This fixes one of the last major layer violations in the networking stack.
The network side is no longer in charge of message serialization, so it is now
decoupled from Bitcoin structures. Only the header is serialized and attached
to the payload.
CVectorWriter is useful for overwriting or appending an existing byte vector.
CNetMsgMaker is a shortcut for creating messages on-the-fly which are suitable
for pushing to CConnman.
Also, send reject messages earlier in SendMessages(), so that disconnections are
processed earlier.
These changes combined should ensure that no message is ever sent after
fDisconnect is set.
This way we're not relying on messages going out after fDisconnect has been
set.
This should not cause any real behavioral changes, though feelers should
arguably disconnect earlier in the process. That can be addressed in a later
functional change.
Make splash screen queue its own deletion when it receives the finished
command, instead of relying on WA_DeleteOnClose which doesn't work under
these circumstances.
Make ownership of the QThread object clear, so that the RPCConsole
can wait for the executor thread to quit before shutdown is called. This
increases overall thread safety, and prevents some objects from leaking
on exit.
None of these are very serious, and are leaks in objects that are
created at most one time.
In most cases this means properly using the QObject parent hierarchy,
except for BanTablePriv/PeerTablePriv which are not QObject,
so use a std::unique_ptr instead.
Make sure that the count is a zero modulo the new mask before
scaling, otherwise the next time until a measure triggers
will take only 1/2 as long as accounted for. This caused
the 'min time' to be potentially off by as much as 100%.
When a BIP152 HB-mode peer is in the least preferred position and
disconnects, they will not be by ForNode on the next loop. They
will continue to sit in that position and prevent deactivating
HB mode for peers that are still connected.
There is no reason for them to stay in the list if already gone,
so drop the first element unconditionally if there are too many.
Fixes issue #9163
- Do sorting for date, amount and confirmations column as longlong, not
unsigned longlong.
- Use `UserRole` to store our own data. This makes it treated as
ancillary data prevents it from being displayed.
- Get rid of `getMappedColumn` `strPad` - these are no longer necessary.
- Get rid of hidden `_INT64` columns.
- Start enumeration from 0 (otherwise values are undefined).
Make a copy of the boost time-point to wait for, otherwise the head of
the queue may be deleted by another thread while this one is waiting,
while the boost function still has a reference to it.
Although this problem is in non-test code, this is not an actual problem
outside of the tests because we use the thread scheduler with only one
service thread, so there will never be threads fighting at the head of
the queue.
The old boost fallback escapes this problem because it passes a scalar
value to wait_until instead of a const object reference.
Found by running the tests in LLVM-4.0-master asan.
This further decouples "main" and "net" processing logic by moving
orphan processing out of the chain-connecting cs_main lock and
into its own cs_main lock, beside all of the other chain callbacks.
Once further decoupling of net and main processing logic occurs,
orphan handing should move to its own lock, out of cs_main.
Note that this will introduce a race if there are any cases where
we assume the orphan map to be consistent with the current chain
tip, however I am confident there is no such case (ATMP will fail
without DoS score in all such cases).