Patrick Monnerat [Tue, 28 Jan 2020 10:56:41 +0000 (11:56 +0100)]
silly web server: silent a compilation warning
Recent gcc warns when byte count of strncpy() equals the destination
buffer size. Since the destination buffer is previously cleared and
the source string is always shorter, reducing the byte count by one
silents the warning without affecting the result.
Daniel Stenberg [Sat, 7 Mar 2020 12:37:50 +0000 (13:37 +0100)]
lib1564: reduce number of mid-wait wakeup calls
This test does A LOT of *wakeup() calls and then calls curl_multi_poll()
twice. The first *poll() is then expected to return early and the second
not - as the first is supposed to drain the socketpair pipe.
It turns out however that when given "excessive" amounts of writes to
the pipe, some operating systems (the Solaris based are known) will
return EAGAIN before the pipe is drained, which in our test case causes
the second *poll() call to also abort early.
This change attempts to avoid the OS-specific behaviors in the test by
reducing the amount of wakeup calls from 1234567 to 10.
Reported-by: Andy Fiddaman
Fixes #5037
Closes #5058
Input buffer filling may delay the data sending if data reads are slow.
To overcome this problem, file and callback data reads do not accumulate
in buffer anymore. All other data (memory data and mime framing) are
considered as fast and still concatenated in buffer.
As this may highly impact performance in terms of data overhead, an early
end of part data check is added to spare a read call.
When encoding a part's data, an encoder may require more bytes than made
available by a single read. In this case, the above rule does not apply
and reads are performed until the encoder is able to deliver some data.
Tests 643, 644, 645, 650 and 654 have been adapted to the output data
changes, with test data size reduced to avoid the boredom of long lists of
1-byte chunks in verification data.
New test 667 checks mimepost using single-byte read callback with encoder.
New test 668 checks the end of part data early detection.
In case a read callback returns a status (pause, abort, eof,
error) instead of a byte count, drain the bytes read so far but
remember this status for further processing.
Takes care of not losing data when pausing, and properly resume a
paused mime structure when requested.
New tests 670-673 check unpausing cases, with easy or multi
interface and mime or form api.
Jay Satiro [Sun, 23 Feb 2020 23:37:09 +0000 (18:37 -0500)]
libssh: Fix matching user-specified MD5 hex key
Prior to this change a match would never be successful because it
was mistakenly coded to compare binary data from libssh to a
user-specified hex string (ie CURLOPT_SSH_HOST_PUBLIC_KEY_MD5).
Jay Satiro [Fri, 6 Mar 2020 04:35:32 +0000 (23:35 -0500)]
multi: skip EINTR check on wakeup socket if it was closed
- Don't check errno on wakeup socket if sread returned 0 since sread
doesn't set errno in that case.
This is a follow-up to cf7760a from several days ago which fixed
Curl_multi_wait to stop busy looping sread on the non-blocking wakeup
socket if it was closed (ie sread returns 0). Due to a logic error it
was still possible to busy loop in that case if errno == EINTR.
Jay Satiro [Tue, 3 Mar 2020 05:25:44 +0000 (00:25 -0500)]
cmake: Show HTTPS-proxy in the features output
- Show HTTPS-proxy in the features output for those backends that
support it: OpenSSL, GnuTLS and NSS.
Prior to this change HTTPS-proxy was missing from the cmake features
output even if curl was built with it. Only cmake output was affected.
Both the library and tool correctly reported the feature.
Bug: https://curl.haxx.se/mail/lib-2020-03/0008.html Reported-by: David Lopes
Closes https://github.com/curl/curl/pull/5025
Steve Holme [Mon, 2 Mar 2020 01:51:49 +0000 (01:51 +0000)]
configure.ac: Disable metalink if mbedTLS is specified
Follow up to cdcc9df1 and #5006. Even though I mentioned mbedTLS as
being one of the backends that metalink needs to be disabled for, I
seem to have included it in the list of allowed SSL/TLS backends in
comnfigure.ac :(
Jay Satiro [Mon, 2 Mar 2020 07:59:40 +0000 (02:59 -0500)]
curl_getenv.3: Fix the memory handling description
- Tell the user to call curl_free() to free the pointer returned by
curl_getenv().
Prior to this change the user was directed to call free(), but that
would not work in cases where the library and application use separate C
runtimes and therefore have separate heap memory management.
Daniel Stenberg [Sat, 29 Feb 2020 11:00:18 +0000 (12:00 +0100)]
http: mark POSTs with no body as "upload done" from the start
As we have logic that checks if we get a >= 400 reponse code back before
the upload is done, which then got confused since it wasn't "done" but
yet there was no data to send!
Reported-by: IvanoG on github
Fixes #4996
Closes #5002
Patrick Monnerat [Wed, 29 Jan 2020 04:28:26 +0000 (05:28 +0100)]
mime: do not perform more than one read in a row
Input buffer filling may delay the data sending if data reads are slow.
To overcome this problem, file and callback data reads do not accumulate
in buffer anymore. All other data (memory data and mime framing) are
considered as fast and still concatenated in buffer.
As this may highly impact performance in terms of data overhead, an early
end of part data check is added to spare a read call.
When encoding a part's data, an encoder may require more bytes than made
available by a single read. In this case, the above rule does not apply
and reads are performed until the encoder is able to deliver some data.
Tests 643, 644, 645, 650 and 654 have been adapted to the output data
changes, with test data size reduced to avoid the boredom of long lists of
1-byte chunks in verification data.
New test 664 checks mimepost using single-byte read callback with encoder.
New test 665 checks the end of part data early detection.
Patrick Monnerat [Tue, 21 Jan 2020 00:33:04 +0000 (01:33 +0100)]
mime: latch last read callback status.
In case a read callback returns a status (pause, abort, eof,
error) instead of a byte count, drain the bytes read so far but
remember this status for further processing.
Takes care of not losing data when pausing, and properly resume a
paused mime structure when requested.
New tests 670-673 check unpausing cases, with easy or multi
interface and mime or form api.
Fixes #4813 Reported-by: MrdUkk on github
Closes #4833
Steve Holme [Sat, 29 Feb 2020 21:44:10 +0000 (21:44 +0000)]
configure.ac: Disable metalink support if an incompatible SSL/TLS specified
tool_metalink only supports cryptography from OpenSSL, GnuTLS, NSS,
The Win32 Crypto library and Apple's Common Crypto library.
If an TLS backend such as mbedTLS or WolfSSL is specified then the
following error is given during compilation along, with a load of
unresolved extern errors:
Can't compile METALINK support without a crypto library.
- FindLibSSH2:
- pass version to FPHSA to show it in the output
- use LIBSSH2_VERSION define to extract the version number in
one shot. This variable exists in the header for 10 years.
- remove unneeded code
- FindNGHTTP2.cmake:
- drop needless FPHSA argument
- mark found variables as advanced
- FindNSS.cmake:
- show version number
- FindCARES.cmake:
- drop default paths
- use FPHSA instead of checking things by hand
- remove needless explict variable dereference
- simplify count_true()
- allow all policies up to version 3.16 to be set to NEW
- do not rerun check for -Wstrict-aliasing=3 every time
In contrast to every other compiler flag this has a = in it, which CMake
can't have in a variable name.
- only read the interesting strings from curlver.h
Reviewed-by: Peter Wu
Closes https://github.com/curl/curl/pull/4975
Jay Satiro [Sat, 29 Feb 2020 08:11:06 +0000 (03:11 -0500)]
runtests: fix output to command log
- Record only the command of the most recently ran test in the command
log.
This is a follow-up to 02988b7 from several weeks ago which fixed
writing to the command log, however it saved all commands for all tests
instead of just the most recently ran test as we would now expect.
projects / thirdparty / curl.git / log
