This removes a lot of the craziness that was added during the 0.21
rebase 240f03f586 (Bitcoin PR #17290)
as well as a lot of other upstream craziness. It's a very positive
change but will probably be a bear to review.
There are two places where I (purposely) nontrivially changed the
logic beyond what the upstream PR did:
1. When setting the change output positions, I used a uniform
distribution for all change outputs and also obeyed the user's
`change_position` choice. The old code was non-uniform and
had a bug where it would offset the change position.
For non-policy assets, I do not create zero-sized change
since we would just remove them later. For the policy asset,
I do sometimes create zero-sized change since the upstream
logic expects it to exist.
2. Rather than "reblinding whenever something changes" we blind
once for size/fee estimation purposes and then again at the
end after all potential adjustments have been made. The
resulting code should be closer to upstream.
We should really redo this code though because our use of the
`BlindDetails` structure is very confusing and stateful.
3. We consider whether or not we're using CT/CA when estimating
the size of change outpust. This is because we now use branch
and bound far more often (yay!) which selects coins such that
there will be no change output. This is implemented however
by adding a change output then dumping it to fees ... meaning
that if we dramatically overestimate the size of a change out
then we'll end up dramatically dumping too many coins to fees.
There are another couple small things, which I apologize for .. this
PR took me 2 hours to get compiling and 11 hours to get working(!!)
and some things got away from me.
One fun thing is that I changed CWallet::SelectCoinsMinConf to gate
an addition on coin_selection_params.m_subtract_fee_outputs ... this
is actually an upstream bug that I'll file whenever I get around to
producing a test that triggers it. (A bit hard as I have to gin up
a scenario where branch-and-bound fails to hit this codepath.)
Also, in test/functional/wallet_bumpfee.py I had to change the feerate
required to force selection to add another input from 500 to 800. I
don't know why and I don't care to figure it out.
This is a really non-trivial wallet refactor "effective value" which has been
in the Core queue for a pretty long time. There were a ton of merge conflicts
but they were mostly mechanical (all the benchmarks) or "take the Core side
of the diff" (most of wallet.cpp).
The two nontrivial things were:
* In `SelectCoinsMinConf` in wallet.cpp I had to slightly move the asset
filtering logic (but I didn't change it)
* In `OutputGroup::Insert' in coinselection.cpp I changed the effective
value computation to only consider policyAsset values
96c2c9520e scripted-diff: Rename SelectCoinsMinConf to AttemptSelection (Andrew Chow)
b583f73354 Move vin filling to before final fee setting (Andrew Chow)
d39cac0547 Set m_subtract_fee_outputs during recipients vector loop (Andrew Chow)
364e0698a5 Move variable initializations to where they are used (Andrew Chow)
32ab430651 Move recipients vector checks to beginning of CreateTransaction (Andrew Chow)
cd1d6d3324 Rename nSubtractFeeFromAmount in CreateTransaction (Andrew Chow)
dac21c793f Rename nValue and nValueToSelect (Andrew Chow)
d2aee3bbc7 Remove extraneous scope in CreateTransactionInternal (Andrew Chow)
b2995963b5 Move cs_wallet lock in CreateTransactionInternal to top of function (Andrew Chow)
Pull request description:
#17331 did some refactors and cleanup of `CreateTransactionInternal` to make it easier to understand, however it is still a bit convoluted even though it doesn't have to be. This PR does additional cleanup and refactoring to `CreateTransactionInternal` so that it is easier to understand. Some unnecessary code was removed, some variables moved around to where they matter, and several indents removed.
ACKs for top commit:
glozow:
reACK 96c2c95
ryanofsky:
Code review ACK 96c2c9520e also acked previously (was reverted).
meshcollider:
re-utACK 96c2c9520e
Tree-SHA512: 3dba67ed436968a07bfd82d435d566ad74e116c6e50ac9baed7144a46ad5c0f630b1ba59d91e8e8972ac2af559d7c0576f0560f09684d2ab20fad6689902866f
44d05d0a69 test: remove sanitizer suppression for nanobench (Martin Ankerl)
e3c866e3ca test: update nanobench from release 4.0.0 to 4.3.4 (Martin Ankerl)
Pull request description:
This updates the third-party library nanobench with the latest release. It contains mostly minor bugfixes, a new pyperf output format, ability to suppress warnings with environment variable `NANOBENCH_SUPPRESS_WARNINGS`. Full changelog:
v4.0.2
* Changed `doNotOptimizeAway` to what google benchmark is doing. The old code did not work on some machines.
* fix: display correct "total" value
* minor Documentation updates
v4.1.0
* Updated link to new pyperf home
* Adds ability to configure console output time unit
* Add support for environment variable `NANOBENCH_SUPPRESS_WARNINGS`
* Nanobench is now usable with CMake's FetchContent (see documentation: https://nanobench.ankerl.com/tutorial.html#cmake-integration)
v4.2.0
* Ability to store and later compare results added, through `pyperf`.
* See https://nanobench.ankerl.com/tutorial.html#pyperf-python-pyperf-module-output
* Added lots of build targets to travis, similar to bitcoin's build.
* Some minor API & documentation improvements
v4.3.0
* `ankerl::nanobench::Rng` can now return the state with `std::vector<uint64_t> Rng::state()`, and this can also be used to initialize the Rng.
v4.3.1
* Minor cmake improvements when integrationg as a third-party library: add alias `nanobench::nanobench`, default to C++17
v4.3.2
* Fixed a MSVC 2015 build problem
* updates license to 2021.
* build should now work with very old linux headers
* Also disable UBSAN (bitcoin needed to add a suppression)
v4.3.3
* Do not use locale-dependent `std::to_string`
v4.3.4
* Add missing sanitizer suppression to `rotl`
ACKs for top commit:
MarcoFalke:
review ACK 44d05d0a69
Tree-SHA512: 3291c85057720cfc84a44bfaa305a7d0df4dc35779169d20de73d32e40d4cdbf3f005bf343f79710eca517441de2459e8118c195c5f5136f99d1f50ebd5dfd08
7a799c9c2b index: refactor-only: Reuse CChain ref (Carl Dong)
db33cde80f index: Add chainstate member to BaseIndex (Carl Dong)
f4a47a1feb bench: Use existing chainman in AssembleBlock (Carl Dong)
91226eb917 bench: Use existing NodeContext in DuplicateInputs (Carl Dong)
e6b4aa6eb5 miner: Pass in chainman to RegenerateCommitments (Carl Dong)
9ecade1425 rest: Add GetChainman function and use it (Carl Dong)
fc1c282845 rpc/blockchain: Use existing blockman in gettxoutsetinfo (Carl Dong)
Pull request description:
Overall PR: #20158 (tree-wide: De-globalize ChainstateManager)
The first 2 commits are fixups addressing review for the last bundle: #21391
NEW note:
1. I have opened #21766 which keeps track of potential improvements where the flaws already existed before the de-globalization work, please post on that issue about these improvements, thanks!
Note to reviewers:
1. This bundle may _apparently_ introduce usage of `g_chainman` or `::Chain(state|)Active()` globals, but these are resolved later on in the overall PR. [Commits of overall PR](https://github.com/bitcoin/bitcoin/pull/20158/commits)
2. There may be seemingly obvious local references to `ChainstateManager` or other validation objects which are not being used in callers of the current function in question, this is done intentionally to **_keep each commit centered around one function/method_** to ease review and to make the overall change systematic. We don't assume anything about our callers. Rest assured that once we are considering that particular caller in later commits, we will use the obvious local references. [Commits of overall PR](https://github.com/bitcoin/bitcoin/pull/20158/commits)
3. When changing a function/method that has many callers (e.g. `LookupBlockIndex` with 55 callers), it is sometimes easier (and less error-prone) to use a scripted-diff. When doing so, there will be 3 commits in sequence so that every commit compiles like so:
1. Add `new_function`, make `old_function` a wrapper of `new_function`, divert all calls to `old_function` to `new_function` **in the local module only**
2. Scripted-diff to divert all calls to `old_function` to `new_function` **in the rest of the codebase**
3. Remove `old_function`
ACKs for top commit:
jarolrod:
ACK 7a799c9
ariard:
Code Review ACK 7a799c9
fjahr:
re-ACK 7a799c9c2b
MarcoFalke:
review ACK 7a799c9c2b🌠
ryanofsky:
Code review ACK 7a799c9c2b. Basically no change since last review except fixed rebase conflicts and a new comment about REST Ensure()
jamesob:
conditional ACK 7a799c9c2b ([`jamesob/ackr/21767.1.dongcarl.bundle_6_n_prune_g_chai`](https://github.com/jamesob/bitcoin/tree/ackr/21767.1.dongcarl.bundle_6_n_prune_g_chai))
Tree-SHA512: 531c00ddcb318817457db2812d9a9d930bc664e58e6f7f1c746350732b031dd624270bfa6b9f49d8056aeb6321d973f0e38e4ff914acd6768edd8602c017d10e
This updates the third-party library nanobench with the latest release. It contains mostly minor bugfixes, a new pyperf output format, ability to suppress warnings with environment variable `NANOBENCH_SUPPRESS_WARNINGS`. Full changelog:
v4.0.2
* Changed `doNotOptimizeAway` to what google benchmark is doing. The old code did not work on some machines.
* fix: display correct "total" value
* minor Documentation updates
v4.1.0
* Updated link to new pyperf home
* Adds ability to configure console output time unit
* Add support for environment variable `NANOBENCH_SUPPRESS_WARNINGS`
* Nanobench is now usable with CMake's FetchContent (see documentation: https://nanobench.ankerl.com/tutorial.html#cmake-integration)
v4.2.0
* Ability to store and later compare results added, through `pyperf`.
* See https://nanobench.ankerl.com/tutorial.html#pyperf-python-pyperf-module-output
* Added lots of build targets to travis, similar to bitcoin's build.
* Some minor API & documentation improvements
v4.3.0
* `ankerl::nanobench::Rng` can now return the state with `std::vector<uint64_t> Rng::state()`, and this can also be used to initialize the Rng.
v4.3.1
* Minor cmake improvements when integrationg as a third-party library: add alias `nanobench::nanobench`, default to C++17
v4.3.2
* Fixed a MSVC 2015 build problem
* updates license to 2021.
* build should now work with very old linux headers
* Also disable UBSAN (bitcoin needed to add a suppression)
v4.3.3
* Do not use locale-dependent `std::to_string`
v4.3.4
* Add missing sanitizer suppression to `rotl`
SelectCoinsMinConf is a bit of a misnomer now since it really just does
all of the coin selection given some parameters. So rename this to
something less annoying to say and makes a bit more sense.
-BEGIN VERIFY SCRIPT-
sed -i 's/SelectCoinsMinConf/AttemptSelection/g' $(git grep -l SelectCoinsMinConf ./src)
-END VERIFY SCRIPT-
51a3ac242c Have OutputGroup determine the value to use (Andrew Chow)
6d6d278475 Change SelectCoins_test to actually test SelectCoins (Andrew Chow)
9d3bd74ab4 Remove CreateTransaction while loop and some related variables (Andrew Chow)
6f0d5189af Remove use_bnb and bnb_used (Andrew Chow)
de26eb0e1f Do both BnB and Knapsack coin selection in SelectCoinsMinConf (Andrew Chow)
01dc8ebda5 Have KnapsackSolver actually use effective values (Andrew Chow)
bf26e018de Roll static tx fees into nValueToSelect instead of having it be separate (Andrew Chow)
cc3f14b27c Move output reductions for fee to after coin selection (Andrew Chow)
d97d25d950 Make cost_of_change part of CoinSelectionParams (Andrew Chow)
af5867c896 Move some calculations to common code in SelectCoinsMinConf (Andrew Chow)
1bf4a62cb6 scripted-diff: rename some variables (Andrew Chow)
Pull request description:
Changes `KnapsackSolver` to use effective values instead of just the nominal txout value. Since fees are taken into account during the selection itself, we finally get rid of the `CreateTransaction` loop as well as a few other things that only were only necessary because of that loop.
This should not change coin selection behavior at all (except maybe remove weird edge cases that were caused by the loop). In order to keep behavior the same, `KnapsackSolver` will select outputs with a negative effective value (as it did before).
ACKs for top commit:
ryanofsky:
Code review ACK 51a3ac242c. Looks good to go!
instagibbs:
review ACK https://github.com/bitcoin/bitcoin/pull/17331/commits/51a3ac242c92e69b59df26f8f9e287b31e5c3b0f
meshcollider:
re-light-utACK 51a3ac242c
Tree-SHA512: 372c27e00edcd5dbf85177421ba88f20bfdaf1791b6e3dc022c44876ecc379403e2375ed69e71c512c49e6af87641001ff385c4b25ab93684b3a08a53bf3824e
ce6bca88e8 doc: release note for getnodeaddresses by network (Jon Atack)
3f89c0e990 test: improve getnodeaddresses coverage, test by network (Jon Atack)
6c98c09991 rpc: enable filtering getnodeaddresses by network (Jon Atack)
80ba294854 p2p: allow CConnman::GetAddresses() by network, add doxygen (Jon Atack)
a49f3ddbba p2p: allow CAddrMan::GetAddr() by network, add doxygen (Jon Atack)
c38981e748 p2p: pull time call out of loop in CAddrMan::GetAddr_() (João Barbosa)
d35ddca91e p2p: enable CAddrMan::GetAddr_() by network, add doxygen (Jon Atack)
Pull request description:
This patch allows passing a network argument to CAddrMan::GetAddr(), CConnman::GetAddresses(), and rpc getnodeaddresses to return only addresses of that network.
It also contains a performance optimisation by promag.
ACKs for top commit:
laanwj:
Code review and lightly tested ACK ce6bca88e8
vasild:
ACK ce6bca88e8
Tree-SHA512: 40e700d97091248429c73cbc0639a1f03ab7288e636a7b9026ad253e9708253c6b2ec98e7d9fb2d56136c0f762313dd648915ac98d723ee330d713813a43f99d
The fees for transaction overhead and recipient outputs are now included
in nTargetValue instead of being a separate parameter. For the coin
selection algorithms, it doesn't matter that these are separate as in
either case, the algorithm needs to select enough to cover these fees.
Note that setting nValueToSelect is changed as it now includes
not_input_fees. Without the change to how nValueToSelect is increased
for KnapsackSolver, this would result in overpaying fees. The change to
increase by the difference between nFeeRet and not_input_fees allows
this to have the same behavior as previously.
Additionally, because we assume that KnapsackSolver will always find a
solution that requires change (we assume that BnB always finds a
non-change solution), we also include the fee for the change output in
KnapsackSolver's target. As part of this, we also use the changeless
nFeeRet when iterating for KnapsackSolver. This is because we include
the change fee when doing KnapsackSolver, so nFeeRet on further
iterations won't include the change fee.
725d7ae049 Use PrecomputedTransactionData in signet check (Pieter Wuille)
497718b467 Treat amount<0 also as missing data for P2WPKH/P2WSH (Pieter Wuille)
3820090bd6 Make all SignatureChecker explicit about missing data (Pieter Wuille)
b77b0cc507 Add MissingDataBehavior and make TransactionSignatureChecker handle it (Pieter Wuille)
Pull request description:
Currently we have 2 levels of potentially-missing data in the transaction signature hashes:
* P2WPKH/P2WSH hashes need the spent amount
* P2TR hashes need all spent outputs (amount + scriptPubKey)
Missing amounts are treated as -1 (thus leading to unexpected signature failures), while missing outputs in P2TR validation cause assertion failure. This is hard to extend for signing support, and also quite ugly in general.
In this PR, an explicit configuration option to {Mutable,}TransactionSignatureChecker is added (MissingDataBehavior enum class) to either select ASSERT_FAIL or FAIL. Validation code passes ASSERT_FAIL (as at validation time all data should always be passed, and anything else is a serious bug in the code), while signing code uses FAIL.
The existence of the ASSERT_FAIL option is really just an abundance of caution. Always using FAIL should be just fine, but if there were for some reason a code path in consensus code was introduced that misses certain data, I think we prefer as assertion failure over silently introducing a consensus change.
Potentially useful follow-ups (not for this PR, in my preference):
* Having an explicit script validation error code for missing data.
* Having a MissingDataBehavior::SUCCEED option as well, for use in script/sign.cpp DataFromTransaction (if a signature is present in a witness, and we don't have enough data to fully validate it, we should probably treat it as valid and not touch it).
ACKs for top commit:
sanket1729:
reACK 725d7ae049
Sjors:
ACK 725d7ae049
achow101:
re-ACK 725d7ae049
benthecarman:
ACK 725d7ae049
fjahr:
Code review ACK 725d7ae049
Tree-SHA512: d67dc51bae9ca7ef6eb9acccefd682529f397830f77d74cd305500a081ef55aede0e9fa380648c3a8dd4857aa7eeb1ab54fe808979d79db0784ac94ceb31b657
Includes a memory leak in the checkqueue unit test (but not in
the actual code). WE really need to switch our checkqueue to use
std::unique_pointer rather than bare pointers. But this would be
invasive enough that I want to do it in a followup PR.
Also pretty-much disable the validation_flush_cache unit test.
This is a stupid and irritating test which tries to unit-test
exact memory usage of std containers. It already has at least
one "remove wrong assumptions" update upstream and after many
tries I was unable to change all the magic numbers in a way
that'd consistently pass CI for Elements.
Also adds a couple ubsan suppressions about perfectly-legitimate
conversions of integer types.
Several conflicts in the C++ code related to the new `flags` parameter
to `CheckSignature` and the corresponding function being renamed upstream
to `CheckSignatureECDSA`.
Several conflicts in the test harness as Steven sorta pulled the new
upstream ECKey module into the Python code, and the actual upstream
code was slightly different. Also needed to update the feature_taproot
code to always use the non-RANGEPROOF sighash since dynafed is not
enabled in the Taproot test.
Also had to pull the `set_wif` method out of `ECKey` and inline it because
otherwise it triggers a "circular inclusion" error between script.py (which
would pull in `base58_to_bytes` from address.py) and address.py (which now
pulls in some taproot EC related stuff from script.py).
Noticed that #960 does not test the "sighash rangeproof flag set but no
witnesses" case.
ebc4ab721b refactor: post Optional<> removal cleanups (fanquake)
57e980d13c scripted-diff: remove Optional & nullopt (fanquake)
Pull request description:
Same rationale & motivation as #21404, which turned out to be quite low in the number of potential conflicts. Lets see what the bot has to say here.
ACKs for top commit:
practicalswift:
cr ACK ebc4ab721b: patch looks correct
jnewbery:
utACK ebc4ab721b
laanwj:
Code review ACK ebc4ab721b
Tree-SHA512: 550fbeef09b9d35ddefaa805d1755c18c8fd499c4b0f77ebfece8c20296a7abd1cf6c699e2261f92fe3552deeb7555ec2a2287ffe3ab9e98bb9f8612a4d43be3
