This rearranges some of the function definitions that pertain to initial
peer version negotiation and bringup so they are more consistent with
the preferred order used throughout the codebase. In particular, the
functions are defined before they're first used and generally as close
as possible to the first use when they're defined in the same file.
There are no functional changes.
Backported from Decred.
In this commit, we patch a goroutine leak within the peer struct. This
goroutine leak can happen, if the remote side fails to actually finish
negotiation the protocol before our timeout ticker ticks. In this case,
the goroutine will be blocked on a send, as the channel is unfired. An
example trace from a btcd node I had on testnet showed:
```
3183 @ 0x42e78a 0x42e83e 0x40540b 0x4051a5 0x872f76 0x45bfc1
```
So, all instances of the goroutine failing to exit due to the remote
peer not finishing the p2p version negotiation handshake.
Our fix is simple: make the `negotiateErr' channel unbuffered. With this
simple change, we ensure that the goroutine will always exit even in the
case that the parent goroutine exists due to a timeout. # Please enter
the commit message for your changes. Lines starting
The cfilter BIP specifies that the filter type is a uint8. The
current code encodes it correctly on the wire, but everywhere else,
it's treated as a boolean (false for basic filter, true for
extended). This commit corrects that to account for possible
additional filter types in the future. All package changes are
done in one commit as they're all interdependent. The following
packages are updated:
* blockchain/indexers
* btcjson
* peer
* wire
* main (server.go and rpcserver.go)
This commit modifies the base peer struct to ascertain when a peer is
able to understand the new witness encoding, and specify the peer’s
supported encoding explicitly before/after the version handshake.
The btclog package has been changed to defining its own logging
interface (rather than seelog's) and provides a default implementation
for callers to use.
There are two primary advantages to the new logger implementation.
First, all log messages are created before the call returns. Compared
to seelog, this prevents data races when mutable variables are logged.
Second, the new logger does not implement any kind of artifical rate
limiting (what seelog refers to as "adaptive logging"). Log messages
are outputted as soon as possible and the application will appear to
perform much better when watching standard output.
Because log rotation is not a feature of the btclog logging
implementation, it is handled by the main package by importing a file
rotation package that provides an io.Reader interface for creating
output to a rotating file output. The rotator has been configured
with the same defaults that btcd previously used in the seelog config
(10MB file limits with maximum of 3 rolls) but now compresses newly
created roll files. Due to the high compressibility of log text, the
compressed files typically reduce to around 15-30% of the original
10MB file.
The github markdown interpreter has been changed such that it no longer
allows spaces in between the brackets and parenthesis of links and now
requires a newline in between anchors and other formatting. This
updates all of the markdown files accordingly.
While here, it also corrects a couple of inconsistencies in some of the
README.md files.
This changes the nonce generated to detect self connections over to use
pseudo randoms instead of a cryptographically random nonce.
There is really not a good reason for it to be cryptographically strong,
using the prng is much faster, and the prng also doesn't burn entropy.
This removes the field that tracks whether the version has been sent
since it is no longer used after the version negotiation was separated
from the main read and write code.
Replace assignments to individual fields of wire.NetAddress with
creating the entire object at once, as one would do if the type was
immutable.
In some places this replaces the creation of a NetAddress with a
high-precision timestamp with a call to a 'constructor' that converts
the timestamp to single second precision. For consistency, the tests
have also been changed to use single-precision timestamps.
Lastly, the number of allocations in readNetAddress have been reduced by
reading the services directly into the NetAddress instead of first into
a temporary variable.
This commit adds all of the infrastructure needed to support BIP0009
soft forks.
The following is an overview of the changes:
- Add new configuration options to the chaincfg package which allows the
rule deployments to be defined per chain
- Implement code to calculate the threshold state as required by BIP0009
- Use threshold state caches that are stored to the database in order
to accelerate startup time
- Remove caches that are invalid due to definition changes in the
params including additions, deletions, and changes to existing
entries
- Detect and warn when a new unknown rule is about to activate or has
been activated in the block connection code
- Detect and warn when 50% of the last 100 blocks have unexpected
versions.
- Remove the latest block version from wire since it no longer applies
- Add a version parameter to the wire.NewBlockHeader function since the
default is no longer available
- Update the miner block template generation code to use the calculated
block version based on the currently defined rule deployments and
their threshold states as of the previous block
- Add tests for new error type
- Add tests for threshold state cache
This modifies the error handling path in the peer read loop such that it
will no longer log an error when the error is io.ErrUnexpectedEOF. This
is being done because that error is almost always the result of a peer
being remotely disconnected and thus it isn't useful information to log.
However, since it might actually be due to a malformed message, a reject
message is still queued up to be sent back to the peer (which will
simply be discarded if the peer disconnected) before it is disconnected.
While it would be ideal to only log if it's not due to a disconnect, and
the code already attempts to handle that, it's not 100% possible to
detect upon the read returning an error without attempting to read again
which will not happen until the next loop iteration.
The inability for a peer to negotiate is not something that should be a
warning which implies something is wrong. On the contrary, it is quite
expected that various peers will connect (or be connected to) that are
unable to properly negotiate for a variety of reasons. One example would
be a peer that is too old.
Also, while here take care of a few style nits.
This adds a new field to the peer struct which stores the protocol
version advertised by the remote peer and updates the StatsSnapshot to
return the advertised version instead of the negotiated version.
