From a13c61e92620cdf67744b6f203b9e717ed72ab67 Mon Sep 17 00:00:00 2001
From: Moinak Ghosh <moinakg@gmail.com>
Date: Mon, 2 Jul 2012 22:08:03 +0530
Subject: [PATCH] Change rabin index encoding scheme for better metadata
 compression.

---
 rabin/rabin_polynomial.c | 13 +++++++++----
 rabin/rabin_polynomial.h | 15 +++++++++++++++
 2 files changed, 24 insertions(+), 4 deletions(-)

diff --git a/rabin/rabin_polynomial.c b/rabin/rabin_polynomial.c
index 22a0b65..3fbd9e5 100755
--- a/rabin/rabin_polynomial.c
+++ b/rabin/rabin_polynomial.c
@@ -101,6 +101,11 @@ create_rabin_context(uint64_t chunksize) {
 	if (chunksize % rabin_polynomial_min_block_size)
 		blknum++;
 
+	if (blknum > RABIN_MAX_BLOCKS) {
+		fprintf(stderr, "Chunk size too large for dedup.\n");
+		destroy_rabin_context(ctx);
+		return (NULL);
+	}
 	ctx = (rabin_context_t *)slab_alloc(NULL, sizeof (rabin_context_t));
 	current_window_data = slab_alloc(NULL, RAB_POLYNOMIAL_WIN_SIZE);
 	ctx->blocks = (rabin_blockentry_t *)slab_alloc(NULL,
@@ -345,7 +350,7 @@ rabin_dedup(rabin_context_t *ctx, uchar_t *buf, ssize_t *size, ssize_t offset)
 						prev_length = 0;
 						rabin_index[blk] = htonl(be->length);
 					} else {
-						if (prev_length + be->length <= RAB_POLYNOMIAL_MAX_BLOCK_SIZE) {
+						if (prev_length + be->length <= RABIN_MAX_BLOCK_SIZE) {
 							prev_length += be->length;
 							rabin_index[prev_index] = htonl(prev_length);
 							rabin_index[blk] = 0;
@@ -359,7 +364,7 @@ rabin_dedup(rabin_context_t *ctx, uchar_t *buf, ssize_t *size, ssize_t offset)
 			} else {
 				prev_index = 0;
 				prev_length = 0;
-				rabin_index[blk] = htonl(RAB_POLYNOMIAL_MAX_BLOCK_SIZE + be->index + 1);
+				rabin_index[blk] = htonl(be->index | RABIN_INDEX_FLAG);
 			}
 		}
 
@@ -437,13 +442,13 @@ rabin_inverse_dedup(rabin_context_t *ctx, uchar_t *buf, ssize_t *size)
 			ctx->blocks[blk].length = 0;
 			ctx->blocks[blk].index = 0;
 
-		} else if (len <= RAB_POLYNOMIAL_MAX_BLOCK_SIZE) {
+		} else if (!(len & RABIN_INDEX_FLAG)) {
 			ctx->blocks[blk].length = len;
 			ctx->blocks[blk].offset = pos1;
 			pos1 += len;
 		} else {
 			ctx->blocks[blk].length = 0;
-			ctx->blocks[blk].index = len - RAB_POLYNOMIAL_MAX_BLOCK_SIZE - 1;
+			ctx->blocks[blk].index = len & RABIN_INDEX_VALUE;
 		}
 	}
 	for (blk = 0; blk < blknum; blk++) {
diff --git a/rabin/rabin_polynomial.h b/rabin/rabin_polynomial.h
index 71ac0d3..9248e2d 100644
--- a/rabin/rabin_polynomial.h
+++ b/rabin/rabin_polynomial.h
@@ -99,6 +99,21 @@ typedef struct {
 // size of deduped data, size of compressed data
 #define	RABIN_HDR_SIZE (sizeof (unsigned int) + sizeof (ssize_t) + sizeof (ssize_t) + sizeof (ssize_t) + sizeof (ssize_t))
 
+// Maximum number of dedup blocks supported (2^31 - 1)
+#define	RABIN_MAX_BLOCKS (0x7fffffff)
+
+// Maximum possible block size for a single rabin block. This is a hard limit much
+// larger than RAB_POLYNOMIAL_MAX_BLOCK_SIZE. Useful when merging non-duplicate blocks.
+// This is also 2^31 - 1.
+#define RABIN_MAX_BLOCK_SIZE (RABIN_MAX_BLOCKS)
+
+// Mask to determine whether Rabin index entry is a length value or index value.
+// MSB = 1 : Index
+// MSB = 0 : Length
+#define RABIN_INDEX_FLAG (0x80000000)
+// Mask to extract value from a rabin index entry
+#define RABIN_INDEX_VALUE (0x7fffffff)
+
 typedef struct {
 	unsigned char *current_window_data;
 	rabin_blockentry_t *blocks;