a340d9500a ci: add int128_struct tests (Jonas Nick)
dceaa1f579 int128: Tidy #includes of int128.h and int128_impl.h (Tim Ruffing)
2914bccbc0 Simulated int128 type. (Russell O'Connor)
Pull request description:
Abstracts the int128 type and provides an native version, if available, or a implements it using a pair of int64_t's.
This is activated by setting the configuration flag `--with-test-override-wide-multiply=int128_struct`.
The primary purpose of this PR is to take advantage of MSVC's [umulh](https://docs.microsoft.com/en-us/cpp/intrinsics/umulh?view=msvc-170) intrinsic that we can use to simulate an int128 type which MSVC does not have (AFAIU). This PR lays out the groundwork for this level of MSVC support, but doesn't include the configuration logic to enable it yet.
For completeness, and implementation of `umulh` and `mulh` are also provided for compilers that support neither the intrinsic nor the int128 type (such as CompCert?). This also opens up the possibility of removing the 32-bit field and scalar implementations should that ever be desired.
ACKs for top commit:
sipa:
ACK a340d9500a
jonasnick:
ACK a340d9500a
Tree-SHA512: b4f2853fa3ab60ce9d77b4eaee1fd20c4b612850e19fcb3179d7e36986f420c6c4589ff72f0cf844f989584ace49a1cd23cca3f4e405dabefc8da647a0df679d
6a965b6b98 Remove usage of CHECK from non-test file (Tobin C. Harding)
Pull request description:
Currently CHECK is used only in test and bench mark files except for one usage in `ecmult_impl.h`.
We would like to move the definition of CHECK out of `util.h` so that `util.h` no longer has a hard dependency on `stdio.h`.
Done as part of an effort to allow secp256k1 to be compiled to WASM as part of `rust-secp256k1`.
### Note to reviewers
Please review carefully, I don't actually know if this patch is correct. Done while working on #1095. I'm happy to make any changes both in concept and execution - I'm super rusty at C programming.
cc real-or-random
ACKs for top commit:
sipa:
utACK 6a965b6b98
real-or-random:
utACK 6a965b6b98
Tree-SHA512: 6bfb456bdb92a831acd3bc202607e80f6d0a194d6b2cf745c8eceb12ba675d03a319d6d105332b0cbca474e443969295e5a8e938635453e21e057d0ee597440b
After this commit, int128.h and int128_impl.h are included as follows:
- .c files which use int128 include int128_impl.h (after util.h)
- .h files which use int128 include int128.h (after util.h)
This list is exhaustive. util.h needs to included first because it sets
up necessary #defines.
Currently CHECK is used only in test and bench mark files except for one
usage in `ecmult_impl.h`.
We would like to move the definition of CHECK out of `util.h` so that
`util.h` no longer has a hard dependency on `stdio.h`.
Done in preparation for moving the definition of `CHECK` as part of an
effort to allow secp256k1 to be compiled to WASM as part of
`rust-secp256k1`.
cabe085bb4 configure: Remove pkgconfig macros again (reintroduced by mismerge) (Tim Ruffing)
Pull request description:
We had removed `PKG_PROG_PKG_CONFIG` in 21b2ebaf74
(#1090). But then then the not rebased (!) merge of 2be6ba0fed
(#1084) brought that macro back at another location, without git
complaining about a conflict.
Fixes#1127.
ACKs for top commit:
fanquake:
ACK cabe085bb4
hebasto:
ACK cabe085bb4
jonasnick:
ACK cabe085bb4
Tree-SHA512: ba497503db3a11e631b15c4fe875e62d892971c2c708d90b2f6be684e85d164043ea97c13af0452831eef41f3cf8230cd8a9eafa332dc5b5ae18e118b87c3828
We had removed `PKG_PROG_PKG_CONFIG` in 21b2ebaf74
(#1090). But then then the not rebased (!) merge of 2be6ba0fed
(#1084) brought that macro back at another location, without git
complaining about a conflict.
Fixes#1127.
17065f48ae tests: Randomize the context with probability 15/16 instead of 1/4 (Tim Ruffing)
Pull request description:
ACKs for top commit:
sipa:
ACK 17065f48ae
jonasnick:
ACK 17065f48ae
Tree-SHA512: 3b7005770007b922a294be610f23da60b0dde74dfd7585d64a2cb04eaa6ec879de8d21a0ade31c1857019a8dd97260fa3aa167ae16fc55027ef280a3e3feaa6d
55f8bc99dc ecmult_gen: Improve comments about projective blinding (Tim Ruffing)
7a86955800 ecmult_gen: Simplify code (no observable change) (Tim Ruffing)
4cc0b1b669 ecmult_gen: Skip RNG when creating blinding if no seed is available (Tim Ruffing)
Pull request description:
Running the RNG is pointless if no seed is available because the key
will be fixed. The computation just wastes time.
Previously, users could avoid this computation at least by asking for
a context without signing capabilities. But since 3b0c218 we always
build an ecmult_gen context, ignoring the context flags. Moreover,
users could never avoid this pointless computation when asking for
the creation of a signing context.
This fixes one item in #1065.
ACKs for top commit:
sipa:
ACK 55f8bc99dc
apoelstra:
ACK 55f8bc99dc
Tree-SHA512: 5ccba56041f94fa8f40a8a56ce505369ff2e0ed20cd7f0bfc3fdfffa5fa7bf826a93602b9b2455a352865a9548ab4928e858c19bb5af7ec221594a3bf25c4f3d
Whenever I read this code, I first think that rescaling ctx->initial is
a dead store because we overwrite it later with gb. But that's wrong.
The rescaling blinds the computation of gb and affects its result.
Running the RNG is pointless if no seed is available because the key
will be fixed. The computation just wastes time.
Previously, users could avoid this computation at least by asking for
a context without signing capabilities. But since 3b0c218 we always
build an ecmult_gen context, ignoring the context flags. Moreover,
users could never avoid this pointless computation when asking for
the creation of a signing context.
40a3473a9d build: Fix #include "..." paths to get rid of further -I arguments (Tim Ruffing)
Pull request description:
This simplifies building without a build system.
This is in line with #925; the paths fixed here were either forgotten
there or only introduced later. This commit also makes the Makefile
stricter so that further "wrong" #include paths will lead to build
errors even in autotools builds.
This belongs to #929.
ACKs for top commit:
hebasto:
ACK 40a3473a9d
Tree-SHA512: 6f4d825ea3cf86b13f294e2ec19fafc29660fa99450e6b579157d7a6e9bdb3404d761edf89c1135fa89b984d6431a527beeb97031dc90f2fae9761528f4d06d1
This simplifies building without a build system.
This is in line with #925; the paths fixed here were either forgotten
there or only introduced later. This commit also makes the Makefile
stricter so that further "wrong" #include paths will lead to build
errors even in autotools builds.
This belongs to #929.
Co-authored-by: Hennadii Stepanov <32963518+hebasto@users.noreply.github.com>
49e2acd927 configure: Improve rationale for WERROR_CFLAGS (Tim Ruffing)
8dc4b03341 ci: Add a C++ job that compiles the public headers without -fpermissive (Tim Ruffing)
51f296a46c ci: Run persistent wineserver to speed up wine (Tim Ruffing)
3fb3269c22 ci: Add 32-bit MinGW64 build (Tim Ruffing)
9efc2e5221 ci: Add MSVC builds (Tim Ruffing)
2be6ba0fed configure: Convince autotools to work with MSVC's archiver lib.exe (Tim Ruffing)
bd81f4140a schnorrsig bench: Suppress a stupid warning in MSVC (Tim Ruffing)
09f3d71c51 configure: Add a few CFLAGS for MSVC (Tim Ruffing)
3b4f3d0d46 build: Reject C++ compilers in the preprocessor (Tim Ruffing)
1cc0941414 configure: Don't abort if the compiler does not define __STDC__ (Tim Ruffing)
cca8cbbac8 configure: Output message when checking for valgrind (Tim Ruffing)
1a6be5745f bench: Make benchmarks compile on MSVC (Tim Ruffing)
Pull request description:
ACKs for top commit:
jonasnick:
ACK 49e2acd927
Tree-SHA512: 986c498fb218231fff3519167d34a92e11dea6a4383788a9723be105c20578cd483c6b06ba5686c6669e3a02cfeebc29b8e5f1428552ebf4ec67fa7a86957548
This commit also raises the TEST_ITERS for wine tasks to the default.
The overhead of wine is negligible, so we can certainly afford the same
number of iterations as for native Linux tests.
This adds MSVC builds built on Linux using wine. This requires some
settings of tools and flags because the autotools support for MSVC is
naturally somewhat limited.
The advantage of this approach is that it is compatible with our
existing CI scripts, so there's no need to write a Windows CI script
(in PowerShell or similar). If we want to test building and running on
Windows native (e.g., as supported by Cirrus CI) we could still do this
in the future.
Another advantage of this approach is that contributors can simply use
the docker image if they need a MSVC installation in a non-Windows
environment.
This commit also improves the Dockerfile by grouping RUN commands
according to Docker docs:
https://docs.docker.com/develop/develop-images/dockerfile_best-practices/#run
6f6cab9989 abi: Don't export symbols in static Windows libraries (Cory Fields)
Pull request description:
For context, Bitcoin Core has recently merged [libbitcoin-kernel](https://github.com/bitcoin/bitcoin/pull/24322), a small library that intends to eventually minimally encompass Core's validation engine. This kernel lib includes a static libsecp256k1. Without this change, because libsecp256k1.a ends up with exported symbols, we end up with libsecp256k1 symbols exported by our libbitcoin-kernel library (which causes unrelated problems not worth getting into here).
libtool takes care of building both object versions, and it automatically builds objects for shared libs with -DDLL_EXPORT. We just need to opt-in to its functionality.
I can't imagine this having any negative impact on any current statically-linking applications, if anything they'll just be a tiny bit smaller because they can now strip unused symbols.
ACKs for top commit:
real-or-random:
utACK 6f6cab9989
theuni:
> Not sure what other changes made compilation on CI fail but Concept ACK [6f6cab9](6f6cab9989). This should be entirely harmless.
sipa:
utACK 6f6cab9989
laanwj:
utACK 6f6cab9989
Tree-SHA512: 39f240046639738f7a8c01068e728b2f9ceac2754cc4b0a5fa46c28f6f57a8c4124653b56dfbf5c13106b07c11ac599cc41b508e16862d539ce1af6c3365a205
Pieter Wuille
Jonas Nick
Tim Ruffing
Russell O'Connor
Tobin C. Harding
Hennadii Stepanov
henopied
siv2r
Cory Fields