libsecp256k1

This commit simply updates a GitHub Actions helper used to build Docker containers from version 5 to version 6. It is a routine maintenance change to the project's continuous integration (CI) configuration and does not alter the cryptographic library code, build scripts, or any software that end users run.

This is a routine cleanup of the project's automated CI (Continuous Integration) configuration. It removes an unused 'tag' input from GitHub Actions Docker build steps and switches to referencing the most recently built Docker image by ID instead of by tag. There is no security issue here.

This commit changes how Bitcoin Core's libsecp256k1 project caches Docker images in its automated testing system. It adds a 'scope' label to GitHub Actions cache settings so that separate caches are kept for different CPU architectures (like x86_64 vs ARM). Previously, all architectures shared one cache, meaning only the most recently used image was kept. This is a CI infrastructure improvement, not a security fix.

This commit adds a simple command-line option to the project's test runner that prints a list of all available tests and modules. It is a pure test-infrastructure convenience feature with no effect on the cryptographic library or any production code paths.

This commit adds a new optional '--log=1' command-line flag to the project's internal test runner so users can see when each test starts, finishes, and how long it took. It also adds a small safety check to make sure the test framework is initialized before tests are run. There is no change to cryptographic code, no network code, and no behavior that affects normal users of the library.

This commit adds a new command-line option (-t or --target) to the project's internal test runner so developers can run a single test or a whole module instead of the entire test suite. It is a testing convenience feature with no apparent security relevance.

This commit adds a --help / -h command-line option to the project's internal test suite. It simply prints usage instructions and exits. There is no security-relevant change to any cryptographic code, network handling, or production library behavior.

This commit is a pure test-infrastructure change. It introduces a small internal unit-test framework for the secp256k1 cryptographic library, replacing a long hand-written main() function with a registry of test cases and adding optional parallel test execution. It does not change any cryptographic code, public API, or production behavior, and it does not fix or introduce any security vulnerability.

This commit is purely a test-code refactoring. It reorganizes how the project's test cases are registered and run, allowing individual tests to be selected and run in parallel. No production cryptographic code is changed, and there is no indication this fixes or introduces a security vulnerability.

This commit only changes documentation comments in a header file. It rewords descriptions of error and illegal-argument callbacks, fixes typos, and clarifies wording. No executable code, build system, or API behavior was changed, so it has no security impact on its own.

This commit only changes a documentation comment in a header file. It clarifies that if a user-supplied error callback returns instead of aborting, the API call's results are undefined and the callback may be triggered more than once. No code behavior was changed.

This commit fixes compiler warnings in build-system tests that detect whether x86_64 assembly can be used. It initializes a variable and adds a return statement to a small test program. These changes do not affect the actual cryptographic code, runtime behavior, or security of the library.

This is a small cleanup in a benchmark test file for the ECDH module. It replaces a dynamically created cryptographic context with a pre-built static one and removes the matching cleanup call. The change is about making the benchmark follow the intended API more cleanly; it does not fix an exploitable vulnerability in production code.

This commit only updates the documentation comments for a function in a header file. It clarifies what the return value of secp256k1_ecdsa_recover means and explains a subtle detail about signature normalization after converting a recoverable signature. No code logic was changed, so there is no security vulnerability or fix here.

This commit is a simple code cleanup: it moves a small helper function that reads the system clock from one internal test/benchmark header file to a new shared test header file. There is no change to the cryptographic library, no change to how Bitcoin transactions or keys are handled, and no security fix or vulnerability introduced.

This commit is a simple cleanup in the test code. It replaces several copies of the same 32-byte secp256k1 curve group-order constant with a single shared constant. There is no change to the actual cryptographic library or to how tests behave; it only removes duplicated code.

This commit is a code-quality and testing-hardening change inside a widely used cryptographic library. It splits one internal memory-wiping helper into two: one that guarantees zeros (used where the code later reads the buffer) and one that intentionally marks memory as undefined in test builds (used where the buffer should never be read again). The only functional change visible in the diff is replacing a plain memset with the guaranteed-zero helper in one MuSig partial-signing path, which makes the code's intent clearer and avoids a potential test-only false positive. There is no direct evidence this fixes an exploitable vulnerability in production.

This change removes a fallback path in an internal ABI-checking shell script. It is a cleanup of build tooling with no effect on the cryptographic library or its security.

This is a one-line documentation update. It clarifies that a specific function should not be called with a special read-only context object named secp256k1_context_static. There is no code change and no security fix.

This commit fixes a small but real security hygiene issue in the Schnorr signature code. A 32-byte buffer that holds the secret signing nonce (the random-like value 'k') was not being wiped from memory after use. The patch renames the buffer to make its sensitive role obvious and adds an explicit clear operation so the nonce does not linger in stack memory after the function returns. If an attacker could read leftover process memory, the un-cleared nonce could help forge signatures for that message/key. The fix is straightforward and aligns the Schnorr code with the existing ECDSA behavior.

This commit fixes a variable type mismatch in the RFC6979 nonce-generation code. The code copies a `size_t` length value into an `int` variable. On platforms where `int` is 32 bits and `size_t` is 64 bits, requesting more than about 2 billion bytes of output would cause the `int` to wrap around to a negative value. That negative value would then be treated as a very large positive number in later comparisons, potentially causing the loop to run far too long or to access memory out of bounds, which could crash or hang the program. In practice, callers inside libsecp256k1 request only 32 bytes, so the bug is unlikely to be reachable from normal use.

This commit only expands the project's automated testing on Windows by adding more clang-cl build configurations to GitHub Actions. It does not change any library code, cryptographic logic, or user-facing behavior. There is no security issue here.

This commit is a large repository import or merge that brings in the entire libsecp256k1 codebase, including build systems, CI configuration, documentation, and source files. The commit title says it only updates Windows build documentation to recommend clang-cl, but the actual diff shows a wholesale addition of 175 files. There is no code change that fixes or introduces a security vulnerability in the cryptographic library itself.

This commit is a test-only cleanup. It moves a helper function that checks SHA256 'tagged hash' setup from one test file to a shared test file, renames it, and updates several test modules to use the shared helper. There are no changes to the actual cryptographic library code that users rely on, and no security bug is fixed or introduced.

This commit only fixes a broken hyperlink in a code comment, replacing an outdated academic paper URL with a working one. No code behavior changes.