a9ecbdfcaa test: add more inactive filter tests to p2p_filter.py (Sebastian Falbesoner)
5eae034996 net: limit BIP37 filter lifespan (active between 'filterload' and 'filterclear') (Sebastian Falbesoner)
Pull request description:
This PR fixes https://github.com/bitcoin/bitcoin/issues/18483. On the master branch, there is currently _always_ a BIP37 filter set for every peer: if not a specific filter is set through a `filterload` message, a default match-everything filter is instanciated and pointed to via the `CBloomFilter` default constructor; that happens both initially, when the containing structure `TxRelay` is constructed:
c0b389b335/src/net.h (L812)
and after a loaded filter is removed again through a `filterclear` message:
c0b389b335/src/net_processing.cpp (L3201)
The behaviour was introduced by commit 37c6389c5a (an intentional covert fix for [CVE-2013-5700](https://github.com/bitcoin/bitcoin/pull/18515), according to gmaxwell).
This default match-everything filter leads to some unintended side-effects:
1. `getdata` request for filtered blocks (i.e. type `MSG_FILTERED_BLOCK`) are always responded to with `merkleblock`s, even if no filter was set by the peer, see issue #18483 (strictly speaking, this is a violation of BIP37) c0b389b335/src/net_processing.cpp (L1504-L1507)
2. if a peer sends a `filteradd` message without having loaded a filter via `filterload` before, the intended increasing of the banscore never happens (triggered if `bad` is set to true, a few lines below) c0b389b335/src/net_processing.cpp (L3182-L3186)
This PR basically activates the `else`-branch code paths for all checks of `pfilter` again (on the master branch, they are dead code) by limiting the pointer's lifespan: instead of always having a filter set, the `pfilter` is only pointing to a `CBloomFilter`-instance after receiving a `filterload` message and the instance is destroyed again (and the pointer nullified) after receiving a `filterclear` message.
Here is a before/after comparison in behaviour:
| code part / scenario | master branch | PR branch |
| --------------------------------------------- | ---------------------------------- | ---------------------------------------------------- |
| `getdata` processing for `MSG_FILTERED_BLOCK` | always responds with `merkleblock` | only responds if filter was set via `filterload` |
| `filteradd` processing, no filter was loaded | nothing | peer's banscore increases by 100 (i.e. disconnect) |
On the other code parts where `pfilter` is checked there is no change in the logic behaviour (except that `CBloomFilter::IsRelevantAndUpdate()` is unnecessarily called and immediately returned in the master branch).
Note that the default constructor of `CBloomFilter` is only used for deserializing the received `filterload` message and nowhere else. The PR also contains a functional test checking that sending `getdata` for filtered blocks is ignored by the node if no bloom filter is set.
ACKs for top commit:
MarcoFalke:
re-ACK a9ecbdfcaa, only change is in test code 🕙
Tree-SHA512: 1a656a6d74ccaf628e7fdca063ba63fbab2089e0b6d0a11be9bbd387c2ee6d3230706ff8ffc1a55711481df3d4547137dd7c9d9184d89eaa43ade4927792d0b6
by calling wait_for_cookie_credentials() to ensure the cookie file is written
and auth credentials available for testing the CLI -rpcwait option before the
RPC connection is up.
9df32e820d scripted-diff: test: replace command with msgtype (Sebastian Falbesoner)
Pull request description:
This is a follow-up PR to https://github.com/bitcoin/bitcoin/pull/18533, which changed the naming of `strCommand` to `msg_type` in the network processing code. The same approach is done here for the function test framework, to get rid of the wrong "command" terminology for network mesage types. (Commands are usually used in the CLI or RPC context, so using the same name in the network message context would only be confusing.)
The commit was created through the following steps:
1. search for all occurences of the string "command" within the folder `test/functional`
```git grep -i command test/functional > command_finds```
2. manually sort out all false-positives, i.e. occurences of "command" which describe commands in the correct sense (mostly CLI or RPC related, also some with Socks5)
3. put the remaining occurences into a scripted-diff (a quite simple one, actually) that renames "command" to "msgtype" in the concerned files.
The name `msgtype` was intentionally chosen without the underscore `_` as classes beginning with `msg_` define concrete types of messages.
ACKs for top commit:
MarcoFalke:
ACK 9df32e820d . Makes sense that tests use the same naming as Bitcoin Core. See `NetMsgType` here: https://doxygen.bitcoincore.org/namespace_net_msg_type.html
Tree-SHA512: cd0ee08a382910b7f10ce583acdaf4f8a39f9ba4a22434a914415727eedd98bac538de9bf6633574d5eb86f62558bc8dcb638a3289d99b04f8481f34e7a9a0c7
0d32d66148 Remove -upgradewallet startup option (Andrew Chow)
92263cce5b Add upgradewallet RPC (Andrew Chow)
1e48796c99 Make UpgradeWallet a member function of CWallet (Andrew Chow)
c988f27937 Have UpgradeWallet take the version to upgrade to and an error message out parameter (Andrew Chow)
1833237123 Only run UpgradeWallet if the wallet needs to be upgraded (Andrew Chow)
9c16b1735f Move wallet upgrading to its own function (Andrew Chow)
Pull request description:
`-upgradewallet` is largely incompatible with many recent wallet features and versions. For example, it was disabled if multiple wallets were used and would not work with encrypted wallets that were being upgraded to HD.
This PR does away with the old method of upgrading upon startup and instead allows users to upgrade their wallets via an `upgradewallet` RPC. This does largely the same thing as the old `-upgradewallet` option but because the wallet is loaded, it can be unlocked to upgrade to HD. Furthermore it is compatible with multiwallet as it works on the individual wallet that is specified by the RPC.
ACKs for top commit:
meshcollider:
Code review ACK 0d32d66148
darosior:
ACK 0d32d66148
MarcoFalke:
ACK 0d32d66148🚵
Tree-SHA512: b425bf6f5d605e26506889d63c780895482f07cbc086193218e031e8504d3072d41e90d65cd41bcc98ee4c1eb048954bc5d4ac85435f7394892373aac89a3b0a
fa03713e13 test: Properly raise FailedToStartError when rpc shutdown before warmup finished (take 2) (MarcoFalke)
Pull request description:
actually (?) fix#18561
See most recent traceback https://travis-ci.org/github/bitcoin/bitcoin/jobs/674668692#L7062
I believe the reason the error is still there is that ConnectionResetError is derived from OSError:
ConnectionResetError(ConnectionError(OSError))
And IOError is an alias for OSError since python 3.3, see https://docs.python.org/3/library/exceptions.html#IOError
So fix that by renaming IOError to the alias OSError and move the less specific catch clause down a few lines.
ACKs for top commit:
jonatack:
ACK fa03713e13
Tree-SHA512: 6e5b214ed9101bf8ebe7472dcc1f9e9d128e2575c93ec00c8d0774ae1a9b52a8c2a653a45a0eab8d881570b08dd5ffeddf5aca88a10438c366e1f633253cb0b5
fabfcad876 test: Bump timeout in wallet_import_rescan (MarcoFalke)
Pull request description:
Avoid timeouts when starting the node, also make error message more verbose
ACKs for top commit:
practicalswift:
ACK fabfcad876 -- patch looks correct
Tree-SHA512: 8fd60a05380349f521d0e814d2f268702dfbe57c7567a4f6e94435498dfdd32909179d75fded44757ecb1a93a4045842bc6d00bfd6cd18ba751513461359c7b0
92bcd70808 [wallet] allow transaction without change if keypool is empty (Sjors Provoost)
709f8685ac [wallet] CreateTransaction: simplify change address check (Sjors Provoost)
5efc25f963 [wallet] translate "Keypool ran out" message (Sjors Provoost)
Pull request description:
Extracted from #16944
First this PR simplifies the check when generating a change address, by dropping `CanGetAddresses` and just letting `reservedest.GetReservedDestination` do this check.
Second, when the keypool is empty, instead of immediately giving up, we create a dummy change address and pass that to coin selection. If we didn't need the change address (e.g. when spending the entire balance), then it's all good. If we did need a change address, we throw the original error.
ACKs for top commit:
fjahr:
Code review ACK 92bcd70808
jonasschnelli:
utACK 92bcd70808
achow101:
ACK 92bcd70808
meshcollider:
Code review ACK 92bcd70808
Tree-SHA512: 07b8c8251f57061c58a85ebf0359be63583c23bac7a2c4cefdc14820c0cdebcc90a2bb218e5ede0db11d1e204cda149e056dfd18614642070b3d56efe2735006
fa32097541 test: Create cached blocks not in the future (MarcoFalke)
Pull request description:
This avoids test failures when tests assume blocks are not from the future, like in wallet_dump: https://cirrus-ci.com/task/6607130193035264?command=ci#L3306
ACKs for top commit:
jonatack:
ACK fa32097541
Tree-SHA512: 60b6882e0e1df8c5d67f034533407a45d3685983891b67ff4631072bfd0a93a325c7ca18758d7a2df252e4fcdb7c87321cb1e84458b22782e57e719eec634c22
fa168d7542 rpc: Document all aliases for first arg of listtransactions (MarcoFalke)
fa5b1f067f rpc: Document all aliases for second arg of getblock (MarcoFalke)
fa86a4bbfc rpc: Rename first arg of generateblock RPC to "output" (MarcoFalke)
Pull request description:
This fixes a bug found with #18531:
* Currently the named argument for `generateblock` is documented as `address/descriptor`, but the server only accepts a named argument of `address`. Fix it by changing the name to `output` for both the documentation and the server code. Also, add tests to prove the server understands the new name `output`.
* Unrelated to that, there have been a bunch of aliases in the server code that are not present in the source code of the documentation. Fix that by adding the alias to the source code of the documentation. Only the first alias is displayed in the rendered documentation. Also, add tests to prove the server actually understands all aliases.
ACKs for top commit:
pierreN:
Tested ACK fa168d7 tests, help messages
Tree-SHA512: 05e15628e3a667b296f3783d20f764b450b959451b5360c7eaf5993156582d47a0f5882330ca2493b851eb46324d504953b90c875bc88a15c9e8c89eb3ef8d92
fa4632c417 test: Move boost/stdlib includes last (MarcoFalke)
fa488f131f scripted-diff: Bump copyright headers (MarcoFalke)
fac5c37300 scripted-diff: Sort test includes (MarcoFalke)
Pull request description:
When writing tests, often includes need to be added or removed. Currently the list of includes is not sorted, so developers that write tests and have `clang-format` installed will either have an unrelated change (sorting) included in their commit or they will have to manually undo the sort.
This pull preempts both issues by just sorting all includes in one commit.
Please be aware that this is **NOT** a change to policy to enforce clang-format or any other developer guideline or process. Developers are free to use whatever tool they want, see also #18651.
Edit: Also includes a commit to bump the copyright headers, so that the touched files don't need to be touched again for that.
ACKs for top commit:
practicalswift:
ACK fa4632c417
jonatack:
ACK fa4632c417, light review and sanity checks with gcc build and clang fuzz build
Tree-SHA512: 130a8d073a379ba556b1e64104d37c46b671425c0aef0ed725fd60156a95e8dc83fb6f0b5330b2f8152cf5daaf3983b4aca5e75812598f2626c39fd12b88b180
38677274f9 rpc: settxfee respects -maxtxfee wallet setting (Fabian Jahr)
bda84a08a0 rpc: Add documentation for deactivating settxfee (Fabian Jahr)
Pull request description:
~~Closes 18315~~
`settxfee` can be deactivated by passing 0 as the fee. That does not seem to be documented for the user so this PR adds it in the description. The return value of a simple boolean seems also too simplified given the multiple dimensions that this deactivation feature enables. I.e. it does not seem intuitive if the returned boolean shows that the call succeeded or if means that `settxfee` is active. My suggested solution is to change the return value to a JSON object that included the "active" state and the currently set fee rate.
Examples:
```
$ src/bitcoin-cli settxfee 0.0000000
{
"active": false,
"fee_rate": "0.00000000 BTC/kB"
}
$ src/bitcoin-cli settxfee 0.0001
{
"active": true,
"fee_rate": "0.00010000 BTC/kB"
}
```
ACKs for top commit:
MarcoFalke:
ACK 38677274f9, seems useful to error out early instead of later #16257🕍
jonatack:
ACK 38677274f9
meshcollider:
LGTM, utACK 38677274f9
Tree-SHA512: 642813b5cf6612abb4b6cb63728081a6bd1659d809e0149c8f56060b6da7253fee989b3b202854f3051df3773c966799af30b612648c466b099f00590f356548
a2324e4d3f test: Improve naming and logging of avoid_reuse tests (Fabian Jahr)
1abbdac677 wallet: Prefer full destination groups in coin selection (Fabian Jahr)
Pull request description:
Fixes#17603 (together with #17843)
In the case of destination groups of >10 outputs existing in a wallet with `avoid_reuse` enabled, the grouping algorithm is adding left-over outputs as an "incomplete" group to the list of groups even when a full group has already been added. This leads to the strange behavior that if there are >10 outputs for a destination the transaction spending from that will effectively use `len(outputs) % 10` as inputs for that transaction.
From the original PR and the code comment I understand the correct behavior should be the usage of 10 outputs. I opted for minimal changes in the current code although there maybe optimizations possible for cases with >20 outputs on a destination this sounds like too much of an edge case right now.
ACKs for top commit:
jonatack:
Re-ACK a2324e4
achow101:
ACK a2324e4d3f
kallewoof:
ACK a2324e4d3f
meshcollider:
Tested ACK a2324e4d3f (verified the new test fails on master without this change)
Tree-SHA512: 4743779c5d469fcd16df5baf166024b1d3c8eaca151df1e8281b71df62b29541cf7bfee3f8ab48d83e3b34c9256e53fd38a7b146a54c79f9caa44cce3636971a
This is the functional test framework pendant for
7777e3624f, which renamed "strCommand" with
"msg_type" in the network processing code.
-BEGIN VERIFY SCRIPT-
# Rename in test framework
sed -i 's/command/msgtype/g' ./test/functional/test_framework/messages.py ./test/functional/test_framework/mininode.py
# Rename in individual tests
sed -i 's/command/msgtype/g' ./test/functional/p2p_invalid_messages.py ./test/functional/p2p_leak.py
-END VERIFY SCRIPT-
check the following expected behaviors if no filter is set:
-> filtered block requests are ignored by the node
-> sending a 'filteradd' message is treated as misbehavior
(i.e. the peer's banscore increases by 100)
also fixes a bug in the on_inv() callback method, which
directly modified the type from BLOCK to FILTERED_BLOCK
in the received 'inv' message rather than just for the reply
Co-authored-by: MarcoFalke <falke.marco@gmail.com>
When a wallet uses avoid_reuse and has a large number of outputs in
a single destination, it groups these outputs in OutputGroups that
are no larger than OUTPUT_GROUP_MAX_ENTRIES. The goal is to spend
as many outputs as possible from the destination while not breaking
consensus due to a huge number of inputs and also not surprise the
use with high fees. If there are n outputs in a destination and
n > OUTPUT_GROUP_MAX_ENTRIES then this results in one or many groups
of size OUTPUT_GROUP_MAX_ENTRIES and possibly one group of size
< OUTPUT_GROUP_MAX_ENTRIES.
Prior to this commit the coin selection in the case where
n > OUTPUT_GROUP_MAX_ENTRIES was skewed towards the one group of
size < OUTPUT_GROUP_MAX_ENTRIES if it exists and the amount to be
spent by the transaction is smaller than the aggregate of those
of the group size < OUTPUT_GROUP_MAX_ENTRIES. The reason is that
the coin selection decides between the different groups based on
fees and mostly the smaller group will cause smaller fees.
The behavior that users of the avoid_reuse flag seek is that the
full groups of size OUTPUT_GROUP_MAX_ENTRIES get used first. This
commit implements this by pretending that the small group has
a large number of ancestors (one smallet than the maximum allowed
for this wallet). This dumps the small group to the bottom of the
list of priorities in the coin selection algorithm.
0306d78cb4 Use getbalances in wallet_address_types tests (Jon Atack)
7eacdc5167 Shift coverage from getunconfirmedbalance to getbalances in wallet_abandonconflict tests (Jon Atack)
3e6f7377f6 Improve getbalances coverage in wallet_balance tests (Jon Atack)
Pull request description:
<strike>This PR updates several tests and then removes the `getunconfirmedbalance` RPC which was deprecated in facfb4111d a year ago.
Next steps: remove the deprecated `getwalletinfo` fields and the `getbalance` RPC in follow-ups, if there seems to be consensus on those removals.</strike>
Update:
`getunconfirmedbalance` RPC was deprecated in facfb4111d a year ago, but following the review comments below, this PR now only updates the test coverage to use `getbalances` while still leaving basic coverage for `getunconfirmedbalance` in wallet_balance.py.
That said, I've seen 3 regular contributors confused in the past 10 days by "DEPRECATED" warnings in the code that are not following the deprecation policy in [JSON-RPC-interface.md#versioning](https://github.com/bitcoin/bitcoin/blob/master/doc/JSON-RPC-interface.md#versioning).
ISTM these warnings should either be removed, or the calls deprecated (`-deprecatedrpc`), or the policy updated to describe these warnings as a pre-deprecation practice.
ACKs for top commit:
jnewbery:
utACK 0306d78cb
Tree-SHA512: 692e43e9bed5afa97d905740666e365f0b64e559e1c75a6a398236d9e943894e3477947fc11324f420a6feaffa0c0c1532aa983c50090ca39d06551399e6ddd1
9cdddae3b4 test: add rpc_signrawtransaction logging (Jon Atack)
4d6cde38ce test: refactor rpc_signrawtransaction witness script tests (Jon Atack)
Pull request description:
As a follow-up to #18484, the new tests are good but bury the one non-duplicate line in each test that sets the witness script, and there is no logging in the testfile. This PR makes it easy to see what is unique to each of the new tests and adds logging.
ACKs for top commit:
theStack:
ACK 9cdddae3b4🥚🐰
Tree-SHA512: 7b1ca303326658afb90b7635abc9fe8bb65f0be004124d4dcf38702bb6f38bc06ce33c0642be4ad5d511453d003cdefeea691e66e3b963a4feb66f6237a3c241
fab9899204 test: Try once more when RPC connection fails on Windows (MarcoFalke)
faa655731e test: Document why connection is re-constructed on windows (MarcoFalke)
fa9f4f663c test: Remove python 3.4 workaround (MarcoFalke)
fae760f2b2 cirrus: Bump freebsd to 12.1 (MarcoFalke)
Pull request description:
Fixes: #18548
ACKs for top commit:
hebasto:
ACK fab9899204, I have reviewed the code and it looks OK, I agree it can be merged.
Tree-SHA512: c4e9ed8d995b63a820ca66984f152ac216c83ba1f318b61b15c6d375c0e936c08f6bc3d38c255dddf3ee8952f848c7ababf684854e07a7c1b1d8504e6b7208ba
555567ace9 test: Extend wallet_dump test to cover comments (MarcoFalke)
Pull request description:
ACKs for top commit:
ryanofsky:
Code review ACK 555567ace9. Nice new checks in this test. I confirmed this catches the missing FormatISO8601DateTime call you discovered in https://github.com/bitcoin/bitcoin/pull/17954#discussion_r406891999
Tree-SHA512: 71aa23dd039f3bcdee642b01151edd1a0d44f48cedd070f5858148c8cb8abd6f5edfd212daeba38e35c843da5ea6c799e5a952105fdecedac355a5a843c05a84
fa404f1e47 test: Check that the version message does not leak the local address of the node (MarcoFalke)
Pull request description:
Add test for #8740
ACKs for top commit:
theStack:
ACK fa404f1e47
Tree-SHA512: 4d1c10d1c02fba4b51bd8b9eb3a0d9a682b6aac8c3f6924e295fdca3faefa5ecc3eaa87d347cfec5d2b2bc49963c10fe0a37c463f36088ed0304a2e3716b963b
The interfaces for the methods wait_for_header() and wait_for_merkleblock() are
changed to take a hex string instead of an integer, improving type safety and
removing the burden from the caller to always do the transformation via
`int(...)`. As suggested by MarcoFalke in
https://github.com/bitcoin/bitcoin/pull/18593#discussion_r407062253
Implements the missing initialization/serialization methods for
msg_merkleblock, based on the already present class CMerkleBlock.
Also changes the method wait_for_merkleblock() to be more precise by waiting
for a merkleblock with a specified blockhash instead of an arbitrary one.
In the BIP37 test p2p_filter.py, this new method is used to make the test of
receiving merkleblock and tx if a filter is set to be more precise, by checking
if they also arrive in the right order.
7524b6479c Add tests for generateblock (Andrew Toth)
dcc8332543 Add generateblock rpc (Andrew Toth)
Pull request description:
The existing block generation rpcs for regtest, `generatetoaddress` and `generatetodescriptor`, mine everything in the mempool up to the block weight limit. This makes it difficult to test a system for several scenarios where a different set of transactions are mined. For example:
- Testing the common scenario where a transaction is replaced in the mempool but the replaced transaction is mined instead.
- Testing for a double-spent transaction where a transaction that conflicts with the mempool is mined.
- Testing for non-standard transactions that are mined.
- Testing the scenario where several blocks are mined without a specific transaction in the mempool being included in a block.
This PR introduces a new rpc, `generateblock`, that takes an array of raw transactions and txids and mines only those and the coinbase. Any txids must be in the mempool, but the raw txs can be anything conforming to consensus rules. The coinbase can be specified as either an address or descriptor.
This reopens#17653 since it was closed by mistake.
Thanks to instagibbs for code suggestions that I used here.
ACKs for top commit:
MarcoFalke:
re-ACK 7524b6479c📁
Tree-SHA512: 857106007465b5b9b8a84b6d07c17cbf8378a33a72d32ff79abea1d5ab4babb4d53a11ddbb14595aa1fac9dfa1391e3a11403d742f69951beea2f683e8a01cd4
5df0877f91 test: update and harden interface_bitcoin_cli tests (Jon Atack)
75019774c9 cli -getinfo: use getbalances instead of deprecated getwalletinfo balance (Jon Atack)
Pull request description:
Extracted from #18453 to preserve that PR as a discussion on multiwallet RPC/CLI.
This PR updates `bitcoin-cli -getinfo` to fetch the wallet balance from `getbalances` in order to no longer depend on `getwalletinfo.balance` which was deprecated a year ago in facfb41.
I found this when removing the getwalletinfo() `balance`, `unconfirmed_balance`, and `immature_balance` fields to see what broke from depending on them.
I didn't see any perceivable change in `-getinfo` run time from the change.
Test coverage for this change is provided by `test/functional/interface_bitcoin_cli.py`, which the second commit updates to (a) no longer depend on getwalletinfo.balances and (b) test the -getinfo blockcount and balance fields against non-default, non-zero values.
ACKs for top commit:
robot-visions:
ACK 5df0877
MarcoFalke:
ACK 5df0877
vasild:
re-ACK 5df0877f9
promag:
Code review ACK 5df0877f91.
theStack:
ACK 5df0877f91
Tree-SHA512: 0dd8c62f915b1c0112e42b132dcf74a141bdd1f51e7c17d4a698b374ec296f4f9836f7058dbe237cf24f9bfb32ea5000e14f7089e2e86472d9c6a175be26e910
MarcoFalke