Bernhard Walle [Wed, 17 Nov 2021 10:40:41 +0000 (11:40 +0100)]
configure: fix runtime-lib detection on macOS
With a non-standard installation of openssl we get this error:
checking run-time libs availability... failed
configure: error: one or more libs available at link-time are not available run-time. Libs used at link-time: -lnghttp2 -lssl -lcrypto -lssl -lcrypto -lz
There's already code to set LD_LIBRARY_PATH on Linux, so set
DYLD_LIBRARY_PATH equivalent on macOS.
Daniel Stenberg [Mon, 15 Nov 2021 16:13:17 +0000 (17:13 +0100)]
docs/cmdline-opts: do not say "protocols: all"
Remove the lines saying "protocols: all". It makes the output in the
manpage look funny, and the expectation is probably by default that if
not anything is mentioned about protocols the option apply to them all.
Patrick Monnerat [Mon, 25 Oct 2021 10:58:37 +0000 (12:58 +0200)]
mime: use percent-escaping for multipart form field and file names
Until now, form field and file names where escaped using the
backslash-escaping algorithm defined for multipart mails. This commit
replaces this with the percent-escaping method for URLs.
As this may introduce incompatibilities with server-side applications, a
new libcurl option CURLOPT_MIME_OPTIONS with bitmask
CURLMIMEOPT_FORMESCAPE is introduced to revert to legacy use of
backslash-escaping. This is controlled by new cli tool option
--form-escape.
New tests and documentation are provided for this feature.
Daniel Stenberg [Sat, 13 Nov 2021 10:44:34 +0000 (11:44 +0100)]
configure: better diagnostics if hyper is built wrong
If hyper is indeed present in the specified directory but couldn't be
used to find the correct symbol, then offer a different error message to
better help the user understand the issue.
Suggested-by: Jacob Hoffman-Andrews
Fixes #8001
Closes #8005
When we're reading out plaintext from rustls' internal buffers, we might
get a read of zero bytes (meaning a clean TCP close, including
close_notify). However, we shouldn't return immediately when that
happens, since we may have already copied out some plaintext bytes.
Break out of the loop when we get a read of zero bytes, and figure out
which path we're dealing with.
The update to rustls-ffi 0.8.0 changed handling of EOF and close_notify.
From the CHANGELOG:
> Handling of unclean close and the close_notify TLS alert. Mirroring
> upstream changes, a rustls_connection now tracks TCP closed state like
> so: rustls_connection_read_tls considers a 0-length read from its
> callback to mean "TCP stream was closed by peer." If that happens
> before the peer sent close_notify, rustls_connection_read will return
> RUSTLS_RESULT_UNEXPECTED_EOF once the available plaintext bytes are
> exhausted. This is useful to protect against truncation attacks. Note:
> some TLS implementations don't send close_notify. If you are already
> getting length information from your protocol (e.g. Content-Length in
> HTTP) you may choose to ignore UNEXPECTED_EOF so long as the number of
> plaintext bytes was as expected.
That means we don't need to check for unclean EOF in `cr_recv()`,
because `process_new_packets()` will give us an error if appropriate.
Daniel Stenberg [Sat, 13 Nov 2021 10:50:59 +0000 (11:50 +0100)]
azure: make the "w/o HTTP/SMTP/IMAP" build disable SSL proper
The configure line would previously depend on a configure mistake using
--without-openssl that is fixed and now this configure line needs
adjusting to use --without-ssl.
Note: The list of libraries that rustc tells us we need to include is
longer, but also includes some more platform-specific libraries that I
am not sure how to effectively incorporate. Adding just -lm seems to
solve an immediate problem, so I'm adding just that.
When torture testing 1021, it turns out the Curl_connect_done function
might be called twice and that previously then wrongly cleared the HTTP
pointer in the second invoke.
Stan Hu [Fri, 12 Nov 2021 00:24:57 +0000 (16:24 -0800)]
configure: don't enable TLS when --without-* flags are used
Previously specifying `--without-gnutls` would unexpectedly attempt to
compile with GnuTLS, effectively interpreting this as
`--with-gnutls`. This caused a significant amount of confusion when
`libcurl` was built with SSL disabled since GnuTLS wasn't present.
68d89f24 dropped the `--without-*` options from the configure help, but
`AC_ARG_WITH` still defines these flags automatically. As
https://www.gnu.org/software/autoconf/manual/autoconf-2.60/html_node/External-Software.html
describes, the `action-if-given` is called when the user specifies
`--with-*` or `--without-*` options.
To prevent this confusion, we make the `--without` flag do the right
thing by ignoring the value if it set to "no".
Kevin Burke [Wed, 3 Nov 2021 05:14:02 +0000 (22:14 -0700)]
vtls/rustls: update to compile with rustls-ffi v0.8.0
Some method names, as well as the generated library name, were changed
in a recent refactoring.
Further, change the default configuration instructions to check for
Hyper in either "target/debug" or "target/release" - the latter
contains an optimized build configuration.
Marc Hoersken [Sat, 6 Nov 2021 15:14:17 +0000 (16:14 +0100)]
tests: add Schannel-specific tests and disable unsupported ones
Adds Schannel variants of SSLpinning tests that include the option
--ssl-revoke-best-effort to ignore certificate revocation check
failures which is required due to our custom test CA certificate.
Disable the original variants if the Schannel backend is enabled.
Also skip all IDN tests which are broken while using an msys shell.
This is a step to simplify test exclusions for Windows and MinGW.
Reviewed-by: Jay Satiro Reviewed-by: Marcel Raad Reviewed-by: Daniel Stenberg
Closes #7968
QUIC Transport Parameters Extension has been changed between draft-29
and latest RFC9001. Most notably, its identifier has been updated from
0xffa5 to 0x0039. The version is selected through the QUIC TLS library
via the legacy codepoint.
Disable the usage of legacy codepoint in curl to switch to latest
RFC9001. This is required to be able to keep up with latest QUIC
implementations.
Daniel Stenberg [Tue, 2 Nov 2021 14:16:18 +0000 (15:16 +0100)]
tool_operate: a failed etag save now only fails that transfer
When failing to create the output file for saving an etag, only fail
that particular single transfer and allow others to follow.
In a serial transfer setup, if no transfer at all is done due to them
all being skipped because of this error, curl will output an error
message and return exit code 26.
Added test 369 and 370 to verify.
Reported-by: Earnestly on github
Ref: #7942
Closes #7945
Kevin Burke [Wed, 3 Nov 2021 23:41:57 +0000 (16:41 -0700)]
.github: retry macos "brew install" command on failure
Previously we saw errors attempting to run "brew install", see
https://github.com/curl/curl/runs/4095721123?check_suite_focus=true for
an example, since this command is idempotent, it is safe to run again.
Daniel Stenberg [Wed, 3 Nov 2021 13:25:49 +0000 (14:25 +0100)]
zuul: pin the quiche build to use an older cmake-rs
The latest cmake-rs assumes cmake's --parallel works. That was added in
cmake 3.12, but a lot of our CI builds run on Ubuntu Bionic which only
has cmake 3.10.
Jay Satiro [Tue, 2 Nov 2021 19:34:04 +0000 (15:34 -0400)]
schannel: fix memory leak due to failed SSL connection
- Call schannel_shutdown if the SSL connection fails.
Prior to this change schannel_shutdown (which shuts down the SSL
connection as well as memory cleanup) was not called when the SSL
connection failed (eg due to failed handshake).
Co-authored-by: Gisle Vanem
Fixes https://github.com/curl/curl/issues/7877
Closes https://github.com/curl/curl/pull/7878
Roy Li [Tue, 26 Apr 2016 05:13:01 +0000 (13:13 +0800)]
configure.ac: replace krb5-config with pkg-config
The rationale is that custom *-config tools don't work well when
cross-compiling or using sysroots (such as when using Yocto project) and
require custom fixing for each of them; pkg-config on the other hand
works similarly everywhere.
Signed-off-by: Roy Li <rongqing.li@windriver.com> Signed-off-by: Alexander Kanavin <alex@linutronix.de>
Closes #7916
Marc Hoersken [Sat, 23 Oct 2021 19:30:03 +0000 (21:30 +0200)]
github: add support for Hacktoberfest using labels
Automatically add hacktoberfest-accepted label to PRs opened between
September 30th and November 1st once a commit with a close reference
to it is pushed onto the master branch.
With this workflow we can participate in Hacktoberfest while not
relying on GitHub to identify PRs as merged due to our rebasing.
Requires hacktoberfest-accepted labels to exist for PRs on the
participating repository. Also requires hacktoberfest topic on
the participating repository to avoid applying to forked repos.
Reviewed-by: Daniel Stenberg
Fixes #7865
Closes #7897
Daniel Stenberg [Tue, 26 Oct 2021 15:47:14 +0000 (17:47 +0200)]
http: reject HTTP response codes < 100
... which then also includes negative ones as test 1430 uses.
This makes native + hyper backend act identically on this and therefore
test 1430 can now be enabled when building with hyper. Adjust test 1431
as well.
Patrick Monnerat [Mon, 25 Oct 2021 11:54:57 +0000 (13:54 +0200)]
sendf: accept zero-length data in Curl_client_write()
Historically, Curl_client_write() used a length value of 0 as a marker
for a null-terminated data string. This feature has been removed in
commit f4b85d2. To detect leftover uses of the feature, a DEBUGASSERT
statement rejecting a length with value 0 was introduced, effectively
precluding use of this function with zero-length data.
The current commit removes the DEBUGASSERT and makes the function to
return immediately if length is 0.
A direct effect is to fix trying to output a zero-length distinguished
name in openldap.
Another DEBUGASSERT statement is also rephrased for better readability.
Viktor Szakats [Mon, 25 Oct 2021 19:10:55 +0000 (19:10 +0000)]
Makefile.m32: fix to not require OpenSSL with -libssh2 or -rtmp options
Previously, -libssh2/-rtmp options assumed that OpenSSL is also enabled
(and then failed with an error when not finding expected OpenSSL headers),
but this isn't necessarly true, e.g. when building both libssh2 and curl
against Schannel. This patch makes sure to only enable the OpenSSL backend
with -libssh2/-rtmp, when there was no SSL option explicitly selected.
- Re-implement the logic as a single block of script.
- Also fix an indentation while there.
projects / thirdparty / curl.git / log
