-
Notifications
You must be signed in to change notification settings - Fork 33
/
slabs.c
517 lines (433 loc) · 16 KB
/
slabs.c
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
/* -*- Mode: C; tab-width: 4; c-basic-offset: 4; indent-tabs-mode: nil -*- */
/*
* Slabs memory allocation, based on powers-of-N. Slabs are up to 1MB in size
* and are divided into chunks. The chunk sizes start off at the size of the
* "item" structure plus space for a small key and value. They increase by
* a multiplier factor from there, up to half the maximum slab size. The last
* slab size is always 1MB, since that's the maximum item size allowed by the
* memcached protocol.
*/
#include "memcached.h"
#include <errno.h>
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <assert.h>
#include <pthread.h>
#include "log.h"
/* powers-of-N allocation structures */
typedef struct {
unsigned int size; /* sizes of items */
unsigned int perslab; /* how many items per slab */
void **slots; /* list of item ptrs */
unsigned int sl_total; /* size of previous array */
unsigned int sl_curr; /* first free slot */
void *end_page_ptr; /* pointer to next free item at end of page, or 0 */
unsigned int end_page_free; /* number of items remaining at end of last alloced page */
unsigned int slabs; /* how many slabs were allocated for this class */
void **slab_list; /* array of slab pointers */
unsigned int list_size; /* size of prev array */
unsigned int killing; /* index+1 of dying slab, or zero if none */
size_t requested; /* The number of requested bytes */
} slabclass_t;
static slabclass_t slabclass[MAX_NUMBER_OF_SLAB_CLASSES];
static size_t mem_limit = 0;
static size_t mem_malloced = 0;
static size_t power_largest;
static void *mem_base = NULL;
static void *mem_current = NULL;
static size_t mem_avail = 0;
/**
* Access to the slab allocator is protected by this lock
*/
static pthread_mutex_t slabs_lock = PTHREAD_MUTEX_INITIALIZER;
/*
* Forward Declarations
*/
static int do_slabs_newslab(const unsigned int id);
static void *memory_allocate(size_t size);
#ifndef DONT_PREALLOC_SLABS
/* Preallocate as many slab pages as possible (called from slabs_init)
on start-up, so users don't get confused out-of-memory errors when
they do have free (in-slab) space, but no space to make new slabs.
if maxslabs is 18 (POWER_LARGEST - POWER_SMALLEST + 1), then all
slab types can be made. if max memory is less than 18 MB, only the
smaller ones will be made. */
static void slabs_preallocate (const unsigned int maxslabs);
#endif
/*
* Figures out which slab class (chunk size) is required to store an item of
* a given size.
*
* Given object size, return id to use when allocating/freeing memory for object
* 0 means error: can't store such a large object
*/
unsigned int slabs_clsid(const size_t size) {
size_t res = POWER_SMALLEST;
if (size == 0)
return 0;
while (size > slabclass[res].size)
if (res++ == power_largest) /* won't fit in the biggest slab */
return 0;
return res;
}
/**
* Determines the chunk sizes and initializes the slab class descriptors
* accordingly.
*/
void slabs_init(const size_t limit, const double factor, const bool prealloc) {
int i = POWER_SMALLEST - 1;
unsigned int size = sizeof(item) + settings.chunk_size;
/* Factor of 2.0 means use the default memcached behavior */
if (factor == 2.0 && size < 128)
size = 128;
mem_limit = limit;
if (prealloc) {
/* Allocate everything in a big chunk with malloc */
mem_base = malloc(mem_limit);
if (mem_base != NULL) {
mem_current = mem_base;
mem_avail = mem_limit;
} else {
moxi_log_write("Warning: Failed to allocate requested memory in"
" one large chunk.\nWill allocate in smaller chunks\n");
}
}
memset(slabclass, 0, sizeof(slabclass));
while (++i < POWER_LARGEST && size <= POWER_BLOCK / 2) {
/* Make sure items are always n-byte aligned */
if (size % CHUNK_ALIGN_BYTES)
size += CHUNK_ALIGN_BYTES - (size % CHUNK_ALIGN_BYTES);
slabclass[i].size = size;
slabclass[i].perslab = POWER_BLOCK / slabclass[i].size;
size *= factor;
if (settings.verbose > 1) {
moxi_log_write("slab class %3d: chunk size %6u perslab %5u\n",
i, slabclass[i].size, slabclass[i].perslab);
}
}
power_largest = i;
slabclass[power_largest].size = POWER_BLOCK;
slabclass[power_largest].perslab = 1;
/* for the test suite: faking of how much we've already malloc'd */
{
char *t_initial_malloc = getenv("T_MEMD_INITIAL_MALLOC");
if (t_initial_malloc) {
mem_malloced = (size_t)atol(t_initial_malloc);
}
}
#ifndef DONT_PREALLOC_SLABS
{
char *pre_alloc = getenv("T_MEMD_SLABS_ALLOC");
if (pre_alloc == NULL || atoi(pre_alloc) != 0) {
slabs_preallocate(power_largest);
}
}
#endif
}
#ifndef DONT_PREALLOC_SLABS
static void slabs_preallocate (const unsigned int maxslabs) {
int i;
unsigned int prealloc = 0;
/* pre-allocate a 1MB slab in every size class so people don't get
confused by non-intuitive "SERVER_ERROR out of memory"
messages. this is the most common question on the mailing
list. if you really don't want this, you can rebuild without
these three lines. */
for (i = POWER_SMALLEST; i <= POWER_LARGEST; i++) {
if (++prealloc > maxslabs)
return;
do_slabs_newslab(i);
}
}
#endif
static int grow_slab_list (const unsigned int id) {
slabclass_t *p = &slabclass[id];
if (p->slabs == p->list_size) {
size_t new_size = (p->list_size != 0) ? p->list_size * 2 : 16;
void *new_list = realloc(p->slab_list, new_size * sizeof(void *));
if (new_list == 0) return 0;
p->list_size = new_size;
p->slab_list = new_list;
}
return 1;
}
static int do_slabs_newslab(const unsigned int id) {
slabclass_t *p = &slabclass[id];
#ifdef ALLOW_SLABS_REASSIGN
int len = POWER_BLOCK;
#else
int len = p->size * p->perslab;
#endif
char *ptr;
if ((mem_limit && mem_malloced + len > mem_limit && p->slabs > 0) ||
(grow_slab_list(id) == 0) ||
((ptr = memory_allocate((size_t)len)) == 0)) {
MEMCACHED_SLABS_SLABCLASS_ALLOCATE_FAILED(id);
return 0;
}
memset(ptr, 0, (size_t)len);
p->end_page_ptr = ptr;
p->end_page_free = p->perslab;
p->slab_list[p->slabs++] = ptr;
mem_malloced += len;
MEMCACHED_SLABS_SLABCLASS_ALLOCATE(id);
return 1;
}
/*@null@*/
static void *do_slabs_alloc(const size_t size, unsigned int id) {
slabclass_t *p;
void *ret = NULL;
if (id < POWER_SMALLEST || id > power_largest) {
MEMCACHED_SLABS_ALLOCATE_FAILED(size, 0);
return NULL;
}
p = &slabclass[id];
assert(p->sl_curr == 0 || ((item *)p->slots[p->sl_curr - 1])->slabs_clsid == 0);
#ifdef USE_SYSTEM_MALLOC
if (mem_limit && mem_malloced + size > mem_limit) {
MEMCACHED_SLABS_ALLOCATE_FAILED(size, id);
return 0;
}
mem_malloced += size;
ret = malloc(size);
MEMCACHED_SLABS_ALLOCATE(size, id, 0, ret);
return ret;
#endif
/* fail unless we have space at the end of a recently allocated page,
we have something on our freelist, or we could allocate a new page */
if (! (p->end_page_ptr != 0 || p->sl_curr != 0 ||
do_slabs_newslab(id) != 0)) {
/* We don't have more memory available */
ret = NULL;
} else if (p->sl_curr != 0) {
/* return off our freelist */
ret = p->slots[--p->sl_curr];
} else {
/* if we recently allocated a whole page, return from that */
assert(p->end_page_ptr != NULL);
ret = p->end_page_ptr;
if (--p->end_page_free != 0) {
p->end_page_ptr = ((caddr_t)p->end_page_ptr) + p->size;
} else {
p->end_page_ptr = 0;
}
}
if (ret) {
p->requested += size;
MEMCACHED_SLABS_ALLOCATE(size, id, p->size, ret);
} else {
MEMCACHED_SLABS_ALLOCATE_FAILED(size, id);
}
return ret;
}
static void do_slabs_free(void *ptr, const size_t size, unsigned int id) {
slabclass_t *p;
assert(((item *)ptr)->slabs_clsid == 0);
assert(id >= POWER_SMALLEST && id <= power_largest);
if (id < POWER_SMALLEST || id > power_largest)
return;
MEMCACHED_SLABS_FREE(size, id, ptr);
p = &slabclass[id];
#ifdef USE_SYSTEM_MALLOC
mem_malloced -= size;
free(ptr);
return;
#endif
if (p->sl_curr == p->sl_total) { /* need more space on the free list */
int new_size = (p->sl_total != 0) ? p->sl_total * 2 : 16; /* 16 is arbitrary */
void **new_slots = realloc(p->slots, new_size * sizeof(void *));
if (new_slots == 0)
return;
p->slots = new_slots;
p->sl_total = new_size;
}
p->slots[p->sl_curr++] = ptr;
p->requested -= size;
return;
}
static int nz_strcmp(int nzlength, const char *nz, const char *z) {
int zlength=strlen(z);
return (zlength == nzlength) && (strncmp(nz, z, zlength) == 0) ? 0 : -1;
}
bool get_stats(const char *stat_type, int nkey, ADD_STAT add_stats, void *c) {
bool ret = true;
if (add_stats != NULL) {
if (!stat_type) {
/* prepare general statistics for the engine */
APPEND_STAT("bytes", "%llu", (unsigned long long)stats.curr_bytes);
APPEND_STAT("curr_items", "%u", stats.curr_items);
APPEND_STAT("total_items", "%u", stats.total_items);
APPEND_STAT("evictions", "%llu",
(unsigned long long)stats.evictions);
} else if (nz_strcmp(nkey, stat_type, "items") == 0) {
item_stats(add_stats, c);
} else if (nz_strcmp(nkey, stat_type, "slabs") == 0) {
slabs_stats(add_stats, c);
} else if (nz_strcmp(nkey, stat_type, "sizes") == 0) {
item_stats_sizes(add_stats, c);
} else {
ret = false;
}
} else {
ret = false;
}
return ret;
}
/*@null@*/
static void do_slabs_stats(ADD_STAT add_stats, void *c) {
int i;
int total;
/* Get the per-thread stats which contain some interesting aggregates */
struct thread_stats thread_stats;
threadlocal_stats_aggregate(&thread_stats);
total = 0;
for (i = POWER_SMALLEST; i <= (int) power_largest; i++) {
slabclass_t *p = &slabclass[i];
if (p->slabs != 0) {
uint32_t perslab, slabs;
slabs = p->slabs;
perslab = p->perslab;
char key_str[STAT_KEY_LEN];
char val_str[STAT_VAL_LEN];
int klen = 0, vlen = 0;
APPEND_NUM_STAT(i, "chunk_size", "%u", p->size);
APPEND_NUM_STAT(i, "chunks_per_page", "%u", perslab);
APPEND_NUM_STAT(i, "total_pages", "%u", slabs);
APPEND_NUM_STAT(i, "total_chunks", "%u", slabs * perslab);
APPEND_NUM_STAT(i, "used_chunks", "%u",
slabs*perslab - p->sl_curr - p->end_page_free);
APPEND_NUM_STAT(i, "free_chunks", "%u", p->sl_curr);
APPEND_NUM_STAT(i, "free_chunks_end", "%u", p->end_page_free);
APPEND_NUM_STAT(i, "mem_requested", "%llu",
(unsigned long long)p->requested);
APPEND_NUM_STAT(i, "get_hits", "%llu",
(unsigned long long)thread_stats.slab_stats[i].get_hits);
APPEND_NUM_STAT(i, "cmd_set", "%llu",
(unsigned long long)thread_stats.slab_stats[i].set_cmds);
APPEND_NUM_STAT(i, "delete_hits", "%llu",
(unsigned long long)thread_stats.slab_stats[i].delete_hits);
APPEND_NUM_STAT(i, "incr_hits", "%llu",
(unsigned long long)thread_stats.slab_stats[i].incr_hits);
APPEND_NUM_STAT(i, "decr_hits", "%llu",
(unsigned long long)thread_stats.slab_stats[i].decr_hits);
APPEND_NUM_STAT(i, "cas_hits", "%llu",
(unsigned long long)thread_stats.slab_stats[i].cas_hits);
APPEND_NUM_STAT(i, "cas_badval", "%llu",
(unsigned long long)thread_stats.slab_stats[i].cas_badval);
total++;
}
}
/* add overall slab stats and append terminator */
APPEND_STAT("active_slabs", "%d", total);
APPEND_STAT("total_malloced", "%llu", (unsigned long long)mem_malloced);
add_stats(NULL, 0, NULL, 0, c);
}
#ifdef ALLOW_SLABS_REASSIGN
/* Blows away all the items in a slab class and moves its slabs to another
class. This is only used by the "slabs reassign" command, for manual tweaking
of memory allocation. It's disabled by default since it requires that all
slabs be the same size (which can waste space for chunk size mantissas of
other than 2.0).
1 = success
0 = fail
-1 = tried. busy. send again shortly. */
int do_slabs_reassign(unsigned char srcid, unsigned char dstid) {
void *slab, *slab_end;
slabclass_t *p, *dp;
void *iter;
bool was_busy = false;
if (srcid < POWER_SMALLEST || srcid > power_largest ||
dstid < POWER_SMALLEST || dstid > power_largest)
return 0;
p = &slabclass[srcid];
dp = &slabclass[dstid];
/* fail if src still populating, or no slab to give up in src */
if (p->end_page_ptr || ! p->slabs)
return 0;
/* fail if dst is still growing or we can't make room to hold its new one */
if (dp->end_page_ptr || ! grow_slab_list(dstid))
return 0;
if (p->killing == 0) p->killing = 1;
slab = p->slab_list[p->killing - 1];
slab_end = (char*)slab + POWER_BLOCK;
for (iter = slab; iter < slab_end; (char*)iter += p->size) {
item *it = (item *)iter;
if (it->slabs_clsid) {
if (it->refcount) was_busy = true;
item_unlink(it);
}
}
/* go through free list and discard items that are no longer part of this slab */
{
int fi;
for (fi = p->sl_curr - 1; fi >= 0; fi--) {
if (p->slots[fi] >= slab && p->slots[fi] < slab_end) {
p->sl_curr--;
if (p->sl_curr > fi) p->slots[fi] = p->slots[p->sl_curr];
}
}
}
if (was_busy) return -1;
/* if good, now move it to the dst slab class */
p->slab_list[p->killing - 1] = p->slab_list[p->slabs - 1];
p->slabs--;
p->killing = 0;
dp->slab_list[dp->slabs++] = slab;
dp->end_page_ptr = slab;
dp->end_page_free = dp->perslab;
/* this isn't too critical, but other parts of the code do asserts to
make sure this field is always 0. */
for (iter = slab; iter < slab_end; (char*)iter += dp->size) {
((item *)iter)->slabs_clsid = 0;
}
return 1;
}
int slabs_reassign(unsigned char srcid, unsigned char dstid) {
int ret;
pthread_mutex_lock(&slabs_lock);
ret = do_slabs_reassign(srcid, dstid);
pthread_mutex_unlock(&slabs_lock);
return ret;
}
#endif
static void *memory_allocate(size_t size) {
void *ret;
if (mem_base == NULL) {
/* We are not using a preallocated large memory chunk */
ret = malloc(size);
} else {
ret = mem_current;
if (size > mem_avail) {
return NULL;
}
/* mem_current pointer _must_ be aligned!!! */
if (size % CHUNK_ALIGN_BYTES) {
size += CHUNK_ALIGN_BYTES - (size % CHUNK_ALIGN_BYTES);
}
mem_current = ((char*)mem_current) + size;
if (size < mem_avail) {
mem_avail -= size;
} else {
mem_avail = 0;
}
}
return ret;
}
void *slabs_alloc(size_t size, unsigned int id) {
void *ret;
pthread_mutex_lock(&slabs_lock);
ret = do_slabs_alloc(size, id);
pthread_mutex_unlock(&slabs_lock);
return ret;
}
void slabs_free(void *ptr, size_t size, unsigned int id) {
pthread_mutex_lock(&slabs_lock);
do_slabs_free(ptr, size, id);
pthread_mutex_unlock(&slabs_lock);
}
void slabs_stats(ADD_STAT add_stats, void *c) {
pthread_mutex_lock(&slabs_lock);
do_slabs_stats(add_stats, c);
pthread_mutex_unlock(&slabs_lock);
}