From 6ecc4005714534dbefb0ddf23b2f3c0f12715149 Mon Sep 17 00:00:00 2001
From: Moinak Ghosh <moinakg@gmail.com>
Date: Sun, 5 May 2013 18:50:52 +0530
Subject: [PATCH] Fix segment offset sorting. Get rid of incorrect duplicate
 checks in index.

---
 rabin/global/index.c | 11 +++++------
 rabin/rabin_dedup.c  |  2 +-
 2 files changed, 6 insertions(+), 7 deletions(-)

diff --git a/rabin/global/index.c b/rabin/global/index.c
index 4254c6a..d9ca942 100644
--- a/rabin/global/index.c
+++ b/rabin/global/index.c
@@ -485,10 +485,11 @@ db_lookup_insert_s(archive_config_t *cfg, uchar_t *sim_cksum, int interval,
 
 	pent = &(htab[htab_entry]);
 	ent = htab[htab_entry];
-	if (cfg->pct_interval == 0) { // Global dedupe with simple index
+	if (cfg->pct_interval == 0) { // Global dedupe with simple index.
+		assert(cfg->similarity_cksum_sz == cfg->chunk_cksum_sz);
 		while (ent) {
 			if (mycmp(sim_cksum, ent->cksum, cfg->similarity_cksum_sz) == 0 &&
-			    ent->item_size == item_size && ent->item_offset != item_offset) {
+			    ent->item_size == item_size) {
 				return (ent);
 			}
 			pent = &(ent->next);
@@ -496,8 +497,7 @@ db_lookup_insert_s(archive_config_t *cfg, uchar_t *sim_cksum, int interval,
 		}
 	} else if (cfg->similarity_cksum_sz == 8) {// Fast path for 64-bit keys
 		while (ent) {
-			if (*((uint64_t *)sim_cksum) == *((uint64_t *)ent->cksum) &&
-			    ent->item_offset != item_offset) {
+			if (*((uint64_t *)sim_cksum) == *((uint64_t *)ent->cksum)) {
 				return (ent);
 			}
 			pent = &(ent->next);
@@ -505,8 +505,7 @@ 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 &&
-			    ent->item_offset != item_offset) {
+			if (mycmp(sim_cksum, ent->cksum, cfg->similarity_cksum_sz) == 0) {
 				return (ent);
 			}
 			pent = &(ent->next);
diff --git a/rabin/rabin_dedup.c b/rabin/rabin_dedup.c
index d0fa12d..07e1f78 100755
--- a/rabin/rabin_dedup.c
+++ b/rabin/rabin_dedup.c
@@ -989,7 +989,7 @@ process_blocks:
 					 * file. Sort the offsets to avoid subsequent random access.
 					 */
 					tgt = src + 1;
-					isort_uint64((uint64_t *)tgt, k);
+					isort_uint64((uint64_t *)tgt, sub_i);
 
 					/*
 					 * Now eliminate duplicate offsets and UINT64_MAX offset entries which