Jay Satiro [Mon, 18 Mar 2019 20:33:58 +0000 (16:33 -0400)]
os400: Disable Alt-Svc by default since it's experimental
Follow-up to 520f0b4 which added Alt-Svc support and enabled it by
default for OS400. Since the feature is experimental, it should be
disabled by default.
Chris Young [Fri, 8 Mar 2019 00:06:59 +0000 (00:06 +0000)]
configure: add --with-amissl
AmiSSL is an Amiga native library which provides a wrapper over OpenSSL.
It also requires all programs using it to use bsdsocket.library
directly, rather than accessing socket functions through clib, which
libcurl was not necessarily doing previously. Configure will now check
for the headers and ensure they are included if found.
* Adjusted unit tests 2056, 2057
* do not generally close connections with CURLAUTH_NEGOTIATE after every request
* moved negotiatedata from UrlState to connectdata
* Added stream rewind logic for CURLAUTH_NEGOTIATE
* introduced negotiatedata::GSS_AUTHDONE and negotiatedata::GSS_AUTHSUCC
* Consider authproblem state for CURLAUTH_NEGOTIATE
* Consider reuse_forbid for CURLAUTH_NEGOTIATE
* moved and adjusted negotiate authentication state handling from
output_auth_headers into Curl_output_negotiate
* Curl_output_negotiate: ensure auth done is always set
* Curl_output_negotiate: Set auth done also if result code is
GSS_S_CONTINUE_NEEDED/SEC_I_CONTINUE_NEEDED as this result code may
also indicate the last challenge request (only works with disabled
Expect: 100-continue and CURLOPT_KEEP_SENDING_ON_ERROR -> 1)
* Consider "Persistent-Auth" header, detect if not present;
Reset/Cleanup negotiate after authentication if no persistent
authentication
* apply changes introduced with #2546 for negotiate rewind logic
Daniel Stenberg [Tue, 12 Mar 2019 07:37:18 +0000 (08:37 +0100)]
memdebug: log pointer before freeing its data
Coverity warned for two potentional "Use after free" cases. Both are false
positives because the memory wasn't used, it was only the actual pointer
value that was logged.
The fix still changes the order of execution to avoid the warnings.
Marcel Raad [Sat, 9 Mar 2019 20:15:18 +0000 (21:15 +0100)]
travis: actually use updated compiler versions
For the Linux builds, GCC 8 and 7 and clang 7 were installed, but the
new GCC versions were only used for the coverage build and for building
nghttp2, while the new clang version was not used at all.
BoringSSL needs to use the default GCC as it respects CC, but not CXX,
so it would otherwise pass gcc 8 options to g++ 4.8 and fail.
examples/externalsocket: add missing close socket calls
.. and for Windows also call WSACleanup since we call WSAStartup.
The example is to demonstrate handling the socket independently of
libcurl. In this case libcurl is not responsible for creating, opening
or closing the socket, it is handled by the application (our example).
Daniel Stenberg [Mon, 11 Mar 2019 10:49:09 +0000 (11:49 +0100)]
multi: removed unused code for request retries
This code was once used for the non multi-interface using code path, but
ever since easy_perform was turned into a wrapper around the multi
interface, this code path never runs.
Jay Satiro [Sat, 9 Mar 2019 08:16:33 +0000 (03:16 -0500)]
doh: inherit some SSL options from user's easy handle
- Inherit SSL options for the doh handle but not SSL client certs,
SSL ALPN/NPN, SSL engine, SSL version, SSL issuer cert,
SSL pinned public key, SSL ciphers, SSL id cache setting,
SSL kerberos or SSL gss-api settings.
- Fix inheritance of verbose setting.
- Inherit NOSIGNAL.
There is no way for the user to set options for the doh (DNS-over-HTTPS)
handles and instead we inherit some options from the user's easy handle.
My thinking for the SSL options not inherited is they are most likely
not intended by the user for the DOH transfer. I did inherit insecure
because I think that should still be in control of the user.
Prior to this change doh did not work for me because CAINFO was not
inherited. Also verbose was set always which AFAICT was a bug (#3660).
Daniel Stenberg [Mon, 4 Mar 2019 15:17:10 +0000 (16:17 +0100)]
ssh: loop the state machine if not done and not blocking
If the state machine isn't complete, didn't fail and it didn't return
due to blocking it can just as well loop again.
This addresses the problem with SFTP directory listings where we would
otherwise return back to the parent and as the multi state machine
doesn't have any code for using CURLM_CALL_MULTI_PERFORM for as long the
doing phase isn't complete, it would return out when in reality there
was more data to deal with.
Jay Satiro [Tue, 26 Feb 2019 07:17:03 +0000 (02:17 -0500)]
multi: support verbose conncache closure handle
- Change closure handle to receive verbose setting from the easy handle
most recently added via curl_multi_add_handle.
The closure handle is a special easy handle used for closing cached
connections. It receives limited settings from the easy handle most
recently added to the multi handle. Prior to this change that did not
include verbose which was a problem because on connection shutdown
verbose mode was not acknowledged.
Ref: https://github.com/curl/curl/pull/3598
Co-authored-by: Daniel Stenberg
Closes https://github.com/curl/curl/pull/3618
The main change here is the timer value that was wrong, it was given in
usecs (ms * 1000), while the itimerspec struct wants nsecs (ms * 1000 *
1000). This resulted in the callback being invoked WAY TOO OFTEN.
As a quick check you can run this command before and after applying this
commit:
Daniel Stenberg [Thu, 28 Feb 2019 19:34:36 +0000 (20:34 +0100)]
threaded-resolver: shutdown the resolver thread without error message
When a transfer is done, the resolver thread will be brought down. That
could accidentally generate an error message in the error buffer even
though this is not an error situationand the transfer would still return
OK. An application that still reads the error buffer could find a
"Could not resolve host: [host name]" message there and get confused.
Reported-by: Michael Schmid
Fixes #3629
Closes #3630
Daniel Stenberg [Wed, 27 Feb 2019 21:30:32 +0000 (22:30 +0100)]
http: set state.infilesize when sending formposts
Without it set, we would unwillingly triger the "HTTP error before end
of send, stop sending" condition even if the entire POST body had been
sent (since it wouldn't know the expected size) which would
unnecessarily log that message and close the connection when it didn't
have to.
Reported-by: Matt McClure
Bug: https://curl.haxx.se/mail/archive-2019-02/0023.html
Closes #3624
Marcel Raad [Tue, 26 Feb 2019 16:38:14 +0000 (17:38 +0100)]
AppVeyor: add classic MinGW build
But use the MSYS2 shell rather than the default MSYS shell because of
POSIX path conversion issues. Classic MinGW is only available on the
Visual Studio 2015 image.
Marcel Raad [Wed, 10 Oct 2018 20:22:06 +0000 (22:22 +0200)]
AppVeyor: add MinGW-w64 build
Add a MinGW-w64 build using CMake's MSYS Makefiles generator.
Use the Visual Studio 2015 image as it has GCC 8, while the
Visual Studio 2017 image only has GCC 7.2.
If the cookieinfo pointer is NULL there really is nothing to save.
Without this fix, we got a problem when a handle was using shared object
with cookies and is told to "FLUSH" it to file (which worked) and then
the share object was removed and when the easy handle was closed just
afterwards it has no cookieinfo and no cookies so it decided to save an
empty jar (overwriting the file just flushed).
Test 1905 now verifies that this works.
Assisted-by: Michael Wallner Assisted-by: Marcel Raad
Closes #3621
Daniel Stenberg [Mon, 25 Feb 2019 10:17:53 +0000 (11:17 +0100)]
urldata: convert bools to bitfields and move to end
This allows the compiler to pack and align the structs better in
memory. For a rather feature-complete build on x86_64 Linux, gcc 8.1.2
makes the Curl_easy struct 4.9% smaller. From 6312 bytes to 6000.
Daniel Stenberg [Mon, 25 Feb 2019 17:12:51 +0000 (18:12 +0100)]
strerror: make the strerror function use local buffers
Instead of using a fixed 256 byte buffer in the connectdata struct.
In my build, this reduces the size of the connectdata struct by 11.8%,
from 2160 to 1904 bytes with no functionality or performance loss.
This also fixes a bug in schannel's Curl_verify_certificate where it
called Curl_sspi_strerror when it should have called Curl_strerror for
string from GetLastError. the only effect would have been no text or the
wrong text being shown for the error.
Daniel Stenberg [Sun, 24 Feb 2019 15:32:04 +0000 (16:32 +0100)]
multi: call multi_done on connect timeouts
Failing to do so would make the CURLINFO_TOTAL_TIME timeout to not get
updated correctly and could end up getting reported to the application
completely wrong (way too small).
Reported-by: accountantM on github
Fixes #3602
Closes #3605
Daniel Stenberg [Fri, 22 Feb 2019 12:44:41 +0000 (13:44 +0100)]
examples: remove recursive calls to curl_multi_socket_action
From within the timer callbacks. Recursive is problematic for several
reasons. They should still work, but this way the examples and the
documentation becomes simpler. I don't think we need to encourage
recursive calls.
Jay Satiro [Thu, 21 Feb 2019 20:30:10 +0000 (15:30 -0500)]
url: change conn shutdown order to unlink data as last step
- Split off connection shutdown procedure from Curl_disconnect into new
function conn_shutdown.
- Change the shutdown procedure to close the sockets before
disassociating the transfer.
Prior to this change the sockets were closed after disassociating the
transfer so SOCKETFUNCTION wasn't called since the transfer was already
disassociated. That likely came about from recent work started in
Jan 2019 (#3442) to separate transfers from connections.
Bug: https://curl.haxx.se/mail/lib-2019-02/0101.html Reported-by: Pavel Löbl
Closes https://github.com/curl/curl/issues/3597
Closes https://github.com/curl/curl/pull/3598
Daniel Stenberg [Tue, 19 Feb 2019 09:02:27 +0000 (10:02 +0100)]
curl: remove MANUAL from -M output
... and remove it from the dist tarball. It has served its time, it
barely gets updated anymore and "everything curl" is now convering all
this document once tried to include, and does it more and better.
In the compressed scenario, this removes ~15K data from the binary,
which is 25% of the -M output.
It remains in the git repo for now for as long as the web site builds a
page using that as source. It renders poorly on the site (especially for
mobile users) so its not even good there.
projects / curl / log