6c81dfad61
This commit fixes a race condition in the notifyBlockEpochs detected by the race condition detector. Previously the notifyBlockEpochs function could cause a race condition when a new caller was either cancelling an existing notification intent or creating a new one. We fix this issue by making the call to notifyBlockEpochs synchronous rather than asynchronous. An alternative would be to add a mutex guarding the map state. The channel itself is buffered with a good margin, so there shouldn’t be a huge impact. |
||
|---|---|---|
| .. | ||
| btcdnotify | ||
| interface_test.go | ||
| interface.go | ||
| log.go | ||
| README.md | ||
chainntnfs
[
]
(https://travis-ci.org/lightningnetwork/lnd)
[
]
(https://github.com/lightningnetwork/lnd/blob/master/LICENSE)
[
]
(http://godoc.org/github.com/lightningnetwork/lnd/chainntnfs)
The chainntnfs package implements a set of interfaces which allow callers to receive notifications in response to specific on-chain events. The set of notifications available include:
- Notifications for each new block connected to the current best chain.
- Notifications once a
txidhas reached a specified number of confirmations. - Notifications once a target outpoint (
txid:index) has been spent.
These notifications are used within lnd in order to properly handle the
workflows for: channel funding, cooperative channel closures, forced channel
closures, channel contract breaches, sweeping time-locked outputs, and finally
pruning the channel graph.
This package is intentionally general enough to be applicable outside the
specific use cases within lnd outlined above. The current sole concrete
implementation of the ChainNotifier interface depends on btcd.
Installation and Updating
$ go get -u github.com/lightningnetwork/lnd/chainntnfs