From 2c4024792a2e776b39c219df32a6b3b7e2df5786 Mon Sep 17 00:00:00 2001
From: Moinak Ghosh <moinakg@gmail.com>
Date: Mon, 22 Apr 2013 22:07:07 +0530
Subject: [PATCH] Several bugfixes. Avoid matching with self during hash
 lookup.

---
 rabin/global/index.c |  5 +++--
 rabin/rabin_dedup.c  | 12 ++++--------
 2 files changed, 7 insertions(+), 10 deletions(-)

diff --git a/rabin/global/index.c b/rabin/global/index.c
index 060f82a..9a96357 100644
--- a/rabin/global/index.c
+++ b/rabin/global/index.c
@@ -432,7 +432,7 @@ db_lookup_insert_s(archive_config_t *cfg, uchar_t *sim_cksum, int interval,
 	if (cfg->pct_interval == 0) { // Global dedupe with simple index
 		while (ent) {
 			if (mycmp(sim_cksum, ent->cksum, cfg->similarity_cksum_sz) == 0 &&
-			    ent->item_size == item_size) {
+			    ent->item_size == item_size && ent->item_offset != item_offset) {
 				return (ent);
 			}
 			pent = &(ent->next);
@@ -440,7 +440,8 @@ db_lookup_insert_s(archive_config_t *cfg, uchar_t *sim_cksum, int interval,
 		}
 	} else {
 		while (ent) {
-			if (mycmp(sim_cksum, ent->cksum, cfg->similarity_cksum_sz) == 0) {
+			if (mycmp(sim_cksum, ent->cksum, cfg->similarity_cksum_sz) == 0 &&
+			    ent->item_offset != item_offset) {
 				return (ent);
 			}
 			pent = &(ent->next);
diff --git a/rabin/rabin_dedup.c b/rabin/rabin_dedup.c
index 401a5b3..2818616 100755
--- a/rabin/rabin_dedup.c
+++ b/rabin/rabin_dedup.c
@@ -858,14 +858,14 @@ process_blocks:
 					qsort(seg_heap, length/8, 8, cmpint);
 
 					/*
-					 * Compute the range similarity minhashes.
+					 * Compute the range similarity hashes.
 					 */
 					sim_ck = ctx->similarity_cksums;
 					crc = 0;
 					sub_i = cfg->sub_intervals;
 					increment = (length / cfg->intervals) / sub_i;
 					tgt = seg_heap;
-					while (increment < cfg->chunk_cksum_sz/4 && sub_i > 0) {
+					while (increment < cfg->chunk_cksum_sz/4 && sub_i > 1) {
 						sub_i--;
 						increment = (length / cfg->intervals) / sub_i;
 					}
@@ -880,7 +880,7 @@ process_blocks:
 
 					increment = length / cfg->intervals;
 					for (j=0; j<cfg->intervals-1; j++) {
-						crc = lzma_crc64(tgt, increment/2, 0);
+						crc = lzma_crc64(tgt, increment/8, 0);
 						*((uint64_t *)sim_ck) = crc;
 						tgt += increment;
 						len -= increment;
@@ -958,11 +958,7 @@ process_blocks:
 						hash_entry_t *he;
 
 						he = db_lookup_insert_s(cfg, sim_ck, 0, seg_offset, 0, 1);
-
-						/*
-						 * If match found also check that match is not with self!
-						 */
-						if (he && he->item_offset != seg_offset) {
+						if (he) {
 							/*
 							 * Match found. Load segment metadata from disk and perform
 							 * identity deduplication with the segment chunks.