010eed3ce0 doc: warn that incoming conns are unlikely when not using default ports (Adam Jonas)
Pull request description:
Closes#5150.
This was mostly copied from #5285 by sulks, who has since quit GitHub.
The issue has remained open for 6 years, but the extra explanation still seems useful.
ACKs for top commit:
laanwj:
re-ACK 010eed3ce0
Tree-SHA512: d240fb06bba41ad8898ced59356c10adefc09f3abb33e277f8e2c5980b40678f2d237f286b476451bb29d2b94032a7dee2ada3b2efe004ed1c2509e70b48e40f
ea36a453e3 [net] Make p2p recv buffer timeout 20 minutes for all peers (John Newbery)
Pull request description:
The timeout interval for the send and recv buffers was changed from 90
minutes to 20 minutes in commit f1920e86 in 2013, except for peers that
did not support the pong message (where the recv buffer timeout remained
at 90 minutes). A few observations:
- for peers that support BIP 31 (pong messages), this recv buffer
timeout is almost redundant with the ping timeout. We send a ping
message every two minutes, and set a timeout of twenty minutes to
receive the pong response. If the recv buffer was really timing out,
then the pong response would also time out.
- BIP 31 is supported by all nodes of p2p version 60000 and higher, and
has been in widespread use since 2013. I'd be very surprised if there
are many nodes on the network that don't support pong messages.
- The recv buffer timeout is not specified in any p2p BIP. We're free to
set it at any value we want.
- A peer that doesn't support BIP 31 and hasn't sent any message to us
at all in 90 minutes is unlikely to be useful for us, and is more likely
to be evicted AttemptToEvictConnection() since it'll have the worst
possible ping time and isn't providing blocks/transactions.
Therefore, we remove this check, and set the recv buffer timeout to 20
minutes for all peers. This removes the final p2p version dependent
logic from the net layer, so all p2p version data can move into the
net_processing layer.
Alternative approaches:
- Set the recv buffer timeout to 90 minutes for all peers. This almost
wouldn't be a behaviour change at all (pre-BIP 31 peers would still
have the same recv buffer timeout, and we can't ever reach a recv buffer
timeout higher than 21 minutes for post-BIP31 peers, because the pong
timeout would be hit first).
- Stop supporting peers that don't support BIP 31. BIP 31 has been in
use since 2012, and implementing it is trivial.
ACKs for top commit:
MarcoFalke:
review ACK ea36a453e3
promag:
Code review ACK ea36a453e3.
practicalswift:
cr ACK ea36a453e3: patch looks correct
ajtowns:
ACK ea36a453e3
sipa:
utACK ea36a453e3
jonatack:
Code review ACK ea36a453e3
Tree-SHA512: df290bb32d2b5d9e59a0125bb215baa92787f9d01542a7437245f1c478c7f9b9831e5f170d3cd0db2811e1b11b857b3e8b2e03376476b8302148e480d81aab19
fa86217e97 doc: Move add relay comment in net to correct place (MarcoFalke)
Pull request description:
The comment was previously attached to `m_addr_known`, but now it is attached to `id`, which is wrong.
Fix that by moving the comment to `RelayAddrsWithConn`.
ACKs for top commit:
practicalswift:
cr ACK fa86217e97: patch looks correct
jnewbery:
ACK fa86217e97
theStack:
Code review ACK fa86217e97🌳
Tree-SHA512: ec3d5f1996aded38947d2a5fd0bb63539e88f83964cd3254984002edfd51abb4dde813c7c81619a8a3a5c55b7e9ae83c8c5be8ad6c84b4593ed3bbf463fe8979
The timeout interval for the send and recv buffers was changed from 90
minutes to 20 minutes in commit f1920e86 in 2013, except for peers that
did not support the pong message (where the recv buffer timeout remained
at 90 minutes). A few observations:
- for peers that support BIP 31 (pong messages), this recv buffer
timeout is almost redundant with the ping timeout. We send a ping
message every two minutes, and set a timeout of twenty minutes to
receive the pong response. If the recv buffer was really timing out,
then the pong response would also time out.
- BIP 31 is supported by all nodes of p2p version 60000 and higher, and
has been in widespread use since 2013. I'd be very surprised if there
are many nodes on the network that don't support pong messages.
- The recv buffer timeout is not specified in any p2p BIP. We're free to
set it at any value we want.
- A peer that doesn't support BIP 31 and hasn't sent any message to us
at all in 90 minutes is unlikely to be useful for us, and is more likely
to be evicted AttemptToEvictConnection() since it'll have the worst
possible ping time and isn't providing blocks/transactions.
Therefore, we remove this check, and sent the recv buffer timeout to 20
minutes for all peers. This removes the final p2p version dependent
logic from the net layer, so all p2p version data can move into the
net_processing layer.
Alternative approaches:
- Set the recv buffer timeout to 90 minutes for all peers. This almost
wouldn't be a behaviour change at all (pre-BIP 31 peers would still
have the same recv buffer timeout, and we can't ever reach a recv buffer
timeout higher than 21 minutes for post-BIP31 peers, because the pong
timeout would be hit first).
- Stop supporting peers that don't support BIP 31. BIP 31 has been in
use since 2012, and implementing it is trivial.
It's not actually possible to change this value, so remove the
indirection of it being a conn option.
DEFAULT_MAX_UPLOAD_TIMEFRAME is a compile time constant.
To make eclipse attacks more difficult, regularly initiate outbound connections
and stay connected long enough to sync headers and potentially learn of new
blocks. If we learn a new block, rotate out an existing block-relay peer in
favor of the new peer.
This augments the existing outbound peer rotation that exists -- currently we
make new full-relay connections when our tip is stale, which we disconnect
after waiting a small time to see if we learn a new block. As block-relay
connections use minimal bandwidth, we can make these connections regularly and
not just when our tip is stale.
Like feeler connections, these connections are not aggressive; whenever our
timer fires (once every 5 minutes on average), we'll try to initiate a new
block-relay connection as described, but if we fail to connect we just wait for
our timer to fire again before repeating with a new peer.
343dc4760f test: add test for high-bandwidth mode states in getpeerinfo (Sebastian Falbesoner)
dab6583307 doc: release note for new getpeerinfo fields "bip152_hb_{from,to}" (Sebastian Falbesoner)
a7ed00f8bb rpc: expose high-bandwidth mode states via getpeerinfo (Sebastian Falbesoner)
30bc8fab68 net: save high-bandwidth mode states in CNodeStats (Sebastian Falbesoner)
Pull request description:
Fixes#19676, "_For every peer expose through getpeerinfo RPC whether or not we selected them as HB peers, and whether or not they selected us as HB peers._" See [BIP152](https://github.com/bitcoin/bips/blob/master/bip-0152.mediawiki), in particular the [protocol flow diagram](https://github.com/bitcoin/bips/raw/master/bip-0152/protocol-flow.png). The newly introduced states are changed on the following places in the code:
* on reception of a `SENDCMPCT` message with valid version, the field `m_highbandwidth_from` is changed depending on the first integer parameter in the message (1=high bandwidth, 0=low bandwidth), i.e. it just mirrors the field `CNodeState.fPreferHeaderAndIDs`.
* after adding a `SENDCMPCT` message to the send queue, the field `m_highbandwidth_to` is changed depending on how the first integer parameter is set (same as above)
Note that after receiving `VERACK`, the node also sends `SENDCMPCT`, but that is only to announce the preferred version and never selects high-bandwidth mode, hence there is no need to change the state variables there, which are initialized to `false` anyways.
ACKs for top commit:
naumenkogs:
reACK 343dc4760f
jonatack:
re-ACK 343dc4760f per `git range-diff 7ea6499 4df1d12 343dc47`
Tree-SHA512: f4999e6a935266812c2259a9b5dc459710037d3c9e938006d282557cc225e56128f72965faffb207fc60c6531fab1206db976dd8729a69e8ca29d4835317b99f
fa5ed3b4ca net: Use Span in ReceiveMsgBytes (MarcoFalke)
Pull request description:
Pass a data pointer and a size as span in `ReceiveMsgBytes` to get the benefits of a span
ACKs for top commit:
jonatack:
ACK fa5ed3b4ca code review, rebased to current master 12a1c3ad1a, debug build, unit tests, ran bitcoind/-netinfo/getpeerinfo
theStack:
ACK fa5ed3b4ca
Tree-SHA512: 89bf111323148d6e6e50185ad20ab39f73ab3a58a27e46319e3a08bcf5dcf9d6aa84faff0fd6afb90cb892ac2f557a237c144560986063bc736a69ace353ab9d
Outbound peer logic prevents connecting to addresses that we're already
connected to, so prevent inadvertent eviction of current peers via
test-before-evict by checking this condition and marking current peer's
addresses as Good().
Co-authored-by: John Newbery <john@johnnewbery.com>
a490d074b3 doc: Add anchors.dat to files.md (Hennadii Stepanov)
0a85e5a7bc p2p: Try to connect to anchors once (Hennadii Stepanov)
5543c7ab28 p2p: Fix off-by-one error in fetching address loop (Hennadii Stepanov)
4170b46544 p2p: Integrate DumpAnchors() and ReadAnchors() into CConnman (Hennadii Stepanov)
bad16aff49 p2p: Add CConnman::GetCurrentBlockRelayOnlyConns() (Hennadii Stepanov)
c29272a157 p2p: Add ReadAnchors() (Hennadii Stepanov)
567008d2a0 p2p: Add DumpAnchors() (Hennadii Stepanov)
Pull request description:
This is an implementation of #17326:
- all (currently 2) outbound block-relay-only connections (#15759) are dumped to `anchors.dat` file
- on restart a node tries to connect to the addresses from `anchors.dat`
This PR prevents a type of eclipse attack when an attacker exploits a victim node restart to force it to connect to new, probably adversarial, peers.
ACKs for top commit:
jnewbery:
code review ACK a490d074b3
laanwj:
Code review ACK a490d074b3
Tree-SHA512: 0f5098a3882f2814be1aa21de308cd09e6654f4e7054b79f3cfeaf26bc02b814ca271497ed00018d199ee596a8cb9b126acee8b666a29e225b08eb2a49b02ddd
96571b3d4c doc: Update onion service target port numbers in tor.md (Hennadii Stepanov)
bb145c9050 net: Extend -bind config option with optional network type (Hennadii Stepanov)
92bd3c1da4 net, refactor: Move AddLocal call one level up (Hennadii Stepanov)
57f17e57c8 net: Pass onion service target to Tor controller (Hennadii Stepanov)
e3f07851f0 refactor: Rename TorController::target to m_tor_control_center (Hennadii Stepanov)
fdd3ae4d26 net, refactor: Refactor CBaseChainParams::RPCPort function (Hennadii Stepanov)
a5266d4546 net: Add alternative port for onion service (Hennadii Stepanov)
b3273cf403 net: Use network byte order for in_addr.s_addr (Hennadii Stepanov)
Pull request description:
This PR adds ability to label incoming Tor connections as different from normal localhost connections.
Closes#8973.
Closes#16693.
Default onion service target ports are:
- 8334 on mainnnet
- 18334 on testnet
- 38334 on signet
- 18445 on regtest
To set the onion service target socket manually the extended `-bind` config option could be used:
```
$ src/bitcoind -help | grep -A 6 -e '-bind'
-bind=<addr>[:<port>][=onion]
Bind to given address and always listen on it (default: 0.0.0.0). Use
[host]:port notation for IPv6. Append =onion to tag any incoming
connections to that address and port as incoming Tor connections
(default: 127.0.0.1:8334=onion, testnet: 127.0.0.1:18334=onion,
signet: 127.0.0.1:38334=onion, regtest: 127.0.0.1:18445=onion)
```
Since [pr19991.02 update](https://github.com/bitcoin/bitcoin/pull/19991#issuecomment-698882284) this PR is an alternative to #19043.
ACKs for top commit:
Sjors:
re-utACK 96571b3d4c
vasild:
ACK 96571b3d4
laanwj:
Re-ACK 96571b3d4c
Tree-SHA512: cb0eade80f4b3395f405f775e1b89c086a1f09d5a4464df6cb4faf808d9c2245474e1720b2b538f203f6c1996507f69b09f5a6e35ea42633c10e22bd733d4438
2ea62cae48 Improve docs about feeler connections (Gleb Naumenko)
Pull request description:
"feeler" and "test-before-evict" are two different strategies suggest in [Eclipse Attacks on Bitcoin’s Peer-to-Peer Network](https://www.usenix.org/system/files/conference/usenixsecurity15/sec15-paper-heilman.pdf). In our codebase, we use `ConnType::FEELER` to implement both.
It is confusing, up to the point that our documentation was just incorrect.
This PR:
- ~clarifies this aspect by renaming "ConnType::FEELER" to "ConnType::PROBE", meaning that this connections only probes that the node is operational, and then disconnects.~
- fixes the documentation
ACKs for top commit:
amitiuttarwar:
ACK 2ea62cae48. thank you!
practicalswift:
ACK 2ea62cae48
Tree-SHA512: c9c03c09eefeacec28ea199cc3f697b0a98723f2f849f7a8115edc43791f8165e296e0e25a82f0b5a4a781a7de38c8954b48bf74c714eba02cdc21f7460673e5
deb52711a1 Remove header checks out of net_processing (Troy Giorshev)
52d4ae46ab Give V1TransportDeserializer CChainParams& member (Troy Giorshev)
5bceef6b12 Change CMessageHeader Constructor (Troy Giorshev)
1ca20c1af8 Add doxygen comment for ReceiveMsgBytes (Troy Giorshev)
890b1d7c2b Move checksum check from net_processing to net (Troy Giorshev)
2716647ebf Give V1TransportDeserializer an m_node_id member (Troy Giorshev)
Pull request description:
Inspired by #15206 and #15197, this PR moves all message header verification from the message processing layer and into the network/transport layer.
In the previous PRs there is a change in behavior, where we would disconnect from peers upon a single failed checksum check. In various discussions there was concern over whether this was the right choice, and some expressed a desire to see how this would look if it was made to be a pure refactor.
For more context, see https://bitcoincore.reviews/15206.html#l-81.
This PR improves the separation between the p2p layers, helping improvements like [BIP324](https://github.com/bitcoin/bitcoin/pull/18242) and #18989.
ACKs for top commit:
ryanofsky:
Code review ACK deb52711a1 just rebase due to conflict on adjacent line
jnewbery:
Code review ACK deb52711a1.
Tree-SHA512: 1a3b7ae883b020cfee1bef968813e04df651ffdad9dd961a826bd80654f2c98676ce7f4721038a1b78d8790e4cebe8060419e3d8affc97ce2b9b4e4b72e6fa9f
This moves header size and netmagic checking out of net_processing and
into net. This check now runs in ReadHeader, so that net can exit early
out of receiving bytes from the peer. IsValid is now slimmed down, so
it no longer needs a MessageStartChars& parameter.
Additionally this removes the rest of the m_valid_* members from
CNetMessage.
This adds a CChainParams& member to V1TransportDeserializer member, and
use it in place of many Params() calls. In addition to reducing the
number of calls to a global, this removes a parameter from GetMessage
(and will later allow us to remove one from CMessageHeader::IsValid())
This commit removes the single-parameter contructor of CMessageHeader
and replaces it with a default constructor.
The single parameter contructor isn't used anywhere except for tests.
There is no reason to initialize a CMessageHeader with a particular
messagestart. This messagestart should always be replaced when
deserializing an actual message header so that we can run checks on it.
The default constructor initializes it to zero, just like the command
and checksum.
This also removes a parameter of a V1TransportDeserializer constructor,
as it was only used for this purpose.
This removes the m_valid_checksum member from CNetMessage. Instead,
GetMessage() returns an Optional.
Additionally, GetMessage() has been given an out parameter to be used to
hold error information. For now it is specifically a uint32_t used to
hold the raw size of the corrupt message.
The checksum check is now done in GetMessage.
This is intended to only be used for logging.
This will allow log messages in the following commits to keep recording
the peer's ID, even when logging is moved into V1TransportDeserializer.
In addition to adding more specificity to the log statement about the type of
connection, this change also consolidates two statements into one. Previously,
the second one should have never been hit, since block-relay connections would
match the "!IsInboundConn()" condition and return early.
