5252f86eb6 fuzz: Reduce maintenance requirements by allowing RPC annotations also for conditionally available RPC commands (such as wallet commands) without the fragility of #ifdef forests (practicalswift)
54549dda31 fuzz: RPC fuzzer post-merge follow-ups. Remove unused includes. Update list of fuzzed RPC commands. (practicalswift)
Pull request description:
Various RPC fuzzer follow-ups:
* Remove unused includes.
* Update list of fuzzed RPC commands.
* Reduce maintenance requirements by allowing RPC annotations also for conditionally available RPC commands (such as wallet commands) without the fragility of `#ifdef` forests.
Context: https://github.com/bitcoin/bitcoin/pull/21169#pullrequestreview-646723483
ACKs for top commit:
MarcoFalke:
Concept ACK 5252f86eb6
Tree-SHA512: 286d70798131706ffb157758e1c73f7f00ed96ce120c7d9dc849e672b283f1362df47b206cfec9da44d5debb5869225e721761dcd5c38a7d5d1019dc6c912ab2
We already attest to the relevant dist-archive in inputs.SHA256SUMS,
which is recorded at build-time.
We use a SKIPATTEST.TAG file to indicate output directories which do not
require attestation (much like the CACHEDIR.TAG specification).
Generally, it's better to have build scripts declare properties of
directories instead of introducing name-based special cases in attest
scripts since build scripts have a more detailed context of what is
going on.
At build/codesigning-time, hash build inputs and output the digest to
${OUTDIR}/inputs.SHA256SUMS, which gets included in the final SHA256SUMS
constructed by guix-attest.
Example final SHA256SUMS:
ee832d2a35b7701bff581dea05a536118b118e3ad0a587a2855b6ee8cd6fba20 inputs/bitcoin-78199266af7b.tar.gz
ca765e70a0c12866dd63c0be228b675278a26329e5f8f5b5c52fd09200fedf21 bitcoin-78199266af7b-powerpc64le-linux-gnu-debug.tar.gz
dae95327d7f2c324e2728c4b73627be6cb2c0d2f2e5bea940d1d5e6463939327 bitcoin-78199266af7b-powerpc64le-linux-gnu.tar.gz
While files are being output to $OUTDIR, it will be under
${DISTSRC}/output, and only when everything is done, will
${DISTSRC}/output be moved to the actual $OUTDIR.
This makes it so that a Ctrl-C in the middle of a build is less likely
to result in a partially-constructed $OUTDIR. In fact, if I understand
correctly, if $OUTDIR and $DISTSRC reside on the same filesystem, the
move (rename) is likely atomic.
Also, since the "working $OUTDIR" is under ${DISTSRC}/output, it will be
cleaned properly by the guix-clean script.
48bd9ebc5d script: update REVIEWERS (Adam Jonas)
Pull request description:
Meta: `git show a06eb03` indicates the commit was first made one year ago and the PR was merged in September 2020.
It might be time for an update, if automated review requests via DrahtBot are operational.
"Regular contributors are free to add their names to specific directories or files provided that they are willing to provide a review."
Perhaps we can compile and squash suggested updates here, per that guideline.
ACKs for top commit:
practicalswift:
ACK 48bd9ebc5d: patch looks correct :)
adamjonas:
ACK 48bd9eb. I ran through the new files added with `*`s using `git ls-files`.
hebasto:
ACK 48bd9ebc5d, I have reviewed the code and it looks OK, I agree it can be merged.
Tree-SHA512: 770fd60d78bd64100814c72211e4bd97718c87a55a4c91e136fababa0149d12cdc1b3e89462dee3a55e5f51d3263420cc6d856868469eb9a72ff8c930859c54f
It does not matter if the tests fail due to a BOOST_CHECK failure or
due to a thrown exception. Prefer the exception because it is less
code.
Example fail with the throwing accessor:
unknown location(0): fatal error: in "script_standard_tests/script_standard_ExtractDestinations": std::bad_variant_access: std::get: wrong index for variant
test/script_standard_tests.cpp(314): last checkpoint
*** 1 failure is detected in the test module "Bitcoin Core Test Suite"
fac96d0265 p2p: Limit m_block_inv_mutex (MarcoFalke)
Pull request description:
Keeping the lock longer than needed is confusing to reviewers and thread analysis. For example, keeping the lock while appending tx-invs, which requires the mempool lock, will tell thread analysis tools an incorrect lock order of `(1) m_block_inv_mutex, (2) pool.cs`.
ACKs for top commit:
Crypt-iQ:
crACK fac96d0265
jnewbery:
utACK fac96d0265
theStack:
Code-Review ACK fac96d0265
Tree-SHA512: fcfac0f1f8b16df7522513abf716b2eed3d2fc9153f231c8cb61f451e342f29c984a5c872deca6bab3e601e5d651874cc229146c9370e46811b4520747a21f2b
9096b13a47 net: remove unnecessary check of CNode::cs_vSend (Vasil Dimov)
Pull request description:
It is not possible to have a node in `CConnman::vNodesDisconnected` and
its reference count to be incremented - all `CNode::AddRef()` are done
either before the node is added to `CConnman::vNodes` or while holding
`CConnman::cs_vNodes` and the object being in `CConnman::vNodes`.
So, the object being in `CConnman::vNodesDisconnected` and its reference
count being zero means that it is not and will not start to be used by
other threads.
So, the lock of `CNode::cs_vSend` in `CConnman::DisconnectNodes()` will
always succeed and is not necessary.
Indeed all locks of `CNode::cs_vSend` are done either when the reference
count is >0 or under the protection of `CConnman::cs_vNodes` and the
node being in `CConnman::vNodes`.
ACKs for top commit:
MarcoFalke:
review ACK 9096b13a47🏧
jnewbery:
utACK 9096b13a47
Tree-SHA512: 910899cdcdc8934642eb0c40fcece8c3b01b7e20a0b023966b9d6972db6a885cb3a9a04e9562bae14d5833967e45e2ecb3687b94d495060c3da4b1f2afb0ac8f
a5550f877a build: use -stdlib++-isystem with Clang 10 (fanquake)
51d9d1607f guix: use Clang 10 for the macOS cross compile (fanquake)
b80a6af9e5 build: no longer patch threading out of ld64 (fanquake)
c29cba44b3 build: Xcode 12.1, macOS SDK 10.15.6 (fanquake)
9ed2f19d38 build: native cctools 973.0.1, ld64 609 (fanquake)
f48f187cce build: Clang 10.0.1 (Hennadii Stepanov)
9b193cd2a3 build: libtapi 1100.0.11 (fanquake)
Pull request description:
Bumps our macOS toolchain to be using the following:
* Clang 10.0.1 (gitian) & Clang 10.0.0 (Guix)
* ld64 609
* libtapi 1100.0.11
* cctools 973.0.1
* Xcode 12.1
* macOS SDK 10.15.6
which are currently the most recent releases available as open source. See upstream [`cctools`](https://github.com/tpoechtrager/cctools-port/commits/973.0.1-ld64-609) and [`libtapi`](https://github.com/tpoechtrager/apple-libtapi/tree/1100.0.11).
This should improve the possibility of Apple ARM cross-compilation in depends.
This also removes our [patching out of pthreads usage](https://github.com/bitcoin/bitcoin/blob/master/depends/patches/native_cctools/ld64_disable_threading.patch) in `ld64`. There have been multiple changes since `ld64 450.3`, which have likely fixed the non-determinism we were working around. i.e from [InputFiles.cpp](https://opensource.apple.com/source/ld64/ld64-609/src/ld/InputFiles.cpp.auto.html):
```cpp
// <rdar://problem/15002251> make implicit dylib order be deterministic by sorting by install_name
std::sort(implicitDylibs.begin(), implicitDylibs.end(), DylibByInstallNameSorter());
```
```cpp
// <rdar://problem/42675402> ld64 output is not deterministic due to dylib processing order
std::sort(unprocessedDylibs.begin(), unprocessedDylibs.end(), [](const ld::dylib::File* lhs, const ld::dylib::File* rhs) {
return strcmp(lhs->path(), rhs->path()) < 0;
});
```
Guix Build:
```bash
find guix-build-$(git rev-parse --short=12 HEAD)/output/ -type f -print0 | env LC_ALL=C sort -z | xargs -r0 sha256sum
f6c3817b8fe5f7370299d1ae2533e4a3acd313ba9f9aa8d423a8956117e52dd5 guix-build-a5550f877a2c/output/dist-archive/bitcoin-a5550f877a2c.tar.gz
4954dcf563c2d496b8d9fecd48f8e3f7fba2f319ffa254a5bc8ee12cfee6acf0 guix-build-a5550f877a2c/output/x86_64-apple-darwin18/bitcoin-a5550f877a2c-osx-unsigned.dmg
8f6095b445c7f1a8e6accd86bb7f0696d5849402084927d2b726b7d557831c3a guix-build-a5550f877a2c/output/x86_64-apple-darwin18/bitcoin-a5550f877a2c-osx-unsigned.tar.gz
cc40f25477b4defc1617ae694313d80f307ddf6742fe6cc85c6bc0e215ef8be0 guix-build-a5550f877a2c/output/x86_64-apple-darwin18/bitcoin-a5550f877a2c-osx64.tar.gz
```
Gitian Build:
```bash
Generating report
506a8abdefe559999b43dd9f14905b9b2b5a3363b1cd013d45ae47acc1f7ef6c bitcoin-a5550f877a2c-osx-unsigned.dmg
f606997f74026dd12d110d683c6f116b40df324836904ef507dd7ac787e6ebe2 bitcoin-a5550f877a2c-osx-unsigned.tar.gz
5b495ef15f2c3260c2950921b61326912a9bf533cccd51e13818809fd225489e bitcoin-a5550f877a2c-osx64.tar.gz
f6c3817b8fe5f7370299d1ae2533e4a3acd313ba9f9aa8d423a8956117e52dd5 src/bitcoin-a5550f877a2c.tar.gz
9eb0221e962d2839770963bd03c6c9e98e8bf3078566bee2ae42f06233a710fa bitcoin-core-osx-22-res.yml
Done.
```
ACKs for top commit:
hebasto:
ACK a5550f877a
Tree-SHA512: 504c4b0f9cd3b939714a322298320c5bde07e9356a48a9a000060b36f8dce4d6134ed60c3a5188810476a28ec5b108733eabbc6fb8053231b9ea8a494cc91b12
ebd4be43cc doc: add release notes for 20867 (Antoine Poinsot)
5aa50ab9cc rpc/util: multisig: only check redeemScript size is <= 520 for P2SH (Antoine Poinsot)
063df9e897 test/functional: standardness sanity checks for P2(W)SH multisig (Antoine Poinsot)
ae0429d3af script: allow up to 20 keys in wsh() descriptors (Antoine Poinsot)
9fc68faf35 script: match multisigs with up to MAX_PUBKEYS_PER_MULTISIG keys (Antoine Poinsot)
Pull request description:
As described in https://github.com/bitcoin/bitcoin/issues/20620 multisigs are currently limited to 16 keys in descriptors and RPC helpers, even for P2WSH and P2SH-P2WSH.
This adds support for multisig with up to 20 keys (which are already standard) for Segwit v0 context for descriptors (`wsh()`, `sh(wsh())`) and RPC helpers.
Fixes https://github.com/bitcoin/bitcoin/issues/20620
ACKs for top commit:
meshcollider:
re-utACK ebd4be43cc
instagibbs:
re-ACK ebd4be43cc
Tree-SHA512: 36141f10a8288010d17d5c4fe8d24878bcd4533b88a8aba3a44fa8f74ceb3182d70fee01427e0ab7f53ce7fab46c88c1cd3ac3b18ab8a10bd4a6b8b74ed79e46
0a331456e4 Remove Visual Studio 2017 reference from readme (Aaron Clauson)
Pull request description:
This PR was motivated by a comment in GUI PR (257) regarding a suggested improvement not being supported by VS2017.
When checking whether master can still be built with the VS2017 toolset ABI issues were encountered. Most likely due to the pre-compiled Qt binaries that are used.
It does not seem worth the effort to try and support VS2017, which would most likely require additional Qt binaries, or lengthy instructions on how to build static Qt binaries on Windows (which is very error prone and tedious).
ACKs for top commit:
jarolrod:
ACK 0a331456e4
hebasto:
ACK 0a331456e4, I have reviewed the code and it looks OK, I agree it can be merged.
Tree-SHA512: 17e2326cd55a5cc3fc13622ba72bb44f9b9d55070cb77941df5fd943cd3f1baf02b9ac9504dfae5941c522748ad7b584c3c8da03fb323a87b3639eb926ce1699
+ Detailed error messages for invalid address
+ Used `IsValidDestination` instead of `IsValidDestinationString`
+ Referred to https://github.com/bitcoin/bitcoin/pull/20832 for solution
This PR was motivated by a comment in GUI PR (257) regarding a suggested improvement not being supported by VS2017.
When checking whether master can still be built with the VS2017 toolset ABI issues were encountered. Most likely due to the pre-compiled Qt binaries that are used.
It does not seem worth the effort to try and support VS2017, which would most likely require additional Qt binaries, or lengthy instructions on how to build static Qt binaries on Windows (which is very error prone and tedious).
Added advisory note about build not working with earlier Visual Studio versions.
Fixed grammar.
c5bb142817 test: resolve bug in test/functional/interface_bitcoin_cli.py - Test -getinfo with -rpcwallet=unloaded wallet returns no balances (klementtan)
Pull request description:
I think there is a bug in this test case where the new value of `cli_get_info` is not asserted.
ACKs for top commit:
jonatack:
ACK c5bb142817
Tree-SHA512: 50c0c2c8fe63c95f951dee892fbacedf92208f47efe5ed481fbb255f15137c799d9200fa3ff31a442df0691248d7ff04d899842722c3032cd7f35553622ba38c
cf971c9ea0 build: use -isysroot over --sysroot on macOS (fanquake)
Pull request description:
Not only does this seem to be the more correct behaviour when targeting Darwin, but if you use `-isysroot`, Clangs Darwin driver will [infer the deployment target](3e037f8f0e/clang/lib/Driver/ToolChains/Darwin.cpp (L1652)) from the SDK and use other SDK info when parsing arguments to the linker. In the case of [`-platform_version`](3e037f8f0e/clang/lib/Driver/ToolChains/Darwin.cpp (L2656)), which is added if the linker is [new enough](3e037f8f0e/clang/lib/Driver/ToolChains/Darwin.cpp (L342)), the version tuple is constructed from the SDKInfo, and SDKInfo, as far as I can tell, only exists when `-isysroot` has been passed, see [parseSDKSettings](3e037f8f0e/clang/lib/Driver/ToolChains/Darwin.cpp (L1785))
As a result, the SDK version field in the `LC_BUILD_VERSION` command is filled out. i.e when building master:
```bash
cmd LC_BUILD_VERSION
cmdsize 32
platform 1
minos 10.14
sdk n/a
ntools 1
tool 3
version 650.9
```
vs this PR:
```bash
cmd LC_BUILD_VERSION
cmdsize 32
platform 1
minos 10.14
sdk 11.3
ntools 1
tool 3
version 650.9
```
This, from what I understand, will fix the issue we are having with Qt deciding wether or not to enable features like "Dark mode" on macOS, see #21771, however I have not tested that. Thus this is an alternative to #21782.
Our usage of `--sysroot` was added in #17118.
```bash
-isysroot <dir> Set the system root directory (usually /)
```
ACKs for top commit:
Sjors:
tACK cf971c9ea0
hebasto:
re-ACK cf971c9ea0, only rebased and addressed comments since my [previous](https://github.com/bitcoin/bitcoin/pull/21793#pullrequestreview-647321518) review.
Tree-SHA512: f01138179fb85083b5505bbaa48810451098ffa4da5d3c9b673785448790aa76f2e64b2aab6e698f6ee378a21f70626445a3fabee7c61dbfc44e96f3e3964656
e94920a0bb qt: peertableview alternating row colors (randymcmillan)
Pull request description:
peers-tab: enable alternating row colors for peer table and banned table
ACKs for top commit:
Bosch-0:
tACK e94920a0bb on Windows 10 - works as intended. Before / after below:
jarolrod:
tACK e94920a0bb
Tree-SHA512: 05ba18e1db9700bbd68644fe02292409f4e5c52e301b1b2977c335d1ff16456a93fb0b15c8c8385d1b15f648141341990706d530f6b08ecb33098fa941b9af1f
