-
Notifications
You must be signed in to change notification settings - Fork 1
/
logstor.c
2075 lines (1825 loc) · 49.5 KB
/
logstor.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
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
/*
Author: Wuyang Chung
e-mail: wuyang.chung1@gmail.com
*/
#include <stdio.h>
#include <stdlib.h>
#include <stddef.h>
#include <string.h>
#include <stdbool.h>
#include <unistd.h>
#include <fcntl.h>
#include <errno.h>
//#include <assert.h>
#include <time.h>
#include <sys/queue.h>
#include <sys/param.h>
#include <sys/stat.h>
#include <sys/ioctl.h>
#if __linux
#include <linux/fs.h>
#include <sys/ioctl.h>
#endif
#if __BSD_VISIBLE
#include <sys/disk.h>
#include "ggate.h"
#endif
#include "logstor.h"
#if defined(MY_DEBUG)
#if defined(FBUF_DEBUG)
static void fbuf_mod_dump(void);
#endif
void my_break(void) {}
void my_debug(const char * fname, int line_num, bool bl_panic)
{
const char *type[] = {"break", "panic"};
printf("*** %s *** %s %d\n", type[bl_panic], fname, line_num);
perror("");
#if defined(FBUF_DEBUG)
fbuf_mod_dump();
#endif
my_break();
if (bl_panic)
#if defined(EXIT_ON_PANIC)
exit(1);
#else
;
#endif
}
#endif
#define MAX_FBUF_COUNT 4096
//#define MAX_FBUF_COUNT 500 //wyctest
#define CLEAN_AGE_LIMIT 3
//#define CLEAN_WINDOW 4
#define SIG_LOGSTOR 0x4C4F4753 // "LOGS": Log-Structured Storage
#define VER_MAJOR 0
#define VER_MINOR 1
#define SEG_DATA_START 1 // the data segment starts here
#define SEG_SUM_OFFSET (SECTORS_PER_SEG - 1) // segment summary offset in segment
#define SEG_SIZE 0x400000 // 4M
#define SECTORS_PER_SEG (SEG_SIZE/SECTOR_SIZE) // 1024
#define SA2SEGA_SHIFT 10
#define BLOCKS_PER_SEG (SEG_SIZE/SECTOR_SIZE - 1)
#define META_BASE 0xFF000000u // metadata block start address
#define META_INVALID 0 // invalid metadata address
#define SECTOR_NULL 0 // this sector address can not map to any block address
#define SECTOR_DELETE 1 // delete marker for a block
#define IS_META_ADDR(x) (((x) & META_BASE) == META_BASE)
#define META_LEAF_DEPTH 2
#define RAM_DISK_SIZE 0x70000000 // 1.75G the maximum size for i386 FreeBSD 12
#define FILE_BUCKET_COUNT 4099
/*
The file descriptor for the forward map files
*/
enum {
FD_BASE, // file descriptor for base map
FD_ACTIVE, // file descriptor for active map
FD_DELTA, // file descriptor for delta map
FD_COUNT // the number of file descriptors
};
struct _superblock {
uint32_t sig; // signature
uint8_t ver_major;
uint8_t ver_minor;
uint16_t sb_gen; // the generation number. Used for redo after system crash
uint32_t max_block_cnt; // max number of blocks supported
/*
The segments are treated as circular buffer
*/
int32_t seg_cnt; // total number of segments
int32_t seg_free_cnt; // number of free segments
int32_t sega_alloc; // allocate this segment
int32_t sega_reclaim; // clean this segment
/*
The files for forward mapping file
Mapping is always updated in @FD_ACTIVE. When snapshot command is issued
@FD_ACTIVE is copied to @FD_DELTA and then cleaned.
Backup program then backs up the delta by reading @FD_DELTA.
After backup is finished, @FD_DELTA is merged into @FD_BASE and then cleaned.
If reduced to reboot restore usage, only @FD_ACTIVE and @FD_BASE are needed.
Each time a PC is rebooted @FD_ACTIVE is cleaned so all data are restored.
So the actual mapping is @FD_ACTIVE || @FD_DELTA || @FD_BASE.
The first mapping that is not empty is used.
*/
uint32_t ftab[FD_COUNT]; // the file table
uint8_t sb_seg_age[]; // the starting address to store seg_age in superblock
};
#if !defined(WYC)
_Static_assert(sizeof(struct _superblock) < SECTOR_SIZE, "The size of the super block must be smaller than SECTOR_SIZE");
#endif
/*
The last sector in a segment is the segment summary. It stores the reverse mapping table
*/
struct _seg_sum {
uint32_t ss_rm[SECTORS_PER_SEG - 1]; // reverse map
// reverse map SECTORS_PER_SEG - 1 is not used so we store something here
uint16_t ss_gen; // sequence number. used for redo after system crash
uint16_t ss_alloc_p; // allocate sector at this location
// below are not stored on disk
uint32_t sega; // the segment address of the segment summary
unsigned live_count;
//LIST_ENTRY(_seg_sum) queue;
};
_Static_assert(offsetof(struct _seg_sum, sega) == SECTOR_SIZE,
"The size of segment summary must be equal to SECTOR_SIZE");
/*
File data and its indirect blocks are also stored in the downstream disk.
The sectors used to store the file data and its indirect blocks are called metadata.
Each metadata block has a corresponding metadata address.
Below is the format of the metadata address.
The metadata address occupies a small part of buffer address space. For buffer address
that is >= META_BASE, it is actually a metadata address.
*/
union meta_addr { // metadata address for file data and its indirect blocks
uint32_t uint32;
struct {
uint32_t index :20; // index for indirect block
uint32_t depth :2; // depth of the node
uint32_t fd :2; // file descriptor
uint32_t meta :8; // 0xFF for metadata address
};
};
_Static_assert(sizeof(union meta_addr) == 4, "The size of emta_addr must be 4");
/*
Metadata is cached in memory. The access unit of metadata is block so each cache line
stores a block of metadata
*/
struct _fbuf { // file buffer
union {
LIST_ENTRY(_fbuf) indir_queue; // for the indirect queue
struct {
struct _fbuf *next;
struct _fbuf *prev;
} cir_queue; // list entry for the circular queue
};
uint16_t ref_cnt; // only used for fbufs on indirect queue
bool on_cir_queue; // on circular queue
bool accessed; // only used for fbufs on circular queue
bool modified; // the fbuf is dirty
LIST_ENTRY(_fbuf) buffer_bucket_queue;// the pointer for bucket chain
struct _fbuf *parent;
union meta_addr ma; // the metadata address
#if defined(MY_DEBUG)
uint32_t sa; // the sector address of the @data
#endif
#if defined(FBUF_DEBUG)
uint16_t index;
struct _fbuf *child[SECTOR_SIZE/sizeof(uint32_t)];
#endif
// the metadata is cached here
uint32_t data[SECTOR_SIZE/sizeof(uint32_t)];
};
/*
logstor soft control
*/
struct g_logstor_softc {
struct _seg_sum seg_sum_cold;// segment summary for the cold segment
struct _seg_sum seg_sum_hot;// segment summary for the hot segment
//struct _seg_sum clean_candidate[CLEAN_WINDOW];
//LIST_HEAD(, _seg_sum) cc_head; // clean candidate
unsigned char cleaner_disabled;
uint32_t clean_low_water;
uint32_t clean_high_water;
int fbuf_count;
int fbuf_modified_count; // This field is for debug use.
struct _fbuf *fbuf;
#if 0
uint32_t *fbuf_accessed;
uint32_t *fbuf_modified;
uint32_t *fbuf_on_cir_queue;
#endif
// buffer hash queue
LIST_HEAD(_fbuf_bucket, _fbuf) fbuf_bucket[FILE_BUCKET_COUNT];
struct _fbuf *fbuf_cir_head; // head of the circular queue
LIST_HEAD(, _fbuf) fbuf_ind_head[META_LEAF_DEPTH]; // indirect queue
#if defined(MY_DEBUG)
int cir_queue_cnt;
#endif
// statistics
unsigned data_write_count; // data block write to disk
unsigned other_write_count; // other write to disk, such as metadata write and segment cleaning
unsigned fbuf_hit;
unsigned fbuf_miss;
bool sb_modified; // the super block is dirty
uint32_t sb_sa; // superblock's sector address
/*
The macro RAM_DISK is used for debug.
By using RAM as the storage device, the test can run way much faster.
*/
#if defined(RAM_DISK)
uint8_t seg_age[448];
#else
int disk_fd;
uint8_t *seg_age;
#endif
struct _superblock superblock;
};
#if defined(MY_DEBUG)
uint32_t sa_rw; // the sector address for _logstor_read_one/_logstor_write_one
#endif
uint32_t gdb_cond0;
uint32_t gdb_cond1;
#if defined(RAM_DISK)
static char *ram_disk;
#endif
static struct g_logstor_softc sc;
static int _logstor_read(unsigned ba, char *data, int size);
static int _logstor_read_one(unsigned ba, char *data);
static int _logstor_write(uint32_t ba, char *data, int size, struct _seg_sum *seg_sum);
static int _logstor_write_one(uint32_t ba, char *data, struct _seg_sum *seg_sum, uint32_t *sa_out);
static void seg_alloc(struct _seg_sum *seg_sum);
static void seg_reclaim_init(struct _seg_sum *seg_sum);
static void seg_sum_write(struct _seg_sum *seg_sum);
static void seg_clean(struct _seg_sum *seg_sum);
static void seg_live_count(struct _seg_sum *seg_sum);
static void clean_check(void);
static void clean_metadata(uint32_t cur_sa, union meta_addr cur_ma);
static void clean_data(uint32_t cur_sa, uint32_t cur_ba);
static int superblock_init_read(void);
static void superblock_init_write(int fd);
static void superblock_write(void);
static uint32_t file_read_4byte(uint8_t fh, uint32_t ba);
static void file_write_4byte(uint8_t fh, uint32_t ba, uint32_t sa);
static struct _fbuf *file_access(uint8_t fd, uint32_t offset, uint32_t *buf_off, bool bl_write);
static void fbuf_mod_init(void);
static void fbuf_mod_fini(void);
static void fbuf_mod_flush(void);
static struct _fbuf *fbuf_get(union meta_addr ma);
static struct _fbuf *fbuf_read_and_hash(uint32_t sa, union meta_addr ma);
static struct _fbuf *fbuf_search(union meta_addr ma);
static bool fbuf_flush(struct _fbuf *buf);
static void fbuf_hash_insert(struct _fbuf *buf, unsigned key);
static union meta_addr ma2pma(union meta_addr ma, unsigned *pindex_out);
static uint32_t ma2sa(union meta_addr ma);
static void my_read (uint32_t sa, void *buf, unsigned size);
static void my_write(uint32_t sa, const void *buf, unsigned size);
#if __BSD_VISIBLE
static off_t
get_mediasize(int fd)
{
off_t mediasize;
struct stat sb;
if (fstat(fd, &sb) == -1) {
printf("fstat(): %s.", strerror(errno));
exit(1);
}
if (S_ISCHR(sb.st_mode)) {
if (ioctl(fd, DIOCGMEDIASIZE, &mediasize) == -1) {
printf("Can't get media size: %s.", strerror(errno));
exit(1);
}
} else if (S_ISREG(sb.st_mode)) {
mediasize = sb.st_size;
} else {
printf("Unsupported file system object.");
exit(1);
}
return (mediasize);
}
#else
static off_t
get_mediasize(int fd)
{
off_t mediasize;
struct stat sb;
int rc;
rc = fstat(fd, &sb);
MY_ASSERT(rc != -1);
if (S_ISCHR(sb.st_mode))
MY_ASSERT(ioctl(fd, BLKGETSIZE64, &mediasize) != -1);
else if (S_ISREG(sb.st_mode))
mediasize = sb.st_size;
else
MY_PANIC(); // Unsupported file system object
return (mediasize);
}
#endif
/*******************************
* logstor *
*******************************/
/*
Description:
segment address to sector address
*/
static uint32_t
sega2sa(uint32_t sega)
{
return sega << SA2SEGA_SHIFT;
}
void logstor_superblock_init(const char *disk_file)
{
int disk_fd;
disk_fd = open(disk_file, O_WRONLY);
MY_ASSERT(disk_fd > 0);
superblock_init_write(disk_fd);
}
void logstor_init(void)
{
#if defined(RAM_DISK)
ram_disk = malloc(RAM_DISK_SIZE);
MY_ASSERT(ram_disk != NULL);
superblock_init_write(-1);
#endif
}
void
logstor_fini(void)
{
#if defined(RAM_DISK)
free(ram_disk);
#endif
}
int
logstor_open(const char *disk_file)
{
int error;
bzero(&sc, sizeof(sc));
#if !defined(RAM_DISK)
#if __BSD_VISIBLE
sc.disk_fd = open(disk_file, O_RDWR | O_DIRECT | O_FSYNC);
#else
sc.disk_fd = open(disk_file, O_RDWR);
#endif
MY_ASSERT(sc.disk_fd > 0);
#endif
error = superblock_init_read();
MY_ASSERT(error == 0);
// the order of the two statements below is important
seg_alloc(&sc.seg_sum_cold);
seg_alloc(&sc.seg_sum_hot);
sc.data_write_count = sc.other_write_count = 0;
sc.clean_low_water = 8;
sc.clean_high_water = sc.clean_low_water + 4;
fbuf_mod_init();
//clean_check();
return 0;
}
void
logstor_close(void)
{
fbuf_mod_fini();
seg_sum_write(&sc.seg_sum_cold);
seg_sum_write(&sc.seg_sum_hot);
superblock_write();
#if !defined(RAM_DISK)
free(sc.seg_age);
close(sc.disk_fd);
#endif
}
/*
Description:
Read blocks from logstor
Parameters:
@offset: disk offset
@data: data buffer
@length: data length
*/
int
logstor_read(off_t offset, void *data, off_t length)
{
unsigned size;
uint32_t ba;
int error;
MY_ASSERT((offset & (SECTOR_SIZE - 1)) == 0);
MY_ASSERT((length & (SECTOR_SIZE - 1)) == 0);
ba = offset / SECTOR_SIZE;
size = length / SECTOR_SIZE;
if (size == 1) {
error = _logstor_read_one(ba, data);
} else {
error = _logstor_read(ba, data, size);
}
return error;
}
/*
Description:
Write blocks to logstor
Parameters:
@offset: disk offset
@data: data buffer
@length: data length
*/
int
logstor_write(off_t offset, void *data, off_t length)
{
uint32_t ba; // block address
int size; // number of remaining sectors to process
int error;
MY_ASSERT((offset & (SECTOR_SIZE - 1)) == 0);
MY_ASSERT((length & (SECTOR_SIZE - 1)) == 0);
ba = offset / SECTOR_SIZE;
size = length / SECTOR_SIZE;
if (size == 1) {
error = _logstor_write_one(ba, data, &sc.seg_sum_hot, NULL);
} else {
error = _logstor_write(ba, data, size, &sc.seg_sum_hot);
}
return error;
}
// To enable TRIM, the following statement must be added
// in "case BIO_GETATTR" of g_gate_start() of g_gate.c
// if (g_handleattr_int(pbp, "GEOM::candelete", 1))
// return;
// and the command below must be executed before mounting the device
// tunefs -t enabled /dev/ggate0
int logstor_delete(off_t offset, void *data, off_t length)
{
uint32_t ba; // block address
int size; // number of remaining sectors to process
int i;
MY_ASSERT((offset & (SECTOR_SIZE - 1)) == 0);
MY_ASSERT((length & (SECTOR_SIZE - 1)) == 0);
ba = offset / SECTOR_SIZE;
size = length / SECTOR_SIZE;
MY_ASSERT(ba < sc.superblock.max_block_cnt);
if (size == 1) {
file_write_4byte(FD_ACTIVE, ba, SECTOR_DELETE);
} else {
for (i = 0; i<size; i++)
file_write_4byte(FD_ACTIVE, ba + i, SECTOR_DELETE);
}
// file_read_4byte/file_write_4byte might trigger fbuf_write and
// clean check cannot be done in fbuf_write
// so need to do a clean check here
clean_check();
return (0);
}
int
logstor_read_test(uint32_t ba, void *data)
{
//wyctest return _logstor_read(ba, data, 1);
return _logstor_read_one(ba, data);
}
int
logstor_write_test(uint32_t ba, void *data)
{
//wyctest return _logstor_write(ba, data, 1, &sc.seg_sum_hot);
return _logstor_write_one(ba, data, &sc.seg_sum_hot, NULL);
}
/*
Description:
Read blocks from the logstor
Parameters:
@ba: block address
@data: data buffer
@size: size in unit of block
*/
static int
_logstor_read(unsigned ba, char *data, int size)
{
unsigned i, count;
uint32_t start_sa, pre_sa, sa; // sector address
MY_ASSERT(ba < sc.superblock.max_block_cnt);
start_sa = pre_sa = file_read_4byte(FD_ACTIVE, ba);
count = 1;
for (i = 1; i < size; i++) {
sa = file_read_4byte(FD_ACTIVE, ba + i);
if (sa == pre_sa + 1) {
count++;
pre_sa = sa;
} else {
if (start_sa == SECTOR_NULL || start_sa == SECTOR_DELETE)
bzero(data, SECTOR_SIZE);
else {
my_read(start_sa, data, count);
}
// set the values for the next write
data += count * SECTOR_SIZE;
start_sa = pre_sa = sa;
count = 1;
}
}
if (start_sa == SECTOR_NULL || start_sa == SECTOR_DELETE)
bzero(data, SECTOR_SIZE);
else {
my_read(start_sa, data, count);
}
// file_read_4byte/file_write_4byte might trigger fbuf_write and
// clean check cannot be done in fbuf_write
// so need to do a clean check here
clean_check();
return 0;
}
static int
_logstor_read_one(unsigned ba, char *data)
{
uint32_t start_sa; // sector address
MY_ASSERT(ba < sc.superblock.max_block_cnt);
start_sa = file_read_4byte(FD_ACTIVE, ba);
#if defined(MY_DEBUG)
sa_rw = start_sa; //wyctest
#endif
if (start_sa == SECTOR_NULL || start_sa == SECTOR_DELETE)
bzero(data, SECTOR_SIZE);
else {
my_read(start_sa, data, 1);
}
// file_read_4byte/file_write_4byte might trigger fbuf_write and
// clean check cannot be done in fbuf_write
// so need to do a clean check here
clean_check();
return 0;
}
/*
Description:
Write blocks to logstor
Parameters:
@ba: block address
@data: data buffer
@size: size in unit of block
*/
static int
_logstor_write(uint32_t ba, char *data, int size, struct _seg_sum *seg_sum)
{
uint32_t sa; // sector address
int sec_remain; // number of remaining sectors to process
int sec_free; // number of free sectors in current segment
int i, count;
MY_ASSERT(ba < sc.superblock.max_block_cnt);
MY_ASSERT(seg_sum->ss_alloc_p < SEG_SUM_OFFSET);
sec_remain = size;
while (sec_remain > 0) {
sec_free = SEG_SUM_OFFSET - seg_sum->ss_alloc_p;
count = sec_remain <= sec_free? sec_remain: sec_free; // min(sec_remain, sec_free)
sa = sega2sa(seg_sum->sega) + seg_sum->ss_alloc_p;
MY_ASSERT(sa + count < sc.superblock.seg_cnt * SECTORS_PER_SEG);
my_write(sa, data, count);
data += count * SECTOR_SIZE;
if (sc.cleaner_disabled) // if doing segment cleaning
sc.other_write_count += count;
else
sc.data_write_count += count;
// record the reverse mapping immediately after the data have been written
for (i = 0; i < count; i++)
seg_sum->ss_rm[seg_sum->ss_alloc_p++] = ba + i;
if (seg_sum->ss_alloc_p == SEG_SUM_OFFSET)
{ // current segment is full
seg_sum_write(seg_sum);
seg_alloc(seg_sum);
clean_check();
}
// record the forward mapping
// the forward mapping must be recorded after
// the segment summary block write
for (i = 0; i < count; i++)
file_write_4byte(FD_ACTIVE, ba++, sa++);
sec_remain -= count;
}
return 0;
}
/*
Note:
sa_out is not NULL only when called from clean_metadata
*/
static int
_logstor_write_one(uint32_t ba, char *data, struct _seg_sum *seg_sum, uint32_t *sa_out)
{
uint32_t sa; // sector address
MY_ASSERT(ba < sc.superblock.max_block_cnt);
MY_ASSERT(seg_sum->ss_alloc_p < SEG_SUM_OFFSET);
sa = sega2sa(seg_sum->sega) + seg_sum->ss_alloc_p;
#if defined(MY_DEBUG)
sa_rw = sa; //wyctest
#endif
MY_ASSERT(sa < sc.superblock.seg_cnt * SECTORS_PER_SEG);
my_write(sa, data, 1);
if (sc.cleaner_disabled) // if doing segment cleaning
sc.other_write_count++;
else
sc.data_write_count ++;
// record the reverse mapping
seg_sum->ss_rm[seg_sum->ss_alloc_p++] = ba;
if (seg_sum->ss_alloc_p == SEG_SUM_OFFSET)
{ // current segment is full
seg_sum_write(seg_sum);
seg_alloc(seg_sum);
clean_check();
}
if (IS_META_ADDR(ba)) {
// the forwarding mapping is returned to the caller
MY_ASSERT(sa_out != NULL);
*sa_out = sa;
} else {
// record the forward mapping
// the forward mapping must be recorded after
// the segment summary block write
MY_ASSERT(sa_out == NULL);
file_write_4byte(FD_ACTIVE, ba, sa);
}
return 0;
}
uint32_t
logstor_get_block_cnt(void)
{
return sc.superblock.max_block_cnt;
}
unsigned
logstor_get_data_write_count(void)
{
return sc.data_write_count;
}
unsigned
logstor_get_other_write_count(void)
{
return sc.other_write_count;
}
unsigned
logstor_get_fbuf_hit(void)
{
return sc.fbuf_hit;
}
unsigned
logstor_get_fbuf_miss(void)
{
return sc.fbuf_miss;
}
static void
seg_sum_read(struct _seg_sum *seg_sum, uint32_t sega)
{
uint32_t sa;
seg_sum->sega = sega;
sa = sega2sa(sega) + SEG_SUM_OFFSET;
my_read(sa, seg_sum, 1);
}
/*
write out the segment summary
*/
static void
seg_sum_write(struct _seg_sum *seg_sum)
{
uint32_t sa;
// segment summary is at the end of a segment
sa = sega2sa(seg_sum->sega) + SEG_SUM_OFFSET;
seg_sum->ss_gen = sc.superblock.sb_gen;
my_write(sa, (void *)seg_sum, 1);
sc.other_write_count++; // the write for the segment summary
}
/*
Segment 0 is used to store superblock so there are SECTORS_PER_SEG sectors
for storing superblock. Each time the superblock is synced, it is stored
in the next sector. When it reachs the end of segment 0, it wraps around
to sector 0.
*/
static int
superblock_init_read(void)
{
int i;
uint16_t sb_gen;
struct _superblock *sb_in;
char buf[2][SECTOR_SIZE];
_Static_assert(sizeof(sb_gen) == sizeof(sc.superblock.sb_gen), "sb_gen");
// get the superblock
sb_in = (struct _superblock *)buf[0];
#if defined(RAM_DISK)
memcpy(sb_in, ram_disk, SECTOR_SIZE);
#else
MY_ASSERT(pread(sc.disk_fd, sb_in, SECTOR_SIZE, 0) == SECTOR_SIZE);
#endif
if (sb_in->sig != SIG_LOGSTOR ||
sb_in->sega_alloc >= sb_in->seg_cnt ||
sb_in->sega_reclaim >= sb_in->seg_cnt)
return EINVAL;
sb_gen = sb_in->sb_gen;
for (i = 1 ; i < SECTORS_PER_SEG; i++) {
sb_in = (struct _superblock *)buf[i%2];
#if defined(RAM_DISK)
memcpy(sb_in, ram_disk + i * SECTOR_SIZE, SECTOR_SIZE);
#else
MY_ASSERT(pread(sc.disk_fd, sb_in, SECTOR_SIZE, i * SECTOR_SIZE) == SECTOR_SIZE);
#endif
if (sb_in->sig != SIG_LOGSTOR)
break;
if (sb_in->sb_gen != (uint16_t)(sb_gen + 1)) // IMPORTANT type cast
break;
sb_gen = sb_in->sb_gen;
}
sc.sb_sa = (i - 1);
sb_in = (struct _superblock *)buf[(i-1)%2];
if (sb_in->sega_alloc >= sb_in->seg_cnt ||
sb_in->sega_reclaim >= sb_in->seg_cnt)
return EINVAL;
#if defined(RAM_DISK)
MY_ASSERT(sizeof(sc.seg_age) >= sc.superblock.seg_cnt);
#else
sc.seg_age = malloc(sc.superblock.seg_cnt);
MY_ASSERT(sc.seg_age != NULL);
#endif
memcpy(sc.seg_age, sb_in->sb_seg_age, sb_in->seg_cnt);
memcpy(&sc.superblock, sb_in, sizeof(sc.superblock));
sc.sb_modified = false;
return 0;
}
/*
Description:
Write the initialized supeblock to the downstream disk
*/
static void
superblock_init_write(int fd)
{
int i;
uint32_t sector_cnt;
struct _superblock *sb_out;
off_t media_size;
char buf[SECTOR_SIZE] __attribute__ ((aligned));
#if defined(RAM_DISK)
media_size = RAM_DISK_SIZE;
#else
media_size = get_mediasize(fd);
#endif
sector_cnt = media_size / SECTOR_SIZE;
sb_out = (struct _superblock *)buf;
sb_out->sig = SIG_LOGSTOR;
sb_out->ver_major = VER_MAJOR;
sb_out->ver_minor = VER_MINOR;
#if __BSD_VISIBLE
sb_out->sb_gen = arc4random();
#else
sb_out->sb_gen = random();
#endif
sb_out->seg_cnt = sector_cnt / SECTORS_PER_SEG;
if (sizeof(struct _superblock) + sb_out->seg_cnt > SECTOR_SIZE) {
printf("%s: size of superblock %d seg_cnt %d\n",
__func__, (int)sizeof(struct _superblock), (int)sb_out->seg_cnt);
printf(" the size of the disk must be less than %lld\n",
(SECTOR_SIZE - sizeof(struct _superblock)) * (long long)SEG_SIZE);
MY_PANIC();
}
sb_out->seg_free_cnt = sb_out->seg_cnt - SEG_DATA_START;
// the physical disk must have at least the space for the metadata
MY_ASSERT(sb_out->seg_free_cnt * BLOCKS_PER_SEG >
(sector_cnt / (SECTOR_SIZE / 4)) * FD_COUNT);
sb_out->max_block_cnt =
sb_out->seg_free_cnt * BLOCKS_PER_SEG -
(sector_cnt / (SECTOR_SIZE / 4)) * FD_COUNT;
sb_out->max_block_cnt *= 0.9;
#if defined(MY_DEBUG)
printf("%s: sector_cnt %u max_block_cnt %u\n",
__func__, sector_cnt, sb_out->max_block_cnt);
#endif
// the root sector address for the files
for (i = 0; i < FD_COUNT; i++) {
sb_out->ftab[i] = SECTOR_NULL; // SECTOR_NULL means not allocated yet
}
sb_out->sega_alloc = SEG_DATA_START; // start allocate from here
sb_out->sega_reclaim = SEG_DATA_START + 1; // start reclaim from here
bzero(sb_out->sb_seg_age, SECTOR_SIZE - sizeof(struct _superblock));
// write out super block
#if defined(RAM_DISK)
memcpy(ram_disk, sb_out, SECTOR_SIZE);
#else
MY_ASSERT(pwrite(fd, sb_out, SECTOR_SIZE, 0) == SECTOR_SIZE);
#endif
// clear the rest of the supeblock's segment
bzero(buf, SECTOR_SIZE);
for ( i = 1; i < SECTORS_PER_SEG; i++) {
#if defined(RAM_DISK)
memcpy(ram_disk + i * SECTOR_SIZE, buf, SECTOR_SIZE);
#else
MY_ASSERT(pwrite(fd, buf, SECTOR_SIZE, i * SECTOR_SIZE) == SECTOR_SIZE);
#endif
}
}
static void
superblock_write(void)
{
struct _superblock *sb_out;
char buf[SECTOR_SIZE];
sc.superblock.sb_gen++;
if (++sc.sb_sa == SECTORS_PER_SEG)
sc.sb_sa = 0;
sb_out = (struct _superblock *)buf;
memcpy(sb_out, &sc.superblock, sizeof(sc.superblock));
memcpy(sb_out->sb_seg_age, sc.seg_age, sb_out->seg_cnt);
my_write(sc.sb_sa, sb_out, 1);
sc.other_write_count++;
}
#if defined(MY_DEBUG)
uint32_t sa_watch = 0;
#endif
#if defined(RAM_DISK)
static void
my_read(uint32_t sa, void *buf, unsigned size)
{
MY_ASSERT(sa < sc.superblock.seg_cnt * SECTORS_PER_SEG);
memcpy(buf, ram_disk + (off_t)sa * SECTOR_SIZE, size * SECTOR_SIZE);
//MY_BREAK(sa == sa_watch);
}
static void
my_write(uint32_t sa, const void *buf, unsigned size)
{
MY_ASSERT(sa < sc.superblock.seg_cnt * SECTORS_PER_SEG);
memcpy(ram_disk + (off_t)sa * SECTOR_SIZE , buf, size * SECTOR_SIZE);
//MY_BREAK(sa==sa_watch);
}
#else
static void
my_read(uint32_t sa, void *buf, unsigned size)
{
ssize_t bc; // byte count
MY_ASSERT(sa < sc.superblock.seg_cnt * SECTORS_PER_SEG);
bc = pread(sc.disk_fd, buf, size * SECTOR_SIZE, (off_t)sa * SECTOR_SIZE);
MY_ASSERT(bc == size * SECTOR_SIZE);
}
static void
my_write(uint32_t sa, const void *buf, unsigned size)
{
ssize_t bc; // byte count
MY_ASSERT(sa < sc.superblock.seg_cnt * SECTORS_PER_SEG);
bc = pwrite(sc.disk_fd, buf, size * SECTOR_SIZE, (off_t)sa * SECTOR_SIZE);
MY_ASSERT(bc == size * SECTOR_SIZE);
}
#endif
/*
Description:
Allocate a segment for writing
Output:
Store the segment address into @seg_sum->sega
Initialize @seg_sum->sum.alloc_p to 0
*/
static void
seg_alloc(struct _seg_sum *seg_sum)
{
uint32_t sega;
#if defined(MY_DEBUG)
uint32_t sega_cold = sc.seg_sum_cold.sega;
uint32_t sega_hot = sc.seg_sum_hot.sega;
#endif
again:
sega = sc.superblock.sega_alloc;
if (++sc.superblock.sega_alloc == sc.superblock.seg_cnt)
sc.superblock.sega_alloc = SEG_DATA_START;
MY_ASSERT(sc.superblock.sega_alloc < sc.superblock.seg_cnt);
MY_ASSERT(sc.superblock.sega_alloc + 1 != sc.superblock.sega_reclaim);
MY_ASSERT(sega != sega_hot);
if (sc.seg_age[sega] != 0) // this segment is not free
goto again;
MY_ASSERT(sega != sega_cold);
sc.seg_age[sega] = 1;
seg_sum->sega = sega;
seg_sum->ss_alloc_p = 0;
sc.superblock.seg_free_cnt--;
MY_ASSERT(sc.superblock.seg_free_cnt > 0 &&
sc.superblock.seg_free_cnt < sc.superblock.seg_cnt);
}