- Support i2cp.destination.sigType option in TunnelController and
I2PSocketManagerFactory
- Fixup of Destination.create() and Destination.size()
- Add generic off/len methods in DSAEngine, needed for streaming
- Fixup of sign/verify in streaming Packet
- Javadocs
of signing keys and signatures in RouterIdentities and Destinations.
Untested, not even for regressions, except with command line
using PrivateKeyFile.
Based on preliminary spec at http://zzz.i2p/topics/1442?page=1#p7524
Not done:
- Transport handshake signing
- Configuration of default type
- Specification of type in options to I2PSocketManagerFactory
- Specification of type in i2ptunnel
- Fix up caching of SigningPublicKey and P256 key cert
- Any non-default crypto type in the key cert
- Documentation
- Fix several 'increment of volatile is not atomic' all over
Remaining: UDP PeerState.java, to be checked in separately
- Comment out all of unused MessageStateMonitor
- Check blacklist/whitelist before connection limits, so
a blacklisted peer does not increment the counters
- Don't increment total throttle if peer is throttled
- Fix blacklist NPE after config change (ticket #1070)
- Make I2PSocketFull.close() nonblocking; it will now cause any user-side
writes blocked in I/O (Connection.packetSendChoke()) to throw
an exception (tickets #629, #1041)
- Don't ignore InterruptedExceptions; throw InterruptedIOException
- Back out static disconnect exception
- MessageInputStream locking fixes
- Cleanups
I2PSnark:
- Close socket before closing output stream to avoid blocking in
Peer.disconnect(), and prevent Peer.disconnect() loop
- Prevent multiple calls or reentrancy in disconnect() (ticket #1041)
- Implement processing of close to skip TIME-WAIT, and
wait for all packets to be acked (not just the CLOSE) before
doing so, if possible (ticket #1042)
- Don't call disconnect() or disconnectComplete() from I2PSocketFull.destroy()
so retransmissions and acks can still happen (removes some close loops)
- Don't call disconnect() until we have both sent and received a CLOSE (ticket #1040)
- Don't reset the connection from CPH just because we sent a CLOSE
and it was acked (ticket #1040)
- Ack packets even if we sent a CLOSE (ticket #1040)
- Retransmit CLOSE if not acked (ticket #1040)
- Send received packets to the MessageInputStream even if we haven't received a SYN
- Don't call MessageInputStream.messageReceived() for ack-only packets, that was pointless
- Don't send a RESET after timeout of an outbound connection
- Work around bugs on other end by limiting retransmission of CLOSE packets
- Make _isInbound final
- More cleanups, javadocs, log tweaks
- Cleanups
- Remove setCloseReceivedOn(), unused outside Connection
- OR the isFlagSet parameter instead of multiple calls
- Remove acked packets from _outboundPackets inside synced iterator
- Short-circuit _outboundPackets iterator if empty
- Small optimization if not logging in ConnectionPacketHandler
- Stub out processing of close ack (ticket #1042)
