more checksum stuff

Makefile

@@ -6,6 +6,7 @@ SOURCE=\
 	checksum.cc \
 	endian_utils.cc \
 	error_set.cc \
+	hex_dump.cc \
 	metadata.cc \
 	metadata_disk_structures.cc \
 	space_map_disk.cc \
@@ -15,7 +16,7 @@ OBJECTS=$(subst .cc,.o,$(SOURCE))
 TOP_DIR:=$(PWD)
 CPPFLAGS=-Wall -g -I$(TOP_DIR)
 #CPPFLAGS=-Wall -std=c++0x -g -I$(TOP_DIR)
-LIBS=-lz -lstdc++
+LIBS=-lstdc++
 
 .PHONEY: test-programs
 

btree.h

@@ -38,6 +38,8 @@ namespace persistent_data {
 		using namespace std;
 		using namespace boost;
 
+		uint32_t const BTREE_CSUM_XOR = 121107;
+
 		//------------------------------------------------
 		// On disk data layout for btree nodes
 		enum node_flags {
@@ -74,6 +76,8 @@ namespace persistent_data {
 		public:
 			explicit node_ref(block_address b, disk_node *raw);
 
+			uint32_t get_checksum() const;
+
 			block_address get_location() const {
 				return location_;
 			}
@@ -125,8 +129,11 @@ namespace persistent_data {
 			template <typename RefCounter>
 			void inc_children(RefCounter &rc);
 
-			// FIXME: remove
-			void *raw() {
+			disk_node *raw() {
+				return raw_;
+			}
+
+			disk_node const *raw() const {
 				return raw_;
 			}
 

btree.tcc

@@ -16,6 +16,13 @@ node_ref<ValueTraits>::node_ref(block_address location, disk_node *raw)
 {
 }
 
+template <typename ValueTraits>
+uint32_t
+node_ref<ValueTraits>::get_checksum() const
+{
+	return to_cpu<uint32_t>(raw_->header.csum);
+}
+
 template <typename ValueTraits>
 block_address
 node_ref<ValueTraits>::get_block_nr() const

btree_checker.h

@@ -3,7 +3,9 @@
 
 #include "btree.h"
 
+#include "checksum.h"
 #include "error_set.h"
+#include "hex_dump.h"
 
 #include <sstream>
 #include <map>
@@ -78,6 +80,8 @@ namespace persistent_data {
 			if (already_visited(n))
 				return false;
 
+			check_sum(n);
+
 			if (!key)
 			  new_root(level);
 
@@ -95,6 +99,8 @@ namespace persistent_data {
 			if (already_visited(n))
 				return false;
 
+			check_sum(n);
+
 			if (!key)
 			  new_root(level);
 
@@ -114,6 +120,8 @@ namespace persistent_data {
 			if (already_visited(n))
 				return false;
 
+			check_sum(n);
+
 			if (!key)
 			  new_root(level);
 
@@ -149,6 +157,22 @@ namespace persistent_data {
 			return false;
 		}
 
+		template <typename node>
+		void check_sum(node const &n) const {
+			crc32c sum(BTREE_CSUM_XOR);
+
+			disk_node const *data = n.raw();
+			sum.append(&data->header.flags, MD_BLOCK_SIZE - sizeof(uint32_t));
+			if (sum.get_sum() != n.get_checksum()) {
+				std::ostringstream out;
+				out << "checksum error for block " << n.get_block_nr()
+				    << ", sum was " << sum.get_sum()
+				    << ", expected " << n.get_checksum();
+				errs_->add_child(out.str());
+				throw runtime_error(out.str());
+			}
+		}
+
 		template <typename node>
 		void check_block_nr(node const &n) const {
 			if (n.get_location() != n.get_block_nr()) {

checksum.cc

@@ -1,25 +1,31 @@
 #include "checksum.h"
 
-#include <zlib.h>
+#include <boost/crc.hpp>
 
 using namespace base;
 
 //----------------------------------------------------------------
 
-crc32::crc32(uint32_t seed)
-	: sum_(seed) {
+crc32c::crc32c(uint32_t xor_value)
+	: xor_value_(xor_value),
+	  sum_(0)
+{
 }
 
 void
-crc32::append(void const *buffer, unsigned len)
+crc32c::append(void const *buffer, unsigned len)
 {
-	sum_ = ::crc32(sum_, reinterpret_cast<Bytef const *>(buffer), len);
+	uint32_t const powers = 0x1EDC6F41;
+
+	boost::crc_basic<32> crc(powers, 0xffffffff, 0, true, true);
+	crc.process_bytes(buffer, len);
+	sum_ = crc.checksum();
 }
 
 uint32_t
-crc32::get_sum() const
+crc32c::get_sum() const
 {
-	return sum_;
+	return sum_ ^ xor_value_;
 }
 
 //----------------------------------------------------------------

checksum.h

@@ -6,13 +6,15 @@
 //----------------------------------------------------------------
 
 namespace base {
-	class crc32 {
-		crc32(uint32_t seed);
+	class crc32c {
+	public:
+		crc32c(uint32_t xor_value);
 
 		void append(void const *buffer, unsigned len);
 		uint32_t get_sum() const;
 
 	private:
+		uint32_t xor_value_;
 		uint32_t sum_;
 	};
 }

hex_dump.cc

@@ -0,0 +1,24 @@
+#include "hex_dump.h"
+
+#include <iostream>
+#include <iomanip>
+
+using namespace std;
+
+//----------------------------------------------------------------
+
+void base::hex_dump(ostream &out, void const *data_, size_t len)
+{
+	unsigned char const *data = reinterpret_cast<unsigned char const *>(data_),
+		*end = data + len;
+	out << hex;
+
+	while (data < end) {
+		for (unsigned i = 0; i < 16 && data < end; i++, data++)
+			out << setw(2) << setfill('0') << (unsigned) *data << " ";
+		out << endl;
+	}
+	out << dec;
+}
+
+//----------------------------------------------------------------

hex_dump.h

@@ -0,0 +1,14 @@
+#ifndef HEX_DUMP_H
+#define HEX_DUMP_H
+
+#include <iosfwd>
+
+//----------------------------------------------------------------
+
+namespace base {
+	void hex_dump(std::ostream &out, void const *data, size_t len);
+}
+
+//----------------------------------------------------------------
+
+#endif

metadata.cc

@@ -74,6 +74,17 @@ namespace {
 		superblock sb;
 		block_manager<>::read_ref r = bm->read_lock(SUPERBLOCK_LOCATION);
 		superblock_disk const *sbd = reinterpret_cast<superblock_disk const *>(&r.data());
+
+		crc32c sum(160774);
+		sum.append(&sbd->flags_, MD_BLOCK_SIZE - sizeof(uint32_t));
+		if (sum.get_sum() != to_cpu<uint32_t>(sbd->csum_)) {
+			ostringstream out;
+			out << "bad checksum in superblock, calculated "
+			    << sum.get_sum()
+			    << ", superblock contains " << to_cpu<uint32_t>(sbd->csum_);
+			throw runtime_error(out.str());
+		}
+
 		superblock_traits::unpack(*sbd, sb);
 		return sb;
 	}

space_map_disk.cc

@@ -81,6 +81,11 @@ namespace {
 		}
 
 	private:
+		void check_crc(transaction_manager::read_ref &rr) {
+			crc32c sum(240779);
+			// sum.append(reinterpret_cast<bitmap_header const *>(&rr.data()[0]);
+		}
+
 		void *bitmap_data(transaction_manager::write_ref &wr) {
 			bitmap_header *h = reinterpret_cast<bitmap_header *>(&wr.data()[0]);
 			return h + 1;