* Clear votes which were created before spork15 activation
* Reject incoming votes which were created pre-DIP3
* Only use voting keys for VOTE_SIGNAL_FUNDING
The other vote signals are meant to be emitted by sentinel and must thus
be signed with the operator key.
* Simplify GetMinVoteTime
* Review suggestions/fixes
* Add missing mutex in CGovernanceObject::RemoveOldVotes
* Do not run scheduled DoMaintenance if shutdown was requested
* Don't dump cache files if the load sequence was interrupted
* Stop privatesend and release keys before wallet flush
* fix typo
* Implement RemoveAskFor to indicate that we're not interested in an item anymore
When an INV item is received from the first node, the item is requested
immediately. If the same item is received from another node, an entry is
added to mapAskFor which marks the item for re-requesting in case the first
node did not respond. When the item is received from the first node,
the item was previously never removed from mapAskFor. Only the later getdata
loop in SendMessages would then gradually remove items from the map. This
is quite delayed however as the entries in mapAskFor have a timeout value.
RemoveAskFor allows to remove all entries from mapAskFor and setAskFor
when we are not interested in the item anymore (e.g. because we received
it already).
* Call RemoveAskFor whenever we receive a message
* Only pass hash instead of CInv object to RemoveAskFor
* remove vector, extra loop in cleanup function
This commit removes 2 loops and a vector which I don't believe are necessary in
CMasternode::FlagGovernanceItemsAsDirty. I could be missing something, but
can't think of any good reason why this was implemented this way.
* use range operator to range over vectors
* remove deprecated wire message types
* mn: simplify govobj map mgmt a bit
* remove extra semicolons
* invert if/else condition and drop else
* remove if/else logic from Qt
This is the entire purpose of the Get<X>String methods on MNP class.
* Revert "remove deprecated wire message types"
This reverts commit 9de88a3fda999555c00cb829d60e322b2fd3800d.
* Revert "remove if/else logic from Qt"
This reverts commit c0f43c9b85274b13a3ac9625775817bc5b53bb9d.
* Split keyIDMasternode into keyIDOwner/keyIDOperator/keyIDVoting
keyIDOwner is the key used for things which should stay in control of the
collateral owner, like proposal voting.
keyIDOperator is the key used for operational things, like signing network
messages, signing trigger/watchdog objects and trigger votes.
keyIDVoting is the key used for proposal voting
Legacy masternodes will always have the same key for all 3 to keep
compatibility.
Using different keys is only allowed after spork15 activation.
* Forbid reusing collateral keys for operator/owner keys and vice versa
* Bump SERIALIZATION_VERSION_STRING in CMasternodeMan
* No need for pointers to gobjects, can simply use hashes in RequestGovernanceObjectVotes
* Drop `swap` and `=` in GovernanceObject
Were used for shuffling, no longer needed
* fix whitespace
* added zmq stuff for governance objects and votes
it seems that zmq is currently not in a working state, need to figure that out.
Need to:
plug in the new methods added
possibly plug in the old methods, as it doesn't look like they are.
* formatting fix. Will probably need to revert for this PR
* continue linking new zmq messages in
* added comment, might need to revert
* fixes error of it not knowing about the classes
* Actually link in, all new govobjects and govvotes should be broadcast over zmq now.
* fix compile error, forgot to change params when copying
* fix compile error
* add imports to the header files in zmqconfig.h
* fixing linking(i think) error
* Revert "added comment, might need to revert"
This reverts commit 2918ea40fe9a96834c4bd89e13cb458cde6814f2.
* Revert "formatting fix. Will probably need to revert for this PR"
This reverts commit ca10558866ab61e3dd0c70541fdcfee6f5115157.
* fix tabs etc
* EOL added
* optimization of hash.begin() @nmarley thoughts?
* remove formatting changes
* iterator i -> it and removal of notifier
* typo in df879f57
* use auto for the iterators
* introduce hash prefix
* implement changes in zmq_sub.py, update the doc, and change argument name to fix typo
* deref iterators before calling methods
* continued e8a4c505
* missed one... continued e8a4c505
* killing some tabs
* fix spacing for setting or comparing iterators
* change order of new variables to match current setup
* re-add elif's I didn't realize got removed
* Revert "fix spacing for setting or comparing iterators"
This reverts commit 8ce2068148dcd275ebba7ee6038d0db1c582b9f3.
* Revert "use auto for the iterators"
This reverts commit cb16cf0760bfaf68c56684877898611802bf2303.
* Revert "missed one... continued e8a4c505"
This reverts commit 2087cf894f7e9682508b4692b89897b4fa4e4b7a.
* Revert "continued e8a4c505"
This reverts commit a78c8ad2c9bb1602242a8f040e17ef958982348c.
* Revert "deref iterators before calling methods"
This reverts commit e8a4c505d1d34360eaf882d92cd8fbc55436cfcc.
* Revert "iterator i -> it and removal of notifier"
This reverts commit 29574248b1a0d05c18d60454d07c77979aae6fb2.
* Revert "fix whitespace"
This reverts commit 612be48d963000b4cbc5f64981a88a3225428b37.
* Revert "typo in df879f5"
* Revert "Optimization of hash.begin()"
* fixes problem with revert
* Udjin complain's a lot ;)
* help text, init.cpp
* add signals in validationinterface.cpp
* Change location of vote notification call.
* remain consistent
* remove unneeded include due to change of notification location
* implement raw notifications
* Mark an object for deletion when the corresponding trigger is deleted
* Mark objects for deletion if creation of corresponding triggers failed
* NULL -> nullptr
* Make sure pObj is not null
And send REJECT msg back to the peer we received the original message from.
Same logic as for MIN_PEER_PROTO_VERSION but using specific proto versions for each submodule.
* Avoid locking cs_main in CMasternode/Ping/Broadcast
Deligate this to CMasternodeMan and use AssertLockHeld instead
* Ensure consistent locking order (cs_main, cs_wallet, mempool.cs, cs_instantsend) in CInstantSend while avoiding potential deadlocks at the same time
* Add missing locks in wallet
* Add missing locks in governance
* Fix cs_vPendingMasternodes vs cs_main potential deadlock
SendVerifyRequest no longer opens connection directly, so no cs_main is needed here
* Drop nBudgetProposalEstablishingTime
* Refactor: replace `== COutPoint()` with `.IsNull()`
* Refactor: add `operator bool()` to CMasternodePing
* Refactor: actually use `operator bool()` for CPendingDsaRequest
* Refactor: fixing code style in TrafficGraphData
* Fix some comments and whitespaces
* Drop CGovernanceVote::GetTypeHash()
* Drop legacy X11 code
No longer used... if it ever was used at all.
* Move `<boost/foreach.hpp>` out of coins.h
* Simplify CMasternodeBlockPayees::GetRequiredPaymentsString()
Also less of boost::lexical_cast
* Drop CTxDSIn::nSentTimes
* Fix few warnings
* fix warning (timer)
* fix nit
* Drop BOOST_FOREACH in Dash-specific code (only), no changes in behavior
* Use references in loops (use const references where applicable) in Dash-specific code (only)
And adjust related code to make it compilable.
* Loop through nodes via connman.ForEachNode in dsq Relay instead of creating a copy
* CacheMap::Insert should not update existing item, should return `false`
* Use `emplace` operator instead of `[]` to add new items in Cache*Map implementation
* Use prefix cm/cmm for CacheMap/CacheMultiMap maps respectively to distinguish them from each other and from std::map-s
* Avoid excessive processing of already known valid votes
Also prettify debug log output a bit
* Drop nCurrentSize from Cache*Map classes
Use `size()` and `empty()` of std classes instead, they have constant complexity since c++11
* Do not create an explicit iterator if it's never used later (in Cache*Map)
* Do not prune last item in CacheMultiMapInsert when trying to insert a duplicate one
0729102 Net: pass interruptMsgProc as const where possible (Jorge Timón)
fc7f2ff Net: Make CNetMsgMaker more const (Jorge Timón)
d45955f Net: CConnman: Make some methods const (Jorge Timón)
And use this where needed. This avoids warnings about unset version fields
and is also generally a good thing (we really should only communicate with
nodes we know are good)
c7be56d net: push only raw data into CConnman (Cory Fields)
2ec935d net: add CVectorWriter and CNetMsgMaker (Cory Fields)
b7695c2 net: No need to check individually for disconnection anymore (Cory Fields)
fedea8a net: don't send any messages before handshake or after requested disconnect (Cory Fields)
d74e352 net: Set feelers to disconnect at the end of the version message (Cory Fields)
* Do not relay `mnp`, `mnb`, `mnw`, `govobj` and `govobjvote` until fully synced
This should massively lower outgoing traffic and load during sync process (especially on mainnet)
* Apply "no sync from me until I'm synced" rule to all nodes
* Make sure to clear setAskFor in Dash submodules
Thanks @sidhujag for finding the bug
* partially revert previous commit
* skip mnv until mn list is synced
This monstrous change eliminates all remaining uses of
g_connman global variable in Dash-specific code.
Unlike previous changes eliminating g_connman use
that were isolated to particular modules, this one covers
multiple modules simultaneously because they are so interdependent
that change in one module was quickly spreading to others.
This is mostly invariant change that was done by
* changing all functions using g_connman to use connman argument,
* changing all functions calling these functions to use connman argument,
* repeating previous step until there's nothing to change.
After multiple iterations, this process converged to final result,
producing code that is mostly equivalent to original one, but passing
CConnman instance through arguments instead of global variable.
The only exception to equivalence of resulting code is that I had to
create overload of CMasternodeMan::CheckAndRemove() method without arguments
that does nothing just for use in CFlatDB<CMasternodeMan>::Dump() and
CFlatDB<CMasternodeMan>::Load() methods.
Normal CMasternodeMan::CheckAndRemove() overload now has argument of
CConnman& type and is used everywhere else.
The normal overload has this code in the beginning:
if(!masternodeSync.IsMasternodeListSynced()) return;
Masternode list is not synced yet when we load "mncache.dat" file,
and we save "mncache.dat" file on shutdown, so I presume that it's OK
to use overload that does nothing in both cases.
Signed-off-by: Oleg Girko <ol@infoserver.lv>
* Store masternodes in a map instead of a vector, drop unused functions in CMasternodeMan
* CTxIn vin -> COutPoint outpoint
* do not use CMasternodeMan::Find outside of the class
* update GetMasternodeInfo
* safe version of GetNextMasternodeInQueueForPayment
* fix ProcessMasternodeConnections
* bump CMasternodeMan::SERIALIZATION_VERSION_STRING
* fixed an issue with MasternodeRateCheck always returns true
* additioanal fixes and refactoring (rebase)
* slightly improved CRateBuffer synchronization for newly started nodes
* Change sync process:
- IsBlockchainSynced(): drop CheckNodeHeight() and all complicated code, use fInitialDownload in UpdatedBlockTip() to switch initial states
- ProcessTick(): detect sleep mode like it was in IsBlockchainSynced(), not by number of masternodes
* Changes for sync in governance:
- do not keep sync alive on ConfirmInventoryRequest()
- skip some governance actions until we are synced to some level
* do not run CMasternodeMan::UpdateLastPaid() until winners list is synced
* start syncing mn list on the same node right after requesting sporks
* replace nTimeLast<Asset> with the unified nTimeLastBumped, bump on UpdatedBlockTip
* fix comments and LogPrintf-s
* remove excessive MASTERNODE_SYNC_IBD
* a bit more descriptive BumpAssetLastTime in few cases
