[era] switch from bloom filters to writesets

This commit is contained in:
Joe Thornber 2014-01-29 21:37:25 +00:00
parent e7bbfbc1fd
commit 8531a2befa
20 changed files with 439 additions and 160 deletions

View File

@ -27,6 +27,7 @@ PROGRAMS=\
\
era_check \
era_dump \
era_invalidate \
\
thin_check \
thin_dump \
@ -53,7 +54,7 @@ SOURCE=\
era/era_array.cc \
era/era_detail.cc \
era/superblock.cc \
era/bloom_tree.cc \
era/writeset_tree.cc \
era/metadata.cc \
era/metadata_dump.cc \
era/xml_format.cc \
@ -305,7 +306,7 @@ ERA_CHECK_SOURCE=\
base/error_state.cc \
base/endian_utils.cc \
\
era/bloom_tree.cc \
era/writeset_tree.cc \
era/era_detail.cc \
era/era_array.cc \
era/metadata.cc \
@ -335,7 +336,7 @@ ERA_DUMP_SOURCE=\
base/error_state.cc \
base/endian_utils.cc \
\
era/bloom_tree.cc \
era/writeset_tree.cc \
era/era_detail.cc \
era/era_array.cc \
era/metadata.cc \
@ -362,6 +363,38 @@ era_dump: $(ERA_DUMP_OBJECTS) era/era_dump.o
@echo " [LD] $@"
$(V) $(CXX) $(CXXFLAGS) $(LDFLAGS) -o $@ $+ $(LIBS)
ERA_INVALIDATE_SOURCE=\
base/base64.cc \
base/error_state.cc \
base/endian_utils.cc \
\
era/writeset_tree.cc \
era/era_detail.cc \
era/era_array.cc \
era/metadata.cc \
era/metadata_dump.cc \
era/superblock.cc \
era/xml_format.cc \
\
persistent-data/checksum.cc \
persistent-data/error_set.cc \
persistent-data/file_utils.cc \
persistent-data/hex_dump.cc \
persistent-data/lock_tracker.cc \
persistent-data/data-structures/btree.cc \
persistent-data/data-structures/bitset.cc \
persistent-data/space_map.cc \
persistent-data/space-maps/disk.cc \
persistent-data/space-maps/recursive.cc \
persistent-data/space-maps/careful_alloc.cc \
persistent-data/transaction_manager.cc \
ERA_INVALIDATE_OBJECTS=$(subst .cc,.o,$(ERA_INVALIDATE_SOURCE))
era_invalidate: $(ERA_INVALIDATE_OBJECTS) era/era_invalidate.o
@echo " [LD] $@"
$(V) $(CXX) $(CXXFLAGS) $(LDFLAGS) -o $@ $+ $(LIBS)
#----------------------------------------------------------------
DEPEND_FILES=\

View File

@ -20,9 +20,9 @@ namespace era {
uint32_t current_era) = 0;
virtual void end_superblock() = 0;
virtual void begin_bloom(uint32_t era, uint32_t nr_bits, pd::block_address nr_blocks) = 0;
virtual void bloom_bit(uint32_t bit, bool value) = 0;
virtual void end_bloom() = 0;
virtual void begin_writeset(uint32_t era, uint32_t nr_bits) = 0;
virtual void writeset_bit(uint32_t bit, bool value) = 0;
virtual void end_writeset() = 0;
virtual void begin_era_array() = 0;
virtual void era(pd::block_address block, uint32_t era) = 0;

View File

@ -14,7 +14,7 @@
#include "base/error_state.h"
#include "base/nested_output.h"
#include "era/bloom_tree.h"
#include "era/writeset_tree.h"
#include "era/era_array.h"
#include "era/superblock.h"
#include "persistent-data/block.h"
@ -92,14 +92,14 @@ namespace {
using reporter_base::get_error;
};
class bloom_tree_reporter : public bloom_tree_detail::damage_visitor, reporter_base {
class writeset_tree_reporter : public writeset_tree_detail::damage_visitor, reporter_base {
public:
bloom_tree_reporter(nested_output &o)
writeset_tree_reporter(nested_output &o)
: reporter_base(o) {
}
void visit(bloom_tree_detail::missing_eras const &d) {
out() << "missing eras from bloom tree" << end_message();
void visit(writeset_tree_detail::missing_eras const &d) {
out() << "missing eras from writeset tree" << end_message();
{
nested_output::nest _ = push();
out() << d.get_desc() << end_message();
@ -110,8 +110,8 @@ namespace {
mplus_error(FATAL);
}
void visit(bloom_tree_detail::damaged_bloom_filter const &d) {
out() << "damaged bloom filter" << end_message();
void visit(writeset_tree_detail::damaged_writeset const &d) {
out() << "damaged writeset" << end_message();
{
nested_output::nest _ = push();
out() << d.get_desc() << end_message();
@ -216,22 +216,22 @@ namespace {
superblock sb = read_superblock(bm);
transaction_manager::ptr tm = open_tm(bm);
bloom_tree_reporter bt_rep(out);
writeset_tree_reporter wt_rep(out);
{
era_detail_traits::ref_counter rc(tm);
bloom_tree bt(tm, sb.bloom_tree_root, rc);
check_bloom_tree(tm, bt, bt_rep);
writeset_tree wt(tm, sb.writeset_tree_root, rc);
check_writeset_tree(tm, wt, wt_rep);
}
era_array_reporter ea_rep(out);
{
uint32_traits::ref_counter rc;
era_array ea(tm, rc, sb.era_array_root, sb.nr_blocks);
check_era_array(ea, ea_rep);
check_era_array(ea, sb.current_era, ea_rep);
}
return combine_errors(sb_rep.get_error(),
combine_errors(bt_rep.get_error(),
combine_errors(wt_rep.get_error(),
ea_rep.get_error()));
}

View File

@ -7,37 +7,18 @@ using namespace era;
//----------------------------------------------------------------
namespace {
#if 0
le32 pack_hash_detail(uint32_t hash1, uint32_t hash2, uint32_t nr_probes) {
throw std::runtime_error("not implemented");
}
void unpack_hash_detail(le32 packed, uint32_t &hash1, uint32_t &hash2, uint32_t &nr_probes) {
throw std::runtime_error("not implemented");
}
#endif
}
void
era_detail_traits::unpack(disk_type const &disk, value_type &value)
{
value.nr_blocks = to_cpu<uint32_t>(disk.nr_blocks);
value.nr_bits = to_cpu<uint32_t>(disk.nr_bits);
value.nr_set = to_cpu<uint32_t>(disk.nr_set);
value.bloom_root = to_cpu<uint64_t>(disk.bloom_root);
//unpack_hash_detail(disk.hash_fns_and_probes, value.hash1, value.hash2, value.nr_probes);
value.writeset_root = to_cpu<uint64_t>(disk.writeset_root);
}
void
era_detail_traits::pack(value_type const &value, disk_type &disk)
{
disk.nr_blocks = to_disk<le32>(value.nr_blocks);
disk.nr_bits = to_disk<le32>(value.nr_bits);
disk.nr_set = to_disk<le32>(value.nr_set);
disk.bloom_root = to_disk<le64>(value.bloom_root);
// disk.hash_fns_and_probes = pack_hash_detail(value.hash1, value.hash2, value.nr_probes);
disk.writeset_root = to_disk<le64>(value.writeset_root);
}
//----------------------------------------------------------------

View File

@ -8,22 +8,13 @@
namespace era {
struct era_detail_disk {
base::le32 nr_blocks;
base::le32 nr_bits;
base::le32 nr_set;
base::le64 bloom_root;
base::le64 writeset_root;
} __attribute__ ((packed));
struct era_detail {
uint32_t nr_blocks;
uint32_t nr_bits;
uint32_t nr_set;
uint32_t hash1;
uint32_t hash2;
uint32_t nr_probes;
uint64_t bloom_root;
uint64_t writeset_root;
};
// FIXME: implement

View File

@ -5,7 +5,7 @@
#include "version.h"
#include "era/era_array.h"
#include "era/bloom_tree.h"
#include "era/writeset_tree.h"
#include "era/metadata.h"
#include "era/metadata_dump.h"
#include "era/xml_format.h"

114
era/era_invalidate.cc Normal file
View File

@ -0,0 +1,114 @@
#include <fstream>
#include <getopt.h>
#include <libgen.h>
#include <iostream>
#include "version.h"
#include "era/era_array.h"
#include "era/writeset_tree.h"
#include "era/metadata.h"
#include "era/xml_format.h"
#include "persistent-data/file_utils.h"
using namespace era;
using namespace std;
//----------------------------------------------------------------
namespace {
struct flags {
flags() {
}
};
//--------------------------------
void emit_blocks() {
}
//--------------------------------
string const STDOUT_PATH("-");
bool want_stdout(string const &output) {
return output == STDOUT_PATH;
}
int invalidate(string const &dev, string const &output, flags const &fs) {
try {
block_manager<>::ptr bm = open_bm(dev, block_io<>::READ_ONLY);
metadata::ptr md(new metadata(bm, metadata::OPEN));
if (want_stdout(output)) {
emitter::ptr e = create_xml_emitter(cout);
//emit_blocks(md, e, fs);
} else {
ofstream out(output.c_str());
emitter::ptr e = create_xml_emitter(out);
//emit_blocks(md, e, fs);
}
} catch (std::exception &e) {
cerr << e.what() << endl;
return 1;
}
return 0;
}
void usage(ostream &out, string const &cmd) {
out << "Usage: " << cmd << " [options] {device|file}" << endl
<< "Options:" << endl
<< " {-h|--help}" << endl
<< " {-o <xml file>}" << endl
<< " {-V|--version}" << endl;
}
}
//----------------------------------------------------------------
int main(int argc, char **argv)
{
int c;
flags fs;
string output("-");
char const shortopts[] = "ho:V";
option const longopts[] = {
{ "help", no_argument, NULL, 'h' },
{ "output", required_argument, NULL, 'o' },
{ "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 'o':
output = optarg;
break;
case 'V':
cout << THIN_PROVISIONING_TOOLS_VERSION << endl;
return 0;
default:
usage(cerr, basename(argv[0]));
return 1;
}
}
if (argc == optind) {
cerr << "No input file provided." << endl;
usage(cerr, basename(argv[0]));
return 1;
}
return invalidate(argv[optind], output, fs);
}
//----------------------------------------------------------------

View File

@ -47,8 +47,8 @@ metadata::open_metadata(block_manager<>::ptr bm)
tm_ = open_tm(bm);
sb_ = read_superblock(tm_->get_bm());
bloom_tree_ = bloom_tree::ptr(new bloom_tree(tm_,
sb_.bloom_tree_root,
writeset_tree_ = writeset_tree::ptr(new writeset_tree(tm_,
sb_.writeset_tree_root,
era_detail_traits::ref_counter(tm_)));
era_array_ = era_array::ptr(new era_array(tm_,

View File

@ -10,7 +10,7 @@
#include "persistent-data/transaction_manager.h"
#include "era/superblock.h"
#include "era/bloom_tree.h"
#include "era/writeset_tree.h"
#include "era/era_array.h"
//----------------------------------------------------------------
@ -34,7 +34,7 @@ namespace era {
tm::ptr tm_;
superblock sb_;
checked_space_map::ptr metadata_sm_;
bloom_tree::ptr bloom_tree_;
writeset_tree::ptr writeset_tree_;
era_array::ptr era_array_;
private:

View File

@ -17,42 +17,42 @@ namespace {
"perhaps you wanted to run with --repair ?");
}
class bloom_tree_emitter : public bloom_tree_detail::bloom_visitor {
class writeset_tree_emitter : public writeset_tree_detail::writeset_visitor {
public:
bloom_tree_emitter(emitter::ptr e)
writeset_tree_emitter(emitter::ptr e)
: e_(e) {
}
virtual void bloom_begin(uint32_t era, uint32_t nr_blocks, uint32_t nr_bits, uint32_t nr_set) {
e_->begin_bloom(era, nr_bits, nr_blocks);
virtual void writeset_begin(uint32_t era, uint32_t nr_bits) {
e_->begin_writeset(era, nr_bits);
}
virtual void bit(uint32_t bit, bool value) {
e_->bloom_bit(bit, value);
e_->writeset_bit(bit, value);
}
virtual void bloom_end() {
e_->end_bloom();
virtual void writeset_end() {
e_->end_writeset();
}
private:
emitter::ptr e_;
};
struct ignore_bloom_tree_damage : public bloom_tree_detail::damage_visitor {
void visit(bloom_tree_detail::missing_eras const &d) {
struct ignore_writeset_tree_damage : public writeset_tree_detail::damage_visitor {
void visit(writeset_tree_detail::missing_eras const &d) {
}
void visit(bloom_tree_detail::damaged_bloom_filter const &d) {
void visit(writeset_tree_detail::damaged_writeset const &d) {
}
};
struct fatal_bloom_tree_damage : public bloom_tree_detail::damage_visitor {
void visit(bloom_tree_detail::missing_eras const &d) {
struct fatal_writeset_tree_damage : public writeset_tree_detail::damage_visitor {
void visit(writeset_tree_detail::missing_eras const &d) {
raise_metadata_damage();
}
void visit(bloom_tree_detail::damaged_bloom_filter const &d) {
void visit(writeset_tree_detail::damaged_writeset const &d) {
raise_metadata_damage();
}
};
@ -104,15 +104,15 @@ era::metadata_dump(metadata::ptr md, emitter::ptr e, bool repair)
sb.current_era);
{
{
bloom_tree_emitter visitor(e);
writeset_tree_emitter visitor(e);
ignore_bloom_tree_damage ignore;
fatal_bloom_tree_damage fatal;
bloom_tree_detail::damage_visitor &dv = repair ?
static_cast<bloom_tree_detail::damage_visitor &>(ignore) :
static_cast<bloom_tree_detail::damage_visitor &>(fatal);
ignore_writeset_tree_damage ignore;
fatal_writeset_tree_damage fatal;
writeset_tree_detail::damage_visitor &dv = repair ?
static_cast<writeset_tree_detail::damage_visitor &>(ignore) :
static_cast<writeset_tree_detail::damage_visitor &>(fatal);
walk_bloom_tree(md->tm_, *md->bloom_tree_, visitor, dv);
walk_writeset_tree(md->tm_, *md->writeset_tree_, visitor, dv);
}
e->begin_era_array();

View File

@ -34,7 +34,7 @@ namespace {
le32 current_era;
era_detail_disk current_detail;
le64 bloom_tree_root;
le64 writeset_tree_root;
le64 era_array_root;
} __attribute__ ((packed));
@ -116,7 +116,7 @@ superblock::superblock()
metadata_block_size(8),
nr_blocks(0),
current_era(0),
bloom_tree_root(0),
writeset_tree_root(0),
era_array_root(0)
{
memset(uuid, 0, sizeof(uuid));
@ -141,7 +141,7 @@ superblock_traits::unpack(disk_type const &disk, value_type &value)
value.nr_blocks = to_cpu<uint32_t>(disk.nr_blocks);
value.current_era = to_cpu<uint32_t>(disk.current_era);
era_detail_traits::unpack(disk.current_detail, value.current_detail);
value.bloom_tree_root = to_cpu<uint64_t>(disk.bloom_tree_root);
value.writeset_tree_root = to_cpu<uint64_t>(disk.writeset_tree_root);
value.era_array_root = to_cpu<uint64_t>(disk.era_array_root);
}
@ -161,7 +161,7 @@ superblock_traits::pack(value_type const &value, disk_type &disk)
disk.nr_blocks = to_disk<le32>(value.nr_blocks);
disk.current_era = to_disk<le32>(value.current_era);
era_detail_traits::pack(value.current_detail, disk.current_detail);
disk.bloom_tree_root = to_disk<le64>(value.bloom_tree_root);
disk.writeset_tree_root = to_disk<le64>(value.writeset_tree_root);
disk.era_array_root = to_disk<le64>(value.era_array_root);
}
@ -282,8 +282,8 @@ era::check_superblock(superblock const &sb,
visitor.visit(superblock_invalid(msg.str()));
}
if (sb.bloom_tree_root == SUPERBLOCK_LOCATION) {
string msg("bloom tree root points back to the superblock");
if (sb.writeset_tree_root == SUPERBLOCK_LOCATION) {
string msg("writeset tree root points back to the superblock");
visitor.visit(superblock_invalid(msg));
}
@ -292,9 +292,9 @@ era::check_superblock(superblock const &sb,
visitor.visit(superblock_invalid(msg));
}
if (sb.bloom_tree_root == sb.era_array_root) {
if (sb.writeset_tree_root == sb.era_array_root) {
ostringstream msg;
msg << "bloom tree root and era array both point to the same block: "
msg << "writeset tree root and era array both point to the same block: "
<< sb.era_array_root;
visitor.visit(superblock_invalid(msg.str()));
}

View File

@ -59,7 +59,7 @@ namespace era {
era_detail current_detail;
// A btree of undigested era_details
uint64_t bloom_tree_root;
uint64_t writeset_tree_root;
// Big array holding the digested era/block info.
uint64_t era_array_root;

View File

@ -1,10 +1,10 @@
#include "era/bloom_tree.h"
#include "era/writeset_tree.h"
#include "persistent-data/data-structures/btree_damage_visitor.h"
#include "persistent-data/data-structures/bitset.h"
using namespace boost;
using namespace era;
using namespace bloom_tree_detail;
using namespace writeset_tree_detail;
using namespace persistent_data;
using namespace std;
@ -22,7 +22,7 @@ missing_eras::visit(damage_visitor &v) const {
v.visit(*this);
}
damaged_bloom_filter::damaged_bloom_filter(string const &desc,
damaged_writeset::damaged_writeset(string const &desc,
uint32_t era,
run<uint32_t> missing_bits)
: damage(desc),
@ -32,7 +32,7 @@ damaged_bloom_filter::damaged_bloom_filter(string const &desc,
}
void
damaged_bloom_filter::visit(damage_visitor &v) const
damaged_writeset::visit(damage_visitor &v) const
{
v.visit(*this);
}
@ -40,39 +40,39 @@ damaged_bloom_filter::visit(damage_visitor &v) const
//----------------------------------------------------------------
namespace {
class ll_bloom_visitor : public bitset_detail::bitset_visitor {
class ll_writeset_visitor : public bitset_detail::bitset_visitor {
public:
typedef persistent_data::transaction_manager::ptr tm_ptr;
ll_bloom_visitor(tm_ptr tm,
bloom_tree_detail::bloom_visitor &bloom_v,
bloom_tree_detail::damage_visitor &dv)
ll_writeset_visitor(tm_ptr tm,
writeset_tree_detail::writeset_visitor &writeset_v,
writeset_tree_detail::damage_visitor &dv)
: tm_(tm),
bloom_v_(bloom_v),
writeset_v_(writeset_v),
dv_(dv) {
}
void visit(btree_path const &path, era_detail const &era) {
era_ = path[0];
bitset bs(tm_, era.bloom_root, era.nr_bits);
bloom_v_.bloom_begin(era_, era.nr_blocks, era.nr_bits, era.nr_set);
bitset bs(tm_, era.writeset_root, era.nr_bits);
writeset_v_.writeset_begin(era_, era.nr_bits);
bs.walk_bitset(*this);
bloom_v_.bloom_end();
writeset_v_.writeset_end();
}
void visit(uint32_t index, bool value) {
bloom_v_.bit(index, value);
writeset_v_.bit(index, value);
}
void visit(bitset_detail::missing_bits const &d) {
dv_.visit(bloom_tree_detail::damaged_bloom_filter("missing bits", era_, d.keys_));
dv_.visit(writeset_tree_detail::damaged_writeset("missing bits", era_, d.keys_));
}
private:
tm_ptr tm_;
uint64_t era_;
bloom_tree_detail::bloom_visitor &bloom_v_;
bloom_tree_detail::damage_visitor &dv_;
writeset_tree_detail::writeset_visitor &writeset_v_;
writeset_tree_detail::damage_visitor &dv_;
};
class ll_damage_visitor {
@ -98,38 +98,37 @@ namespace {
}
void
era::walk_bloom_tree(persistent_data::transaction_manager::ptr tm,
bloom_tree const &tree,
bloom_tree_detail::bloom_visitor &bloom_v,
bloom_tree_detail::damage_visitor &dv)
era::walk_writeset_tree(persistent_data::transaction_manager::ptr tm,
writeset_tree const &tree,
writeset_tree_detail::writeset_visitor &writeset_v,
writeset_tree_detail::damage_visitor &dv)
{
ll_bloom_visitor ll_bv(tm, bloom_v, dv);
ll_writeset_visitor ll_bv(tm, writeset_v, dv);
ll_damage_visitor ll_dv(dv);
btree_visit_values(tree, ll_bv, ll_dv);
}
namespace {
class noop_bloom_visitor : public bloom_tree_detail::bloom_visitor {
class noop_writeset_visitor : public writeset_tree_detail::writeset_visitor {
public:
void bloom_begin(uint32_t era, uint32_t nr_blocks,
uint32_t nr_bits, uint32_t nr_set) {
void writeset_begin(uint32_t era, uint32_t nr_bits) {
}
void bit(uint32_t index, bool value) {
}
void bloom_end() {
void writeset_end() {
}
};
};
void
era::check_bloom_tree(persistent_data::transaction_manager::ptr tm,
bloom_tree const &tree,
bloom_tree_detail::damage_visitor &dv)
era::check_writeset_tree(persistent_data::transaction_manager::ptr tm,
writeset_tree const &tree,
writeset_tree_detail::damage_visitor &dv)
{
noop_bloom_visitor bv;
walk_bloom_tree(tm, tree, bv, dv);
noop_writeset_visitor bv;
walk_writeset_tree(tm, tree, bv, dv);
}
//----------------------------------------------------------------

View File

@ -1,5 +1,5 @@
#ifndef ERA_BLOOM_TREE_H
#define ERA_BLOOM_TREE_H
#ifndef ERA_WRITESET_TREE_H
#define ERA_WRITESET_TREE_H
#include "era/era_detail.h"
#include "persistent-data/data-structures/btree.h"
@ -7,7 +7,7 @@
//----------------------------------------------------------------
namespace era {
namespace bloom_tree_detail {
namespace writeset_tree_detail {
class damage_visitor;
class damage {
@ -34,8 +34,8 @@ namespace era {
run<uint32_t> eras_;
};
struct damaged_bloom_filter : public damage {
damaged_bloom_filter(std::string const &desc,
struct damaged_writeset : public damage {
damaged_writeset(std::string const &desc,
uint32_t era,
run<uint32_t> missing_bits);
virtual void visit(damage_visitor &v) const;
@ -55,32 +55,31 @@ namespace era {
}
virtual void visit(missing_eras const &d) = 0;
virtual void visit(damaged_bloom_filter const &d) = 0;
virtual void visit(damaged_writeset const &d) = 0;
};
class bloom_visitor {
class writeset_visitor {
public:
typedef boost::shared_ptr<bloom_visitor> ptr;
typedef boost::shared_ptr<writeset_visitor> ptr;
virtual ~bloom_visitor() {}
virtual ~writeset_visitor() {}
virtual void bloom_begin(uint32_t era, uint32_t nr_blocks,
uint32_t nr_bits, uint32_t nr_set) = 0;
virtual void writeset_begin(uint32_t era, uint32_t nr_bits) = 0;
virtual void bit(uint32_t index, bool value) = 0;
virtual void bloom_end() = 0;
virtual void writeset_end() = 0;
};
}
typedef persistent_data::btree<1, era_detail_traits> bloom_tree;
typedef persistent_data::btree<1, era_detail_traits> writeset_tree;
void walk_bloom_tree(persistent_data::transaction_manager::ptr tm,
bloom_tree const &tree,
bloom_tree_detail::bloom_visitor &bloom_v,
bloom_tree_detail::damage_visitor &dv);
void walk_writeset_tree(persistent_data::transaction_manager::ptr tm,
writeset_tree const &tree,
writeset_tree_detail::writeset_visitor &writeset_v,
writeset_tree_detail::damage_visitor &dv);
void check_bloom_tree(persistent_data::transaction_manager::ptr tm,
bloom_tree const &tree,
bloom_tree_detail::damage_visitor &dv);
void check_writeset_tree(persistent_data::transaction_manager::ptr tm,
writeset_tree const &tree,
writeset_tree_detail::damage_visitor &dv);
}
//----------------------------------------------------------------

View File

@ -33,25 +33,23 @@ namespace {
out_ << "</superblock>" << endl;
}
void begin_bloom(uint32_t era, uint32_t nr_bits,
pd::block_address nr_blocks) {
void begin_writeset(uint32_t era, uint32_t nr_bits) {
indent();
out_ << "<bloom era=\"" << era << "\""
<< " nr_bits=\"" << nr_bits << "\""
<< " nr_blocks=\"" << nr_blocks << "\">" << endl;
out_ << "<writeset era=\"" << era << "\""
<< " nr_bits=\"" << nr_bits << "\">" << endl;
inc();
}
void bloom_bit(uint32_t bit, bool value) {
void writeset_bit(uint32_t bit, bool value) {
indent();
// FIXME: collect all the bits, then uuencode
out_ << "<bit bit=\"" << bit << "\" value=\"" << value << "\">" << endl;
}
void end_bloom() {
void end_writeset() {
dec();
indent();
out_ << "</bloom>" << endl;
out_ << "</writeset>" << endl;
}
void begin_era_array() {

View File

@ -343,7 +343,7 @@ namespace persistent_data {
unsigned max_concurrent_blocks,
typename block_io<BlockSize>::mode mode)
: io_(new block_io<BlockSize>(path, nr_blocks, mode)),
cache_(max(64u, max_concurrent_blocks)),
cache_(max(1024u, max_concurrent_blocks)),
tracker_(0, nr_blocks)
{
}

View File

@ -20,6 +20,7 @@
#include "persistent-data/transaction_manager.h"
#include "persistent-data/space-maps/core.h"
#include "persistent-data/data-structures/array.h"
#include "persistent-data/data-structures/simple_traits.h"
#include <vector>

View File

@ -40,17 +40,26 @@ namespace {
using namespace boost::random;
mt19937 rng;
uniform_int_distribution<uint64_t> uniform_dist(0, max);
while (r.size() < count) {
block_address b = uniform_dist(rng);
block_address b = uniform_dist(rng_);
r.insert(b);
}
return r;
}
set<block_address> generate_linear_blocks(unsigned count,
block_address max = std::numeric_limits<uint64_t>::max()) {
set<block_address> r;
for (unsigned i = 0; i < count; i++)
r.insert(i);
return r;
}
void commit() {
block_manager<>::write_ref superblock(bm_->superblock(SUPERBLOCK));
}
@ -65,6 +74,8 @@ namespace {
block_manager<>::ptr bm_;
space_map::ptr sm_;
transaction_manager::ptr tm_;
boost::random::mt19937 rng_;
};
}
@ -119,18 +130,75 @@ TEST_F(BloomFilterTests, reload_works)
}
}
TEST_F(BloomFilterTests, count_false_positives)
unsigned next_power(unsigned n)
{
block_address nr_blocks = 1024 * 1024;
block_address written_blocks = nr_blocks / 100;
unsigned r = 1;
while (r < n)
r <<= 1;
unsigned shift = 1;
return r;
}
while ((1ull << shift) < (16 * written_blocks))
shift++;
cerr << "bitset size: " << ((1 << shift) / (8 * 1024)) << "k" << endl;
unsigned calc_nr_bits(double false_positive_rate, unsigned dirty_blocks_per_era, unsigned nr_probes)
{
double k = (double) nr_probes;
double kth_root = exp(log(false_positive_rate) / k); // can be precomputed
bloom_filter f(tm_, 1 << shift, 6);
// FIXME: we need a way to calulate this in kernel? or should we
// just pass in the bloom params on the target line?
double tmp = log(1.0 - kth_root);
double n = (- k * (double) dirty_blocks_per_era) / tmp;
return next_power(ceil(n));
}
unsigned calc_m(double fp, unsigned nr_probes, unsigned n)
{
double k = (double) nr_probes;
double kth_root = exp(log(fp) / k);
double tmp = log(1.0 - kth_root);
double m = (- ((double) n) / k) * tmp;
return ceil(m);
}
void print_nr_bits_table(double fp, unsigned nr_probes)
{
cout << "fp = " << fp << ", k = " << nr_probes << endl;
for (unsigned long long m = 1024; m < (1ull << 25); m *= 2) {
unsigned n = calc_nr_bits(fp, m, nr_probes);
unsigned actual_m = calc_m(fp, nr_probes, n);
cout << " m = " << m << ", n = " << n << ", " << n / (1024 * 8)
<< "k, actual_m = " << actual_m << endl;
}
}
// Not really a test
TEST_F(BloomFilterTests, nr_bits_table)
{
print_nr_bits_table(0.001, 4);
print_nr_bits_table(0.001, 6);
print_nr_bits_table(0.001, 8);
print_nr_bits_table(0.001, 16);
}
TEST_F(BloomFilterTests, count_false_positives_with_random_inserts)
{
block_address nr_blocks = 1 << 27;
block_address written_blocks = nr_blocks / 1024;
unsigned nr_probes = 6;
unsigned n = calc_nr_bits(0.001, written_blocks, nr_probes);
cerr << "bitset size: " << (n / (8 * 1024)) << "k" << endl;
double ideal_k = log(2) * ((double) n / (double) written_blocks);
cerr << "Ideal k = " << ideal_k << endl;
bloom_filter f(tm_, n, nr_probes);
set<block_address> bs = generate_random_blocks(written_blocks, nr_blocks);
set<block_address>::const_iterator it;
@ -138,8 +206,6 @@ TEST_F(BloomFilterTests, count_false_positives)
for (it = bs.begin(); it != bs.end(); ++it)
f.set(*it);
// f.print_debug(cerr);
unsigned count = 0;
for (unsigned i = 0; i < nr_blocks; i++)
if (!bs.count(i) && f.test(i))
@ -150,4 +216,99 @@ TEST_F(BloomFilterTests, count_false_positives)
<< "%" << endl;
}
TEST_F(BloomFilterTests, count_false_positives_with_linear_inserts)
{
block_address nr_blocks = 1 << 25;
block_address written_blocks = nr_blocks / 100;
double fp = 0.001;
unsigned nr_probes = 6;
unsigned n = calc_nr_bits(fp, written_blocks, nr_probes);
cerr << "bitset size: " << (n / (8 * 1024)) << "k" << endl;
double ideal_k = log(2) * ((double) n / (double) written_blocks);
cerr << "Ideal k = " << ideal_k << endl;
bloom_filter f(tm_, n, nr_probes);
set<block_address> bs = generate_linear_blocks(written_blocks, nr_blocks);
set<block_address>::const_iterator it;
for (it = bs.begin(); it != bs.end(); ++it)
f.set(*it);
unsigned count = 0;
for (unsigned i = 0; i < nr_blocks; i++)
if (!bs.count(i) && f.test(i))
count++;
double actual_fp = static_cast<double>(count) / static_cast<double>(nr_blocks);
ASSERT_THAT(actual_fp, Lt(fp));
cerr << count << " false positives out of " << nr_blocks << ", "
<< actual_fp * 100.0 << "%" << endl;
}
TEST_F(BloomFilterTests, false_positives_over_multiple_eras)
{
unsigned nr_eras = 10;
block_address nr_blocks = 1 << 20;
block_address written_blocks = nr_blocks / nr_eras;
double fp = 0.001;
unsigned nr_probes = 6;
unsigned n = calc_nr_bits(fp, written_blocks, nr_probes);
cerr << "bitset size: " << (n / (8 * 1024)) << "k" << endl;
double ideal_k = log(2) * ((double) n / (double) written_blocks);
cerr << "Ideal k = " << ideal_k << endl;
vector<set<block_address> > writes(nr_eras);
vector<bloom_filter::ptr> filters(nr_eras);
for (unsigned era = 0; era < writes.size(); era++) {
cerr << "inserting era " << era << endl;
writes[era] = generate_random_blocks(written_blocks, nr_blocks);
set<block_address> const &bs = writes[era];
filters[era] = bloom_filter::ptr(new bloom_filter(tm_, n, nr_probes));
bloom_filter::ptr &f = filters[era];
set<block_address>::const_iterator it;
for (it = bs.begin(); it != bs.end(); ++it)
f->set(*it);
}
set<block_address> write_sum;
set<block_address> filter_sum;
for (unsigned era_plus_1 = writes.size(); era_plus_1 > 0; era_plus_1--) {
unsigned era = era_plus_1 - 1;
set<block_address> const &era_writes = writes[era];
write_sum.insert(era_writes.begin(), era_writes.end());
for (unsigned i = 0; i < nr_blocks; i++)
if (filters[era]->test(i))
filter_sum.insert(i);
unsigned count = 0;
for (unsigned i = 0; i < nr_blocks; i++) {
if (write_sum.count(i) > 0)
ASSERT_THAT(filter_sum.count(i), Gt(0ull));
else if (filter_sum.count(i))
count++;
}
cerr << "blocks >= era " << era << ", false positives = "
<< static_cast<double>(count * 100) / static_cast<double>(nr_blocks)
<< "%" << endl;
}
}
//----------------------------------------------------------------

View File

@ -5,6 +5,7 @@
#include "persistent-data/data-structures/btree.h"
#include "persistent-data/data-structures/btree_counter.h"
#include "persistent-data/space-maps/core.h"
#include "persistent-data/data-structures/simple_traits.h"
using namespace base;
using namespace std;

View File

@ -20,6 +20,7 @@
#include "persistent-data/transaction_manager.h"
#include "persistent-data/space-maps/core.h"
#include "persistent-data/data-structures/btree.h"
#include "persistent-data/data-structures/simple_traits.h"
using namespace std;
using namespace persistent_data;