Age | Commit message (Collapse) | Author |
|
|
|
d5a715536e497c160a2520f81334aab6c7490213 build: remove boost::process dependency for building external signer support (Sebastian Falbesoner)
70434b1c443d9251a880d0193af771f574c40617 external_signer: replace boost::process with cpp-subprocess (Sebastian Falbesoner)
cc8b9875b104c31f0a5b5e4195a8278ec55f35f7 Add `cpp-subprocess` header-only library (Hennadii Stepanov)
Pull request description:
Closes https://github.com/bitcoin/bitcoin/issues/24907.
This PR is based on **theStack**'s [work](https://github.com/bitcoin/bitcoin/issues/24907#issuecomment-1466087049).
The `subprocess.hpp` header has been sourced from the [upstream repo](https://github.com/arun11299/cpp-subprocess) with the only modification being the removal of convenience functions, which are not utilized in our codebase.
Windows-related changes will be addressed in subsequent follow-ups.
ACKs for top commit:
achow101:
reACK d5a715536e497c160a2520f81334aab6c7490213
Sjors:
re-tACK d5a715536e497c160a2520f81334aab6c7490213
theStack:
Light re-ACK d5a715536e497c160a2520f81334aab6c7490213
fanquake:
ACK d5a715536e497c160a2520f81334aab6c7490213 - with the expectation that this code is going to be maintained as our own. Next PRs should:
Tree-SHA512: d7fb6fecc3f5792496204190afb7d85b3e207b858fb1a75efe483c05260843b81b27d14b299323bb667c990e87a07197059afea3796cf218ed8b614086bd3611
|
|
timingsafe_bcmp rather than libc's
2d1819455cb4c516f6cdf81c11e869a23dee3e6b crypto: chacha20: always use our fallback timingsafe_bcmp rather than libc's (Cory Fields)
Pull request description:
Looking at libc sources, apple and openbsd implementations match our naive fallback. Only FreeBSD (and only x86_64) seems to [implement an optimized version](https://github.com/freebsd/freebsd-src/blob/main/lib/libc/amd64/string/timingsafe_bcmp.S).
It's not worth the hassle of using a platform-specific function for such little gain.
Additionally, as mentioned below, this is the only case outside of sha2 that requires an autoconf check, and I have upcoming PRs to remove the sha2 ones.
Apple's [impl is unoptimized](https://opensource.apple.com/source/Libc/Libc-1244.1.7/string/FreeBSD/timingsafe_bcmp.c.auto.html).
As-is [OpenBSD's impl](https://github.com/openbsd/src/blob/master/lib/libc/string/timingsafe_bcmp.c).
Relevant IRC conversation with sipa:
> \<cfields\> sipa: chacha20poly1305.cpp uses libc's timingsafe_bcmp when possible. But looking around at apple/freebsd/openbsd, I don't see any impl that doesn't use the naive implementation that matches our fallback...
> \<cfields\> is there any reason to belive there's an optimized impl somewhere that we're actually hitting?
> \<cfields\> asking because after cleaning up sha2, timingsafe_bcmp is the last autoconf check that remains in all of crypto. It'd make life easy if we could just always use our internal one.
> \<cfields\> *all of crypto/
> \<sipa\> cfields: let's get rid of the dependency then
> \<sipa\> it's a trivial function
> \<sipa\> and if we need it for some platforms, no real reason not to use it on all
After the above discusstion, I did end up finding the x86_64-optimized FreeBSD impl, but I don't think that's all that significant.
ACKs for top commit:
sipa:
utACK 2d1819455cb4c516f6cdf81c11e869a23dee3e6b
fanquake:
ACK 2d1819455cb4c516f6cdf81c11e869a23dee3e6b
TheCharlatan:
ACK 2d1819455cb4c516f6cdf81c11e869a23dee3e6b
theStack:
ACK 2d1819455cb4c516f6cdf81c11e869a23dee3e6b
Tree-SHA512: b9583e19ac2f77c5d572aa5b95bc4b53669d5717e5708babef930644980de7c5d06a9c7decd5c2b559d70b8597328ecfe513375e3d8c3ef523db80012dfe9266
|
|
fa9f36babaceba6ab2f88e64bc4bc2956f58871f build: Remove HAVE_GMTIME_R (MarcoFalke)
fa72dcbfa56177ca878375bae7c7bca6ca6a1f40 refactor: FormatISO8601* without gmtime* (MarcoFalke)
fa2c486afc8501f2678cc19c9e9518a23c4ebcbd Revert "time: add runtime sanity check" (MarcoFalke)
Pull request description:
Now that the `ChronoSanityCheck` has passed for everyone with C++17 and is guaranteed by C++20 to always pass, remove it.
Also, remove `gmtime_r` and `gmtime_s` and replace them with `year_month_day`+`hh_mm_ss` from C++20.
ACKs for top commit:
sipa:
utACK fa9f36babaceba6ab2f88e64bc4bc2956f58871f
fanquake:
ACK fa9f36babaceba6ab2f88e64bc4bc2956f58871f - more std lib & even less stuff to port.
Tree-SHA512: a9e7e805b757b7dade0bcc3f95273a7dc4f68622630d74838339789dd203ad7542d36b2e090a93b2bc5a7ecc383207dd7ec82c68147108bdac7ce44f088c8c9a
|
|
Looking at apple/freebsd/openbsd sources, their implementations match our naive
fallback. It's not worth the hassle of using a platform-specific function for
no gain.
|
|
|
|
This was deprecated in v27.0, for removal in v28.0.
See discussion in PR #29189.
|
|
|
|
-onion
567cec9a05e1261e955535f734826b12341684b6 doc: add release notes and help text for unix sockets (Matthew Zipkin)
bfe51928911daf484ae07deb52a7ff0bcb2526ae test: cover UNIX sockets in feature_proxy.py (Matthew Zipkin)
c65c0d01630b44fa71321ea7ad68d5f9fbb7aefb init: allow UNIX socket path for -proxy and -onion (Matthew Zipkin)
c3bd43142eba77dcf1acd4984e437759f65e237a gui: accomodate unix socket Proxy in updateDefaultProxyNets() (Matthew Zipkin)
a88bf9dedd1d8c1db0a9c8b663dab3e3c2f0f030 i2p: construct Session with Proxy instead of CService (Matthew Zipkin)
d9318a37ec09fe0b002815a7e48710e530620ae2 net: split ConnectToSocket() from ConnectDirectly() for unix sockets (Matthew Zipkin)
ac2ecf3182fb5ad9bcd41540b19382376114d6ee proxy: rename randomize_credentials to m_randomize_credentials (Matthew Zipkin)
a89c3f59dc44eaf4f59912c1accfc0ce5d61933a netbase: extend Proxy class to wrap UNIX socket as well as TCP (Matthew Zipkin)
3a7d6548effa6cd9a4a5413b690c2fd85da4ef65 net: move CreateSock() calls from ConnectNode() to netbase methods (Matthew Zipkin)
74f568cb6fd5c74b7b9bf0ce69876430746a53b1 netbase: allow CreateSock() to create UNIX sockets if supported (Matthew Zipkin)
bae86c8d318d06818aa75a9ebe3db864197f0bc6 netbase: refactor CreateSock() to accept sa_family_t (Matthew Zipkin)
adb3a3e51de205cc69b1a58647c65c04fa6c6362 configure: test for unix domain sockets (Matthew Zipkin)
Pull request description:
Closes https://github.com/bitcoin/bitcoin/issues/27252
UNIX domain sockets are a mechanism for inter-process communication that are faster than local TCP ports (because there is no need for TCP overhead) and potentially more secure because access is managed by the filesystem instead of serving an open port on the system.
There has been work on [unix domain sockets before](https://github.com/bitcoin/bitcoin/pull/9979) but for now I just wanted to start on this single use-case which is enabling unix sockets from the client side, specifically connecting to a local Tor proxy (Tor can listen on unix sockets and even enforces strict curent-user-only access permission before binding) configured by `-onion=` or `-proxy=`
I copied the prefix `unix:` usage from Tor. With this patch built locally you can test with your own filesystem path (example):
`tor --SocksPort unix:/Users/matthewzipkin/torsocket/x`
`bitcoind -proxy=unix:/Users/matthewzipkin/torsocket/x`
Prep work for this feature includes:
- Moving where and how we create `sockaddr` and `Sock` to accommodate `AF_UNIX` without disturbing `CService`
- Expanding `Proxy` class to represent either a `CService` or a UNIX socket (by its file path)
Future work:
- Enable UNIX sockets for ZMQ (https://github.com/bitcoin/bitcoin/pull/27679)
- Enable UNIX sockets for I2P SAM proxy (some code is included in this PR but not tested or exposed to user options yet)
- Enable UNIX sockets on windows where supported
- Update Network Proxies dialog in GUI to support UNIX sockets
ACKs for top commit:
Sjors:
re-ACK 567cec9a05e1261e955535f734826b12341684b6
tdb3:
re ACK for 567cec9a05e1261e955535f734826b12341684b6.
achow101:
ACK 567cec9a05e1261e955535f734826b12341684b6
vasild:
ACK 567cec9a05e1261e955535f734826b12341684b6
Tree-SHA512: de81860e56d5de83217a18df4c35297732b4ad491e293a0153d2d02a0bde1d022700a1131279b187ef219651487537354b9d06d10fde56225500c7e257df92c1
|
|
These come from GUI code, and haven't/aren't being fixed, see discussion
in https://github.com/bitcoin-core/gui/issues/112. For now, just ignore
them entirely. Note that this only applies to ObjCXX code, so will not
hide any relevant warnings coming from C or CXX code (and they would be
unlikely in any case).
Alternative to #29362, which disables all compiler warnings, for macOS
builds in the CI.
Relevant output:
```bash
qt/macnotificationhandler.mm:27:9: warning: 'NSUserNotification' is deprecated: first deprecated in macOS 11.0 - All NSUserNotifications API should be replaced with UserNotifications.frameworks API [-Wdeprecated-declarations]
NSUserNotification* userNotification = [[NSUserNotification alloc] init];
^
/Library/Developer/CommandLineTools/SDKs/MacOSX.sdk/System/Library/Frameworks/Foundation.framework/Headers/NSUserNotification.h:24:12: note: 'NSUserNotification' has been explicitly marked deprecated here
@interface NSUserNotification : NSObject <NSCopying> {
^
qt/macnotificationhandler.mm:27:50: warning: 'NSUserNotification' is deprecated: first deprecated in macOS 11.0 - All NSUserNotifications API should be replaced with UserNotifications.frameworks API [-Wdeprecated-declarations]
NSUserNotification* userNotification = [[NSUserNotification alloc] init];
^
/Library/Developer/CommandLineTools/SDKs/MacOSX.sdk/System/Library/Frameworks/Foundation.framework/Headers/NSUserNotification.h:24:12: note: 'NSUserNotification' has been explicitly marked deprecated here
@interface NSUserNotification : NSObject <NSCopying> {
^
qt/macnotificationhandler.mm:30:11: warning: 'NSUserNotificationCenter' is deprecated: first deprecated in macOS 11.0 - All NSUserNotifications API should be replaced with UserNotifications.frameworks API [-Wdeprecated-declarations]
[[NSUserNotificationCenter defaultUserNotificationCenter] deliverNotification: userNotification];
^
/Library/Developer/CommandLineTools/SDKs/MacOSX.sdk/System/Library/Frameworks/Foundation.framework/Headers/NSUserNotification.h:118:12: note: 'NSUserNotificationCenter' has been explicitly marked deprecated here
@interface NSUserNotificationCenter : NSObject {
^
3 warnings generated.
```
|
|
|
|
Copied from https://github.com/bitcoin/bitcoin/pull/9979
Co-authored-by: laanwj <126646+laanwj@users.noreply.github.com>
|
|
modernization
86b7f28d6c507155a9d3a15487ee883989b88943 serialization: use internal endian conversion functions (Cory Fields)
432b18ca8d0654318a8d882b28b20af2cb2d2e5d serialization: detect byteswap builtins without autoconf tests (Cory Fields)
297367b3bb062c57142747719ac9bf2e12717ce9 crypto: replace CountBits with std::bit_width (Cory Fields)
52f9bba889fd9b50a0543fd9fedc389592cdc7e5 crypto: replace non-standard CLZ builtins with c++20's bit_width (Cory Fields)
Pull request description:
This replaces #28674, #29036, and #29057. Now ready for testing and review.
Replaces platform-specific endian and byteswap functions. This is especially useful for kernel, as it means that our deep serialization code no longer requires bitcoin-config.h.
I apologize for the size of the last commit, but it's hard to avoid making those changes at once.
All platforms now use our internal functions rather than libc or platform-specific ones, with the exception of MSVC.
Sadly, benchmarking showed that not all compilers are capable of detecting and optimizing byteswap functions, so compiler builtins are instead used where possible. However, they're now detected via macros rather than autoconf checks.
This[ matches how libc++ implements std::byteswap for c++23](https://github.com/llvm/llvm-project/blob/main/libcxx/include/__bit/byteswap.h#L26).
I suggest we move/rename `compat/endian.h`, but I left that out of this PR to avoid bikeshedding.
#29057 pointed out some irregularities in benchmarks. After messing with various compilers and configs for a few weeks with these changes, I'm of the opinion that we can't win on every platform every time, so we should take the code that makes sense going forward. That said, if any real-world slowdowns are caused here, we should obviously investigate.
ACKs for top commit:
maflcko:
ACK 86b7f28d6c507155a9d3a15487ee883989b88943 📘
fanquake:
ACK 86b7f28d6c507155a9d3a15487ee883989b88943 - we can finish pruning out the __builtin_clz* checks/usage once the minisketch code has been updated. This is more good cleanup pre-CMake & for the kernal.
Tree-SHA512: 715a32ec190c70505ffbce70bfe81fc7b6aa33e376b60292e801f60cf17025aabfcab4e8c53ebb2e28ffc5cf4c20b74fe3dd8548371ad772085c13aec8b7970e
|
|
1. It didn't actually disable asm usage in our code. Regardless of the setting,
asm is used in random.cpp and support/cleanse.cpp.
2. The value wasn't forwarded to libsecp as a user might have reasonably
expected.
3. We now have the DISABLE_OPTIMIZED_SHA256 define which is what disable-asm
actually did in practice.
If there is any desire, we can hook DISABLE_OPTIMIZED_SHA256 up to a new
configure option that actually does what it says.
|
|
detection with c++20 concept
ad7584d8b60119ca3717117a1eb6a16d753c5d74 serialization: replace char-is-int8_t autoconf detection with c++20 concept (Cory Fields)
Pull request description:
Doesn't depend on #29263, but it's really only relevant after that one's merged.
This removes the only remaining autoconf macro in our serialization code (after #29263), so it can now be used trivially and safely out-of-tree.
~Our code does not currently contain any concepts, but couldn't find any discussion or docs about avoiding them. I guess we'll see if this blows up our c-i.~
Edit: Ignore this. ajtowns pointed out that we're already using a few concepts.
This was introduced in #13580. Please check my logic on this as I'm unable to test on a SmartOS system. Even better would be a confirmation from someone who can build there.
ACKs for top commit:
Empact:
Code review ACK https://github.com/bitcoin/bitcoin/commit/ad7584d8b60119ca3717117a1eb6a16d753c5d74
Tree-SHA512: 1faf65c900700efb1cf3092c607a2230321b393cb2f029fbfb94bc8e50df1dabd7a9e4b91e3b34f0d2f3471aaf18ee7e56d91869db5c5f4bae84da95443e1120
|
|
These replace our platform-specific mess in favor of c++20 endian detection
via std::endian and internal byteswap functions when necessary.
They no longer rely on autoconf detection.
|
|
Rather than a complicated set of tests to decide which bswap functions to
use, always prefer the compiler built-ins when available.
These builtins and fallbacks can all be removed once we're using c++23, which
adds std::byteswap.
|
|
This removes the only remaining autoconf macro in our serialization code,
so it can now be used trivially and safely out-of-tree.
|
|
Now that --enable-suppress-external-warnings is on by default, we can
drop it.
|
|
`libsecp256k1` code
cbea49c0d32badb975fbf22d44f8e25cc7972af7 build: Pass sanitize flags to instrument `libsecp256k1` code (Hennadii Stepanov)
Pull request description:
This PR is a revived https://github.com/bitcoin/bitcoin/pull/27991 with an addressed [comment](https://github.com/bitcoin/bitcoin/pull/27991#discussion_r1252148488).
Fixes https://github.com/bitcoin/bitcoin/issues/27990.
Might be tested as follows:
```
$ ./autogen.sh && ./configure --enable-fuzz --with-sanitizers=fuzzer CC=clang-13 CXX=clang++-13
$ make clean > /dev/null && make
$ objdump --disassemble=secp256k1_xonly_pubkey_serialize src/test/fuzz/fuzz | grep __sanitizer_cov
1953bd0:e8 bb c6 05 ff call 9b0290 <__sanitizer_cov_trace_const_cmp8>
1953d32:e8 69 c4 05 ff call 9b01a0 <__sanitizer_cov_trace_pc_indir>
1953d58:e8 43 c4 05 ff call 9b01a0 <__sanitizer_cov_trace_pc_indir>
1953d82:e8 19 c4 05 ff call 9b01a0 <__sanitizer_cov_trace_pc_indir>
```
ACKs for top commit:
fanquake:
ACK cbea49c0d32badb975fbf22d44f8e25cc7972af7
dergoegge:
reACK cbea49c0d32badb975fbf22d44f8e25cc7972af7
Tree-SHA512: 801994e75b711d20eaf0d675f378da07d693f4a7de026efd93860f5f1deabed855a83eca3561725263e4fe605fcc5f91eb73c021ec91c831864e6deb575e3885
|
|
Also a new UBSan suppression has been added.
|
|
-O0 is just overriding -Og.
|
|
|
|
This avoids cases of missing -O2, when *FLAGS has been overriden.
Removes the need for duplicate code to clear autoconf defaults.
Also, move CORE_CXXFLAGS before DEBUG_CXXFLAGS, so that -O2 is always
overriden if debugging etc.
|
|
2d1b1c7daeeada3f737e62ceb2db7484cde5ff4e build: remove --enable-lto (fanquake)
Pull request description:
This has outlived its usefulness, doesn't gel well with newer compilers & `-flto` related options, i.e thin vs full, or `=auto`, and having `-flto` as the only option means that sometimes this just needs to be worked around, i.e in oss-fuzz:
https://github.com/google/oss-fuzz/blob/master/projects/bitcoin-core/build.sh.
While it was convenient when `-flto` was newer, support for `-flto` is now in all compilers we use, and there's also no-longer any real need for us to treat `-flto` different to any other optimization option.
Remove it, to remove build complexity, and so there's no need to port a similar option to CMake.
Note that the LTO option remains in depends, because we still a way to build packages that have LTO specific patches/options.
ACKs for top commit:
TheCharlatan:
ACK 2d1b1c7daeeada3f737e62ceb2db7484cde5ff4e
hebasto:
ACK 2d1b1c7daeeada3f737e62ceb2db7484cde5ff4e.
Tree-SHA512: 91812de7da35346f51850714a188fcffbac478bc8b348bf756c2555fcbde86ba622ac2fb77d294dea0378c741d3656f06121ef3a795aeed63fd170fc31bfa5af
|
|
aaaace2fd1299939c755c281b787df0bbf1747a0 fuzz: Assume presence of __builtin_*_overflow, without checks (MarcoFalke)
fa223ba5eb764fe822229a58d4d44d3ea83d0793 Revert "build: Fix undefined reference to __mulodi4" (MarcoFalke)
fa7c751bd923cd9fb4790fe7fb51fafa2faa1db6 build: Bump clang minimum supported version to 14 (MarcoFalke)
Pull request description:
Most supported operating systems ship with clang-14 (or later), so bump the minimum to that and allow new code to drop workarounds for previous clang bugs.
For reference:
* https://packages.debian.org/bookworm/clang (`clang-14`)
* https://packages.ubuntu.com/jammy/clang (`clang-14`)
* CentOS-like 8/9 Stream: All Clang versions from 15 to 17
* FreeBSD 12/13: All Clang versions from 15 to 16
* OpenSuse Tumbleweed ships with https://software.opensuse.org/package/clang (`clang17`); No idea about OpenSuse Leap
On operating systems where the clang version is not shipped by default, the user would have to use GCC, or install clang in a different way. For example:
* https://packages.debian.org/bullseye/g++ (g++-10)
* https://packages.ubuntu.com/focal/g++-10
* https://apt.llvm.org/, or nix, or guix, or compile clang from source, ...
ACKs for top commit:
fanquake:
ACK aaaace2fd1299939c755c281b787df0bbf1747a0
Tree-SHA512: 81d066b14cc568d27312f1cc814b09540b038a10a0a8e9d71fc9745b024fb6c32a959af673e6819b817ea7cef98da4abfa63dff16cffb7821b40083016b0291f
|
|
|
|
This reverts commit e4c8bb62e4a6873c45f42d0d2a24927cb241a0ea.
|
|
LLVM Clang >=16.0 and Apple Clang >=15.0 do not recognize
`-Xclang -internal-isystem/usr/local/include` anymore.
For example, see: https://github.com/llvm/llvm-project/commit/cbbe1d44546db52c71c9a2b18f85b87ae82df9e7
|
|
This has outlived its usefulness, doesn't gel well with
newer compilers & `-flto` related options, i.e thin vs full, or `=auto`,
and having `-flto` as the only option means that sometimes this just
needs to be worked around, i.e in oss-fuzz:
https://github.com/google/oss-fuzz/blob/master/projects/bitcoin-core/build.sh.
While it was convenient when `-flto` was newer, support for `-flto` is now
in all compilers we use, and there's also no-longer any real need
for us to treat `-flto` different to any other optimization option.
Remove it, to remove build complexity, and so there's no need
to port a similar option to CMake.
Note that the LTO option remains in depends, because we still a way to
build packages that have LTO specific patches/options.
If we decide to merge this, I'll follow up downstream in oss-fuzz first,
to make sure we don't break the build.
|
|
migration to C++20
49a90915aa3ee8e3a7e163f23a55de931faf8523 build: Bump minimum required Boost to 1.73.0 to support C++20 (Hennadii Stepanov)
Pull request description:
Boost versions <1.73 have C++20-specific bugs that were fixed in the following commits:
- https://github.com/boostorg/signals2/commit/15fcf213563718d2378b6b83a1614680a4fa8cec
- https://github.com/boostorg/test/commit/495c095dc063052ce54f2fe9217fe0fc69ced5f1
I tested [`libboost1.71-dev`](https://packages.ubuntu.com/focal/libboost1.71-dev) in Ubuntu 20.04 and Boost 1.71, 1.72, 1.73 in our depends build system.
Closes https://github.com/bitcoin/bitcoin/issues/29063.
ACKs for top commit:
fanquake:
ACK 49a90915aa3ee8e3a7e163f23a55de931faf8523
Tree-SHA512: b8ebc08af85abfa3fda70961bd1136ee9e5149dd76a3f901e43acba624d231971873cba5cbf30837f9e5ab58790b8330f241a76cb76d8cf5dce5ad0cca33fba8
|
|
308aec3e5655327d98e0428d8205d246f24d6af5 build: disable external-signer for Windows (fanquake)
35537318a19360ddf1ea8f0c1e6d8ad49e635516 ci: remove --enable-external-signer from win64 job (fanquake)
Pull request description:
It's come to light that Boost ASIO (a Boost Process sub dep) has in some
instances, been quietly initialising our network stack on Windows (see
PR https://github.com/bitcoin/bitcoin/pull/28486 and discussion in https://github.com/bitcoin/bitcoin/issues/28940).
This has been shielding a bug in our own code, but the larger issue
is that Boost Process/ASIO is running code before main, and doing things
like setting up networking. This undermines our own assumptions about
how our binary works, happens before we run any sanity checks,
and before we call our own code to setup networking. Note that ASIO also
calls WSAStartup with version `2.0`, whereas we call with `2.2`.
It's also not clear why a feature like external signer would have a
dependency that would be doing anything network/socket related,
given it only exists to spawn a local process.
See also the discussion in https://github.com/bitcoin/bitcoin/issues/24907. Note that the maintaince of Boost Process in general,
has not really improved. For example, rather than fixing bugs like https://github.com/boostorg/process/issues/111,
i.e, https://github.com/boostorg/process/pull/317, the maintainer chooses to just wrap exception causing overflows
in try-catch blocks: https://github.com/boostorg/process/commit/0c42a58eacab6a96b19196e399307bad8a938a27. These changes get merged in large,
unreviewed PRs, i.e https://github.com/boostorg/process/pull/319.
This PR disables external-signer on Windows for now. If, in future, someone
changes how Boost Process works, or replaces it entirely with some
properly reviewed and maintained code, we could reenable this feature on
Windows.
ACKs for top commit:
hebasto:
re-ACK 308aec3e5655327d98e0428d8205d246f24d6af5.
TheCharlatan:
ACK 308aec3e5655327d98e0428d8205d246f24d6af5
Tree-SHA512: 7405f7fc9833eeaacd6836c4e5b1c1a7845a40c1fdd55c1060152f8d8189e4777464fde650e11eb1539556a75dddf49667105987078b1457493ee772945da66e
|
|
Boost versions <1.73 have C++20-specific bugs that were fixed in the
following commits:
- https://github.com/boostorg/signals2/commit/15fcf213563718d2378b6b83a1614680a4fa8cec
- https://github.com/boostorg/test/commit/495c095dc063052ce54f2fe9217fe0fc69ced5f1
|
|
fa8adbe7c17b16cf7ecd16eb9f3f1792e9da2876 build: Enable -Wunreachable-code (MarcoFalke)
Pull request description:
It seems a bit confusing to write code after a `return`. This can even lead to bugs, or incorrect code, such as https://github.com/bitcoin/bitcoin/pull/28830/files#r1415372320 . (Edit: The linked instance is not found by clang's `-Wunreachable-code`).
Fix all issues by enabling `-Wunreachable-code`.
This flag also enables `-Wunreachable-code-loop-increment`, according to https://clang.llvm.org/docs/DiagnosticsReference.html#wunreachable-code, so remove that.
ACKs for top commit:
ajtowns:
> ACK [fa8adbe](https://github.com/bitcoin/bitcoin/commit/fa8adbe7c17b16cf7ecd16eb9f3f1792e9da2876)
stickies-v:
ACK fa8adbe7c17b16cf7ecd16eb9f3f1792e9da2876
jonatack:
ACK fa8adbe7c17b16cf7ecd16eb9f3f1792e9da2876 tested with arm64 clang 17.0.6
Tree-SHA512: 12a2f74b69ae002e62ae08038f7458837090a12051a4c154d05ae4bb26fb19fc1fa76c63aedf2b3fbb36f048c593ca3b8c0efe03fe93cf07a0fd114fc84ce1e7
|
|
|
|
|
|
It's come to light that Boost ASIO (a Boost Process sub dep) has in some
instances, been queitly initialising our network stack on Windows (see
PR #28486 and discussion in #28940).
This has been shielding a bug in our own code, but the larger issue
is that Boost Process/ASIO is running code before main, and doing things
like setting up networking. This undermines our own assumptions about
how our binary works, happens before we get to run any sanity checks,
and also runs before we call our own code to setup networking.
It's also not clear why a feature like external signer would have a
dependency that would be doing anything network/socket related, given it
only exists to spawn a local process.
|
|
intrinsics" test
228d6a2969e4fcee573c9df7aad31550eab9c8d4 build: Fix regression in "ARMv8 CRC32 intrinsics" test (Hennadii Stepanov)
Pull request description:
In the master branch, the `aarch64` binaries lack support for CRC32 intrinsics.
The `vmull_p64` is a part of the Crypto extensions from the ACLE. They are optional extensions, so they get enabled with a `+crypto` for architecture flags.
The regression was introduced in https://github.com/bitcoin/bitcoin/pull/26183 (v25.0).
The `./configure` script log excerpts:
- the master branch @ d752349029ec7a76f1fd440db2ec2e458d0f3c99:
```
checking whether C++ compiler accepts -march=armv8-a+crc... yes
checking whether C++ compiler accepts -march=armv8-a+crypto... yes
checking for ARMv8 CRC32 intrinsics... no
checking for ARMv8 SHA-NI intrinsics... yes
```
- this PR:
```
checking whether C++ compiler accepts -march=armv8-a+crc+crypto... yes
checking whether C++ compiler accepts -march=armv8-a+crypto... yes
checking for ARMv8 CRC32 intrinsics... yes
checking for ARMv8 SHA-NI intrinsics... yes
```
Guix build:
```
x86_64
2afd81f540c6d3b36ff305e88bafe935e4272cd3efef3130aa69d49a0522541b guix-build-228d6a2969e4/output/aarch64-linux-gnu/SHA256SUMS.part
6c704d6d30d495adb3fb86befdb500eb389a02c1167163f14ab5c3c3e630e6b3 guix-build-228d6a2969e4/output/aarch64-linux-gnu/bitcoin-228d6a2969e4-aarch64-linux-gnu-debug.tar.gz
e4419963c9c0d99adc4e38538900b648f2c14f793b60c8ee2e6f5acc9d3fadd3 guix-build-228d6a2969e4/output/aarch64-linux-gnu/bitcoin-228d6a2969e4-aarch64-linux-gnu.tar.gz
7d11052b6bd28cdf26d5f2a4987f02d32c93a061907bcd048fb6d161a0466ca9 guix-build-228d6a2969e4/output/dist-archive/bitcoin-228d6a2969e4.tar.gz
```
ACKs for top commit:
TheCharlatan:
ACK 228d6a2969e4fcee573c9df7aad31550eab9c8d4
Tree-SHA512: 4c27ca8acb953bf56e972d907a282ee19e3f30f7a4bf8a9822395fe0e28977cd6233e8b65b4a25cc1d3d5ff6a796d7af07653e18531c44ee3efaff1563d96d32
|
|
f95af98128f17002bf137a48441167020f3ef9bb guix: default ssp for Windows GCC (fanquake)
95d55b96c2cfd2a0d5a246d4a9eff9d0744ba223 guix: remove ssp workaround from Windows GCC (fanquake)
8f43302a0a1bb79129933e4cc174bf8d8d59ec15 build: remove explicit libssp linking from Windows build (fanquake)
Pull request description:
I was expecting this to fail to compile somewhere, maybe in the CI, but that doesn't seem to be the case?
Seems workable given the SSP related changes in the newer mingw-w64 headers (which are in Guix):
> Implement some of the stack protector functions/variables so -lssp is now optional when _FORTIFY_SOURCE or -fstack-protector-strong is used.
However I think this would still be broken in some older environments, so we might have to wait for a compiler bump, or similar. The optional -lssp also seems to work when using older headers, which doesn't make sense.
Would fix #28104.
ACKs for top commit:
hebasto:
ACK f95af98128f17002bf137a48441167020f3ef9bb, I've verified binaries from `bitcoin-f95af98128f1-win64.zip` on Windows 11 Pro 23H2.
TheCharlatan:
ACK f95af98128f17002bf137a48441167020f3ef9bb
Tree-SHA512: 71169ec513cfe692dfa7741d2bf37b45da05627c0af1cbd50cf8c3c04cc21c4bf88f3284532bddc1e3e648391ec78dbaca5170987a13c21ac204a7bcaf27f349
|
|
The `vmull_p64` is a part of the Crypto extensions from the ACLE. They
are optional extensions, so they get enabled with a `+crypto` for
architecture flags.
|
|
We no-longer need this, as we no-longer build python packages.
Effectively reverts de619a37fd18a17225c8a10b828fc61958abe4cf.
|
|
|
|
Similar to a98356fee8a44d7d1cb37f22c876fff8f244365e.
|
|
This is deprecated on macOS:
```bash
ld: warning: -bind_at_load is deprecated on macOS
```
and likely redundant anyways, given the behaviour of dyld3.
Unfortunately libtool is still injecting a `-bind_at_load`:
```bash
# Don't allow lazy linking, it breaks C++ global constructors
# But is supposedly fixed on 10.4 or later (yay!).
if test CXX = "$tagname"; then
case ${MACOSX_DEPLOYMENT_TARGET-10.0} in
10.[0123])
func_append compile_command " $wl-bind_at_load"
func_append finalize_command " $wl-bind_at_load"
;;
esac
fi
```
so this doesn't remove all the warnings, but removes us as a potential
source of them.
Note that anywhere the ld64 warnings are being emitted, we are already
not adding this flag to our hardened ldflags, because of `-Wl,-fatal_warnings`.
|
|
Consolidate the lib checking logic to be the same as miniupnpc.
|
|
Having the link check in the header check loop means we get `-lminiupnpc
-lminiupnpc -lminiupnpc` on the link line. This is unnecessary, and
results in warnings, i.e:
```bash
ld: warning: ignoring duplicate libraries: '-levent', '-lminiupnpc'
ld: warning: ignoring duplicate libraries: '-levent', '-lminiupnpc'
ld: warning: ignoring duplicate libraries: '-levent', '-lminiupnpc'
```
These warnings have been occurring since the new linker released with
Xcode 15, and also came up in https://github.com/hebasto/bitcoin/pull/34.
|
|
|
|
fa25e8b0a1610553014c786428f146ef9c694678 doc: Recommend lint image build on every call (MarcoFalke)
faf70c1f330a92612cf381d32c791e9ba445d3f2 Bump python minimum version to 3.9 (MarcoFalke)
fa8996b930886da712c09ffe4b58016b36c2ae5b ci: Bump i686_multiprocess.sh to latest Ubuntu LTS (MarcoFalke)
Pull request description:
All supported operating systems ship with python 3.9 (or later), so bumping the minimum should not cause any issues. A bump will allow new code to use new python 3.9 features.
For reference:
* https://packages.debian.org/bullseye/python3
* https://packages.ubuntu.com/focal/python3.9
* FreeBSD 12/13 also ships with 3.9
* CentOS-like 8/9 also ships with 3.9 (and 3.11)
* OpenSuse Leap also ships with 3.9 (and 3.11) https://software.opensuse.org/package/python311-base
This is for Bitcoin Core 27.0 in 2024 (next year), not the soon upcoming 26.0 next month.
ACKs for top commit:
Sjors:
ACK fa25e8b0a1610553014c786428f146ef9c694678
jamesob:
ACK fa25e8b0a1610553014c786428f146ef9c694678 ([`jamesob/ackr/28211.1.MarcoFalke.bump_python_minimum_supp`](https://github.com/jamesob/bitcoin/tree/ackr/28211.1.MarcoFalke.bump_python_minimum_supp))
Tree-SHA512: 86c9f6ac4b5ba94a62ee6a6062dd48a8295d8611a39cdb5829f4f0dbc77aaa1a51edccc7a99275bf699143ad3a6fe826de426d413e5a465e3b0e82b86d10c32e
|
|
|
|
This is not a hardening specific flag, it should be used at all times,
regardless of if hardening is enabled or not. Note that this was
still the case here, but having this exist in the hardening flags is
confusing, and may lead someone to move it inside one of the `use_hardening`
blocks, where it would become unused, with `--disable-hardening`.
|