This commit updates the version of btcwallet that and is pinned against
to point to a version that includes a bug fix that was noticed in the
latest upstream PR’s we’ve included. The culprit bug would attempt to
create a write transaction inside of a greater read transaction which
would cause boltdb to block indefinitely internally. roasbeef’s fork of
btcwallet has been updated to include this fix.
The prior methods we employed to handle persistent connections could
result in the following situation: both peers come up, and
_concurrently_ establish connection to each other. With the prior
logic, at this point, both connections would be terminated as each peer
would go to kill the connection of the other peer. In order to resolve
this issue in this commit, we’ve re-written the way we handle
persistent connections.
The eliminate the issue described above, in the case of concurrent peer
connection, we now use a deterministic method to decide _which_
connection should be closed. The following rule governs which
connection should be closed: the connection of the peer with the
“smaller” public key should be closed. With this rule we now avoid the
issue described above.
Additionally, each peer now gains a peerTerminationWatcher which waits
until a peer has been disconnected, and then cleans up all resources
allocated to the peer, notifies relevant sub-systems of its demise, and
finally handles re-connecting to the peer if it's persistent. This
replaces the goroutine that was spawned in the old version of
peer.Disconnect().
This commit adds a new method to the peer struct: WaitForDisconnect().
This method is put in place to be used by wallers to synchronize the
ending of a peer’s lifetime. A follow up commit will utilize this new
method to re-write the way we handle persistent peer connections.
This commit modifies the way we go about unlocking the wallet. With the
latest changes to the API of btcwallet, we can on longer directly
access the waddrmgr struct. As a result, we’re now forced to go
_directly_ via the wallet to unlock the waddrmgr. The root
LightingWallet has been modified to not request the root key until we
finish starting the underlying wallet, so we can unlock the wallet in
the Start() method.
This commit modifies the initialization logic of the LightningWallet to
fetch the root key during startup rather than during creation. We make
this change in order to give enough time for the underlying
WalletController to properly boot up before we ask it to do any work.
We now pin gRPC against a particular commit version as it seems that
glide has some trouble properly resolving the semantic versioning
constraints for the latest version of gRPC.
This commit updates the version of btcwallet to the newest version
available in roasbeef’s fork. This new version includes the following
changes:
* a fix for the ping/pong deadlock issue with an expiring session
* and a preliminary merging of r btcsuite/btcwallet#469 into
roasbeef’s fork
The first change should solve an issue of lnd’s internal wallet
(btcwallet) being disconnected from the local btcd node. And the second
change should improve the reliability/correctness of the wallet as the
wtxmgr (tx/utxo store) and the waddrmgr (key store) are now updated
under a _single_ database transaction.
This commit modifies both readMessage and writeMessage to be further
message oriented. This means that message will be read and written _as
a whole_ rather than piece wise. This also fixes two bugs: the
readHandler could be blocked due to an sync read, and the writeHandler
would unnecessarily chunk up wire messages into distinct crypto
messages rather than writing it in one swoop.
Also with these series of changes, we’re now able to properly parse
messages that have been padded out with additional data as is allowed
by the current specification draft.
This commit modifies ReadMessage to no longer return the total bytes
read as this value will now be calculated at a higher level. The
io.Reader that’s passed to ReadMessage is expected to contain the
_entire_ message rather than be a pointer into a stream that contains
the message itself.
This commit adds a new message to the brontide.Conn struct which allows
callers to read an _entire_ message from the stream. As defined now,
brontide is a message crypto messaging protocol. Previously the only
method that allowed callers to read attempted to hide this feature with
a stream-like abstraction. However, having this as the sole interface
is at odds with the message oriented Lightning wire protocol, and isn’t
sufficient to allow parsing messages that have been padded as is
allowed by the protocol.
This new ReadNextMessage is intended to be used by higher level systems
which implement the Lightning p2p protocol.
This commit does away with all the old manual message equality tests
and replace it with a single property-based test that uses the
testing/quick package. This test uses a single scenario which MUST hold
for all the messages type and all possible messages generated for those
types. As a result we are able to do away with all the prior manually
generated test data as the fuzzer to scan the input space looking for a
message that violates the scenario.
This commit abandons our old bitcoin inspired message header and
replaces it with the bare type-only message headers that’s currently
used within the draft specification. As a result the message header now
consists of only 2-bytes for the message type, then actual payload
itself. With this change, the daemon will now need to switch to a
purely message based wire protocol in order to be able to handle the
extra data that can be extended to arbitrary messages.
This commit fixes a bug lingering in the decoding of the feature
vectors which was masked by the prior method of reading the _entire_
message from the stream before parsing it. The issue was that
performing a zero-byte Read on an io.Reader that’s purely streaming
will result in an indefinite block. We fix this bug by properly using
io.ReadFull in this context.
This commit modifies the Message interface to convert the Command
method to a MsgType method that uses a new set of message type for all
the defined messages. These new messages types nearly exactly match the
message types used within the current draft of the BOLT specifications.
This commit revues the Validate method from the Message interface as
the method is no longer used and is a relic from an older version of
the codebase.
This commit is tied to the prior commit and it patches up a lingering
race condition and deadlock that can arise due to now properly waiting
for all goroutine to exit before concluding the shutdown process.
This commit fixes a race condition detected by the race condition
detector that can be triggered by the lightnignNode exiting while the
underlying node is still active. Since the wait group wasn’t tied to
this cog routine, when the main process was exiting, it wouldn’t also
wait for this grouting to exit, thus triggering a race condition of
modifying the channel reference while reading for it.
The fix for this is straightforward: we now ensure that the goroutine
is factored into the struct level wait group.
This commit implements the new ping/pong messages along with their new
behavior. The new set of ping/pong messages allow clients to generate
fake cover traffic as the ping messages tells the pong message how many
bytes to included and can also be padded itself.
This commit fixes a deadlock bug within the readHandler of the peer.
Previously, once a channel was pending opening, _no_ other message
would be processed by the readHandler as it would be blocked waiting
for the channel to open. On testnet this would be manifsted as a node
locking up, until the channel was detected as being open.
We fix this bug by tracking which channel streams are active. If a
channel stream is active, then we can send the update directly to it.
Otherwise, we launch a goroutine that’ll block until the channel is
open, then in a synchronized manner, update the channel stream as being
active and send the update to the channel.
This commit modifies the order of operations after a channel has been
detected as fully opened. Rather than waiting until we receive the
FundingLocked message to announce the channel, we now do so immediately
after we detect the channel has been opened.
Additionally, we also patch a possible dead-lock bug by ensuring that
the RevokeAndAck message is always sent _after_ the FundingLocked
message. We do this by only passing the newly created channel too the
relevant sub-systems _after_ we receive the FundingLocked message. This
movement also serves to pave the way for our switch to the spec’s
funding workflow, as once we remove the initial revocation window,
there’ll be no way for us to initiate a state transition until we
receive the FundingLocked message from the remote party.
