thin-provisioning-tools/thin-provisioning/thin_show_duplicates.cc

339 lines
8.4 KiB
C++
Raw Normal View History

// Copyright (C) 2015 Red Hat, Inc. All rights reserved.
//
// This file is part of the thin-provisioning-tools source.
//
// thin-provisioning-tools is free software: you can redistribute it
// and/or modify it under the terms of the GNU General Public License
// as published by the Free Software Foundation, either version 3 of
// the License, or (at your option) any later version.
//
// thin-provisioning-tools is distributed in the hope that it will be
// useful, but WITHOUT ANY WARRANTY; without even the implied warranty
// of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU General Public License for more details.
//
// You should have received a copy of the GNU General Public License along
// with thin-provisioning-tools. If not, see
// <http://www.gnu.org/licenses/>.
#include <iostream>
#include <getopt.h>
#include <libgen.h>
#include "version.h"
#include "base/application.h"
#include "base/error_state.h"
#include "base/progress_monitor.h"
2015-08-21 17:40:49 +05:30
#include "persistent-data/data-structures/btree_damage_visitor.h"
#include "persistent-data/file_utils.h"
#include "persistent-data/space-maps/core.h"
#include "persistent-data/space-maps/disk.h"
#include "thin-provisioning/cache_stream.h"
#include "thin-provisioning/pool_stream.h"
#include "thin-provisioning/commands.h"
#include "thin-provisioning/device_tree.h"
#include "thin-provisioning/mapping_tree.h"
2015-08-21 17:40:49 +05:30
#include "thin-provisioning/rmap_visitor.h"
#include "thin-provisioning/superblock.h"
2015-09-03 17:32:29 +05:30
#include "thin-provisioning/variable_chunk_stream.h"
2015-08-19 17:02:57 +05:30
#include <boost/uuid/sha1.hpp>
#include <boost/lexical_cast.hpp>
2015-08-21 17:40:49 +05:30
#include <boost/optional.hpp>
#include <deque>
2015-08-19 17:02:57 +05:30
#include <vector>
using namespace base;
2015-08-21 17:40:49 +05:30
using namespace boost;
using namespace persistent_data;
using namespace std;
using namespace thin_provisioning;
//----------------------------------------------------------------
namespace {
2015-08-21 17:40:49 +05:30
bool factor_of(block_address f, block_address n) {
cerr << n << " % " << f << "\n";
return (n % f) == 0;
}
block_manager<>::ptr
open_bm(string const &path) {
block_address nr_blocks = get_nr_blocks(path);
block_manager<>::mode m = block_manager<>::READ_ONLY;
return block_manager<>::ptr(new block_manager<>(path, nr_blocks, 1, m));
}
transaction_manager::ptr
open_tm(block_manager<>::ptr bm) {
space_map::ptr sm(new core_map(bm->get_nr_blocks()));
sm->inc(superblock_detail::SUPERBLOCK_LOCATION);
transaction_manager::ptr tm(new transaction_manager(bm, sm));
return tm;
}
uint64_t parse_int(string const &str, string const &desc) {
try {
return boost::lexical_cast<uint64_t>(str);
} catch (...) {
ostringstream out;
out << "Couldn't parse " << desc << ": '" << str << "'";
exit(1);
}
return 0; // never get here
}
//--------------------------------
struct flags {
2015-08-19 17:02:57 +05:30
flags()
: cache_mem(64 * 1024 * 1024),
content_based_chunks(false) {
}
2015-08-19 17:02:57 +05:30
2015-08-21 17:40:49 +05:30
string data_dev;
optional<string> metadata_dev;
optional<unsigned> block_size;
2015-08-19 17:02:57 +05:30
unsigned cache_mem;
bool content_based_chunks;
};
2015-08-21 17:40:49 +05:30
using namespace mapping_tree_detail;
2015-08-19 17:02:57 +05:30
class duplicate_counter {
public:
2015-09-03 17:32:29 +05:30
duplicate_counter()
: non_zero_dups_(0),
2015-08-25 13:44:40 +05:30
zero_dups_(0) {
2015-08-19 17:02:57 +05:30
}
2015-09-03 17:32:29 +05:30
void add_duplicate(block_address len) {
non_zero_dups_ += len;
2015-08-19 17:02:57 +05:30
}
2015-09-03 17:32:29 +05:30
void add_zero_duplicate(block_address len) {
zero_dups_ += len;
2015-08-25 13:44:40 +05:30
}
2015-08-19 17:02:57 +05:30
block_address get_total() const {
2015-08-25 13:44:40 +05:30
return non_zero_dups_ + zero_dups_;
}
block_address get_non_zeroes() const {
return non_zero_dups_;
}
block_address get_zeroes() const {
return zero_dups_;
2015-08-19 17:02:57 +05:30
}
void display_results(chunk_stream const &stream) const {
block_address meg = 1024 * 1024;
cout << "\n\n"
<< stream.size() / meg << "m examined, "
<< get_non_zeroes() / meg << "m duplicates, "
<< get_zeroes() / meg << "m zeroes\n";
}
2015-08-19 17:02:57 +05:30
private:
2015-08-25 13:44:40 +05:30
block_address non_zero_dups_;
block_address zero_dups_;
2015-08-19 17:02:57 +05:30
};
class duplicate_detector {
public:
void scan(chunk_stream &stream) {
block_address total_seen(0);
auto_ptr<progress_monitor> pbar = create_progress_bar("Examining data");
do {
// FIXME: use a wrapper class to automate the put()
chunk const &c = stream.get();
examine(c);
stream.put(c);
total_seen += c.len_;
pbar->update_percent((total_seen * 100) / stream.size());
} while (stream.next());
pbar->update_percent(100);
results_.display_results(stream);
}
void scan_with_variable_sized_chunks(chunk_stream &stream) {
variable_chunk_stream vstream(stream, 4096);
scan(vstream);
}
duplicate_counter const &get_results() const {
return results_;
}
private:
2015-08-25 13:44:40 +05:30
void examine(chunk const &c) {
2015-09-03 17:32:29 +05:30
if (all_zeroes(c))
results_.add_zero_duplicate(c.len_);
2015-08-25 13:44:40 +05:30
2015-09-03 17:32:29 +05:30
else {
digestor_.reset();
digestor_.process_bytes(c.mem_.begin, c.mem_.end - c.mem_.begin);
2015-08-19 17:02:57 +05:30
2015-09-03 17:32:29 +05:30
unsigned int digest[5];
digestor_.get_digest(digest);
2015-08-25 13:44:40 +05:30
2015-09-03 17:32:29 +05:30
// hack
vector<unsigned int> v(5);
for (unsigned i = 0; i < 5; i++)
v[i] = digest[i];
2015-08-25 13:44:40 +05:30
fingerprint_map::const_iterator it = fm_.find(v);
if (it != fm_.end()) {
2015-09-03 17:32:29 +05:30
results_.add_duplicate(c.len_);
2015-08-25 13:44:40 +05:30
} else
2015-09-03 17:32:29 +05:30
fm_.insert(make_pair(v, c.offset_));
2015-08-25 13:44:40 +05:30
}
2015-08-19 17:02:57 +05:30
}
2015-09-03 17:32:29 +05:30
bool all_zeroes(chunk const &c) const {
for (uint8_t *ptr = c.mem_.begin; ptr != c.mem_.end; ptr++) {
if (*ptr != 0)
return false;
2015-09-03 17:32:29 +05:30
}
2015-09-03 17:32:29 +05:30
return true;
2015-08-19 17:02:57 +05:30
}
typedef map<vector<unsigned int>, block_address> fingerprint_map;
unsigned block_size_;
boost::uuids::detail::sha1 digestor_;
fingerprint_map fm_;
duplicate_counter results_;
};
2015-08-21 17:40:49 +05:30
int show_dups_pool(flags const &fs) {
block_manager<>::ptr bm = open_bm(*fs.metadata_dev);
transaction_manager::ptr tm = open_tm(bm);
superblock_detail::superblock sb = read_superblock(bm);
block_address block_size = sb.data_block_size_ * 512;
block_address nr_blocks = get_nr_blocks(fs.data_dev, block_size);
cache_stream stream(fs.data_dev, block_size, fs.cache_mem);
pool_stream pstream(stream, tm, sb, nr_blocks);
2015-08-21 17:40:49 +05:30
duplicate_detector detector;
if (fs.content_based_chunks)
detector.scan_with_variable_sized_chunks(pstream);
else
detector.scan(pstream);
2015-08-21 17:40:49 +05:30
return 0;
}
int show_dups_linear(flags const &fs) {
if (!fs.block_size)
// FIXME: this check should be moved to the switch parsing
throw runtime_error("--block-sectors or --metadata-dev must be supplied");
block_address block_size = *fs.block_size;
2015-08-21 17:40:49 +05:30
block_address nr_blocks = get_nr_blocks(fs.data_dev, *fs.block_size);
cerr << "path = " << fs.data_dev << "\n";
2015-08-19 17:02:57 +05:30
cerr << "nr_blocks = " << nr_blocks << "\n";
cerr << "block size = " << block_size << "\n";
2015-08-19 17:02:57 +05:30
cache_stream stream(fs.data_dev, block_size, fs.cache_mem);
duplicate_detector dd;
if (fs.content_based_chunks)
dd.scan_with_variable_sized_chunks(stream);
else
dd.scan(stream);
2015-08-19 17:02:57 +05:30
return 0;
}
2015-08-21 17:40:49 +05:30
int show_dups(flags const &fs) {
if (fs.metadata_dev)
return show_dups_pool(fs);
else {
cerr << "No metadata device provided, so treating data device as a linear device\n";
return show_dups_linear(fs);
2015-08-21 17:40:49 +05:30
}
}
void usage(ostream &out, string const &cmd) {
2015-08-19 17:02:57 +05:30
out << "Usage: " << cmd << " [options] {device|file}\n"
<< "Options:\n"
<< " {--block-sectors} <integer>\n"
<< " {--content-based-chunks}\n"
2015-08-21 17:40:49 +05:30
<< " {--metadata-dev} <path>\n"
2015-08-19 17:02:57 +05:30
<< " {-h|--help}\n"
<< " {-V|--version}" << endl;
}
}
int thin_show_dups_main(int argc, char **argv)
{
int c;
flags fs;
char const shortopts[] = "qhV";
option const longopts[] = {
{ "block-sectors", required_argument, NULL, 1},
{ "content-based-chunks", no_argument, NULL, 2},
{ "metadata-dev", required_argument, NULL, 3},
{ "help", no_argument, NULL, 'h'},
{ "version", no_argument, NULL, 'V'},
{ NULL, no_argument, NULL, 0 }
};
while ((c = getopt_long(argc, argv, shortopts, longopts, NULL)) != -1) {
switch(c) {
case 'h':
usage(cout, basename(argv[0]));
return 0;
case 'V':
cout << THIN_PROVISIONING_TOOLS_VERSION << endl;
return 0;
case 1:
fs.block_size = 512 * parse_int(optarg, "block sectors");
break;
2015-08-21 17:40:49 +05:30
case 2:
fs.content_based_chunks = true;
break;
case 3:
2015-08-21 17:40:49 +05:30
fs.metadata_dev = optarg;
break;
default:
usage(cerr, basename(argv[0]));
return 1;
}
}
if (argc == optind) {
2015-08-21 17:40:49 +05:30
cerr << "No data device/file provided." << endl;
usage(cerr, basename(argv[0]));
exit(1);
}
2015-08-21 17:40:49 +05:30
fs.data_dev = argv[optind];
return show_dups(fs);
}
base::command thin_provisioning::thin_show_dups_cmd("thin_show_duplicates", thin_show_dups_main);
//----------------------------------------------------------------