This patch improves performance and resource usage around IP
addresses that are banned for misbehavior. They're already not
actually banned, as connections from them are still allowed,
but they are preferred for eviction if the inbound connection
slots are full.
Stop treating these like manually banned IP ranges, and instead
just keep them in a rolling Bloom filter of misbehaving nodes,
which isn't persisted to disk or exposed through the ban
framework. The effect remains the same: preferred for eviction,
avoided for outgoing connections, and not relayed to other peers.
Also change the name of this mechanism to better reflect reality;
they're not banned, just discouraged.
Contains release notes and several interface improvements by
John Newbery.
fa525e4d1c net: Avoid wasting inv traffic during IBD (MarcoFalke)
fa06d7e934 refactor: block import implies IsInitialBlockDownload (MarcoFalke)
faba65e696 Add ChainstateManager::ActiveChainstate (MarcoFalke)
fabf3d64ff test: Add FeeFilterRounder test (MarcoFalke)
Pull request description:
Tx-inv messages are ignored during IBD, so it would be nice if we told peers to not send them in the first place. Do that by sending two `feefilter` messages: One when the connection is made (and the node is in IBD), and another one when the node leaves IBD.
ACKs for top commit:
jamesob:
ACK fa525e4d1c ([`jamesob/ackr/19204.1.MarcoFalke.p2p_reduce_inv_traffic_d`](https://github.com/jamesob/bitcoin/tree/ackr/19204.1.MarcoFalke.p2p_reduce_inv_traffic_d))
naumenkogs:
utACK fa525e4
gzhao408:
ACK fa525e4d1c
jonatack:
re-ACK fa525e4 checked diff `git range-diff 19612ca fa8a66c fa525e4`, re-reviewed, ran tests, ran a custom p2p IBD behavior test at 9321e0f223.
hebasto:
re-ACK fa525e4d1c, only rebased since the [previous](https://github.com/bitcoin/bitcoin/pull/19204#pullrequestreview-429519667) review (verified with `git range-diff`).
Tree-SHA512: 2c22a5def9822396fca45d808b165b636f1143c4bdb2eaa5c7e977f1f18e8b10c86d4c180da488def38416cf3076a26de15014dfd4d86b2a7e5af88c74afb8eb
56010f9256 test: hoist p2p values to test framework constants (Jon Atack)
75447f0893 test: improve msg sends and p2p disconnections in p2p_invalid_messages (Jon Atack)
57960192a5 test: refactor test_large_inv() into 3 tests with common method (Jon Atack)
e2b21d8a59 test: add p2p_invalid_messages logging (Jon Atack)
9fa494dc09 net: update misbehavior logging for oversized messages (Jon Atack)
Pull request description:
...seen while reviewing #19264, #19252, #19304 and #19107:
in `net_processing.cpp`
- make the debug logging for oversized message size misbehavior the same for `addr`, `getdata`, `headers` and `inv` messages
in `p2p_invalid_messages`
- add missing logging
- improve assertions/message sends, move cleanup disconnections outside the assertion scopes
- split a slowish 3-part test into 3 order-independent tests
- add a few p2p constants to the test framework
ACKs for top commit:
troygiorshev:
reACK 56010f9256
MarcoFalke:
ACK 56010f9256 🎛
Tree-SHA512: db67b70278f8d4c318907e105af54b54eb3afd15500f9aa0c98034f6fd4bd1cf9ad1663037bd9b237ff4890f3059b37291a6498d8d6ae2cc38efb9f045f73310
fa1904e5f0 net: Remove dead logging code (MarcoFalke)
fac12ebf4f net: Avoid redundant and confusing FAILED log (MarcoFalke)
Pull request description:
Remove a redundant and confusing "FAILED" log message and gets rid of the unused return type in `ProcessMessage`
ACKs for top commit:
jnewbery:
utACK fa1904e5f0
gzhao408:
utACK fa1904e5f0
troygiorshev:
ACK fa1904e5f0
naumenkogs:
utACK fa1904e
Tree-SHA512: bfa553d5efa022727ed17877fb7c08c14849d804fe6d6a7ce172d513857beba35de41ea40b27ff1aedf68b81e2cda7b2a948ac985fcaaf1b6cfb96cce4837c90
PushInventory() is currently called with a CInv object, which can be a
MSG_TX or MSG_BLOCK. PushInventory() only uses the type to determine
whether to add the hash to setInventoryTxToSend or
vInventoryBlockToSend.
Since the caller always knows what type of inventory they're pushing,
the CInv is wastefully constructed and thrown away, and tx/block relay
is being split out, we split the function into PushTxInventory() and
PushBlockInventory().
Every `return false` is preceeded by a detailed debug log message to
explain that a disconnect or misbehavior happened. Logging another
generic "FAILED" message seems redundant.
Also, the size of the message and the message type has already been
logged and is thus redundant as well.
Finally, claiming that message processing FAILED seems odd, because the
message was fully processed to the point where it was concluded that the
peer should be either disconnected or marked as misbehaving.
-Increasing the banscore and/or banning is too harsh,
just disconnecting is enough.
-Return true from ProcessMessage because we already log
receipt of filterclear and disconnect.
-nodes not serving bloomfilters should disconnect peers
that send filterclear, just like filteradd and filterload
-nodes that want to enable/disable txrelay should use
feefilter
9e36067d8c [test] Add test for cfilters. (Jim Posen)
11106a4722 [net processing] Message handling for getcfilters. (Jim Posen)
e535670726 [indexes] Fix default [de]serialization of BlockFilter. (Jim Posen)
bb911ae7f5 [refactor] Pass CNode and CConnman by reference (John Newbery)
Pull request description:
Support `getcfilters` requests when `-peerblockfilters` is set.
Does not advertise compact filter support in version messages.
ACKs for top commit:
Empact:
re-Code Review ACK 9e36067d8c
MarcoFalke:
re-ACK 9e36067d8c , only change is adding commit "[refactor] Pass CNode and CConnman by reference" 🥑
jkczyz:
ACK 9e36067d8c
fjahr:
Code review ACK 9e36067d8c
Tree-SHA512: b45b42a25905ef0bd9e195029185300c86856c87f78cbe17921f4a25e159ae0f6f003e61714fa43779017eb97cd89d3568419be88e47d19dc8095562939e7887
9e1cb1adf1 [trivial/doc] Fix comment type (Amiti Uttarwar)
8f30260a67 [doc] Update unbroadcast description in RPC results (Amiti Uttarwar)
750456d6f2 [trivial] Remove misleading 'const' (Amiti Uttarwar)
fa32e676e5 [test] Manage node connections better in mempool persist test (Amiti Uttarwar)
1f94bb0c74 [doc] Provide rationale for randomization in scheduling. (Amiti Uttarwar)
9c8a55d9cb [mempool] Don't throw expected error message when upgrading (Amiti Uttarwar)
ba54983182 [test] Test that wallet transactions aren't rebroadcast before 12 hours (Amiti Uttarwar)
00d44a534b [test] P2P connection behavior should meet expectations (Amiti Uttarwar)
bd093ca15d [test] updates to unbroadcast test (Amiti Uttarwar)
dab298d9ab [docs] add release notes (Amiti Uttarwar)
Pull request description:
This PR is a follow up to #18038 which introduced the idea of an unbroadcast set & focuses mostly on documentation updates and test fixes. One small functionality update to not throw an expected error in `LoadMempool` when you upgrade software versions.
#18895 is another follow up to that addresses other functionality updates.
Background context:
The unbroadcast set is a mechanism for the mempool to track locally submitted transactions (via wallet or RPC). The node does a best-effort of delivering the transactions to the network via retries every 10-15 minutes until either a `GETDATA` is received or the transaction is removed from the mempool.
ACKs for top commit:
MarcoFalke:
ACK 9e1cb1adf1 👁
gzhao408:
ACK [`9e1cb1a`](9e1cb1adf1)
Tree-SHA512: 0cd51c4ca368b9dce92d50d73ec6e9df278a259e609eef2858f24cb8595ad07acc3db781d9eb0c351715f18fca5a2b4526838981fdb34a522427e9dc868bdaa6
5308c97cca [test] Add test for cfheaders (Jim Posen)
f6b58c1506 [net processing] Message handling for getcfheaders. (Jim Posen)
3bdc7c2d39 [doc] Add comment for m_headers_cache (John Newbery)
Pull request description:
Support `getcfheaders` requests when `-peerblockfilters` is set.
Does not advertise compact filter support in version messages.
ACKs for top commit:
jkczyz:
ACK 5308c97cca
MarcoFalke:
re-ACK 5308c97cca , only change is doc related 🗂
theStack:
ACK 5308c97cca🚀
Tree-SHA512: 240fc654f6f634c191d9f7628b6c4801f87ed514a1dd55c7de5d454d4012d1c09509a2d5a246bc7da445cd920252b4cd56a493c060cdb207b04af4ffe53b95f7
fab6b9d18f validation: Mark g_chainman DEPRECATED (MarcoFalke)
fa1d97b256 validation: Make ProcessNewBlock*() members of ChainstateManager (MarcoFalke)
fa24d49098 validation: Make PruneOneBlockFile() a member of ChainstateManager (MarcoFalke)
fa84b1cd84 validation: Make LoadBlockIndex() a member of ChainstateManager (MarcoFalke)
fa05fdf0f1 net: Pass chainman into PeerLogicValidation (MarcoFalke)
fa7b626d7a node: Add chainman alias for g_chainman (MarcoFalke)
Pull request description:
The global `g_chainman` has recently been introduced in #17737. The chainstate manager is primarily needed for the assumeutxo feature, but it can also simplify testing in the future.
The goal of this pull is to make the global chainstate manager internal to validation, so that all external code does not depend on globals and that unit or fuzz tests can pass in their (potentially mocked) chainstate manager.
I suggest reviewing the pull request commit-by-commit. It should be relatively straightforward refactoring that does not change behavior at all.
ACKs for top commit:
ryanofsky:
Code review ACK fab6b9d18f. Had to be rebased but still looks good
Tree-SHA512: dcbf114aeef4f8320d466369769f22ce4dd8f46a846870354df176c3de9ff17c64630fbd777e7121d7470d7a8564ed8d37b77168746e8df7489c6877e55d7b4f
651f1d816f [test] wait for inital broadcast before comparing mempool entries (gzhao408)
9d3f7eb986 [mempool] sanity check that all unbroadcast txns are in mempool (gzhao408)
a7ebe48b94 [rpc] add unbroadcast info to mempool entries and getmempoolinfo (gzhao408)
d160069604 [wallet] remove nLastResend logic (gzhao408)
Pull request description:
Followup to #18038 by amitiuttarwar which introduces the unbroadcast set: "a mechanism for the mempool to track locally submitted transactions" and decreases the frequency of rebroadcast from 10-15 minutes to 12-36 hours.
This PR addresses some of the outstanding TODOs building on top of it:
- remove `nLastResend` logic, which is used to ensure rebroadcast doesn't happen again if no new block has been mined (makes sense in 10-15 min period, but not necessary for 12-36 hour period). (#18038 [comment](https://github.com/bitcoin/bitcoin/pull/18038#discussion_r416826914))
- expose unbroadcast info via RPCs, for more informative queries and testing (#18038 [comment](https://github.com/bitcoin/bitcoin/pull/18038#discussion_r416837980))
- add sanity check to verify unbroadcast transactions still exist in mempool before rebroadcasting (#18038 [comment](https://github.com/bitcoin/bitcoin/pull/18038#discussion_r416861609))
ACKs for top commit:
naumenkogs:
Code review ACK 651f1d816f
amitiuttarwar:
ACK 651f1d816f🎉
MarcoFalke:
Review ACK 651f1d816f
Tree-SHA512: d5327e95ef39d44152b48df5c610502ae11c168f43dbbfb2885340c93d1ba9426eb3a5794573f5fc843502109cb3ffb63efa3f2db4f8f112efcde8f76d9a8845
0187d4c118 [indexes] Add compact block filter headers cache (John Newbery)
Pull request description:
Cache block filter headers at heights of multiples of 1000 in memory.
Block filter headers at height 1000x are checkpointed, and will be the most frequently requested. Cache them in memory to avoid costly disk reads.
ACKs for top commit:
jkczyz:
ACK 0187d4c118
theStack:
ACK 0187d4c118🎉
fjahr:
re-utACK 0187d4c118
laanwj:
code review ACK 0187d4c118
ariard:
Code Review ACK 0187d4c.
Tree-SHA512: 2075ae36901ebcdc4a217eae5203ebc8582181a0831fb7a53a119f031c46bca960a610a38a3d0636a9a405f713efcf4200c85f10c8559fd80139036d89473c56
0ea5d70b47 Updated comment for the condition where a transaction relay is denied (glowang)
be01449cc8 Add test for param interaction b/w -blocksonly and -whitelistforcerelay (glowang)
Pull request description:
Related to: #18428
When -blocksonly is turned on, a node would still relay transactions from whitelisted peers. This funcitonality has not been tested.
ACKs for top commit:
MarcoFalke:
ACK 0ea5d70b47
Tree-SHA512: 4e99c88281cb518cc67f5f3be7171a7b413933047b5d24a04bb3ff2210a82e914d69079f64cd5bac9206ec435e21a622c8e69cedbc2ccb39d2328ac5c01668e5
- before reattempting broadcast for unbroadcast txns, check they are in mempool and remove if not
- this protects from memory leaks and network spam just in case unbroadcast set (incorrectly) has extra txns
- check that tx is in mempool before adding to unbroadcast set to try to prevent this from happening
2896c412fa Do not answer GETDATA for to-be-announced tx (Pieter Wuille)
f2f32a3dee Push down use of cs_main into FindTxForGetData (Pieter Wuille)
c6131bf407 Abstract logic to determine whether to answer tx GETDATA (Pieter Wuille)
Pull request description:
This PR intends to improve transaction-origin privacy.
In general, we should try to not leak information about what transactions we have (recently) learned about before deciding to announce them to our peers. There is a controlled transaction dissemination process that reveals our transactions to peers that has various safeguards for privacy (it's rate-limited, delayed & batched, deterministically sorted, ...), and ideally there is no way to test which transactions we have before that controlled process reveals them. The handling of the `mempool` BIP35 message has protections in this regard as well, as it would be an obvious way to bypass these protections (handled asynchronously after a delay, also deterministically sorted).
However, currently, if we receive a GETDATA for a transaction that we have not yet announced to the requester, we will still respond to it if it was announced to *some* other peer already (because it needs to be in `mapRelay`, which only happens on the first announcement). This is a slight privacy leak.
Thankfully, this seems easy to solve: `setInventontoryTxToSend` keeps track of the txids we have yet to announce to a peer - which almost(*) exactly corresponds to the transactions we know of that we haven't revealed to that peer. By checking whether a txid is in that set before responding to a GETDATA, we can filter these out.
(*) Locally resubmitted or rebroadcasted transactions may end up in setInventoryTxToSend while the peer already knows we have them, which could result in us incorrectly claiming we don't have such transactions if coincidentally requested right after we schedule reannouncing them, but before they're actually INVed. This is made even harder by the fact that filterInventoryKnown will generally keep known reannouncements out of setInventoryTxToSend unless it overflows (which needs 50000 INVs in either direction before it happens).
The condition for responding now becomes:
```
(not in setInventoryTxToSend) AND
(
(in relay map) OR
(
(in mempool) AND
(old enough that it could have expired from relay map) AND
(older than our last getmempool response)
)
)
```
ACKs for top commit:
naumenkogs:
utACK 2896c41
ajtowns:
ACK 2896c412fa
amitiuttarwar:
code review ACK 2896c412fa
jonatack:
ACK 2896c412fa per `git diff 2b3f101 2896c41` only change since previous review is moving the recency check up to be verified first in `FindTxForGetData`, as it was originally in 353a391 (good catch), before looking up the transaction in the relay pool.
jnewbery:
code review ACK 2896c412fa
Tree-SHA512: e7d5bc006e626f60a2c108a9334f3bbb67205ace04a7450a1e4d4db1d85922a7589e0524500b7b4953762cf70554c4a08eec62c7b38b486cbca3d86321600868
Cache block filter headers at heights of multiples of 1000 in memory.
Block filter headers at height 1000x are checkpointed, and will be the
most frequently requested. Cache them in memory to avoid costly disk
reads.
Headers-first is the primary method of announcement on the network. If a
node fell back sending blocks by inv, it's probably for a re-org. The
final block hash provided should be the highest, so send a getheaders
and then fetch the blocks we need to catch up.
23083856a5 [test] Add test for cfcheckpt (Jim Posen)
f9e00bb25a [net processing] Message handling for getcfcheckpt. (Jim Posen)
9ccaaba11e [init] Add -peerblockfilters option (Jim Posen)
Pull request description:
Serve cfcheckpt messages if basic block filter index is enabled and `-peercfilters` is set.
`NODE_COMPACT_FILTERS` is not signaled to peers, but functionality can be used for testing and serving pre-configured clients.
ACKs for top commit:
jonatack:
Code review re-ACK 23083856a5 the only change since my review @ 967e2b1 is an update required for #16224 that was merged yesterday.
fjahr:
re-ACK 23083856a5
jkczyz:
re-ACK 23083856a5
ariard:
re-Code Review ACK 2308385
clarkmoody:
Tested ACK 23083856a
MarcoFalke:
re-ACK 23083856a5🌳
theStack:
ACK 23083856a5
Tree-SHA512: 8c751bbd7d1c31a413096462ae025c3d2f3163c7016cbec472a5f5ec267f8dd19a2dfc4d749876d7409c1db546e6fdd16461c6863effcfa0d3e993edcfa92a08
9847e205bf [docs] Improve commenting in ProcessGetData() (John Newbery)
2f032556e0 [test] test that an invalid GETDATA doesn't prevent processing of future messages (Amiti Uttarwar)
e257cf71c8 [net processing] ignore unknown INV types in GETDATA messages (Amiti Uttarwar)
047ceac142 [net processing] ignore tx GETDATA from blocks-only peers (Amiti Uttarwar)
Pull request description:
Currently we'll stall peers that send us an unknown INV type in a GETDATA message. Be a bit more friendly and just drop the invalid request.
Ditto for blocks-relay-only peers that send us a GETDATA for a transaction.
There's a test for the first part. The second is difficult to test in the functional test framework since we aren't able to make blocks-relay-only connections.
ACKs for top commit:
sipa:
utACK 9847e205bf
brakmic:
ACK 9847e205bf
luke-jr:
utACK 9847e205bf
naumenkogs:
utACK 9847e20
ajtowns:
utACK 9847e205bf
Tree-SHA512: 6007f2fd839ffe737727f6fb8e8f083b2d9e05a510748f1d40b8f9be8fdf7b5419a36d8f1039923eec1ba2983e8f6f0436ec5fc196d9f6dcb0657f2ff8ff8e4c
1ad8ea2b73 net: remove is{Empty,Full} flags from CBloomFilter, clarify CVE fix (Sebastian Falbesoner)
Pull request description:
The BIP37 bloom filter class `CBloomFilter` contains two flags `isEmpty`/`isFull` together with an update method with the purpose to, according to the comments, "avoid wasting cpu", i.e. the mechanism should serve as an optimization for the trivial cases of empty (all bits zero) or full (all bits one) filters.
However, the real reason of adding those flags (introduced with commit 37c6389c5a by gmaxwell) was a _covert fix_ of [CVE-2013-5700](https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2013-5700), a vulnerability that allowed a divide-by-zero remote node crash.
According to gmaxwell himself (https://github.com/bitcoin/bitcoin/pull/9060#issuecomment-257749165):
> the IsEmpty/IsFull optimizations were largely a pretextual optimization intended to make unexploitable a remote crash vulnerability (integer division by zero) that existed in the original bloom filtering code without disclosing it. I'm doubtful that they are all that useful. :)
For more information on how to trigger this crash, see PR https://github.com/bitcoin/bitcoin/pull/18515 which contains a detailled description and a regression test. It has also been discussed on a [recent PR club meeting on fuzzing](https://bitcoincore.reviews/18521.html).
The covert fix code already led to issues and PR based on the wrong assumption that the flags are there for optimization reasons (see #16886 and #16922). This PR gets rid of the flags and the update method and just focuses on the CVE fix itself, i.e. it can be seen as a revert of the covert fix commit modulo the actual fix.
ACKs for top commit:
meshcollider:
utACK 1ad8ea2b73
laanwj:
Concept and code review ACK 1ad8ea2b73
jkczyz:
ACK 1ad8ea2b73
MarcoFalke:
ACK 1ad8ea2b73
fjahr:
Code review ACK 1ad8ea2b73
Tree-SHA512: 29f7ff9faece0285e11e16c024851f5bcb772dec64118ccc3f9067ec256267ec8e1b1e3105c7de2a72fd122c3b085e8fc840ab8f4e49813f1cc7a444df1867f7
50fc4df6c4 [mempool] Persist unbroadcast set to mempool.dat (Amiti Uttarwar)
297a178536 [test] Integration tests for unbroadcast functionality (Amiti Uttarwar)
6851502472 [refactor/test] Extract P2PTxInvStore into test framework (Amiti Uttarwar)
dc1da48dc5 [wallet] Update the rebroadcast frequency to be ~1/day. (Amiti Uttarwar)
e25e42f20a [p2p] Reattempt initial send of unbroadcast transactions (Amiti Uttarwar)
7e93eecce3 [util] Add method that returns random time in milliseconds (Amiti Uttarwar)
89eeb4a333 [mempool] Track "unbroadcast" transactions (Amiti Uttarwar)
Pull request description:
This PR introduces mempool tracking of unbroadcast transactions and periodic reattempts at initial broadcast. This is a part of the rebroadcast project, and a standalone privacy win.
The current rebroadcast logic is terrible for privacy because 1. only the source wallet rebroadcasts transactions and 2. it does so quite frequently. In the current system, if a user submits a transaction that does not immediately get broadcast to the network (eg. they are offline), this "rebroadcast" behavior is the safety net that can actually serve as the initial broadcast. So, keeping the attempts frequent is important for initial delivery within a reasonable timespan.
This PR aims to improve # 2 by reducing the wallet rebroadcast frequency to ~1/day from ~1/15 min. It achieves this by separating the notion of initial broadcast from rebroadcasts. With these changes, the mempool tracks locally submitted transactions & periodically reattempts initial broadcast. Transactions submitted via the wallet or RPC are added to an "unbroadcast" set & are removed when a peer sends a `getdata` request, or the transaction is removed from the mempool. Every 10-15 minutes, the node reattempts an initial broadcast. This enables reducing the wallet rebroadcast frequency while ensuring the transactions will be propagated to the network.
For privacy improvements around # 1, please see #16698.
Thank you to gmaxwell for the idea of how to break out this subset of functionality (https://github.com/bitcoin/bitcoin/pull/16698#issuecomment-571399346)
ACKs for top commit:
fjahr:
Code review ACK 50fc4df6c4
MarcoFalke:
ACK 50fc4df6c4, I think this is ready for merge now 👻
amitiuttarwar:
The current tip `50fc4df` currently has 6 ACKs on it, so I've opened #18807 to address the last bits.
jnewbery:
utACK 50fc4df6c4.
ariard:
Code Review ACK 50fc4df (minor points no need to invalid other ACKs)
robot-visions:
ACK 50fc4df6c4
sipa:
utACK 50fc4df6c4
naumenkogs:
utACK 50fc4df
Tree-SHA512: 2dd935d645d5e209f8abf87bfaa3ef0e4492705ce7e89ea64279cb27ffd37f4727fa94ad62d41be331177332f8edbebf3c7f4972f8cda10dd951b80a28ab3c0f
- Mempool tracks locally submitted transactions (wallet or rpc)
- Transactions are removed from set when the node receives a GETDATA request
from a peer, or if the transaction is removed from the mempool.
