preceeded and followed by either
|
---|
2055 | inuse chunks or the ends of memory.
|
---|
2056 |
|
---|
2057 | Chunks in bins are kept in size order, with ties going to the
|
---|
2058 | approximately least recently used chunk. Ordering isn't needed
|
---|
2059 | for the small bins, which all contain the same-sized chunks, but
|
---|
2060 | facilitates best-fit allocation for larger chunks. These lists
|
---|
2061 | are just sequential. Keeping them in order almost never requires
|
---|
2062 | enough traversal to warrant using fancier ordered data
|
---|
2063 | structures.
|
---|
2064 |
|
---|
2065 | Chunks of the same size are linked with the most
|
---|
2066 | recently freed at the front, and allocations are taken from the
|
---|
2067 | back. This results in LRU (FIFO) allocation order, which tends
|
---|
2068 | to give each chunk an equal opportunity to be consolidated with
|
---|
2069 | adjacent freed chunks, resulting in larger free chunks and less
|
---|
2070 | fragmentation.
|
---|
2071 |
|
---|
2072 | To simplify use in double-linked lists, each bin header acts
|
---|
2073 | as a malloc_chunk. This avoids special-casing for headers.
|
---|
2074 | But to conserve space and improve locality, we allocate
|
---|
2075 | only the fd/bk pointers of bins, and then use repositioning tricks
|
---|
2076 | to treat these as the fields of a malloc_chunk*.
|
---|
2077 | */
|
---|
2078 |
|
---|
2079 | typedef struct chunkinfo* mbinptr;
|
---|
2080 |
|
---|
2081 | /* addressing -- note that bin_at(0) does not exist */
|
---|
2082 | #define bin_at(m, i) (&(m)->bins[i])
|
---|
2083 |
|
---|
2084 | /* analog of ++bin */
|
---|
2085 | #define next_bin(b) (b+1)
|
---|
2086 |
|
---|
2087 | /* Reminders about list directionality within bins */
|
---|
2088 | #define first(b) ((b)->fd)
|
---|
2089 | #define last(b) ((b)->bk)
|
---|
2090 |
|
---|
2091 | /* Take a chunk off a bin list */
|
---|
2092 | #define unlink(P, BK, FD) { \
|
---|
2093 | FD = P->fd; \
|
---|
2094 | BK = P->bk; \
|
---|
2095 | FD->bk = BK; \
|
---|
2096 | BK->fd = FD; \
|
---|
2097 | }
|
---|
2098 |
|
---|
2099 | /*
|
---|
2100 | Indexing
|
---|
2101 |
|
---|
2102 | Bins for sizes < 512 bytes contain chunks of all the same size, spaced
|
---|
2103 | 8 bytes apart. Larger bins are approximately logarithmically spaced:
|
---|
2104 |
|
---|
2105 | 64 bins of size 8
|
---|
2106 | 32 bins of size 64
|
---|
2107 | 16 bins of size 512
|
---|
2108 | 8 bins of size 4096
|
---|
2109 | 4 bins of size 32768
|
---|
2110 | 2 bins of size 262144
|
---|
2111 | 1 bin of size what's left
|
---|
2112 |
|
---|
2113 | The bins top out around 1MB because we expect to service large
|
---|
2114 | requests via mmap.
|
---|
2115 | */
|
---|
2116 |
|
---|
2117 | #define NBINS 96
|
---|
2118 | #define NSMALLBINS 32
|
---|
2119 | #define SMALLBIN_WIDTH 8
|
---|
2120 | #define MIN_LARGE_SIZE 256
|
---|
2121 |
|
---|
2122 | #define in_smallbin_range(sz) \
|
---|
2123 | ((CHUNK_SIZE_T)(sz) < (CHUNK_SIZE_T)MIN_LARGE_SIZE)
|
---|
2124 |
|
---|
2125 | #define smallbin_index(sz) (((unsigned)(sz)) >> 3)
|
---|
2126 |
|
---|
2127 | /*
|
---|
2128 | Compute index for size. We expect this to be inlined when
|
---|
2129 | compiled with optimization, else not, which works out well.
|
---|
2130 | */
|
---|
2131 | static int largebin_index(size_t sz) {
|
---|
2132 |
|
---|
2133 | unsigned long xx = sz >> SMALLBIN_WIDTH;
|
---|
2134 |
|
---|
2135 | if (xx < 0x10000)
|
---|
2136 | {
|
---|
2137 | unsigned int m; /* bit position of highest set bit of m */
|
---|
2138 |
|
---|
2139 | /* On intel, use BSRL instruction to find highest bit */
|
---|
2140 | #if defined(__GNUC__) && defined(i386) && !defined(USE_UNO)
|
---|
2141 |
|
---|
2142 | unsigned int x = (unsigned int) xx;
|
---|
2143 |
|
---|
2144 | __asm__("bsrl %1,%0\n\t"
|
---|
2145 | : "=r" (m)
|
---|
2146 | : "rm" (x));
|
---|
2147 |
|
---|
2148 | #elif defined(__GNUC__) && defined(x86_64) && !defined(USE_UNO)
|
---|
2149 |
|
---|
2150 | __asm__("bsrq %1,%0\n\t"
|
---|
2151 | : "=r" (m)
|
---|
2152 | : "rm" (xx));
|
---|
2153 |
|
---|
2154 | #else
|
---|
2155 |
|
---|
2156 | /* Taken from Bit Twiddling Hacks
|
---|
2157 | * http://graphics.stanford.edu/~seander/bithacks.html
|
---|
2158 | * public domain
|
---|
2159 | */
|
---|
2160 | unsigned int v = (unsigned int) xx;
|
---|
2161 | register unsigned int shift;
|
---|
2162 |
|
---|
2163 | m = (v > 0xFFFF) << 4; v >>= m;
|
---|
2164 | shift = (v > 0xFF ) << 3; v >>= shift; m |= shift;
|
---|
2165 | shift = (v > 0xF ) << 2; v >>= shift; m |= shift;
|
---|
2166 | shift = (v > 0x3 ) << 1; v >>= shift; m |= shift;
|
---|
2167 | m |= (v >> 1);
|
---|
2168 |
|
---|
2169 | #endif
|
---|
2170 |
|
---|
2171 | /* Use next 2 bits to create finer-granularity bins */
|
---|
2172 | return NSMALLBINS + (m << 2) + ((sz >> (m + 6)) & 3);
|
---|
2173 | }
|
---|
2174 | else
|
---|
2175 | {
|
---|
2176 | return NBINS-1;
|
---|
2177 | }
|
---|
2178 | }
|
---|
2179 |
|
---|
2180 | #define bin_index(sz) \
|
---|
2181 | ((in_smallbin_range(sz)) ? smallbin_index(sz) : largebin_index(sz))
|
---|
2182 |
|
---|
2183 | /*
|
---|
2184 | FIRST_SORTED_BIN_SIZE is the chunk size corresponding to the
|
---|
2185 | first bin that is maintained in sorted order. This must
|
---|
2186 | be the smallest size corresponding to a given bin.
|
---|
2187 |
|
---|
2188 | Normally, this should be MIN_LARGE_SIZE. But you can weaken
|
---|
2189 | best fit guarantees to sometimes speed up malloc by increasing value.
|
---|
2190 | Doing this means that malloc may choose a chunk that is
|
---|
2191 | non-best-fitting by up to the width of the bin.
|
---|
2192 |
|
---|
2193 | Some useful cutoff values:
|
---|
2194 | 512 - all bins sorted
|
---|
2195 | 2560 - leaves bins <= 64 bytes wide unsorted
|
---|
2196 | 12288 - leaves bins <= 512 bytes wide unsorted
|
---|
2197 | 65536 - leaves bins <= 4096 bytes wide unsorted
|
---|
2198 | 262144 - leaves bins <= 32768 bytes wide unsorted
|
---|
2199 | -1 - no bins sorted (not recommended!)
|
---|
2200 | */
|
---|
2201 |
|
---|
2202 | /* #define FIRST_SORTED_BIN_SIZE 65536 */
|
---|
2203 |
|
---|
2204 | #define FIRST_SORTED_BIN_SIZE MIN_LARGE_SIZE
|
---|
2205 |
|
---|
2206 |
|
---|
2207 | /*
|
---|
2208 | Unsorted chunks
|
---|
2209 |
|
---|
2210 | All remainders from chunk splits, as well as all returned chunks,
|
---|
2211 | are first placed in the "unsorted" bin. They are then placed
|
---|
2212 | in regular bins after malloc gives them ONE chance to be used before
|
---|
2213 | binning. So, basically, the unsorted_chunks list acts as a queue,
|
---|
2214 | with chunks being placed on it in free (and malloc_consolidate),
|
---|
2215 | and taken off (to be either used or placed in bins) in malloc.
|
---|
2216 | */
|
---|
2217 |
|
---|
2218 | /* The otherwise unindexable 1-bin is used to hold unsorted chunks. */
|
---|
2219 | #define unsorted_chunks(M) (bin_at(M, 1))
|
---|
2220 |
|
---|
2221 | /*
|
---|
2222 | Top
|
---|
2223 |
|
---|
2224 | The top-most available chunk (i.e., the one bordering the end of
|
---|
2225 | available memory) is treated specially. It is never included in
|
---|
2226 | any bin, is used only if no other chunk is available, and is
|
---|
2227 | released back to the system if it is very large (see
|
---|
2228 | M_TRIM_THRESHOLD). Because top initially
|
---|
2229 | points to its own bin with initial zero size, thus forcing
|
---|
2230 | extension on the first malloc request, we avoid having any special
|
---|
2231 | code in malloc to check whether it even exists yet. But we still
|
---|
2232 | need to do so when getting memory from system, so we make
|
---|
2233 | initial_top treat the bin as a legal but unusable chunk during the
|
---|
2234 | interval between initialization and the first call to
|
---|
2235 | sYSMALLOc. (This is somewhat delicate, since it relies on
|
---|
2236 | the 2 preceding words to be zero during this interval as well.)
|
---|
2237 | */
|
---|
2238 |
|
---|
2239 | /* Conveniently, the unsorted bin can be used as dummy top on first call */
|
---|
2240 | #define initial_top(M) (unsorted_chunks(M))
|
---|
2241 |
|
---|
2242 | /*
|
---|
2243 | Binmap
|
---|
2244 |
|
---|
2245 | To help compensate for the large number of bins, a one-level index
|
---|
2246 | structure is used for bin-by-bin searching. `binmap' is a
|
---|
2247 | bitvector recording whether bins are definitely empty so they can
|
---|
2248 | be skipped over during during traversals. The bits are NOT always
|
---|
2249 | cleared as soon as bins are empty, but instead only
|
---|
2250 | when they are noticed to be empty during traversal in malloc.
|
---|
2251 | */
|
---|
2252 |
|
---|
2253 | /* Conservatively use 32 bits per map word, even if on 64bit system */
|
---|
2254 | #define BINMAPSHIFT 5
|
---|
2255 | #define BITSPERMAP (1U << BINMAPSHIFT)
|
---|
2256 | #define BINMAPSIZE (NBINS / BITSPERMAP)
|
---|
2257 |
|
---|
2258 | #define idx2block(i) ((i) >> BINMAPSHIFT)
|
---|
2259 | #define idx2bit(i) ((1U << ((i) & ((1U << BINMAPSHIFT)-1))))
|
---|
2260 |
|
---|
2261 | #define mark_bin(m,i) ((m)->binmap[idx2block(i)] |= idx2bit(i))
|
---|
2262 | #define unmark_bin(m,i) ((m)->binmap[idx2block(i)] &= ~(idx2bit(i)))
|
---|
2263 | #define get_binmap(m,i) ((m)->binmap[idx2block(i)] & idx2bit(i))
|
---|
2264 |
|
---|
2265 | /*
|
---|
2266 | Fastbins
|
---|
2267 |
|
---|
2268 | An array of lists holding recently freed small chunks. Fastbins
|
---|
2269 | are not doubly linked. It is faster to single-link them, and
|
---|
2270 | since chunks are never removed from the middles of these lists,
|
---|
2271 | double linking is not necessary. Also, unlike regular bins, they
|
---|
2272 | are not even processed in FIFO order (they use faster LIFO) since
|
---|
2273 | ordering doesn't much matter in the transient contexts in which
|
---|
2274 | fastbins are normally used.
|
---|
2275 |
|
---|
2276 | Chunks in fastbins keep their inuse bit set, so they cannot
|
---|
2277 | be consolidated with other free chunks. malloc_consolidate
|
---|
2278 | releases all chunks in fastbins and consolidates them with
|
---|
2279 | other free chunks.
|
---|
2280 | */
|
---|
2281 |
|
---|
2282 | typedef struct chunkinfo* mfastbinptr;
|
---|
2283 |
|
---|
2284 | /* offset 2 to use otherwise unindexable first 2 bins */
|
---|
2285 | #define fastbin_index(sz) ((((unsigned int)(sz)) >> 3) - 2)
|
---|
2286 |
|
---|
2287 | /* The maximum fastbin request size we support */
|
---|
2288 | #define MAX_FAST_SIZE 80
|
---|
2289 |
|
---|
2290 | #define NFASTBINS (fastbin_index(request2size(MAX_FAST_SIZE))+1)
|
---|
2291 |
|
---|
2292 | /*
|
---|
2293 | FASTBIN_CONSOLIDATION_THRESHOLD is the size of a chunk in free()
|
---|
2294 | that triggers automatic consolidation of possibly-surrounding
|
---|
2295 | fastbin chunks. This is a heuristic, so the exact value should not
|
---|
2296 | matter too much. It is defined at half the default trim threshold as a
|
---|
2297 | compromise heuristic to only attempt consolidation if it is likely
|
---|
2298 | to lead to trimming. However, it is not dynamically tunable, since
|
---|
2299 | consolidation reduces fragmentation surrounding loarge chunks even
|
---|
2300 | if trimming is not used.
|
---|
2301 | */
|
---|
2302 |
|
---|
2303 | #define FASTBIN_CONSOLIDATION_THRESHOLD \
|
---|
2304 | ((unsigned long)(DEFAULT_TRIM_THRESHOLD) >> 1)
|
---|
2305 |
|
---|
2306 | /*
|
---|
2307 | Since the lowest 2 bits in max_fast don't matter in size comparisons,
|
---|
2308 | they are used as flags.
|
---|
2309 | */
|
---|
2310 |
|
---|
2311 | /*
|
---|
2312 | ANYCHUNKS_BIT held in max_fast indicates that there may be any
|
---|
2313 | freed chunks at all. It is set true when entering a chunk into any
|
---|
2314 | bin.
|
---|
2315 | */
|
---|
2316 |
|
---|
2317 | #define ANYCHUNKS_BIT (1U)
|
---|
2318 |
|
---|
2319 | #define have_anychunks(M) (((M)->max_fast & ANYCHUNKS_BIT))
|
---|
2320 | #define set_anychunks(M) ((M)->max_fast |= ANYCHUNKS_BIT)
|
---|
2321 | #define clear_anychunks(M) ((M)->max_fast &= ~ANYCHUNKS_BIT)
|
---|
2322 |
|
---|
2323 | /*
|
---|
2324 | FASTCHUNKS_BIT held in max_fast indicates that there are probably
|
---|
2325 | some fastbin chunks. It is set true on entering a chunk into any
|
---|
2326 | fastbin, and cleared only in malloc_consolidate.
|
---|
2327 | */
|
---|
2328 |
|
---|
2329 | #define FASTCHUNKS_BIT (2U)
|
---|
2330 |
|
---|
2331 | #define have_fastchunks(M) (((M)->max_fast & FASTCHUNKS_BIT))
|
---|
2332 | #define set_fastchunks(M) ((M)->max_fast |= (FASTCHUNKS_BIT|ANYCHUNKS_BIT))
|
---|
2333 | #define clear_fastchunks(M) ((M)->max_fast &= ~(FASTCHUNKS_BIT))
|
---|
2334 |
|
---|
2335 | /*
|
---|
2336 | Set value of max_fast.
|
---|
2337 | Use impossibly small value if 0.
|
---|
2338 | */
|
---|
2339 |
|
---|
2340 | #define set_max_fast(M, s) \
|
---|
2341 | (M)->max_fast = (((s) == 0)? SMALLBIN_WIDTH: request2size(s)) | \
|
---|
2342 | ((M)->max_fast & (FASTCHUNKS_BIT|ANYCHUNKS_BIT))
|
---|
2343 |
|
---|
2344 | #define get_max_fast(M) \
|
---|
2345 | ((M)->max_fast & ~(FASTCHUNKS_BIT | ANYCHUNKS_BIT))
|
---|
2346 |
|
---|
2347 |
|
---|
2348 | /*
|
---|
2349 | morecore_properties is a status word holding dynamically discovered
|
---|
2350 | or controlled properties of the morecore function
|
---|
2351 | */
|
---|
2352 |
|
---|
2353 | #define MORECORE_CONTIGUOUS_BIT (1U)
|
---|
2354 |
|
---|
2355 | #define contiguous(M) \
|
---|
2356 | (((M)->morecore_properties & MORECORE_CONTIGUOUS_BIT))
|
---|
2357 | #define noncontiguous(M) \
|
---|
2358 | (((M)->morecore_properties & MORECORE_CONTIGUOUS_BIT) == 0)
|
---|
2359 | #define set_contiguous(M) \
|
---|
2360 | ((M)->morecore_properties |= MORECORE_CONTIGUOUS_BIT)
|
---|
2361 | #define set_noncontiguous(M) \
|
---|
2362 | ((M)->morecore_properties &= ~MORECORE_CONTIGUOUS_BIT)
|
---|
2363 |
|
---|
2364 | #define MORECORE_32BIT_BIT (2U)
|
---|
2365 |
|
---|
2366 | #define morecore32bit(M) \
|
---|
2367 | (((M)->morecore_properties & MORECORE_32BIT_BIT))
|
---|
2368 | #define nonmorecore32bit(M) \
|
---|
2369 | (((M)->morecore_properties & MORECORE_32BIT_BIT) == 0)
|
---|
2370 | #define set_morecore32bit(M) \
|
---|
2371 | ((M)->morecore_properties |= MORECORE_32BIT_BIT)
|
---|
2372 | #define set_nonmorecore32bit(M) \
|
---|
2373 | ((M)->morecore_properties &= ~MORECORE_32BIT_BIT)
|
---|
2374 |
|
---|
2375 |
|
---|
2376 |
|
---|
2377 | /* ----------------- dnmalloc -------------------- */
|
---|
2378 |
|
---|
2379 | /* size of pages */
|
---|
2380 | #define PGSIZE malloc_getpagesize
|
---|
2381 | /* pointer size */
|
---|
2382 | #define PTRSIZE sizeof(long)
|
---|
2383 |
|
---|
2384 |
|
---|
2385 |
|
---|
2386 | /* TODO: mmapped chunks are always multiples of pagesize -> we're wasting
|
---|
2387 | address space: the hashtable has granularity of 16*8, set it to something
|
---|
2388 | closer to pagesize for mmapped chunks (current waste: 32 positions/mmapped
|
---|
2389 | page)
|
---|
2390 | */
|
---|
2391 |
|
---|
2392 | /* The maximum heap size that dnmalloc can operate with
|
---|
2393 | * represented in hex to avoid annoying gcc warning
|
---|
2394 | *
|
---|
2395 | * Avoid integer overflow, cover complete 32bit address
|
---|
2396 | * space for portability. With deferred allocation, the
|
---|
2397 | * hashtable size is a non-issue.
|
---|
2398 | */
|
---|
2399 | #define HEAPMAXSIZE_HALF 0x80000000UL
|
---|
2400 |
|
---|
2401 | /* How many elements are stored in the linked list */
|
---|
2402 | #define LINKEDLSTELS 8
|
---|
2403 |
|
---|
2404 | /* Minimum size of a chunk */
|
---|
2405 |
|
---|
2406 | #if (SIZEOF_UNSIGNED_LONG == 8) || defined(__arch64__) || defined(__ia64__) || defined(__x86_64__) || defined(__LP64__) || defined(__64BIT__) || defined(_LP64) || defined(_M_IA64) || (defined(_MIPS_SZLONG) && (_MIPS_SZLONG == 64))
|
---|
2407 | # define MINCHUNKSIZE 32
|
---|
2408 | #else
|
---|
2409 | # define MINCHUNKSIZE 16
|
---|
2410 | #endif
|
---|
2411 |
|
---|
2412 |
|
---|
2413 | /* The amount of hashtable entries for each page:
|
---|
2414 | Pagesize divded by the numer of elements in the linkedlists
|
---|
2415 | divided by the minimum chunk size
|
---|
2416 | */
|
---|
2417 | #define CHUNKINFOPAGE (PGSIZE / LINKEDLSTELS / MINCHUNKSIZE)
|
---|
2418 |
|
---|
2419 | /* The amount of hashtable entries needed to manage the memory:
|
---|
2420 | Maximum heap size divided by page size multiplied by the amount
|
---|
2421 | of chunk info's per page
|
---|
2422 | */
|
---|
2423 | #define AMOUNTHASH ((HEAPMAXSIZE_HALF / PGSIZE) * CHUNKINFOPAGE * 2)
|
---|
2424 |
|
---|
2425 | /* Initial size of the map for the hashtable
|
---|
2426 | Amount of entries muliplied by pointer size
|
---|
2427 | */
|
---|
2428 | #define HASHTABLESIZE (AMOUNTHASH * PTRSIZE)
|
---|
2429 |
|
---|
2430 | /* Amount of free chunks that the system should allocate at the start */
|
---|
2431 | #define NUMBER_FREE_CHUNKS 32768
|
---|
2432 |
|
---|
2433 | /* Initial size of the chunk info region,
|
---|
2434 | also used when growing the region */
|
---|
2435 | #define CIREGSIZE (NUMBER_FREE_CHUNKS * sizeof(struct chunkinfo))
|
---|
2436 |
|
---|
2437 | /* Start address of the heap */
|
---|
2438 | char *startheap;
|
---|
2439 |
|
---|
2440 | /* pointer to the hashtable: struct chunkinfo **hashtable -> *hashtable[] */
|
---|
2441 | chunkinfoptr *hashtable;
|
---|
2442 |
|
---|
2443 | /* Current chunkinfo region */
|
---|
2444 | struct cireginfo *currciinfo = 0;
|
---|
2445 | struct cireginfo *firstciinfo = 0;
|
---|
2446 |
|
---|
2447 | unsigned long totalcictr = 0;
|
---|
2448 |
|
---|
2449 |
|
---|
2450 | /* Initialize the area for chunkinfos and the hashtable and protect
|
---|
2451 | * it with non-writable pages
|
---|
2452 | */
|
---|
2453 | static void
|
---|
2454 | dnmalloc_init ()
|
---|
2455 | {
|
---|
2456 | void *hashtb;
|
---|
2457 | int mprot;
|
---|
2458 | int flags = MAP_PRIVATE;
|
---|
2459 |
|
---|
2460 | /* Allocate the malloc_state struct */
|
---|
2461 | malloc_mmap_state();
|
---|
2462 |
|
---|
2463 | /* Use MAP_NORESERVE if available (Solaris, HP-UX; most other
|
---|
2464 | * systems use defered allocation anyway.
|
---|
2465 | */
|
---|
2466 | #ifdef MAP_NORESERVE
|
---|
2467 | flags |= MAP_NORESERVE;
|
---|
2468 | #endif
|
---|
2469 |
|
---|
2470 | /* Always start at 0, hashtable covers whole 32bit address space
|
---|
2471 | */
|
---|
2472 | #define STARTHEAP_IS_ZERO
|
---|
2473 | startheap = 0;
|
---|
2474 |
|
---|
2475 | /* Map space for the hashtable */
|
---|
2476 | #if PARANOIA > 1
|
---|
2477 | hashtb = MMAP(0, HASHTABLESIZE+(2*PGSIZE), PROT_READ|PROT_WRITE, flags);
|
---|
2478 | #else
|
---|
2479 | hashtb = MMAP(0, HASHTABLESIZE+PGSIZE, PROT_READ|PROT_WRITE, flags);
|
---|
2480 | #endif
|
---|
2481 |
|
---|
2482 | #ifdef NDEBUG
|
---|
2483 | if (hashtb == MAP_FAILED) {
|
---|
2484 | fprintf (stderr, "Couldn't mmap hashtable: %s\n", strerror (errno));
|
---|
2485 | abort ();
|
---|
2486 | }
|
---|
2487 | #else
|
---|
2488 | assert(hashtb != MAP_FAILED);
|
---|
2489 | #endif
|
---|
2490 |
|
---|
2491 | /* Protect the hashtable with non-writable pages */
|
---|
2492 | mprot = mprotect(hashtb, (size_t) PGSIZE, PROT_NONE);
|
---|
2493 | #ifdef NDEBUG
|
---|
2494 | if (mprot == -1) {
|
---|
2495 | fprintf (stderr, "Couldn't mprotect first non-rw page for hashtable: %s\n",
|
---|
2496 | strerror (errno));
|
---|
2497 | abort ();
|
---|
2498 | }
|
---|
2499 | #else
|
---|
2500 | assert(mprot != -1);
|
---|
2501 | #endif
|
---|
2502 |
|
---|
2503 | /* HP-UX: Cannot do arithmetic with pointers to objects of unknown size. */
|
---|
2504 | hashtable = (chunkinfoptr *) (((char*)hashtb) + PGSIZE);
|
---|
2505 |
|
---|
2506 | /* Protect the hashtable with non-writable pages */
|
---|
2507 | #if PARANOIA > 1
|
---|
2508 | mprot = mprotect((void*)((char*)hashtb+HASHTABLESIZE+PGSIZE), (size_t) PGSIZE, PROT_NONE);
|
---|
2509 | #ifdef NDEBUG
|
---|
2510 | if (mprot == -1) {
|
---|
2511 | fprintf (stderr, "Couldn't mprotect last non-rw page for hashtable: %s\n",
|
---|
2512 | strerror (errno));
|
---|
2513 | abort ();
|
---|
2514 | }
|
---|
2515 | #else
|
---|
2516 | assert(mprot != -1);
|
---|
2517 | #endif
|
---|
2518 | #endif
|
---|
2519 | }
|
---|
2520 |
|
---|
2521 |
|
---|
2522 |
|
---|
2523 | /* Extend the region for chunk infos by mapping more memory before the region */
|
---|
2524 | static void
|
---|
2525 | cireg_extend ()
|
---|
2526 | {
|
---|
2527 | void *newcireg;
|
---|
2528 | int mprot;
|
---|
2529 | struct cireginfo *tempciinfo = 0;
|
---|
2530 |
|
---|
2531 | #if PARANOIA > 1
|
---|
2532 | newcireg = MMAP(0, CIREGSIZE+(2*PGSIZE), PROT_READ|PROT_WRITE, MAP_PRIVATE);
|
---|
2533 | #else
|
---|
2534 | newcireg = MMAP(0, CIREGSIZE+PGSIZE, PROT_READ|PROT_WRITE, MAP_PRIVATE);
|
---|
2535 | #endif
|
---|
2536 |
|
---|
2537 | #ifdef NDEBUG
|
---|
2538 | if (newcireg == MAP_FAILED)
|
---|
2539 | {
|
---|
2540 | fprintf (stderr, "Couldn't extend chunkinfo region: %s\n",
|
---|
2541 | strerror (errno));
|
---|
2542 | abort ();
|
---|
2543 | }
|
---|
2544 | #else
|
---|
2545 | assert(newcireg != MAP_FAILED);
|
---|
2546 | #endif
|
---|
2547 | mprot = mprotect(newcireg, PGSIZE, PROT_NONE);
|
---|
2548 | #ifdef NDEBUG
|
---|
2549 | if (mprot == -1) {
|
---|
2550 | fprintf (stderr, "Couldn't mprotect first non-rw page for extended region: %s\n",
|
---|
2551 | strerror (errno));
|
---|
2552 | abort ();
|
---|
2553 | }
|
---|
2554 | #else
|
---|
2555 | assert(mprot != -1);
|
---|
2556 | #endif
|
---|
2557 | newcireg = ((char*)newcireg)+PGSIZE;
|
---|
2558 |
|
---|
2559 | #if PARANOIA > 1
|
---|
2560 | mprot = mprotect((void*)((char*)newcireg+CIREGSIZE), (size_t) PGSIZE, PROT_NONE);
|
---|
2561 | #ifdef NDEBUG
|
---|
2562 | if (mprot == -1) {
|
---|
2563 | fprintf (stderr, "Couldn't mprotect last non-rw page for extended region: %s\n",
|
---|
2564 | strerror (errno));
|
---|
2565 | abort ();
|
---|
2566 | }
|
---|
2567 | #else
|
---|
2568 | assert(mprot != -1);
|
---|
2569 | #endif
|
---|
2570 | #endif
|
---|
2571 |
|
---|
2572 | tempciinfo = currciinfo;
|
---|
2573 | currciinfo = (struct cireginfo *) newcireg;
|
---|
2574 | if (tempciinfo)
|
---|
2575 | tempciinfo->next = currciinfo;
|
---|
2576 | currciinfo->position = 1;
|
---|
2577 | currciinfo->freecounter = NUMBER_FREE_CHUNKS;
|
---|
2578 | if (!firstciinfo)
|
---|
2579 | firstciinfo = currciinfo;
|
---|
2580 | totalcictr++;
|
---|
2581 | VALGRIND_CREATE_MEMPOOL(newcireg, 0, 0);
|
---|
2582 | }
|
---|
2583 |
|
---|
2584 |
|
---|
2585 | /* Get a free chunkinfo */
|
---|
2586 | static chunkinfoptr
|
---|
2587 | cireg_getfree ()
|
---|
2588 | {
|
---|
2589 | chunkinfoptr freeci;
|
---|
2590 | chunkinfoptr freelst = 0;
|
---|
2591 | struct cireginfo *newciinfo = firstciinfo;
|
---|
2592 |
|
---|
2593 | if (newciinfo) {
|
---|
2594 | freelst = newciinfo->freelist;
|
---|
2595 |
|
---|
2596 | if (!freelst && newciinfo->next) {
|
---|
2597 | do {
|
---|
2598 | newciinfo = newciinfo->next;
|
---|
2599 | freelst = newciinfo->freelist;
|
---|
2600 | } while (!freelst && newciinfo->next);
|
---|
2601 | }
|
---|
2602 | }
|
---|
2603 |
|
---|
2604 | /* Check if there are any free chunkinfos on the list of free chunkinfos */
|
---|
2605 | if (freelst)
|
---|
2606 | {
|
---|
2607 | freeci = freelst;
|
---|
2608 | newciinfo->freecounter--;
|
---|
2609 | newciinfo->freelist = freelst->fd;
|
---|
2610 |
|
---|
2611 | VALGRIND_MEMPOOL_ALLOC((char*)currciinfo, (char*)freeci,
|
---|
2612 | sizeof(struct chunkinfo));
|
---|
2613 |
|
---|
2614 | freeci->prev_size = 0;
|
---|
2615 | freeci->size = 0;
|
---|
2616 | freeci->req = 0;
|
---|
2617 | freeci->hash_next = NULL;
|
---|
2618 | freeci->fd = NULL;
|
---|
2619 | freeci->bk = NULL;
|
---|
2620 | freeci->chunk = NULL;
|
---|
2621 | return (freeci);
|
---|
2622 | }
|
---|
2623 | else
|
---|
2624 | {
|
---|
2625 | /* No free chunkinfos, check if chunkinfo region still has place
|
---|
2626 | * for a chunkinfo. If not, extend the region.
|
---|
2627 | */
|
---|
2628 | if (UNLIKELY(!currciinfo || currciinfo->position == NUMBER_FREE_CHUNKS))
|
---|
2629 | cireg_extend ();
|
---|
2630 | /* Get a chunkinfo from the chunkinfo region */
|
---|
2631 | freeci = (chunkinfoptr) currciinfo + currciinfo->position;
|
---|
2632 | currciinfo->freecounter--;
|
---|
2633 | currciinfo->position++;
|
---|
2634 |
|
---|
2635 | VALGRIND_MEMPOOL_ALLOC((char*)currciinfo, (char*)freeci,
|
---|
2636 | sizeof(struct chunkinfo));
|
---|
2637 |
|
---|
2638 | return (freeci);
|
---|
2639 | }
|
---|
2640 | }
|
---|
2641 |
|
---|
2642 | static void freeciregion(struct cireginfo *freeme) {
|
---|
2643 | /* free the chunkinfo region */
|
---|
2644 | struct cireginfo *newciinfo = firstciinfo;
|
---|
2645 | struct cireginfo *prevciinfo = firstciinfo;
|
---|
2646 | void *unmapme;
|
---|
2647 |
|
---|
2648 | while (newciinfo && newciinfo != freeme) {
|
---|
2649 | prevciinfo = newciinfo;
|
---|
2650 | newciinfo = newciinfo->next;
|
---|
2651 | }
|
---|
2652 | assert(freeme == newciinfo); /* rw */
|
---|
2653 | assert(newciinfo != NULL); /* rw */
|
---|
2654 | if (newciinfo)
|
---|
2655 | prevciinfo->next = newciinfo->next;
|
---|
2656 | unmapme = (void *) ((char*)freeme - PGSIZE);
|
---|
2657 | VALGRIND_DESTROY_MEMPOOL((char*)freeme);
|
---|
2658 | #if PARANOIA > 1
|
---|
2659 | munmap(unmapme, CIREGSIZE+(2*PGSIZE));
|
---|
2660 | #else
|
---|
2661 | munmap(unmapme, CIREGSIZE+PGSIZE);
|
---|
2662 | #endif
|
---|
2663 | }
|
---|
2664 |
|
---|
2665 |
|
---|
2666 | static void freecilst_add(chunkinfoptr p) {
|
---|
2667 |
|
---|
2668 | struct cireginfo *newciinfo;
|
---|
2669 |
|
---|
2670 | newciinfo = currciinfo;
|
---|
2671 | if (((chunkinfoptr) newciinfo < p) && (p < (chunkinfoptr) (newciinfo+NUMBER_FREE_CHUNKS))) {
|
---|
2672 | p->fd = newciinfo->freelist;
|
---|
2673 | newciinfo->freelist = p;
|
---|
2674 | newciinfo->freecounter++;
|
---|
2675 | VALGRIND_MEMPOOL_FREE((char*)newciinfo, (char*)p);
|
---|
2676 | VALGRIND_MAKE_MEM_DEFINED(p,sizeof(struct chunkinfo));
|
---|
2677 | VALGRIND_MAKE_MEM_NOACCESS(p->size, sizeof(INTERNAL_SIZE_T));
|
---|
2678 | VALGRIND_MAKE_MEM_NOACCESS(p->req, sizeof(INTERNAL_SIZE_T));
|
---|
2679 | VALGRIND_MAKE_MEM_NOACCESS(p->bk, sizeof(struct chunkinfo*));
|
---|
2680 | VALGRIND_MAKE_MEM_NOACCESS(p->chunk, sizeof(mchunkptr));
|
---|
2681 | } else {
|
---|
2682 | newciinfo = firstciinfo;
|
---|
2683 | if (newciinfo) {
|
---|
2684 | do {
|
---|
2685 | if (((chunkinfoptr) newciinfo < p) && (p < (chunkinfoptr) (newciinfo+NUMBER_FREE_CHUNKS))) {
|
---|
2686 | p->fd = newciinfo->freelist;
|
---|
2687 | newciinfo->freelist = p;
|
---|
2688 | newciinfo->freecounter++;
|
---|
2689 | VALGRIND_MEMPOOL_FREE((char*)newciinfo, (char*)p);
|
---|
2690 | VALGRIND_MAKE_MEM_DEFINED(p,sizeof(struct chunkinfo));
|
---|
2691 | VALGRIND_MAKE_MEM_NOACCESS(p->size, sizeof(INTERNAL_SIZE_T));
|
---|
2692 | VALGRIND_MAKE_MEM_NOACCESS(p->req, sizeof(INTERNAL_SIZE_T));
|
---|
2693 | VALGRIND_MAKE_MEM_NOACCESS(p->bk, sizeof(struct chunkinfo*));
|
---|
2694 | VALGRIND_MAKE_MEM_NOACCESS(p->chunk, sizeof(mchunkptr));
|
---|
2695 | if (UNLIKELY(newciinfo->freecounter == NUMBER_FREE_CHUNKS))
|
---|
2696 | freeciregion(newciinfo);
|
---|
2697 | break;
|
---|
2698 | }
|
---|
2699 | newciinfo = newciinfo->next;
|
---|
2700 | } while (newciinfo);
|
---|
2701 | }
|
---|
2702 | }
|
---|
2703 | }
|
---|
2704 |
|
---|
2705 | /* Calculate the hash table entry for a chunk */
|
---|
2706 | #ifdef STARTHEAP_IS_ZERO
|
---|
2707 | #define hash(p) (((unsigned long) p) >> 7)
|
---|
2708 | #else
|
---|
2709 | #define hash(p) (((unsigned long) p - (unsigned long) startheap) >> 7)
|
---|
2710 | #endif
|
---|
2711 |
|
---|
2712 | static void
|
---|
2713 | hashtable_add (chunkinfoptr ci)
|
---|
2714 | {
|
---|
2715 | chunkinfoptr temp, next;
|
---|
2716 | unsigned long hashval;
|
---|
2717 | mchunkptr cic = chunk (ci);
|
---|
2718 |
|
---|
2719 | hashval = hash (cic);
|
---|
2720 |
|
---|
2721 | if (hashval < AMOUNTHASH) {
|
---|
2722 |
|
---|
2723 | temp = hashtable[hashval];
|
---|
2724 |
|
---|
2725 | #ifdef DNMALLOC_DEBUG
|
---|
2726 | fprintf(stderr, "hashtable_add: %p, %lu\n", chunk(ci), hashval);
|
---|
2727 | #endif
|
---|
2728 |
|
---|
2729 | /* If no pointer to a chunk info list is stored at this location
|
---|
2730 | * in the hashtable or if the chunk's address is smaller than the
|
---|
2731 | * one present, add the chunk to the front of the linked list
|
---|
2732 | */
|
---|
2733 | if (temp == 0 || chunk (temp) > cic)
|
---|
2734 | {
|
---|
2735 | ci->hash_next = temp;
|
---|
2736 | hashtable[hashval] = ci;
|
---|
2737 | if (!temp) /* more likely case */
|
---|
2738 | goto out;
|
---|
2739 | temp->prev_size = chunksize(ci);
|
---|
2740 | return;
|
---|
2741 | }
|
---|
2742 | else
|
---|
2743 | {
|
---|
2744 | /* We must place the chunk in the linked list for this hashentry
|
---|
2745 | * Loop to end of list or to a position where temp's chunk's address
|
---|
2746 | * is larger than the new chunkinfo's chunk's address
|
---|
2747 | */
|
---|
2748 | if (!temp->hash_next || (chunk (temp->hash_next) > cic))
|
---|
2749 | {
|
---|
2750 | ci->hash_next = temp->hash_next;
|
---|
2751 | temp->hash_next = ci;
|
---|
2752 | }
|
---|
2753 | else
|
---|
2754 | {
|
---|
2755 | while ((temp->hash_next != 0) && (chunk (temp->hash_next) < cic))
|
---|
2756 | {
|
---|
2757 | temp = temp->hash_next;
|
---|
2758 | }
|
---|
2759 | /* Place in linked list if not already there */
|
---|
2760 | if (!temp->hash_next || !(chunk (temp->hash_next) == cic))
|
---|
2761 | {
|
---|
2762 | ci->hash_next = temp->hash_next;
|
---|
2763 | temp->hash_next = ci;
|
---|
2764 | }
|
---|
2765 | }
|
---|
2766 | }
|
---|
2767 | }
|
---|
2768 | else {
|
---|
2769 | #ifdef DNMALLOC_CHECKS
|
---|
2770 | if (hashval >= AMOUNTHASH) {
|
---|
2771 | fprintf(stderr, "Dnmalloc error: trying to write outside of the bounds of the hashtable, this is definitely a bug, please email dnmalloc@fort-knox.org (hashval: %lu, AMOUNTHASH: %lu, HEAPMAXSIZE_HALF %lu PGSIZE %ld CHUNKINFOPAGE %ld chunk: %p, chunkinfo: %p, startheap: %p).\n", hashval, AMOUNTHASH, HEAPMAXSIZE_HALF, PGSIZE, CHUNKINFOPAGE, chunk(ci), ci, startheap);
|
---|
2772 | abort();
|
---|
2773 | }
|
---|
2774 | #else
|
---|
2775 | assert(hashval < AMOUNTHASH);
|
---|
2776 | #endif
|
---|
2777 | }
|
---|
2778 |
|
---|
2779 | out:
|
---|
2780 | next = next_chunkinfo(ci);
|
---|
2781 | if (!next)
|
---|
2782 | return;
|
---|
2783 | next->prev_size = chunksize(ci);
|
---|
2784 | }
|
---|
2785 |
|
---|
2786 | static void
|
---|
2787 | hashtable_insert (chunkinfoptr ci_orig, chunkinfoptr ci_insert)
|
---|
2788 | {
|
---|
2789 | chunkinfoptr next;
|
---|
2790 |
|
---|
2791 | #ifdef DNMALLOC_DEBUG
|
---|
2792 | fprintf(stderr, "hashtable_ins: %p, %lu\n", chunk(ci_insert),
|
---|
2793 | (unsigned long)hash(chunk(ci_insert));
|
---|
2794 | #endif
|
---|
2795 |
|
---|
2796 | if (hash(chunk(ci_orig)) != hash(chunk(ci_insert))) {
|
---|
2797 | hashtable_add(ci_insert);
|
---|
2798 | }
|
---|
2799 | else {
|
---|
2800 |
|
---|
2801 | ci_insert->hash_next = ci_orig->hash_next;
|
---|
2802 | ci_orig->hash_next = ci_insert;
|
---|
2803 |
|
---|
2804 | /* added for prevsize */
|
---|
2805 | if (!(ci_insert->hash_next))
|
---|
2806 | next = next_chunkinfo(ci_insert);
|
---|
2807 | else
|
---|
2808 | next = ci_insert->hash_next;
|
---|
2809 |
|
---|
2810 | if (!next)
|
---|
2811 | {
|
---|
2812 | ci_insert->prev_size = chunksize(ci_orig);
|
---|
2813 | }
|
---|
2814 | else
|
---|
2815 | {
|
---|
2816 | next->prev_size = chunksize(ci_insert);
|
---|
2817 | ci_insert->prev_size = chunksize(ci_orig);
|
---|
2818 | }
|
---|
2819 | }
|
---|
2820 | }
|
---|
2821 |
|
---|
2822 | static void
|
---|
2823 | hashtable_remove (mchunkptr p)
|
---|
2824 | {
|
---|
2825 | chunkinfoptr prevtemp, temp;
|
---|
2826 | unsigned long hashval;
|
---|
2827 |
|
---|
2828 | hashval = hash (p);
|
---|
2829 | #ifdef DNMALLOC_DEBUG
|
---|
2830 | fprintf(stderr, "hashtable_rem: %p, %lu\n", p, hashval);
|
---|
2831 | #endif
|
---|
2832 | assert(hashval < AMOUNTHASH); /* rw */
|
---|
2833 | prevtemp = temp = hashtable[hashval];
|
---|
2834 | if (chunk (temp) == p) {
|
---|
2835 | hashtable[hashval] = temp->hash_next;
|
---|
2836 | }
|
---|
2837 | else
|
---|
2838 | {
|
---|
2839 | if (temp && chunk (temp) != p) {
|
---|
2840 | do
|
---|
2841 | {
|
---|
2842 | prevtemp = temp;
|
---|
2843 | temp = temp->hash_next;
|
---|
2844 | } while (temp && chunk (temp) != p);
|
---|
2845 | }
|
---|
2846 | #ifdef DNMALLOC_CHECKS
|
---|
2847 | if (!temp) {
|
---|
2848 | fprintf (stderr,
|
---|
2849 | "Dnmalloc error (hash_rm): could not find a chunkinfo for the chunk %p in the hashtable at entry %lu\n This is definitely a bug, please report it to dnmalloc@fort-knox.org.\n",
|
---|
2850 | p, hashval);
|
---|
2851 | abort();
|
---|
2852 | }
|
---|
2853 | #else
|
---|
2854 | assert(temp != NULL);
|
---|
2855 | #endif
|
---|
2856 | prevtemp->hash_next = temp->hash_next;
|
---|
2857 | }
|
---|
2858 | }
|
---|
2859 |
|
---|
2860 | /* mmapped chunks are multiples of pagesize, no hash_nexts,
|
---|
2861 | * just remove from the hashtable
|
---|
2862 | */
|
---|
2863 | #define hashtable_remove_mmapped(p) hashtable[hash(p)] = 0;
|
---|
2864 |
|
---|
2865 | static void
|
---|
2866 | hashtable_skiprm (chunkinfoptr ci_orig, chunkinfoptr ci_todelete)
|
---|
2867 | {
|
---|
2868 | unsigned long hashval;
|
---|
2869 | chunkinfoptr next;
|
---|
2870 |
|
---|
2871 | #ifdef DNMALLOC_DEBUG
|
---|
2872 | fprintf(stderr, "hashtable_skiprm: %p, %lu\n", chunk(ci_todelete), hash(chunk(ci_todelete)));
|
---|
2873 | #endif
|
---|
2874 |
|
---|
2875 | if (ci_orig->hash_next != ci_todelete) {
|
---|
2876 | hashval = hash(chunk(ci_todelete));
|
---|
2877 | assert(hashval < AMOUNTHASH); /* rw */
|
---|
2878 | #ifdef DNMALLOC_CHECKS
|
---|
2879 | if (hashtable[hashval] != ci_todelete ) {
|
---|
2880 | fprintf(stderr, "Dnmalloc error: trying to delete wrong value (hash: %lu): ci_todelete: %p (%p), hashtable[hashval]: %p (%p)\n This is definitely a bug, please report it to dnmalloc@fort-knox.org.\n", hashval, ci_todelete, chunk(ci_todelete), hashtable[hashval], chunk(hashtable[hashval]));
|
---|
2881 | }
|
---|
2882 | #else
|
---|
2883 | assert(hashtable[hashval] == ci_todelete);
|
---|
2884 | #endif
|
---|
2885 | hashtable[hashval] = ci_todelete->hash_next;
|
---|
2886 | }
|
---|
2887 |
|
---|
2888 | else {
|
---|
2889 | ci_orig->hash_next = ci_todelete->hash_next;
|
---|
2890 | if (!ci_orig->hash_next) {
|
---|
2891 | next = next_chunkinfo(ci_orig);
|
---|
2892 | } else {
|
---|
2893 | next = ci_orig->hash_next;
|
---|
2894 | }
|
---|
2895 | if (next)
|
---|
2896 | next->prev_size = chunksize(ci_orig);
|
---|
2897 |
|
---|
2898 | }
|
---|
2899 | }
|
---|
2900 |
|
---|
2901 |
|
---|
2902 | static chunkinfoptr
|
---|
2903 | hashtable_lookup (mchunkptr p)
|
---|
2904 | {
|
---|
2905 | chunkinfoptr ci;
|
---|
2906 | unsigned long hashval;
|
---|
2907 |
|
---|
2908 | /* if we were called wrongly
|
---|
2909 | * if ((char *) p < startheap) return 0;
|
---|
2910 | */
|
---|
2911 | if ((char *) p >= startheap)
|
---|
2912 | {
|
---|
2913 | hashval = hash (p);
|
---|
2914 | assert(hashval < AMOUNTHASH); /* rw */
|
---|
2915 | ci = hashtable[hashval];
|
---|
2916 | if (ci && chunk (ci) == p)
|
---|
2917 | return ci;
|
---|
2918 |
|
---|
2919 | if (ci) {
|
---|
2920 | do {
|
---|
2921 | ci = ci->hash_next;
|
---|
2922 | } while (ci && chunk (ci) != p);
|
---|
2923 | }
|
---|
2924 | #ifdef DNMALLOC_CHECKS
|
---|
2925 | /* This should never occur but if it does, we'd like to know */
|
---|
2926 | if (!ci) {
|
---|
2927 | fprintf (stderr,
|
---|
2928 | "Dnmalloc error: could not find a chunkinfo for the chunk %p in the hashtable at entry %lu\n This is definitely a bug, please report it to dnmalloc@fort-knox.org.\n",
|
---|
2929 | p, hashval);
|
---|
2930 | abort();
|
---|
2931 | }
|
---|
2932 | #else
|
---|
2933 | assert(ci != NULL);
|
---|
2934 | #endif
|
---|
2935 | return ci;
|
---|
2936 | }
|
---|
2937 | return 0;
|
---|
2938 | }
|
---|
2939 |
|
---|
2940 |
|
---|
2941 |
|
---|
2942 | /*
|
---|
2943 | ----------- Internal state representation and initialization -----------
|
---|
2944 | */
|
---|
2945 |
|
---|
2946 | struct malloc_state {
|
---|
2947 |
|
---|
2948 | /* The maximum chunk size to be eligible for fastbin */
|
---|
2949 | INTERNAL_SIZE_T max_fast; /* low 2 bits used as flags */
|
---|
2950 |
|
---|
2951 | /* Fastbins */
|
---|
2952 | mfastbinptr fastbins[NFASTBINS];
|
---|
2953 |
|
---|
2954 | /* Base of the topmost chunk -- not otherwise kept in a bin */
|
---|
2955 | chunkinfoptr top;
|
---|
2956 |
|
---|
2957 | /* The remainder from the most recent split of a small request */
|
---|
2958 | chunkinfoptr last_remainder;
|
---|
2959 |
|
---|
2960 | /* Normal bins */
|
---|
2961 | struct chunkinfo bins[NBINS];
|
---|
2962 |
|
---|
2963 | /* Bitmap of bins. Trailing zero map handles cases of largest binned size */
|
---|
2964 | unsigned int binmap[BINMAPSIZE+1];
|
---|
2965 |
|
---|
2966 | /* Tunable parameters */
|
---|
2967 | CHUNK_SIZE_T trim_threshold;
|
---|
2968 | INTERNAL_SIZE_T top_pad;
|
---|
2969 | INTERNAL_SIZE_T mmap_threshold;
|
---|
2970 |
|
---|
2971 | /* Memory map support */
|
---|
2972 | int n_mmaps;
|
---|
2973 | int n_mmaps_max;
|
---|
2974 | int max_n_mmaps;
|
---|
2975 |
|
---|
2976 | /* Cache malloc_getpagesize */
|
---|
2977 | unsigned int pagesize;
|
---|
2978 |
|
---|
2979 | /* Canary */
|
---|
2980 | char guard_stored[GUARD_SIZE];
|
---|
2981 |
|
---|
2982 | /* Track properties of MORECORE */
|
---|
2983 | unsigned int morecore_properties;
|
---|
2984 |
|
---|
2985 | /* Statistics */
|
---|
2986 | INTERNAL_SIZE_T mmapped_mem;
|
---|
2987 | INTERNAL_SIZE_T sbrked_mem;
|
---|
2988 | INTERNAL_SIZE_T max_sbrked_mem;
|
---|
2989 | INTERNAL_SIZE_T max_mmapped_mem;
|
---|
2990 | INTERNAL_SIZE_T max_total_mem;
|
---|
2991 | };
|
---|
2992 |
|
---|
2993 | typedef struct malloc_state *mstate;
|
---|
2994 |
|
---|
2995 | /*
|
---|
2996 | There is exactly one instance of this struct in this malloc.
|
---|
2997 | If you are adapting this malloc in a way that does NOT use a static
|
---|
2998 | malloc_state, you MUST explicitly zero-fill it before using. This
|
---|
2999 | malloc relies on the property that malloc_state is initialized to
|
---|
3000 | all zeroes (as is true of C statics).
|
---|
3001 | */
|
---|
3002 |
|
---|
3003 | static struct malloc_state * av_ = NULL; /* never directly referenced */
|
---|
3004 |
|
---|
3005 | /*
|
---|
3006 | All uses of av_ are via get_malloc_state().
|
---|
3007 | At most one "call" to get_malloc_state is made per invocation of
|
---|
3008 | the public versions of malloc and free, but other routines
|
---|
3009 | that in turn invoke malloc and/or free may call more then once.
|
---|
3010 | Also, it is called in check* routines if DEBUG is set.
|
---|
3011 | */
|
---|
3012 |
|
---|
3013 | #define get_malloc_state() (av_)
|
---|
3014 |
|
---|
3015 | /*
|
---|
3016 | Initialize a malloc_state struct.
|
---|
3017 |
|
---|
3018 | This is called only from within malloc_consolidate, which needs
|
---|
3019 | be called in the same contexts anyway. It is never called directly
|
---|
3020 | outside of malloc_consolidate because some optimizing compilers try
|
---|
3021 | to inline it at all call points, which turns out not to be an
|
---|
3022 | optimization at all. (Inlining it in malloc_consolidate is fine though.)
|
---|
3023 | */
|
---|
3024 |
|
---|
3025 | #if __STD_C
|
---|
3026 | static void malloc_mmap_state(void)
|
---|
3027 | #else
|
---|
3028 | static void malloc_mmap_state()
|
---|
3029 | #endif
|
---|
3030 | {
|
---|
3031 | int mprot;
|
---|
3032 | unsigned long pagesize = malloc_getpagesize;
|
---|
3033 | size_t size = (sizeof(struct malloc_state) + pagesize - 1) & ~(pagesize - 1);
|
---|
3034 |
|
---|
3035 | void * foo = MMAP(0, size+(2*pagesize), PROT_READ|PROT_WRITE, MAP_PRIVATE);
|
---|
3036 |
|
---|
3037 |
|
---|
3038 | #ifdef NDEBUG
|
---|
3039 | if (foo == MAP_FAILED) {
|
---|
3040 | fprintf (stderr, "Couldn't mmap struct malloc_state: %s\n", strerror (errno));
|
---|
3041 | abort ();
|
---|
3042 | }
|
---|
3043 | #else
|
---|
3044 | assert(foo != MAP_FAILED);
|
---|
3045 | #endif
|
---|
3046 |
|
---|
3047 | mprot = mprotect(foo, pagesize, PROT_NONE);
|
---|
3048 | #ifdef NDEBUG
|
---|
3049 | if (mprot == -1) {
|
---|
3050 | fprintf (stderr, "Couldn't mprotect first non-rw page for struct malloc_state: %s\n",
|
---|
3051 | strerror (errno));
|
---|
3052 | abort ();
|
---|
3053 | }
|
---|
3054 | #else
|
---|
3055 | assert(mprot != -1);
|
---|
3056 | #endif
|
---|
3057 |
|
---|
3058 | av_ = (struct malloc_state *) ((char*)foo + pagesize);
|
---|
3059 |
|
---|
3060 | MALLOC_ZERO(av_, sizeof(struct malloc_state));
|
---|
3061 |
|
---|
3062 | mprot = mprotect((void*)((char*)foo + size + pagesize), (size_t) pagesize, PROT_NONE);
|
---|
3063 | #ifdef NDEBUG
|
---|
3064 | if (mprot == -1) {
|
---|
3065 | fprintf (stderr,
|
---|
3066 | "Couldn't mprotect last non-rw page for struct malloc_state: %s\n",
|
---|
3067 | strerror (errno));
|
---|
3068 | abort ();
|
---|
3069 | }
|
---|
3070 | #else
|
---|
3071 | assert(mprot != -1);
|
---|
3072 | #endif
|
---|
3073 | }
|
---|
3074 |
|
---|
3075 | #if __STD_C
|
---|
3076 | static void malloc_init_state(mstate av)
|
---|
3077 | #else
|
---|
3078 | static void malloc_init_state(av) mstate av;
|
---|
3079 | #endif
|
---|
3080 | {
|
---|
3081 | int i;
|
---|
3082 | mbinptr bin;
|
---|
3083 |
|
---|
3084 | void * morecore_test = MORECORE(0);
|
---|
3085 | unsigned long hashval;
|
---|
3086 |
|
---|
3087 | /* Test morecore function
|
---|
3088 | */
|
---|
3089 | set_morecore32bit(av);
|
---|
3090 |
|
---|
3091 | if (morecore_test == MORECORE_FAILURE)
|
---|
3092 | {
|
---|
3093 | set_nonmorecore32bit(av);
|
---|
3094 | }
|
---|
3095 | else
|
---|
3096 | {
|
---|
3097 | /* On 64bit systems, the heap may be located above the
|
---|
3098 | * 32bit address space. Since mmap() probably still can be
|
---|
3099 | * convinced to map within 32bit, we don't use sbrk().
|
---|
3100 | */
|
---|
3101 | hashval = hash (morecore_test);
|
---|
3102 | if (hashval >= AMOUNTHASH)
|
---|
3103 | {
|
---|
3104 | set_nonmorecore32bit(av);
|
---|
3105 | }
|
---|
3106 | }
|
---|
3107 |
|
---|
3108 |
|
---|
3109 | /* Establish circular links for normal bins */
|
---|
3110 | for (i = 1; i < NBINS; ++i) {
|
---|
3111 | bin = bin_at(av,i);
|
---|
3112 | bin->fd = bin->bk = bin;
|
---|
3113 | }
|
---|
3114 |
|
---|
3115 | av->top_pad = DEFAULT_TOP_PAD;
|
---|
3116 | av->n_mmaps_max = DEFAULT_MMAP_MAX;
|
---|
3117 | av->mmap_threshold = DEFAULT_MMAP_THRESHOLD;
|
---|
3118 | av->trim_threshold = DEFAULT_TRIM_THRESHOLD;
|
---|
3119 |
|
---|
3120 | #if MORECORE_CONTIGUOUS
|
---|
3121 | set_contiguous(av);
|
---|
3122 | #else
|
---|
3123 | set_noncontiguous(av);
|
---|
3124 | #endif
|
---|
3125 |
|
---|
3126 | set_max_fast(av, DEFAULT_MXFAST);
|
---|
3127 |
|
---|
3128 | av->top = cireg_getfree ();
|
---|
3129 | av->top->chunk = (mchunkptr) startheap;
|
---|
3130 | av->top->size = 0;
|
---|
3131 | set_previnuse(av->top);
|
---|
3132 | clear_inuse(av->top);
|
---|
3133 | hashtable[0] = av->top;
|
---|
3134 | av->pagesize = malloc_getpagesize;
|
---|
3135 |
|
---|
3136 | memcpy(av->guard_stored, dnmalloc_arc4random(), GUARD_SIZE);
|
---|
3137 |
|
---|
3138 | }
|
---|
3139 |
|
---|
3140 | /*
|
---|
3141 | Other internal utilities operating on mstates
|
---|
3142 | */
|
---|
3143 |
|
---|
3144 | #if __STD_C
|
---|
3145 | static Void_t* sYSMALLOc(INTERNAL_SIZE_T, mstate);
|
---|
3146 | static int sYSTRIm(size_t, mstate);
|
---|
3147 | static void malloc_consolidate(mstate);
|
---|
3148 | #else
|
---|
3149 | static Void_t* sYSMALLOc();
|
---|
3150 | static int sYSTRIm();
|
---|
3151 | static void malloc_consolidate();
|
---|
3152 | #endif
|
---|
3153 |
|
---|
3154 | /* dnmalloc functions */
|
---|
3155 | /* needs mstate so moved here */
|
---|
3156 |
|
---|
3157 | static chunkinfoptr
|
---|
3158 | next_chunkinfo (chunkinfoptr ci)
|
---|
3159 | {
|
---|
3160 | mchunkptr nextp;
|
---|
3161 | unsigned long hashval;
|
---|
3162 | chunkinfoptr cinfonextp;
|
---|
3163 | mstate av = get_malloc_state();
|
---|
3164 |
|
---|
3165 | /* ci is not the last element in the linked list, just
|
---|
3166 | return the next chunkinfo from the list
|
---|
3167 | */
|
---|
3168 | if (!ci->hash_next)
|
---|
3169 | {
|
---|
3170 | /* ci is the last element, find the next chunkinfo by
|
---|
3171 | * looking up the chunkinfo for the chunk that is after p's chunk
|
---|
3172 | */
|
---|
3173 | nextp = (mchunkptr) (((char *) (ci->chunk)) + chunksize (ci));
|
---|
3174 |
|
---|
3175 | if (!(nextp == av->top->chunk))
|
---|
3176 | {
|
---|
3177 | hashval = hash (nextp);
|
---|
3178 | /* assert(hashval < AMOUNTHASH); *//* major bottleneck */
|
---|
3179 | cinfonextp = hashtable[hashval];
|
---|
3180 | if (cinfonextp && chunk (cinfonextp) == nextp)
|
---|
3181 | return cinfonextp;
|
---|
3182 |
|
---|
3183 | #ifdef DNMALLOC_CHECKS_EXTRA
|
---|
3184 | /* This seems bogus; a chunkinfo may legally have no nextp if
|
---|
3185 | * it's the last one allocated (?)
|
---|
3186 | */
|
---|
3187 | else {
|
---|
3188 | if (cinfonextp)
|
---|
3189 | fprintf (stderr,
|
---|
3190 | "Dnmalloc error: could not find a next chunkinfo for the chunk %p in the hashtable at entry %lu, cinfonextp: %p, chunk(cinfonextp): %p, nextp: %p\n This is definitely a bug, please report it to dnmalloc@fort-knox.org.\n",
|
---|
3191 | chunk(ci), hashval, cinfonextp, chunk(cinfonextp), nextp);
|
---|
3192 | else
|
---|
3193 | fprintf (stderr,
|
---|
3194 | "Dnmalloc error: could not find a next chunkinfo for the chunk %p in the hashtable at entry %lu, cinfonextp: %s, chunk(cinfonextp): %s, nextp: %p\n This is definitely a bug, please report it to dnmalloc@fort-knox.org.\n",
|
---|
3195 | chunk(ci), hashval, "null", "null", nextp);
|
---|
3196 | }
|
---|
3197 | #endif
|
---|
3198 |
|
---|
3199 | return NULL;
|
---|
3200 | }
|
---|
3201 | else
|
---|
3202 | {
|
---|
3203 | return av->top;
|
---|
3204 | }
|
---|
3205 |
|
---|
3206 | }
|
---|
3207 | else
|
---|
3208 | {
|
---|
3209 | return (ci->hash_next);
|
---|
3210 | }
|
---|
3211 | }
|
---|
3212 |
|
---|
3213 | static int is_next_chunk(chunkinfoptr oldp, chunkinfoptr newp) {
|
---|
3214 | mchunkptr nextp;
|
---|
3215 | if (oldp->hash_next == newp)
|
---|
3216 | return 1;
|
---|
3217 | nextp = (mchunkptr) (((char *) (oldp->chunk)) + chunksize (oldp));
|
---|
3218 | if (nextp == chunk(newp))
|
---|
3219 | return 1;
|
---|
3220 | return 0;
|
---|
3221 | }
|
---|
3222 |
|
---|
3223 |
|
---|
3224 |
|
---|
3225 | /* Get the chunkinfo of the physically previous chunk */
|
---|
3226 | /* Since we disposed of prev_size, we need this function to find the previous */
|
---|
3227 |
|
---|
3228 | static chunkinfoptr
|
---|
3229 | prev_chunkinfo (chunkinfoptr ci)
|
---|
3230 | {
|
---|
3231 | unsigned int i;
|
---|
3232 | chunkinfoptr prev;
|
---|
3233 | mchunkptr prevchunk = 0;
|
---|
3234 | /* chunkinfoptr temp; */
|
---|
3235 |
|
---|
3236 | /* Get the hashtable location of the chunkinfo */
|
---|
3237 | i = hash (chunk (ci));
|
---|
3238 | assert(i < AMOUNTHASH); /* rw */
|
---|
3239 |
|
---|
3240 | /* Get the first element of the linked list of chunkinfo's that contains p */
|
---|
3241 | prev = hashtable[i];
|
---|
3242 |
|
---|
3243 | if (ci == prev) {
|
---|
3244 | prevchunk = (mchunkptr) (((char *) (ci->chunk)) - (ci->prev_size));
|
---|
3245 | i = hash(prevchunk);
|
---|
3246 | assert(i < AMOUNTHASH); /* rw */
|
---|
3247 | /* Loop over the linked list until we reach the last element */
|
---|
3248 | for (prev = hashtable[i]; prev->hash_next != 0; prev = prev->hash_next) ;
|
---|
3249 | } else {
|
---|
3250 | /* p is not the first element in the linked list, we can just
|
---|
3251 | loop over the list and return the previous
|
---|
3252 | */
|
---|
3253 | for (prev = hashtable[i]; prev->hash_next != ci; prev = prev->hash_next);
|
---|
3254 | }
|
---|
3255 |
|
---|
3256 | return prev;
|
---|
3257 | }
|
---|
3258 |
|
---|
3259 |
|
---|
3260 | /*
|
---|
3261 | Debugging support
|
---|
3262 | Dnmalloc broke dlmallocs debugging functions, should fix them some
|
---|
3263 | time in the future, for now leave them undefined.
|
---|
3264 | */
|
---|
3265 |
|
---|
3266 | #define check_chunk(P)
|
---|
3267 | #define check_free_chunk(P)
|
---|
3268 | #define check_inuse_chunk(P)
|
---|
3269 | #define check_remalloced_chunk(P,N)
|
---|
3270 | #define check_malloced_chunk(P,N)
|
---|
3271 | #define check_malloc_state()
|
---|
3272 |
|
---|
3273 |
|
---|
3274 | /* ----------- Routines dealing with system allocation -------------- */
|
---|
3275 |
|
---|
3276 | /*
|
---|
3277 | sysmalloc handles malloc cases requiring more memory from the system.
|
---|
3278 | On entry, it is assumed that av->top does not have enough
|
---|
3279 | space to service request for nb bytes, thus requiring that av->top
|
---|
3280 | be extended or replaced.
|
---|
3281 | */
|
---|
3282 |
|
---|
3283 | #if __STD_C
|
---|
3284 | static Void_t* sYSMALLOc(INTERNAL_SIZE_T nb, mstate av)
|
---|
3285 | #else
|
---|
3286 | static Void_t* sYSMALLOc(nb, av) INTERNAL_SIZE_T nb; mstate av;
|
---|
3287 | #endif
|
---|
3288 | {
|
---|
3289 | chunkinfoptr old_top; /* incoming value of av->top */
|
---|
3290 | INTERNAL_SIZE_T old_size; /* its size */
|
---|
3291 | char* old_end; /* its end address */
|
---|
3292 |
|
---|
3293 | long size; /* arg to first MORECORE or mmap call */
|
---|
3294 | char* brk; /* return value from MORECORE */
|
---|
3295 |
|
---|
3296 | long correction; /* arg to 2nd MORECORE call */
|
---|
3297 | char* snd_brk; /* 2nd return val */
|
---|
3298 |
|
---|
3299 | INTERNAL_SIZE_T front_misalign; /* unusable bytes at front of new space */
|
---|
3300 | INTERNAL_SIZE_T end_misalign; /* partial page left at end of new space */
|
---|
3301 | char* aligned_brk; /* aligned offset into brk */
|
---|
3302 |
|
---|
3303 | chunkinfoptr p; /* the allocated/returned chunk */
|
---|
3304 | chunkinfoptr remainder; /* remainder from allocation */
|
---|
3305 | chunkinfoptr fencepost; /* fencepost */
|
---|
3306 | CHUNK_SIZE_T remainder_size; /* its size */
|
---|
3307 |
|
---|
3308 | CHUNK_SIZE_T sum; /* for updating stats */
|
---|
3309 |
|
---|
3310 | size_t pagemask = av->pagesize - 1;
|
---|
3311 |
|
---|
3312 | #ifdef DNMALLOC_DEBUG
|
---|
3313 | fprintf(stderr, "Enter sysmalloc\n");
|
---|
3314 | #endif
|
---|
3315 | /*
|
---|
3316 | If there is space available in fastbins, consolidate and retry
|
---|
3317 | malloc from scratch rather than getting memory from system. This
|
---|
3318 | can occur only if nb is in smallbin range so we didn't consolidate
|
---|
3319 | upon entry to malloc. It is much easier to handle this case here
|
---|
3320 | than in malloc proper.
|
---|
3321 | */
|
---|
3322 |
|
---|
3323 |
|
---|
3324 | if (have_fastchunks(av)) {
|
---|
3325 | Void_t * retval;
|
---|
3326 | assert(in_smallbin_range(nb));
|
---|
3327 | malloc_consolidate(av);
|
---|
3328 | #ifdef DNMALLOC_DEBUG
|
---|
3329 | fprintf(stderr, "Return sysmalloc have_fastchunks\n");
|
---|
3330 | #endif
|
---|
3331 | retval = mALLOc(nb - MALLOC_ALIGN_MASK);
|
---|
3332 | VALGRIND_FREELIKE_BLOCK(retval, 0);
|
---|
3333 | return retval;
|
---|
3334 | }
|
---|
3335 |
|
---|
3336 |
|
---|
3337 | /*
|
---|
3338 | If have mmap, and the request size meets the mmap threshold, and
|
---|
3339 | the system supports mmap, and there are few enough currently
|
---|
3340 | allocated mmapped regions, try to directly map this request
|
---|
3341 | rather than expanding top.
|
---|
3342 | */
|
---|
3343 |
|
---|
3344 | if (UNLIKELY((CHUNK_SIZE_T)(nb) >= (CHUNK_SIZE_T)(av->mmap_threshold) &&
|
---|
3345 | (av->n_mmaps < av->n_mmaps_max))) {
|
---|
3346 |
|
---|
3347 | char* mm; /* return value from mmap call*/
|
---|
3348 |
|
---|
3349 | /*
|
---|
3350 | Round up size to nearest page. For mmapped chunks, the overhead
|
---|
3351 | is one SIZE_SZ unit larger than for normal chunks, because there
|
---|
3352 | is no following chunk whose prev_size field could be used.
|
---|
3353 | */
|
---|
3354 | size = (nb + MALLOC_ALIGN_MASK + pagemask) & ~pagemask;
|
---|
3355 |
|
---|
3356 | /* Don't try if size wraps around 0 */
|
---|
3357 | if ((CHUNK_SIZE_T)(size) > (CHUNK_SIZE_T)(nb)) {
|
---|
3358 |
|
---|
3359 |
|
---|
3360 | mm = (char*)(MMAP(0, size, PROT_READ|PROT_WRITE, MAP_PRIVATE));
|
---|
3361 |
|
---|
3362 | if (mm != (char*)(MORECORE_FAILURE)) {
|
---|
3363 |
|
---|
3364 | VALGRIND_MAKE_MEM_NOACCESS(mm,size);
|
---|
3365 |
|
---|
3366 | /*
|
---|
3367 | The offset to the start of the mmapped region is stored
|
---|
3368 | in the prev_size field of the chunk. This allows us to adjust
|
---|
3369 | returned start address to meet alignment requirements here
|
---|
3370 | and in memalign(), and still be able to compute proper
|
---|
3371 | address argument for later munmap in free() and realloc().
|
---|
3372 | */
|
---|
3373 |
|
---|
3374 | front_misalign = (INTERNAL_SIZE_T) mm & MALLOC_ALIGN_MASK;
|
---|
3375 | p = cireg_getfree();
|
---|
3376 |
|
---|
3377 | if (front_misalign > 0) {
|
---|
3378 | correction = MALLOC_ALIGNMENT - front_misalign;
|
---|
3379 | p->chunk = (mchunkptr)(mm + correction);
|
---|
3380 | p->hash_next = (chunkinfoptr) correction;
|
---|
3381 | set_head(p, (size - correction) |INUSE|IS_MMAPPED);
|
---|
3382 | }
|
---|
3383 | else {
|
---|
3384 | p->chunk = (mchunkptr)mm;
|
---|
3385 | p->hash_next = 0;
|
---|
3386 | set_head(p, size|INUSE|IS_MMAPPED);
|
---|
3387 | }
|
---|
3388 | hashtable_add(p);
|
---|
3389 | /* update statistics */
|
---|
3390 |
|
---|
3391 | if (++av->n_mmaps > av->max_n_mmaps)
|
---|
3392 | av->max_n_mmaps = av->n_mmaps;
|
---|
3393 |
|
---|
3394 | sum = av->mmapped_mem += size;
|
---|
3395 | if (sum > (CHUNK_SIZE_T)(av->max_mmapped_mem))
|
---|
3396 | av->max_mmapped_mem = sum;
|
---|
3397 | sum += av->sbrked_mem;
|
---|
3398 | if (sum > (CHUNK_SIZE_T)(av->max_total_mem))
|
---|
3399 | av->max_total_mem = sum;
|
---|
3400 |
|
---|
3401 | check_chunk(p);
|
---|
3402 |
|
---|
3403 | #ifdef DNMALLOC_DEBUG
|
---|
3404 | fprintf(stderr, "Return mmapped (%lu, total %lu)\n",
|
---|
3405 | size, (unsigned long)/* size_t */av->max_total_mem );
|
---|
3406 | #endif
|
---|
3407 | return chunk(p);
|
---|
3408 | }
|
---|
3409 | }
|
---|
3410 | }
|
---|
3411 |
|
---|
3412 | /* Record incoming configuration of top */
|
---|
3413 |
|
---|
3414 | old_top = av->top;
|
---|
3415 | old_size = chunksize(old_top);
|
---|
3416 | old_end = (char*)(chunk_at_offset(chunk(old_top), old_size));
|
---|
3417 |
|
---|
3418 | brk = snd_brk = (char*)(MORECORE_FAILURE);
|
---|
3419 |
|
---|
3420 | /*
|
---|
3421 | If not the first time through, we require old_size to be
|
---|
3422 | at least MINSIZE and to have prev_inuse set.
|
---|
3423 | */
|
---|
3424 |
|
---|
3425 | /* assert((old_top == initial_top(av) && old_size == 0) ||
|
---|
3426 | ((CHUNK_SIZE_T) (old_size) >= MINSIZE &&
|
---|
3427 | prev_inuse(old_top))); */
|
---|
3428 |
|
---|
3429 | /* Precondition: not enough current space to satisfy nb request */
|
---|
3430 | assert((CHUNK_SIZE_T)(old_size) < (CHUNK_SIZE_T)(nb + MINSIZE));
|
---|
3431 |
|
---|
3432 | /* Precondition: all fastbins are consolidated */
|
---|
3433 | assert(!have_fastchunks(av));
|
---|
3434 |
|
---|
3435 | /* Request enough space for nb + pad + overhead */
|
---|
3436 | size = nb + av->top_pad + MINSIZE;
|
---|
3437 |
|
---|
3438 | /*
|
---|
3439 | If contiguous, we can subtract out existing space that we hope to
|
---|
3440 | combine with new space. We add it back later only if
|
---|
3441 | we don't actually get contiguous space.
|
---|
3442 | */
|
---|
3443 | if (contiguous(av))
|
---|
3444 | size -= old_size;
|
---|
3445 |
|
---|
3446 | /*
|
---|
3447 | Round to a multiple of page size.
|
---|
3448 | If MORECORE is not contiguous, this ensures that we only call it
|
---|
3449 | with whole-page arguments. And if MORECORE is contiguous and
|
---|
3450 | this is not first time through, this preserves page-alignment of
|
---|
3451 | previous calls. Otherwise, we correct to page-align below.
|
---|
3452 | */
|
---|
3453 |
|
---|
3454 | size = (size + pagemask) & ~pagemask;
|
---|
3455 |
|
---|
3456 | /*
|
---|
3457 | Don't try to call MORECORE if argument is so big as to appear
|
---|
3458 | negative. Note that since mmap takes size_t arg, it may succeed
|
---|
3459 | below even if we cannot call MORECORE.
|
---|
3460 | */
|
---|
3461 | if (size > 0 && morecore32bit(av))
|
---|
3462 | brk = (char*)(MORECORE(size));
|
---|
3463 |
|
---|
3464 | /*
|
---|
3465 | If have mmap, try using it as a backup when MORECORE fails or
|
---|
3466 | cannot be used. This is worth doing on systems that have "holes" in
|
---|
3467 | address space, so sbrk cannot extend to give contiguous space, but
|
---|
3468 | space is available elsewhere. Note that we ignore mmap max count
|
---|
3469 | and threshold limits, since the space will not be used as a
|
---|
3470 | segregated mmap region.
|
---|
3471 | */
|
---|
3472 | if (brk != (char*)(MORECORE_FAILURE)) {
|
---|
3473 | av->sbrked_mem += size;
|
---|
3474 | VALGRIND_MAKE_MEM_NOACCESS(brk,size);
|
---|
3475 | }
|
---|
3476 |
|
---|
3477 | else {
|
---|
3478 |
|
---|
3479 | #ifdef DNMALLOC_DEBUG
|
---|
3480 | fprintf(stderr, "Morecore failure in sysmalloc\n");
|
---|
3481 | #endif
|
---|
3482 |
|
---|
3483 | /* Cannot merge with old top, so add its size back in */
|
---|
3484 | if (contiguous(av))
|
---|
3485 | size = (size + old_size + pagemask) & ~pagemask;
|
---|
3486 |
|
---|
3487 | /* If we are relying on mmap as backup, then use larger units */
|
---|
3488 | if ((CHUNK_SIZE_T)(size) < (CHUNK_SIZE_T)(MMAP_AS_MORECORE_SIZE))
|
---|
3489 | size = MMAP_AS_MORECORE_SIZE;
|
---|
3490 |
|
---|
3491 | /* Don't try if size wraps around 0 */
|
---|
3492 | if ((CHUNK_SIZE_T)(size) > (CHUNK_SIZE_T)(nb)) {
|
---|
3493 |
|
---|
3494 | #ifdef DNMALLOC_DEBUG
|
---|
3495 | fprintf(stderr, "Try mmap in sysmalloc\n");
|
---|
3496 | #endif
|
---|
3497 | brk = (char*)(MMAP(0, size, PROT_READ|PROT_WRITE, MAP_PRIVATE));
|
---|
3498 |
|
---|
3499 | if (brk != (char*)(MORECORE_FAILURE)) {
|
---|
3500 |
|
---|
3501 | VALGRIND_MAKE_MEM_NOACCESS(brk,size);
|
---|
3502 |
|
---|
3503 | av->mmapped_mem += size;
|
---|
3504 | #ifdef DNMALLOC_DEBUG
|
---|
3505 | fprintf(stderr, "Mmapped successfully in sysmalloc %p\n", brk);
|
---|
3506 | #endif
|
---|
3507 |
|
---|
3508 | /* We do not need, and cannot use, another sbrk call to find end */
|
---|
3509 | snd_brk = brk + size;
|
---|
3510 |
|
---|
3511 | /*
|
---|
3512 | Record that we no longer have a contiguous sbrk region.
|
---|
3513 | After the first time mmap is used as backup, we do not
|
---|
3514 | ever rely on contiguous space since this could incorrectly
|
---|
3515 | bridge regions.
|
---|
3516 | */
|
---|
3517 | set_noncontiguous(av);
|
---|
3518 | }
|
---|
3519 | }
|
---|
3520 | }
|
---|
3521 |
|
---|
3522 | if (brk != (char*)(MORECORE_FAILURE)) {
|
---|
3523 | #ifdef DNMALLOC_DEBUG
|
---|
3524 | fprintf(stderr, "Success path %lu allocated, sbrked %lu\n",
|
---|
3525 | size, (unsigned long)av->sbrked_mem);
|
---|
3526 | #endif
|
---|
3527 | /* av->sbrked_mem += size; moved up */
|
---|
3528 |
|
---|
3529 | /*
|
---|
3530 | If MORECORE extends previous space, we can likewise extend top size.
|
---|
3531 | */
|
---|
3532 |
|
---|
3533 | if (brk == old_end && snd_brk == (char*)(MORECORE_FAILURE)) {
|
---|
3534 | set_head(old_top, (size + old_size) | PREV_INUSE);
|
---|
3535 | #ifdef DNMALLOC_DEBUG
|
---|
3536 | fprintf(stderr, "Previous space extended\n");
|
---|
3537 | #endif
|
---|
3538 | }
|
---|
3539 |
|
---|
3540 | /*
|
---|
3541 | Otherwise, make adjustments:
|
---|
3542 |
|
---|
3543 | * If the first time through or noncontiguous, we need to call sbrk
|
---|
3544 | just to find out where the end of memory lies.
|
---|
3545 |
|
---|
3546 | * We need to ensure that all returned chunks from malloc will meet
|
---|
3547 | MALLOC_ALIGNMENT
|
---|
3548 |
|
---|
3549 | * If there was an intervening foreign sbrk, we need to adjust sbrk
|
---|
3550 | request size to account for fact that we will not be able to
|
---|
3551 | combine new space with existing space in old_top.
|
---|
3552 |
|
---|
3553 | * Almost all systems internally allocate whole pages at a time, in
|
---|
3554 | which case we might as well use the whole last page of request.
|
---|
3555 | So we allocate enough more memory to hit a page boundary now,
|
---|
3556 | which in turn causes future contiguous calls to page-align.
|
---|
3557 | */
|
---|
3558 |
|
---|
3559 | else {
|
---|
3560 | front_misalign = 0;
|
---|
3561 | end_misalign = 0;
|
---|
3562 | correction = 0;
|
---|
3563 | aligned_brk = brk;
|
---|
3564 |
|
---|
3565 | /*
|
---|
3566 | If MORECORE returns an address lower than we have seen before,
|
---|
3567 | we know it isn't really contiguous. This and some subsequent
|
---|
3568 | checks help cope with non-conforming MORECORE functions and
|
---|
3569 | the presence of "foreign" calls to MORECORE from outside of
|
---|
3570 | malloc or by other threads. We cannot guarantee to detect
|
---|
3571 | these in all cases, but cope with the ones we do detect.
|
---|
3572 | */
|
---|
3573 | if (contiguous(av) && old_size != 0 && brk < old_end) {
|
---|
3574 | set_noncontiguous(av);
|
---|
3575 | }
|
---|
3576 |
|
---|
3577 | /* handle contiguous cases */
|
---|
3578 | if (contiguous(av)) {
|
---|
3579 |
|
---|
3580 | #ifdef DNMALLOC_DEBUG
|
---|
3581 | fprintf(stderr, "Handle contiguous cases\n");
|
---|
3582 | #endif
|
---|
3583 | /*
|
---|
3584 | We can tolerate forward non-contiguities here (usually due
|
---|
3585 | to foreign calls) but treat them as part of our space for
|
---|
3586 | stats reporting.
|
---|
3587 | */
|
---|
3588 | if (old_size != 0)
|
---|
3589 | av->sbrked_mem += brk - old_end;
|
---|
3590 |
|
---|
3591 | /* Guarantee alignment of first new chunk made from this space */
|
---|
3592 |
|
---|
3593 | front_misalign = (INTERNAL_SIZE_T) brk & MALLOC_ALIGN_MASK;
|
---|
3594 | if (front_misalign > 0) {
|
---|
3595 |
|
---|
3596 | /*
|
---|
3597 | Skip over some bytes to arrive at an aligned position.
|
---|
3598 | We don't need to specially mark these wasted front bytes.
|
---|
3599 | They will never be accessed anyway because
|
---|
3600 | prev_inuse of av->top (and any chunk created from its start)
|
---|
3601 | is always true after initialization.
|
---|
3602 | */
|
---|
3603 |
|
---|
3604 | correction = MALLOC_ALIGNMENT - front_misalign;
|
---|
3605 | aligned_brk += correction;
|
---|
3606 | }
|
---|
3607 |
|
---|
3608 | /*
|
---|
3609 | If this isn't adjacent to existing space, then we will not
|
---|
3610 | be able to merge with old_top space, so must add to 2nd request.
|
---|
3611 | */
|
---|
3612 |
|
---|
3613 | correction += old_size;
|
---|
3614 |
|
---|
3615 | /* Extend the end address to hit a page boundary */
|
---|
3616 | end_misalign = (INTERNAL_SIZE_T)(brk + size + correction);
|
---|
3617 | correction += ((end_misalign + pagemask) & ~pagemask) - end_misalign;
|
---|
3618 |
|
---|
3619 | assert(correction >= 0);
|
---|
3620 | snd_brk = (char*)(MORECORE(correction));
|
---|
3621 |
|
---|
3622 | if (snd_brk == (char*)(MORECORE_FAILURE)) {
|
---|
3623 | /*
|
---|
3624 | If can't allocate correction, try to at least find out current
|
---|
3625 | brk. It might be enough to proceed without failing.
|
---|
3626 | */
|
---|
3627 | correction = 0;
|
---|
3628 | snd_brk = (char*)(MORECORE(0));
|
---|
3629 | }
|
---|
3630 | else if (snd_brk < brk) {
|
---|
3631 | /*
|
---|
3632 | If the second call gives noncontiguous space even though
|
---|
3633 | it says it won't, the only course of action is to ignore
|
---|
3634 | results of second call, and conservatively estimate where
|
---|
3635 | the first call left us. Also set noncontiguous, so this
|
---|
3636 | won't happen again, leaving at most one hole.
|
---|
3637 |
|
---|
3638 | Note that this check is intrinsically incomplete. Because
|
---|
3639 | MORECORE is allowed to give more space than we ask for,
|
---|
3640 | there is no reliable way to detect a noncontiguity
|
---|
3641 | producing a forward gap for the second call.
|
---|
3642 | */
|
---|
3643 | snd_brk = brk + size;
|
---|
3644 | correction = 0;
|
---|
3645 | set_noncontiguous(av);
|
---|
3646 | }
|
---|
3647 | else {
|
---|
3648 | VALGRIND_MAKE_MEM_NOACCESS(snd_brk,correction);
|
---|
3649 | }
|
---|
3650 |
|
---|
3651 | }
|
---|
3652 |
|
---|
3653 | /* handle non-contiguous cases */
|
---|
3654 | else {
|
---|
3655 |
|
---|
3656 | #ifdef DNMALLOC_DEBUG
|
---|
3657 | fprintf(stderr, "Handle non-contiguous cases\n");
|
---|
3658 | #endif
|
---|
3659 |
|
---|
3660 | /* MORECORE/mmap must correctly align */
|
---|
3661 | assert(aligned_OK(brk));
|
---|
3662 |
|
---|
3663 | /* Find out current end of memory */
|
---|
3664 | if (snd_brk == (char*)(MORECORE_FAILURE)) {
|
---|
3665 | snd_brk = (char*)(MORECORE(0));
|
---|
3666 | av->sbrked_mem += snd_brk - brk - size;
|
---|
3667 | }
|
---|
3668 | #ifdef DNMALLOC_DEBUG
|
---|
3669 | fprintf(stderr, "Sbrked now %lu\n", (unsigned long)av->sbrked_mem);
|
---|
3670 | #endif
|
---|
3671 | }
|
---|
3672 |
|
---|
3673 | /* Adjust top based on results of second sbrk.
|
---|
3674 | *
|
---|
3675 | * If mmap() has been used as backup for failed morecore(),
|
---|
3676 | * we end up in this branch as well.
|
---|
3677 | */
|
---|
3678 | if (snd_brk != (char*)(MORECORE_FAILURE)) {
|
---|
3679 | #ifdef DNMALLOC_DEBUG
|
---|
3680 | fprintf(stderr, "Adjust top, correction %lu\n", correction);
|
---|
3681 | #endif
|
---|
3682 | /* hashtable_remove(chunk(av->top)); *//* rw 19.05.2008 removed */
|
---|
3683 | av->top = cireg_getfree();
|
---|
3684 | av->top->chunk = (mchunkptr)aligned_brk;
|
---|
3685 | set_head(av->top, (snd_brk - aligned_brk + correction) | PREV_INUSE);
|
---|
3686 | #ifdef DNMALLOC_DEBUG
|
---|
3687 | fprintf(stderr, "Adjust top, top %p size %lu\n",
|
---|
3688 | av->top, (unsigned long)chunksize(av->top));
|
---|
3689 | #endif
|
---|
3690 | hashtable_add(av->top);
|
---|
3691 | av->sbrked_mem += correction;
|
---|
3692 |
|
---|
3693 | /*
|
---|
3694 | If not the first time through, we either have a
|
---|
3695 | gap due to foreign sbrk or a non-contiguous region. Insert a
|
---|
3696 | double fencepost at old_top to prevent consolidation with space
|
---|
3697 | we don't own. These fenceposts are artificial chunks that are
|
---|
3698 | marked as inuse. Original dlmalloc had two of these but too
|
---|
3699 | small to use. To ensure that the linked lists contain a maximum
|
---|
3700 | of 8 elements we only use 1. Inuse is determined by the
|
---|
3701 | current rather than the next chunk anyway.
|
---|
3702 | */
|
---|
3703 |
|
---|
3704 | if (old_size != 0) {
|
---|
3705 | #ifdef DNMALLOC_DEBUG
|
---|
3706 | fprintf(stderr, "Shrink old_top to insert fenceposts\n");
|
---|
3707 | #endif
|
---|
3708 | /*
|
---|
3709 | Shrink old_top to insert fenceposts, keeping size a
|
---|
3710 | multiple of MALLOC_ALIGNMENT. We know there is at least
|
---|
3711 | enough space in old_top to do this.
|
---|
3712 | */
|
---|
3713 | #ifdef DNMALLOC_DEBUG
|
---|
3714 | fprintf(stderr, "Adjust top, old_top %p old_size before %lu\n",
|
---|
3715 | old_top, (unsigned long)old_size);
|
---|
3716 | #endif
|
---|
3717 | old_size = (old_size - 4*SIZE_SZ) & ~MALLOC_ALIGN_MASK;
|
---|
3718 | set_head(old_top, old_size | PREV_INUSE);
|
---|
3719 | #ifdef DNMALLOC_DEBUG
|
---|
3720 | fprintf(stderr, "Adjust top, old_size after %lu\n",
|
---|
3721 | (unsigned long)old_size);
|
---|
3722 | #endif
|
---|
3723 |
|
---|
3724 | /*
|
---|
3725 | Note that the following assignments completely overwrite
|
---|
3726 | old_top when old_size was previously MINSIZE. This is
|
---|
3727 | intentional. We need the fencepost, even if old_top otherwise gets
|
---|
3728 | lost.
|
---|
3729 | */
|
---|
3730 | /* dnmalloc, we need the fencepost to be 16 bytes, however since
|
---|
3731 | it's marked inuse it will never be coalesced
|
---|
3732 | */
|
---|
3733 | fencepost = cireg_getfree();
|
---|
3734 | fencepost->chunk = (mchunkptr) chunk_at_offset(chunk(old_top),
|
---|
3735 | old_size);
|
---|
3736 | fencepost->size = 16|INUSE|PREV_INUSE;
|
---|
3737 | hashtable_add(fencepost);
|
---|
3738 | /*
|
---|
3739 | If possible, release the rest, suppressing trimming.
|
---|
3740 | */
|
---|
3741 | if (old_size >= MINSIZE) {
|
---|
3742 | INTERNAL_SIZE_T tt = av->trim_threshold;
|
---|
3743 | #ifdef DNMALLOC_DEBUG
|
---|
3744 | fprintf(stderr, "Release\n");
|
---|
3745 | #endif
|
---|
3746 | av->trim_threshold = (INTERNAL_SIZE_T)(-1);
|
---|
3747 | set_head(old_top, old_size | PREV_INUSE | INUSE);
|
---|
3748 | guard_set(av->guard_stored, old_top, 0, old_size);
|
---|
3749 | VALGRIND_MALLOCLIKE_BLOCK(chunk(old_top), old_size, 0, 0);
|
---|
3750 | fREe(chunk(old_top));
|
---|
3751 | av->trim_threshold = tt;
|
---|
3752 | #ifdef DNMALLOC_DEBUG
|
---|
3753 | fprintf(stderr, "Release done\n");
|
---|
3754 | #endif
|
---|
3755 | }
|
---|
3756 |
|
---|
3757 | #ifdef DNMALLOC_DEBUG
|
---|
3758 | fprintf(stderr, "Adjust top, size %lu\n",
|
---|
3759 | (unsigned long)chunksize(av->top));
|
---|
3760 | #endif
|
---|
3761 |
|
---|
3762 | } /* fenceposts */
|
---|
3763 | } /* adjust top */
|
---|
3764 | } /* not extended previous region */
|
---|
3765 |
|
---|
3766 | /* Update statistics */ /* FIXME check this */
|
---|
3767 | sum = av->sbrked_mem;
|
---|
3768 | if (sum > (CHUNK_SIZE_T)(av->max_sbrked_mem))
|
---|
3769 | av->max_sbrked_mem = sum;
|
---|
3770 |
|
---|
3771 | sum += av->mmapped_mem;
|
---|
3772 | if (sum > (CHUNK_SIZE_T)(av->max_total_mem))
|
---|
3773 | av->max_total_mem = sum;
|
---|
3774 |
|
---|
3775 | check_malloc_state();
|
---|
3776 |
|
---|
3777 | /* finally, do the allocation */
|
---|
3778 |
|
---|
3779 | p = av->top;
|
---|
3780 | size = chunksize(p);
|
---|
3781 |
|
---|
3782 | #ifdef DNMALLOC_DEBUG
|
---|
3783 | fprintf(stderr, "Size: %lu nb+MINSIZE: %lu\n",
|
---|
3784 | (CHUNK_SIZE_T)(size), (CHUNK_SIZE_T)(nb + MINSIZE));
|
---|
3785 | #endif
|
---|
3786 |
|
---|
3787 | /* check that one of the above allocation paths succeeded */
|
---|
3788 | if ((CHUNK_SIZE_T)(size) >= (CHUNK_SIZE_T)(nb + MINSIZE)) {
|
---|
3789 | remainder_size = size - nb;
|
---|
3790 | remainder = cireg_getfree();
|
---|
3791 | remainder->chunk = chunk_at_offset(chunk(p), nb);
|
---|
3792 | av->top = remainder;
|
---|
3793 | set_head(p, nb | PREV_INUSE | INUSE);
|
---|
3794 | set_head(remainder, remainder_size | PREV_INUSE);
|
---|
3795 | hashtable_insert (p, av->top);
|
---|
3796 | check_malloced_chunk(p, nb);
|
---|
3797 | #ifdef DNMALLOC_DEBUG
|
---|
3798 | fprintf(stderr, "Return any (total %lu)\n",
|
---|
3799 | (unsigned long)/* size_t */av->max_total_mem );
|
---|
3800 | #endif
|
---|
3801 | return chunk(p);
|
---|
3802 | }
|
---|
3803 |
|
---|
3804 | }
|
---|
3805 |
|
---|
3806 | #ifdef DNMALLOC_DEBUG
|
---|
3807 | fprintf(stderr, "Return failed (total %lu)\n",
|
---|
3808 | (unsigned long)/* size_t */av->max_total_mem );
|
---|
3809 | #endif
|
---|
3810 |
|
---|
3811 | /* catch all failure paths */
|
---|
3812 | MALLOC_FAILURE_ACTION;
|
---|
3813 | return 0;
|
---|
3814 | }
|
---|
3815 |
|
---|
3816 |
|
---|
3817 |
|
---|
3818 |
|
---|
3819 | /*
|
---|
3820 | sYSTRIm is an inverse of sorts to sYSMALLOc. It gives memory back
|
---|
3821 | to the system (via negative arguments to sbrk) if there is unused
|
---|
3822 | memory at the `high' end of the malloc pool. It is called
|
---|
3823 | automatically by free() when top space exceeds the trim
|
---|
3824 | threshold. It is also called by the public malloc_trim routine. It
|
---|
3825 | returns 1 if it actually released any memory, else 0.
|
---|
3826 | */
|
---|
3827 |
|
---|
3828 | #if __STD_C
|
---|
3829 | static int sYSTRIm(size_t pad, mstate av)
|
---|
3830 | #else
|
---|
3831 | static int sYSTRIm(pad, av) size_t pad; mstate av;
|
---|
3832 | #endif
|
---|
3833 | {
|
---|
3834 | long top_size; /* Amount of top-most memory */
|
---|
3835 | long extra; /* Amount to release */
|
---|
3836 | long released; /* Amount actually released */
|
---|
3837 | char* current_brk; /* address returned by pre-check sbrk call */
|
---|
3838 | char* new_brk; /* address returned by post-check sbrk call */
|
---|
3839 | size_t pagesz;
|
---|
3840 |
|
---|
3841 | pagesz = av->pagesize;
|
---|
3842 | top_size = chunksize(av->top);
|
---|
3843 |
|
---|
3844 | /* Release in pagesize units, keeping at least one page */
|
---|
3845 | extra = ((top_size - pad - MINSIZE + (pagesz-1)) / pagesz - 1) * pagesz;
|
---|
3846 |
|
---|
3847 | if (extra > 0) {
|
---|
3848 |
|
---|
3849 | /*
|
---|
3850 | Only proceed if end of memory is where we last set it.
|
---|
3851 | This avoids problems if there were foreign sbrk calls.
|
---|
3852 | */
|
---|
3853 | current_brk = (char*)(MORECORE(0));
|
---|
3854 | if (current_brk == (char*)(av->top) + top_size) {
|
---|
3855 |
|
---|
3856 | /*
|
---|
3857 | Attempt to release memory. We ignore MORECORE return value,
|
---|
3858 | and instead call again to find out where new end of memory is.
|
---|
3859 | This avoids problems if first call releases less than we asked,
|
---|
3860 | of if failure somehow altered brk value. (We could still
|
---|
3861 | encounter problems if it altered brk in some very bad way,
|
---|
3862 | but the only thing we can do is adjust anyway, which will cause
|
---|
3863 | some downstream failure.)
|
---|
3864 | */
|
---|
3865 |
|
---|
3866 | MORECORE(-extra);
|
---|
3867 | new_brk = (char*)(MORECORE(0));
|
---|
3868 |
|
---|
3869 | if (new_brk != (char*)MORECORE_FAILURE) {
|
---|
3870 | released = (long)(current_brk - new_brk);
|
---|
3871 |
|
---|
3872 | if (released != 0) {
|
---|
3873 | /* Success. Adjust top. */
|
---|
3874 | av->sbrked_mem -= released;
|
---|
3875 | set_head(av->top, (top_size - released) | PREV_INUSE);
|
---|
3876 | check_malloc_state();
|
---|
3877 | return 1;
|
---|
3878 | }
|
---|
3879 | }
|
---|
3880 | }
|
---|
3881 | }
|
---|
3882 | return 0;
|
---|
3883 | }
|
---|
3884 |
|
---|
3885 | /*
|
---|
3886 | ------------------------------ malloc ------------------------------
|
---|
3887 | */
|
---|
3888 |
|
---|
3889 |
|
---|
3890 | #if __STD_C
|
---|
3891 | DL_STATIC Void_t* mALLOc(size_t bytes)
|
---|
3892 | #else
|
---|
3893 | DL_STATIC Void_t* mALLOc(bytes) size_t bytes;
|
---|
3894 | #endif
|
---|
3895 | {
|
---|
3896 | mstate av = get_malloc_state();
|
---|
3897 |
|
---|
3898 | INTERNAL_SIZE_T nb; /* normalized request size */
|
---|
3899 | unsigned int idx; /* associated bin index */
|
---|
3900 | mbinptr bin; /* associated bin */
|
---|
3901 | mfastbinptr* fb; /* associated fastbin */
|
---|
3902 |
|
---|
3903 | chunkinfoptr victim; /* inspected/selected chunk */
|
---|
3904 | INTERNAL_SIZE_T size; /* its size */
|
---|
3905 | int victim_index; /* its bin index */
|
---|
3906 |
|
---|
3907 | chunkinfoptr remainder; /* remainder from a split */
|
---|
3908 | CHUNK_SIZE_T remainder_size; /* its size */
|
---|
3909 |
|
---|
3910 | unsigned int block; /* bit map traverser */
|
---|
3911 | unsigned int bit; /* bit map traverser */
|
---|
3912 | unsigned int map; /* current word of binmap */
|
---|
3913 |
|
---|
3914 | chunkinfoptr fwd; /* misc temp for linking */
|
---|
3915 | chunkinfoptr bck; /* misc temp for linking */
|
---|
3916 |
|
---|
3917 | Void_t* retval;
|
---|
3918 |
|
---|
3919 | /* chunkinfoptr next; */
|
---|
3920 |
|
---|
3921 |
|
---|
3922 | /*
|
---|
3923 | Convert request size to internal form by adding SIZE_SZ bytes
|
---|
3924 | overhead plus possibly more to obtain necessary alignment and/or
|
---|
3925 | to obtain a size of at least MINSIZE, the smallest allocatable
|
---|
3926 | size. Also, checked_request2size traps (returning 0) request sizes
|
---|
3927 | that are so large that they wrap around zero when padded and
|
---|
3928 | aligned.
|
---|
3929 | */
|
---|
3930 | #if defined(SH_CUTEST)
|
---|
3931 | extern int malloc_count;
|
---|
3932 | ++malloc_count;
|
---|
3933 | #endif
|
---|
3934 |
|
---|
3935 | checked_request2size(bytes, nb);
|
---|
3936 |
|
---|
3937 | /*
|
---|
3938 | Bypass search if no frees yet
|
---|
3939 | */
|
---|
3940 | if (av && have_anychunks(av)) {
|
---|
3941 | goto av_initialized;
|
---|
3942 | }
|
---|
3943 | else {
|
---|
3944 | if (!av || av->max_fast == 0) { /* initialization check */
|
---|
3945 | malloc_consolidate(av);
|
---|
3946 | av = get_malloc_state();
|
---|
3947 | }
|
---|
3948 | goto use_top;
|
---|
3949 | }
|
---|
3950 |
|
---|
3951 | av_initialized:
|
---|
3952 |
|
---|
3953 | /*
|
---|
3954 | If the size qualifies as a fastbin, first check corresponding bin.
|
---|
3955 | */
|
---|
3956 | if ((CHUNK_SIZE_T)(nb) <= (CHUNK_SIZE_T)(av->max_fast)) {
|
---|
3957 | fb = &(av->fastbins[(fastbin_index(nb))]);
|
---|
3958 | if ( (victim = *fb) != 0) {
|
---|
3959 | *fb = victim->fd;
|
---|
3960 | check_remalloced_chunk(victim, nb);
|
---|
3961 | guard_set(av->guard_stored, victim, bytes, nb);
|
---|
3962 | VALGRIND_MALLOCLIKE_BLOCK(chunk(victim), bytes, 0, 0);
|
---|
3963 | return chunk(victim);
|
---|
3964 | }
|
---|
3965 | }
|
---|
3966 |
|
---|
3967 | /*
|
---|
3968 | If a small request, check regular bin. Since these "smallbins"
|
---|
3969 | hold one size each, no searching within bins is necessary.
|
---|
3970 | (For a large request, we need to wait until unsorted chunks are
|
---|
3971 | processed to find best fit. But for small ones, fits are exact
|
---|
3972 | anyway, so we can check now, which is faster.)
|
---|
3973 | */
|
---|
3974 |
|
---|
3975 | if (in_smallbin_range(nb)) {
|
---|
3976 | idx = smallbin_index(nb);
|
---|
3977 | bin = bin_at(av,idx);
|
---|
3978 |
|
---|
3979 | if ((victim = last(bin)) != bin) {
|
---|
3980 | bck = victim->bk;
|
---|
3981 | bin->bk = bck;
|
---|
3982 | bck->fd = bin;
|
---|
3983 |
|
---|
3984 | set_all_inuse(victim);
|
---|
3985 |
|
---|
3986 | check_malloced_chunk(victim, nb);
|
---|
3987 | guard_set(av->guard_stored, victim, bytes, nb);
|
---|
3988 | VALGRIND_MALLOCLIKE_BLOCK(chunk(victim), bytes, 0, 0);
|
---|
3989 | return chunk(victim);
|
---|
3990 | }
|
---|
3991 | }
|
---|
3992 |
|
---|
3993 | /*
|
---|
3994 | If this is a large request, consolidate fastbins before continuing.
|
---|
3995 | While it might look excessive to kill all fastbins before
|
---|
3996 | even seeing if there is space available, this avoids
|
---|
3997 | fragmentation problems normally associated with fastbins.
|
---|
3998 | Also, in practice, programs tend to have runs of either small or
|
---|
3999 | large requests, but less often mixtures, so consolidation is not
|
---|
4000 | invoked all that often in most programs. And the programs that
|
---|
4001 | it is called frequently in otherwise tend to fragment.
|
---|
4002 | */
|
---|
4003 |
|
---|
4004 | else {
|
---|
4005 | idx = largebin_index(nb);
|
---|
4006 | if (have_fastchunks(av))
|
---|
4007 | malloc_consolidate(av);
|
---|
4008 | }
|
---|
4009 |
|
---|
4010 | /*
|
---|
4011 | Process recently freed or remaindered chunks, taking one only if
|
---|
4012 | it is exact fit, or, if this a small request, the chunk is remainder from
|
---|
4013 | the most recent non-exact fit. Place other traversed chunks in
|
---|
4014 | bins. Note that this step is the only place in any routine where
|
---|
4015 | chunks are placed in bins.
|
---|
4016 | */
|
---|
4017 |
|
---|
4018 | while ( (victim = unsorted_chunks(av)->bk) != unsorted_chunks(av)) {
|
---|
4019 | bck = victim->bk;
|
---|
4020 | size = chunksize(victim);
|
---|
4021 |
|
---|
4022 | /*
|
---|
4023 | If a small request, try to use last remainder if it is the
|
---|
4024 | only chunk in unsorted bin. This helps promote locality for
|
---|
4025 | runs of consecutive small requests. This is the only
|
---|
4026 | exception to best-fit, and applies only when there is
|
---|
4027 | no exact fit for a small chunk.
|
---|
4028 | */
|
---|
4029 |
|
---|
4030 | if (UNLIKELY(in_smallbin_range(nb) &&
|
---|
4031 | bck == unsorted_chunks(av) &&
|
---|
4032 | victim == av->last_remainder &&
|
---|
4033 | (CHUNK_SIZE_T)(size) > (CHUNK_SIZE_T)(nb + MINSIZE))) {
|
---|
4034 |
|
---|
4035 | /* split and reattach remainder */
|
---|
4036 | remainder_size = size - nb;
|
---|
4037 | remainder = cireg_getfree();
|
---|
4038 | remainder->chunk = chunk_at_offset(chunk(victim), nb);
|
---|
4039 | unsorted_chunks(av)->bk = unsorted_chunks(av)->fd = remainder;
|
---|
4040 | av->last_remainder = remainder;
|
---|
4041 | remainder->bk = remainder->fd = unsorted_chunks(av);
|
---|
4042 |
|
---|
4043 | set_head(victim, nb | PREV_INUSE|INUSE);
|
---|
4044 | set_head(remainder, remainder_size | PREV_INUSE);
|
---|
4045 | hashtable_insert(victim, remainder);
|
---|
4046 |
|
---|
4047 | check_malloced_chunk(victim, nb);
|
---|
4048 | guard_set(av->guard_stored, victim, bytes, nb);
|
---|
4049 | VALGRIND_MALLOCLIKE_BLOCK(chunk(victim), bytes, 0, 0);
|
---|
4050 | return chunk(victim);
|
---|
4051 | }
|
---|
4052 |
|
---|
4053 | /* remove from unsorted list */
|
---|
4054 | unsorted_chunks(av)->bk = bck;
|
---|
4055 | bck->fd = unsorted_chunks(av);
|
---|
4056 |
|
---|
4057 | /* Take now instead of binning if exact fit */
|
---|
4058 |
|
---|
4059 | if (UNLIKELY(size == nb)) {
|
---|
4060 | set_all_inuse(victim)
|
---|
4061 | check_malloced_chunk(victim, nb);
|
---|
4062 | guard_set(av->guard_stored, victim, bytes, nb);
|
---|
4063 | VALGRIND_MALLOCLIKE_BLOCK(chunk(victim), bytes, 0, 0);
|
---|
4064 | return chunk(victim);
|
---|
4065 | }
|
---|
4066 |
|
---|
4067 | /* place chunk in bin */
|
---|
4068 |
|
---|
4069 | if (in_smallbin_range(size)) {
|
---|
4070 |
|
---|
4071 | victim_index = smallbin_index(size);
|
---|
4072 | bck = bin_at(av, victim_index);
|
---|
4073 | fwd = bck->fd;
|
---|
4074 | }
|
---|
4075 | else {
|
---|
4076 | victim_index = largebin_index(size);
|
---|
4077 | bck = bin_at(av, victim_index);
|
---|
4078 | fwd = bck->fd;
|
---|
4079 |
|
---|
4080 | if (UNLIKELY(fwd != bck)) {
|
---|
4081 | /* if smaller than smallest, place first */
|
---|
4082 | if ((CHUNK_SIZE_T)(size) < (CHUNK_SIZE_T)(bck->bk->size)) {
|
---|
4083 | fwd = bck;
|
---|
4084 | bck = bck->bk;
|
---|
4085 | }
|
---|
4086 | else if ((CHUNK_SIZE_T)(size) >=
|
---|
4087 | (CHUNK_SIZE_T)(FIRST_SORTED_BIN_SIZE)) {
|
---|
4088 |
|
---|
4089 | /* maintain large bins in sorted order */
|
---|
4090 | size |= PREV_INUSE|INUSE; /* Or with inuse bits to speed comparisons */
|
---|
4091 | while ((CHUNK_SIZE_T)(size) < (CHUNK_SIZE_T)(fwd->size))
|
---|
4092 | fwd = fwd->fd;
|
---|
4093 | bck = fwd->bk;
|
---|
4094 | }
|
---|
4095 | }
|
---|
4096 | }
|
---|
4097 |
|
---|
4098 | mark_bin(av, victim_index);
|
---|
4099 | victim->bk = bck;
|
---|
4100 | victim->fd = fwd;
|
---|
4101 | fwd->bk = victim;
|
---|
4102 | bck->fd = victim;
|
---|
4103 | }
|
---|
4104 |
|
---|
4105 | /*
|
---|
4106 | If a large request, scan through the chunks of current bin to
|
---|
4107 | find one that fits. (This will be the smallest that fits unless
|
---|
4108 | FIRST_SORTED_BIN_SIZE has been changed from default.) This is
|
---|
4109 | the only step where an unbounded number of chunks might be
|
---|
4110 | scanned without doing anything useful with them. However the
|
---|
4111 | lists tend to be short.
|
---|
4112 | */
|
---|
4113 |
|
---|
4114 | if (!in_smallbin_range(nb)) {
|
---|
4115 | bin = bin_at(av, idx);
|
---|
4116 |
|
---|
4117 | victim = last(bin);
|
---|
4118 |
|
---|
4119 | if (UNLIKELY(victim != bin)) {
|
---|
4120 |
|
---|
4121 | do {
|
---|
4122 | size = chunksize(victim);
|
---|
4123 |
|
---|
4124 | if ((CHUNK_SIZE_T)(size) >= (CHUNK_SIZE_T)(nb)) {
|
---|
4125 | remainder_size = size - nb;
|
---|
4126 | unlink(victim, bck, fwd);
|
---|
4127 |
|
---|
4128 | /* Split */
|
---|
4129 | if (remainder_size >= MINSIZE) {
|
---|
4130 | remainder = cireg_getfree();
|
---|
4131 | remainder->chunk = chunk_at_offset(chunk(victim), nb);
|
---|
4132 | unsorted_chunks(av)->bk = unsorted_chunks(av)->fd = remainder;
|
---|
4133 | remainder->bk = remainder->fd = unsorted_chunks(av);
|
---|
4134 | set_head(victim, nb | PREV_INUSE | INUSE);
|
---|
4135 | set_head(remainder, remainder_size | PREV_INUSE);
|
---|
4136 | hashtable_insert(victim, remainder);
|
---|
4137 | check_malloced_chunk(victim, nb);
|
---|
4138 | guard_set(av->guard_stored, victim, bytes, nb);
|
---|
4139 | VALGRIND_MALLOCLIKE_BLOCK(chunk(victim), bytes, 0, 0);
|
---|
4140 | return chunk(victim);
|
---|
4141 | }
|
---|
4142 | /* Exhaust */
|
---|
4143 | else {
|
---|
4144 | set_all_inuse(victim);
|
---|
4145 | check_malloced_chunk(victim, nb);
|
---|
4146 | guard_set(av->guard_stored, victim, bytes, nb);
|
---|
4147 | VALGRIND_MALLOCLIKE_BLOCK(chunk(victim), bytes, 0, 0);
|
---|
4148 | return chunk(victim);
|
---|
4149 | }
|
---|
4150 | }
|
---|
4151 | victim = victim->bk;
|
---|
4152 | } while(victim != bin);
|
---|
4153 | }
|
---|
4154 | }
|
---|
4155 |
|
---|
4156 | /*
|
---|
4157 | Search for a chunk by scanning bins, starting with next largest
|
---|
4158 | bin. This search is strictly by best-fit; i.e., the smallest
|
---|
4159 | (with ties going to approximately the least recently used) chunk
|
---|
4160 | that fits is selected.
|
---|
4161 |
|
---|
4162 | The bitmap avoids needing to check that most blocks are nonempty.
|
---|
4163 | */
|
---|
4164 |
|
---|
4165 |
|
---|
4166 | ++idx;
|
---|
4167 | bin = bin_at(av,idx);
|
---|
4168 | block = idx2block(idx);
|
---|
4169 | map = av->binmap[block];
|
---|
4170 | bit = idx2bit(idx);
|
---|
4171 |
|
---|
4172 | for (;;) {
|
---|
4173 |
|
---|
4174 | /* Skip rest of block if there are no more set bits in this block. */
|
---|
4175 | if (bit > map || bit == 0) {
|
---|
4176 | do {
|
---|
4177 | if (++block >= BINMAPSIZE) /* out of bins */
|
---|
4178 | goto use_top;
|
---|
4179 | } while ( (map = av->binmap[block]) == 0);
|
---|
4180 |
|
---|
4181 | bin = bin_at(av, (block << BINMAPSHIFT));
|
---|
4182 | bit = 1;
|
---|
4183 | }
|
---|
4184 |
|
---|
4185 | /* Advance to bin with set bit. There must be one. */
|
---|
4186 | while ((bit & map) == 0) {
|
---|
4187 | bin = next_bin(bin);
|
---|
4188 | bit <<= 1;
|
---|
4189 | assert(bit != 0);
|
---|
4190 | }
|
---|
4191 |
|
---|
4192 | /* Inspect the bin. It is likely to be non-empty */
|
---|
4193 | victim = last(bin);
|
---|
4194 |
|
---|
4195 | /* If a false alarm (empty bin), clear the bit. */
|
---|
4196 | if (victim == bin) {
|
---|
4197 | av->binmap[block] = map &= ~bit; /* Write through */
|
---|
4198 | bin = next_bin(bin);
|
---|
4199 | bit <<= 1;
|
---|
4200 | }
|
---|
4201 |
|
---|
4202 | else {
|
---|
4203 | size = chunksize(victim);
|
---|
4204 |
|
---|
4205 | /* We know the first chunk in this bin is big enough to use. */
|
---|
4206 | assert((CHUNK_SIZE_T)(size) >= (CHUNK_SIZE_T)(nb));
|
---|
4207 |
|
---|
4208 | remainder_size = size - nb;
|
---|
4209 |
|
---|
4210 | /* unlink */
|
---|
4211 | bck = victim->bk;
|
---|
4212 | bin->bk = bck;
|
---|
4213 | bck->fd = bin;
|
---|
4214 |
|
---|
4215 | /* Split */
|
---|
4216 | if (remainder_size >= MINSIZE) {
|
---|
4217 | remainder = cireg_getfree();
|
---|
4218 | remainder->chunk = chunk_at_offset(chunk(victim), nb);
|
---|
4219 |
|
---|
4220 | unsorted_chunks(av)->bk = unsorted_chunks(av)->fd = remainder;
|
---|
4221 | remainder->bk = remainder->fd = unsorted_chunks(av);
|
---|
4222 | /* advertise as last remainder */
|
---|
4223 | if (in_smallbin_range(nb))
|
---|
4224 | av->last_remainder = remainder;
|
---|
4225 |
|
---|
4226 | set_head(victim, nb | PREV_INUSE | INUSE);
|
---|
4227 | set_head(remainder, remainder_size | PREV_INUSE);
|
---|
4228 | hashtable_insert(victim, remainder);
|
---|
4229 | check_malloced_chunk(victim, nb);
|
---|
4230 | guard_set(av->guard_stored, victim, bytes, nb);
|
---|
4231 | VALGRIND_MALLOCLIKE_BLOCK(chunk(victim), bytes, 0, 0);
|
---|
4232 | return chunk(victim);
|
---|
4233 | }
|
---|
4234 | /* Exhaust */
|
---|
4235 | else {
|
---|
4236 | set_all_inuse(victim);
|
---|
4237 | check_malloced_chunk(victim, nb);
|
---|
4238 | guard_set(av->guard_stored, victim, bytes, nb);
|
---|
4239 | VALGRIND_MALLOCLIKE_BLOCK(chunk(victim), bytes, 0, 0);
|
---|
4240 | return chunk(victim);
|
---|
4241 | }
|
---|
4242 |
|
---|
4243 | }
|
---|
4244 | }
|
---|
4245 |
|
---|
4246 | use_top:
|
---|
4247 |
|
---|
4248 |
|
---|
4249 | /*
|
---|
4250 | If large enough, split off the chunk bordering the end of memory
|
---|
4251 | (held in av->top). Note that this is in accord with the best-fit
|
---|
4252 | search rule. In effect, av->top is treated as larger (and thus
|
---|
4253 | less well fitting) than any other available chunk since it can
|
---|
4254 | be extended to be as large as necessary (up to system
|
---|
4255 | limitations).
|
---|
4256 |
|
---|
4257 | We require that av->top always exists (i.e., has size >=
|
---|
4258 | MINSIZE) after initialization, so if it would otherwise be
|
---|
4259 | exhuasted by current request, it is replenished. (The main
|
---|
4260 | reason for ensuring it exists is that we may need MINSIZE space
|
---|
4261 | to put in fenceposts in sysmalloc.)
|
---|
4262 | */
|
---|
4263 |
|
---|
4264 | victim = av->top;
|
---|
4265 | size = chunksize(victim);
|
---|
4266 |
|
---|
4267 | if ((CHUNK_SIZE_T)(size) >= (CHUNK_SIZE_T)(nb + MINSIZE)) {
|
---|
4268 | remainder = cireg_getfree();
|
---|
4269 | remainder_size = size - nb;
|
---|
4270 | remainder->chunk = chunk_at_offset(chunk(victim), nb);
|
---|
4271 | av->top = remainder;
|
---|
4272 | set_head(victim, nb | PREV_INUSE | INUSE);
|
---|
4273 | set_head(remainder, remainder_size | PREV_INUSE);
|
---|
4274 | hashtable_insert(victim, remainder);
|
---|
4275 | check_malloced_chunk(victim, nb);
|
---|
4276 | guard_set(av->guard_stored, victim, bytes, nb);
|
---|
4277 | VALGRIND_MALLOCLIKE_BLOCK(chunk(victim), bytes, 0, 0);
|
---|
4278 | return chunk(victim);
|
---|
4279 | }
|
---|
4280 |
|
---|
4281 | /*
|
---|
4282 | If no space in top, relay to handle system-dependent cases
|
---|
4283 | */
|
---|
4284 | retval = sYSMALLOc(nb, av);
|
---|
4285 | if (retval) {
|
---|
4286 | victim = mem2chunk(retval);
|
---|
4287 | guard_set(av->guard_stored, victim, bytes, nb);
|
---|
4288 | }
|
---|
4289 | VALGRIND_MALLOCLIKE_BLOCK(retval, bytes, 0, 0);
|
---|
4290 | return retval;
|
---|
4291 | }
|
---|
4292 |
|
---|
4293 | /*
|
---|
4294 | ------------------------------ free ------------------------------
|
---|
4295 | */
|
---|
4296 |
|
---|
4297 | #if __STD_C
|
---|
4298 | DL_STATIC void fREe(Void_t* mem)
|
---|
4299 | #else
|
---|
4300 | DL_STATIC void fREe(mem) Void_t* mem;
|
---|
4301 | #endif
|
---|
4302 | {
|
---|
4303 | mstate av = get_malloc_state();
|
---|
4304 |
|
---|
4305 | chunkinfoptr p; /* chunk corresponding to mem */
|
---|
4306 | INTERNAL_SIZE_T size; /* its size */
|
---|
4307 | mfastbinptr* fb; /* associated fastbin */
|
---|
4308 | chunkinfoptr prevchunk; /* previous physical chunk */
|
---|
4309 | chunkinfoptr nextchunk; /* next contiguous chunk */
|
---|
4310 | INTERNAL_SIZE_T nextsize; /* its size */
|
---|
4311 | INTERNAL_SIZE_T prevsize; /* size of previous contiguous chunk */
|
---|
4312 | chunkinfoptr bck; /* misc temp for linking */
|
---|
4313 | chunkinfoptr fwd; /* misc temp for linking */
|
---|
4314 | chunkinfoptr next;
|
---|
4315 | #if defined(SH_CUTEST)
|
---|
4316 | extern int malloc_count;
|
---|
4317 | --malloc_count;
|
---|
4318 | #endif
|
---|
4319 |
|
---|
4320 | /* free(0) has no effect */
|
---|
4321 | if (mem != 0) {
|
---|
4322 | p = hashtable_lookup(mem);
|
---|
4323 | /* check that memory is managed by us
|
---|
4324 | * and is inuse
|
---|
4325 | */
|
---|
4326 | if (UNLIKELY(!p || !inuse(p)))
|
---|
4327 | {
|
---|
4328 | #ifdef DNMALLOC_CHECKS
|
---|
4329 | if (p) {
|
---|
4330 | fprintf(stderr, "Attempt to free memory not in use\n");
|
---|
4331 | abort();
|
---|
4332 | } else {
|
---|
4333 | fprintf(stderr, "Attempt to free memory not allocated\n");
|
---|
4334 | abort();
|
---|
4335 | }
|
---|
4336 | #endif
|
---|
4337 | assert(p && inuse(p));
|
---|
4338 | return;
|
---|
4339 | }
|
---|
4340 |
|
---|
4341 | VALGRIND_FREELIKE_BLOCK(mem, 0);
|
---|
4342 |
|
---|
4343 | guard_check(av->guard_stored, p);
|
---|
4344 |
|
---|
4345 | size = chunksize(p);
|
---|
4346 |
|
---|
4347 | check_inuse_chunk(p);
|
---|
4348 |
|
---|
4349 | /*
|
---|
4350 | If eligible, place chunk on a fastbin so it can be found
|
---|
4351 | and used quickly in malloc.
|
---|
4352 | */
|
---|
4353 |
|
---|
4354 | if ((CHUNK_SIZE_T)(size) <= (CHUNK_SIZE_T)(av->max_fast)
|
---|
4355 |
|
---|
4356 | #if TRIM_FASTBINS
|
---|
4357 | /*
|
---|
4358 | If TRIM_FASTBINS set, don't place chunks
|
---|
4359 | bordering top into fastbins
|
---|
4360 | */
|
---|
4361 | && (chunk_at_offset(chunk(p), size) != av->top)
|
---|
4362 | #endif
|
---|
4363 | ) {
|
---|
4364 |
|
---|
4365 | set_fastchunks(av);
|
---|
4366 | fb = &(av->fastbins[fastbin_index(size)]);
|
---|
4367 | p->fd = *fb;
|
---|
4368 | *fb = p;
|
---|
4369 | }
|
---|
4370 |
|
---|
4371 | /*
|
---|
4372 | Consolidate other non-mmapped chunks as they arrive.
|
---|
4373 | */
|
---|
4374 |
|
---|
4375 | else if (!chunk_is_mmapped(p)) {
|
---|
4376 | set_anychunks(av);
|
---|
4377 |
|
---|
4378 | nextchunk = next_chunkinfo(p);
|
---|
4379 | if (nextchunk)
|
---|
4380 | nextsize = chunksize(nextchunk);
|
---|
4381 | else
|
---|
4382 | nextsize = 0;/* gcc doesn't notice that it's only used if (nextchunk)*/
|
---|
4383 |
|
---|
4384 | /* consolidate backward */
|
---|
4385 | if (UNLIKELY(!prev_inuse(p))) {
|
---|
4386 | prevchunk = prev_chunkinfo(p);
|
---|
4387 | prevsize = chunksize(prevchunk);
|
---|
4388 | #ifdef DNMALLOC_CHECKS
|
---|
4389 | if (inuse(prevchunk)) {
|
---|
4390 | fprintf(stderr, "Dnmalloc error: trying to unlink an inuse chunk: %p (chunk: %p)\n This is definitely a bug, please report it to dnmalloc@fort-knox.org.\n", prevchunk, chunk(prevchunk));
|
---|
4391 | abort();
|
---|
4392 | }
|
---|
4393 | #else
|
---|
4394 | assert(!inuse(prevchunk));
|
---|
4395 | #endif
|
---|
4396 | size += prevsize;
|
---|
4397 | unlink(prevchunk, bck, fwd);
|
---|
4398 | set_head(p, size | PREV_INUSE);
|
---|
4399 | hashtable_skiprm(prevchunk,p);
|
---|
4400 | /* This chunk no longer exists in any form: release the chunkinfoptr
|
---|
4401 | */
|
---|
4402 | freecilst_add(p);
|
---|
4403 | p = prevchunk;
|
---|
4404 | }
|
---|
4405 |
|
---|
4406 | if (nextchunk) {
|
---|
4407 | if (nextchunk != av->top) {
|
---|
4408 | /* get and clear inuse bit */
|
---|
4409 | clear_previnuse(nextchunk);
|
---|
4410 |
|
---|
4411 | /* consolidate forward */
|
---|
4412 | if (!inuse(nextchunk)) {
|
---|
4413 | /* if mmap is used instead of sbrk, we may have a
|
---|
4414 | * chunk with !nextchunk->fd && !nextchunk->bk
|
---|
4415 | */
|
---|
4416 | if (nextchunk->fd && nextchunk->fd)
|
---|
4417 | {
|
---|
4418 | unlink(nextchunk, bck, fwd);
|
---|
4419 | size += nextsize;
|
---|
4420 | set_head(p, size | PREV_INUSE);
|
---|
4421 | hashtable_skiprm(p, nextchunk);
|
---|
4422 | freecilst_add (nextchunk);
|
---|
4423 | }
|
---|
4424 | }
|
---|
4425 |
|
---|
4426 | set_head(p, size | PREV_INUSE);
|
---|
4427 | next = next_chunkinfo(p);
|
---|
4428 | if (next)
|
---|
4429 | next->prev_size = size;
|
---|
4430 |
|
---|
4431 | /*
|
---|
4432 | Place the chunk in unsorted chunk list. Chunks are
|
---|
4433 | not placed into regular bins until after they have
|
---|
4434 | been given one chance to be used in malloc.
|
---|
4435 | */
|
---|
4436 |
|
---|
4437 | bck = unsorted_chunks(av);
|
---|
4438 | fwd = bck->fd;
|
---|
4439 | p->bk = bck;
|
---|
4440 | p->fd = fwd;
|
---|
4441 | bck->fd = p;
|
---|
4442 | fwd->bk = p;
|
---|
4443 |
|
---|
4444 | nextchunk = next_chunkinfo(p);
|
---|
4445 | if (nextchunk)
|
---|
4446 | nextchunk->prev_size = chunksize(p);
|
---|
4447 |
|
---|
4448 | check_free_chunk(p);
|
---|
4449 | }
|
---|
4450 |
|
---|
4451 | /*
|
---|
4452 | If the chunk borders the current high end of memory,
|
---|
4453 | consolidate into top
|
---|
4454 | */
|
---|
4455 |
|
---|
4456 | else {
|
---|
4457 | size += nextsize;
|
---|
4458 | set_head(p, size | PREV_INUSE);
|
---|
4459 | hashtable_remove(chunk(av->top));
|
---|
4460 | freecilst_add(av->top);
|
---|
4461 | av->top = p;
|
---|
4462 | check_chunk(p);
|
---|
4463 | }
|
---|
4464 | } /* if (nextchunk) */
|
---|
4465 |
|
---|
4466 | /*
|
---|
4467 | If freeing a large space, consolidate possibly-surrounding
|
---|
4468 | chunks. Then, if the total unused topmost memory exceeds trim
|
---|
4469 | threshold, ask malloc_trim to reduce top.
|
---|
4470 |
|
---|
4471 | Unless max_fast is 0, we don't know if there are fastbins
|
---|
4472 | bordering top, so we cannot tell for sure whether threshold
|
---|
4473 | has been reached unless fastbins are consolidated. But we
|
---|
4474 | don't want to consolidate on each free. As a compromise,
|
---|
4475 | consolidation is performed if FASTBIN_CONSOLIDATION_THRESHOLD
|
---|
4476 | is reached.
|
---|
4477 | */
|
---|
4478 |
|
---|
4479 | if (UNLIKELY((CHUNK_SIZE_T)(size) >= FASTBIN_CONSOLIDATION_THRESHOLD)) {
|
---|
4480 | if (have_fastchunks(av))
|
---|
4481 | malloc_consolidate(av);
|
---|
4482 |
|
---|
4483 | #ifndef MORECORE_CANNOT_TRIM
|
---|
4484 | if ((CHUNK_SIZE_T)(chunksize(av->top)) >=
|
---|
4485 | (CHUNK_SIZE_T)(av->trim_threshold))
|
---|
4486 | {
|
---|
4487 | if (morecore32bit(av))
|
---|
4488 | {
|
---|
4489 | #ifdef DNMALLOC_DEBUG
|
---|
4490 | fprintf(stderr, "Calling systrim from free()\n");
|
---|
4491 | #endif
|
---|
4492 | sYSTRIm(av->top_pad, av);
|
---|
4493 | #ifdef DNMALLOC_DEBUG
|
---|
4494 | fprintf(stderr, "Systrim done\n");
|
---|
4495 | #endif
|
---|
4496 | }
|
---|
4497 | }
|
---|
4498 | #endif
|
---|
4499 | }
|
---|
4500 |
|
---|
4501 | }
|
---|
4502 | /*
|
---|
4503 | If the chunk was allocated via mmap, release via munmap()
|
---|
4504 | Note that if HAVE_MMAP is false but chunk_is_mmapped is
|
---|
4505 | true, then user must have overwritten memory. There's nothing
|
---|
4506 | we can do to catch this error unless DEBUG is set, in which case
|
---|
4507 | check_inuse_chunk (above) will have triggered error.
|
---|
4508 | */
|
---|
4509 |
|
---|
4510 | else {
|
---|
4511 | #if PARANOIA > 0
|
---|
4512 | int ret;
|
---|
4513 | #endif
|
---|
4514 | INTERNAL_SIZE_T offset = (INTERNAL_SIZE_T) p->hash_next;
|
---|
4515 | av->n_mmaps--;
|
---|
4516 | av->mmapped_mem -= (size + offset);
|
---|
4517 | #if PARANOIA > 0
|
---|
4518 | ret = munmap((char*) chunk(p) - offset, size + offset);
|
---|
4519 | #else
|
---|
4520 | munmap((char*) chunk(p) - offset, size + offset);
|
---|
4521 | #endif
|
---|
4522 | hashtable_remove_mmapped(chunk(p));
|
---|
4523 | freecilst_add(p);
|
---|
4524 | /* munmap returns non-zero on failure */
|
---|
4525 | assert(ret == 0);
|
---|
4526 | }
|
---|
4527 | }
|
---|
4528 | }
|
---|
4529 |
|
---|
4530 | /*
|
---|
4531 | ------------------------- malloc_consolidate -------------------------
|
---|
4532 |
|
---|
4533 | malloc_consolidate is a specialized version of free() that tears
|
---|
4534 | down chunks held in fastbins. Free itself cannot be used for this
|
---|
4535 | purpose since, among other things, it might place chunks back onto
|
---|
4536 | fastbins. So, instead, we need to use a minor variant of the same
|
---|
4537 | code.
|
---|
4538 |
|
---|
4539 | Also, because this routine needs to be called the first time through
|
---|
4540 | malloc anyway, it turns out to be the perfect place to trigger
|
---|
4541 | initialization code.
|
---|
4542 | */
|
---|
4543 |
|
---|
4544 | #if __STD_C
|
---|
4545 | static void malloc_consolidate(mstate av)
|
---|
4546 | #else
|
---|
4547 | static void malloc_consolidate(av) mstate av;
|
---|
4548 | #endif
|
---|
4549 | {
|
---|
4550 | mfastbinptr* fb; /* current fastbin being consolidated */
|
---|
4551 | mfastbinptr* maxfb; /* last fastbin (for loop control) */
|
---|
4552 | chunkinfoptr p; /* current chunk being consolidated */
|
---|
4553 | chunkinfoptr nextp; /* next chunk to consolidate */
|
---|
4554 | chunkinfoptr prevp;
|
---|
4555 | chunkinfoptr unsorted_bin; /* bin header */
|
---|
4556 | chunkinfoptr first_unsorted; /* chunk to link to */
|
---|
4557 |
|
---|
4558 | /* These have same use as in free() */
|
---|
4559 | chunkinfoptr nextchunk;
|
---|
4560 | INTERNAL_SIZE_T size;
|
---|
4561 | INTERNAL_SIZE_T nextsize;
|
---|
4562 | INTERNAL_SIZE_T prevsize;
|
---|
4563 | chunkinfoptr bck;
|
---|
4564 | chunkinfoptr fwd;
|
---|
4565 | chunkinfoptr next;
|
---|
4566 |
|
---|
4567 | /*
|
---|
4568 | If max_fast is 0, we know that av hasn't
|
---|
4569 | yet been initialized, in which case do so below
|
---|
4570 | */
|
---|
4571 | if (av && av->max_fast != 0) {
|
---|
4572 |
|
---|
4573 | clear_fastchunks(av);
|
---|
4574 |
|
---|
4575 | unsorted_bin = unsorted_chunks(av);
|
---|
4576 |
|
---|
4577 | /*
|
---|
4578 | Remove each chunk from fast bin and consolidate it, placing it
|
---|
4579 | then in unsorted bin. Among other reasons for doing this,
|
---|
4580 | placing in unsorted bin avoids needing to calculate actual bins
|
---|
4581 | until malloc is sure that chunks aren't immediately going to be
|
---|
4582 | reused anyway.
|
---|
4583 | */
|
---|
4584 |
|
---|
4585 | maxfb = &(av->fastbins[fastbin_index(av->max_fast)]);
|
---|
4586 | fb = &(av->fastbins[0]);
|
---|
4587 | do {
|
---|
4588 | if ( UNLIKELY((p = *fb) != 0)) {
|
---|
4589 | *fb = 0;
|
---|
4590 | do {
|
---|
4591 | check_inuse_chunk(p);
|
---|
4592 | nextp = p->fd;
|
---|
4593 |
|
---|
4594 | /*
|
---|
4595 | * Slightly streamlined version of consolidation code in free()
|
---|
4596 | */
|
---|
4597 |
|
---|
4598 | size = chunksize(p);
|
---|
4599 | nextchunk = next_chunkinfo(p);
|
---|
4600 |
|
---|
4601 | /* gcc doesn't notice that it's only used if (nextchunk) */
|
---|
4602 | if (nextchunk)
|
---|
4603 | nextsize = chunksize(nextchunk);
|
---|
4604 | else
|
---|
4605 | nextsize = 0;
|
---|
4606 |
|
---|
4607 | if (!prev_inuse(p)) {
|
---|
4608 | prevp = prev_chunkinfo(p);
|
---|
4609 | prevsize = chunksize(prevp);
|
---|
4610 | size += prevsize;
|
---|
4611 | #ifdef DNMALLOC_CHECKS
|
---|
4612 | if (inuse(prevp)) {
|
---|
4613 | fprintf(stderr, "Dnmalloc error: trying to unlink an inuse chunk (2): %p (chunk: %p)\n This is definitely a bug, please report it to dnmalloc@fort-knox.org.\n", prevp, chunk(prevp));
|
---|
4614 | abort();
|
---|
4615 | }
|
---|
4616 | #else
|
---|
4617 | assert(!inuse(prevp));
|
---|
4618 | #endif
|
---|
4619 | unlink(prevp, bck, fwd);
|
---|
4620 | set_head(p, size | PREV_INUSE);
|
---|
4621 | hashtable_skiprm(prevp,p);
|
---|
4622 | freecilst_add(p);
|
---|
4623 | p=prevp;
|
---|
4624 | }
|
---|
4625 |
|
---|
4626 | if (nextchunk) {
|
---|
4627 | if (nextchunk != av->top) {
|
---|
4628 |
|
---|
4629 | clear_previnuse(nextchunk);
|
---|
4630 |
|
---|
4631 | /* if mmap is used instead of sbrk, we may have a
|
---|
4632 | * chunk with !nextchunk->fd && !nextchunk->bk
|
---|
4633 | */
|
---|
4634 | if (!inuse(nextchunk)) {
|
---|
4635 | if( nextchunk->fd && nextchunk->bk) {
|
---|
4636 | size += nextsize;
|
---|
4637 | unlink(nextchunk, bck, fwd);
|
---|
4638 | set_head(p, size | PREV_INUSE);
|
---|
4639 | hashtable_skiprm(p,nextchunk);
|
---|
4640 | freecilst_add(nextchunk);
|
---|
4641 | }
|
---|
4642 | }
|
---|
4643 |
|
---|
4644 | first_unsorted = unsorted_bin->fd;
|
---|
4645 | unsorted_bin->fd = p;
|
---|
4646 | first_unsorted->bk = p;
|
---|
4647 |
|
---|
4648 | set_head(p, size | PREV_INUSE);
|
---|
4649 | p->bk = unsorted_bin;
|
---|
4650 | p->fd = first_unsorted;
|
---|
4651 | next = next_chunkinfo(p);
|
---|
4652 | if (next)
|
---|
4653 | next->prev_size = size;
|
---|
4654 |
|
---|
4655 |
|
---|
4656 | }
|
---|
4657 |
|
---|
4658 | else if (nextchunk == av->top) {
|
---|
4659 | size += nextsize;
|
---|
4660 | set_head(p, size | PREV_INUSE);
|
---|
4661 | hashtable_remove(chunk(av->top));
|
---|
4662 | freecilst_add(av->top);
|
---|
4663 | av->top = p;
|
---|
4664 | }
|
---|
4665 | } /* if (nextchunk) */
|
---|
4666 |
|
---|
4667 | } while ( (p = nextp) != 0);
|
---|
4668 |
|
---|
4669 | }
|
---|
4670 | } while (fb++ != maxfb);
|
---|
4671 | }
|
---|
4672 | else {
|
---|
4673 | /* Initialize dnmalloc */
|
---|
4674 | dnmalloc_init();
|
---|
4675 | malloc_init_state(get_malloc_state());
|
---|
4676 | check_malloc_state();
|
---|
4677 | }
|
---|
4678 | }
|
---|
4679 |
|
---|
4680 | /*
|
---|
4681 | ------------------------------ realloc ------------------------------
|
---|
4682 | */
|
---|
4683 |
|
---|
4684 |
|
---|
4685 | #if __STD_C
|
---|
4686 | DL_STATIC Void_t* rEALLOc(Void_t* oldmem, size_t bytes)
|
---|
4687 | #else
|
---|
4688 | DL_STATIC Void_t* rEALLOc(oldmem, bytes) Void_t* oldmem; size_t bytes;
|
---|
4689 | #endif
|
---|
4690 | {
|
---|
4691 | mstate av = get_malloc_state();
|
---|
4692 |
|
---|
4693 | INTERNAL_SIZE_T nb; /* padded request size */
|
---|
4694 |
|
---|
4695 | chunkinfoptr oldp; /* chunk corresponding to oldmem */
|
---|
4696 | INTERNAL_SIZE_T oldsize; /* its size */
|
---|
4697 |
|
---|
4698 | chunkinfoptr newp; /* chunk to return */
|
---|
4699 | INTERNAL_SIZE_T newsize; /* its size */
|
---|
4700 | Void_t* newmem; /* corresponding user mem */
|
---|
4701 |
|
---|
4702 | chunkinfoptr next; /* next contiguous chunk after oldp */
|
---|
4703 |
|
---|
4704 | chunkinfoptr remainder; /* extra space at end of newp */
|
---|
4705 | CHUNK_SIZE_T remainder_size; /* its size */
|
---|
4706 |
|
---|
4707 | chunkinfoptr bck; /* misc temp for linking */
|
---|
4708 | chunkinfoptr fwd; /* misc temp for linking */
|
---|
4709 |
|
---|
4710 | CHUNK_SIZE_T copysize; /* bytes to copy */
|
---|
4711 | unsigned int ncopies; /* INTERNAL_SIZE_T words to copy */
|
---|
4712 | INTERNAL_SIZE_T* s; /* copy source */
|
---|
4713 | INTERNAL_SIZE_T* d; /* copy destination */
|
---|
4714 |
|
---|
4715 |
|
---|
4716 | #ifdef REALLOC_ZERO_BYTES_FREES
|
---|
4717 | if (UNLIKELY(bytes == 0)) {
|
---|
4718 | fREe(oldmem);
|
---|
4719 | return 0;
|
---|
4720 | }
|
---|
4721 | #endif
|
---|
4722 |
|
---|
4723 | if (UNLIKELY(!av || av->max_fast == 0)) {
|
---|
4724 | malloc_consolidate(av);
|
---|
4725 | av = get_malloc_state();
|
---|
4726 | }
|
---|
4727 |
|
---|
4728 | /* realloc of null is supposed to be same as malloc */
|
---|
4729 | if (UNLIKELY(oldmem == 0))
|
---|
4730 | return mALLOc(bytes);
|
---|
4731 |
|
---|
4732 | checked_request2size(bytes, nb);
|
---|
4733 |
|
---|
4734 | oldp = hashtable_lookup(oldmem);
|
---|
4735 |
|
---|
4736 | if (UNLIKELY(!oldp || !inuse(oldp))){
|
---|
4737 | /* attempt to either realloc memory not managed by us
|
---|
4738 | * or memory that is not in use
|
---|
4739 | */
|
---|
4740 | #ifdef DNMALLOC_CHECKS
|
---|
4741 | if (oldp) {
|
---|
4742 | fprintf(stderr, "Attempt to free memory not in use\n");
|
---|
4743 | abort();
|
---|
4744 | } else {
|
---|
4745 | fprintf(stderr, "Attempt to free memory not allocated\n");
|
---|
4746 | abort();
|
---|
4747 | }
|
---|
4748 | #endif
|
---|
4749 | assert(oldp && inuse(oldp));
|
---|
4750 | return 0;
|
---|
4751 | }
|
---|
4752 |
|
---|
4753 | VALGRIND_FREELIKE_BLOCK(oldmem, 0);
|
---|
4754 | guard_check(av->guard_stored, oldp);
|
---|
4755 |
|
---|
4756 | oldsize = chunksize(oldp);
|
---|
4757 |
|
---|
4758 | check_inuse_chunk(oldp);
|
---|
4759 |
|
---|
4760 | if (!chunk_is_mmapped(oldp)) {
|
---|
4761 |
|
---|
4762 | if (UNLIKELY((CHUNK_SIZE_T)(oldsize) >= (CHUNK_SIZE_T)(nb))) {
|
---|
4763 | /* already big enough; split below */
|
---|
4764 | newp = oldp;
|
---|
4765 | newsize = oldsize;
|
---|
4766 | }
|
---|
4767 |
|
---|
4768 | else {
|
---|
4769 | next = next_chunkinfo(oldp);
|
---|
4770 | if (next)
|
---|
4771 | next->prev_size = oldsize;
|
---|
4772 | /* Try to expand forward into top */
|
---|
4773 | if (next && next == av->top &&
|
---|
4774 | (CHUNK_SIZE_T)(newsize = oldsize + chunksize(next)) >=
|
---|
4775 | (CHUNK_SIZE_T)(nb + MINSIZE)) {
|
---|
4776 | set_head_size(oldp, nb);
|
---|
4777 | hashtable_remove(chunk(av->top));
|
---|
4778 | av->top->chunk = chunk_at_offset(chunk(oldp), nb);
|
---|
4779 | set_head(av->top, (newsize - nb) | PREV_INUSE);
|
---|
4780 | /* av->top->chunk has been moved move in hashtable */
|
---|
4781 | hashtable_insert(oldp, av->top);
|
---|
4782 | guard_set(av->guard_stored, oldp, bytes, nb);
|
---|
4783 | VALGRIND_MALLOCLIKE_BLOCK(chunk(oldp), bytes, 0, 0);
|
---|
4784 | return chunk(oldp);
|
---|
4785 | }
|
---|
4786 |
|
---|
4787 | /* Try to expand forward into next chunk; split off remainder below */
|
---|
4788 | else if (next && next != av->top &&
|
---|
4789 | !inuse(next) &&
|
---|
4790 | (CHUNK_SIZE_T)(newsize = oldsize + chunksize(next)) >=
|
---|
4791 | (CHUNK_SIZE_T)(nb)) {
|
---|
4792 | newp = oldp;
|
---|
4793 | unlink(next, bck, fwd);
|
---|
4794 | hashtable_remove(chunk(next));
|
---|
4795 | freecilst_add(next);
|
---|
4796 | next = next_chunkinfo(oldp);
|
---|
4797 | if (next)
|
---|
4798 | next->prev_size = newsize;
|
---|
4799 | }
|
---|
4800 |
|
---|
4801 | /* allocate, copy, free */
|
---|
4802 | else {
|
---|
4803 |
|
---|
4804 | newmem = mALLOc(nb - MALLOC_ALIGN_MASK);
|
---|
4805 | if (newmem == 0)
|
---|
4806 | return 0; /* propagate failure */
|
---|
4807 |
|
---|
4808 | newp = hashtable_lookup(newmem);
|
---|
4809 | newsize = chunksize(newp);
|
---|
4810 |
|
---|
4811 |
|
---|
4812 | /* next = next_chunkinfo(oldp); *//* 'next' never used rw 19.05.2008 */
|
---|
4813 | /*
|
---|
4814 | Avoid copy if newp is next chunk after oldp.
|
---|
4815 | */
|
---|
4816 | if (UNLIKELY(is_next_chunk(oldp, newp))) {
|
---|
4817 | newsize += oldsize;
|
---|
4818 | set_head_size(oldp, newsize);
|
---|
4819 | hashtable_skiprm(oldp, newp);
|
---|
4820 | freecilst_add(newp);
|
---|
4821 | newp = oldp;
|
---|
4822 | }
|
---|
4823 | else {
|
---|
4824 | /*
|
---|
4825 | Unroll copy of <= 40 bytes (80 if 8byte sizes)
|
---|
4826 | We know that contents have an even number of
|
---|
4827 | INTERNAL_SIZE_T-sized words; minimally 4 (2 on amd64).
|
---|
4828 | */
|
---|
4829 |
|
---|
4830 | VALGRIND_MALLOCLIKE_BLOCK(chunk(oldp), chunksize(oldp), 0, 0);
|
---|
4831 |
|
---|
4832 | copysize = oldsize;
|
---|
4833 | s = (INTERNAL_SIZE_T*)(oldmem);
|
---|
4834 | d = (INTERNAL_SIZE_T*)(newmem);
|
---|
4835 | ncopies = copysize / sizeof(INTERNAL_SIZE_T);
|
---|
4836 | assert(ncopies >= 2);
|
---|
4837 |
|
---|
4838 | if (ncopies > 10)
|
---|
4839 | MALLOC_COPY(d, s, copysize);
|
---|
4840 |
|
---|
4841 | else {
|
---|
4842 | *(d+0) = *(s+0);
|
---|
4843 | *(d+1) = *(s+1);
|
---|
4844 | if (ncopies > 2) {
|
---|
4845 | *(d+2) = *(s+2);
|
---|
4846 | *(d+3) = *(s+3);
|
---|
4847 | if (ncopies > 4) {
|
---|
4848 | *(d+4) = *(s+4);
|
---|
4849 | *(d+5) = *(s+5);
|
---|
4850 | if (ncopies > 6) {
|
---|
4851 | *(d+6) = *(s+6);
|
---|
4852 | *(d+7) = *(s+7);
|
---|
4853 | if (ncopies > 8) {
|
---|
4854 | *(d+8) = *(s+8);
|
---|
4855 | *(d+9) = *(s+9);
|
---|
4856 | }
|
---|
4857 | }
|
---|
4858 | }
|
---|
4859 | }
|
---|
4860 | }
|
---|
4861 |
|
---|
4862 | fREe(oldmem);
|
---|
4863 | check_inuse_chunk(newp);
|
---|
4864 | guard_set(av->guard_stored, newp, bytes, nb);
|
---|
4865 | return chunk(newp);
|
---|
4866 | }
|
---|
4867 | }
|
---|
4868 | }
|
---|
4869 |
|
---|
4870 | /* If possible, free extra space in old or extended chunk */
|
---|
4871 |
|
---|
4872 | assert((CHUNK_SIZE_T)(newsize) >= (CHUNK_SIZE_T)(nb));
|
---|
4873 |
|
---|
4874 | remainder_size = newsize - nb;
|
---|
4875 |
|
---|
4876 | if (remainder_size >= MINSIZE) { /* split remainder */
|
---|
4877 | remainder = cireg_getfree();
|
---|
4878 | remainder->chunk = chunk_at_offset(chunk(newp), nb);
|
---|
4879 | set_head_size(newp, nb);
|
---|
4880 | set_head(remainder, remainder_size | PREV_INUSE | INUSE);
|
---|
4881 | remainder->prev_size = nb;
|
---|
4882 | hashtable_insert(newp, remainder);
|
---|
4883 | /* Mark remainder as inuse so free() won't complain */
|
---|
4884 | set_all_inuse(remainder);
|
---|
4885 | guard_set(av->guard_stored, remainder, 0, remainder_size);
|
---|
4886 | VALGRIND_MALLOCLIKE_BLOCK(chunk(remainder), remainder_size, 0, 0);
|
---|
4887 | fREe(chunk(remainder));
|
---|
4888 | }
|
---|
4889 | else { /* not enough extra to split off */
|
---|
4890 | set_head_size(newp, newsize);
|
---|
4891 | set_all_inuse(newp);
|
---|
4892 | }
|
---|
4893 |
|
---|
4894 | check_inuse_chunk(newp);
|
---|
4895 | guard_set(av->guard_stored, newp, bytes, nb);
|
---|
4896 | VALGRIND_MALLOCLIKE_BLOCK(chunk(newp), bytes, 0, 0);
|
---|
4897 | return chunk(newp);
|
---|
4898 | }
|
---|
4899 |
|
---|
4900 | /*
|
---|
4901 | Handle mmap cases
|
---|
4902 | */
|
---|
4903 |
|
---|
4904 | else {
|
---|
4905 |
|
---|
4906 | #if HAVE_MREMAP
|
---|
4907 | INTERNAL_SIZE_T offset = (INTERNAL_SIZE_T) oldp->hash_next;
|
---|
4908 | size_t pagemask = av->pagesize - 1;
|
---|
4909 | char *cp;
|
---|
4910 | CHUNK_SIZE_T sum;
|
---|
4911 |
|
---|
4912 |
|
---|
4913 | /* Note the extra SIZE_SZ overhead */
|
---|
4914 | /* newsize = (nb + offset + SIZE_SZ + pagemask) & ~pagemask; */
|
---|
4915 | newsize = (nb + offset + pagemask) & ~pagemask;
|
---|
4916 |
|
---|
4917 | /* don't need to remap if still within same page */
|
---|
4918 | if (oldsize == newsize - offset)
|
---|
4919 | {
|
---|
4920 | guard_set(av->guard_stored, oldp, bytes, nb);
|
---|
4921 | VALGRIND_FREELIKE_BLOCK(oldmem, 0);
|
---|
4922 | VALGRIND_MALLOCLIKE_BLOCK(oldmem, bytes, 0, 0);
|
---|
4923 | return oldmem;
|
---|
4924 | }
|
---|
4925 |
|
---|
4926 | cp = (char*)mremap((char*)chunk(oldp) - offset, oldsize + offset, newsize, 1);
|
---|
4927 |
|
---|
4928 | if (cp != (char*)MORECORE_FAILURE) {
|
---|
4929 |
|
---|
4930 | hashtable_remove_mmapped(chunk(oldp));
|
---|
4931 |
|
---|
4932 | oldp->chunk = (mchunkptr)(cp + offset);
|
---|
4933 | set_head(oldp, (newsize - offset)|IS_MMAPPED|INUSE);
|
---|
4934 |
|
---|
4935 | hashtable_add(oldp);
|
---|
4936 |
|
---|
4937 | assert(aligned_OK(chunk(oldp))); /* rw fix: newp -> oldp */
|
---|
4938 | assert(( ((INTERNAL_SIZE_T) oldp->hash_next) == offset));
|
---|
4939 |
|
---|
4940 | /* update statistics */
|
---|
4941 | sum = av->mmapped_mem += newsize - oldsize;
|
---|
4942 | if (sum > (CHUNK_SIZE_T)(av->max_mmapped_mem))
|
---|
4943 | av->max_mmapped_mem = sum;
|
---|
4944 | sum += av->sbrked_mem;
|
---|
4945 | if (sum > (CHUNK_SIZE_T)(av->max_total_mem))
|
---|
4946 | av->max_total_mem = sum;
|
---|
4947 |
|
---|
4948 | guard_set(av->guard_stored, oldp, bytes, nb);
|
---|
4949 | VALGRIND_FREELIKE_BLOCK(oldmem, 0);
|
---|
4950 | VALGRIND_MALLOCLIKE_BLOCK(chunk(oldp), bytes, 0, 0);
|
---|
4951 | return chunk(oldp);
|
---|
4952 | }
|
---|
4953 | #endif /* have MREMAP */
|
---|
4954 |
|
---|
4955 | /* Note the extra SIZE_SZ overhead. */
|
---|
4956 | if ((CHUNK_SIZE_T)(oldsize) >= (CHUNK_SIZE_T)(nb + SIZE_SZ))
|
---|
4957 | newmem = oldmem; /* do nothing */
|
---|
4958 | else {
|
---|
4959 | /* Must alloc, copy, free. */
|
---|
4960 | newmem = mALLOc(nb - MALLOC_ALIGN_MASK);
|
---|
4961 | if (newmem != 0) {
|
---|
4962 | MALLOC_COPY(newmem, oldmem, oldsize);
|
---|
4963 | fREe(oldmem);
|
---|
4964 | }
|
---|
4965 | }
|
---|
4966 | VALGRIND_MALLOCLIKE_BLOCK(newmem, bytes, 0, 0);
|
---|
4967 | guard_set(av->guard_stored, mem2chunk(newmem), bytes, nb);
|
---|
4968 | return newmem;
|
---|
4969 | }
|
---|
4970 | }
|
---|
4971 |
|
---|
4972 | /*
|
---|
4973 | ---------------------------posix_memalign ----------------------------
|
---|
4974 | */
|
---|
4975 |
|
---|
4976 | #if __STD_C
|
---|
4977 | DL_STATIC int posix_mEMALIGn(Void_t** memptr, size_t alignment, size_t bytes)
|
---|
4978 | #else
|
---|
4979 | DL_STATIC int posix_mEMALIGn(memptr, alignment, bytes) Void_t** memptr; size_t alignment; size_t bytes;
|
---|
4980 | #endif
|
---|
4981 | {
|
---|
4982 | mstate av;
|
---|
4983 |
|
---|
4984 | if (alignment % sizeof(void *) != 0)
|
---|
4985 | return EINVAL;
|
---|
4986 | if ((alignment & (alignment - 1)) != 0)
|
---|
4987 | return EINVAL;
|
---|
4988 |
|
---|
4989 | av = get_malloc_state();
|
---|
4990 | if (!av || av->max_fast == 0) malloc_consolidate(av);
|
---|
4991 | *memptr = mEMALIGn(alignment, bytes);
|
---|
4992 |
|
---|
4993 | return (*memptr != NULL ? 0 : ENOMEM);
|
---|
4994 | }
|
---|
4995 |
|
---|
4996 | /*
|
---|
4997 | ------------------------------ memalign ------------------------------
|
---|
4998 | */
|
---|
4999 |
|
---|
5000 | #if __STD_C
|
---|
5001 | DL_STATIC Void_t* mEMALIGn(size_t alignment, size_t bytes)
|
---|
5002 | #else
|
---|
5003 | DL_STATIC Void_t* mEMALIGn(alignment, bytes) size_t alignment; size_t bytes;
|
---|
5004 | #endif
|
---|
5005 | {
|
---|
5006 | INTERNAL_SIZE_T nb; /* padded request size */
|
---|
5007 | char* m; /* memory returned by malloc call */
|
---|
5008 | chunkinfoptr p; /* corresponding chunk */
|
---|
5009 | char* brk; /* alignment point within p */
|
---|
5010 | chunkinfoptr newp; /* chunk to return */
|
---|
5011 | INTERNAL_SIZE_T newsize; /* its size */
|
---|
5012 | INTERNAL_SIZE_T leadsize; /* leading space before alignment point */
|
---|
5013 | chunkinfoptr remainder; /* spare room at end to split off */
|
---|
5014 | CHUNK_SIZE_T remainder_size; /* its size */
|
---|
5015 | INTERNAL_SIZE_T size;
|
---|
5016 | #if PARANOIA > 2
|
---|
5017 | mstate av;
|
---|
5018 | #endif
|
---|
5019 |
|
---|
5020 | /* If need less alignment than we give anyway, just relay to malloc */
|
---|
5021 |
|
---|
5022 | if (UNLIKELY(alignment <= MALLOC_ALIGNMENT)) return mALLOc(bytes);
|
---|
5023 |
|
---|
5024 | /* Otherwise, ensure that it is at least a minimum chunk size */
|
---|
5025 |
|
---|
5026 | if (alignment < MINSIZE) alignment = MINSIZE;
|
---|
5027 |
|
---|
5028 | /* Make sure alignment is power of 2 (in case MINSIZE is not). */
|
---|
5029 | if (UNLIKELY((alignment & (alignment - 1)) != 0)) {
|
---|
5030 | size_t a = MALLOC_ALIGNMENT * 2;
|
---|
5031 | while ((CHUNK_SIZE_T)a < (CHUNK_SIZE_T)alignment) a <<= 1;
|
---|
5032 | alignment = a;
|
---|
5033 | }
|
---|
5034 |
|
---|
5035 | checked_request2size(bytes, nb);
|
---|
5036 |
|
---|
5037 | /*
|
---|
5038 | Strategy: find a spot within that chunk that meets the alignment
|
---|
5039 | request, and then possibly free the leading and trailing space.
|
---|
5040 | */
|
---|
5041 |
|
---|
5042 |
|
---|
5043 | /* Call malloc with worst case padding to hit alignment. */
|
---|
5044 |
|
---|
5045 | m = (char*)(mALLOc(nb + alignment + MINSIZE));
|
---|
5046 |
|
---|
5047 | if (m == 0) return 0; /* propagate failure */
|
---|
5048 |
|
---|
5049 | #if PARANOIA > 2
|
---|
5050 | av = get_malloc_state();
|
---|
5051 | #endif
|
---|
5052 |
|
---|
5053 | p = hashtable_lookup((mchunkptr) m);
|
---|
5054 |
|
---|
5055 | if ((((PTR_UINT)(m)) % alignment) != 0) { /* misaligned */
|
---|
5056 |
|
---|
5057 | /*
|
---|
5058 | Find an aligned spot inside chunk. Since we need to give back
|
---|
5059 | leading space in a chunk of at least MINSIZE, if the first
|
---|
5060 | calculation places us at a spot with less than MINSIZE leader,
|
---|
5061 | we can move to the next aligned spot -- we've allocated enough
|
---|
5062 | total room so that this is always possible.
|
---|
5063 | */
|
---|
5064 |
|
---|
5065 | brk = (char*) ((PTR_UINT)(((PTR_UINT)(m + alignment - 1)) &
|
---|
5066 | -((signed long) alignment)));
|
---|
5067 | if ((CHUNK_SIZE_T)(brk - (char*)(chunk(p))) < MINSIZE)
|
---|
5068 | brk += alignment;
|
---|
5069 |
|
---|
5070 | newp = cireg_getfree();
|
---|
5071 | newp->chunk = (mchunkptr)brk;
|
---|
5072 | leadsize = brk - (char*)(chunk(p));
|
---|
5073 | newsize = chunksize(p) - leadsize;
|
---|
5074 |
|
---|
5075 | /* For mmapped chunks, just adjust offset */
|
---|
5076 | if (UNLIKELY(chunk_is_mmapped(p))) {
|
---|
5077 | newp->hash_next = (chunkinfoptr) (((INTERNAL_SIZE_T) p->hash_next) + leadsize);
|
---|
5078 | set_head(newp, newsize|IS_MMAPPED|INUSE);
|
---|
5079 | hashtable_remove_mmapped(chunk(p));
|
---|
5080 | freecilst_add(p);
|
---|
5081 | hashtable_add(newp);
|
---|
5082 | guard_set(av->guard_stored, newp, bytes, nb);
|
---|
5083 | return chunk(newp);
|
---|
5084 | }
|
---|
5085 |
|
---|
5086 | /* Otherwise, give back leader, use the rest */
|
---|
5087 | set_head(newp, newsize | PREV_INUSE | INUSE);
|
---|
5088 | set_head_size(p, leadsize);
|
---|
5089 | set_all_inuse(newp);
|
---|
5090 | hashtable_add(newp); /* 20.05.2008 rw */
|
---|
5091 | guard_set(av->guard_stored, p, 0, leadsize);
|
---|
5092 | fREe(chunk(p));
|
---|
5093 | p = newp;
|
---|
5094 |
|
---|
5095 | assert (newsize >= nb &&
|
---|
5096 | (((PTR_UINT)(chunk(p))) % alignment) == 0);
|
---|
5097 | }
|
---|
5098 |
|
---|
5099 | /* Also give back spare room at the end */
|
---|
5100 | if (!chunk_is_mmapped(p)) {
|
---|
5101 | size = chunksize(p);
|
---|
5102 | if ((CHUNK_SIZE_T)(size) > (CHUNK_SIZE_T)(nb + MINSIZE)) {
|
---|
5103 | remainder = cireg_getfree();
|
---|
5104 | remainder_size = size - nb;
|
---|
5105 | remainder->chunk = chunk_at_offset(chunk(p), nb);
|
---|
5106 | set_head(remainder, remainder_size | PREV_INUSE | INUSE);
|
---|
5107 | set_head_size(p, nb);
|
---|
5108 | hashtable_add(remainder); /* 20.05.2008 rw */
|
---|
5109 | guard_set(av->guard_stored, remainder, 0, remainder_size);
|
---|
5110 | fREe(chunk(remainder));
|
---|
5111 | }
|
---|
5112 | }
|
---|
5113 |
|
---|
5114 | check_inuse_chunk(p);
|
---|
5115 | guard_set(av->guard_stored, p, bytes, nb);
|
---|
5116 | return chunk(p);
|
---|
5117 | }
|
---|
5118 |
|
---|
5119 | /*
|
---|
5120 | ------------------------------ calloc ------------------------------
|
---|
5121 | */
|
---|
5122 |
|
---|
5123 | #if __STD_C
|
---|
5124 | DL_STATIC Void_t* cALLOc(size_t n_elements, size_t elem_size)
|
---|
5125 | #else
|
---|
5126 | DL_STATIC Void_t* cALLOc(n_elements, elem_size) size_t n_elements; size_t elem_size;
|
---|
5127 | #endif
|
---|
5128 | {
|
---|
5129 | chunkinfoptr p;
|
---|
5130 | CHUNK_SIZE_T clearsize;
|
---|
5131 | CHUNK_SIZE_T nclears;
|
---|
5132 | INTERNAL_SIZE_T* d;
|
---|
5133 | Void_t* mem;
|
---|
5134 |
|
---|
5135 |
|
---|
5136 | mem = mALLOc(n_elements * elem_size);
|
---|
5137 |
|
---|
5138 | if (mem != 0) {
|
---|
5139 | p = hashtable_lookup(mem);
|
---|
5140 |
|
---|
5141 | if (!chunk_is_mmapped(p))
|
---|
5142 | {
|
---|
5143 | /*
|
---|
5144 | Unroll clear of <= 40 bytes (80 if 8byte sizes)
|
---|
5145 | We know that contents have an even number of
|
---|
5146 | INTERNAL_SIZE_T-sized words; minimally 4 (2 on amd64).
|
---|
5147 | */
|
---|
5148 |
|
---|
5149 | d = (INTERNAL_SIZE_T*)mem;
|
---|
5150 | clearsize = chunksize(p);
|
---|
5151 | nclears = clearsize / sizeof(INTERNAL_SIZE_T);
|
---|
5152 | assert(nclears >= 2);
|
---|
5153 |
|
---|
5154 | if (nclears > 10) {
|
---|
5155 | MALLOC_ZERO(d, clearsize);
|
---|
5156 | }
|
---|
5157 |
|
---|
5158 | else {
|
---|
5159 | *(d+0) = 0;
|
---|
5160 | *(d+1) = 0;
|
---|
5161 | if (nclears > 2) {
|
---|
5162 | *(d+2) = 0;
|
---|
5163 | *(d+3) = 0;
|
---|
5164 | if (nclears > 4) {
|
---|
5165 | *(d+4) = 0;
|
---|
5166 | *(d+5) = 0;
|
---|
5167 | if (nclears > 6) {
|
---|
5168 | *(d+6) = 0;
|
---|
5169 | *(d+7) = 0;
|
---|
5170 | if (nclears > 8) {
|
---|
5171 | *(d+8) = 0;
|
---|
5172 | *(d+9) = 0;
|
---|
5173 | }
|
---|
5174 | }
|
---|
5175 | }
|
---|
5176 | }
|
---|
5177 | }
|
---|
5178 | }
|
---|
5179 | #if ! MMAP_CLEARS
|
---|
5180 | else
|
---|
5181 | {
|
---|
5182 | d = (INTERNAL_SIZE_T*)mem;
|
---|
5183 | clearsize = chunksize(p);
|
---|
5184 | MALLOC_ZERO(d, clearsize);
|
---|
5185 | }
|
---|
5186 | #endif
|
---|
5187 | /* Set guard again, since we just cleared it
|
---|
5188 | */
|
---|
5189 | guard_set(get_malloc_state()->guard_stored, p, (n_elements * elem_size), p->size);
|
---|
5190 | }
|
---|
5191 |
|
---|
5192 | return mem;
|
---|
5193 | }
|
---|
5194 |
|
---|
5195 | /*
|
---|
5196 | ------------------------------ valloc ------------------------------
|
---|
5197 | */
|
---|
5198 |
|
---|
5199 | #if __STD_C
|
---|
5200 | DL_STATIC Void_t* vALLOc(size_t bytes)
|
---|
5201 | #else
|
---|
5202 | DL_STATIC Void_t* vALLOc(bytes) size_t bytes;
|
---|
5203 | #endif
|
---|
5204 | {
|
---|
5205 | /* Ensure initialization */
|
---|
5206 | mstate av = get_malloc_state();
|
---|
5207 | if (!av || av->max_fast == 0) {
|
---|
5208 | malloc_consolidate(av);
|
---|
5209 | av = get_malloc_state();
|
---|
5210 | }
|
---|
5211 | return mEMALIGn(av->pagesize, bytes);
|
---|
5212 | }
|
---|
5213 |
|
---|
5214 | /*
|
---|
5215 | ------------------------------ pvalloc ------------------------------
|
---|
5216 | */
|
---|
5217 |
|
---|
5218 |
|
---|
5219 | #if __STD_C
|
---|
5220 | DL_STATIC Void_t* pVALLOc(size_t bytes)
|
---|
5221 | #else
|
---|
5222 | DL_STATIC Void_t* pVALLOc(bytes) size_t bytes;
|
---|
5223 | #endif
|
---|
5224 | {
|
---|
5225 | mstate av = get_malloc_state();
|
---|
5226 | size_t pagesz;
|
---|
5227 |
|
---|
5228 | /* Ensure initialization */
|
---|
5229 | if (!av || av->max_fast == 0) {
|
---|
5230 | malloc_consolidate(av);
|
---|
5231 | av = get_malloc_state();
|
---|
5232 | }
|
---|
5233 | pagesz = av->pagesize;
|
---|
5234 | return mEMALIGn(pagesz, (bytes + pagesz - 1) & ~(pagesz - 1));
|
---|
5235 | }
|
---|
5236 |
|
---|
5237 |
|
---|
5238 | /*
|
---|
5239 | ------------------------------ malloc_trim ------------------------------
|
---|
5240 | */
|
---|
5241 |
|
---|
5242 | #if __STD_C
|
---|
5243 | DL_STATIC int mTRIm(size_t pad)
|
---|
5244 | #else
|
---|
5245 | DL_STATIC int mTRIm(pad) size_t pad;
|
---|
5246 | #endif
|
---|
5247 | {
|
---|
5248 | mstate av = get_malloc_state();
|
---|
5249 | /* Ensure initialization/consolidation */
|
---|
5250 | malloc_consolidate(av);
|
---|
5251 | av = get_malloc_state();
|
---|
5252 | #ifndef MORECORE_CANNOT_TRIM
|
---|
5253 | if (morecore32bit(av))
|
---|
5254 | return sYSTRIm(pad, av);
|
---|
5255 | else
|
---|
5256 | return 0;
|
---|
5257 | #else
|
---|
5258 | return 0;
|
---|
5259 | #endif
|
---|
5260 | }
|
---|
5261 |
|
---|
5262 |
|
---|
5263 |
|
---|
5264 | /*
|
---|
5265 | ------------------------- malloc_usable_size -------------------------
|
---|
5266 | */
|
---|
5267 |
|
---|
5268 | #if __STD_C
|
---|
5269 | DL_STATIC size_t mUSABLe(Void_t* mem)
|
---|
5270 | #else
|
---|
5271 | DL_STATIC size_t mUSABLe(mem) Void_t* mem;
|
---|
5272 | #endif
|
---|
5273 | {
|
---|
5274 | chunkinfoptr p;
|
---|
5275 | if (mem != 0) {
|
---|
5276 | p = hashtable_lookup(mem);
|
---|
5277 | if (p && inuse(p)) return chunksize(p);
|
---|
5278 | }
|
---|
5279 | return 0;
|
---|
5280 | }
|
---|
5281 |
|
---|
5282 | /*
|
---|
5283 | ------------------------------ mallinfo ------------------------------
|
---|
5284 | */
|
---|
5285 |
|
---|
5286 | DL_STATIC struct mallinfo mALLINFo()
|
---|
5287 | {
|
---|
5288 | mstate av = get_malloc_state();
|
---|
5289 | struct mallinfo mi;
|
---|
5290 | unsigned int i;
|
---|
5291 | mbinptr b;
|
---|
5292 | chunkinfoptr p;
|
---|
5293 | INTERNAL_SIZE_T avail;
|
---|
5294 | INTERNAL_SIZE_T fastavail;
|
---|
5295 | int nblocks;
|
---|
5296 | int nfastblocks;
|
---|
5297 |
|
---|
5298 | /* Ensure initialization */
|
---|
5299 | if (!av || av->top == 0) {
|
---|
5300 | malloc_consolidate(av);
|
---|
5301 | av = get_malloc_state();
|
---|
5302 | }
|
---|
5303 | check_malloc_state();
|
---|
5304 |
|
---|
5305 | /* Account for top */
|
---|
5306 | avail = chunksize(av->top);
|
---|
5307 | nblocks = 1; /* top always exists */
|
---|
5308 |
|
---|
5309 | /* traverse fastbins */
|
---|
5310 | nfastblocks = 0;
|
---|
5311 | fastavail = 0;
|
---|
5312 |
|
---|
5313 | for (i = 0; i < NFASTBINS; ++i) {
|
---|
5314 | for (p = av->fastbins[i]; p != 0; p = p->fd) {
|
---|
5315 | ++nfastblocks;
|
---|
5316 | fastavail += chunksize(p);
|
---|
5317 | }
|
---|
5318 | }
|
---|
5319 |
|
---|
5320 | avail += fastavail;
|
---|
5321 |
|
---|
5322 | /* traverse regular bins */
|
---|
5323 | for (i = 1; i < NBINS; ++i) {
|
---|
5324 | b = bin_at(av, i);
|
---|
5325 | for (p = last(b); p != b; p = p->bk) {
|
---|
5326 | ++nblocks;
|
---|
5327 | avail += chunksize(p);
|
---|
5328 | }
|
---|
5329 | }
|
---|
5330 |
|
---|
5331 | mi.smblks = nfastblocks;
|
---|
5332 | mi.ordblks = nblocks;
|
---|
5333 | mi.fordblks = avail;
|
---|
5334 | mi.uordblks = av->sbrked_mem - avail;
|
---|
5335 | mi.arena = av->sbrked_mem;
|
---|
5336 | mi.hblks = av->n_mmaps;
|
---|
5337 | mi.hblkhd = av->mmapped_mem;
|
---|
5338 | mi.fsmblks = fastavail;
|
---|
5339 | mi.keepcost = chunksize(av->top);
|
---|
5340 | mi.usmblks = av->max_total_mem;
|
---|
5341 | return mi;
|
---|
5342 | }
|
---|
5343 |
|
---|
5344 | /*
|
---|
5345 | ------------------------------ malloc_stats ------------------------------
|
---|
5346 | */
|
---|
5347 |
|
---|
5348 | DL_STATIC void mSTATs()
|
---|
5349 | {
|
---|
5350 | struct mallinfo mi = mALLINFo();
|
---|
5351 |
|
---|
5352 | fprintf(stderr, "hashtable = %10lu MB\n",
|
---|
5353 | (CHUNK_SIZE_T)(HASHTABLESIZE / (1024*1024)));
|
---|
5354 | fprintf(stderr, "max system bytes = %10lu\n",
|
---|
5355 | (CHUNK_SIZE_T)(mi.usmblks));
|
---|
5356 | fprintf(stderr, "system bytes = %10lu (%10lu sbrked, %10lu mmaped)\n",
|
---|
5357 | (CHUNK_SIZE_T)(mi.arena + mi.hblkhd),
|
---|
5358 | (CHUNK_SIZE_T)(mi.arena),
|
---|
5359 | (CHUNK_SIZE_T)(mi.hblkhd));
|
---|
5360 | fprintf(stderr, "in use bytes = %10lu\n",
|
---|
5361 | (CHUNK_SIZE_T)(mi.uordblks + mi.hblkhd));
|
---|
5362 |
|
---|
5363 | }
|
---|
5364 |
|
---|
5365 |
|
---|
5366 | /*
|
---|
5367 | ------------------------------ mallopt ------------------------------
|
---|
5368 | */
|
---|
5369 |
|
---|
5370 | #if __STD_C
|
---|
5371 | DL_STATIC int mALLOPt(int param_number, int value)
|
---|
5372 | #else
|
---|
5373 | DL_STATIC int mALLOPt(param_number, value) int param_number; int value;
|
---|
5374 | #endif
|
---|
5375 | {
|
---|
5376 | mstate av = get_malloc_state();
|
---|
5377 | /* Ensure initialization/consolidation */
|
---|
5378 | malloc_consolidate(av);
|
---|
5379 | av = get_malloc_state();
|
---|
5380 |
|
---|
5381 | switch(param_number) {
|
---|
5382 | case M_MXFAST:
|
---|
5383 | if (value >= 0 && value <= MAX_FAST_SIZE) {
|
---|
5384 | set_max_fast(av, value);
|
---|
5385 | return 1;
|
---|
5386 | }
|
---|
5387 | else
|
---|
5388 | return 0;
|
---|
5389 |
|
---|
5390 | case M_TRIM_THRESHOLD:
|
---|
5391 | av->trim_threshold = value;
|
---|
5392 | return 1;
|
---|
5393 |
|
---|
5394 | case M_TOP_PAD:
|
---|
5395 | av->top_pad = value;
|
---|
5396 | return 1;
|
---|
5397 |
|
---|
5398 | case M_MMAP_THRESHOLD:
|
---|
5399 | av->mmap_threshold = value;
|
---|
5400 | return 1;
|
---|
5401 |
|
---|
5402 | case M_MMAP_MAX:
|
---|
5403 | if (value != 0)
|
---|
5404 | return 0;
|
---|
5405 | av->n_mmaps_max = value;
|
---|
5406 | return 1;
|
---|
5407 |
|
---|
5408 | default:
|
---|
5409 | return 0;
|
---|
5410 | }
|
---|
5411 | }
|
---|
5412 |
|
---|
5413 |
|
---|
5414 | /* $OpenBSD: arc4random.c,v 1.19 2008/06/04 00:50:23 djm Exp $ */
|
---|
5415 |
|
---|
5416 | /*
|
---|
5417 | * Copyright (c) 1996, David Mazieres <dm@uun.org>
|
---|
5418 | * Copyright (c) 2008, Damien Miller <djm@openbsd.org>
|
---|
5419 | *
|
---|
5420 | * Permission to use, copy, modify, and distribute this software for any
|
---|
5421 | * purpose with or without fee is hereby granted, provided that the above
|
---|
5422 | * copyright notice and this permission notice appear in all copies.
|
---|
5423 | *
|
---|
5424 | * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
|
---|
5425 | * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
|
---|
5426 | * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
|
---|
5427 | * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
|
---|
5428 | * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
|
---|
5429 | * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
|
---|
5430 | * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
|
---|
5431 | */
|
---|
5432 |
|
---|
5433 | /*
|
---|
5434 | * Arc4 random number generator for OpenBSD.
|
---|
5435 | *
|
---|
5436 | * This code is derived from section 17.1 of Applied Cryptography,
|
---|
5437 | * second edition, which describes a stream cipher allegedly
|
---|
5438 | * compatible with RSA Labs "RC4" cipher (the actual description of
|
---|
5439 | * which is a trade secret). The same algorithm is used as a stream
|
---|
5440 | * cipher called "arcfour" in Tatu Ylonen's ssh package.
|
---|
5441 | *
|
---|
5442 | * Here the stream cipher has been modified always to include the time
|
---|
5443 | * when initializing the state. That makes it impossible to
|
---|
5444 | * regenerate the same random sequence twice, so this can't be used
|
---|
5445 | * for encryption, but will generate good random numbers.
|
---|
5446 | *
|
---|
5447 | * RC4 is a registered trademark of RSA Laboratories.
|
---|
5448 | */
|
---|
5449 |
|
---|
5450 | /* Moved u_int8_t -> unsigned char (portability)
|
---|
5451 | * Eliminated unneeded functions, added read from /dev/urandom taken from:
|
---|
5452 | $MirOS: contrib/code/Snippets/arc4random.c,v 1.3 2008-03-04 22:53:14 tg Exp $
|
---|
5453 | * Modified by Robert Connolly from OpenBSD lib/libc/crypt/arc4random.c v1.11.
|
---|
5454 | * This is arc4random(3) using urandom.
|
---|
5455 | */
|
---|
5456 |
|
---|
5457 | #include <fcntl.h>
|
---|
5458 | #include <limits.h>
|
---|
5459 | #include <stdlib.h>
|
---|
5460 | #include <sys/param.h>
|
---|
5461 | #include <sys/time.h>
|
---|
5462 |
|
---|
5463 | struct arc4_stream {
|
---|
5464 | unsigned char i;
|
---|
5465 | unsigned char j;
|
---|
5466 | unsigned char s[256];
|
---|
5467 | };
|
---|
5468 |
|
---|
5469 | static int rs_initialized;
|
---|
5470 | static struct arc4_stream rs;
|
---|
5471 | static pid_t arc4_stir_pid;
|
---|
5472 | static int arc4_count;
|
---|
5473 |
|
---|
5474 | static unsigned char arc4_getbyte(void);
|
---|
5475 |
|
---|
5476 | static void
|
---|
5477 | arc4_init(void)
|
---|
5478 | {
|
---|
5479 | int n;
|
---|
5480 |
|
---|
5481 | for (n = 0; n < 256; n++)
|
---|
5482 | rs.s[n] = n;
|
---|
5483 | rs.i = 0;
|
---|
5484 | rs.j = 0;
|
---|
5485 | }
|
---|
5486 |
|
---|
5487 | static inline void
|
---|
5488 | arc4_addrandom(unsigned char *dat, int datlen)
|
---|
5489 | {
|
---|
5490 | int n;
|
---|
5491 | unsigned char si;
|
---|
5492 |
|
---|
5493 | rs.i--;
|
---|
5494 | for (n = 0; n < 256; n++) {
|
---|
5495 | rs.i = (rs.i + 1);
|
---|
5496 | si = rs.s[rs.i];
|
---|
5497 | rs.j = (rs.j + si + dat[n % datlen]);
|
---|
5498 | rs.s[rs.i] = rs.s[rs.j];
|
---|
5499 | rs.s[rs.j] = si;
|
---|
5500 | }
|
---|
5501 | rs.j = rs.i;
|
---|
5502 | }
|
---|
5503 |
|
---|
5504 | #ifdef HAVE_SCHED_H
|
---|
5505 | #include <sched.h>
|
---|
5506 | #endif
|
---|
5507 |
|
---|
5508 | static void
|
---|
5509 | arc4_stir(void)
|
---|
5510 | {
|
---|
5511 | int i;
|
---|
5512 | struct {
|
---|
5513 | struct timeval tv1;
|
---|
5514 | struct timeval tv2;
|
---|
5515 | u_int rnd[(128 - 2*sizeof(struct timeval)) / sizeof(u_int)];
|
---|
5516 | } rdat;
|
---|
5517 | #if !defined(__FreeBSD__) && !defined(__OpenBSD__) && !defined(__NetBSD__)
|
---|
5518 | size_t sz = 0;
|
---|
5519 | int fd;
|
---|
5520 | #endif
|
---|
5521 |
|
---|
5522 | gettimeofday(&rdat.tv1, NULL);
|
---|
5523 |
|
---|
5524 |
|
---|
5525 | if (!rs_initialized) {
|
---|
5526 | arc4_init();
|
---|
5527 | rs_initialized = 1;
|
---|
5528 | }
|
---|
5529 |
|
---|
5530 | #if !defined(__FreeBSD__) && !defined(__OpenBSD__) && !defined(__NetBSD__)
|
---|
5531 |
|
---|
5532 | #ifdef HAVE_SCHED_YIELD
|
---|
5533 | /* Yield the processor to introduce some random delay. */
|
---|
5534 | (void) sched_yield();
|
---|
5535 | #endif
|
---|
5536 |
|
---|
5537 | /*
|
---|
5538 | * Pthread problem in multithreaded code on *BSD.
|
---|
5539 | */
|
---|
5540 | fd = open("/dev/urandom", O_RDONLY);
|
---|
5541 | if (fd != -1) {
|
---|
5542 | sz = (size_t)read(fd, rdat.rnd, sizeof (rdat.rnd));
|
---|
5543 | close(fd);
|
---|
5544 | }
|
---|
5545 | if (sz > sizeof (rdat.rnd))
|
---|
5546 | sz = 0;
|
---|
5547 | #endif
|
---|
5548 |
|
---|
5549 | arc4_stir_pid = getpid();
|
---|
5550 | gettimeofday(&rdat.tv2, NULL);
|
---|
5551 |
|
---|
5552 | arc4_addrandom((void *)&rdat, sizeof(rdat));
|
---|
5553 |
|
---|
5554 | /*
|
---|
5555 | * Discard early keystream, as per recommendations in:
|
---|
5556 | * http://www.wisdom.weizmann.ac.il/~itsik/RC4/Papers/Rc4_ksa.ps
|
---|
5557 | */
|
---|
5558 | for (i = 0; i < 256; i++)
|
---|
5559 | (void)arc4_getbyte();
|
---|
5560 | arc4_count = 1600000;
|
---|
5561 | }
|
---|
5562 |
|
---|
5563 | static unsigned char
|
---|
5564 | arc4_getbyte(void)
|
---|
5565 | {
|
---|
5566 | unsigned char si, sj;
|
---|
5567 |
|
---|
5568 | rs.i = (rs.i + 1);
|
---|
5569 | si = rs.s[rs.i];
|
---|
5570 | rs.j = (rs.j + si);
|
---|
5571 | sj = rs.s[rs.j];
|
---|
5572 | rs.s[rs.i] = sj;
|
---|
5573 | rs.s[rs.j] = si;
|
---|
5574 | return (rs.s[(si + sj) & 0xff]);
|
---|
5575 | }
|
---|
5576 |
|
---|
5577 |
|
---|
5578 | /* Changed to return char* */
|
---|
5579 | static char *
|
---|
5580 | dnmalloc_arc4random(void)
|
---|
5581 | {
|
---|
5582 | static char val[4];
|
---|
5583 |
|
---|
5584 | /* We only call this once, hence no need for locking. */
|
---|
5585 |
|
---|
5586 | /* _ARC4_LOCK(); */
|
---|
5587 | arc4_count -= 4;
|
---|
5588 | if (arc4_count <= 0 || !rs_initialized || arc4_stir_pid != getpid())
|
---|
5589 | arc4_stir();
|
---|
5590 |
|
---|
5591 | val[0] = (char) arc4_getbyte();
|
---|
5592 | val[1] = (char) arc4_getbyte();
|
---|
5593 | val[2] = (char) arc4_getbyte();
|
---|
5594 | val[3] = (char) arc4_getbyte();
|
---|
5595 |
|
---|
5596 | arc4_stir();
|
---|
5597 | /* _ARC4_UNLOCK(); */
|
---|
5598 | return val;
|
---|
5599 | }
|
---|
5600 |
|
---|
5601 | #else
|
---|
5602 | int dnmalloc_pthread_init() { return 0; }
|
---|
5603 | #endif /* ! USE_SYSTEM_MALLOC */
|
---|