curl requires `stdint.h` from C99, and no longer builds without it since
v8.18.0 (after dropping VS2008 support). Assume it's available, drop
feature checks.
Also:
- drop duplicate `stdint.h` includes.
- introduce internal `HAVE_UINTPTR_T`, enabled by default.
- OS400: disable `HAVE_UINTPTR_T`.
- build: keep cmake pre-fill and `cmp-config.pl` exception because cmake
and autotools both detect `stdint.h` implicitly.
Co-authored-by: Dan Fandrich
Ref: #20405
Ref: #20384
Follow-up to 2e1a045d89#17931Closes#20406
The test for restarting the server during ongoing transfers does not
work reliably for HTTP/3. This seems due to the nature of UDP/QUIC where
the client may learn about a closed connection at any time, not only
when starting a new request.
Remove the test.
Closes#20458
Report UDP packets with SOCKEMSGSIZE as being "sent" to progress
the send buffer properly on PMTUD probes.
Reported-by: Daniil Gentili
Fixes#20440Closes#20448
For HTTP/2, add error code description to close failures.
For HTTP/3, add special handling like in HTTP/2 when streams
have been rejected or an error comes during the response body
and we are not interested in the body.
Closes#20207
Replacing `_WIN32`.
Also:
- tool_doswin: guard possibly non-portable socket code with
`USE_WINSOCK`. The socket is cast to `HANDLE` and passed to
win32 API `SetStdHandle(STD_INPUT_HANDLE, ...)`.
- lib/setup-win32.h: move `#undef`s before their `#define` pair.
Closes#20455
For general readability. Also to match the rest of the source code.
- bump `send()` result type from `int` to `ssize_t`.
- fix an `int` to be `curl_socklen_t`.
- `.S_un.S_addr` -> `.s_addr`.
- `SD_RECEIVE` -> `SHUT_RD`.
- `SD_SEND` -> `SHUT_WR`.
Follow-up to a81ab3e6db#20452
Follow-up to 9a2663322c#17572Closes#20457
A cache entry created by windows-2022 is not picked up by
windows-11-arm. Also a cache created by windows-11-arm is not picked up
by windows-2022. Possibly related to this filed in 2025 June:
https://github.com/actions/cache/issues/1622. Also tried
`enableCrossOsArchive` to no avail. Unclear if these two runners count
as distinct operating systems, I'd guess not. Cache entries are
identical on the web UI. Via GH API they show up with the same cache key
bot different "version" (hash) and different sizes, possibly due to the
zstd vs. gzip bug above.
Fixing (identical error text on either runner):
```
Error: Failed to restore cache entry. Exiting as fail-on-cache-miss is set. Input key: Windows-stunnel-5.76-amd64
```
Also fix a silly typo in the shell value.
Follow-up to 0f54ca6150#20454Closes#20456
Since 9d8998c994, the setopt code changes input DEFAULT to an
actual more specific TLS version (1.2) for the backends to use and check
for.
This means that the default value (0L) cannot and should not actually be
used when the TLS backends run. This change adds asserts to verify that
and removes code that accepts the DEFAULT value as a valid version with
the TLS version functions' logic.
Applications can still set a specific lower version if they want (1, 1.0
or 1.1).
Closes#20453
For general readability. Also to match the rest of the source code.
- `SOCKADDR` -> `struct sockaddr`
- `SOCKADDR_IN` -> `struct sockaddr_in`
- `== SOCKET_ERROR` -> `== -1` or silent `!= 0`
Follow-up to 9a2663322c#17572Closes#20452
OpenSSL records its peer verification status inside its SSL_SESSION
objects. When a session is later reused, the SSL connection inherits
this verify status.
Session keys prevent reuse of sessions between connections that verify
the peer and those who do not. However, when Apple SecTrust is used
to verify a connection, this does not update the Sessions verify
status (and there is no setter). On session reuse, OpenSSL fails
the verification and Apple SecTrust cannot verify either since the
certificate peer chain is not available.
Fix this by checking the verification status on session reuse and
remove the session again if the peer needs to be verified, but the
session is not.
Reported-by: Christian Schmitza
Fixes#20435Closes#20446
... between curl_multi_perform and curl_multi_wait/poll documentation
The `curl_multi_perform` documentation uses integer contextual
conversion to `bool` to test the resulting `CURLMcode`, while other
functions like `curl_multi_wait` and `curl_multi_poll` test against
`CURLM_OK`. (I was initially confused by this as it looked like some
docs call curl_multi_wait/poll on error, while some called only on
success. But that was my misread, not a docs problem.)
Also fixed the example to print which function call failed; previously
an error reported by `curl_multi_perform` was printed as a failure of
`curl_multi_wait`.
Closes#20444
To avoid noise due to local C files when using automatic local checksrc
checks (e.g. via CMake `-DCURL_LINT=ON` option, or `curl-lint` target).
Also replace single-quote with double-quote in external git command, for
portability.
Follow-up to 88ff396549#17882
Follow-up to e785e898a6#17376Closes#20439
GCC 15.2 warns when assigning computed "all" bitmask macros to 32-bit
flag types because negated masks expand to the full width of unsigned
long/long on 64-bit platforms.
Mask these macros to a 32-bit domain so they do not set high bits and
avoid -Woverflow/-Wconversion warnings in callers.
Reported-by: Patrick Monnerat
Fixes#20242Closes#20416
The -J / --remote-header-name logic now records the file name part used
in the redirects so that it can use the last one as a name if no
Content-Disposition header arrives.
Add tests to verify:
1641: -J with a redirect and extract the CD contents in the second
response
1642: -J with a redirect but no Content-Disposition, use the name from
the Location: header
1643: -J with two redirects, using the last file name and also use
queries and fragments to verify them stripped off
Closes#20430
For consistency.
Also:
- one remaining in `src/tool_writeout.c`.
- replace casting an `int` to `CURLcode`.
- lib758: rename `CURLMcode` `result` to `mresult`.
- move literals to the right side of if expressions.
Follow-up to d0dc6e2ec0#20426
Follow-up to 56f600ec23Closes#20432
Via options:
- `BROTLI_USE_STATIC_LIBS`
- `CARES_USE_STATIC_LIBS`
- `LIBSSH_USE_STATIC_LIBS`
- `LIBSSH2_USE_STATIC_LIBS`
- `MBEDTLS_USE_STATIC_LIBS`
- `NGHTTP2_USE_STATIC_LIBS`
- `NGHTTP3_USE_STATIC_LIBS`
- `NGTCP2_USE_STATIC_LIBS`
- `ZSTD_USE_STATIC_LIBS`
When enabled, make a "best effort" finding static libs first and set
the "build static" macro (on Windows) as required by the dependency.
When doing `pkg-config`-based detections, make curl select the static
configuration, which shall set the "build static" macro also.
These options resemble CMake's `OPENSSL_USE_STATIC_LIBS` and
`ZLIB_USE_STATIC_LIBS` (the latter does not support `pkg-config` as of
CMake v4.2.2).
Shared/static library selection based on loose filename conventions is
fragile and prone to break if the non-static-suffixed library is found
and happens to be a shared library, or, if the linker decides to pick up
a shared copy (e.g. `.a.dll`) that shadows the static one. It may help
to provide either static or shared, but not both, on the disk, and match
that with this setting.
Experimental.
Ref: #20013Closes#20015
To build all targets in a single go, meaning curl and libcurl as normal,
and tests and examples in addition. To build-test everything without
running multiple cmake commands.
Enable with:
- `-DCURL_BUILD_EVERYTHING=ON`
Special values: `QUICK` to build examples quickly for build test,
`NOEXAMPLES` to not build examples.
A well-equipped build takes 2.8s of configure time, and 1.7s to build
everything (shared, unity, ninja, prefill), 1.4s with `QUICK`. Without
this option it takes <1s to build curl/libcurl.
Also: streamline `CURL_LINT` internal logic.
Closes#20429
While working #16973, the binutils ld lib order workaround logic
regressed so that it modified the wrong target, writing into the system
`ZLIB::ZLIB` and `OpenSSL::Crypto` ones a `INTERFACE_LINK_LIBRARIES`
property, instead of creating CURL-namespaced targets. Oddly enough,
this also fixed the binutils ld lib ordering issue. It seems this
property makes CMake insert each referenced library in two more
positions (not at the very end though), which allows ld to resolve all
symbols in the cases tested in CI.
Fix by creating the indented namespaced targets, and also creating
these in `curl-config.cmake` to be available when consuming libcurl.
Note that the logic continues doing `get_target_property()` on the two
system targets above. If these targets are defined manually and miss
the `LOCATION` propery, or are defined as aliases, this command may
fail. curl expects these targets be created by CMake's `FindZLIB` and
`FindOpenSSL` built-in Find modules (or ones compatible). Ref: #20419
The binutils ld issue is reproduced by these CI jobs:
- Linux gcc glibc (amd64, arm64)
- Windows gcc zlib-classic (x64)
Currently using this curl-for-win revision:
7d12669daf
Examples:
https://github.com/curl/curl/actions/runs/21332437230/job/61399234023?pr=20427https://github.com/curl/curl/actions/runs/21332437230/job/61399234033?pr=20427
Comparison of lib orders, as passed by CMake to the linker:
without workaround (possibly breaking binutils `ld`):
```diff
-framework [...]
libssl.dylib
libcrypto.dylib
libz.tbd
-lssh2 -lidn2
libldap.tbd
liblber.tbd
-lbrotlidec -lbrotlicommon -lzstd -lnghttp2 -lpsl -lrtmp
-lz
-lssl
-lcrypto
```
before this patch:
```diff
-framework [...]
libssl.dylib
libcrypto.dylib
libz.tbd
+libcrypto.dylib <== inserted via `INTERFACE_LINK_LIBRARIES`
+libz.tbd <== inserted via `INTERFACE_LINK_LIBRARIES`
-lssh2 -lidn2
libldap.tbd
liblber.tbd
+libcrypto.dylib <== inserted via `INTERFACE_LINK_LIBRARIES`
+ibz.tbd <== inserted via `INTERFACE_LINK_LIBRARIES`
-lbrotlidec -lbrotlicommon -lzstd -lnghttp2 -lpsl -lrtmp
-lz
-lssl
-lcrypto
```
after this patch:
```diff
-framework [...]
libssl.dylib
libcrypto.dylib
libz.tbd
-lssh2 -lidn2
libldap.tbd
liblber.tbd
-lbrotlidec -lbrotlicommon -lzstd -lnghttp2 -lpsl -lrtmp
-lz
-lssl
-lcrypto
+libcrypto.dylib <== inserted via `CURL::OpenSSL_Crypto`
+libz.tbd <== inserted via `CURL::ZLIB`
```
Bug: https://github.com/curl/curl/pull/20382#discussion_r2716660108
Reverts: 615c43eae8
Follow-up to 16f073ef49#16973Closes#20427
And a few variables around.
There remain cases where the accepted pointer is const, yet the returned
pointer is written to.
Partly addressing (glibc 2.43):
```
* For ISO C23, the functions bsearch, memchr, strchr, strpbrk, strrchr,
strstr, wcschr, wcspbrk, wcsrchr, wcsstr and wmemchr that return
pointers into their input arrays now have definitions as macros that
return a pointer to a const-qualified type when the input argument is
a pointer to a const-qualified type.
```
Ref: https://lists.gnu.org/archive/html/info-gnu/2026-01/msg00005.html
Reported-by: Rudi Heitbaum
Ref: #20420Closes#20421
- set `CURL_DROP_UNUSED=ON` for Windows (MSVC) to make the analysis step
faster: 1m30s -> 1m15s
- enable `CURL_WERROR=ON` in all builds, to catch potential build issues
in addition to running CodeQL. To make these builds useful as regular
build tests too.
- add links to CodeQL Actions documentation.
- delete test data C files after checkout in an attempt to remove them
from CodeQL code coverage stats.
Closes#20418
- Use upper limit INT_MAX instead of UINT_MAX.
UINT_MAX doesn't work as the max value for the variable since it is
passed as a long and becomes -1 on platforms that have same sized
int and long, like Windows.
Closes https://github.com/curl/curl/pull/20414
Replace Chocolatey install with direct download and unpack. To reduce
CI dependencies (Chocolatey, NuGet), improve install performance
(10s -> 1s) and hopefully reliability. Last but not least to enable it
for the Cygwin CI job.
Caveats:
- Need to bump stunnel versions manually (2-3 times a year).
Renovate could likely do it, but I failed to understand its
documentation and miss tooling/interface to make tests.
- FIPS not enabled. (can be done if necessary)
- Possibly losing checksum verification (not sure if Chocolatey did it
automatically for this package.)
Also:
- Increase minimum tests by 100 for the Cygwin job.
Ref: #16819 (earlier attempt)
Ref: https://www.stunnel.org/archive/
Ref: https://www.githubstatus.com/incidents/cqb5hcy0gx18
Follow-up to d176f58a20#20413
Follow-up to 19b1e44660#20409Closes#20410
Fixing (Seen when enabling stunnel for Cygwin in CI):
```
=== Start of file https_stunnel.log
[ ] Initializing inetd mode configuration
[ ] Running on Windows 6.2
[...]
[.] Reading configuration from file /cygdrive/d/a/curl/curl/bld/tests/log/6/server/https_stunnel.conf
[!] Cannot open configuration file
```
Ref: https://github.com/curl/curl/actions/runs/21289677523/job/61279662459?pr=20410
Cherry-picked from #20410Closes#20413
