Use class template argument deduction for locks

This commit is contained in:
seelabs
2019-08-06 14:16:01 -07:00
parent 4076b6d92e
commit 5d1728cc96
104 changed files with 563 additions and 635 deletions

View File

@@ -129,7 +129,7 @@ PeerImp::run()
{
// Operations on closedLedgerHash_ and previousLedgerHash_ must be
// guarded by recentLock_.
std::lock_guard<std::mutex> sl(recentLock_);
std::lock_guard sl(recentLock_);
closedLedgerHash_ = hello_.ledgerclosed();
@@ -304,7 +304,7 @@ PeerImp::json()
}
{
std::lock_guard<std::mutex> sl (recentLock_);
std::lock_guard sl (recentLock_);
if (latency_)
ret[jss::latency] = static_cast<Json::UInt> (latency_->count());
}
@@ -337,7 +337,7 @@ PeerImp::json()
uint256 closedLedgerHash;
protocol::TMStatusChange last_status;
{
std::lock_guard<std::mutex> sl(recentLock_);
std::lock_guard sl(recentLock_);
closedLedgerHash = closedLedgerHash_;
last_status = last_status_;
}
@@ -384,7 +384,7 @@ bool
PeerImp::hasLedger (uint256 const& hash, std::uint32_t seq) const
{
{
std::lock_guard<std::mutex> sl(recentLock_);
std::lock_guard sl(recentLock_);
if ((seq != 0) && (seq >= minLedger_) && (seq <= maxLedger_) &&
(sanity_.load() == Sanity::sane))
return true;
@@ -401,7 +401,7 @@ void
PeerImp::ledgerRange (std::uint32_t& minSeq,
std::uint32_t& maxSeq) const
{
std::lock_guard<std::mutex> sl(recentLock_);
std::lock_guard sl(recentLock_);
minSeq = minLedger_;
maxSeq = maxLedger_;
@@ -410,7 +410,7 @@ PeerImp::ledgerRange (std::uint32_t& minSeq,
bool
PeerImp::hasShard (std::uint32_t shardIndex) const
{
std::lock_guard<std::mutex> l {shardInfoMutex_};
std::lock_guard l {shardInfoMutex_};
auto const it {shardInfo_.find(publicKey_)};
if (it != shardInfo_.end())
return boost::icl::contains(it->second.shardIndexes, shardIndex);
@@ -420,7 +420,7 @@ PeerImp::hasShard (std::uint32_t shardIndex) const
bool
PeerImp::hasTxSet (uint256 const& hash) const
{
std::lock_guard<std::mutex> sl(recentLock_);
std::lock_guard sl(recentLock_);
return std::find (recentTxSets_.begin(),
recentTxSets_.end(), hash) != recentTxSets_.end();
}
@@ -430,7 +430,7 @@ PeerImp::cycleStatus ()
{
// Operations on closedLedgerHash_ and previousLedgerHash_ must be
// guarded by recentLock_.
std::lock_guard<std::mutex> sl(recentLock_);
std::lock_guard sl(recentLock_);
previousLedgerHash_ = closedLedgerHash_;
closedLedgerHash_.zero ();
}
@@ -444,7 +444,7 @@ PeerImp::supportsVersion (int version)
bool
PeerImp::hasRange (std::uint32_t uMin, std::uint32_t uMax)
{
std::lock_guard<std::mutex> sl(recentLock_);
std::lock_guard sl(recentLock_);
return (sanity_ != Sanity::insane) &&
(uMin >= minLedger_) &&
(uMax <= maxLedger_);
@@ -511,7 +511,7 @@ PeerImp::fail(std::string const& name, error_code ec)
boost::optional<RangeSet<std::uint32_t>>
PeerImp::getShardIndexes() const
{
std::lock_guard<std::mutex> l {shardInfoMutex_};
std::lock_guard l {shardInfoMutex_};
auto it{shardInfo_.find(publicKey_)};
if (it != shardInfo_.end())
return it->second.shardIndexes;
@@ -521,7 +521,7 @@ PeerImp::getShardIndexes() const
boost::optional<hash_map<PublicKey, PeerImp::ShardInfo>>
PeerImp::getPeerShardInfo() const
{
std::lock_guard<std::mutex> l {shardInfoMutex_};
std::lock_guard l {shardInfoMutex_};
if (!shardInfo_.empty())
return shardInfo_;
return boost::none;
@@ -611,7 +611,7 @@ PeerImp::onTimer (error_code const& ec)
// Operations on lastPingSeq_, lastPingTime_, no_ping_, and latency_
// must be guarded by recentLock_.
{
std::lock_guard<std::mutex> sl(recentLock_);
std::lock_guard sl(recentLock_);
if (no_ping_++ >= Tuning::noPing)
{
failedNoPing = true;
@@ -714,7 +714,7 @@ void PeerImp::doAccept()
{
// Operations on closedLedgerHash_ and previousLedgerHash_ must be
// guarded by recentLock_.
std::lock_guard<std::mutex> sl(recentLock_);
std::lock_guard sl(recentLock_);
closedLedgerHash_ = hello_.ledgerclosed();
@@ -988,7 +988,7 @@ PeerImp::onMessage (std::shared_ptr <protocol::TMPing> const& m)
{
// Operations on lastPingSeq_, lastPingTime_, no_ping_, and latency_
// must be guarded by recentLock_.
std::lock_guard<std::mutex> sl(recentLock_);
std::lock_guard sl(recentLock_);
if (m->has_seq() && m->seq() == lastPingSeq_)
{
@@ -1324,7 +1324,7 @@ PeerImp::onMessage(std::shared_ptr <protocol::TMPeerShardInfo> const& m)
publicKey = publicKey_;
{
std::lock_guard<std::mutex> l {shardInfoMutex_};
std::lock_guard l {shardInfoMutex_};
auto it {shardInfo_.find(publicKey)};
if (it != shardInfo_.end())
{
@@ -1709,7 +1709,7 @@ PeerImp::onMessage (std::shared_ptr <protocol::TMStatusChange> const& m)
m->set_networktime (app_.timeKeeper().now().time_since_epoch().count());
{
std::lock_guard<std::mutex> sl(recentLock_);
std::lock_guard sl(recentLock_);
if (!last_status_.has_newstatus () || m->has_newstatus ())
last_status_ = *m;
else
@@ -1727,7 +1727,7 @@ PeerImp::onMessage (std::shared_ptr <protocol::TMStatusChange> const& m)
{
// Operations on closedLedgerHash_ and previousLedgerHash_ must be
// guarded by recentLock_.
std::lock_guard<std::mutex> sl(recentLock_);
std::lock_guard sl(recentLock_);
if (!closedLedgerHash_.isZero ())
{
outOfSync = true;
@@ -1750,7 +1750,7 @@ PeerImp::onMessage (std::shared_ptr <protocol::TMStatusChange> const& m)
{
// Operations on closedLedgerHash_ and previousLedgerHash_ must be
// guarded by recentLock_.
std::lock_guard<std::mutex> sl(recentLock_);
std::lock_guard sl(recentLock_);
if (peerChangedLedgers)
{
closedLedgerHash_ = m->ledgerhash();
@@ -1786,7 +1786,7 @@ PeerImp::onMessage (std::shared_ptr <protocol::TMStatusChange> const& m)
if (m->has_firstseq () && m->has_lastseq())
{
std::lock_guard<std::mutex> sl (recentLock_);
std::lock_guard sl (recentLock_);
minLedger_ = m->firstseq ();
maxLedger_ = m->lastseq ();
@@ -1856,7 +1856,7 @@ PeerImp::onMessage (std::shared_ptr <protocol::TMStatusChange> const& m)
{
uint256 closedLedgerHash {};
{
std::lock_guard<std::mutex> sl(recentLock_);
std::lock_guard sl(recentLock_);
closedLedgerHash = closedLedgerHash_;
}
j[jss::ledger_hash] = to_string (closedLedgerHash);
@@ -1886,7 +1886,7 @@ PeerImp::checkSanity (std::uint32_t validationSeq)
{
// Extract the seqeuence number of the highest
// ledger this peer has
std::lock_guard<std::mutex> sl (recentLock_);
std::lock_guard sl (recentLock_);
serverSeq = maxLedger_;
}
@@ -1912,7 +1912,7 @@ PeerImp::checkSanity (std::uint32_t seq1, std::uint32_t seq2)
if ((diff > Tuning::insaneLedgerLimit) && (sanity_.load() != Sanity::insane))
{
// The peer's ledger sequence is way off the validation's
std::lock_guard<std::mutex> sl(recentLock_);
std::lock_guard sl(recentLock_);
sanity_ = Sanity::insane;
insaneTime_ = clock_type::now();
@@ -1928,7 +1928,7 @@ void PeerImp::check ()
clock_type::time_point insaneTime;
{
std::lock_guard<std::mutex> sl(recentLock_);
std::lock_guard sl(recentLock_);
insaneTime = insaneTime_;
}
@@ -1968,7 +1968,7 @@ PeerImp::onMessage (std::shared_ptr <protocol::TMHaveTransactionSet> const& m)
if (m->status () == protocol::tsHAVE)
{
std::lock_guard<std::mutex> sl(recentLock_);
std::lock_guard sl(recentLock_);
if (std::find (recentTxSets_.begin (),
recentTxSets_.end (), hash) != recentTxSets_.end ())
@@ -2830,7 +2830,7 @@ PeerImp::getScore (bool haveItem) const
boost::optional<std::chrono::milliseconds> latency;
{
std::lock_guard<std::mutex> sl (recentLock_);
std::lock_guard sl (recentLock_);
latency = latency_;
}
@@ -2845,7 +2845,7 @@ PeerImp::getScore (bool haveItem) const
bool
PeerImp::isHighLatency() const
{
std::lock_guard<std::mutex> sl (recentLock_);
std::lock_guard sl (recentLock_);
return latency_ >= Tuning::peerHighLatency;
}