Behavior might have recently changed in #17331 (it is not clear) but not
noticed because there is no test coverage.
This adds test coverage for current subtract from recipient behavior
without changing it.
Co-authored-by: Andrew Chow <achow101-github@achow101.com>
fbf485c9b2 Allow tr() import only when Taproot is active (Andrew Chow)
Pull request description:
To avoid issues around fund loss, only allow descriptor wallets to import `tr()` descriptors after taproot has activated.
ACKs for top commit:
sipa:
utACK fbf485c9b2
fjahr:
Code review ACK fbf485c9b2
laanwj:
Code review ACK fbf485c9b2
prayank23:
utACK fbf485c9b2
Tree-SHA512: 83c43376515eea523dbc89bc5a0fde53e54aec492e49a40c2a33d80fc94aac459e232ae07b024b4bd75b58078c8d090bc7a2d69541c5d3d4834d2f4cfc9c8208
No change in behavior. This just moves some code from the ListCoins test
setup to a reusable util function, so it can be reused in a new test in
the next commit.
f47e802839 Rearrange fillPSBT arguments (Russell Yanofsky)
Pull request description:
Move fillPSBT inout argument before output-only arguments. This is a nice thing to do to keep the interface style [consistent](https://google.github.io/styleguide/cppguide.html#Inputs_and_Outputs). But motivation is to work around a current limitation of the libmultiprocess code generator (which figures out order of inout parameters by looking at input list, but more ideally would use the output list).
---
This PR is part of the [process separation project](https://github.com/bitcoin/bitcoin/projects/10). The commit was first part of larger PR #10102.
ACKs for top commit:
achow101:
ACK f47e802839
theStack:
Code-review ACK f47e802839
Tree-SHA512: 1787af3031ff7ed6b519f3b93054d8b257af96a3380a476a6dab0f759329039ecc5d624b785c5c2d14d594fc852dd81c626880c775c691ec9c79b7b3dbcfb257
Move fillPSBT input-output argument before output-only arguments. This is a
temporary workaround which can go away with improvements to libmultiprocess
code generator. Currently code generator figures out order of input-output
parameters by looking at input list, but it would make more sense for it to
take order from output list, so input-only parameters still have to be first
but there is more flexibility for the other parameters.
e60cd26ad4 Do not load external signers wallets when unsupported (Andrew Chow)
Pull request description:
When external signer support is not compiled, do not load external signer wallets.
Alternative to #22168.
ACKs for top commit:
promag:
Tested ACK e60cd26ad4.
meshcollider:
Code review ACK e60cd26ad4
Tree-SHA512: aed2d0038f448c2f89c6b48f412b106e63c9ed20e748e69aae21fb58c33fc7e4fa73375a52372c73788669eb2b968a8da6b022c65658fa4484f5bbcf205b1b15
d44a261acf Fix issues when `walletdir` is root directory (unknown)
Pull request description:
+ Remove one character less from wallet path
+ After testing lot of random strings with special chars in `wallet_name`, I found that the issue was not related to special characters in the name. Reviewing PR https://github.com/bitcoin/bitcoin/pull/21907 helped me resolve the issue.
**Real issue**: If the path mentioned in `walletdir` is a root directory, first character of the wallet name or path is removed
**Solution**: `if` statement to check `walletdir` is a root directory
Fixes: https://github.com/bitcoin/bitcoin/issues/21510https://github.com/bitcoin/bitcoin/issues/21501
Related PR: https://github.com/bitcoin/bitcoin/pull/20080
Consider the wallet directories `w1` and `w2` saved in `D:\`. Run `bitcoind.exe -walletdir=D:\`, Results for `bitcoin-cli.exe listwalletdir`:
Before this PR:
```
{
"wallets": [
{
"name": "1"
},
{
"name": "2"
}
]
}
```
After this PR:
```
"wallets": [
{
"name": "w1"
},
{
"name": "w2"
}
]
}
```
ACKs for top commit:
ryanofsky:
Code review ACK d44a261acf
meshcollider:
utACK d44a261acf
Tree-SHA512: b09b00f727407e3771c8694861dae1bfd29d97a0d51ddcb5d9c0111dc618b3fff2f75829cbb4361c54457ee564e94fcefd9e2928262a1c918a2b6bbad724eb55
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
1c4b456e1a gui: send using external signer (Sjors Provoost)
24815c6309 gui: wallet creation detects external signer (Sjors Provoost)
3f845ea299 node: add externalSigners to interface (Sjors Provoost)
62ac119f91 gui: display address on external signer (Sjors Provoost)
450cb40a34 wallet: add displayAddress to interface (Sjors Provoost)
eef8d64529 gui: create wallet with external signer (Sjors Provoost)
6cdbc83e93 gui: add external signer path to options dialog (Sjors Provoost)
Pull request description:
Big picture overview in [this gist](https://gist.github.com/Sjors/29d06728c685e6182828c1ce9b74483d).
This PR adds GUI support for external signers, based on the since merged bitcoin/bitcoin#16546 (RPC).
The UX isn't amazing - especially the blocking calls - but it works.
First we adds a GUI setting for the signer script (e.g. path to HWI):
<img width="625" alt="Schermafbeelding 2019-08-05 om 19 32 59" src="https://user-images.githubusercontent.com/10217/62483415-e1ff1680-b7b7-11e9-97ca-8d2ce54ca1cb.png">
Then we add an external signer checkbox to the wallet creation dialog:
<img width="374" alt="Schermafbeelding 2019-11-07 om 19 17 23" src="https://user-images.githubusercontent.com/10217/68416387-b57ee000-0194-11ea-9730-127d60273008.png">
It's checked by default if HWI detects a device. It also grabs the name. It then creates a fresh wallet and imports the keys.
You can verify an address on the device (blocking...):
<img width="673" alt="Schermafbeelding 2019-08-05 om 19 29 22" src="https://user-images.githubusercontent.com/10217/62483560-43bf8080-b7b8-11e9-9902-8a036116dc4b.png">
Sending, including coin selection, Just Works(tm) as long the device is present.
~External signer support is enabled by default when the GUI is configured and Boost::Process is present.~
External signer support remains disabled by default, see https://github.com/bitcoin/bitcoin/pull/21935.
ACKs for top commit:
achow101:
Code Review ACK 1c4b456e1a
hebasto:
ACK 1c4b456e1a, tested on Linux Mint 20.1 (Qt 5.12.8) with HWW `2.0.2-rc.1`.
promag:
Tested ACK 1c4b456e1a but rebased with e033ca1379, with HWI 2.0.2, with Nano S and Nano X.
meshcollider:
re-code-review ACK 1c4b456e1a
Tree-SHA512: 3503113c5c69d40adb6ce364d8e7cae23ce82d032a00474ba9aeb6202eb70f496ef4a6bf2e623e5171e524ad31ade7941a4e0e89539c64518aaec74f4562d86b
2667366aaa tests: check derivation of P2TR (Pieter Wuille)
7cedafc541 Add tr() descriptor (derivation only, no signing) (Pieter Wuille)
90fcac365e Add TaprootBuilder class (Pieter Wuille)
5f6cc8daa8 Add XOnlyPubKey::CreateTapTweak (Pieter Wuille)
2fbfb1becb Make consensus checking of tweaks in pubkey.* Taproot-specific (Pieter Wuille)
a4bf84039c Separate WitnessV1Taproot variant in CTxDestination (Pieter Wuille)
41839bdb89 Avoid dependence on CTxDestination index order (Pieter Wuille)
31df02a070 Change Solver() output for WITNESS_V1_TAPROOT (Pieter Wuille)
4b1cc08f9f Make XOnlyPubKey act like byte container (Pieter Wuille)
Pull request description:
This is a subset of #21365, to aide review.
This adds support `tr(KEY)` or `tr(KEY,SCRIPT)` or `tr(KEY,{{S1,{{S2,S3},...}},...})` descriptors, describing Taproot outputs with specified internal key, and optionally any number of scripts, in nested groups of 2 inside `{`/`}` if there are more than one. While it permits importing `tr(KEY)`, anything beyond that is just laying foundations for more features later.
Missing:
* Signing support (see #21365)
* Support for more interesting scripts inside the tree (only `pk(KEY)` is supported for now). In particular, a multisig policy based on the new `OP_CHECKSIGADD` opcode would be very useful.
* Inferring `tr()` descriptors from outputs (given sufficient information).
* `getaddressinfo` support.
* MuSig support. Standardizing that is still an ongoing effort, and is generally kind of useless without corresponding PSBT support.
* Convenient ways of constructing descriptors without spendable internal key (especially ones that arent't trivially recognizable as such).
ACKs for top commit:
Sjors:
utACK 2667366 (based on https://github.com/bitcoin/bitcoin/pull/21365#issuecomment-846945215 review, plus the new functional test)
achow101:
Code Review ACK 2667366aaa
lsilva01:
Tested ACK 2667366aaa
meshcollider:
utACK 2667366aaa
Tree-SHA512: 61046fef22c561228338cb178422f0b782ef6587ec8208d3ce2bd07afcff29a664b54b35c6b01226eb70b6540b43f6dd245043d09aa6cb6db1381b6042667e75
f5ba424cd4 wallet: Add IsAddressUsed / SetAddressUsed methods (Russell Yanofsky)
62252c95e5 interfaces: Stop exposing wallet destdata to gui (Russell Yanofsky)
985430d9b2 test: Add gui test for wallet receive requests (Russell Yanofsky)
Pull request description:
Stop giving GUI access to destdata rows in database. Replace with narrow API just for saving and reading receive request information.
This simplifies code and should prevent the GUI from interfering with other destdata like address-used status. It also adds some more GUI test coverage.
There are no changes in behavior.
ACKs for top commit:
jarolrod:
tACK f5ba424cd4
laanwj:
Code review ACK f5ba424cd4
Tree-SHA512: 5423df4786e537a59013cb5bfb9e1bc29a7ca4b8835360c00cc2165a59f925fdc355907a4ceb8bca0285bb4946ba235bffa7645537a951ad03fd3b4cee17b6b0
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-
It's unnecessary to fill in the vin with dummy inputs, calculate the
fee, then fill in the vin with the actual inputs. Just fill the vin with
the actual inputs the first time.
- txNew nLockTime setting to txNew init
- FeeCalc to the fee estimation fetching
- setCoins to prior to SelectCoins
- nBytes to CalculateMaximumSignedTxSize call
- tx_sizes to CalculateMaximumSignedTxSize call
- coin_selection_params.m_avoid_partial_spends to params init
Ensuring that the recipients vector is not empty and that the amounts
are non-negative can be done in CreateTransaction rather than
CreateTransactionInternal. Additionally, these checks should happen as
soon as possible, so they are done at the beginning of
CreateTransaction.
nValue is the sum of the intended recipient amounts, so name it that for
clarity.
nValueToSelect is the coin selection target value, so name it
selection_target for clarity.
c7bd5842e4 MOVEONLY: CWallet transaction code out of wallet.cpp/.h (Russell Yanofsky)
Pull request description:
This commit just moves function without making any changes. It can be reviewed with `git log -p -n1 --color-moved=dimmed_zebra`
Motivation for this change is to make `wallet.cpp/h` less monolithic and start to make wallet transaction state tracking comprehensible so bugs in https://github.com/bitcoin-core/bitcoin-devwiki/wiki/Wallet-Transaction-Conflict-Tracking can be fixed safely without introducing new problems.
This moves wallet classes and methods that deal with transactions out of `wallet.cpp/.h` into better organized files:
- `transaction.cpp/.h` - CWalletTx and CMerkleTx class definitions
- `receive.cpp/.h` - functions checking received transactions and computing balances
- `spend.cpp/.h` - functions creating transactions and finding spendable coins
After #20773, when loading is separated from syncing it will also be possible to move more `wallet.cpp/.h` functions to:
- `sync.cpp/.h` - functions handling chain notifications and rescanning
This commit arranges `receive.cpp` and `spend.cpp` functions in dependency order so it's possible to skim `receive.cpp` and get an idea of how computing balances works, and skim `spend.cpp` and get an idea of how transactions are created, without having to jump all over `wallet.cpp` where functions are not in order and there is a lot of unrelated code.
Followup commit "refactor: Detach wallet transaction methods" in https://github.com/bitcoin/bitcoin/pull/21206 follows up this PR and tweaks function names and arguments to reflect new locations. The two commits are split into separate PRs because this commit is more work to maintain and less work to review, while the other commit is less work to maintain and more work to review, so hopefully this commit can be merged earlier.
ACKs for top commit:
Sjors:
re-utACK c7bd5842e4
fjahr:
utACK c7bd5842e4
promag:
Code review ACK c7bd5842e4, verified move only claim.
meshcollider:
Dimmed-zebra-check and functional test run ACK c7bd5842e4
Tree-SHA512: 4981de6911cb1196774db375494355cc9af59b52456129c002d264a77cd9ed6175f8ecbb6b2f492a59a4d5a0def21a39d96fa79c9f4d99be0992985f553be32f
This commit just moves functions without making any changes. It can be
reviewed with `git log -p -n1 --color-moved=dimmed_zebra`
Motivation for this change is to make wallet.cpp/h less monolithic and
start to make wallet transaction state tracking comprehensible so bugs
in
https://github.com/bitcoin-core/bitcoin-devwiki/wiki/Wallet-Transaction-Conflict-Tracking
can be fixed safely without introducing new problems.
This commit moves wallet classes and methods that deal with transactions
out of wallet.cpp/.h into better organized files:
- transaction.cpp/.h - CWalletTx and CMerkleTx class definitions
- receive.cpp/.h - functions checking received transactions and computing balances
- spend.cpp/.h - functions creating transactions and finding spendable coins
After #20773, when loading is separated from syncing it will also be
possible to move more wallet.cpp/.h functions to:
- sync.cpp/.h - functions handling chain notifications and rescanning
This commit arranges receive.cpp and spend.cpp functions in dependency
order so it's possible to skim receive.cpp and get an idea of how
computing balances works, and skim spend.cpp and get an idea of how
transactions are created, without having to jump all over wallet.cpp
where functions are not in order and there is a lot of unrelated code.
Followup commit "refactor: Detach wallet transaction methods" in
https://github.com/bitcoin/bitcoin/pull/21206 follows up this PR and
tweaks function names and arguments to reflect new locations. The two
commits are split into separate PRs because this commit is more work to
maintain and less work to review, while the other commit is less work to
maintain and more work to review, so hopefully this commit can be merged
earlier.
e6fe1c37d0 rpc: Improve avoidpartialspends and avoid_reuse documentation (Fabian Jahr)
8f073076b1 wallet: Increase OUTPUT_GROUP_MAX_ENTRIES to 100 (Fabian Jahr)
Pull request description:
Follow-up to #17824.
This increases OUTPUT_GROUP_MAX_ENTRIES to 100 which means that OutputGroups will now be up to 100 outputs large, up from previously 10. The main motivation for this change is that during the PR review club on #17824 [several participants signaled](https://bitcoincore.reviews/17824.html#l-339) that 100 might be a better value here.
I think fees should be manageable for users but more importantly, users should know what they can expect when using the wallet with this configuration, so I also tried to clarify the documentation on `-avoidpartialspends` and `avoid_reuse` a bit. If there are other additional ways how or docs where users can be made aware of the potential consequences of using these parameters, please let me know. Another small upside is that [there seem to be a high number of batching transactions with 100 and 200 inputs](https://miro.medium.com/max/3628/1*sZ5eaBSbsJsHx-J9iztq2g.png)([source](https://medium.com/@hasufly/an-analysis-of-batching-in-bitcoin-9bdf81a394e0)) giving these transactions a bit of a larger anonymity set, although that is probably a very weak argument.
ACKs for top commit:
jnewbery:
ACK e6fe1c37d0
Xekyo:
retACK e6fe1c37d0
rajarshimaitra:
tACK `e6fe1c3`
achow101:
ACK e6fe1c37d0
glozow:
code review ACK e6fe1c37d0
Tree-SHA512: 79685c58bafa64ed8303b0ecd616fce50fc9a2b758aa79833e4ad9f15760e09ab60c007bc16ab4cbc4222e644cfd154f1fa494b0f3a5d86faede7af33a6f2826
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 51a3ac242c
meshcollider:
re-light-utACK 51a3ac242c
Tree-SHA512: 372c27e00edcd5dbf85177421ba88f20bfdaf1791b6e3dc022c44876ecc379403e2375ed69e71c512c49e6af87641001ff385c4b25ab93684b3a08a53bf3824e
4f504f826b rpc: fix code comment for bumpfee/psbtbumpfee output (Jon Atack)
5cb7ac23fb rpc: fix docs for bumpfee psbt update (Jon Atack)
Pull request description:
Follow-up to #21544 and #20891 for the `bumpfee_helper` used for RPCs bumpfee and psbtbumpfee:
- "psbt" field is only returned in psbtbumpfee and not bumpfee
- bumpfee raises if private keys are disabled, so the txid help "Only returned when wallet private keys are enabled." no longer makes sense; remove it
- add missing space in RPC examples ("Bump the fee, get the new transaction'stxid")
- update txid/psbt code comments
ACKs for top commit:
klementtan:
ACK [`4f504f8`](4f504f826b)
Tree-SHA512: 194faf8af52383eb8ac5cd22825265931bcde135dac79d8ecc4f84f698070da9b9373c00eef8623961881bb293157c7c9a0d71d1bcccf481ae3605a2d1444ed8
- "psbt" field is only returned in psbtbumpfee and not bumpfee
- bumpfee raises if privkeys are disabled, so drop "Only returned when wallet private keys are enabled."
- add missing space in RPC example
6e2eb0d63b rpc/wallet: use OMITTED_NAMED_ARG instead of Default(VNULL) (Karl-Johan Alm)
4983f4cba4 rpc/createwallet: omitted named arguments (Karl-Johan Alm)
dc4db23b30 rpc: address:amount dictionaries are OBJ_USER_KEYS (Karl-Johan Alm)
c8cf0a3d51 rpc/getpeerinfo: bytesrecv_per_msg is a dynamic dictionary (Karl-Johan Alm)
eb4fb7e507 rpc/gettxoutsetinfo: hash_or_height is a named argument (Karl-Johan Alm)
Pull request description:
This is a follow-up to #21897, and I believe covers the remaining cases, at least that I could find.
Edited to remove unrelated information about a side project.
ACKs for top commit:
laanwj:
Documentation diff ACK 6e2eb0d63b
promag:
Code review ACK 6e2eb0d63b.
Tree-SHA512: d26f6e074e13d64bbca2a114a0adc7f905d47d238c4e9bc49f70ca0b775afbebf9879fc3794ab29dc316a6dbd00ba8cbeb01197e236ee4ab2e9854db25f23f04
Instead of hijacking the effective_feerate to use the correct value
during coin selection, have OutputGroup be aware of whether we are
subtracting the fee from the outputs and provide the correct value to
use for selection.
To do this, OutputGroup now takes CoinSelectionParams and has a new
function GetSelectionAmount().
This was originally modified to use SelectCoinsMinConf in order to test
both BnB and Knapsack at the same time. But since SelectCoins does both
now, this is no longer necessary and we can revert back to actually
testing SelectCoins.
Remove the CreateTransaction while loop. Removes variables that were
only needed because of that loop. Also renames a few variables and
moves their declarations to where they are used.
Some subtractFeeFromOutputs handling is moved to after coin selection
in order to reduce their amounts once the fee is known.
If subtracting the fee reduces the change to dust, we will also now
remove the change output
Although the CreateTransaction loop currently remains, it should be
largely unused. KnapsackSolver will now account for transaction fees
when doing its selection.
In the previous commit, SelectCoinsMinConf was refactored to have some
calculations become shared for KnapsackSolver and SelectCoinsBnB. In
this commit, KnapsackSolver will now use the not_input_fees and
effective_feerate so that it include the fee for non-input things
(excluding a change output) so that the algorithm will select enough to
cover those fees. This is necessary for selecting on effective values.
Additionally, the OutputGroups
created for KnapsackSolver will actually have their effective values
calculated and set, and KnapsackSolver will do its selection on those
effective values.
Lastly, SelectCoins is modified to use the same value for preselected
inputs for BnB and KnapsackSolver. While it will still use the real
value when subtracting the fee from outputs, this behavior will be
the same regardless of the algo used for selecting additional inputs.
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.
Simplifies CreateTransactionInternal without changing behavior. Removes
the pick_new_inputs variable by moving the subtract fee from amount
implementation to later in the loop to where it is possible to calculate
the fee for the transaction. This allows the fee to be subtracted from
the outputs within a single iteration, instead of calculating the fee in
the first iteration, and subtracting the fee in the second.
This also removes another scenario where a second iteration of the loop
finds a smaller input set (and thus smaller fees than the first
iteration) with no change and so a third iteration of the loop is done in order to make
a change output that contains the excess fees.
To handle these cases, we always create a change output which contains
the difference between selected input values and the recipient amounts.
Once the transaction fee is calculated, the change output is reduced (in
the normal case) or the recipient amounts are reduced (in the subtract
fee from amount case). All of this is done in a single iteration of the
loop.
