1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
|
/*
Copyright (c) 2008-2010 Gluster, Inc. <http://www.gluster.com>
This file is part of GlusterFS.
GlusterFS is free software; you can redistribute it and/or modify
it under the terms of the GNU Affero General Public License as published
by the Free Software Foundation; either version 3 of the License,
or (at your option) any later version.
GlusterFS is distributed in the hope that it will be useful, but
WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
Affero General Public License for more details.
You should have received a copy of the GNU Affero General Public License
along with this program. If not, see
<http://www.gnu.org/licenses/>.
*/
#include "mem-pool.h"
#include "logging.h"
#include "xlator.h"
#include <stdlib.h>
#include <stdarg.h>
#define GF_MEM_POOL_LIST_BOUNDARY (sizeof(struct list_head))
#define GF_MEM_POOL_PAD_BOUNDARY (GF_MEM_POOL_LIST_BOUNDARY + sizeof(int))
#define mem_pool_chunkhead2ptr(head) ((head) + GF_MEM_POOL_PAD_BOUNDARY)
#define mem_pool_ptr2chunkhead(ptr) ((ptr) - GF_MEM_POOL_PAD_BOUNDARY)
#define is_mem_chunk_in_use(ptr) (*ptr == 1)
#define GF_MEM_HEADER_SIZE (4 + sizeof (size_t) + sizeof (xlator_t *) + 4 + 8)
#define GF_MEM_TRAILER_SIZE 8
#define GF_MEM_HEADER_MAGIC 0xCAFEBABE
#define GF_MEM_TRAILER_MAGIC 0xBAADF00D
#define GLUSTERFS_ENV_MEM_ACCT_STR "GLUSTERFS_DISABLE_MEM_ACCT"
static int gf_mem_acct_enable = 0;
int
gf_mem_acct_is_enabled ()
{
return gf_mem_acct_enable;
}
void
gf_mem_acct_enable_set ()
{
char *opt = NULL;
long val = -1;
#ifdef DEBUG
gf_mem_acct_enable = 1;
return;
#endif
opt = getenv (GLUSTERFS_ENV_MEM_ACCT_STR);
if (!opt)
return;
val = strtol (opt, NULL, 0);
if (val)
gf_mem_acct_enable = 0;
else
gf_mem_acct_enable = 1;
}
void
gf_mem_set_acct_info (xlator_t *xl, char **alloc_ptr,
size_t size, uint32_t type)
{
char *ptr = NULL;
if (!alloc_ptr)
return;
ptr = (char *) (*alloc_ptr);
if (!xl) {
GF_ASSERT (0);
}
if (!(xl->mem_acct.rec)) {
GF_ASSERT (0);
}
if (type > xl->mem_acct.num_types) {
GF_ASSERT (0);
}
LOCK(&xl->mem_acct.rec[type].lock);
{
xl->mem_acct.rec[type].size += size;
xl->mem_acct.rec[type].num_allocs++;
xl->mem_acct.rec[type].max_size =
max (xl->mem_acct.rec[type].max_size,
xl->mem_acct.rec[type].size);
xl->mem_acct.rec[type].max_num_allocs =
max (xl->mem_acct.rec[type].max_num_allocs,
xl->mem_acct.rec[type].num_allocs);
}
UNLOCK(&xl->mem_acct.rec[type].lock);
*(uint32_t *)(ptr) = type;
ptr = ptr + 4;
memcpy (ptr, &size, sizeof(size_t));
ptr += sizeof (size_t);
memcpy (ptr, &xl, sizeof(xlator_t *));
ptr += sizeof (xlator_t *);
*(uint32_t *)(ptr) = GF_MEM_HEADER_MAGIC;
ptr = ptr + 4;
ptr = ptr + 8; //padding
*(uint32_t *) (ptr + size) = GF_MEM_TRAILER_MAGIC;
*alloc_ptr = (void *)ptr;
return;
}
void *
__gf_calloc (size_t nmemb, size_t size, uint32_t type)
{
size_t tot_size = 0;
size_t req_size = 0;
char *ptr = NULL;
xlator_t *xl = NULL;
if (!gf_mem_acct_enable)
return CALLOC (nmemb, size);
xl = THIS;
req_size = nmemb * size;
tot_size = req_size + GF_MEM_HEADER_SIZE + GF_MEM_TRAILER_SIZE;
ptr = calloc (1, tot_size);
if (!ptr) {
gf_log_nomem ("", GF_LOG_ALERT, tot_size);
return NULL;
}
gf_mem_set_acct_info (xl, &ptr, req_size, type);
return (void *)ptr;
}
void *
__gf_malloc (size_t size, uint32_t type)
{
size_t tot_size = 0;
char *ptr = NULL;
xlator_t *xl = NULL;
if (!gf_mem_acct_enable)
return MALLOC (size);
xl = THIS;
tot_size = size + GF_MEM_HEADER_SIZE + GF_MEM_TRAILER_SIZE;
ptr = malloc (tot_size);
if (!ptr) {
gf_log_nomem ("", GF_LOG_ALERT, tot_size);
return NULL;
}
gf_mem_set_acct_info (xl, &ptr, size, type);
return (void *)ptr;
}
void *
__gf_realloc (void *ptr, size_t size)
{
size_t tot_size = 0;
char *orig_ptr = NULL;
xlator_t *xl = NULL;
uint32_t type = 0;
if (!gf_mem_acct_enable)
return REALLOC (ptr, size);
tot_size = size + GF_MEM_HEADER_SIZE + GF_MEM_TRAILER_SIZE;
orig_ptr = (char *)ptr - 8 - 4;
GF_ASSERT (*(uint32_t *)orig_ptr == GF_MEM_HEADER_MAGIC);
orig_ptr = orig_ptr - sizeof(xlator_t *);
xl = *((xlator_t **)orig_ptr);
orig_ptr = (char *)ptr - GF_MEM_HEADER_SIZE;
type = *(uint32_t *)orig_ptr;
ptr = realloc (orig_ptr, tot_size);
if (!ptr) {
gf_log_nomem ("", GF_LOG_ALERT, tot_size);
return NULL;
}
gf_mem_set_acct_info (xl, (char **)&ptr, size, type);
return (void *)ptr;
}
int
gf_vasprintf (char **string_ptr, const char *format, va_list arg)
{
va_list arg_save;
char *str = NULL;
int size = 0;
int rv = 0;
if (!string_ptr || !format)
return -1;
va_copy (arg_save, arg);
size = vsnprintf (NULL, 0, format, arg);
size++;
str = GF_MALLOC (size, gf_common_mt_asprintf);
if (str == NULL) {
/* log is done in GF_MALLOC itself */
return -1;
}
rv = vsnprintf (str, size, format, arg_save);
*string_ptr = str;
return (rv);
}
int
gf_asprintf (char **string_ptr, const char *format, ...)
{
va_list arg;
int rv = 0;
va_start (arg, format);
rv = gf_vasprintf (string_ptr, format, arg);
va_end (arg);
return rv;
}
void
__gf_free (void *free_ptr)
{
size_t req_size = 0;
char *ptr = NULL;
uint32_t type = 0;
xlator_t *xl = NULL;
if (!gf_mem_acct_enable) {
FREE (free_ptr);
return;
}
if (!free_ptr)
return;
ptr = (char *)free_ptr - 8 - 4;
if (GF_MEM_HEADER_MAGIC != *(uint32_t *)ptr) {
//Possible corruption, assert here
GF_ASSERT (0);
}
*(uint32_t *)ptr = 0;
ptr = ptr - sizeof(xlator_t *);
memcpy (&xl, ptr, sizeof(xlator_t *));
if (!xl) {
//gf_free expects xl to be available
GF_ASSERT (0);
}
if (!xl->mem_acct.rec) {
ptr = (char *)free_ptr - GF_MEM_HEADER_SIZE;
goto free;
}
ptr = ptr - sizeof(size_t);
memcpy (&req_size, ptr, sizeof (size_t));
ptr = ptr - 4;
type = *(uint32_t *)ptr;
if (GF_MEM_TRAILER_MAGIC != *(uint32_t *)
((char *)free_ptr + req_size)) {
// This points to a memory overrun
GF_ASSERT (0);
}
*(uint32_t *) ((char *)free_ptr + req_size) = 0;
LOCK (&xl->mem_acct.rec[type].lock);
{
xl->mem_acct.rec[type].size -= req_size;
xl->mem_acct.rec[type].num_allocs--;
}
UNLOCK (&xl->mem_acct.rec[type].lock);
free:
FREE (ptr);
}
struct mem_pool *
mem_pool_new_fn (unsigned long sizeof_type,
unsigned long count)
{
struct mem_pool *mem_pool = NULL;
unsigned long padded_sizeof_type = 0;
void *pool = NULL;
int i = 0;
struct list_head *list = NULL;
if (!sizeof_type || !count) {
gf_log ("mem-pool", GF_LOG_ERROR, "invalid argument");
return NULL;
}
padded_sizeof_type = sizeof_type + GF_MEM_POOL_PAD_BOUNDARY;
mem_pool = GF_CALLOC (sizeof (*mem_pool), 1, gf_common_mt_mem_pool);
if (!mem_pool)
return NULL;
LOCK_INIT (&mem_pool->lock);
INIT_LIST_HEAD (&mem_pool->list);
mem_pool->padded_sizeof_type = padded_sizeof_type;
mem_pool->cold_count = count;
mem_pool->real_sizeof_type = sizeof_type;
pool = GF_CALLOC (count, padded_sizeof_type, gf_common_mt_long);
if (!pool) {
GF_FREE (mem_pool);
return NULL;
}
for (i = 0; i < count; i++) {
list = pool + (i * (padded_sizeof_type));
INIT_LIST_HEAD (list);
list_add_tail (list, &mem_pool->list);
}
mem_pool->pool = pool;
mem_pool->pool_end = pool + (count * (padded_sizeof_type));
return mem_pool;
}
void*
mem_get0 (struct mem_pool *mem_pool)
{
void *ptr = NULL;
if (!mem_pool) {
gf_log ("mem-pool", GF_LOG_ERROR, "invalid argument");
return NULL;
}
ptr = mem_get(mem_pool);
if (ptr)
memset(ptr, 0, mem_pool->real_sizeof_type);
return ptr;
}
void *
mem_get (struct mem_pool *mem_pool)
{
struct list_head *list = NULL;
void *ptr = NULL;
int *in_use = NULL;
if (!mem_pool) {
gf_log ("mem-pool", GF_LOG_ERROR, "invalid argument");
return NULL;
}
LOCK (&mem_pool->lock);
{
if (mem_pool->cold_count) {
list = mem_pool->list.next;
list_del (list);
mem_pool->hot_count++;
mem_pool->cold_count--;
ptr = list;
in_use = (ptr + GF_MEM_POOL_LIST_BOUNDARY);
*in_use = 1;
goto fwd_addr_out;
}
/* This is a problem area. If we've run out of
* chunks in our slab above, we need to allocate
* enough memory to service this request.
* The problem is, these indvidual chunks will fail
* the first address range check in __is_member. Now, since
* we're not allocating a full second slab, we wont have
* enough info perform the range check in __is_member.
*
* I am working around this by performing a regular allocation
* , just the way the caller would've done when not using the
* mem-pool. That also means, we're not padding the size with
* the list_head structure because, this will not be added to
* the list of chunks that belong to the mem-pool allocated
* initially.
*
* This is the best we can do without adding functionality for
* managing multiple slabs. That does not interest us at present
* because it is too much work knowing that a better slab
* allocator is coming RSN.
*/
ptr = MALLOC (mem_pool->real_sizeof_type);
/* Memory coming from the heap need not be transformed from a
* chunkhead to a usable pointer since it is not coming from
* the pool.
*/
goto unlocked_out;
}
fwd_addr_out:
ptr = mem_pool_chunkhead2ptr (ptr);
unlocked_out:
UNLOCK (&mem_pool->lock);
return ptr;
}
static int
__is_member (struct mem_pool *pool, void *ptr)
{
if (!pool || !ptr) {
gf_log ("mem-pool", GF_LOG_ERROR, "invalid argument");
return -1;
}
if (ptr < pool->pool || ptr >= pool->pool_end)
return 0;
if ((mem_pool_ptr2chunkhead (ptr) - pool->pool)
% pool->padded_sizeof_type)
return -1;
return 1;
}
void
mem_put (struct mem_pool *pool, void *ptr)
{
struct list_head *list = NULL;
int *in_use = NULL;
void *head = NULL;
if (!pool || !ptr) {
gf_log ("mem-pool", GF_LOG_ERROR, "invalid argument");
return;
}
LOCK (&pool->lock);
{
switch (__is_member (pool, ptr))
{
case 1:
list = head = mem_pool_ptr2chunkhead (ptr);
in_use = (head + GF_MEM_POOL_LIST_BOUNDARY);
if (!is_mem_chunk_in_use(in_use)) {
gf_log_callingfn ("mem-pool", GF_LOG_CRITICAL,
"mem_put called on freed ptr %p of mem "
"pool %p", ptr, pool);
break;
}
pool->hot_count--;
pool->cold_count++;
*in_use = 0;
list_add (list, &pool->list);
break;
case -1:
/* For some reason, the address given is within
* the address range of the mem-pool but does not align
* with the expected start of a chunk that includes
* the list headers also. Sounds like a problem in
* layers of clouds up above us. ;)
*/
abort ();
break;
case 0:
/* The address is outside the range of the mem-pool. We
* assume here that this address was allocated at a
* point when the mem-pool was out of chunks in mem_get
* or the programmer has made a mistake by calling the
* wrong de-allocation interface. We do
* not have enough info to distinguish between the two
* situations.
*/
FREE (ptr);
break;
default:
/* log error */
break;
}
}
UNLOCK (&pool->lock);
}
void
mem_pool_destroy (struct mem_pool *pool)
{
if (!pool)
return;
LOCK_DESTROY (&pool->lock);
GF_FREE (pool->pool);
GF_FREE (pool);
return;
}
|