From 5d383c029334bb8d85dee5c81c40741c2975cf01 Mon Sep 17 00:00:00 2001
From: Joe Thornber <ejt@redhat.com>
Date: Fri, 4 Sep 2015 10:10:41 +0100
Subject: [PATCH] [thin_show_dups] get the backup break working in the rolling
 hash

---
 base/rolling_hash.cc                       | 23 ----------------------
 thin-provisioning/variable_chunk_stream.cc | 20 +++++++------------
 thin-provisioning/variable_chunk_stream.h  |  2 +-
 3 files changed, 8 insertions(+), 37 deletions(-)

diff --git a/base/rolling_hash.cc b/base/rolling_hash.cc
index 1ea362f..9c6e1bf 100644
--- a/base/rolling_hash.cc
+++ b/base/rolling_hash.cc
@@ -82,7 +82,6 @@ content_based_hash::reset()
 optional<unsigned>
 content_based_hash::step(uint8_t byte)
 {
-#if 0
 	optional<unsigned> r;
 
 	rhash_.step(byte);
@@ -114,28 +113,6 @@ content_based_hash::step(uint8_t byte)
 	}
 
 	return r;
-#else
-	optional<unsigned> r;
-
-	rhash_.step(byte);
-	len_++;
-
-	if (len_ < min_len_)
-		return r;
-
-	if (hit_break(div_)) {
-		// found a break
-		r = len_;
-		len_ = 0;
-		backup_break_.reset();
-
-	} else if (len_ >= max_len_) {
-		r = len_;
-		len_ = 0;
-	}
-
-	return r;
-#endif
 }
 
 bool
diff --git a/thin-provisioning/variable_chunk_stream.cc b/thin-provisioning/variable_chunk_stream.cc
index 9a9d11e..41c6c96 100644
--- a/thin-provisioning/variable_chunk_stream.cc
+++ b/thin-provisioning/variable_chunk_stream.cc
@@ -80,7 +80,7 @@ variable_chunk_stream::next_big_chunk()
 		return false;
 
 	big_chunk_ = &stream_.get();
-	little_b_ = little_e_ = big_chunk_->mem_.front().begin;
+	little_b_ = little_e_ = last_hashed_ = big_chunk_->mem_.front().begin;
 	h_.reset();
 
 	return true;
@@ -95,6 +95,7 @@ variable_chunk_stream::advance_one()
 
 	big_e = big_chunk_->mem_.front().end;
 	little_b_ = little_e_;
+	little_e_ = last_hashed_;
 
 	if (little_b_ == big_e) {
 		if (next_big_chunk())
@@ -105,35 +106,28 @@ variable_chunk_stream::advance_one()
 
 	assert(little_e_ >= big_chunk_->mem_.front().begin);
 	assert(little_b_ >= big_chunk_->mem_.front().begin);
-#if 1
-	if (little_e_ > big_e) {
-		cerr << "before -- little_e_: " << (void *) little_e_ << ", big_e: " << (void *) big_e << "\n";
-	}
-#endif
 	assert(little_e_ <= big_e);
 	assert(little_b_ <= big_e);
 
 
 	while (little_e_ != big_e) {
 		optional<unsigned> maybe_break = h_.step(*little_e_);
+		little_e_++;
 
 		if (maybe_break) {
 			// The break is not neccessarily at the current
 			// byte.
+			last_hashed_ = little_e_;
 			little_e_ = little_b_ + *maybe_break;
 			break;
 		}
-
-		little_e_++;
 	}
 
+	if (little_e_ == big_e)
+		last_hashed_ = little_e_;
+
 	assert(little_e_ >= big_chunk_->mem_.front().begin);
 	assert(little_b_ >= big_chunk_->mem_.front().begin);
-#if 1
-	if (little_e_ > big_e) {
-		cerr << "after -- little_e_: " << (void *) little_e_ << ", big_e: " << (void *) big_e << "\n";
-	}
-#endif
 	assert(little_e_ <= big_e);
 	assert(little_b_ <= big_e);
 
diff --git a/thin-provisioning/variable_chunk_stream.h b/thin-provisioning/variable_chunk_stream.h
index 0327f1d..f9c5ec7 100644
--- a/thin-provisioning/variable_chunk_stream.h
+++ b/thin-provisioning/variable_chunk_stream.h
@@ -32,7 +32,7 @@ namespace thin_provisioning {
 		chunk_stream &stream_;
 		chunk const *big_chunk_;
 
-		uint8_t *little_b_, *little_e_;
+		uint8_t *little_b_, *little_e_, *last_hashed_;
 		chunk little_chunk_;
 	};
 }