## Problem Windows builds break with a current local toolchain (Scoop LLVM 22.1.8, CMake 4.4.0, VS 2026), in two independent ways: 1. The build stops at curl's deliberate guard: `#error "no non-blocking method was found/used/set"` in `third-party/curl/lib/nonblock.c`. 2. From the second configure onward, `cmake --build` re-runs CMake in an endless loop (observed 42 consecutive reconfigure cycles in a single build). Likely the same mechanism behind the "endlessly building" VS 2026 note in `docs/setup/dev/vs.md`. ## Root cause 1. `third-party/curl/CMake/CurlTests.c` passes `int *` to `ioctlsocket()`, whose third parameter is `u_long *`. Clang 22 promotes `-Wincompatible-pointer-types` to a hard error in C, so the `HAVE_IOCTLSOCKET_FIONBIO` try_compile silently fails and `curl_config.h` never defines it. Upstream CI does not see this because the windows-2022 runner image ships an older LLVM. GCC 14 promotes the same warning to a hard error, which is very likely the `CurlTests.c.obj` failure reported from MSYS2 in open-goal/jak-project#3551. Upstream curl hit the identical problem with GCC 14 and fixed the probe in curl 8.8.0 (curl/curl#13578). 2. The root CMakeLists copies the build tree's `compile_commands.json` into `<src>/build/` for clangd using `configure_file()`, which registers its input as a configure dependency. CMake rewrites `compile_commands.json` late in every generation, after `CTestTestfile.cmake` and `cmake_install.cmake` (outputs of the same Ninja regen rule), so once the dependency is registered the rule is deterministically dirty and every `ninja` invocation re-runs CMake. A pristine first configure is safe (the file does not exist yet, so the `if(EXISTS ...)` guard skips the copy), which is why the loop looks machine- or IDE-specific. ## Fix 1. Per review, re-vendor `third-party/curl` at the `curl-8_21_0` tag (previously `curl-8_3_0`), which carries the upstream probe fix plus two years of upstream development; `vendor.yaml` updated to match. Adjustments the version jump forced: - curl 8.15 removed the native macOS Secure Transport backend (`CURL_USE_SECTRANSP`), so macOS now builds curl against OpenSSL like Linux. The two macOS workflows install Homebrew `openssl@3` and export `OPENSSL_ROOT_DIR` (keg-only), and the macOS setup docs gained the same two lines. - `CURL_BROTLI` / `CURL_ZSTD` switched to AUTO-detection in curl 8.10; pinned OFF to keep the previous no-compression behavior and avoid silently linking whatever the CI images happen to have. - curl's new top-level `BUILD_EXAMPLES` cache option (default ON) leaked into discord-rpc's identically named option and broke configure at a nonexistent `examples/send-presence` directory; pinned OFF ahead of the third-party subdirectories. The diff is dominated by the mechanical tag-tree swap under `third-party/curl` (linguist-vendored, collapsed in review). The hand-written changes are `CMakeLists.txt`, the two macOS workflows, `docs/setup/system/macos.md`, and `vendor.yaml`. 2. Swap `configure_file()` for `file(COPY ...)`: the same clangd copy with no configure dependency registered. (`file(COPY_FILE ... ONLY_IF_DIFFERENT)` would be cleaner still but requires CMake 3.21, above the declared `cmake_minimum_required(VERSION 3.10)`.) ## Test plan - [x] Fresh `cmake --preset Release-windows-clang` (LLVM 22, no cache seeding) completes and logs `Enabled SSL backends: Schannel`; the FIONBIO probe passes without the previous `#error` - [x] Full Windows Release build from scratch in the branch worktree (all 1422 targets) - [x] goalc-test suite: 1509 passed, 0 failed - [x] Second consecutive configure with `compile_commands.json` present: the regen rule in `build.ninja` has no `compile_commands.json` input; `<src>/build/compile_commands.json` is still refreshed for clangd - [x] Repeated `ninja` invocations after a full build no longer re-run CMake - [x] macOS Intel and ARM CI green (first exercise of the OpenSSL backend switch) --- I work off a self-hosted forge, so this GitHub account is quiet; the configure logs and ninja dirty-node traces from the investigation are available if anyone wants the raw data. (AI-assisted)
4.9 KiB
Vendored
Generated
Anatomy of a curl security advisory
As described in the vulnerability disclosure policy, when a vulnerability has been reported to the project and and it has been confirmed by the team, we author an advisory document for the issue.
This advisory document should ideally be written in cooperation with the reporter to make sure all the angles and details of the problem are gathered and described correctly and succinctly.
New document
A security advisory for curl is created in the docs/ folder in the
curl-www repository. It should be named
$CVEID.md where $CVEID is the full CVE Id that has been registered for the
flaw. Like CVE-2016-0755. The .md extension of course means that the
document is written using markdown.
The standard way to go about this is to first write the VULNERABILITY
section for the document, so that there is description of the flaw available,
then paste this description into the CVE Id request.
vuln.pm
The new issue should be entered at the top of the list in the file vuln.pm
in the same directory. It holds a large array with all published curl
vulnerabilities. All fields should be filled in accordingly, separated by a
pipe character (|).
The fields for every CVE in vuln.pm are, in order:
- HTML page name
- first vulnerable version
- last vulnerable version
- name of the issue
- CVE Id
- announce date (
YYYYMMDD) - report to the project date (
YYYYMMDD) - CWE
- awarded reward amount (USD)
- area (single word)
- C-issue (
-if not a C issue at all,OVERFLOW,OVERREAD,DOUBLE_FREE,USE_AFTER_FREE,NULL_MISTAKE,UNINIT,BAD_FREE) - affected components:
both,libortool - severity:
low,medium,highorcritical - URL to the initial report (often on HackerOne)
Makefile
The new CVE webpage filename needs to be added in the Makefile's CVELIST
macro.
When the markdown is in place and the Makefile and vuln.pm are updated,
all other files and metadata for all curl advisories and versions get
generated automatically using those files.
Document format
The easy way is to start with a recent previously published advisory and blank out old texts and save it using a new name. Save the subtitles and general layout.
Some details and metadata are extracted from this document so it is important to stick to the existing format.
The first list must be the title of the issue.
VULNERABILITY
The first subtitle should be VULNERABILITY. That should then include a
through and detailed description of the flaw. Including how it can be
triggered and maybe something about what might happen if triggered or
exploited.
INFO
The next section is INFO which adds meta data information about the flaw. It
specifically mentions the official CVE Id for the issue and it must list the
CWE Id, starting on its own line. We write CWE identifiers in advisories with
the full (official) explanation on the right side of a colon. Like this:
CWE-305: Authentication Bypass by Primary Weakness
AFFECTED VERSIONS
The third section first lists what versions that are affected, then adds clarity by stressing what versions that are not affected. A third line adds information about which specific git commit that introduced the vulnerability.
The Introduced-in commit should be a full URL that displays the commit, but
should work as a stand-alone commit hash if everything up to the last slash is
cut out.
An example using the correct syntax:
- Affected versions: curl 7.16.1 to and including 7.88.1
- Not affected versions: curl < 7.16.1 and curl >= 8.0.0
- Introduced-in: https://github.com/curl/curl/commit/2147284cad
THE SOLUTION
This section describes and discusses the fix. The only mandatory information here is the link to the git commit that fixes the problem.
The Fixed-in value should be a full URL that displays the commit, but should
work as a stand-alone commit hash if everything up to the last slash is cut
out.
Example:
- Fixed-in: https://github.com/curl/curl/commit/af369db4d3833272b8ed
RECOMMENDATIONS
This section lists the recommended actions for the users in a top to bottom priority order and should ideally contain three items but no less than two.
The top two are almost always upgrade curl to version XXX and apply the patch to your local version.
TIMELINE
Detail when this report was received in the project. When package distributors were notified (via the distros mailing list or similar)
When the advisory and fixed version are released.
CREDITS
Mention the reporter and patch author at least, then everyone else involved you think deserves a mention.
If you want to mention more than one name, separate the names with comma
(,).
- Reported-by: Full Name
- Patched-by: Full Name