greg/pcompress
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

Improve duplicate segment match detection.

Browse source
This commit is contained in:
Moinak Ghosh 2013-04-23 20:51:12 +05:30
parent 6b7d883393
commit b32f4b3f9a
1 changed files with 28 additions and 7 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

33
rabin/rabin_dedup.c
View file

@ -838,7 +838,7 @@ process_blocks:
ary_sz = cfg->segment_sz * sizeof (global_blockentry_t **); ary_sz = cfg->segment_sz * sizeof (global_blockentry_t **);
htab = (global_blockentry_t **)(seg_heap - ary_sz); htab = (global_blockentry_t **)(seg_heap - ary_sz);
for (i=0; i<blknum;) { for (i=0; i<blknum;) {
uint64_t crc; uint64_t crc, off1;
length = 0; length = 0;
/* /*
@ -955,20 +955,41 @@ process_blocks:
} }
/* /*
* Now lookup the similarity hashes starting at the highest * Now lookup all the similarity hashes. We sort the hashes first so that
* significance level. * all duplicates can be easily detected.
*/ */
qsort(ctx->similarity_cksums, cfg->intervals + sub_i - 1, 8, cmpint);
crc = 0;
off1 = UINT64_MAX;
for (j=cfg->intervals + sub_i; j > 0; j--) { for (j=cfg->intervals + sub_i; j > 0; j--) {
hash_entry_t *he = NULL, *he1 = NULL; hash_entry_t *he = NULL;
/*
* Check for duplicate checksum which need not be looked up
* again.
*/
if (crc == *((uint64_t *)sim_ck)) {
he = NULL;
} else {
he = db_lookup_insert_s(cfg, sim_ck, 0, seg_offset, 0, 1); he = db_lookup_insert_s(cfg, sim_ck, 0, seg_offset, 0, 1);
if (he && he != he1) { /*
* Check for different checksum but same segment match.
* This is not a complete check but does help to reduce
* wasted processing.
*/
if (he && off1 == he->item_offset) {
crc = *((uint64_t *)sim_ck);
he = NULL;
}
}
if (he) {
/* /*
* Match found. Load segment metadata from disk and perform * Match found. Load segment metadata from disk and perform
* identity deduplication with the segment chunks. * identity deduplication with the segment chunks.
*/ */
he1 = he; crc = *((uint64_t *)sim_ck);
offset = he->item_offset; offset = he->item_offset;
off1 = offset;
if (db_segcache_map(cfg, ctx->id, &o_blks, &offset, if (db_segcache_map(cfg, ctx->id, &o_blks, &offset,
(uchar_t **)&seg_blocks) == -1) { (uchar_t **)&seg_blocks) == -1) {
fprintf(stderr, "Segment cache mmap failed.\n"); fprintf(stderr, "Segment cache mmap failed.\n");