summaryrefslogtreecommitdiffstats
path: root/src/responder/nss/nsssrv_mmap_cache.c
blob: 8cf27ca69f1afcf551a5aef77980d1bdb925918d (plain)
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
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
/*
   SSSD

   NSS Responder - Mmap Cache

   Copyright (C) Simo Sorce <ssorce@redhat.com>	2011

   This program is free software; you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
   the Free Software Foundation; either version 3 of the License, or
   (at your option) any later version.

   This program 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 General Public License for more details.

   You should have received a copy of the GNU General Public License
   along with this program.  If not, see <http://www.gnu.org/licenses/>.
*/

#include "util/util.h"
#include "confdb/confdb.h"
#include <sys/mman.h>
#include <fcntl.h>
#include "util/mmap_cache.h"
#include "responder/nss/nsssrv.h"
#include "responder/nss/nsssrv_mmap_cache.h"

/* arbitrary (avg of my /etc/passwd) */
#define SSS_AVG_PASSWD_PAYLOAD (MC_SLOT_SIZE * 4)
/* short group name and no gids (private user group */
#define SSS_AVG_GROUP_PAYLOAD (MC_SLOT_SIZE * 3)

#define MC_NEXT_BARRIER(val) ((((val) + 1) & 0x00ffffff) | 0xf0000000)

#define MC_RAISE_BARRIER(m) do { \
    m->b2 = MC_NEXT_BARRIER(m->b1); \
    __sync_synchronize(); \
} while (0)

#define MC_LOWER_BARRIER(m) do { \
    __sync_synchronize(); \
    m->b1 = m->b2; \
} while (0)

#define MC_RAISE_INVALID_BARRIER(m) do { \
    m->b2 = MC_INVALID_VAL; \
    __sync_synchronize(); \
} while (0)

struct sss_mc_ctx {
    char *name;             /* mmap cache name */
    enum sss_mc_type type;  /* mmap cache type */
    char *file;             /* mmap cache file name */
    int fd;                 /* file descriptor */

    uint32_t seed;          /* pseudo-random seed to avoid collision attacks */
    time_t valid_time_slot; /* maximum time the entry is valid in seconds */

    void *mmap_base;        /* base address of mmap */
    size_t mmap_size;       /* total size of mmap */

    uint32_t *hash_table;   /* hash table address (in mmap) */
    uint32_t ht_size;       /* size of hash table */

    uint8_t *free_table;    /* free list bitmaps */
    uint32_t ft_size;       /* size of free table */
    uint32_t next_slot;     /* the next slot after last allocation */

    uint8_t *data_table;    /* data table address (in mmap) */
    uint32_t dt_size;       /* size of data table */
};

#define MC_FIND_BIT(base, num) \
    uint32_t n = (num); \
    uint8_t *b = (base) + n / 8; \
    uint8_t c = 0x80 >> (n % 8);

#define MC_SET_BIT(base, num) do { \
    MC_FIND_BIT(base, num) \
    *b |= c; \
} while (0)

#define MC_CLEAR_BIT(base, num) do { \
    MC_FIND_BIT(base, num) \
    *b &= ~c; \
} while (0)

#define MC_PROBE_BIT(base, num, used) do { \
    MC_FIND_BIT(base, num) \
    if (*b & c) used = true; \
    else used = false; \
} while (0)

static uint32_t sss_mc_hash(struct sss_mc_ctx *mcc,
                            const char *key, size_t len)
{
    return murmurhash3(key, len, mcc->seed) % MC_HT_ELEMS(mcc->ht_size);
}

static void sss_mc_add_rec_to_chain(struct sss_mc_ctx *mcc,
                                    struct sss_mc_rec *rec,
                                    uint32_t hash)
{
    struct sss_mc_rec *cur;
    uint32_t slot;

    if (hash > mcc->ht_size) {
        /* Invalid hash. This should never happen, but better
         * return than trying to access out of bounds memory */
        return;
    }

    slot = mcc->hash_table[hash];
    if (slot == MC_INVALID_VAL) {
        /* no previous record/collision, just add to hash table */
        mcc->hash_table[hash] = MC_PTR_TO_SLOT(mcc->data_table, rec);
        return;
    }

    do {
        cur = MC_SLOT_TO_PTR(mcc->data_table, slot, struct sss_mc_rec);
        if (cur == rec) {
            /* rec already stored in hash chain */
            return;
        }
        slot = cur->next;
    } while (slot != MC_INVALID_VAL);
    /* end of chain, append our record here */

    /* changing a single uint32_t is atomic, so there is no
     * need to use barriers in this case */
    cur->next = MC_PTR_TO_SLOT(mcc->data_table, rec);
}

static void sss_mc_rm_rec_from_chain(struct sss_mc_ctx *mcc,
                                     struct sss_mc_rec *rec,
                                     uint32_t hash)
{
    struct sss_mc_rec *prev = NULL;
    struct sss_mc_rec *cur = NULL;
    uint32_t slot;

    if (hash > mcc->ht_size) {
        /* Invalid hash. This should never happen, but better
         * return than trying to access out of bounds memory */
        return;
    }

    slot = mcc->hash_table[hash];
    cur = MC_SLOT_TO_PTR(mcc->data_table, slot, struct sss_mc_rec);
    if (cur == rec) {
        mcc->hash_table[hash] = rec->next;
    } else {
        slot = cur->next;
        while (slot != MC_INVALID_VAL) {
            prev = cur;
            cur = MC_SLOT_TO_PTR(mcc->data_table, slot, struct sss_mc_rec);
            if (cur == rec) {
                /* changing a single uint32_t is atomic, so there is no
                 * need to use barriers in this case */
                prev->next = cur->next;
                slot = MC_INVALID_VAL;
            } else {
                slot = cur->next;
            }
        }
    }
}

static void sss_mc_free_slots(struct sss_mc_ctx *mcc, struct sss_mc_rec *rec)
{
    uint32_t slot;
    uint32_t num;
    uint32_t i;

    slot = MC_PTR_TO_SLOT(mcc->data_table, rec);
    num = MC_SIZE_TO_SLOTS(rec->len);
    for (i = 0; i < num; i++) {
        MC_CLEAR_BIT(mcc->free_table, slot + i);
    }
}

static void sss_mc_invalidate_rec(struct sss_mc_ctx *mcc,
                                  struct sss_mc_rec *rec)
{
    if (rec->b1 == MC_INVALID_VAL) {
        /* record already invalid */
        return;
    }

    /* Remove from hash chains */
    /* hash chain 1 */
    sss_mc_rm_rec_from_chain(mcc, rec, rec->hash1);
    /* hash chain 2 */
    sss_mc_rm_rec_from_chain(mcc, rec, rec->hash2);

    /* Clear from free_table */
    sss_mc_free_slots(mcc, rec);

    /* Invalidate record fields */
    MC_RAISE_INVALID_BARRIER(rec);
    memset(rec->data, MC_INVALID_VAL8, ((MC_SLOT_SIZE * MC_SIZE_TO_SLOTS(rec->len))
                                        - sizeof(struct sss_mc_rec)));
    rec->len = MC_INVALID_VAL32;
    rec->expire = MC_INVALID_VAL64;
    rec->next = MC_INVALID_VAL32;
    rec->hash1 = MC_INVALID_VAL32;
    rec->hash2 = MC_INVALID_VAL32;
    MC_LOWER_BARRIER(rec);
}

static bool sss_mc_is_valid_rec(struct sss_mc_ctx *mcc, struct sss_mc_rec *rec)
{
    struct sss_mc_rec *self;
    uint32_t slot;

    if (((uint8_t *)rec < mcc->data_table) ||
        ((uint8_t *)rec > (mcc->data_table + mcc->dt_size - MC_SLOT_SIZE))) {
        return false;
    }

    if ((rec->b1 == MC_INVALID_VAL) ||
        (rec->b1 != rec->b2)) {
        return false;
    }

    if ((rec->len == MC_INVALID_VAL32) ||
        (rec->len > (mcc->dt_size - ((uint8_t *)rec - mcc->data_table)))) {
        return false;
    }

    if (rec->expire == MC_INVALID_VAL64) {
        return false;
    }

    /* rec->next can be invalid if there are no next records */

    if (rec->hash1 == MC_INVALID_VAL32) {
        return false;
    } else {
        self = NULL;
        slot = mcc->hash_table[rec->hash1];
        while (slot != MC_INVALID_VAL32 && self != rec) {
            self = MC_SLOT_TO_PTR(mcc->data_table, slot, struct sss_mc_rec);
            slot = self->next;
        }
        if (self != rec) {
            return false;
        }
    }
    if (rec->hash2 != MC_INVALID_VAL32) {
        self = NULL;
        slot = mcc->hash_table[rec->hash2];
        while (slot != MC_INVALID_VAL32 && self != rec) {
            self = MC_SLOT_TO_PTR(mcc->data_table, slot, struct sss_mc_rec);
            slot = self->next;
        }
        if (self != rec) {
            return false;
        }
    }

    /* all tests passed */
    return true;
}

/* FIXME: This is a very simplistic, inefficient, memory allocator,
 * it will just free the oldest entries regardless of expiration if it
 * cycled the whole freebits map and found no empty slot */
static errno_t sss_mc_find_free_slots(struct sss_mc_ctx *mcc,
                                      int num_slots, uint32_t *free_slot)
{
    struct sss_mc_rec *rec;
    uint32_t tot_slots;
    uint32_t cur;
    uint32_t i;
    uint32_t t;
    bool used;

    tot_slots = mcc->ft_size * 8;

    /* Try to find a free slot w/o removing anything first */
    /* FIXME: is it really worth it ? May be it is easier to
     * just recycle the next set of slots ? */
    if ((mcc->next_slot + num_slots) > tot_slots) {
        cur = 0;
    } else {
        cur = mcc->next_slot;
    }

    /* search for enough (num_slots) consecutive zero bits, indicating
     * consecutive empty slots */
    for (i = 0; i < mcc->ft_size; i++) {
        t = cur / 8;
        /* if all full in this byte skip directly to the next */
        if (mcc->free_table[t] == 0xff) {
            cur = ((cur + 8) & ~7);
            if (cur >= tot_slots) {
                cur = 0;
            }
            continue;
        }

        /* at least one bit in this byte is marked as empty */
        for (t = ((cur + 8) & ~7) ; cur < t; cur++) {
            MC_PROBE_BIT(mcc->free_table, cur, used);
            if (!used) break;
        }
        /* check if we have enough slots before hitting the table end */
        if ((cur + num_slots) > tot_slots) {
            cur = 0;
            continue;
        }

        /* check if we have at least num_slots empty starting from the first
         * we found in the previous steps */
        for (t = cur + num_slots; cur < t; cur++) {
            MC_PROBE_BIT(mcc->free_table, cur, used);
            if (used) break;
        }
        if (cur == t) {
            /* ok found num_slots consecutive free bits */
            *free_slot = cur - num_slots;
            return EOK;
        }
    }

    /* no free slots found, free occupied slots after next_slot */
    if ((mcc->next_slot + num_slots) > tot_slots) {
        cur = 0;
    } else {
        cur = mcc->next_slot;
    }
    for (i = 0; i < num_slots; i++) {
        MC_PROBE_BIT(mcc->free_table, cur + i, used);
        if (used) {
            /* the first used slot should be a record header, however we
             * carefully check it is a valid header and hardfail if not */
            rec = MC_SLOT_TO_PTR(mcc->data_table, cur + i, struct sss_mc_rec);
            if (!sss_mc_is_valid_rec(mcc, rec)) {
                /* this is a fatal error, the caller should probaly just
                 * invalidate the whole cache */
                return EFAULT;
            }
            /* next loop skip the whole record */
            i += MC_SIZE_TO_SLOTS(rec->len) - 1;

            /* finally invalidate record completely */
            sss_mc_invalidate_rec(mcc, rec);
        }
    }

    mcc->next_slot = cur + num_slots;
    *free_slot = cur;
    return EOK;
}

static struct sss_mc_rec *sss_mc_find_record(struct sss_mc_ctx *mcc,
                                             struct sized_string *key)
{
    struct sss_mc_rec *rec;
    uint32_t hash;
    uint32_t slot;
    rel_ptr_t name_ptr;
    char *t_key;

    hash = sss_mc_hash(mcc, key->str, key->len);

    slot = mcc->hash_table[hash];
    if (slot > MC_SIZE_TO_SLOTS(mcc->dt_size)) {
        return NULL;
    }

    while (slot != MC_INVALID_VAL) {
        rec = MC_SLOT_TO_PTR(mcc->data_table, slot, struct sss_mc_rec);
        name_ptr = *((rel_ptr_t *)rec->data);

        t_key = (char *)rec->data + name_ptr;
        if (strcmp(key->str, t_key) == 0) {
            break;
        }

        slot = rec->next;
    }

    if (slot == MC_INVALID_VAL) {
        return NULL;
    }

    return rec;
}

static errno_t sss_mc_get_record(struct sss_mc_ctx **_mcc,
                                 size_t rec_len,
                                 struct sized_string *key,
                                 struct sss_mc_rec **_rec)
{
    struct sss_mc_ctx *mcc = *_mcc;
    struct sss_mc_rec *old_rec = NULL;
    struct sss_mc_rec *rec;
    int old_slots;
    int num_slots;
    uint32_t base_slot;
    errno_t ret;
    int i;

    num_slots = MC_SIZE_TO_SLOTS(rec_len);

    old_rec = sss_mc_find_record(mcc, key);
    if (old_rec) {
        old_slots = MC_SIZE_TO_SLOTS(old_rec->len);

        if (old_slots == num_slots) {
            *_rec = old_rec;
            return EOK;
        }

        /* slot size changed, invalidate record and fall through to get a
        * fully new record */
        sss_mc_invalidate_rec(mcc, old_rec);
    }

    /* we are going to use more space, find enough free slots */
    ret = sss_mc_find_free_slots(mcc, num_slots, &base_slot);
    if (ret != EOK) {
        if (ret == EFAULT) {
            DEBUG(SSSDBG_CRIT_FAILURE,
                  ("Fatal internal mmap cache error, invalidating cache!\n"));
            (void)sss_mmap_cache_reinit(talloc_parent(mcc), -1, -1, _mcc);
        }
        return ret;
    }

    rec = MC_SLOT_TO_PTR(mcc->data_table, base_slot, struct sss_mc_rec);

    /* mark as not valid yet */
    MC_RAISE_INVALID_BARRIER(rec);
    rec->len = rec_len;
    rec->next = MC_INVALID_VAL;
    MC_LOWER_BARRIER(rec);

    /* and now mark slots as used */
    for (i = 0; i < num_slots; i++) {
        MC_SET_BIT(mcc->free_table, base_slot + i);
    }

    *_rec = rec;
    return EOK;
}


/***************************************************************************
 * generic invalidation
 ***************************************************************************/

static errno_t sss_mmap_cache_invalidate(struct sss_mc_ctx *mcc,
                                         struct sized_string *key)
{
    struct sss_mc_rec *rec;

    rec = sss_mc_find_record(mcc, key);
    if (rec == NULL) {
        /* nothing to invalidate */
        return ENOENT;
    }

    sss_mc_invalidate_rec(mcc, rec);

    return EOK;
}

/***************************************************************************
 * passwd map
 ***************************************************************************/

errno_t sss_mmap_cache_pw_store(struct sss_mc_ctx **_mcc,
                                struct sized_string *name,
                                struct sized_string *pw,
                                uid_t uid, gid_t gid,
                                struct sized_string *gecos,
                                struct sized_string *homedir,
                                struct sized_string *shell)
{
    struct sss_mc_ctx *mcc = *_mcc;
    struct sss_mc_rec *rec;
    struct sss_mc_pwd_data *data;
    struct sized_string uidkey;
    char uidstr[11];
    size_t data_len;
    size_t rec_len;
    size_t pos;
    int ret;

    ret = snprintf(uidstr, 11, "%ld", (long)uid);
    if (ret > 10) {
        return EINVAL;
    }
    to_sized_string(&uidkey, uidstr);

    data_len = name->len + pw->len + gecos->len + homedir->len + shell->len;
    rec_len = sizeof(struct sss_mc_rec) +
              sizeof(struct sss_mc_pwd_data) +
              data_len;
    if (rec_len > mcc->dt_size) {
        return ENOMEM;
    }

    ret = sss_mc_get_record(_mcc, rec_len, name, &rec);
    if (ret != EOK) {
        return ret;
    }

    data = (struct sss_mc_pwd_data *)rec->data;
    pos = 0;

    MC_RAISE_BARRIER(rec);

    /* header */
    rec->len = rec_len;
    rec->expire = time(NULL) + mcc->valid_time_slot;
    rec->hash1 = sss_mc_hash(mcc, name->str, name->len);
    rec->hash2 = sss_mc_hash(mcc, uidkey.str, uidkey.len);

    /* passwd struct */
    data->name = MC_PTR_DIFF(data->strs, data);
    data->uid = uid;
    data->gid = gid;
    data->strs_len = data_len;
    memcpy(&data->strs[pos], name->str, name->len);
    pos += name->len;
    memcpy(&data->strs[pos], pw->str, pw->len);
    pos += pw->len;
    memcpy(&data->strs[pos], gecos->str, gecos->len);
    pos += gecos->len;
    memcpy(&data->strs[pos], homedir->str, homedir->len);
    pos += homedir->len;
    memcpy(&data->strs[pos], shell->str, shell->len);
    pos += shell->len;

    MC_LOWER_BARRIER(rec);

    /* finally chain the rec in the hash table */
    /* name hash first */
    sss_mc_add_rec_to_chain(mcc, rec, rec->hash1);
    /* then uid */
    sss_mc_add_rec_to_chain(mcc, rec, rec->hash2);

    return EOK;
}

errno_t sss_mmap_cache_pw_invalidate(struct sss_mc_ctx *mcc,
                                     struct sized_string *name)
{
    return sss_mmap_cache_invalidate(mcc, name);
}

errno_t sss_mmap_cache_pw_invalidate_uid(struct sss_mc_ctx *mcc, uid_t uid)
{
    struct sss_mc_rec *rec;
    struct sss_mc_pwd_data *data;
    uint32_t hash;
    uint32_t slot;
    char *uidstr;
    errno_t ret;

    uidstr = talloc_asprintf(NULL, "%ld", (long)uid);
    if (!uidstr) {
        return ENOMEM;
    }

    hash = sss_mc_hash(mcc, uidstr, strlen(uidstr) + 1);

    slot = mcc->hash_table[hash];
    if (slot > MC_SIZE_TO_SLOTS(mcc->dt_size)) {
        ret = ENOENT;
        goto done;
    }

    while (slot != MC_INVALID_VAL) {
        rec = MC_SLOT_TO_PTR(mcc->data_table, slot, struct sss_mc_rec);
        data = (struct sss_mc_pwd_data *)(&rec->data);

        if (uid == data->uid) {
            break;
        }

        slot = rec->next;
    }

    if (slot == MC_INVALID_VAL) {
        ret = ENOENT;
        goto done;
    }

    sss_mc_invalidate_rec(mcc, rec);

    ret = EOK;

done:
    talloc_zfree(uidstr);
    return ret;
}

/***************************************************************************
 * group map
 ***************************************************************************/

int sss_mmap_cache_gr_store(struct sss_mc_ctx **_mcc,
                            struct sized_string *name,
                            struct sized_string *pw,
                            gid_t gid, size_t memnum,
                            char *membuf, size_t memsize)
{
    struct sss_mc_ctx *mcc = *_mcc;
    struct sss_mc_rec *rec;
    struct sss_mc_grp_data *data;
    struct sized_string gidkey;
    char gidstr[11];
    size_t data_len;
    size_t rec_len;
    size_t pos;
    int ret;

    ret = snprintf(gidstr, 11, "%ld", (long)gid);
    if (ret > 10) {
        return EINVAL;
    }
    to_sized_string(&gidkey, gidstr);

    data_len = name->len + pw->len + memsize;
    rec_len = sizeof(struct sss_mc_rec) +
              sizeof(struct sss_mc_grp_data) +
              data_len;
    if (rec_len > mcc->dt_size) {
        return ENOMEM;
    }

    ret = sss_mc_get_record(_mcc, rec_len, name, &rec);
    if (ret != EOK) {
        return ret;
    }

    data = (struct sss_mc_grp_data *)rec->data;
    pos = 0;

    MC_RAISE_BARRIER(rec);

    /* header */
    rec->len = rec_len;
    rec->expire = time(NULL) + mcc->valid_time_slot;
    rec->hash1 = sss_mc_hash(mcc, name->str, name->len);
    rec->hash2 = sss_mc_hash(mcc, gidkey.str, gidkey.len);

    /* group struct */
    data->name = MC_PTR_DIFF(data->strs, data);
    data->gid = gid;
    data->members = memnum;
    data->strs_len = data_len;
    memcpy(&data->strs[pos], name->str, name->len);
    pos += name->len;
    memcpy(&data->strs[pos], pw->str, pw->len);
    pos += pw->len;
    memcpy(&data->strs[pos], membuf, memsize);
    pos += memsize;

    MC_LOWER_BARRIER(rec);

    /* finally chain the rec in the hash table */
    /* name hash first */
    sss_mc_add_rec_to_chain(mcc, rec, rec->hash1);
    /* then gid */
    sss_mc_add_rec_to_chain(mcc, rec, rec->hash2);

    return EOK;
}

errno_t sss_mmap_cache_gr_invalidate(struct sss_mc_ctx *mcc,
                                     struct sized_string *name)
{
    return sss_mmap_cache_invalidate(mcc, name);
}

errno_t sss_mmap_cache_gr_invalidate_gid(struct sss_mc_ctx *mcc, gid_t gid)
{
    struct sss_mc_rec *rec;
    struct sss_mc_grp_data *data;
    uint32_t hash;
    uint32_t slot;
    char *gidstr;
    errno_t ret;

    gidstr = talloc_asprintf(NULL, "%ld", (long)gid);
    if (!gidstr) {
        return ENOMEM;
    }

    hash = sss_mc_hash(mcc, gidstr, strlen(gidstr) + 1);

    slot = mcc->hash_table[hash];
    if (slot > MC_SIZE_TO_SLOTS(mcc->dt_size)) {
        ret = ENOENT;
        goto done;
    }

    while (slot != MC_INVALID_VAL) {
        rec = MC_SLOT_TO_PTR(mcc->data_table, slot, struct sss_mc_rec);
        data = (struct sss_mc_grp_data *)(&rec->data);

        if (gid == data->gid) {
            break;
        }

        slot = rec->next;
    }

    if (slot == MC_INVALID_VAL) {
        ret = ENOENT;
        goto done;
    }

    sss_mc_invalidate_rec(mcc, rec);

    ret = EOK;

done:
    talloc_zfree(gidstr);
    return ret;
}


/***************************************************************************
 * initialization
 ***************************************************************************/

static errno_t sss_mc_set_recycled(int fd)
{
    uint32_t w = SSS_MC_HEADER_RECYCLED;
    struct sss_mc_header h;
    off_t offset;
    off_t pos;
    int ret;

    offset = MC_PTR_DIFF(&h.status, &h);

    pos = lseek(fd, offset, SEEK_SET);
    if (pos == -1) {
        /* What do we do now ? */
        return errno;
    }

    errno = 0;
    ret = sss_atomic_write_s(fd, (uint8_t *)&w, sizeof(h.status));
    if (ret == -1) {
        return errno;
    }

    if (ret != sizeof(h.status)) {
        /* Write error */
        return EIO;
    }

    return EOK;
}

/*
 * When we (re)create a new file we must mark the current file as recycled
 * so active clients will abandon its use asap.
 * We unlink the current file and make a new one
 */
static errno_t sss_mc_create_file(struct sss_mc_ctx *mc_ctx)
{
    mode_t old_mask;
    int ofd;
    int ret, uret;
    useconds_t t = 50000;
    int retries = 3;

    ofd = open(mc_ctx->file, O_RDWR);
    if (ofd != -1) {
        ret = sss_br_lock_file(ofd, 0, 1, retries, t);
        if (ret != EOK) {
            DEBUG(SSSDBG_FATAL_FAILURE,
                  ("Failed to lock file %s.\n", mc_ctx->file));
        }
        ret = sss_mc_set_recycled(ofd);
        if (ret) {
            DEBUG(SSSDBG_FATAL_FAILURE, ("Failed to mark mmap file %s as"
                                         " recycled: %d(%s)\n",
                                         mc_ctx->file, ret, strerror(ret)));
        }

        close(ofd);
    } else if (errno != ENOENT) {
        ret = errno;
        DEBUG(SSSDBG_CRIT_FAILURE,
              ("Failed to open old memory cache file %s: %d(%s).\n",
               mc_ctx->file, ret, strerror(ret)));
    }

    errno = 0;
    ret = unlink(mc_ctx->file);
    if (ret == -1 && errno != ENOENT) {
        ret = errno;
        DEBUG(SSSDBG_TRACE_FUNC, ("Failed to rm mmap file %s: %d(%s)\n",
                                  mc_ctx->file, ret, strerror(ret)));
    }

    /* temporarily relax umask as we need the file to be readable
     * by everyone for now */
    old_mask = umask(0022);

    errno = 0;
    mc_ctx->fd = open(mc_ctx->file, O_CREAT | O_EXCL | O_RDWR, 0644);
    umask(old_mask);
    if (mc_ctx->fd == -1) {
        ret = errno;
        DEBUG(SSSDBG_CRIT_FAILURE, ("Failed to open mmap file %s: %d(%s)\n",
                                    mc_ctx->file, ret, strerror(ret)));
        return ret;
    }

    ret = sss_br_lock_file(mc_ctx->fd, 0, 1, retries, t);
    if (ret != EOK) {
        DEBUG(SSSDBG_FATAL_FAILURE,
              ("Failed to lock file %s.\n", mc_ctx->file));
        close(mc_ctx->fd);
        mc_ctx->fd = -1;

        /* Report on unlink failures but don't overwrite the errno
         * from sss_br_lock_file
         */
        errno = 0;
        uret = unlink(mc_ctx->file);
        if (uret == -1) {
            uret = errno;
            DEBUG(SSSDBG_TRACE_FUNC, ("Failed to rm mmap file %s: %d(%s)\n",
                                    mc_ctx->file, uret, strerror(uret)));
        }

        return ret;
    }

    return ret;
}

static void sss_mc_header_update(struct sss_mc_ctx *mc_ctx, int status)
{
    struct sss_mc_header *h;

    /* update header using barriers */
    h = (struct sss_mc_header *)mc_ctx->mmap_base;
    MC_RAISE_BARRIER(h);
    if (status != SSS_MC_HEADER_RECYCLED) {
        /* no reason to update anything else if the file is recycled */
        h->hash_table = MC_PTR_DIFF(mc_ctx->hash_table, mc_ctx->mmap_base);
        h->free_table = MC_PTR_DIFF(mc_ctx->free_table, mc_ctx->mmap_base);
        h->data_table = MC_PTR_DIFF(mc_ctx->data_table, mc_ctx->mmap_base);
        h->ht_size = mc_ctx->ht_size;
        h->ft_size = mc_ctx->ft_size;
        h->dt_size = mc_ctx->dt_size;
        h->major_vno = SSS_MC_MAJOR_VNO;
        h->minor_vno = SSS_MC_MINOR_VNO;
        h->seed = mc_ctx->seed;
        h->reserved = 0;
    }
    h->status = status;
    MC_LOWER_BARRIER(h);
}

static int mc_ctx_destructor(struct sss_mc_ctx *mc_ctx)
{
    int ret;

    /* Print debug message to logs if munmap() or close()
     * fail but always return 0 */

    if (mc_ctx->mmap_base != NULL) {
        ret = munmap(mc_ctx->mmap_base, mc_ctx->mmap_size);
        if (ret == -1) {
            ret = errno;
            DEBUG(SSSDBG_CRIT_FAILURE,
                  ("Failed to unmap old memory cache file."
                   "[%d]: %s\n", ret, strerror(ret)));
        }
    }

    if (mc_ctx->fd != -1) {
        ret = close(mc_ctx->fd);
        if (ret == -1) {
            ret = errno;
            DEBUG(SSSDBG_CRIT_FAILURE,
                  ("Failed to close old memory cache file."
                   "[%d]: %s\n", ret, strerror(ret)));
        }
    }

    return 0;
}

errno_t sss_mmap_cache_init(TALLOC_CTX *mem_ctx, const char *name,
                            enum sss_mc_type type, size_t n_elem,
                            time_t timeout, struct sss_mc_ctx **mcc)
{
    struct sss_mc_ctx *mc_ctx = NULL;
    unsigned int rseed;
    int payload;
    int ret, dret;

    switch (type) {
    case SSS_MC_PASSWD:
        payload = SSS_AVG_PASSWD_PAYLOAD;
        break;
    case SSS_MC_GROUP:
        payload = SSS_AVG_GROUP_PAYLOAD;
        break;
    default:
        return EINVAL;
    }

    mc_ctx = talloc_zero(mem_ctx, struct sss_mc_ctx);
    if (!mc_ctx) {
        return ENOMEM;
    }
    mc_ctx->fd = -1;
    talloc_set_destructor(mc_ctx, mc_ctx_destructor);

    mc_ctx->name = talloc_strdup(mc_ctx, name);
    if (!mc_ctx->name) {
        ret = ENOMEM;
        goto done;
    }

    mc_ctx->type = type;

    mc_ctx->valid_time_slot = timeout;

    mc_ctx->file = talloc_asprintf(mc_ctx, "%s/%s",
                                   SSS_NSS_MCACHE_DIR, name);
    if (!mc_ctx->file) {
        ret = ENOMEM;
        goto done;
    }

    /* elements must always be multiple of 8 to make things easier to handle,
     * so we increase by the necessary amount if they are not a multiple */
    /* We can use MC_ALIGN64 for this */
    n_elem = MC_ALIGN64(n_elem);

    /* hash table is double the size because it will store both forward and
     * reverse keys (name/uid, name/gid, ..) */
    mc_ctx->ht_size = MC_HT_SIZE(n_elem * 2);
    mc_ctx->dt_size = MC_DT_SIZE(n_elem, payload);
    mc_ctx->ft_size = MC_FT_SIZE(n_elem);
    mc_ctx->mmap_size = MC_HEADER_SIZE +
                        MC_ALIGN64(mc_ctx->dt_size) +
                        MC_ALIGN64(mc_ctx->ft_size) +
                        MC_ALIGN64(mc_ctx->ht_size);


    /* for now ALWAYS create a new file on restart */

    ret = sss_mc_create_file(mc_ctx);
    if (ret) {
        goto done;
    }

    ret = ftruncate(mc_ctx->fd, mc_ctx->mmap_size);
    if (ret == -1) {
        ret = errno;
        DEBUG(SSSDBG_CRIT_FAILURE, ("Failed to resize file %s: %d(%s)\n",
                                    mc_ctx->file, ret, strerror(ret)));
        goto done;
    }

    mc_ctx->mmap_base = mmap(NULL, mc_ctx->mmap_size,
                             PROT_READ | PROT_WRITE,
                             MAP_SHARED, mc_ctx->fd, 0);
    if (mc_ctx->mmap_base == MAP_FAILED) {
        ret = errno;
        DEBUG(SSSDBG_CRIT_FAILURE, ("Failed to mmap file %s(%ld): %d(%s)\n",
                                    mc_ctx->file, mc_ctx->mmap_size,
                                    ret, strerror(ret)));
        goto done;
    }

    mc_ctx->data_table = MC_PTR_ADD(mc_ctx->mmap_base, MC_HEADER_SIZE);
    mc_ctx->free_table = MC_PTR_ADD(mc_ctx->data_table,
                                    MC_ALIGN64(mc_ctx->dt_size));
    mc_ctx->hash_table = MC_PTR_ADD(mc_ctx->free_table,
                                    MC_ALIGN64(mc_ctx->ft_size));

    memset(mc_ctx->data_table, 0xff, mc_ctx->dt_size);
    memset(mc_ctx->free_table, 0x00, mc_ctx->ft_size);
    memset(mc_ctx->hash_table, 0xff, mc_ctx->ht_size);

    /* generate a pseudo-random seed.
     * Needed to fend off dictionary based collision attacks */
    rseed = time(NULL) * getpid();
    mc_ctx->seed = rand_r(&rseed);

    sss_mc_header_update(mc_ctx, SSS_MC_HEADER_ALIVE);

    ret = EOK;

done:
    if (ret) {
        /* Closing the file descriptor and ummaping the file
         * from memory is done in the mc_ctx_destructor. */
        if (mc_ctx && mc_ctx->file && mc_ctx->fd != -1) {
            dret = unlink(mc_ctx->file);
            if (dret == -1) {
                DEBUG(SSSDBG_CRIT_FAILURE,
                      ("Failed to rm mmap file %s: %d(%s)\n", mc_ctx->file,
                       dret, strerror(dret)));
            }
        }

        talloc_free(mc_ctx);
    } else {
        *mcc = mc_ctx;
    }
    return ret;
}

errno_t sss_mmap_cache_reinit(TALLOC_CTX *mem_ctx, size_t n_elem,
                              time_t timeout, struct sss_mc_ctx **mc_ctx)
{
    errno_t ret;
    TALLOC_CTX* tmp_ctx = NULL;
    char *name;
    enum sss_mc_type type;

    if (mc_ctx == NULL || (*mc_ctx) == NULL) {
        DEBUG(SSSDBG_CRIT_FAILURE,
              ("Unable to re-init unitialized memory cache.\n"));
        return EINVAL;
    }

    tmp_ctx = talloc_new(NULL);
    if (tmp_ctx == NULL) {
        DEBUG(SSSDBG_CRIT_FAILURE, ("Out of memory.\n"));
        return ENOMEM;
    }

    name = talloc_strdup(tmp_ctx, (*mc_ctx)->name);
    if (name == NULL) {
        DEBUG(SSSDBG_CRIT_FAILURE, ("Out of memory.\n"));
        ret = ENOMEM;
        goto done;
    }

    type = (*mc_ctx)->type;

    if (n_elem == (size_t)-1) {
        n_elem = (*mc_ctx)->ft_size * 8;
    }

    if (timeout == (time_t)-1) {
        timeout = (*mc_ctx)->valid_time_slot;
    }

    talloc_free(*mc_ctx);

    /* make sure we do not leave a potentially freed pointer around */
    *mc_ctx = NULL;

    ret = sss_mmap_cache_init(mem_ctx, name, type, n_elem, timeout, mc_ctx);
    if (ret != EOK) {
        DEBUG(SSSDBG_CRIT_FAILURE, ("Failed to re-initialize mmap cache.\n"));
        goto done;
    }

done:
    talloc_free(tmp_ctx);
    return ret;
}