From 962a2cae8a00ae64af72a469a0353202fc72f95a Mon Sep 17 00:00:00 2001
From: Moinak Ghosh <moinakg@gmail.com>
Date: Sun, 22 Jul 2012 00:00:41 +0530
Subject: [PATCH] Compress Dedup index only if it is at least 90 bytes to avoid
 expansion. Some minor cleanup.

---
 main.c                   | 24 ++++++++++++++++++------
 rabin/rabin_polynomial.c |  1 -
 rabin/rabin_polynomial.h |  6 +++---
 3 files changed, 21 insertions(+), 10 deletions(-)

diff --git a/main.c b/main.c
index 3471ba3..fa74662 100644
--- a/main.c
+++ b/main.c
@@ -216,10 +216,16 @@ redo:
 				goto cont;
 			}
 
+			rv = 0;
 			cmpbuf = cseg + RABIN_HDR_SIZE;
 			ubuf = tdat->uncompressed_chunk + RABIN_HDR_SIZE;
-			rv = lzma_decompress(cmpbuf, rabin_index_sz_cmp, ubuf, &rabin_index_sz,
-			    tdat->rctx->level, 0, tdat->rctx->lzma_data);
+			if (rabin_index_sz >= 90) {
+				/* Index should be at least 90 bytes to have been compressed. */
+				rv = lzma_decompress(cmpbuf, rabin_index_sz_cmp, ubuf,
+				    &rabin_index_sz, tdat->rctx->level, 0, tdat->rctx->lzma_data);
+			} else {
+				memcpy(ubuf, cmpbuf, rabin_index_sz);
+			}
 		} else {
 			rv = tdat->decompress(cseg, tdat->len_cmp, tdat->uncompressed_chunk,
 			    &_chunksize, tdat->level, HDR, tdat->data);
@@ -630,10 +636,16 @@ redo:
 		index_size_cmp = rabin_index_sz;
 		memcpy(compressed_chunk, tdat->uncompressed_chunk, RABIN_HDR_SIZE);
 
-		/* Compress index. */
-		rv = lzma_compress(tdat->uncompressed_chunk + RABIN_HDR_SIZE,
-		    rabin_index_sz, compressed_chunk + RABIN_HDR_SIZE, &index_size_cmp,
-		    tdat->rctx->level, 0, tdat->rctx->lzma_data);
+		rv = 0;
+		if (rabin_index_sz >= 90) {
+			/* Compress index if it is at least 90 bytes. */
+			rv = lzma_compress(tdat->uncompressed_chunk + RABIN_HDR_SIZE,
+			    rabin_index_sz, compressed_chunk + RABIN_HDR_SIZE,
+			    &index_size_cmp, tdat->rctx->level, 0, tdat->rctx->lzma_data);
+		} else {
+			memcpy(compressed_chunk + RABIN_HDR_SIZE,
+			    tdat->uncompressed_chunk + RABIN_HDR_SIZE, rabin_index_sz);
+		}
 
 		index_size_cmp += RABIN_HDR_SIZE;
 		rabin_index_sz += RABIN_HDR_SIZE;
diff --git a/rabin/rabin_polynomial.c b/rabin/rabin_polynomial.c
index 858fa32..2bf1ef7 100755
--- a/rabin/rabin_polynomial.c
+++ b/rabin/rabin_polynomial.c
@@ -424,7 +424,6 @@ rabin_dedup(rabin_context_t *ctx, uchar_t *buf, ssize_t *size, ssize_t offset, s
 
 				if (blk > 0 && ctx->blocks[blk].refcount == 0 &&
 				    ctx->blocks[blk].cksum_n_offset == prev_cksum) {
-					ssize_t sz1, sz2;
 					ctx->blocks[blk].index = prev_index;
 					ctx->blocks[blk].similar = SIMILAR_PARTIAL;
 					(ctx->blocks[prev_blk].refcount)++;
diff --git a/rabin/rabin_polynomial.h b/rabin/rabin_polynomial.h
index fe67122..91eb54e 100644
--- a/rabin/rabin_polynomial.h
+++ b/rabin/rabin_polynomial.h
@@ -124,9 +124,9 @@
 // Mask to extract value from a rabin index entry
 #define	RABIN_INDEX_VALUE (0x3FFFFFFFUL)
 
-// Tolerance for partial similarity check. We expect 80% similarity for
-// delta compression. See: http://www.armedia.com/wp/SimilarityIndex.pdf
-#define	SIMILARITY_TOLERANCE (0.2f)
+/*
+ * Types of block similarity.
+ */
 #define	SIMILAR_EXACT 1
 #define	SIMILAR_PARTIAL 2