Changes

_ _ ____ _
___| | | | _ \| |
/ __| | | | |_) | |
| (__| |_| | _ <| |___
\___|\___/|_| \_\_____|
Changelog
Version 7.64.0 (6 Feb 2019)
Daniel Stenberg (6 Feb 2019)

- RELEASE-NOTES: 7.64.0
- RELEASE-PROCEDURE: update the release calendar
- THANKS: 7.64.0 status
Daniel Gustafsson (5 Feb 2019)

- ROADMAP: remove already performed item
Commit 7a09b52c98ac8d840a8a9907b1a1d9a9e684bcf5 introduced support

for the draft-ietf-httpbis-cookie-alone-01 cookie draft, and while
the entry was removed from the TODO it was mistakenly left here.
Fix by removing and rewording the entry slightly.
Closes #3530
Reviewed-by: Daniel Stenberg <daniel@haxx.se>
- [Etienne Simard brought this change]
CONTRIBUTE.md: Fix grammatical errors
Fix grammatical errors making the document read better. Also fixes
a typo.
Closes #3525
Reviewed-by: Daniel Gustafsson <daniel@yesql.se>

- [Julian Z brought this change]
docs: use $(INSTALL_DATA) to install man page
Fixes #3518
Closes #3522
Jay Satiro (4 Feb 2019)

- [Ladar Levison brought this change]
runtests.pl: Fix perl call to include srcdir
- Use explicit include opt for perl calls.
Prior to this change some scripts couldn't find their dependencies.
At the top, perl is called using with the "-Isrcdir" option, and it
works:
https://github.com/curl/curl/blob/curl-7_63_0/tests/runtests.pl#L183
But on line 3868, that option is omitted. This caused problems for me,
as the symbol-scan.pl script in particular couldn't find its
dependencies properly:
https://github.com/curl/curl/blob/curl-7_63_0/tests/runtests.pl#L3868
This patch fixes that oversight by making calls to perl sub-shells

uniform.
Closes https://github.com/curl/curl/pull/3496

- [Daniel Gustafsson brought this change]
smtp: avoid risk of buffer overflow in strtol
If the incoming len 5, but the buffer does not have a termination
after 5 bytes, the strtol() call may keep reading through the line
buffer until is exceeds its boundary. Fix by ensuring that we are
using a bounded read with a temporary buffer on the stack.
Bug: https://curl.haxx.se/docs/CVE-2019-3823.html
Reported-by: Brian Carpenter (Geeknik Labs)
CVE-2019-3823
- ntlm: fix *_type3_message size check to avoid buffer overflow
Reported-by: Wenxiang Qian
CVE-2019-3822
- NTLM: fix size check condition for type2 received data
Reported-by: Wenxiang Qian
CVE-2018-16890
Marcel Raad (1 Feb 2019)

- [georgeok brought this change]
spnego_sspi: add support for channel binding
Attempt to add support for Secure Channel binding when negotiate

authentication is used. The problem to solve is that by default IIS
accepts channel binding and curl doesn't utilise them. The result was a
401 response. Scope affects only the Schannel(winssl)-SSPI combination.
Fixes https://github.com/curl/curl/issues/3503

- RELEASE-NOTES: synced
- schannel: stop calling it "winssl"
Stick to "Schannel" everywhere. The configure option --with-winssl is

kept to allow existing builds to work but --with-schannel is added as an
alias.
Closes #3504
- multi: set the EXPIRE_*TIMEOUT timers at TIMER_STARTSINGLE time
To make sure Curl_timeleft() also thinks the timeout has been reached
when one of the EXPIRE_*TIMEOUTs expires.
Bug: https://curl.haxx.se/mail/lib-2019-01/0073.html
Reported-by: Zhao Yisha
Closes #3501
- [John Marshall brought this change]
doc: use meaningless port number in CURLOPT_LOCALPORT example
Use an ephemeral port number here; previously the example had 8080
which could be confusing as the common web server port number might
be misinterpreted as suggesting this option affects the remote port.
URL: https://curl.haxx.se/mail/lib-2019-01/0084.html
Closes #3513
GitHub (29 Jan 2019)

- [Gisle Vanem brought this change]
Escape the '\'
A backslash should be escaped in Roff / Troff.
Jay Satiro (29 Jan 2019)

- TODO: WinSSL: 'Add option to disable client cert auto-send'
By default WinSSL selects and send a client certificate automatically,

but for privacy and consistency we should offer an option to disable the
default auto-send behavior.
Reported-by: Jeroen Ooms
Closes https://github.com/curl/curl/issues/2262
Daniel Stenberg (28 Jan 2019)

- [Jeremie Rapin brought this change]
sigpipe: if mbedTLS is used, ignore SIGPIPE
mbedTLS doesn't have a sigpipe management. If a write/read occurs when

the remote closes the socket, the signal is raised and kills the
application. Use the curl mecanisms fix this behavior.
Signed-off-by: Jeremie Rapin <j.rapin@overkiz.com>
Closes #3502
- unit1653: make it survive torture tests
Jay Satiro (28 Jan 2019)

- [Michael Kujawa brought this change]
timeval: Disable MSVC Analyzer GetTickCount warning
Compiling with msvc /analyze and a recent Windows SDK warns against
using GetTickCount (Suggests to use GetTickCount64 instead.)
Since GetTickCount is only being used when GetTickCount64 isn't

available, I am disabling that warning.

- configure: rewrite --enable-code-coverage
The previously used ax_code_coverage.m4 is not license compatible and

must not be used.
Reported-by: William A. Rowe Jr

Fixes #3497
Closes #3499
- [Felix Hädicke brought this change]
setopt: enable CURLOPT_SSH_KNOWNHOSTS and CURLOPT_SSH_KEYFUNCTION for libssh
CURLOPT_SSH_KNOWNHOSTS and CURLOPT_SSH_KEYFUNCTION are supported for

libssh as well. So accepting these options only when compiling with
libssh2 is wrong here.
Fixes #3493
Closes #3494
- [Felix Hädicke brought this change]
libssh: do not let libssh create socket
By default, libssh creates a new socket, instead of using the socket

created by curl for SSH connections.
Pass the socket created by curl to libssh using ssh_options_set() with

SSH_OPTIONS_FD directly after ssh_new(). So libssh uses our socket
instead of creating a new one.
This approach is very similar to what is done in the libssh2 code, where
the socket created by curl is passed to libssh2 when
libssh2_session_startup() is called.
Fixes #3491
Closes #3495
- [Archangel_SDY brought this change]
schannel: preserve original certificate path parameter
Fixes #3480
Closes #3487
- KNOWN_BUGS: tests not compatible with python3
Closes #3289
[skip ci]
Daniel Gustafsson (20 Jan 2019)

- memcmp: avoid doing single char memcmp
There is no real gain in performing memcmp() comparisons on single

characters, so change these to array subscript inspections which
saves a call and makes the code clearer.
Closes #3486
Reviewed-by: Jay Satiro <raysatiro@yahoo.com>

- COPYING: it's 2019
[skip ci]
- [hhb brought this change]
configure: fix recv/send/select detection on Android
This reverts commit d4f25201fb7da03fc88f90d51101beb3d0026db9.
The overloadable attribute is removed again starting from

NDK17. Actually they only exist in two NDK versions (15 and 16). With
overloadable, the first condition tried will succeed. Results in wrong
detection result.
Closes #3484
Marcel Raad (19 Jan 2019)

- [georgeok brought this change]
ntlm_sspi: add support for channel binding
Windows extended potection (aka ssl channel binding) is required

to login to ntlm IIS endpoint, otherwise the server returns 401
responses.
Fixes #3280
Closes #3321

- schannel: on connection close there might not be a transfer
Reported-by: Marcel Raad

Fixes #3412
Closes #3483
- [Joel Depooter brought this change]
ssh: log the libssh2 error message when ssh session startup fails
When a ssh session startup fails, it is useful to know why it has

failed. This commit changes the message from:
"Failure establishing ssh session"
to something like this, for example:
"Failure establishing ssh session: -5, Unable to exchange encryption keys"
Closes #3481
Alessandro Ghedini (16 Jan 2019)

- Fix typo in manpage

Sergei Nikulov (16 Jan 2019)

- cmake: updated check for HAVE_POLL_FINE to match autotools

- curl-compilers.m4: check for __ibmxl__ to detect xlclang
Follow-up to 2fa0d57e2e3. The __xlc__ symbol is only defined there if a

particular flag is used for legacy macros.
Fixes #3474
Closes #3479
- openssl: fix the SSL_get_tlsext_status_ocsp_resp call
.... to not pass in a const in the second argument as that's not how it
is supposed to be used and might cause compiler warnings.
Reported-by: Pavel Pavlov

Fixes #3477
Closes #3478
- curl-compilers.m4: detect xlclang
Since it isn't totally clang compatible, we detect this IBM clang

front-end and if detected, avoids some clang specific magic.
Reported-by: Kees Dekker

Fixes #3474
Closes #3476
- README: add codacy code quality badge
[skip ci]
- extract_if_dead: follow-up to 54b201b48c90a
extract_if_dead() dead is called from two functions, and only one of

them should get conn->data updated and now neither call path clears it.
scan-build found a case where conn->data would be NULL dereferenced in

ConnectionExists() otherwise.
Closes #3473
- multi: remove "Dead assignment"
Found by scan-build. Follow-up to 4c35574bb785ce.

Closes #3471
- tests: move objnames-* from lib into tests
Since they're used purely for testing purposes, I think they should
rather be stored there.
Closes #3470

- travis: added cmake build for osx

- [Frank Gevaerts brought this change]
cookie: fix comment typo (url_path_len -> uri_path_len)
Closes #3469

- winbuild: conditionally use /DZLIB_WINAPI
zlibwapi.lib (dynamic library) and zlibstat.lib (static library) have

the ZLIB_WINAPI define set by default. Using them requires that define
too.
Ref: https://zlib.net/DLL_FAQ.txt

- src/Makefile: make 'tidy' target work for metalink builds
- extract_if_dead: use a known working transfer when checking connections
Make sure that this function sets a proper "live" transfer for the
connection before calling the protocol-specific connection check
function, and then clear it again afterward as a non-used connection has
no current transfer.

Reviewed-by: Marcel Raad
Reviewed-by: Daniel Gustafsson
Fixes #3463
Closes #3464
- openssl: adapt to 3.0.0, OpenSSL_version_num() is deprecated
OpenSSL_version() replaces OpenSSL_version_num()
Closes #3462

- cmake: added checks for HAVE_VARIADIC_MACROS_C99 and HAVE_VARIADIC_MACROS_GCC

- urldata: rename easy_conn to just conn
We use "conn" everywhere to be a pointer to the connection.
Introduces two functions that "attaches" and "detaches" the connection

to and from the transfer.
Going forward, we should favour using "data->conn" (since a transfer

always only has a single connection or none at all) to "conn->data"
(since a connection can have none, one or many transfers associated with
it and updating conn->data to be correct is error prone and a frequent
reason for internal issues).
Closes #3442
- tool_cb_prg: avoid integer overflow
When calculating the progress bar width.
Reported-by: Peng Li
Fixes #3456
Closes #3458

- travis: turn off copyright year checks in checksrc
Invoking the maintainer intended COPYRIGHTYEAR check for everyone

in the PR pipeline is too invasive, especially at the turn of the
year when many files get affected. Remove and leave it as a tool
for maintainers to verify patches before commits.
This reverts f7bdf4b2e1d81b2652b81b9b3029927589273b41.
After discussion with: Daniel Stenberg

- KNOWN_BUGS: cmake makes unusable tool_hugehelp.c with MinGW
Closes #3125
- KNOWN_BUGS: Improve --data-urlencode space encoding
Closes #3229
Patrick Monnerat (10 Jan 2019)

- os400: add a missing closing bracket
See https://github.com/curl/curl/issues/3453#issuecomment-453054458
Reported-by: jonrumsey on github
- os400: fix extra parameter syntax error.
Reported-by: jonrumsey on github

Closes #3453

- test1558: verify CURLINFO_PROTOCOL on file:// transfer
Attempt to reproduce issue #3444.

Closes #3447
- xattr: strip credentials from any URL that is stored
Both user and password are cleared uncondtitionally.
Added unit test 1621 to verify.
Fixes #3423
Closes #3433
- cookies: allow secure override when done over HTTPS
Added test 1562 to verify.

Fixes #3445
Closes #3450
- multi: multiplexing improvements
Fixes #3436
Closes #3448
Problem 1
After LOTS of scratching my head, I eventually realized that even when doing
10 uploads in parallel, sometimes the socket callback to the application that
tells it what to wait for on the socket, looked like it would reflect the
status of just the single transfer that just changed state.
Digging into the code revealed that this was indeed the truth. When multiple
transfers are using the same connection, the application did not correctly get
the *combined* flags for all transfers which then could make it switch to READ
(only) when in fact most transfers wanted to get told when the socket was
WRITEABLE.
Problem 1b
A separate but related regression had also been introduced by me when I

cleared connection/transfer association better a while ago, as now the logic
couldn't find the connection and see if that was marked as used by more
transfers and then it would also prematurely remove the socket from the socket
hash table even in times other transfers were still using it!
Fix 1
Make sure that each socket stored in the socket hash has a "combined" action
field of what to ask the application to wait for, that is potentially the ORed
action of multiple parallel transfers. And remove that socket hash entry only
if there are no transfers left using it.
Problem 2
The socket hash entry stored an association to a single transfer using that
socket - and when curl_multi_socket_action() was called to tell libcurl about
activities on that specific socket only that transfer was "handled".
This was WRONG, as a single socket/connection can be used by numerous parallel

transfers and not necessarily a single one.
Fix 2
We now store a list of handles in the socket hashtable entry and when libcurl
is told there's traffic for a particular socket, it now iterates over all
known transfers using that single socket.
- test1561: improve test name
[skip ci]
- [Katsuhiko YOSHIDA brought this change]
cookies: skip custom cookies when redirecting cross-site
Closes #3417
- THANKS: fixups and a dedupe
[skip ci]
- timediff: fix math for unsigned time_t
Closes #3449
- [Bernhard M. Wiedemann brought this change]
tests: allow tests to pass by 2037-02-12
similar to commit f508d29f3902104018
Closes #3443
- [Brad Spencer brought this change]
curl_multi_remove_handle() don't block terminating c-ares requests
Added Curl_resolver_kill() for all three resolver modes, which only

blocks when necessary, along with test 1592 to confirm
curl_multi_remove_handle() doesn't block unless it must.
Closes #3428
Fixes #3371
- Revert "http_negotiate: do not close connection until negotiation is completed"
This reverts commit 07ebaf837843124ee670e5b8c218b80b92e06e47.
This also reopens PR #3275 which brought the change now reverted.
Fixes #3384
Closes #3439
- curl/urlapi.h: include "curl.h" first
This allows programs to include curl/urlapi.h directly.

Reported-by: Ben Kohler
Fixes #3438
Closes #3441

- VS projects: fix build warning
Starting with Visual Studio 2017 Update 9, Visual Studio doesn't like
the MinimalRebuild option anymore and warns:
cl : Command line warning D9035: option 'Gm' has been deprecated and
will be removed in a future release
The option can be safely removed so that the default is used.
- schannel: fix compiler warning
When building with Unicode on MSVC, the compiler warns about freeing a
pointer to const in Curl_unicodefree. Fix this by declaring it as
non-const and casting the argument to Curl_convert_UTF8_to_tchar to
non-const too, like we do in all other places.

- [Rikard Falkeborn brought this change]
printf: introduce CURL_FORMAT_TIMEDIFF_T
printf: fix format specifiers
Closes #3426
- libtest/stub_gssapi: use "real" snprintf
... since it doesn't link with libcurl.
Reverts the commit dcd6f81025 changes from this file.
Reported-by: Shlomi Fish
Reviewed-by: Kamil Dudka
Closes #3434
- INTERNALS: correct some outdated function names

Closes #3431
- docs/version.d: mention MultiSSL

Closes #3432

examples: Update .gitignore
Add a few missing examples to make `make examples` not leave the
workspace in a dirty state.
Closes #3427
- THANKS: add more missing names
Add Adrian Burcea who made the artwork for the curl://up 2018 event
which was held in Stockholm, Sweden.
- docs: mention potential leak in curl_slist_append
When a non-empty list is appended to, and used as the returnvalue,

the list pointer can leak in case of an allocation failure in the
curl_slist_append() call. This is correctly handled in curl code
usage but we weren't explicitly pointing it out in the API call
documentation. Fix by extending the RETURNVALUE manpage section
and example code.
Closes #3424
Reported-by: dnivras on github

- tvnow: silence conversion warnings
MinGW-w64 defaults to targeting Windows 7 now, so GetTickCount64 is

used and the milliseconds are represented as unsigned long long,
leading to a compiler warning when implicitly converting them to long.

- THANKS: dedupe more names
Researched-by: Tae Wong

- [Markus Moeller brought this change]
ntlm: update selection of type 3 response
NTLM2 did not work i.e. no NTLMv2 response was created. Changing the
check seems to work.
Ref: https://winprotocoldoc.blob.core.windows.net/productionwindowsarchives/MS-
NLMP/[MS-NLMP].pdf
Daniel Stenberg (31 Dec 2018)

- THANKS: added missing names from year <= 2000
Due to a report of a missing name in THANKS I manually went through an

old CHANGES.0 file and added many previously missing names here.
Daniel Gustafsson (30 Dec 2018)

- urlapi: fix parsing ipv6 with zone index
The previous fix for parsing IPv6 URLs with a zone index was a paddle
short for URLs without an explicit port. This patch fixes that case
and adds a unit test case.
This bug was highlighted by issue #3408, and while it's not the full
fix for the problem there it is an isolated bug that should be fixed
regardless.
Closes #3411
Reported-by: GitYuanQu on github

- THANKS: dedupe Guenter Knauf
Reported-by: Tae Wong
- THANKS: missing name from the 6.3.1 release!

- [Claes Jakobsson brought this change]
hostip: support wildcard hosts
This adds support for wildcard hosts in CURLOPT_RESOLVE. These are

try-last so any non-wildcard entry is resolved first. If specified,
any host not matched by another CURLOPT_RESOLVE config will use this
as fallback.
Example send a.com to 10.0.0.1 and everything else to 10.0.0.2:

curl --resolve *:443:10.0.0.2 --resolve a.com:443:10.0.0.1 \
https://a.com https://b.com
This is probably quite similar to using:

--connect-to a.com:443:10.0.0.1:443 --connect-to :443:10.0.0.2:443
Closes #3406
- url: fix incorrect indentation
Patrick Monnerat (26 Dec 2018)

- os400: upgrade ILE/RPG binding.
- Trailer function support.
- http 0.9 option.
- curl_easy_upkeep.

- FAQ: remove mention of sourceforge for github
The project bug tracker is no longer hosted at sourceforge but is now

hosted on the curl Github page. Update the FAQ to reflect.
Closes #3410
- openvms: fix typos in documentation
- openvms: fix OpenSSL discovery on VAX
The DCL code had a typo in one of the commands which would make the
OpenSSL discovery on VAX fail. The correct syntax is F$ENVIRONMENT.
Closes #3407
Reviewed-by: Viktor Szakats <commit@vszakats.net>

- [Ruslan Baratov brought this change]
cmake: use lowercase for function name like the rest of the code
Reviewed-by: Sergei Nikulov
closes #3196
- Revert "libssh: no data pointer == nothing to do"
This reverts commit c98ee5f67f497195c9 since commit f3ce38739fa fixed the

problem in a more generic way.
- disconnect: set conn->data for protocol disconnect
Follow-up to fb445a1e18d: Set conn->data explicitly to point out the

current transfer when invoking the protocol-specific disconnect function
so that it can work correctly.
Bug: https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=12173
Jay Satiro (23 Dec 2018)

- [Pavel Pavlov brought this change]
timeval: Use high resolution timestamps on Windows
- Use QueryPerformanceCounter on Windows Vista+
There is confusing info floating around that QueryPerformanceCounter

can leap etc, which might have been true long time ago, but no longer
the case nowadays (perhaps starting from WinXP?). Also, boost and
std::chrono::steady_clock use QueryPerformanceCounter in a similar way.
Prior to this change GetTickCount or GetTickCount64 was used, which has

lower resolution. That is still the case for <= XP.

- libssh: no data pointer == nothing to do
- conncache_unlock: avoid indirection by changing input argument type
- disconnect: separate connections and easy handles better
Do not assume/store assocation between a given easy handle and the

connection if it can be avoided.
Long-term, the 'conn->data' pointer should probably be removed as it is a

little too error-prone. Still used very widely though.
Reported-by: masbug on github

Fixes #3391
Closes #3400
- libssh: free sftp_canonicalize_path() data correctly
Assisted-by: Harry Sintonen
Fixes #3402
Closes #3403
- http: added options for allowing HTTP/0.9 responses
Added CURLOPT_HTTP09_ALLOWED and --http0.9 for this purpose.
For now, both the tool and library allow HTTP/0.9 by default.
docs/DEPRECATE.md lays out the plan for when to reverse that default: 6
months after the 7.64.0 release. The options are added already now so
that applications/scripts can start using them already now.
Fixes #2873
Closes #3383
- if2ip: remove unused function Curl_if_is_interface_name
Closes #3401
- http2: clear pause stream id if it gets closed
Reported-by: Florian Pritz
Fixes #3392
Closes #3399

- [David Garske brought this change]
wolfssl: Perform cleanup
This adds a cleanup callback for cyassl. Resolves possible memory leak
when using ECC fixed point cache.
Closes #3395

- mbedtls: follow-up VERIFYHOST fix from f097669248
Fix-by: Eric Rosenquist
Fixes #3376
Closes #3390
- curlver: bump to 7.64.0 for next release

- cookies: extend domain checks to non psl builds
Ensure to perform the checks we have to enforce a sane domain in

the cookie request. The check for non-PSL enabled builds is quite
basic but it's better than nothing.
Closes #2964

- [Matus Uzak brought this change]
smb: fix incorrect path in request if connection reused
Follow-up to 09e401e01bf9. If connection gets reused, then data member

will be copied, but not the proto member. As a result, in smb_do(),
path has been set from the original proto.share data.
Closes #3388
- curl -J: do not append to the destination file
Reported-by: Kamil Dudka

Fixes #3380
Closes #3381
- mbedtls: use VERIFYHOST
Previously, VERIFYPEER would enable/disable all checks.
Reported-by: Eric Rosenquist

Fixes #3376
Closes #3380
- pingpong: change default response timeout to 120 seconds
Previously it was 30 minutes
- pingpong: ignore regular timeout in disconnect phase
The timeout set with CURLOPT_TIMEOUT is no longer used when

disconnecting from one of the pingpong protocols (FTP, IMAP, SMTP,
POP3).
Reported-by: jasal82 on github
Fixes #3264
Closes #3374
- TODO: Windows: set attribute 'archive' for completed downloads
Closes #3354
- http: minor whitespace cleanup from f464535b
- [Ayoub Boudhar brought this change]
http: Implement trailing headers for chunked transfers
This adds the CURLOPT_TRAILERDATA and CURLOPT_TRAILERFUNCTION

options that allow a callback based approach to sending trailing headers
with chunked transfers.
The test server (sws) was updated to take into account the detection of the
end of transfer in the case of trailing headers presence.
Test 1591 checks that trailing headers can be sent using libcurl.
Closes #3350
- darwinssl: accept setting max-tls with default min-tls
Reported-by: Andrei Neculau

Fixes #3367
Closes #3373
- gopher: fix memory leak from 9026083ddb2a9
- [Leonardo Taccari brought this change]
test1201: Add a trailing `?' to the selector
This verify that the `?' in the selector is kept as is.
Verifies the fix in #3370
gopher: always include the entire gopher-path in request
After the migration to URL API all octets in the selector after the
first `?' were interpreted as query and accidentally discarded and not
passed to the server.
Add a gopherpath to always concatenate possible path and query URL

pieces.
Fixes #3369
Closes #3370
urlapi: distinguish possibly empty query
If just a `?' to indicate the query is passed always store a zero length
query instead of having a NULL query.
This permits to distinguish URL with trailing `?'.
Fixes #3369
Closes #3370

- OS400: handle memory error in list conversion
Curl_slist_append_nodup() returns NULL when it fails to create a new

item for the specified list, and since the coding here reassigned the
new list on top of the old list it would result in a dangling pointer
and lost memory. Also, in case we hit an allocation failure at some
point during the conversion, with allocation succeeding again on the
subsequent call(s) we will return a truncated list around the malloc
failure point. Fix by assigning to a temporary list pointer, which can
be checked (which is the common pattern for slist appending), and free
all the resources on allocation failure.
Closes #3372
- cookies: leave secure cookies alone
Only allow secure origins to be able to write cookies with the

'secure' flag set. This reduces the risk of non-secure origins
to influence the state of secure origins. This implements IETF
Internet-Draft draft-ietf-httpbis-cookie-alone-01 which updates
RFC6265.
Closes #2956

- docs: fix the --tls-max description
Reported-by: Tobias Lindgren

Pointed out in #3367
Closes #3368

- urlapi: Fix port parsing of eol colon
A URL with a single colon without a portnumber should use the default
port, discarding the colon. Fix, add a testcase and also do little bit
of comment wordsmithing.
Closes #3365
Version 7.63.0 (12 Dec 2018)

- THANKS: from the curl 7.62.0 cycle
- test1519: use lib1518 and test CURLINFO_REDIRECT_URL more
- Curl_follow: extract the Location: header field unvalidated
... when not actually following the redirect. Otherwise we return error
for this and an application can't extract the value.
Test 1518 added to verify.
Reported-by: Pavel Pavlov

Fixes #3340
Closes #3364
- multi: convert two timeout variables to timediff_t
The time_t type is unsigned on some systems and these variables are used
to hold return values from functions that return timediff_t
already. timediff_t is always a signed type.
Closes #3363
- delta: use --diff-filter on the git diff-tree invokes
Suggested-by: Dave Reisner
Patrick Monnerat (11 Dec 2018)

- documentation: curl_formadd field and file names are now escaped
Prior to 7.56.0, fieldnames and filenames were set in Content-Disposition

header without special processing: this may lead to invalid RFC 822
quoted-strings.
7.56.0 introduces escaping of backslashes and double quotes in these names:
mention it in the documentation.
Reported-by: daboul on github

Closes #3361

- scripts/delta: show repo delta info from last release
... where "last release" should be the git tag in the repo.

- tests: add urlapi unittest
This adds a new unittest intended to cover the internal functions in

the urlapi code, starting with parse_port(). In order to avoid name
collisions in debug builds, parse_port() is renamed Curl_parse_port()
since it will be exported.

Reviewed-by: Marcel Raad <Marcel.Raad@teamviewer.com>
- urlapi: fix portnumber parsing for ipv6 zone index
An IPv6 URL which contains a zone index includes a '%%25<zode id>'

string before the ending ']' bracket. The parsing logic wasn't set
up to cope with the zone index however, resulting in a malformed url
error being returned. Fix by breaking the parsing into two stages
to correctly handle the zone index.
Closes #3355
Closes #3319
Reported-by: tonystz on Github

- [Jay Satiro brought this change]
http: fix HTTP auth to include query in URI
- Include query in the path passed to generate HTTP auth.
Recent changes to use the URL API internally (46e1640, 7.62.0)

inadvertently broke authentication URIs by omitting the query.
Closes #3356
- [Michael Kaufmann brought this change]
http: don't set CURLINFO_CONDITION_UNMET for http status code 204
The http status code 204 (No Content) should not change the "condition
unmet" flag. Only the http status code 304 (Not Modified) should do
this.
Closes #359
- [Samuel Surtees brought this change]
ldap: fix LDAP URL parsing regressions
- Match URL scheme with LDAP and LDAPS

- Retrieve attributes, scope and filter from URL query instead
Regression brought in 46e164069d1a5230 (7.62.0)
Closes #3362
- [Stefan Kanthak brought this change]
(lib)curl.rc: fixup for minor bugs
All resources defined in lib/libcurl.rc and curl.rc are language

neutral.
winbuild/MakefileBuild.vc ALWAYS defines the macro DEBUGBUILD, so the

ifdef's in line 33 of lib/libcurl.rc and src/curl.rc are wrong.
Replace the hard-coded constants in both *.rc files with #define'd
values.
Thumbs-uped-by: Rod Widdowson, Johannes Schindelin

Closes #3348
- test329: verify cookie max-age=0 immediate expiry
- cookies: expire "Max-Age=0" immediately

Fixes #3351
Closes #3352
- [Johannes Schindelin brought this change]
Upon HTTP_1_1_REQUIRED, retry the request with HTTP/1.1
This is a companion patch to cbea2fd2c (NTLM: force the connection to

HTTP/1.1, 2018-12-06): with NTLM, we can switch to HTTP/1.1
preemptively. However, with other (Negotiate) authentication it is not
clear to this developer whether there is a way to make it work with
HTTP/2, so let's try HTTP/2 first and fall back in case we encounter the
error HTTP_1_1_REQUIRED.
Note: we will still keep the NTLM workaround, as it avoids an extra

round trip.
Daniel Stenberg helped a lot with this patch, in particular by

suggesting to introduce the Curl_h2_http_1_1_error() function.
Closes #3349
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
- [Ben Greear brought this change]
openssl: fix unused variable compiler warning with old openssl
Closes #3347
NTLM: force the connection to HTTP/1.1
Since v7.62.0, cURL tries to use HTTP/2 whenever the server announces
the capability. However, NTLM authentication only works with HTTP/1.1,
and will likely remain in that boat (for details, see
https://docs.microsoft.com/en-us/iis/get-started/whats-new-in-iis-10/http2-on-
iis#when-is-http2-not-supported).
When we just found out that we want to use NTLM, and when the current
connection runs in HTTP/2 mode, let's force the connection to be closed
and to be re-opened using HTTP/1.1.
Fixes https://github.com/curl/curl/issues/3341.
Closes #3345
curl_global_sslset(): id == -1 is not necessarily an error
It is allowed to call that function with id set to -1, specifying the

backend by the name instead. We should imitate what is done further down
in that function to allow for that.
Closes #3346
Johannes Schindelin (6 Dec 2018)

- .gitattributes: make tabs in indentation a visible error

- doh: fix memory leak in OOM situation

Closes #3342
- doh: make it work for h2-disabled builds too
Reported-by: dtmsecurity at github

Fixes #3325
Closes #3336
- packages: remove old leftover files and dirs
This subdir has mostly become an attic of never-used cruft from the
past.
Closes #3331
- [Gergely Nagy brought this change]
openssl: do not use file BIOs if not requested
Moves the file handling BIO calls to the branch of the code where they
are actually used.
Closes #3339
- [Paul Howarth brought this change]
nss: Fix compatibility with nss versions 3.14 to 3.15
nss: Improve info message when falling back SSL protocol

Use descriptive text strings rather than decimal numbers.
nss: Fall back to latest supported SSL version
NSS may be built without support for the latest SSL/TLS versions,
leading to "SSL version range is not valid" errors when the library
code supports a recent version (e.g. TLS v1.3) but it has explicitly
been disabled.
This change adjusts the maximum SSL version requested by libcurl to

be the maximum supported version at runtime, as long as that version
is at least as high as the minimum version required by libcurl.
Fixes #3261

- travis: enable COPYRIGHTYEAR extended warning
The extended warning for checking incorrect COPYRIGHTYEAR is quite

expensive to run, so rather than expecting every developer to do it
we ensure it's turned on locally for Travis.
- checksrc: add COPYRIGHTYEAR check
Forgetting to bump the year in the copyright clause when hacking has
been quite common among curl developers, but a traditional checksrc
check isn't a good fit as it would penalize anyone hacking on January
1st (among other things). This adds a more selective COPYRIGHTYEAR
check which intends to only cover the currently hacked on changeset.
The check for updated copyright year is currently not enforced on all
files but only on files edited and/or committed locally. This is due to
the amount of files which aren't updated with their correct copyright
year at the time of their respective commit.
To further avoid running this expensive check for every developer, it

adds a new local override mode for checksrc where a .checksrc file can
be used to turn on extended warnings locally.
Closes #3303

- CHECKSRC.md: document more warnings
Closes #3335
[ci skip]
- SECURITY-PROCESS: bountygraph shuts down
This backpedals back the documents to the state before bountygraph.
Closes #3311
- curl: fix memory leak reading --writeout from file
If another string had been set first, the writout function for reading
the syntax from file would leak the previously allocated memory.
Reported-by: Brian Carpenter

Fixes #3322
Closes #3330
- tool_main: rename function to make it unique and better
... there's already another function in the curl tool named

free_config_fields!
Daniel Gustafsson (29 Nov 2018)

- TODO: remove CURLOPT_DNS_USE_GLOBAL_CACHE entry
Commit 7c5837e79280e6abb3ae143dfc49bca5e74cdd11 deprecated the option

making it a manual code-edit operation to turn it back on. The removal
process has thus started and is now documented in docs/DEPRECATE.md so
remove from the TODO to avoid anyone looking for something to pick up
spend cycles on an already in-progress entry.
Jay Satiro (29 Nov 2018)

- [Sevan Janiyan brought this change]
connect: fix building for recent versions of Minix
EBADIOCTL doesn't exist on more recent Minix.

There have also been substantial changes to the network stack.
Fixes build on Minix 3.4rc
- [Konstantin Kushnir brought this change]
CMake: fix MIT/Heimdal Kerberos detection
- fix syntax error in FindGSS.cmake

- correct krb5 include directory. FindGSS exports
"GSS_INCLUDE_DIR" variable.
Daniel Stenberg (28 Nov 2018)

- test328: verify Content-Encoding: none
Because of issue #3315
Closes #3317
- [James Knight brought this change]
configure: include all libraries in ssl-libs fetch
When compiling a collection of SSL libraries to link against (SSL_LIBS),

ensure all libraries are included. The call `--libs-only-l` can produce
only a subset of found in a `--libs` call (e.x. pthread may be excluded).
Adding `--libs-only-other` ensures other libraries are also included in
the list. This corrects select build environments compiling against a
static version of OpenSSL. Before the change, the following could be
observed:
checking for openssl options with pkg-config... found

configure: pkg-config: SSL_LIBS: "-lssl -lz -ldl -lcrypto -lz -ldl "
configure: pkg-config: SSL_LDFLAGS: "-
L/home/jdknight/<workdir>/staging/usr/lib
-L/home/jdknight/<workdir>/staging/usr/lib "
configure: pkg-config: SSL_CPPFLAGS: "-
I/home/jdknight/<workdir>/staging/usr/include "
checking for HMAC_Update in -lcrypto... no
checking for HMAC_Init_ex in -lcrypto... no
checking OpenSSL linking with -ldl... no
checking OpenSSL linking with -ldl and -lpthread... no
configure: WARNING: SSL disabled, you will not be able to use HTTPS, FTPS,
NTLM and more.
configure: WARNING: Use --with-ssl, --with-gnutls, --with-polarssl, --with-
cyassl, --with-nss, --with-axtls, --with-winssl, or --with-darwinssl to address
this.
...
SSL support: no (--with-
{ssl,gnutls,nss,polarssl,mbedtls,cyassl,axtls,winssl,darwinssl} )
...
And include the other libraries when compiling SSL_LIBS succeeds with:
checking for openssl options with pkg-config... found

configure: pkg-config: SSL_LIBS: "-lssl -lz -ldl -pthread -lcrypto -lz -ldl
-pthread "
configure: pkg-config: SSL_LDFLAGS: "-
L/home/jdknight/<workdir>/staging/usr/lib
-L/home/jdknight/<workdir>/staging/usr/lib "
configure: pkg-config: SSL_CPPFLAGS: "-
I/home/jdknight/<workdir>/staging/usr/include "
checking for HMAC_Update in -lcrypto... yes
checking for SSL_connect in -lssl... yes
...
SSL support: enabled (OpenSSL)
...
Signed-off-by: James Knight <james.d.knight@live.com>

Closes #3193

- doh: fix typo in infof call
- cmdline-opts/gen.pl: define the correct varname
The variable definition had a small typo making it declare another

variable then the intended.
Closes #3304
- curl_easy_perform: fix timeout handling
curl_multi_wait() was erroneously used from within

curl_easy_perform(). It could lead to it believing there was no socket
to wait for and then instead sleep for a while instead of monitoring the
socket and then miss acting on that activity as swiftly as it should
(causing an up to 1000 ms delay).
Reported-by: Antoni Villalonga

Fixes #3305
Closes #3306
Closes #3308
- CURLOPT_WRITEFUNCTION.3: spell out that it gets called many times
- cookies: create the cookiejar even if no cookies to save
Important for when the file is going to be read again and thus must not
contain old contents!
Adds test 327 to verify.
Reported-by: daboul on github

Fixes #3299
Closes #3300
- checksrc: ban snprintf use, add command line flag to override warns
- snprintf: renamed and we now only use msnprintf()
The function does not return the same value as snprintf() normally does,
so readers may be mislead into thinking the code works differently than
it actually does. A different function name makes this easier to detect.
Reported-by: Tomas Hoger

Assisted-by: Daniel Gustafsson
Fixes #3296
Closes #3297
- [Tobias Hintze brought this change]
test: update test20/1322 for eglibc bug workaround
The tests 20 and 1322 are using getaddrinfo of libc for resolving. In
eglibc-2.19 there is a memory leakage and invalid free bug which
surfaces in some special circumstances (PF_UNSPEC hint with invalid or
non-existent names). The valgrind runs in testing fail in these
situations.
As the tests 20/1322 are not specific on either protocol (IPv4/IPv6)

this commit changes the hints to IPv4 protocol by passing `--ipv4` flag
on the tests' command line. This prevents the valgrind failures.
- [Tobias Hintze brought this change]
host names: allow trailing dot in name resolve, then strip it

Delays stripping of trailing dots to after resolving the hostname.
Fixes #3022
Closes #3222
- [UnknownShadow200 brought this change]
CURLOPT_HEADERFUNCTION.3: match 'nitems' name in synopsis and description
Closes #3295

- configure: Fix typo in comment
Michael Kaufmann (21 Nov 2018)

- openssl: support session resume with TLS 1.3
Session resumption information is not available immediately after a TLS 1.3

handshake. The client must wait until the server has sent a session ticket.
Use OpenSSL's "new session" callback to get the session information and put it
into curl's session cache. For TLS 1.3 sessions, this callback will be invoked
after the server has sent a session ticket.
The "new session" callback is invoked only if OpenSSL's session cache is

enabled, so enable it and use the "external storage" mode which lets curl manage
the contents of the session cache.
A pointer to the connection data and the sockindex are now saved as "SSL extra
data" to make them available to the callback.
This approach also works for old SSL/TLS versions and old OpenSSL versions.
Fixes #3202
Closes #3271
- ssl: fix compilation with OpenSSL 0.9.7
- ENGINE_cleanup() was used without including "openssl/engine.h"

- enable engine support for OpenSSL 0.9.7
Closes #3266

- openssl: disable TLS renegotiation with BoringSSL
Since we're close to feature freeze, this change disables this feature
with an #ifdef. Define ALLOW_RENEG at build-time to enable.
This could be converted to a bit for CURLOPT_SSL_OPTIONS to let

applications opt-in this.
Concern-raised-by: David Benjamin

Fixes #3283
Closes #3293
- [Romain Fliedel brought this change]
ares: remove fd from multi fd set when ares is about to close the fd
When using c-ares for asyn dns, the dns socket fd was silently closed
by c-ares without curl being aware. curl would then 'realize' the fd
has been removed at next call of Curl_resolver_getsock, and only then
notify the CURLMOPT_SOCKETFUNCTION to remove fd from its poll set with
CURL_POLL_REMOVE. At this point the fd is already closed.
By using ares socket state callback (ARES_OPT_SOCK_STATE_CB), this

patch allows curl to be notified that the fd is not longer needed
for neither for write nor read. At this point by calling
Curl_multi_closed we are able to notify multi with CURL_POLL_REMOVE
before the fd is actually closed by ares.
In asyn-ares.c Curl_resolver_duphandle we can't use ares_dup anymore

since it does not allow passing a different sock_state_cb_data
Closes #3238
- [Romain Fliedel brought this change]
examples/ephiperfifo: report error when epoll_ctl fails

- [pkubaj brought this change]
ntlm: Remove redundant ifdef USE_OPENSSL
lib/curl_ntlm.c had code that read as follows:
#ifdef USE_OPENSSL
# ifdef USE_OPENSSL
# else
# ..
# endif
#endif
Remove the redundant USE_OPENSSL along with #else (it's not possible to
reach it anyway). The removed construction is a leftover from when the
SSLeay support was removed.
Closes #3269

- [Han Han brought this change]
ssl: replace all internal uses of CURLE_SSL_CACERT
Closes #3291
Han Han (19 Nov 2018)

- docs: add more description to unified ssl error codes
- curle: move deprecated error code to ifndef block

Patrick Monnerat (19 Nov 2018)
- os400: add CURLOPT_CURLU to ILE/RPG binding.
- os400: Add curl_easy_conn_upkeep() to ILE/RPG binding.
- os400: fix return type of curl_easy_pause() in ILE/RPG binding.

- impacket: add LICENSE
The license for the impacket package was not in our tree.
Imported now from upstream's

https://github.com/SecureAuthCorp/impacket/blob/master/LICENSE
Reported-by: infinnovation-dev on github

Fixes #3276
Closes #3277

- tool_doswin: Fix uninitialized field warning
The partial struct initialization in 397664a065abffb7c3445ca9 caused

a warning on uninitialized MODULEENTRY32 struct members:
/src/tool_doswin.c:681:3: warning: missing initializer for field

'th32ModuleID' of 'MODULEENTRY32 {aka struct tagMODULEENTRY32}'
[-Wmissing-field-initializers]
This is sort of a bogus warning as the remaining members will be set

to zero by the compiler, as all omitted members are. Nevertheless,
remove the warning by omitting all members and setting the dwSize
members explicitly.
Closes #3254
- openssl: Remove SSLEAY leftovers
Commit 709cf76f6bb7dbac deprecated USE_SSLEAY, as curl since long isn't

compatible with the SSLeay library. This removes the few leftovers that
were omitted in the less frequently used platform targets.
Closes #3270

- [Elia Tufarolo brought this change]
http_negotiate: do not close connection until negotiation is completed
Fix HTTP POST using CURLAUTH_NEGOTIATE.
Closes #3275
- pop3: only do APOP with a valid timestamp

Brought-by: bobmitchell1956 on github
Fixes #3278
Closes #3279

- [Peter Wu brought this change]
openssl: do not log excess "TLS app data" lines for TLS 1.3
The SSL_CTX_set_msg_callback callback is not just called for the

Handshake or Alert protocols, but also for the raw record header
(SSL3_RT_HEADER) and the decrypted inner record type
(SSL3_RT_INNER_CONTENT_TYPE). Be sure to ignore the latter to avoid
excess debug spam when using `curl -v` against a TLSv1.3-enabled server:
* TLSv1.3 (IN), TLS app data, [no content] (0):
(Following this message, another callback for the decrypted

handshake/alert messages will be be present anyway.)
Marc Hoersken (15 Nov 2018)

- tests: disable SO_EXCLUSIVEADDRUSE for stunnel on Windows
SO_EXCLUSIVEADDRUSE is on by default on Vista or newer,

but does not work together with SO_REUSEADDR being on.
The default changes were made with stunnel 5.34 and 5.35.

- [Kamil Dudka brought this change]
nss: remove version selecting dead code
Closes #3262
- nss: set default max-tls to 1.3/1.2
Fixes #3261

- tool_cb_wrt: Silence function cast compiler warning
Commit 5bfaa86ceb3c2a9ac474a928e748c4a86a703b33 introduced a new

compiler warning on Windows cross compilation with GCC. See below
for an example of the warning from the autobuild logs (whitespace
edited to fit):
/src/tool_cb_wrt.c:175:9: warning: cast from function call of type

'intptr_t {aka long long int}' to non-matching type 'void *'
[-Wbad-function-cast]
(HANDLE) _get_osfhandle(fileno(outs->stream)),
^
Store the return value from _get_osfhandle() in an intermediate

variable and cast the variable in WriteConsoleW() rather than the
function call directly to avoid a compiler warning.
In passing, also add inspection of the MultiByteToWideChar() return
value and return failure in case an error is reported.
Closes #3263

- nss: fix fallthrough comment to fix picky compiler warning
- docs: expanded on some CURLU details
- [Tim Rühsen brought this change]
ftp: avoid two unsigned int overflows in FTP listing parser
Curl_ftp_parselist: avoid unsigned integer overflows
The overflow has no real world impact, just avoid it for "best
practice".
Closes #3225
- curl: --local-port range was not "including"
The end port number in a given range was not included in the range used,
as it is documented to be.

Fixes #3251
Closes #3255
- [Jérémy Rocher brought this change]
openssl: support BoringSSL TLS renegotiation
As per BoringSSL porting documentation [1], BoringSSL rejects peer

renegotiations by default.
curl fails when trying to authenticate to server through client

certificate if it is requested by server after the initial TLS
handshake.
Enable renegotiation by default with BoringSSL to get same behavior as

with OpenSSL. This is done by calling SSL_set_renegotiate_mode [2]
which was introduced in commit 1d5ef3bb1eb9 [3].
1 - https://boringssl.googlesource.com/boringssl/+/HEAD/PORTING.md#tls-
renegotiation
2 - https://boringssl.googlesource.com/boringssl/
+/master/include/openssl/ssl.h#3482
3 - https://boringssl.googlesource.com/boringssl/
+/1d5ef3bb1eb97848617db5e7d633d735a401df86
Signed-off-by: Jérémy Rocher <rocher.jeremy@gmail.com>

Fixes #3258
Closes #3259
- HISTORY: add some milestones
Added a few of the more notable milestones in curl history that were
missing. Primarily more recent ones but I also noted some older that
could be worth mentioning.
[ci skip]
Closes #3257

- KNOWN_BUGS: add --proxy-any connection issue
Add the identified issue with --proxy-any and proxy servers which
advertise authentication schemes other than the supported one.
Closes #876
Closes #3250
Reported-by: NTMan on Github

- [Jim Fuller brought this change]
setopt: add CURLOPT_CURLU
Allows an application to pass in a pre-parsed URL via a URL handle.
Closes #3227
docs: ESCape "\n" codes
Groff / Troff will display a:

printaf("Errno: %ld\n", error);
as:
printf("Errno: %ld0, error);
when a "\n" is not escaped. Use "\\n" instead.
Closes #3246
- curl: --local-port fix followup
Regression by 52db54869e6.

Fixes #3248
Closes #3249
GitHub (7 Nov 2018)

More "\n" ESCaping

- curl: fix --local-port integer overflow

The tool's local port command line range parser didn't check for integer
overflows and could pass "weird" data to libcurl for this option.
libcurl however, has a strict range check for the values so it rejects
anything outside of the accepted range.

Closes #3242
- curl: correct the switch() logic in ourWriteOut
Follow-up to e431daf013, as I did the wrong correction for a compiler

warning. It should be a break and not a fall-through.
Pointed-out-by: Frank Gevaerts
curl: add %{stderr} and %{stdout} for --write-out
Closes #3115

- winssl: be consistent in Schannel capitalization
The productname from Microsoft is "Schannel", but in infof/failf

reporting we use "schannel". This removes different versions.
Closes #3243

- TODO: Have the URL API offer IDN decoding
Similar to how URL decoding/encoding is done, we could have URL

functions to convert IDN host names to punycode.
Suggested-by: Alexey Melnichuk

Closes #3232
- urlapi: only skip encoding the first '=' with APPENDQUERY set
APPENDQUERY + URLENCODE would skip all equals signs but now it only skip
encoding the first to better allow "name=content" for any content.
Reported-by: Alexey Melnichuk

Fixes #3231
Closes #3231
- url: a short host name + port is not a scheme
The function identifying a leading "scheme" part of the URL considered a

few letters ending with a colon to be a scheme, making something like
"short:80" to become an unknown scheme instead of a short host name and
a port number.
Extended test 1560 to verify.
Also fixed test203 to use file_pwd to make it get the correct path on
windows. Removed test 2070 since it was a duplicate of 203.
Assisted-by: Marcel Raad

Reported-by: Hagai Auro
Fixes #3220
Fixes #3233
Closes #3223
Closes #3235
- [Sangamkar brought this change]
libcurl: stop reading from paused transfers
In the transfer loop it would previously not acknwledge the pause bit
and continue until drained or loop ended.
Closes #3240

- tool: add undocumented option --dump-module-paths for win32
- Add an undocumented diagnostic option for Windows to show the full

paths of all loaded modules regardless of whether or not libcurl
initialization succeeds.
This is needed so that in the CI we can get a list of all DLL

dependencies after initialization (when they're most likely to have
finished loading) and then package them as artifacts so that a
functioning build can be downloaded. Also I imagine it may have some use
as a diagnostic for help requests.
Ref: https://github.com/curl/curl/pull/3103
- curl_multibyte: fix a malloc overcalculation
Prior to this change twice as many bytes as necessary were malloc'd when
converting wchar to UTF8. To allay confusion in the future I also
changed the variable name for the amount of bytes from len to bytes.

- netrc: don't ignore the login name specified with "--user"
- for "--netrc", don't ignore the login/password specified with "--user",

only ignore the login/password in the URL.
This restores the netrc behaviour of curl 7.61.1 and earlier.
- fix the documentation of CURL_NETRC_REQUIRED
- improve the detection of login/password changes when reading .netrc
- don't read .netrc if both login and password are already set
Fixes #3213
Closes #3224
Patrick Monnerat (5 Nov 2018)

- OS400: add URL API ccsid wrappers and sync ILE/RPG bindings
- [Yasuhiro Matsumoto brought this change]
curl: fixed UTF-8 in current console code page (Windows)
Fixes #3211
Fixes #3175
Closes #3212
- TODO: 2.6 multi upkeep
Closes #3199

- unittest: make 1652 stable across collations
The previous coding used a format string whose output depended on the
current locale of the environment running the test. Since the gist of
the test is to have a format string, with the actual formatting being
less important, switch to a more stable formatstring with decimals.

Closes #3234

- Revert "url: a short host name + port is not a scheme"
This reverts commit 226cfa8264cd979eff3fd52c0f3585ef095e7cf2.
This commit caused test failures on appveyor/windows. Work on fixing them is

in #3235.
- symbols-in-versions: add missing CURLU_ symbols
...and fix symbol-scan.pl to also scan urlapi.h
Reported-by: Alexey Melnichuk

Fixes #3226
Closes #3230

- infof: clearly indicate truncation
The internal buffer in infof() is limited to 2048 bytes of payload plus

an additional byte for NULL termination. Servers with very long error
messages can however cause truncation of the string, which currently
isn't very clear, and leads to badly formatted output.
This appends a "...\n" (or just "..." in case the format didn't with a
newline char) marker to the end of the string to clearly show
that it has been truncated.
Also include a unittest covering infof() to try and catch any bugs
introduced in this quite important function.
Closes #3216

- tool_getparam: fix some comments

- url: a short host name + port is not a scheme
The function identifying a leading "scheme" part of the URL considered a few
letters ending with a colon to be a scheme, making something like "short:80"
to become an unknown scheme instead of a short host name and a port number.
Extended test 1560 to verify.
Reported-by: Hagai Auro

Fixes #3220
Closes #3223
- URL: fix IPv6 numeral address parser
Regression from 46e164069d1a52. Extended test 1560 to verify.
Reported-by: tpaukrt on github

Fixes #3218
Closes #3219
- travis: remove curl before a normal build
on Linux. To make sure the test suite runs with its newly build tool and
doesn't require an external one present.
Bug: #3198
Closes #3200
- [Tim Rühsen brought this change]
mprintf: avoid unsigned integer overflow warning
The overflow has no real world impact.

Just avoid it for "best practice".
Code change suggested by "The Infinnovation Team" and Daniel Stenberg.

Closes #3184
- Curl_follow: accept non-supported schemes for "fake" redirects
When not actually following the redirect and the target URL is only
stored for later retrieval, curl always accepted "non-supported"
schemes. This was a regression from 46e164069d1a5230.
Reported-by: Brad King

Fixes #3210
Closes #3215

- openvms: fix example name
Commit efc696a2e09225bfeab4 renamed persistant.c to persistent.c to

fix the typo in the name, but missed to update the OpenVMS package
files which still looked for the old name.
Closes #3217

- configure: show CFLAGS, LDFLAGS etc in summary
To make it easier to understand other people's and remote builds etc.
Closes #3207
- version: bump for next cycle
- axtls: removed
As has been outlined in the DEPRECATE.md document, the axTLS code has
been disabled for 6 months and is hereby removed.
Use a better supported TLS library!

Closes #3194
- [marcosdiazr brought this change]
schannel: make CURLOPT_CERTINFO support using Issuer chain
Closes #3197
- travis: build with sanitize=address,undefined,signed-integer-overflow
... using clang
Closes #3190
- schannel: use Curl_ prefix for global private symbols
Curl_verify_certificate() must use the Curl_ prefix since it is globally

available in the lib and otherwise steps outside of our namespace!
Closes #3201
Kamil Dudka (1 Nov 2018)

- tests: drop http_pipe.py script no longer used
It is unused since commit f7208df7d9d5cd5e15e2d89237e828f32b63f135.
Closes #3204
Daniel Stenberg (31 Oct 2018)

- runtests: use the local curl for verifying
... revert the mistaken change brought in commit 8440616f53.
Reported-by: Alessandro Ghedini

Closes #3198
Version 7.62.0 (30 Oct 2018)

Daniel Gustafsson (30 Oct 2018)

- vtls: add MesaLink to curl_sslbackend enum
MesaLink support was added in commit 57348eb97d1b8fc3742e02c but the

backend was never added to the curl_sslbackend enum in curl/curl.h.
This adds the new backend to the enum and updates the relevant docs.
Closes #3195

cmake: Remove unused CURL_CONFIG_HAS_BEEN_RUN_BEFORE variable
Closes #3191
- test2080: verify the fix for CVE-2018-16842
- voutf: fix bad arethmetic when outputting warnings to stderr
CVE-2018-16842
- [Tuomo Rinne brought this change]
cmake: uniform ZLIB to use USE_ variable and clean curl-config.cmake.in
Closes #3123
cmake: add find_dependency call for ZLIB to CMake config file
cmake: add support for transitive ZLIB target
- unit1650: fix "null pointer passed as argument 1 to memcmp"
Detected by UndefinedBehaviorSanitizer
Closes #3187
- travis: add a "make tidy" build that runs clang-tidy
Closes #3182
- unit1300: fix stack-use-after-scope AddressSanitizer warning

Closes #3186
- Curl_auth_create_plain_message: fix too-large-input-check
CVE-2018-16839
Reported-by: Harry Sintonen
- Curl_close: clear data->multi_easy on free to avoid use-after-free
Regression from b46cfbc068 (7.59.0)

CVE-2018-16840
Reported-by: Brian Carpenter (Geeknik Labs)
- [randomswdev brought this change]
system.h: use proper setting with Sun C++ as well
system.h selects the proper Sun settings when __SUNPRO_C is defined. The
Sun compiler does not define it when compiling C++ files. I'm adding a
check also on __SUNPRO_CC to allow curl to work properly also when used
in a C++ project on Sun Solaris.
Closes #3181
- rand: add comment to skip a clang-tidy false positive
- test1651: unit test Curl_extract_certinfo()
The version used for Gskit, NSS, GnuTLS, WolfSSL and schannel.
- x509asn1: always check return code from getASN1Element()
- Makefile: add 'tidy' target that runs clang-tidy
Available in the root, src and lib dirs.
Closes #3163
- RELEASE-PROCEDURE: adjust the release dates
See: https://curl.haxx.se/mail/lib-2018-10/0107.html
Patrick Monnerat (27 Oct 2018)

- x509asn1: suppress left shift on signed value
Use an unsigned variable: as the signed operation behavior is undefined,

this change silents clang-tidy about it.
Reported-By: Daniel Stenberg
Michael Kaufmann (27 Oct 2018)

- multi: Fix error handling in the SENDPROTOCONNECT state
If Curl_protocol_connect() returns an error code,

handle the error instead of switching to the next state.
Closes #3170

- openssl: output the correct cipher list on TLS 1.3 error
When failing to set the 1.3 cipher suite, the wrong string pointer would
be used in the error message. Most often saying "(nil)".
Reported-by: Ricky-Tigg on github

Fixes #3178
Closes #3180
- docs/CIPHERS: fix the TLS 1.3 cipher names
... picked straight from the OpenSSL man page:

https://www.openssl.org/docs/manmaster/man3/SSL_CTX_set_ciphersuites.html
Reported-by: Ricky-Tigg on github

Bug: #3178
Marcel Raad (27 Oct 2018)

- travis: install gnutls-bin package
This is required for gnutls-serv, which enables a few more tests.

- ssh: free the session on init failures
Ensure to clear the session object in case the libssh2 initialization

fails.
It could be argued that the libssh2 error function should be called to

get a proper error message in this case. But since the only error path
in libssh2_knownhost_init() is memory a allocation failure it's safest
to avoid since the libssh2 error handling allocates memory.
Closes #3179

- docs/RELEASE-PROCEDURE: remove old entries, modify the Dec 2018 date
... I'm moving it up one week due to travels. The rest stays.
openssl: make 'done' a proper boolean
Closes #3176
- gtls: Values stored to but never read
Detected by clang-tidy
Closes #3176
- [Alexey Eremikhin brought this change]
curl.1: --ipv6 mutexes ipv4 (fixed typo)
Fixes #3171
Closes #3172
- tool_main: make TerminalSettings static
Reported-by: Gisle Vanem

Bug:
https://github.com/curl/curl/commit/becfe1233ff2b6b0c3e1b6a10048b55b68c2539f#commit
comment-31008819
Closes #3161
- curl-config.in: remove dependency on bc
Reported-by: Dima Pasechnik

Fixes #3143
Closes #3174
rtmp: fix for compiling with lwIP
Compiling on _WIN32 and with USE_LWIPSOCK, causes this error:

curl_rtmp.c(223,3): error: use of undeclared identifier 'setsockopt'
setsockopt(r->m_sb.sb_socket, SOL_SOCKET, SO_RCVTIMEO,
^
curl_rtmp.c(41,32): note: expanded from macro 'setsockopt'
#define setsockopt(a,b,c,d,e) (setsockopt)(a,b,c,(const char *)d,(int)e)
^
Closes #3155
- configure: remove CURL_CONFIGURE_CURL_SOCKLEN_T
Follow-up to #3166 which did the cmake part of this. This type/define is
not used.
Closes #3168
cmake: remove unused variables
Remove variables:
* HAVE_SOCKLEN_T
* CURL_SIZEOF_CURL_SOCKLEN_T
* CURL_TYPEOF_CURL_SOCKLEN_T
Closes #3166

- urldata: Fix comment in header
The "connecting" function is used by multiple protocols, not only FTP

- netrc: free temporary strings if memory allocation fails
- Change the inout parameters after all needed memory has been
allocated. Do not change them if something goes wrong.
- Free the allocated temporary strings if strdup() fails.
Closes #3122

config: Remove unused SIZEOF_VOIDP
Closes #3162
GitHub (23 Oct 2018)

Fix for compiling with lwIP (3)
lwIP on Windows does not have a WSAIoctl() function.

But it do have a SO_SNDBUF option to lwip_setsockopt(). But it currently does
nothing.

- Curl_follow: return better errors on URL problems
... by making the converter function global and accessible.
Closes #3153
- Curl_follow: remove remaining free(newurl)
Follow-up to 05564e750e8f0c. This function no longer frees the passed-in

URL.
Reported-by: Michael Kaufmann

Bug:
https://github.com/curl/curl/commit/05564e750e8f0c79016c680f301ce251e6e86155#commit
comm
ent-30985666

- headers: end all headers with guard comment
Most headerfiles end with a /* <headerguard> */ comment, but it was

missing from some. The comment isn't the most important part of our
code documentation but consistency has an intrinsic value in itself.
This adds header guard comments to the files that were lacking it.
Closes #3158
Jay Satiro (23 Oct 2018)

- CIPHERS.md: Mention the options used to set TLS 1.3 ciphers

- docs/BUG-BOUNTY: the sponsors actually decide the amount
Retract the previous approach as the sponsors will be the ones to set the
final amounts.
Closes #3152
[ci skip]
- multi: avoid double-free
Curl_follow() no longer frees the string. Make sure it happens in the

caller function, like we normally handle allocations.
This bug was introduced with the use of the URL API internally, it has
never been in a release version
Reported-by: Dario Weißer

Closes #3149
- multi: make the closure handle "inherit" CURLOPT_NOSIGNAL
Otherwise, closing that handle can still cause surprises!
Reported-by: Martin Ankerl

Fixes #3138
Closes #3147

- VS projects: add USE_IPV6
The Visual Studio builds didn't use IPv6. Add it to all projects since
Visual Studio 2008, which is verified to build via AppVeyor.
- config_win32: enable LDAPS
As done in the autotools and CMake builds by default.

- travis: add build for "configure --disable-verbose"
Closes #3144
Kamil Dudka (17 Oct 2018)

- tool_cb_hdr: handle failure of rename()
Detected by Coverity.
Closes #3140
Reviewed-by: Jay Satiro

- docs/SECURITY-PROCESS: the hackerone IBB program drops curl
... now there's only BountyGraph.

- [Matthew Whitehead brought this change]
x509asn1: Fix SAN IP address verification
For IP addresses in the subject alternative name field, the length

of the IP address (and hence the number of bytes to perform a
memcmp on) is incorrectly calculated to be zero. The code previously
subtracted q from name.end. where in a successful case q = name.end
and therefore addrlen equalled 0. The change modifies the code to
subtract name.beg from name.end to calculate the length correctly.
The issue only affects libcurl with GSKit SSL, not other SSL backends.
The issue is not a security issue as IP verification would always fail.
Fixes #3102
Closes #3141

- INSTALL: mention mesalink in TLS section
Commit 57348eb97d1b8fc3742e02c6587d2d02ff592da5 added support for the

MesaLink vtls backend, but missed updating the TLS section containing
supported backends in the docs.
Closes #3134

- nonblock: fix unused parameter warning
If USE_BLOCKING_SOCKETS is defined, curlx_nonblock's arguments are not

used.

- Curl_follow: Always free the passed new URL
Closes #3124
Viktor Szakats (12 Oct 2018)

- replace rawgit links [ci skip]
Ref: https://rawgit.com/ "RawGit has reached the end of its useful life"
Ref: https://news.ycombinator.com/item?id=18202481

- docs/BUG-BOUNTY.md: for vulns published since Aug 1st 2018
[ci skip]
- travis: make distcheck scan for BOM markers

and remove BOM from projects/wolfssl_override.props
Closes #3126

- CMake: remove BOM
Accidentally aded in commit 1bb86057ff07083deeb0b00f8ad35879ec4d03ea.
Reported-by: Viktor Szakats

Ref: https://github.com/curl/curl/pull/3120#issuecomment-428673136

- transfer: fix typo in comment

- docs: add "see also" links for SSL options
- link TLS 1.2 and TLS 1.3 options

- link proxy and non-proxy options
Closes #3121

- AppVeyor: remove BDIR variable that sneaked in again
Removed in ae762e1abebe3a5fe75658583c85059a0957ef6e, accidentally added

again in 9f3be5672dc4dda30ab43e0152e13d714a84d762.
- CMake: disable -Wpedantic-ms-format
As done in the autotools build. This is required for MinGW, which

supports only %I64 for printing 64-bit values, but warns about it.

- ldap: show precise LDAP call in error message on Windows
Also add a unique but common text ('bind via') to make it

easy to grep this specific failure regardless of platform.
Ref: https://github.com/curl/curl/pull/878/files#diff-
7a636f08047c4edb53a240f540b4ecf6R468

- docs/DEPRECATE: minor reformat to render nicer on web

- CURLOPT_SSL_VERIFYSTATUS: Fix typo
Changes s/OSCP/OCSP/ and bumps the copyright year due to the change.

- curl_setup: define NOGDI on Windows
This avoids an ERROR macro clash between <wingdi.h> and <arpa/tftp.h>
on MinGW.
- Windows: fixes for MinGW targeting Windows Vista
Classic MinGW has neither InitializeCriticalSectionEx nor

GetTickCount64, independent of the target Windows version.

- TODO: fixed 'API for URL parsing/splitting'

- KNOWN_BUGS: Fix various typos
Closes #3112

- spelling fixes [ci skip]
as detected by codespell 1.14.0

- curl_ntlm_wb: check aprintf() return codes
... when they return NULL we're out of memory and MUST return failure.
closes #3111
- docs/BUG-BOUNTY: proposed additional docs
Bug bounty explainer. See https://bountygraph.com/programs/curl
Closes #3067
- [Rick Deist brought this change]
hostip: fix check on Curl_shuffle_addr return value
Closes #3110
- FILE: fix CURLOPT_NOBODY and CURLOPT_HEADER output
Now FILE transfers send headers to the header callback like HTTP and
other protocols. Also made curl_easy_getinfo(...CURLINFO_PROTOCOL...)
work for FILE in the callbacks.
Makes "curl -i file://.." and "curl -I file://.." work like before

again. Applied the bold header logic to them too.
Regression from c1c2762 (7.61.0)
Reported-by: Shaun Jackman

Fixes #3083
Closes #3101

- gskit: make sure to terminate version string
In case a very small buffer was passed to the version function, it could
result in the buffer not being NULL-terminated since strncpy() doesn't
guarantee a terminator on an overflowed buffer. Rather than adding code
to terminate (and handle zero-sized buffers), move to using snprintf()
instead like all the other vtls backends.
Closes #3105
- TODO: add LD_PRELOAD support on macOS
Add DYLD_INSERT_LIBRARIES support to the TODO list. Reported in #2394.
- runtests: skip ld_preload tests on macOS
The LD_PRELOAD functionality doesn't exist on macOS, so skip any tests

requiring it.
Fixes #2394
Closes #3106
Reported-by: Github user @jakirkham

- AppVeyor: use Debug builds to run tests
This enables more tests.
- AppVeyor: add HTTP_ONLY build
- AppVeyor: add WinSSL builds
Use the oldest and latest Windows SDKs for them.

Also, remove all but one OpenSSL build.
- AppVeyor: add remaining Visual Studio versions
This adds Visual Studio 9 and 10 builds.

There's no 64-bit VC9 compiler on AppVeyor, so use it as the Win32
build. Also, VC9 cannot be used for running the test suite.
- AppVeyor: break long line
- AppVeyor: remove unused BDIR variable

- test2100: test DoH using IPv4-only
To make it only send one DoH request and avoid the race condition that
could lead to the requests getting sent in reversed order and thus
making it hard to compare in the test case.
Fixes #3107
Closes #3108
- tests/FILEFORMAT: mention how to use <fileN> and <stripfileN> too
[ci skip]
- [Dmitry Kostjuchenko brought this change]
timeval: fix use of weak symbol clock_gettime() on Apple platforms
Closes #3048
- doh: keep the IPv4 address in (original) network byte order
Ideally this will fix the reversed order shown in SPARC tests:
resp 8: Expected 127.0.0.1 got 1.0.0.127
Closes #3091

- INTERNALS.md: wrap lines longer than 79

- INTERNALS: escape reference to parameter
The parameter reference <string> was causing rendering issues in the

generated HTML page, as <string> isn't a valid HTML tag. Fix by back-
tick escaping it.
Closes #3099
- checksrc: handle zero scoped ignore commands
If a !checksrc! disable command specified to ignore zero errors, it was

still added to the ignore block even though nothing was ignored. While
there were no blocks ignored that shouldn't be ignored, the processing
ended with with a warning:
<filename>:<line>:<col>: warning: Unused ignore: LONGLINE (UNUSEDIGNORE)
/* !checksrc! disable LONGLINE 0 */
^
Fix by instead treating a zero ignore as a a badcommand and throw a
warning for that one.
Closes #3096
- checksrc: enable strict mode and warnings
Enable strict and warnings mode for checksrc to ensure we aren't missing
anything due to bugs in the checking code. This uncovered a few things
which are all fixed in this commit:
* several variables were used uninitialized

* several variables were not defined in the correct scope
* the whitelist filehandle was read even if the file didn't exist
* the enable_warn() call when a disable counter had expired was passing
incorrect variables, but since the checkwarn() call is unlikely to hit
(the counter is only decremented to zero on actual ignores) it didn't
manifest a problem.
Closes #3090

- CMake: suppress MSVC warning C4127 for libtest
It's issued by older Windows SDKs (prior to version 8.0).
Sergei Nikulov (5 Oct 2018)

- Merge branch 'dmitrykos-fix_missing_CMake_defines'
- [Dmitry Kostjuchenko brought this change]
cmake: test and set missed defines during configuration
Added configuration checks for HAVE_BUILTIN_AVAILABLE and

HAVE_CLOCK_GETTIME_MONOTONIC.
Closes #3097

- AppVeyor: disable test 500
It almost always results in

"starttransfer vs total: 0.000001 0.000000".
I cannot reproduce this locally, so disable it for now.
- AppVeyor: set custom install prefix
CMake's default has spaces and in 32-bit mode parentheses, which result
in syntax errors in curl-config.
- AppVeyor: Remove non-SSL non-test builds
They don't add much value.
- AppVeyor: run test suite
Use the preinstalled MSYS2 bash for that.

Disable test 1139 as the CMake build doesn't generate curl.1.
Ref: https://github.com/curl/curl/issues/3070#issuecomment-425922224
- AppVeyor: use in-tree build
Required to run the tests.

- doh: make sure TTL isn't re-inited by second (discarded?) response
Closes #3092
- test320: strip out more HTML when comparing
To make the test case work with different gnutls-serv versions better.

Fixes #3093
Closes #3094

- runtests: use Windows paths for Windows curl
curl generated by CMake's Visual Studio generator has "Windows" in the

version number.

- [Colin Hogben brought this change]
tests/negtelnetserver.py: fix Python2-ism in neg TELNET server
Fix problems caused by differences in treatment of bytes objects between

python2 and python3.
Fixes #2929
Closes #3080

- memory: ensure to check allocation results
The result of a memory allocation should always be checked, as we may

run under memory pressure where even a small allocation can fail. This
adds checking and error handling to a few cases where the allocation
wasn't checked for success. In the ftp case, the freeing of the path
variable is moved ahead of the allocation since there is little point
in keeping it around across the strdup, and the separation makes for
more readable code. In nwlib, the lock is aslo freed in the error path.
Also bumps the copyright years on affected files.
Closes #3084
- comment: Fix multiple typos in function parameters
Ensure that the parameters in the comment match the actual names in the
prototype.
Closes #3079
- CURLOPT_SSLVERSION.3: fix typos and consistent spelling
Use TLS vX.Y throughout the document, instead of TLS X.Y, as that was
already done in all but a few cases. Also fix a few typos.
Closes #3076
- SECURITY-PROCESS: make links into hyperlinks
Use proper Markdown hyperlink format for the Bountygraph links in order
for the generated website page to be more user friendly. Also link to
the sponsors to give them a little extra credit.
Closes #3082

- CURLOPT_HEADER.3: fix typo
- nss: fix nssckbi module loading on Windows
- Use .DLL extension instead of .so to load modules on Windows.
Reported-by: Maxime Legros
Ref: https://github.com/curl/curl/pull/3016/#issuecomment-423069442
- data-binary.d: clarify default content-type is x-www-form-urlencoded
- Advise user that --data-binary sends a default content type of

x-www-form-urlencoded, and to have the data treated as arbitrary
binary data by the server set the content-type header to octet-stream.
- test1299: use single quotes around asterisk
Ref: https://github.com/curl/curl/issues/1751#issuecomment-321522580

- docs/CIPHERS: mention the colon separation for OpenSSL
Bug: #3077
- runtests: ignore disabled even when ranges are given
runtests.pl support running a range of tests, like "44 to 127". Starting

now, the code makes sure that even such given ranges will ignore tests
that are marked as disabled.
Disabled tests can still be run by explictly specifying that test

number.
Closes #3075
- urlapi: starting with a drive letter on win32 is not an abs url
... and libcurl doesn't support any single-letter URL schemes (if there
even exist any) so it should be fairly risk-free.
Fixes #3070
Closes #3071

- doh: fix curl_easy_setopt argument type
CURLOPT_POSTFIELDSIZE is long. Fixes a compiler warning on 64-bit

MinGW.


CMake: Improve config installation
Use 'GNUInstallDirs' standard module to set destinations of installed

files.
Use uppercase "CURL" names instead of lowercase "curl" to match standard

'FindCURL.cmake' CMake module:
* https://cmake.org/cmake/help/latest/module/FindCURL.html
Meaning:
* Install 'CURLConfig.cmake' instead of 'curl-config.cmake'
* User should call 'find_package(CURL)' instead of 'find_package(curl)'
Use 'configure_package_config_file' function to generate

'CURLConfig.cmake' file. This will make 'curl-config.cmake.in' template
file smaller and handle components better. E.g. current configuration
report no error if user specified unknown components (note: new
configuration expects no components, report error if user will try to
specify any).

- test1650: make it depend on http/2
Follow-up to 570008c99da0ccbb as it gets link errors.

Closes #3068
- [Nate Prewitt brought this change]
MANUAL: minor grammar fix
Noticed a typo reading through the docs.
Closes #3069
- doh: only build if h2 enabled
The DoH spec says "HTTP/2 [RFC7540] is the minimum RECOMMENDED version
of HTTP for use with DoH".

Closes #3066
- test2100: require http2 to run

Fixes #3064
Closes #3065
- multi: fix memory leak in content encoding related error path
... a missing multi_done() call.
Credit to OSS-Fuzz
Closes #3063
- travis: bump the Secure Transport build to use xcode 10
Due to an issue with travis

(https://github.com/travis-ci/travis-ci/issues/9956) we've been using
Xcode 9.2 for darwinssl builds for a while. Now xcode 10 is offered as
an alternative and as it builds curl+darwinssl fine that seems like a
better choice.
Closes #3062
- [Rich Turner brought this change]
curl: enabled Windows VT Support and UTF-8 output
Enabled Console VT support (if running OS supports VT) in tool_main.c.

Fixes #3008
Closes #3011
- multi: fix location URL memleak in error path
Follow-up to #3044 - fix a leak OSS-Fuzz detected

Closes #3057
Sergei Nikulov (28 Sep 2018)

- cmake: fixed path used in generation of docs/tests during curl build through
add_subdicectory(...)
- [Brad King brought this change]
cmake: Backport to work with CMake 3.0 again
Changes in commit 7867aaa9a0 (cmake: link curl to the OpenSSL targets

instead of lib absolute paths, 2018-07-17) and commit f826b4ce98 (cmake:
bumped minimum version to 3.4, 2018-07-19) required CMake 3.4 to fix
issue #2746. This broke support for users on older versions of CMake
even if they just want to build curl and do not care whether transitive
dependencies work.
Backport the logic to work with CMake 3.0 again by implementing the
fix only when the version of CMake is at least 3.4.
Marcel Raad (27 Sep 2018)

- curl_threads: fix classic MinGW compile break
Classic MinGW still has _beginthreadex's return type as unsigned long

instead of uintptr_t [0]. uintptr_t is not even defined because of [1].
[0] https://sourceforge.net/p/mingw/mingw-org-wsl/ci/wsl-5.1-
release/tree/mingwrt/include/process.h#l167
[1] https://sourceforge.net/p/mingw/mingw-org-wsl/ci/wsl-5.1-
release/tree/mingwrt/include/process.h#l90
Bug: https://github.com/curl/curl/issues/2924#issuecomment-424334807
Daniel Stenberg (26 Sep 2018)

- configure: s/AC_RUN_IFELSE/CURL_RUN_IFELSE
fix a few leftovers
Fixes #3006
Closes #3049
- [Doron Behar brought this change]
example/htmltidy: fix include paths of tidy libraries
Closes #3050
- Curl_http2_done: fix memleak in error path

Free 'header_recvbuf' unconditionally even if 'h2' isn't (yet) set, for
early failures.
Detected by OSS-Fuzz
Closes #3046
- http: fix memleak in rewind error path
If the rewind would fail, a strdup() would not get freed.
Closes #3044
Viktor Szakats (24 Sep 2018)

- test320: fix regression in [ci skip]
The value in question is coming directly from `gnutls-serv`, so it cannot

be modified freely.

Ref:
https://github.com/curl/curl/commit/6ae6b2a533e8630afbb21f570305bd4ceece6348#commit
comment-30621004

- Curl_retry_request: fix memory leak
Closes #3042
- openssl: load built-in engines too
Regression since 38203f1
Reported-by: Jean Fabrice

Fixes #3023
Closes #3040
- [Christian Heimes brought this change]
OpenSSL: enable TLS 1.3 post-handshake auth
OpenSSL 1.1.1 requires clients to opt-in for post-handshake

authentication.
Fixes: https://github.com/curl/curl/issues/3026
Signed-off-by: Christian Heimes <christian@python.org>
- [Even Rouault brought this change]
Curl_dedotdotify(): always nul terminate returned string.

This fixes potential out-of-buffer access on "file:./" URL
$ valgrind curl "file:./"

==24516== Memcheck, a memory error detector
==24516== Copyright (C) 2002-2015, and GNU GPL'd, by Julian Seward et al.
==24516== Using Valgrind-3.11.0 and LibVEX; rerun with -h for copyright info
==24516== Command: /home/even/install-curl-git/bin/curl file:./
==24516==
==24516== Conditional jump or move depends on uninitialised value(s)
==24516== at 0x4C31F9C: strcmp (in /usr/lib/valgrind/vgpreload_memcheck-amd64-
linux.so)
==24516== by 0x4EBB315: seturl (urlapi.c:801)
==24516== by 0x4EBB568: parseurl (urlapi.c:861)
==24516== by 0x4EBC509: curl_url_set (urlapi.c:1199)
==24516== by 0x4E644C6: parseurlandfillconn (url.c:2044)
==24516== by 0x4E67AEF: create_conn (url.c:3613)
==24516== by 0x4E68A4F: Curl_connect (url.c:4119)
==24516== by 0x4E7F0A4: multi_runsingle (multi.c:1440)
==24516== by 0x4E808E5: curl_multi_perform (multi.c:2173)
==24516== by 0x4E7558C: easy_transfer (easy.c:686)
==24516== by 0x4E75801: easy_perform (easy.c:779)
==24516== by 0x4E75868: curl_easy_perform (easy.c:798)
Was originally spotted by

https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=10637
Credit to OSS-Fuzz
Closes #3039

- update URLs in tests
- and one in docs/MANUAL as well
- whitespace fixes
- replace tabs with spaces where possible

- remove line ending spaces
- remove double/triple newlines at EOF
- fix a non-UTF-8 character
- cleanup a few indentations/line continuations
in manual examples

- http: add missing return code check
Detected by Coverity. CID 1439610.
Follow-up from 46e164069d1a523
Closes #3034
- ftp: don't access pointer before NULL check

Detected by Coverity. CID 1439611.
Follow-up from 46e164069d1a523
- unit1650: fix out of boundary access
Fixes #2987
Closes #3035

- docs/examples: URL updates
- also update two URLs outside of docs/examples

- fix spelling of filename persistant.c
- fix three long lines that started failing checksrc.pl
- examples/Makefile.m32: sync with core [ci skip]
also:
- fix two warnings in synctime.c (one of them Windows-specific)
- upgrade URLs in synctime.c and remove a broken one

- examples/parseurl.c: show off the URL API a bit
Closes #3030
- SECURITY-PROCESS: mention the bountygraph program [ci skip]
Closes #3032
- url: use the URL API internally as well
... to make it a truly unified URL parser.
Closes #3017

- URL and mailmap updates, remove an obsolete directory [ci skip]

- configure: force-use -lpthreads on HPUX
When trying to detect pthreads use on HPUX the checks will succeed
without the correct -l option but then end up failing at run-time.
Reported-by: Eason-Yu on github

Fixes #2697
Closes #3025
- [Erik Minekus brought this change]

Curl_saferealloc: Fixed typo in docblock
Closes #3029
- urlapi: fix support for address scope in IPv6 numerical addresses
Closes #3024
- [Loganaden Velvindron brought this change]
GnutTLS: TLS 1.3 support
Closes #2971
- TODO: c-ares and CURLOPT_OPENSOCKETFUNCTION
Removed DoH.
Closes #2734
Jay Satiro (20 Sep 2018)

- vtls: fix ssl version "or later" behavior change for many backends
- Treat CURL_SSLVERSION_MAX_NONE the same as

CURL_SSLVERSION_MAX_DEFAULT. Prior to this change NONE would mean use
the minimum version also as the maximum.
This is a follow-up to 6015cef which changed the behavior of setting

the SSL version so that the requested version would only be the minimum
and not the maximum. It appears it was (mostly) implemented in OpenSSL
but not other backends. In other words CURL_SSLVERSION_TLSv1_0 used to
mean use just TLS v1.0 and now it means use TLS v1.0 *or later*.
- Fix CURL_SSLVERSION_MAX_DEFAULT for OpenSSL.
Prior to this change CURL_SSLVERSION_MAX_DEFAULT with OpenSSL was

erroneously treated as always TLS 1.3, and would cause an error if
OpenSSL was built without TLS 1.3 support.
Co-authored-by: Daniel Gustafsson

- certs: generate tests certs with sha256 digest algorithm
As OpenSSL 1.1.1 starts to complain and fail on sha1 CAs:
"SSL certificate problem: CA signature digest algorithm too weak"
Closes #3014
- urlapi: document the error codes, remove two unused ones

Closes #3019
- urlapi: add CURLU_GUESS_SCHEME and fix hostname acceptance
In order for this API to fully work for libcurl itself, it now offers a
CURLU_GUESS_SCHEME flag that makes it "guess" scheme based on the host
name prefix just like libcurl always did. If there's no known prefix, it
will guess "http://".
Separately, it relaxes the check of the host name so that IDN host names
can be passed in as well.
Both these changes are necessary for libcurl itself to use this API.

Closes #3018
Kamil Dudka (19 Sep 2018)

- nss: try to connect even if libnssckbi.so fails to load
One can still use CA certificates stored in NSS database.
Reported-by: Maxime Legros

Closes #3016
Daniel Gustafsson (19 Sep 2018)

- urlapi: don't set value which is never read
In the CURLUPART_URL case, there is no codepath which invokes url

decoding so remove the assignment of the urldecode variable. This
fixes the deadstore bug-report from clang static analysis.
Closes #3015
- todo: Update reference to already done item
TODO item 1.1 was implemented in commit 946ce5b61f, update reference

to it with instead referencing the implemented option.
Closes #3013

- [slodki brought this change]
cmake: don't require OpenSSL if USE_OPENSSL=OFF
User must have OpenSSL installed even if not used by libcurl at all
since 7.61.1 release. Broken at
7867aaa9a01decf93711428462335be8cef70212

Closes #3001
- curl_multi_wait: call getsock before figuring out timeout

.... since getsock may update the expiry timer.
Fixes #2996
Closes #3000
- examples/http2-pushinmemory: receive HTTP/2 pushed files in memory
Closes #3004

- darwinssl: Fix realloc memleak
The reallocation was using the input pointer for the return value, which
leads to a memory leak on reallication failure. Fix by instead use the
safe internal API call Curl_saferealloc().
Closes #3005
Reviewed-by: Nick Zitzmann <nickzman@gmail.com>
- [Kruzya brought this change]
examples: Fix memory leaks from realloc errors
Make sure to not overwrite the reallocated pointer in realloc() calls

to avoid a memleak on memory errors.
- memory: add missing curl_printf header
ftp_send_command() was using vsnprintf() without including the libcurl

*rintf() replacement header. Fix by including curl_printf.h and also
add curl_memory.h while at it since memdebug.h depends on it.
Closes #2999

- [Si brought this change]
curl: update --tlsv* descriptions in --help output
Closes #2994
- http: made Curl_add_buffer functions take a pointer-pointer
... so that they can clear the original pointer on failure, which makes
the error-paths and their cleanups easier.
Closes #2992
- http2: fix memory leaks on error-path
libtest: Add chkdecimalpoint to .gitignore
Closes #2998

- secure Openwall URLs

- openssl: show "proper" version number for libressl builds
Closes #2989
- [Rainer Jung brought this change]
openssl: assume engine support in 0.9.8 or later
Fixes #2983
Closes #2988

- sendf: use failf() rather than Curl_failf()
The failf() macro is the name used for invoking Curl_failf(). While
there isn't a way to turn off failf like there is for infof, but it's
still a good idea to use the macro.
- sendf: Fix whitespace in infof/failf concatenation
Strings broken on multiple rows in the .c file need to have appropriate

whitespace padding on either side of the concatenation point to render
a correct amalgamated string. Fix by adding a space at the occurrences
found.
Closes #2986
- krb5: fix memory leak in krb_auth
The FTP command allocated by aprintf() must be freed after usage.
- ftp: include command in Curl_ftpsend sendbuffer
Commit 8238ba9c5f10414a88f502bf3f5d5a42d632984c inadvertently removed

the actual command to be sent from the send buffer in a refactoring.
Add back copying the command into the buffer. Also add more guards
against malformed input while at it.
Closes #2985
- ntlm_wb: Fix memory leaks in ntlm_wb_response
When erroring out on a request being too large, the existing buffer was
leaked. Fix by explicitly freeing on the way out.
Closes #2966

- [Yiming Jing brought this change]
travis: build the MesaLink vtls backend with MesaLink 0.7.1
runtests.pl: run tests against the MesaLink vtls backend
vtls: add a MesaLink vtls backend
Closes #2984
configure.ac: add a MesaLink vtls backend
- [Dave Reisner brought this change]
curl_url_set.3: properly escape \n in example code
This yields
"the scheme is %s\n"
instead of
"the scheme is %s0
Closes #2970
curl_url_set.3: fix typo in reference to CURLU_APPENDQUERY
- urlglob: improve error message
to help user understand what the problem is
Reported-by: Daniel Shahaf
Fixes #2763
Closes #2977
tests/certs: rebuild certs with 2048-bit RSA keys
The previous test certificates contained RSA keys of only 1024 bits.
However, RSA claims that 1024-bit RSA keys are likely to become
crackable some time before 2010. The NIST recommends at least 2048-bit
keys for RSA for now.
Better use full 2048 also for testing.
Closes #2973

- TODO: fix typo in item
Closes #2968

- anyauthput: fix compiler warning on 64-bit Windows
On Windows, the read function from <io.h> is used, which has its byte
count parameter as unsigned int instead of size_t.

- lib: fix gcc8 warning on Windows
Jay Satiro (12 Sep 2018)

- openssl: fix gcc8 warning
- Use memcpy instead of strncpy to copy a string without termination,

since gcc8 warns about using strncpy to copy as many bytes from a
string as its length.
Suggested-by: Viktor Szakats

- libcurl-url.3: overview man page for the URL API
Closes #2967
- example/asiohiper: insert warning comment about its status
This example is simply not working correctly but there's nobody around
with the skills and energy to fix it.
Closes #2407

- docs/cmdline-opts: update the documentation of --tlsv1.0
... to reflect the changes in 6015cefb1b2cfde4b4850121c42405275e5e77d9
Closes #2955
- docs/examples: do not wait when no transfers are running
Closes #2948

cookies: Move failure case label to end of function
Rather than jumping backwards to where failure cleanup happens

to be performed, move the failure case to end of the function
where it is expected per existing coding convention.
Closes #2965
misc: fix typos in comments
Closes #2963
cookies: fix leak when writing cookies to file
If the formatting fails, we error out on a fatal error and

clean up on the way out. The array was however freed within
the wrong scope and was thus never freed in case the cookies
were written to a file instead of STDOUT.
Closes #2957
cookies: Remove redundant expired check
Expired cookies have already been purged at a later expiration time

before this check, so remove the redundant check.
closes #2962
- ntlm_wb: bail out if the response gets overly large
Exit the realloc() loop if the response turns out ridiculously large to
avoid worse problems.
Reported-by: Harry Sintonen

Closes #2959
url.c: fix comment typo and indentation
Closes #2960
- urlapi: avoid derefencing a possible NULL pointer
Coverity CID 1439134

- test324: fix after 3f3b26d6feb0667714902e836af608094235fca2
The expected error code is now 60. 51 is dead.

- curl_url_set.3: correct description
- curl_url-docs: fix AVAILABILITY as Added in curl 7.62.0

- URL-API
See header file and man pages for API. All documented API details work
and are tested in the 1560 test case.
Closes #2842
- curl_easy_upkeep: removed 'conn' from the name
... including the associated option.
Fixes #2951
Closes #2952
- [Max Dymond brought this change]
upkeep: add a connection upkeep API: curl_easy_conn_upkeep()
Add functionality so that protocols can do custom keepalive on their

connections, when an external API function is called.
Add docs for the new options in 7.62.0
Closes #1641
- [Philipp Waehnert brought this change]
configure: add option to disable automatic OpenSSL config loading
Sometimes it may be considered a security risk to load an external

OpenSSL configuration automatically inside curl_global_init(). The
configuration option --disable-ssl-auto-load-config disables this
automatism. The Windows build scripts winbuild/Makefile.vs provide a
corresponding option ENABLE_SSL_AUTO_LOAD_CONFIG accepting a boolean
value.
Setting neither of these options corresponds to the previous behavior

loading the external OpenSSL configuration automatically.
Fixes #2724
Closes #2791
- doh: minor edits to please Coverity
The gcc typecheck macros and coverity combined made it warn on the 2nd
argument for ERROR_CHECK_SETOPT(). Here's minor rearrange to please it.
Coverity CID 1439115 and CID 1439114.
- schannel: avoid switch-cases that go to default anyway
SEC_E_APPLICATION_PROTOCOL_MISMATCH isn't defined in some versions of

mingw and would require an ifdef otherwise.
Reported-by: Thomas Glanzmann

Approved-by: Marc Hörsken
Closes #2950
- [Nicklas Avén brought this change]
imap: change from "FETCH" to "UID FETCH"
... and add "MAILINDEX".
As described in #2789, this is a suggested solution. Changing UID=xx to

actually get mail with UID xx and add "MAILINDEX" to get a mail with a
special index in the mail box (old behavior). So MAILINDEX=1 gives the
first non deleted mail in the mail box.
Fixes #2789
Closes #2815
- CURLOPT_UPLOAD_BUFFERSIZE: set upload buffer size
This is step 3 of #2888.
Fixes #2888
Closes #2896
- travis: add the DOH tests to the torture testing
- DOH: add test case 1650 and 2100
- curl: --doh-url added
- setopt: add CURLOPT_DOH_URL
Closes #2668
ssl: deprecate CURLE_SSL_CACERT in favour of a unified error code
Long live CURLE_PEER_FAILED_VERIFICATION
x509asn1: return CURLE_PEER_FAILED_VERIFICATION on failure to parse cert
CURLE_PEER_FAILED_VERIFICATION makes more sense because Curl_parseX509

does not allocate memory internally as its first argument is a pointer
to the certificate structure. The same error code is also returned by
Curl_verifyhost when its call to Curl_parseX509 fails so the change
makes error handling more consistent.
openssl: return CURLE_PEER_FAILED_VERIFICATION on failure to parse issuer
Failure to extract the issuer name from the server certificate should
return a more specific error code like on other TLS backends.
schannel: unified error code handling
Closes #2901
darwinssl: more specific and unified error codes
Closes #2901
- CURLOPT_DNS_USE_GLOBAL_CACHE: deprecated
Disable the CURLOPT_DNS_USE_GLOBAL_CACHE option and mark it for

deprecation and complete removal in six months.
Closes #2942
- url: default to CURL_HTTP_VERSION_2TLS if built h2-enabled
Closes #2709
- multiplex: enable by default
Starting 7.62.0, multiplexing is enabled by default in multi handles.
- [Jim Fuller brought this change]
tests: add unit tests for url.c
Approved-by: Daniel Gustafsson

Closes #2937
- test1452: mark as flaky
makes it not run in the CI builds
Closes #2941
- pipelining: deprecated
Transparently. The related curl_multi_setopt() options all still returns

OK when pipelining is selected.
To re-enable the support, the single line change in lib/multi.c needs to

be reverted.
See docs/DEPRECATE.md
Closes #2705
- RELEASE-NOTES: start working on 7.62.0
Version 7.61.1 (4 Sep 2018)

- Curl_getoff_all_pipelines: ignore unused return values

Since scan-build would warn on the dead "Dead store/Dead increment"

- sftp: fix indentation

- [Przemysław Tomaszewski brought this change]
sftp: don't send post-qoute sequence when retrying a connection
Fixes #2939
Closes #2940

- url, vtls: make CURLOPT{,_PROXY}_TLS13_CIPHERS work
This is a follow-up to PR #2607 and PR #2926.
Closes #2936

tool_operate: Add http code 408 to transient list for --retry
- Treat 408 request timeout as transient so that curl will retry the
request if --retry was used.
Closes #2925
openssl: Fix setting TLS 1.3 cipher suites
The flag indicating TLS 1.3 cipher support in the OpenSSL backend was
missing.
Bug: https://github.com/curl/curl/pull/2607#issuecomment-417283187
Closes #2926
- Curl_ntlm_core_mk_nt_hash: return error on too long password
... since it would cause an integer overflow if longer than (max size_t
/ 2).
This is CVE-2018-14618
Closes #2756
Reported-by: Zhaoyang Wu
http2: Use correct format identifier for stream_id
Closes #2928
- test1148: fix precheck output
"precheck command error" is not very helpful.

- all: s/int/size_t cleanup
Assisted-by: Rikard Falkeborn
Closes #2922
- ssh-libssh: use FALLTHROUGH to silence gcc8
Jay Satiro (31 Aug 2018)

- tool_operate: Fix setting proxy TLS 1.3 ciphers
Daniel Stenberg (31 Aug 2018)

cookies: support creation-time attribute for cookies
According to RFC6265 section 5.4, cookies with equal path lengths

SHOULD be sorted by creation-time (earlier first). This adds a
creation-time record to the cookie struct in order to make cookie
sorting more deterministic. The creation-time is defined as the
order of the cookies in the jar, the first cookie read fro the
jar being the oldest. The creation-time is thus not serialized
into the jar. Also remove the strcmp() matching in the sorting as
there is no lexicographic ordering in RFC6265. Existing tests are
updated to match.
Closes #2524
Marcel Raad (31 Aug 2018)

- Don't use Windows path %PWD for SSH tests
All these tests failed on Windows because something like

sftp://%HOSTIP:%SSHPORT%PWD/
expanded to
sftp://127.0.0.1:1234c:/msys64/home/bla/curl
and then curl complained about the port number ending with a letter.
Use the original POSIX path instead of the Windows path created in
checksystem to fix this.

- CURLOPT_SSL_CTX_FUNCTION.3: clarify connection reuse warning
Reported-by: Daniel Stenberg

- THANKS-filter: dedup Daniel Jeliński
- CURLOPT_ACCEPT_ENCODING.3: list them comma-separated [ci skip]
- CURLOPT_SSL_CTX_FUNCTION.3: might cause unintended connection reuse [ci skip]
Added a warning!
Closes #2915
- curl: fix time-of-check, time-of-use race in dir creation
Patch-by: Jay Satiro

Detected by Coverity
Fixes #2739
Closes #2912
- cmdline-opts/page-footer: fix edit mistake
There was a missing newline.
follow-up to a7ba60bb7250
- docs: clarify NO_PROXY env variable functionality
Reported-by: Kirill Marchuk

Fixes #2773
Closes #2911

- lib1522: fix curl_easy_setopt argument type
CURLOPT_POSTFIELDSIZE is a long option.
- curl_threads: silence bad-function-cast warning
As uintptr_t and HANDLE are always the same size, this warning is
harmless. Just silence it using an intermediate uintptr_t variable.

- README: add appveyor build badge [ci skip]
Closes #2913
- [Ihor Karpenko brought this change]
schannel: client certificate store opening fix
1) Using CERT_STORE_OPEN_EXISTING_FLAG ( or CERT_STORE_READONLY_FLAG )

while opening certificate store would be sufficient in this scenario and
less-demanding in sense of required user credentials ( for example,
IIS_IUSRS will get "Access Denied" 0x05 error for existing CertOpenStore
call without any of flags mentioned above ),
2) as 'cert_store_name' is a DWORD, attempt to format its value like a

string ( in "Failed to open cert store" error message ) will throw null
pointer exception
3) adding GetLastError(), in my opinion, will make error message more
useful.
Closes #2909
gopher: Do not translate `?' to `%09'
Since GOPHER support was added in curl `?' character was automatically
translated to `%09' (`\t').
However, this behaviour does not seems documented in RFC 4266 and for
search selectors it is documented to directly use `%09' in the URL.
Apart that several gopher servers in the current gopherspace have CGI
support where `?' is used as part of the selector and translating it to
`%09' often leads to surprising results.
Closes #2910

- cookie tests: treat files as text
Fixes test failures because of wrong line endings on Windows.

- libcurl-thread.3: expand somewhat on the NO_SIGNAL motivation
Multi-threaded applictions basically MUST set CURLOPT_NO_SIGNAL to 1L to

avoid the risk of getting a SIGPIPE.
Either way, a multi-threaded application that uses libcurl/openssl needs

to have a signhandler for or ignore SIGPIPE on its own.
Based on discussions in #2800

Closes #2904

- Tests: fixes for Windows
- test 1268 requires unix sockets

- test 2072 must be disabled also for MSYS/MinGW

- http2: abort the send_callback if not setup yet
When Curl_http2_done() gets called before the http2 data is setup all
the way, we cannot send anything and this should just return an error.
- http2: remove four unused nghttp2 callbacks
Closes #2903
- x509asn1: use FALLTHROUGH
... as no other comments are accepted since 014ed7c22f51463

- test1148: disable if decimal separator is not point
Modifying the locale with environment variables doesn't work for native
Windows applications. Just disable the test in this case if the decimal
separator is something different than a point. Use a precheck with a
small C program to achieve that.
- Enable more GCC warnings
This enables the following additional warnings:

-Wold-style-definition
-Warray-bounds=2 instead of the default 1
-Wformat=2, but only for GCC 4.8+ as Wno-format-nonliteral is not
respected for older versions
-Wunused-const-variable, which enables level 2 instead of the default 1
-Warray-bounds also in debug mode through -ftree-vrp
-Wnull-dereference also in debug mode through
-fdelete-null-pointer-checks
- curl-compilers: enable -Wimplicit-fallthrough=4 for GCC
This enables level 4 instead of the default level 3, which of the

currently used comments only allows /* FALLTHROUGH */ to silence the
warning.
- curl-compilers: enable -Wbad-function-cast on GCC
This warning used to be enabled only for clang as it's a bit stricter
on GCC. Silence the remaining occurrences and enable it on GCC too.
- configure: conditionally enable pedantic-errors
Enable pedantic-errors for GCC >= 5 with --enable-werror. Before GCC 5,

pedantic-errors was synonymous to -Werror=pedantic [0], which is still
the case for clang [1]. With GCC 5, it became complementary [2].
Also fix a resulting error in acinclude.m4 as main's return type was

missing, which is illegal in C99.
[0] https://gcc.gnu.org/onlinedocs/gcc-4.9.0/gcc/Warning-Options.html
[1] https://clang.llvm.org/docs/UsersManual.html#options-to-control-error-and-
warning-messages
- Remove unused definitions

- x509asn1: make several functions static
and remove the private SIZE_T_MAX define and use the generic one.
Closes #2902
- INTERNALS: require GnuTLS >= 2.11.3
Since the public pinning support was brought in e644866caf4. GnuTLS

2.11.3 was released in October 2010.
Figured out in #2890
- http2: avoid set_stream_user_data() before stream is assigned
... before the stream is started, we have it set to -1.
Fixes #2894
Closes #2898
- SSLCERTS: improve the openssl command line
... for extracting certs from a live HTTPS server to make a cacerts.pem
from them.
- docs/SECURITY-PROCESS: now we name the files after the CVE id
- upload: change default UPLOAD_BUFSIZE to 64KB
To make uploads significantly faster in some circumstances.
Part 2 of #2888
Closes #2892
- upload: allocate upload buffer on-demand
Saves 16KB on the easy handle for operations that don't need that
buffer.
Part 1 of #2888
- [Laurent Bonnans brought this change]
vtls: reinstantiate engine on duplicated handles
Handles created with curl_easy_duphandle do not use the SSL engine set
up in the original handle. This fixes the issue by storing the engine
name in the internal url state and setting the engine from its name
inside curl_easy_duphandle.
Reported-by: Anton Gerasimov

Signed-of-by: Laurent Bonnans
Fixes #2829
Closes #2833
- http2: make sure to send after RST_STREAM
If this is the last stream on this connection, the RST_STREAM might not
get pushed to the wire otherwise.
Fixes #2882
Closes #2887
Researched-by: Michael Kaufmann
- test1268: check the stderr output as "text"
Follow-up to 099f37e9c57
Pointed-out-by: Marcel Raad
- urldata: remove unused pipe_broke struct field
This struct field is never set TRUE in any existing code path. This
change removes the field completely.
Closes #2871
- curl: warn the user if a given file name looks like an option
... simply because this is usually a sign of the user having omitted the
file name and the next option is instead "eaten" by the parser as a file
name.
Add test1268 to verify
Closes #2885
- http2: check nghttp2_session_set_stream_user_data return code
Might help bug #2688 debugging
Closes #2880
- travis: revert back to gcc-7 for coverage builds
... since the gcc-8 ones seem to fail frequently.
Follow-up from b85207199544ca
Closes #2886
... and now listed in alphabetical order!
- [Adrien brought this change]
CMake: CMake config files are defining CURL_STATICLIB for static builds
This change allows to use the CMake config files generated by Curl's
CMake scripts for static builds of the library.
The symbol CURL_STATIC lib must be defined to compile downstream,
thus the config package is the perfect place to do so.
Fixes #2817
Closes #2823
Reported-by: adnn on github
- TODO: host name sections in config files
Kamil Dudka (14 Aug 2018)

- ssh-libssh: fix infinite connect loop on invalid private key
Added test 656 (based on test 604) to verify the fix.
Bug: https://bugzilla.redhat.com/1595135
Closes #2879
- ssh-libssh: reduce excessive verbose output about pubkey auth
The verbose message "Authentication using SSH public key file" was
printed each time the ssh_userauth_publickey_auto() was called, which
meant each time a packet was transferred over network because the API
operates in non-blocking mode.
This patch makes sure that the verbose message is printed just once
(when the authentication state is entered by the SSH state machine).

- travis: disable h2 torture tests for "coverage"
Since they started to fail almost 100% since a few days.
Closes #2876

- travis: update to GCC 8

- http: fix for tiny "HTTP/0.9" response
Deal with tiny "HTTP/0.9" (header-less) responses by checking the

status-line early, even before a full "HTTP/" is received to allow
detecting 0.9 properly.
Test 1266 and 1267 added to verify.
Fixes #2420
Closes #2872
Kamil Dudka (13 Aug 2018)

- docs: add disallow-username-in-url.d and haproxy-protocol.d on the list
... to make make the files appear in distribution tarballs

Closes #2856
- .travis.yml: verify that man pages can be regenerated
... when curl is built from distribution tarball
Closes #2856

- Split non-portable part off test 1133
Split off testing file names with double quotes into new test 1158.
Disable it for MSYS using a precheck as it doesn't support file names
with double quotes (but Cygwin does, for example).

- projects: Improve Windows perl detection in batch scripts
- Determine if perl is in the user's PATH by running perl.exe.
Prior to this change detection was done by checking the PATH for perl/
but that did not work in all cases (eg git install includes perl but
not in perl/ path).
Bug: https://github.com/curl/curl/pull/2865
Reported-by: Daniel Jeliński
docs: Improve the manual pages of some callbacks
- CURLOPT_HEADERFUNCTION: add newlines

- CURLOPT_INTERLEAVEFUNCTION: fix the description of 'userdata'
- CURLOPT_READDATA: mention crashes, same as in CURLOPT_WRITEDATA
- CURLOPT_READFUNCTION: rename 'instream' to 'userdata' and explain
how to set it

- GCC: silence -Wcast-function-type uniformly
Pointed-out-by: Rikard Falkeborn

- Silence GCC 8 cast-function-type warnings
On Windows, casting between unrelated function types is fine and

sometimes even necessary, so just use an intermediate cast to
(void (*) (void)) to silence the warning as described in [0].

- CURLINFO_SIZE_UPLOAD: fix missing counter update
Adds test 1522 for verification.
Reported-by: cjmsoregan
Fixes #2847
Closes #2864
- [Daniel Jelinski brought this change]
Documentation: fix CURLOPT_SSH_COMPRESSION copy/paste bug
Closes #2867
- openssl: fix potential NULL pointer deref in is_pkcs11_uri
Follow-up to 298d2565e
Coverity CID 1438387

- travis: execute "set -eo pipefail" for coverage build
Follow-up to 2de63ab179eb78630ee039ad94fb2a5423df522d and

0b87c963252d3504552ee0c8cf4402bd65a80af5.

- lib1502: fix memory leak in torture test

Fixes #2861
Closes #2863
- docs: mention NULL is fine input to several functions
Fixes #2837
Closes #2858
Reported-by: Markus Elfring
- [Bas van Schaik brought this change]
README.md: add LGTM.com code quality grade for C/C++
Closes #2857
test1531: Add timeout
Previously, the macro TEST_HANG_TIMEOUT was unused, but since there is

looping going on, we might as well add timing instead of removing it.
Closes #2853

test1540: Remove unused macro TEST_HANG_TIMEOUT
The macro has never been used, and it there is not really any place
where it would make sense to add timing checks.
Closes #2852
asyn-thread: Remove unused macro
The macro seems to never have been used.
Closes #2852
http_proxy: Remove unused macro SELECT_TIMEOUT
Usage was removed in 5113ad0424044458ac497fa1458ebe0101356b22.
Closes #2852
formdata: Remove unused macro HTTPPOST_CONTENTTYPE_DEFAULT
Its usage was removed in

84ad1fd3047815f9c6e78728bb351b828eac10b1.
Closes #2852
telnet: Remove unused macros TELOPTS and TELCMDS
Their usage was removed in 3a145180cc754a5959ca971ef3cd243c5c83fc51.
Closes #2852
openssl: fix debug messages
Fixes #2806
Closes #2843
- configure: fix for -lpthread detection with OpenSSL and pkg-config
... by making sure it uses the -I provided by pkg-config!
Reported-by: pszemus on github

Fixes #2848
Closes #2850
- windows: follow up to the buffer-tuning 1ba1dba7

Somehow I didn't include the amended version of the previous fix. This
is the missing piece.
Pointed-out-by: Viktor Szakats
windows: implement send buffer tuning
Significantly enhances upload performance on modern Windows versions.
Closes #2762
Fixes #2224
- [Anderson Toshiyuki Sasaki brought this change]
ssl: set engine implicitly when a PKCS#11 URI is provided
This allows the use of PKCS#11 URI for certificates and keys without
setting the corresponding type as "ENG" and the engine as "pkcs11"
explicitly. If a PKCS#11 URI is provided for certificate, key,
proxy_certificate or proxy_key, the corresponding type is set as "ENG"
if not provided and the engine is set to "pkcs11" if not provided.
Acked-by: Nikos Mavrogiannopoulos

Closes #2333
CMake: Respect BUILD_SHARED_LIBS
Use standard CMake variable BUILD_SHARED_LIBS instead of introducing

custom option CURL_STATICLIB.
Use '-DBUILD_SHARED_LIBS=%SHARED%' in appveyor.yml.

Closes #2755
- [John Butterfield brought this change]
cmake: bumped minimum version to 3.4
Closes #2753
- [John Butterfield brought this change]
cmake: link curl to the OpenSSL targets instead of lib absolute paths
Reviewed-by: Jakub Zakrzewski

Closes #2753
- travis: build darwinssl on macos 10.12
... as building on 10.13.x before 10.13.4 leads to link errors.
Assisted-by: Nick Zitzmann

Fixes #2835
Closes #2845
- DEPRECATE: remove release date from 7.62.0
Since it will slip and the version is the important part there, not the
date.
- lib/Makefile: only do symbol hiding if told to
This restores the ability to build a static lib with

--disable-symbol-hiding to keep non-curl_ symbols.
Researched-by: Dan Fandrich

Reported-by: Ran Mozes
Fixes #2830
Closes #2831

- hostip: fix unused variable warning
addresses is only used in an infof call, which is a macro expanding to

nothing if CURL_DISABLE_VERBOSE_STRINGS is set.

- test1307: disabled
Turns out that since we're using the native fnmatch function now when
available, and they simply disagree on a huge number of test patterns
that make it hard to test this function like this...
Fixes #2825
- smb: don't mark it done in smb_do
Follow-up to 09e401e01bf9. The SMB protocol handler needs to use its

doing function too, which requires smb_do() to not mark itself as
done...
Closes #2822
general: fix printf specifiers
Closes #2818
- mailmap: Daniel Jelinski
- [Harry Sintonen brought this change]
HTTP: Don't attempt to needlessly decompress redirect body
This change fixes a regression where redirect body would needlessly be

decompressed even though it was to be ignored anyway. As it happens this
causes secondary issues since there appears to be a bug in apache2 that
it in certain conditions generates a corrupt zlib response. The
regression was created by commit:
dbcced8e32b50c068ac297106f0502ee200a1ebd
Discovered-by: Harry Sintonen

Closes #2798
- curl: use Content-Disposition before the "URL end" for -OJ
Regression introduced in 7.61.0
Reported-by: Thomas Klausner

Fixes #2783
Closes #2813
retry: return error if rewind was necessary but didn't happen
Fixes #2801
Closes #2812
- http2: clear the drain counter in Curl_http2_done
Reported-by: Andrei Virtosu

Fixes #2800
Closes #2809
- smb: fix memory leak on early failure
... by making sure connection related data (->share) is stored in the

connection and not in the easy handle.
Detected by OSS-fuzz
Fixes #2769
Closes #2810
- travis: run a 'make checksrc' too
... to make sure the examples are all checked.
Closes #2811
Jay Satiro (29 Jul 2018)

- examples/ephiperfifo: checksrc compliance
sws: handle EINTR when calling select()
Daniel Stenberg (29 Jul 2018)

- test1157: follow-up to 35ecffb9
Ignore the user-agent line.

Pointed-out-by: Marcel Raad
Michael Kaufmann (29 Jul 2018)

- tests/http_pipe.py: Use /usr/bin/env to find python

- TODO: Support Authority Information Access certificate extension (AIA)
Closes #2793
- conn_free: updated comment to clarify
Let's call it disassociate instead of disconnect since the latter term

is used so much for (TCP) connections already.
- test1157: test -H from empty file
Verifies bugfix #2797
- [Tobias Blomberg brought this change]
curl: Fix segfault when -H @headerfile is empty
The curl binary would crash if the -H command line option was given a
filename to read using the @filename syntax but that file was empty.
Closes #2797
- mime: check Curl_rand_hex's return code
Bug: https://curl.haxx.se/mail/archive-2018-07/0015.html
Reported-by: Jeffrey Walton
Closes #2795
- [Josh Bialkowski brought this change]
docs/examples: add hiperfifo example using linux epoll/timerfd
Closes #2804
- [Darío Hereñú brought this change]
docs/INSTALL.md: minor formatting fixes
Closes #2794
- [Christopher Head brought this change]
docs/CURLOPT_URL: fix indentation
The statement, “The application does not have to keep the string around
after setting this option,” appears to be indented under the RTMP
paragraph. It actually applies to all protocols, not just RTMP.
Eliminate the extra indentation.
Closes #2788
- [Christopher Head brought this change]
docs/CURLOPT_WRITEFUNCTION: size is always 1
For compatibility with `fwrite`, the `CURLOPT_WRITEFUNCTION` callback is

passed two `size_t` parameters which, when multiplied, designate the
number of bytes of data passed in. In practice, CURL always sets the
first parameter (`size`) to 1.
This practice is also enshrined in documentation and cannot be changed

in future. The documentation states that the default callback is
`fwrite`, which means `fwrite` must be a suitable function for this
purpose. However, the documentation also states that the callback must
return the number of *bytes* it successfully handled, whereas ISO C
`fwrite` returns the number of items (each of size `size`) which it
wrote. The only way these numbers can be equal is if `size` is 1.
Since `size` is 1 and can never be changed in future anyway, document

that fact explicitly and let users rely on it.
Closes #2787
- [Carie Pointer brought this change]
wolfSSL/CyaSSL: Fix memory leak in Curl_cyassl_random
RNG structure must be freed by call to FreeRng after its use in

Curl_cyassl_random. This call fixes Valgrind failures when running the
test suite with wolfSSL.
Closes #2784
- [Even Rouault brought this change]
reuse_conn(): free old_conn->options
This fixes a memory leak when CURLOPT_LOGIN_OPTIONS is used, together with

connection reuse.
I found this with oss-fuzz on GDAL and curl master:

I couldn't reproduce with the oss-fuzz original test case, but looking
at curl source code pointed to this well reproducable leak.
Closes #2790
Marcel Raad (25 Jul 2018)

system_win32: fix version checking
In the current version, VERSION_GREATER_THAN_EQUAL 6.3 will return false

when run on windows 10.0. This patch addresses that error.

auth: pick Bearer authentication whenever a token is available
So far, the code tries to pick an authentication method only if

user/password credentials are available, which is not the case for
Bearer authentictation...
Closes #2754
auth: only ever pick CURLAUTH_BEARER if we *have* a Bearer token
The Bearer authentication was added to cURL 7.61.0, but there is a

problem: if CURLAUTH_ANY is selected, and the server supports multiple
authentication methods including the Bearer method, we strongly prefer
that latter method (only CURLAUTH_NEGOTIATE beats it), and if the Bearer
authentication fails, we will never even try to attempt any other
method.
This is particularly unfortunate when we already know that we do not

have any Bearer token to work with.
Such a scenario happens e.g. when using Git to push to Visual Studio
Team Services (which supports Basic and Bearer authentication among
other methods) and specifying the Personal Access Token directly in the
URL (this aproach is frequently taken by automated builds).
Let's make sure that we have a Bearer token to work with before we
select the Bearer authentication among the available authentication
methods.

Closes #2754

- test320: treat curl320.out file as binary
Otherwise, LF line endings are converted to CRLF on Windows,

but no conversion is done for the reply, so the test case fails.

- vtls: set conn->data when closing TLS
Follow-up to 1b76c38904f0. The VTLS backends that close down the TLS
layer for a connection still needs a Curl_easy handle for the session_id
cache etc.
Fixes #2764
Closes #2771

- tests: fixes for Windows line endlings
Set mode="text" when line endings depend on the system representation.
- test214: disable MSYS2's POSIX path conversion for URL
By default, the MSYS2 bash converts all backslashes to forward slashes

in URLs. Disable this with MSYS2_ARG_CONV_EXCL for the test to pass.
Ref https://github.com/msys2/msys2/wiki/Porting#filesystem-namespaces

- http2: several cleanups
- separate easy handle from connections better

- added asserts on a number of places
- added sanity check of pipelines for debug builds
Closes #2751
- smb_getsock: always wait for write socket too
... the protocol is doing read/write a lot, so it needs to write often

even when downloading. A more proper fix could check for eactly when it
wants to write and only ask for it then.
Without this fix, an SMB download could easily get stuck when the event-driven
API was used.
Closes #2768

- test1143: disable MSYS2's POSIX path conversion
By default, the MSYS2 bash interprets http:/%HOSTIP:%HTTPPORT/want/1143

as a POSIX file list and converts it to a Windows file list.
Disable this with MSYS2_ARG_CONV_EXCL for the test to pass.
Ref https://github.com/msys2/msys2/wiki/Porting#filesystem-namespaces

- RELEASE-NOTES: sync
... and work toward 7.61.1
CMake: Update scripts to use consistent style
Closes #2727
- header output: switch off all styles, not just unbold
... the "unbold" sequence doesn't work on the mac Terminal.
Reported-by: Zero King

Fixes #2736
Closes #2738
Nick Zitzmann (14 Jul 2018)

- [Rodger Combs brought this change]
darwinssl: add support for ALPN negotiation

- test1422: add required file feature
curl configured with --enable-debug --disable-file currently complains

on test1422:
Info: Protocol "file" not supported or disabled in libcurl
Make test1422 dependend on enabled FILE protocol to fix this.
Patrick Monnerat (12 Jul 2018)

- content_encoding: accept up to 4 unknown trailer bytes after raw deflate data
Some servers issue raw deflate data that may be followed by an undocumented
trailer. This commit makes curl tolerate such a trailer of up to 4 bytes
before considering the data is in error.
Reported-by: clbr on github

Fixes #2719

- smb: fix memory-leak in URL parse error path
Closes #2740

- schannel: enable CALG_TLS1PRF for w32api >= 5.1
The definition of CALG_TLS1PRF has been fixed in the 5.1 branch:

https://osdn.net/projects/mingw/scm/git/mingw-org-
wsl/commits/73aedcc0f2e6ba370de0d86ab878ad76a0dda7b5

- docs/SECURITY-PROCESS: mention bounty, drop pre-notify
+ The hackerone bounty and its process
- We don't and can't handle pre-notification
- multi: always do the COMPLETED procedure/state
It was previously erroneously skipped in some situations.
libtest/libntlmconnect.c wrongly depended on wrong behavior (that it

would get a zero timeout) when no handles are "running" in a multi
handle. That behavior is no longer present with this fix. Now libcurl
will always return a -1 timeout when all handles are completed.
Closes #2733
- Curl_getoff_all_pipelines: improved for multiplexed
On multiplexed connections, transfers can be removed from anywhere not

just at the head as for pipelines.
- ares: check for NULL in completed-callback

- conn: remove the boolean 'inuse' field
... as the usage needs to be counted.
openssl: assume engine support in 1.0.0 or later
Commit 38203f1585da changed engine detection to be version-based,

with a baseline of openssl 1.0.1. This does in fact break builds
with openssl 1.0.0, which has engine support - the configure script
detects that ENGINE_cleanup() is available - but <openssl/engine.h>
doesn't get included to declare it.
According to upstream documentation, engine support was added to

mainstream openssl builds as of version 0.9.7:
https://github.com/openssl/openssl/blob/master/README.ENGINE
This commit drops the version test down to 1.0.0 as version 1.0.0d
is the oldest version I have to test with.
Closes #2732

- schannel: fix MinGW compile break
Original MinGW's w32api has a sytax error in its definition of

CALG_TLS1PRF [0]. Don't use original MinGW w32api's CALG_TLS1PRF
until this bug [1] is fixed.
[0] https://osdn.net/projects/mingw/scm/git/mingw-org-
wsl/blobs/d1d4a17e51a2b78e252ef0147d483267d56c90cc/w32api/include/wincrypt.h
[1] https://osdn.net/projects/mingw/ticket/38391
Fixes https://github.com/curl/curl/pull/2721#issuecomment-403636043

- examples/crawler.c: move #ifdef to column 0
Apparently the C => HTML converter on the web site doesn't quite like it
otherwise.
Version 7.61.0 (11 Jul 2018)

- release: 7.61.0
- TODO: Configurable loading of OpenSSL configuration file
Closes #2724
- post303.d: clarify that this is an RFC violation
... and not the other way around, which this previously said.
Reported-by: Vasiliy Faronov
Fixes #2723
Closes #2726
CMake: remove redundant and old end-of-block syntax
Reviewed-by: Jakub Zakrzewski

Closes #2715
Jay Satiro (9 Jul 2018)

- lib/curl_setup.h: remove unicode character
Follow-up to 82ce416.
Ref: https://github.com/curl/curl/commit/8272ec5#commitcomment-29646818

- lib/curl_setup.h: remove unicode bom from 8272ec50f02

- schannel: fix -Wsign-compare warning
MinGW warns:
/lib/vtls/schannel.c:219:64: warning: signed and unsigned type in
conditional expression [-Wsign-compare]
Fix this by casting the ptrdiff_t to size_t as we know it's positive.
- schannel: workaround for wrong function signature in w32api
Original MinGW's w32api has CryptHashData's second parameter as BYTE *

instead of const BYTE *.
- schannel: make more cipher options conditional
They are not defined in the original MinGW's <wincrypt.h>.
- curl_setup: include <winerror.h> before <windows.h>
Otherwise, only part of it gets pulled in through <windows.h> on

original MinGW.
- examples: fix -Wformat warnings
When size_t is not a typedef for unsigned long (as usually the case on
Windows), GCC emits -Wformat warnings when using lu and lx format
specifiers with size_t. Silence them with explicit casts to
unsigned long.

- smtp: use the upload buffer size for scratch buffer malloc
... not the read buffer size, as that can be set smaller and thus cause
a buffer overflow! CVE-2018-0500
Reported-by: Peter Wu
Bug: https://curl.haxx.se/docs/adv_2018-70a2.html
scripts: include _curl as part of CLEANFILES
Closes #2718
- [Nick Zitzmann brought this change]
darwinssl: allow High Sierra users to build the code using GCC
...but GCC users lose out on TLS 1.3 support, since we can't weak-link
enumeration constants.
Fixes #2656
Closes #2703
CMake: Remove unused 'output_var' from 'collect_true'
Variable 'output_var' is not used and can be removed.

Function 'collect_true' renamed to 'count_true'.
CMake: Remove unused functions
Closes #2711
- KNOWN_BUGS: Stick to same family over SOCKS proxy
- libssh: goto DISCONNECT state on error, not SSH_SESSION_FREE
... because otherwise not everything get closed down correctly.
Fixes #2708
Closes #2712
- libssh: include line number in state change debug messages
Closes #2713
- KNOWN_BUGS: Borland support is dropped, AIX problem is too old
- [Jeroen Ooms brought this change]
example/crawler.c: simple crawler based on libxml2

Closes #2706
- DEPRECATE: include year when specifying date
- DEPRECATE: linkified
- DEPRECATE: mention the PR that disabled axTLS
- docs/DEPRECATE.md: spelling and minor formatting
- DEPRECATE: new doc describing planned item removals
Closes #2704
telnet: fix clang warnings
telnet.c(1401,28): warning: cast from function call of type 'int' to

non-matching type 'HANDLE' (aka 'void *') [-Wbad-function-cast]
Fixes #2696
Closes #2700
- docs: fix missed option name markups
- [Gaurav Malhotra brought this change]
openssl: Remove some dead code
Closes #2698
- openssl: make the requested TLS version the *minimum* wanted
The code treated the set version as the *exact* version to require in
the TLS handshake, which is not what other TLS backends do and probably
not what most people expect either.
Reported-by: Andreas Olsson

Assisted-by: Gaurav Malhotra
Fixes #2691
Closes #2694
- openssl: allow TLS 1.3 by default
Reported-by: Andreas Olsson

Fixes #2692
Closes #2693
- [Adrian Peniak brought this change]
CURLINFO_TLS_SSL_PTR.3: improve the example
The previous example was a little bit confusing, because SSL* structure
(or other "in use" SSL connection pointer) is not accessible after the
transfer is completed, therefore working with the raw TLS library
specific pointer needs to be done during transfer.
Closes #2690
- travis: add a build using the synchronous name resolver
... since default uses the threaded one and we test the c-ares build
already.
Closes #2689
- configure: remove CURL_CHECK_NI_WITHSCOPEID too
Since it isn't used either and requires the getnameinfo check
Follow-up to 0aeca41702d2
- getnameinfo: not used
Closes #2687
- easy_perform: use *multi_timeout() to get wait times
... and trim the threaded Curl_resolver_getsock() to return zero

millisecond wait times during the first three milliseconds so that
localhost or names in the OS resolver cache gets detected and used
faster.
Closes #2685
Max Dymond (27 Jun 2018)

- configure: Add dependent libraries after crypto
The linker is pretty dumb and processes things left to right, keeping a
tally of symbols it hasn't resolved yet. So, we need -ldl to appear
after -lcrypto otherwise the linker won't find the dl functions.
Closes #2684
Daniel Stenberg (27 Jun 2018)

- GOVERNANCE: linkify, changed some titles
- GOVERNANCE: add maintainer details/duties
- url: check Curl_conncache_add_conn return code
... it was previously unchecked in two places and thus errors could
remain undetected and cause trouble.
Closes #2681
- include/README: remove "hacking" advice, not the right place
- CURLOPT_SSL_VERIFYPEER.3: fix syntax mistake

Follow-up to b6a16afa0aa5
- netrc: use a larger buffer
... to work with longer passwords etc. Grow it from a 256 to a 4096
bytes buffer.
Reported-by: Dario Nieuwenhuis

Fixes #2676
Closes #2680
- [Patrick Schlangen brought this change]
CURLOPT_SSL_VERIFYPEER.3: Add performance note
Closes #2673
- [Javier Blazquez brought this change]
multi: fix crash due to dangling entry in connect-pending list
Fixes #2677
Closes #2679
- ConnectionExists: make sure conn->data is set when "taking" a connection
Follow-up to 2c15693.
Bug #2674
Closes #2675
- [Kevin R. Bulgrien brought this change]
system.h: fix for gcc on 32 bit OpenServer
- [Raphael Gozzo brought this change]
cmake: allow multiple SSL backends
This will make possible to select the SSL backend (using

curl_global_sslset()) even when the libcurl is built using CMake
Closes #2665
- url: fix dangling conn->data pointer
By masking sure to use the *current* easy handle with extracted

connections from the cache, and make sure to NULLify the ->data pointer
when the connection is put into the cache to make this mistake easier to
detect in the future.
Reported-by: Will Dietz

Fixes #2669
Closes #2672
- CURLOPT_INTERFACE.3: interface names not supported on Windows

- travis: run more tests for coverage check
... run a few more tortured based and run all tests event-based.
Closes #2664
- multi: fix memory leak when stopped during name resolve
When the application just started the transfer and then stops it while
the name resolve in the background thread hasn't completed, we need to
wait for the resolve to complete and then cleanup data accordingly.
Enabled test 1553 again and added test 1590 to also check when the host
name resolves successfully.
Detected by OSS-fuzz.
Closes #1968
Viktor Szakats (15 Jun 2018)

- maketgz: delete .bak files, fix indentation

- runtests.pl: remove debug leftover from bb9a340c73f3
- curl-confopts.m4: fix typo from ed224f23d5beb
Fixes my local configure to detect a custom installed c-ares without

pkgconfig.
- docs/RELEASE-PROCEDURE.md: renamed to use .md extension
Closes #2663
- RELEASE-PROCEDURE: gpg sign the tags
- CURLOPT_HTTPAUTH.3: CURLAUTH_BEARER was added in 7.61.0
- [Mamta Upadhyay brought this change]
maketgz: fix sed issues on OSX
maketgz creates release tarballs and removes the -DEV string in curl
version (e.g. 7.58.0-DEV), else -DEV shows up on command line when curl
is run. maketgz works fine on linux but fails on OSX. Problem is with
the sed commands that use option -i without an extension. Maketgz
expects GNU sed instead of BSD and this simply won't work on OSX. Adding
a backup extension .bak after -i fixes this issue
Running the script as if on OSX gives this error:
sed: -e: No such file or directory
Adding a .bak extension resolves it

Closes #2660
- configure: enhance ability to detect/build with static openssl
Fix the -ldl and -ldl + -lpthread checks for OpenSSL, necessary for
building with static libs without pkg-config.

Fixes #2199
Closes #2659
- configure: use pkg-config for c-ares detection
First check if there's c-ares information given as pkg-config info and use
that as first preference.
Reported-by: pszemus on github

Fixes #2203
Closes #2658
- GOVERNANCE.md: explains how this project is run
Closes #2657
- KNOWN_BUGS: NTLM doen't support password with § character
Closes #2120
- KNOWN_BUGS: slow connect to localhost on Windows
Closes #2281
- [Matteo Bignotti brought this change]
mk-ca-bundle.pl: make -u delete certdata.txt if found not changed
certdata.txt should be deleted also when the process is interrupted by

"same certificate downloaded, exiting"
The certdata.txt is currently kept on disk even if you give the -u

option
Closes #2655
- progress: remove a set of unused defines
Reported-by: Peter Wu
Closes #2654
- TODO: "Option to refuse usernames in URLs" done
Implemented by Björn in 946ce5b61f
- [Lyman Epp brought this change]
Curl_init_do: handle NULL connection pointer passed in
Closes #2653
- runtests: support variables in <strippart>
... and make use of that to make 1455 work better without using a fixed
local port number.
Fixes #2649
Closes #2650
- Curl_debug: remove dead printhost code
The struct field is never set (since 5e0d9aea3) so remove the use of it
and remove the connectdata pointer from the prototype.
Reported-by: Tejas
Closes #2647

- schannel: avoid incompatible pointer warning
with clang-6.0:
```
vtls/schannel_verify.c: In function 'add_certs_to_store':
vtls/schannel_verify.c:212:30: warning: passing argument 11 of 'CryptQueryObject'
from incompatible pointer type [-Wincompatible-pointer-types]
&cert_context)) {
^
In file included from /usr/share/mingw-w64/include/schannel.h:10:0,
from /usr/share/mingw-w64/include/schnlsp.h:9,
from vtls/schannel.h:29,
from vtls/schannel_verify.c:40:
/usr/share/mingw-w64/include/wincrypt.h:4437:26: note: expected 'const void **'
but argument is of type 'CERT_CONTEXT ** {aka struct _CERT_CONTEXT **}'
WINIMPM WINBOOL WINAPI CryptQueryObject (DWORD dwObjectType, const void
*pvObject, DWORD dwExpectedContentTypeFlags, DWORD dwExpectedFormatTypeFlags, DWORD
dwFlags,
^~~~~~~~~~~~~~~~
```
Ref: https://msdn.microsoft.com/library/windows/desktop/aa380264

- [Robert Prag brought this change]
schannel: support selecting ciphers
Given the contstraints of SChannel, I'm exposing these as the algorithms

themselves instead; while replicating the ciphersuite as specified by
OpenSSL would have been preferable, I found no way in the SChannel API
to do so.
To use this from the commandline, you need to pass the names of contants
defining the desired algorithms. For example, curl --ciphers
"CALG_SHA1:CALG_RSA_SIGN:CALG_RSA_KEYX:CALG_AES_128:CALG_DH_EPHEM"
https://github.com The specific names come from wincrypt.h
Closes #2630
- [Bernhard M. Wiedemann brought this change]
test 46: make test pass after 2025
shifting the expiry date to 2037 for now

to be before the possibly problematic year 2038
similar in spirit to commit e6293cf8764e9eecb
Closes #2646
- [Marian Klymov brought this change]
cppcheck: fix warnings
- Get rid of variable that was generating false positive warning

(unitialized)
- Fix issues in tests
- Reduce scope of several variables all over
etc
Closes #2631
- openssl: assume engine support in 1.0.1 or later
Previously it was checked for in configure/cmake, but that would then

leave other build systems built without engine support.
While engine support probably existed prior to 1.0.1, I decided to play

safe. If someone experience a problem with this, we can widen the
version check.
Fixes #2641
Closes #2644
- RELEASE-PROCEDURE: update the release calendar for 2019
boringssl + schannel: undef X509_NAME in lib/schannel.h
Fixes the build problem when both boringssl and schannel are enabled.
Fixes #2634
Closes #2643
- [Vladimir Kotal brought this change]
mk-ca-bundle.pl: leave certificate name untouched in decode()
Closes #2640

tests/libtests/Makefile.am: Add lib1521.c to CLEANFILES
This removes the generated lib1521.c when running make clean.
Closes #2633
tests/libtest: Add lib1521 to nodist_SOURCES
Since 467da3af0, lib1521.c is generated instead of checked in. According

to the commit message, the intention was to remove it from the tarball
as well. However, it is still present when running make dist. To remove
it, add it to nodist_lib1521_SOURCES. This also means there is no need
for the manually added dist-rule in the Makefile.
Also update CMakelists.txt to handle the fact that we now may have
nodist_SOURCES.
- [Stephan Mühlstrasser brought this change]
system.h: add support for IBM xlc C compiler
Added a section to system.h guarded with __xlc__ for the IBM xml C
compiler. Before this change the section titled 'generic "safe guess" on
old 32 bit style' was used, which resulted in a wrong definition of
CURL_TYPEOF_CURL_SOCKLEN_T, and for 64-bit also CURL_TYPEOF_CURL_OFF_T
was wrong.
Compilation warnings fixed with this change:
CC libcurl_la-ftp.lo
"ftp.c", line 290.55: 1506-280 (W) Function argument assignment between types
"unsigned long* restrict" and "int*" is not allowed.
CC libcurl_la-connect.lo
"connect.c", line 448.56: 1506-280 (W) Function argument assignment between types
CC libcurl_la-tftp.lo
"tftp.c", line 1115.33: 1506-280 (W) Function argument assignment between types
Closes #2637
- cmdline-opts/cert-type.d: mention "p12" as a recognized type as well

- spelling fixes
Detected using the `codespell` tool (version 1.13.0).
Also secure and fix an URL.

- axtls: follow-up spell fix of comment
- axTLS: not considered fit for use
This is step one. It adds #error statements that require source edits to
make curl build again if asked to use axTLS. At a later stage we might
remove the axTLS specific code completely.
Closes #2628
- build: remove the Borland specific makefiles
According to the user survey 2018, not even one out of 670 users use
them. Nobody on the mailing list spoke up for them either.
Closes #2629
- curl_addrinfo: use same #ifdef conditions in source as header
... for curl_dofreeaddrinfo
- multi: remove a DEBUGF()
... it might call infof() with a NULL first argument that isn't harmful
but makes it not do anything. The infof() line is not very useful
anymore, it has served it purpose. Good riddance!
Fixes #2627
- [Alibek.Jorajev brought this change]
CURLOPT_RESOLVE: always purge old entry first
If there's an existing entry using the selected name.
Closes #2622
- fnmatch: use the system one if available
If configure detects fnmatch to be available, use that instead of our

custom one for FTP wildcard pattern matching. For standard compliance,
to reduce our footprint and to use already well tested and well
exercised code.
A POSIX fnmatch behaves slightly different than the internal function

for a few test patterns currently and the macOS one yet slightly
different. Test case 1307 is adjusted for these differences.
Closes #2626
Patrick Monnerat (31 May 2018)

- os400: add new option in ILE/RPG binding
Follow-up to commit 946ce5b
Daniel Stenberg (31 May 2018)

- tests/libtest/.gitignore: follow-up fix to ignore lib5* too
- KNOWN_BUGS: CURL_GLOBAL_SSL
Closes #2276
- [Bernhard Walle brought this change]
configure: check for declaration of getpwuid_r
On our x86 Android toolchain, getpwuid_r is implemented but the header

is missing:
netrc.c:81:7: error: implicit declaration of function 'getpwuid_r' [-

Werror=implicit-function-declaration]
Unfortunately, the function is used in curl_ntlm_wb.c, too, so I moved

the prototype to curl_setup.h.
Signed-off-by: Bernhard Walle <bernhard@bwalle.de>

Closes #2609
tests: update .gitignore for libtests
Closes #2624
strictness: correct {infof, failf} format specifiers
Closes #2623
- [Björn Stenberg brought this change]
option: disallow username in URL
Adds CURLOPT_DISALLOW_USERNAME_IN_URL and --disallow-username-in-url. Makes

libcurl reject URLs with a username in them.
Closes #2340
- libcurl-security.3: improved layout for two rememdy lists
- libcurl-security.3: refer to URL instead of in-source markdown file
Viktor Szakats (30 May 2018)

- curl.rc: embed manifest for correct Windows version detection
* enable it in `src/Makefile.m32`
* enable it in `winbuild/MakefileBuild.vc` if a custom manifest is
_not_ enabled via the existing `EMBED_MANIFEST` option
* enable it for all Windows CMake builds (also disable the built-in
minimal manifest, added by CMake by default.)
For other build systems, add the `-DCURL_EMBED_MANIFEST` option to

the list of RC (Resource Compiler) flags to enable the manifest
included in `src/curl.rc`. This may require to disable whatever
automatic or other means in which way another manifest is added to
`curl.exe`.
Notice that Borland C doesn't support this method due to a

long-pending resource compiler bug. Watcom C may also not handle
it correctly when the `-zm` `wrc` option is used (this option may
be unnecessary though) and regardless of options in certain earlier
revisions of the 2.0 beta version.

- os400: sync EBCDIC wrappers and ILE/RPG binding with latest options
- os400: implement mime api EBCDIC wrappers
Also sync ILE/RPG binding to define the new functions.

- setopt: add TLS 1.3 ciphersuites
Adds CURLOPT_TLS13_CIPHERS and CURLOPT_PROXY_TLS13_CIPHERS.
curl: added --tls13-ciphers and --proxy-tls13-ciphers
Fixes #2435
Reported-by: zzq1015 on github
Closes #2607
- configure: override AR_FLAGS to silence warning
The automake default ar flags are 'cru', but the 'u' flag in there
causes warnings on many modern Linux distros. Removing 'u' may have a
minor performance impact on older distros but should not cause harm.
Explained on the automake mailing list already back in April 2015:
https://www.mail-archive.com/automake-patches@gnu.org/msg07705.html
Reported-by: elephoenix on github

Fixes #2617
Closes #2619
Sergei Nikulov (29 May 2018)

- cmake: fixed comments in compile checks code

- INSTALL: LDFLAGS=-Wl,-R/usr/local/ssl/lib
... the older description doesn't work

Reported-by: Peter Varga
Fixes #2615
Closes #2616
- [Will Dietz brought this change]
KNOWN_BUGS: restore text regarding #2101.
This was added earlier but appears to have been removed accidentally.
AFAICT this is very much still an issue.
-----
I say "accidentally" because the text seems to have harmlessly snuck

into [1] (which makes no mention of it). [1] was later reverted for
unspecified reasons in [2], presumably because the mentioned issue was
fixed or invalid.
[1] de9fac00c40db321d44fa6fbab6eb62ec4c83998
[2] 16d1f369403cbb04bd7b085eabbeebf159473fc2
Closes #2618
- fnmatch: insist on escaped bracket to match
A non-escaped bracket ([) is for a character group - as documented. It

will *not* match an individual bracket anymore. Test case 1307 updated
accordingly to match.
Problem detected by OSS-Fuzz, although this fix is probably not a final

fix for the notorious timeout issues.
Closes #2614

- psl: use latest psl and refresh it periodically
The latest psl is cached in the multi or share handle. It is refreshed

before use after 72 hours.
New share lock CURL_LOCK_DATA_PSL controls the psl cache sharing.
If the latest psl is not available, the builtin psl is used.
Reported-by: Yaakov Selkowitz

Fixes #2553
Closes #2601

- [Fabrice Fontaine brought this change]
configure: fix ssh2 linking when built with a static mbedtls
The ssh2 pkg-config file could contain the following lines when build
with a static version of mbedtls:
Libs: -L${libdir} -lssh2 /xxx/libmbedcrypto.a
Libs.private: /xxx/libmbedcrypto.a
This static mbedtls library must be used to correctly detect ssh2
support and this library must be copied in libcurl.pc otherwise
compilation of any application (such as upmpdcli) with libcurl will fail
when trying to found mbedtls functions included in libssh2. So, replace
pkg-config --libs-only-l by pkg-config --libs.
Fixes:
-
http://autobuild.buildroot.net/results/43e24b22a77f616d6198c10435dcc23cc3b9088a
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>

Closes #2613
- [Bernhard Walle brought this change]
cmake: check for getpwuid_r
The autotools-based build system does it, so we do it also in CMake.
Bug: #2609
Signed-off-by: Bernhard Walle <bernhard@bwalle.de>
- cmdline-opts/gen.pl: warn if mutexes: or see-also: list non-existing options
curl.1: Fix cmdline-opts reference errors.
--data, --form, and --ntlm were declared to be mutually exclusive with

non-existing options. --data and --form referred to --upload (which is
short for --upload-file and therefore did work, so this one was merely
a bit confusing), --ntlm referred to --negotiated instead of --negotiate.
Closes #2612
docs: fix cmdline-opts metadata headers case consistency.
Almost all headers start with an uppercase letter, but some didn't.
- mailmap: Max Savenkov

- [Max Savenkov brought this change]
Fix the test for fsetxattr and strerror_r tests in CMake to work without
compiling

- mailmap: a Richard Alcock fixup
- [Richard Alcock brought this change]
schannel: add failf calls for client certificate failures
Closes #2604
winbuild: In MakefileBuild.vc fix typo DISTDIR->DIRDIST
Change requirement from $(DISTDIR) to $(DIRDIST)
closes #2603
winbuild: only delete OUTFILE if it exists
This removes the slightly annoying "Could not file LIBCURL_OBJS.inc" and
"Could not find CURL_OBJS.inc.inc" message when building into a clean
folder.
closes #2602
- [Alejandro R. Sedeño brought this change]
content_encoding: handle zlib versions too old for Z_BLOCK
Fallback on Z_SYNC_FLUSH when Z_BLOCK is not available.
Fixes #2606
Closes #2608
- multi: provide a socket to wait for in Curl_protocol_getsock
... even when there's no protocol specific handler setup.
Reported-by: Sean Miller
Closes #2600
- [Linus Lewandowski brought this change]
httpauth: add support for Bearer tokens
Closes #2102
- TODO: CURLINFO_PAUSE_STATE
Closes #2588

- cmake: set -d postfix for debug builds if not specified
using -DCMAKE_DEBUG_POSTFIX explicitly
fixes #2121, obsoletes #2384

- configure: add basic test of --with-ssl prefix
When given a prefix, the $PREFIX_OPENSSL/lib/openssl.pc or

$PREFIX_OPENSSL/include/openssl/ssl.h files must be present or cause an
error. Helps users detect when giving configure the wrong path.
Reported-by: Oleg Pudeyev
Assisted-by: Per Malmberg
Fixes #2580

- http resume: skip body if http code 416 (range error) is ignored.
This avoids appending error data to already existing good data.
Test 92 is updated to match this change.

New test 1156 checks all combinations of --range/--resume, --fail,
Content-Range header and http status code 200/416.
Fixes #1163
Reported-By: Ithubg on github
Closes #2578

- tftp: make sure error is zero terminated before printfing it
- configure: add missing m4/ax_compile_check_sizeof.m4
follow-up to mistake in 6876ccf90b4
Jay Satiro (22 May 2018)

schannel: make CAinfo parsing resilient to CR/LF
OpenSSL has supported --cacert for ages, always accepting LF-only line
endings ("Unix line endings") as well as CR/LF line endings ("Windows
line endings").
When we introduced support for --cacert also with Secure Channel (or in
cURL speak: "WinSSL"), we did not take care to support CR/LF line
endings, too, even if we are much more likely to receive input in that
form when using Windows.
Let's fix that.
Happily, CryptQueryObject(), the function we use to parse the ca-bundle,

accepts CR/LF input already, and the trailing LF before the END
CERTIFICATE marker catches naturally any CR/LF line ending, too. So all
we need to care about is the BEGIN CERTIFICATE marker. We do not
actually need to verify here that the line ending is CR/LF. Just
checking for a CR or an LF is really plenty enough.

- CURLOPT_ACCEPT_ENCODING.3: add brotli and clarify a bit
- KNOWN_BUGS: mention the -O with %-encoded file names
Closes #2573
- checksrc: make sure sizeof() is used *with* parentheses
... and unify the source code to adhere.
Closes #2563
- curl: added --styled-output
It is enabled by default, so --no-styled-output will switch off the

detection/use of bold headers.
Closes #2538
- curl: show headers in bold
The feature is only enabled if the output is believed to be a tty.
-J: There's some minor differences and improvements in -J handling, as

now J should work with -i and it actually creates a file first using the
initial name and then *renames* that to the one found in
Content-Disposition (if any).
-i: only shows headers for HTTP transfers now (as documented).
Previously it would also show for pieces of the transfer that were HTTP
(for example when doing FTP over a HTTP proxy).
-i: now shows trailers as well. Previously they were not shown at all.
--libcurl: the CURLOPT_HEADER is no longer set, as the header output is

now done in the header callback.
- configure: compile-time SIZEOF checks
... instead of exeucting code to get the size. Removes the use of
LD_LIBRARY_PATH for this.
Fixes #2586
Closes #2589
Reported-by: Bernhard Walle
- configure: replace AC_TRY_RUN with CURL_RUN_IFELSE
... and export LD_LIBRARY_PATH properly. This is a follow-up from

2d4c215.
Fixes #2586
Reported-by: Bernhard Walle
- docs: clarify CURLOPT_HTTPGET somewhat
Reported-by: bsammon on github

Fixes #2590
- curl_fnmatch: only allow two asterisks for matching
The previous limit of 5 can still end up in situation that takes a very
long time and consumes a lot of CPU.
If there is still a rare use case for this, a user can provide their own
fnmatch callback for a version that allows a larger set of wildcards.
This commit was triggered by yet another OSS-Fuzz timeout due to this.
Closes #2587
- checksrc: fix too long line
follow-up to e05ad5d
- [Aleks brought this change]
docs: mention HAproxy protocol "version 1"
...as there's also a version 2.
Closes #2579
- examples/progressfunc: make it build on older libcurls
This example was changed in ce2140a8c1 to use the new microsecond based
getinfo option. This change makes it conditionally keep using the older
option so that the example still builds with older libcurl versions.
Closes #2584
- stub_gssapi: fix numerous 'unused parameter' warnings
follow-up to d9e92fd9fd1d
- [Philip Prindeville brought this change]
getinfo: add microsecond precise timers for various intervals
Provide a set of new timers that return the time intervals using integer
number of microseconds instead of floats.
The new info names are as following:
CURLINFO_APPCONNECT_TIME_T
CURLINFO_CONNECT_TIME_T
CURLINFO_NAMELOOKUP_TIME_T
CURLINFO_PRETRANSFER_TIME_T
CURLINFO_REDIRECT_TIME_T
CURLINFO_STARTTRANSFER_TIME_T
CURLINFO_TOTAL_TIME_T
Closes #2495
- openssl: acknowledge --tls-max for default version too
... previously it only used the max setting if a TLS version was also
explicitly asked for.
Reported-by: byte_bucket
Fixes #2571
Closes #2572
- bump: start working on the pending 7.61.0
- [Dagobert Michelsen brought this change]
tests/libtest/Makefile: Do not unconditionally add gcc-specific flags
The warning flag leads e.g. Sun Studio compiler to bail out.
Closes #2576
- schannel_verify: fix build for non-schannel

- rand: fix typo
- schannel: disable manual verify if APIs not available
.. because original MinGW and old compilers do not have the Windows API
definitions needed to support manual verification.
schannel: disable client cert option if APIs not available
Original MinGW targets Windows 2000 by default, which lacks some APIs and
definitions for this feature. Disable it if these APIs are not available.
Version 7.60.0 (15 May 2018)

- RELEASE-NOTES: 7.60.0 release
- THANKS: added people from the curl 7.60.0 release
- docs/libcurl/index.html: removed
The HTML files are long gone from the dist, now remove the last HTML
file pointing to those missing files.
- [steini2000 brought this change]
http2: remove unused variable
Closes #2570
- [steini2000 brought this change]
http2: use easy handle of stream for logging
- gcc: disable picky gcc-8 function pointer warnings in two places
Reported-by: Rikard Falkeborn

Bug: #2560
Closes #2569
- http2: use the correct function pointer typedef
Fixes gcc-8 picky compiler warnings

Bug: #2560
Closes #2568
- CODE_STYLE: mention return w/o parens, but sizeof with
... and remove the github markdown syntax so that it renders better on
the web site. Also, don't use back-ticks inlined to allow the CSS to
highlight source code better.
examples: Fix format specifiers
Closes #2561
tool: Fix format specifiers
ntlm: Fix format specifiers
tests: Fix format specifiers
lib: Fix format specifiers
- contributors.sh: use "on github", not at
- http2: getsock fix for uploads
When there's an upload in progress, make sure to wait for the socket to
become writable.
Detected-by: steini2000 on github

Bug: #2520
Closes #2567
- pingpong: fix response cache memcpy overflow
Response data for a handle with a large buffer might be cached and then
used with the "closure" handle when it has a smaller buffer and then the
larger cache will be copied and overflow the new smaller heap based
buffer.
Reported-by: Dario Weisser

CVE: CVE-2018-1000300
Bug: https://curl.haxx.se/docs/adv_2018-82c2.html
- http: restore buffer pointer when bad response-line is parsed

... leaving the k->str could lead to buffer over-reads later on.
CVE: CVE-2018-1000301
Assisted-by: Max Dymond
Detected by OSS-Fuzz.
Bug: https://curl.haxx.se/docs/adv_2018-b138.html

- cookies: do not take cookie name as a parameter
RFC 6265 section 4.2.1 does not set restrictions on cookie names.
This is a follow-up to commit 7f7fcd0.
Also explicitly check proper syntax of cookie name/value pair.
New test 1155 checks that cookie names are not reserved words.
Reported-By: anshnd at github

Fixes #2564
Closes #2566

- smb: reject negative file sizes
- setup_transfer: deal with both sockets being -1
Detected by Coverity; CID 1435559. Follow-up to f8d608f38d00. It would

index the array with -1 if neither index was a socket.
- travis: add build using NSS
Closes #2558
- [Sunny Purushe brought this change]
openssl: change FILE ops to BIO ops
To make builds with VS2015 work. Recent changes in VS2015 _IOB_ENTRIES

handling is causing problems. This fix changes the OpenSSL backend code
to use BIO functions instead of FILE I/O functions to circumvent those
problems.
Closes #2512
- travis: add a build using WolfSSL
Assisted-by: Dan Fandrich
Closes #2528
- RELEASE-NOTES: typo
URLs: fix one more http url
This file wasn't included in commit 4af40b3646d3b09 which updated all

haxx.se http urls to https. The file was committed prior to that update,
but may have been merged after it and hence didn't get updated.
Closes #2550
- github/lock: auto-lock closed issues after 90 days of inactivity
- vtls: fix missing commas
follow-up to e66cca046cef
- vtls: use unified "supports" bitfield member in backends
... instead of previous separate struct fields, to make it easier to

extend and change individual backends without having to modify them all.
closes #2547
- transfer: don't unset writesockfd on setup of multiplexed conns
Curl_setup_transfer() can be called to setup a new individual transfer

over a multiplexed connection so it shouldn't unset writesockfd.
Bug: #2520
Closes #2549
configure: put CURLDEBUG and DEBUGBUILD in lib/curl_config.h
They are removed from the compiler flags.
This ensures that make dependency tracking will force a rebuild whenever
configure --enable-debug or --enable-curldebug changes.
Closes #2548
- http: don't set the "rewind" flag when not uploading anything
It triggers an assert.
Closes #2546
- travis: add an mbedtls build
Closes #2531
- configure: only check for CA bundle for file-using SSL backends
When only building with SSL backends that don't use the CA bundle file
(by default), skip the check.
Fixes #2543
Fixes #2180
Closes #2545
- ssh-libssh.c: fix left shift compiler warning
ssh-libssh.c:2429:21: warning: result of '1 << 31' requires 33 bits to

represent, but 'int' only has 32 bits [-Wshift-overflow=]
'len' will never be that big anyway so I converted the run-time check to
a regular assert.
URL: fix ASCII dependency in strcpy_url and strlen_url
Commit 3c630f9b0af097663a64e5c875c580aa9808a92b partially reverted the

changes from commit dd7521bcc1b7a6fcb53c31f9bd1192fcc884bd56 because of
the problem that strcpy_url() was modified unilaterally without also
modifying strlen_url(). As a consequence strcpy_url() was again
depending on ASCII encoding.
This change fixes strlen_url() and strcpy_url() in parallel to use a

common host-encoding independent criterion for deciding whether an URL
character must be %-escaped.
Closes #2535
- [Denis Ollier brought this change]
docs: remove extraneous commas in man pages
Closes #2544
- Revert "TODO: remove configure --disable-pthreads"
This reverts commit d5d683a97f9765bddfd964fe32e137aa6e703ed3.
--disable-pthreads can be used to disable pthreads and get the threaded

resolver to use the windows threading when building with mingw.
- vtls: don't define MD5_DIGEST_LENGTH for wolfssl
... as it defines it (too)
- TODO: remove configure --disable-pthreads

- [David Garske brought this change]
wolfssl: Fix non-blocking connect
Daniel Stenberg (30 Apr 2018)

- CURLOPT_URL.3: add ENCODING section [ci skip]
Feedback-by: Michael Kilburn
- KNOWN_BUGS: Client cert with Issuer DN differs between backends
Closes #1411
- KNOWN_BUGS: Passive transfer tries only one IP address
Closes #1508
- KNOWN_BUGS: --upload-file . hang if delay in STDIN
Closes #2051
- KNOWN_BUGS: Connection information when using TCP Fast Open
Closes #1332
- travis: enable libssh2 on both macos and Linux
It seems to not be detected by default anymore (which is a bug I

believe)
Closes #2541
- TODO: Support the clienthello extension
Closes #2299
- TODO: CLOEXEC
Closes #2252
- tests: provide 'manual' as a feature to optionally require
... and make test 1026 rely on that feature so that --disable-manual
builds don't cause test failures.
Reported-by: Max Dymond and Anders Roxell

Fixes #2533
Closes #2540
- CURLINFO_PROTOCOL.3: mention the existing defined names
Jay Satiro (27 Apr 2018)

cookies: remove unused macro
Commit 2bc230de63 made the macro MAX_COOKIE_LINE_TXT become unused,

so remove as it's not part of the published API.

checksrc: force indentation of lines after an else
This extends the INDENTATION case to also handle 'else' statements

and require proper indentation on the following line. Also fixes the
offending cases found in the codebase.
Closes #2532
- http2: fix null pointer dereference in http2_connisdead
This function can get called on a connection that isn't setup enough to
have the 'recv_underlying' function pointer initialized so it would try
to call the NULL pointer.
Follow-up to db1b2c7fe9b093f8 (never shipped in a release)

Closes #2536
- http2: get rid of another strstr()
Follow-up to 1514c44655e12e: replace another strstr() call done on a

buffer that might not be zero terminated - with a memchr() call, even if
we know the substring will be found.
Closes #2534
- cyassl: adapt to libraries without TLS 1.0 support built-in
WolfSSL doesn't enable it by default anymore
- configure: provide --with-wolfssl as an alias for --with-cyassl
os400.c: fix ASSIGNWITHINCONDITION checksrc warnings
All occurrences of assignment within conditional expression in

os400sys.c rewritten into two steps: first assignment and then the check
on the success of the assignment. Also adjust related incorrect brace
positions to match project indentation style.
This was spurred by seeing "if((inp = input_token))", but while in there

all warnings were fixed.
There should be no functional change from these changes.
Closes #2525
cookies: ensure that we have cookies before writing jar

The jar should be written iff there are cookies, so ensure that we still
have cookies after expiration to avoid creating an empty file.
Closes #2529
- strcpy_url: only %-encode values >= 0x80
OSS-Fuzz detected
Broke in dd7521bcc1b7
- mime: avoid NULL pointer dereference risk
Coverity detected, CID 1435120
Closes #2527
ctype: restore character classification for non-ASCII platforms
With commit 4272a0b0fc49a1ac0ceab5c4a365c9f6ab8bf8e2 curl-speficic

character classification macros and functions were introduced in
curl_ctype.[ch] to avoid dependencies on the locale. This broke curl on
non-ASCII, e.g. EBCDIC platforms. This change restores the previous set
of character classification macros when CURL_DOES_CONVERSIONS is
defined.
Closes #2494
- ftplistparser: keep state between invokes
Fixes FTP wildcard parsing when done over a number of read buffers.
Regression from f786d1f14
Reported-by: wncboy on github

Fixes #2445
Closes #2526
- examples/http2-upload: expand buffer to avoid silly warning
http2-upload.c:135:44: error: ‘%02d’ directive output may be truncated

writing between 2 and 11 bytes into a region of size between 8 and 17
- examples/sftpuploadresume: typecast fseek argument to long
/docs/examples/sftpuploadresume.c:102:12: warning: conversion to 'long

int' from 'curl_off_t {aka long long int}' may alter its value
- Revert "ftplistparser: keep state between invokes"
This reverts commit abbc8457d85aca74b7cfda1d394b0844932b2934.
Caused fuzzer problems on travis not seen when this was a PR!
- Curl_memchr: zero length input can't match
Avoids undefined behavior.
Reported-by: Geeknik Labs
Fixes FTP wildcard parsing when doing over a number of read buffers.

Fixes #2445
Closes #2519
- ftplistparser: renamed some members and variables
... to make them better spell out what they're for.
- [Christian Schmitz brought this change]
curl_global_sslset: always provide available backends
Closes #2499
- http2: convert an assert to run-time check
Fuzzing has proven we can reach code in on_frame_recv with status_code

not having been set, so let's detect that in run-time (instead of with
assert) and error error accordingly.
(This should no longer happen with the latest nghttp2)
Closes #2514
- curl.1: clarify that options and URLs can be mixed
Fixes #2515
Closes #2517

CURLOPT_SSLCERT.3: improve WinSSL-specific usage info
schannel: fix build error on targets <= XP
- Use CRYPT_STRING_HEX instead of CRYPT_STRING_HEXRAW since XP doesn't

support the latter.

- Revert "ftplistparser: keep state between invokes"
This reverts commit 8fb78f9ddc6d858d630600059b8ad84a80892fd9.
Unfortunately this fix introduces memory leaks I've not been able to fix
in several days. Reverting this for now to get the leaks fixed.

- tool_help: clarify --max-time unit of time is seconds
Before:
-m, --max-time <time> Maximum time allowed for the transfer
After:
-m, --max-time <seconds> Maximum time allowed for the transfer

- http2: handle GOAWAY properly
When receiving REFUSED_STREAM, mark the connection for close and retry
streams accordingly on another/fresh connection.
Reported-by: Terry Wu
Fixes #2416
Fixes #1618
Closes #2510
- http2: clear the "drain counter" when a stream is closed
This fixes the notorious "httpc->drain_total >= data->state.drain"

assert.
Reported-by: Anders Bakken
Fixes #1680
Closes #2509
- http2: avoid strstr() on data not zero terminated
It's not strictly clear if the API contract allows us to call strstr()
on a string that isn't zero terminated even when we know it will find
the substring, and clang's ASAN check dislikes us for it.
Also added a check of the return code in case it fails, even if I can't
think of a situation how that can trigger.
Closes #2513

openssl: fix subjectAltName check on non-ASCII platforms
Curl_cert_hostcheck operates with the host character set, therefore the

ASCII subjectAltName string retrieved with OpenSSL must be converted to
the host encoding before comparison.
Closes #2493

- openssl: Add support for OpenSSL 1.1.1 verbose-mode trace messages
- Support handling verbose-mode trace messages of type

SSL3_RT_INNER_CONTENT_TYPE, SSL3_MT_ENCRYPTED_EXTENSIONS,
SSL3_MT_END_OF_EARLY_DATA, SSL3_MT_KEY_UPDATE, SSL3_MT_NEXT_PROTO,
SSL3_MT_MESSAGE_HASH
Reported-by: iz8mbw@users.noreply.github.com


Fixes #2445
Closes #2508
- detect_proxy: only show proxy use if it had contents
- http2: handle on_begin_headers() called more than once
This triggered an assert if called more than once in debug mode (and a
memory leak if not debug build). With the right sequence of HTTP/2
headers incoming it can happen.
Closes #2507

- [Dan McNulty brought this change]
schannel: add support for CURLOPT_CAINFO
- Move verify_certificate functionality in schannel.c into a new

file called schannel_verify.c. Additionally, some structure defintions
from schannel.c have been moved to schannel.h to allow them to be
used in schannel_verify.c.
- Make verify_certificate functionality for Schannel available on

all versions of Windows instead of just Windows CE. verify_certificate
will be invoked on Windows CE or when the user specifies
CURLOPT_CAINFO and CURLOPT_SSL_VERIFYPEER.
- In verify_certificate, create a custom certificate chain engine that

exclusively trusts the certificate store backed by the CURLOPT_CAINFO
file.
- doc updates of --cacert/CAINFO support for schannel
- Use CERT_NAME_SEARCH_ALL_NAMES_FLAG when invoking CertGetNameString

when available. This implements a TODO in schannel.c to improve
handling of multiple SANs in a certificate. In particular, all SANs
will now be searched instead of just the first name.
- Update tool_operate.c to not search for the curl-ca-bundle.crt file

when using Schannel to maintain backward compatibility. Previously,
any curl-ca-bundle.crt file found in that search would have been
ignored by Schannel. But, with CAINFO support, the file found by
that search would have been used as the certificate store and
could cause issues for any users that have curl-ca-bundle.crt in
the search path.
- Update url.c to not set the build time CURL_CA_BUNDLE if the selected
SSL backend is Schannel. We allow setting CA location for schannel
only when explicitly specified by the user via CURLOPT_CAINFO /
--cacert.
- Add new test cases 3000 and 3001. These test cases check that the first
and last SAN, respectively, matches the connection hostname. New test
certificates have been added for these cases. For 3000, the certificate
prefix is Server-localhost-firstSAN and for 3001, the certificate
prefix is Server-localhost-secondSAN.
- Remove TODO 15.2 (Add support for custom server certificate

validation), this commit addresses it.
- schannel: fix warning
- Fix warning 'integer from pointer without a cast' on 3rd arg in

CertOpenStore. The arg type HCRYPTPROV may be a pointer or integer
type of the same size.
Follow-up to e35b025.
Caught by Marc's CI builds.
- [Jakub Wilk brought this change]
docs: fix typos


- [Kees Dekker brought this change]
winbuild: Support custom devel paths for each dependency
- Support custom devel paths for c-ares, mbedTLS, nghttp2, libSSH2,

OpenSSL and zlib. Respectively: CARES_PATH, MBEDTLS_PATH,
NGHTTP2_PATH, SSH2_PATH, SSL_PATH and ZLIB_PATH.
- Use lib.exe for making the static library instead of link.exe /lib.
The latter is undocumented and could cause problems as noted in the
comments.
- Remove a dangling URL that no longer worked. (I was not able to find
the IDN download at MSDN/microsoft.com, so it seems to be removed.)
- Remove custom override for release-ssh2-ssl-dll-zlib configuration.

Nobody knows why it was there and as far as we can see is unnecessary.

- [Jess brought this change]
README.md: add backers and sponsors
Closes #2484
schannel: add client certificate authentication
Users can now specify a client certificate in system certificates store

explicitly using expression like `--cert "CurrentUser\MY\<thumbprint>"`
Closes #2376
Marcel Raad (16 Apr 2018)

- [toughengineer brought this change]
ntlm_sspi: fix authentication using Credential Manager
If you pass empty user/pass asking curl to use Windows Credential

Storage (as stated in the docs) and it has valid credentials for the
domain, e.g.
curl -v -u : --ntlm example.com
currently authentication fails.
This change fixes it by providing proper SPN string to the SSPI API
calls.

- configure: keep LD_LIBRARY_PATH changes local
... only set it when we actually have to run tests to reduce its impact
on for example build commands etc.
Fixes #2490
Closes #2492
Reported-by: Dmitry Mikhirev

- urldata: make service names unconditional
The ifdefs have become quite long. Also, the condition for the
definition of CURLOPT_SERVICE_NAME and for setting it from
CURLOPT_SERVICE_NAME have diverged. We will soon also need the two
options for NTLM, at least when using SSPI, for
https://github.com/curl/curl/pull/1660.
Just make the definitions unconditional to make that easier.

- test1148: tolerate progress updates better
Fixes #2446
Closes #2488
- [Christian Schmitz brought this change]
ssh: show libSSH2 error code when closing fails
Closes #2500

vauth: Fix typo
Address various spellings of "credentials".
- [Dagobert Michelsen brought this change]
system.h: Add sparcv8plus to oracle/sunpro 32-bit detection
With specific compiler options selecting the arch like -xarch=sparc on

newer compilers like Oracle Studio 12.4 there is no definition of
__sparcv8 but __sparcv8plus which means the V9 ISA, but limited to the
32ÎíÎñbit subset defined by the V8plus ISA specification, without the
Visual Instruction Set (VIS), and without other implementation-specific
ISA extensions. So it should be the same as __sparcv8.
checksrc: Fix typo
Fix typo in "semicolon" spelling and remove stray tab character.
all: Refactor malloc+memset to use calloc
When a zeroed out allocation is required, use calloc() rather than

malloc() followed by an explicit memset(). The result will be the
same, but using calloc() everywhere increases consistency in the
codebase and avoids the risk of subtle bugs when code is injected
between malloc and memset by accident.

- duphandle: make sure CURLOPT_RESOLVE is duplicated fine too
Verified in test 1502 now
Fixes #2485
Closes #2486
Reported-by: Ernst Sjöstrand
- mailmap: add a monnerat fixup [ci skip]
- proxy: show getenv proxy use in verbose output
... to aid debugging etc as it sometimes isn't immediately obvious why

curl uses or doesn't use a proxy.
Inspired by #2477
Closes #2480
- travis: build libpsl and make builds use it
closes #2471
- travis: bump to clang 6 and gcc 7
Extra-eye-on-this-by: Marcel Raad
Closes #2478

- travis: use trusty for coverage build
This works now and precise is in the process of being decommissioned.
- lib: silence null-dereference warnings
In debug mode, MingGW-w64's GCC 7.3 issues null-dereference warnings

when dereferencing pointers after DEBUGASSERT-ing that they are not
NULL.
Fix this by removing the DEBUGASSERTs.
Suggested-by: Daniel Stenberg

winbuild: fix URL
Follow up on https://github.com/curl/curl/pull/2472.
Now using en-us instead of nl-nl as language code in the URL.

winbuild: updated the documentation
The setenv command no longer exists and visual studio build prompts got
changed. Used Visual Studio 2015/2017 as reference.
Closes #2472
- test1136: fix cookie order after commit c990eadd1277
- build: cleanup to fix clang warnings/errors
unit1309 and vtls/gtls: error: arithmetic on a null pointer treated as a

cast from integer to pointer is a GNU extension
Fixes #2466
Closes #2468

- examples/sftpuploadresmue: Fix Windows large file seek
- Use _fseeki64 instead of fseek (long) to seek curl_off_t in Windows.
- Use CURL_FORMAT_CURL_OFF_T specifier instead of %ld to print

curl_off_t.
Caught by Marc's CI builds.

- curl_setup: provide a CURL_SA_FAMILY_T type if none exists
... and use this type instead of 'sa_family_t' in the code since several
platforms don't have it.
Closes #2463
- [Eric Gallager brought this change]
build: add picky compiler warning flags for gcc 6 and 7
- configure: detect sa_family_t

- [Stefan Agner brought this change]
tool_operate: Fix retry on FTP 4xx to ignore other protocols
Only treat response code as FTP response codes in case the

protocol type is FTP.
This fixes an issue where an HTTP download was treated as FTP

in case libcurl returned with 33. This happens when the
download has already finished and the server responses 416:
HTTP/1.1 416 Requested Range Not Satisfiable
This should not be treated as an FTP error.
Fixes #2464
Closes #2465

- hash: calculate sizes with size_t instead of longs
... since they return size_t anyway!
closes #2462
build-openssl.bat: Refer to VS2017 as VC14.1 instead of VC15
.. and do the same for build-wolfssl.bat.
Because MS calls it VC14.1.
winbuild: make the clean target work without build-type
Due to the check in Makefile.vc and MakefileBuild.vc, no make call can

be invoked unless a build-type was specified. However, a clean target
only existed when a build type was specified. As a result, the clean
target was unreachable. Made clean target unconditional.
Closes #2455
- [patelvivekv1993 brought this change]
build-openssl.bat: allow custom paths for VS and perl
Fixes #2430
Closes #2457
- [Laurie Clark-Michalek brought this change]
FTP: allow PASV on IPv6 connections when a proxy is being used
In the situation of a client connecting to an FTP server using an IPv6

tunnel proxy, the connection info will indicate that the connection is
IPv6. However, because the server behing the proxy is IPv4, it is
permissable to attempt PSV mode. In the case of the FTP server being
IPv4 only, EPSV will always fail, and with the current logic curl will
be unable to connect to the server, as the IPv6 fwdproxy causes curl to
think that EPSV is impossible.
Closes #2432
- [Jon DeVree brought this change]

file: restore old behavior for file:////foo/bar URLs
curl 7.57.0 and up interpret this according to Appendix E.3.2 of RFC

8089 but then returns an error saying this is unimplemented. This is
actually a regression in behavior on both Windows and Unix.
Before curl 7.57.0 this URL was treated as a path of "//foo/bar" and
then passed to the relevant OS API. This means that the behavior of this
case is actually OS dependent.
The Unix path resolution rules say that the OS must handle swallowing
the extra "/" and so this path is the same as "/foo/bar"
The Windows path resolution rules say that this is a UNC path and
automatically handles the SMB access for the program. So curl on Windows
was already doing Appendix E.3.2 without any special code in curl.
Regression
Closes #2438
- [Gaurav Malhotra brought this change]
Revert "openssl: Don't add verify locations when verifypeer==0"
This reverts commit dc85437736e1fc90e689bb1f6c51c8f1aa9430eb.
libcurl (with the OpenSSL backend) performs server certificate verification

even if verifypeer == 0 and the verification result is available using
CURLINFO_SSL_VERIFYRESULT. The commit that is being reverted caused the
CURLINFO_SSL_VERIFYRESULT to not have useful information for the
verifypeer == 0 use case (it would always have
X509_V_ERR_UNABLE_TO_GET_ISSUER_CERT_LOCALLY).
Closes #2451
- [Wyatt O'Day brought this change]
tls: fix mbedTLS 2.7.0 build + handle sha256 failures
(mbedtls 2.70 compiled with MBEDTLS_DEPRECATED_REMOVED)
Closes #2453
- [Lauri Kasanen brought this change]
cookie: case-insensitive hashing for the domains
closes #2458
Patrick Monnerat (4 Apr 2018)

- cookie: fix and optimize 2nd top level domain name extraction
This fixes a segfault occurring when a name of the (invalid) form "domain..tld"
is processed.
test46 updated to cover this case.

Follow-up to commit c990ead.

- openssl: provide defines for argument typecasts to build warning-free
... as OpenSSL >= 1.1.0 and libressl >= 2.7.0 use different argument types.
- [Bernard Spil brought this change]
openssl: fix build with LibreSSL 2.7
- LibreSSL 2.7 implements (most of) OpenSSL 1.1 API
Fixes #2319
Closes #2447
Closes #2448
Signed-off-by: Bernard Spil <brnrd@FreeBSD.org>
cookie: store cookies per top-level-domain-specific hash table
This makes libcurl handle thousands of cookies much better and speedier.
Closes #2440
cookies: when reading from a file, only remove_expired once
This drops the cookie load time for 8k cookies from 178ms to 15ms.
Closes #2441
- test1148: set a fixed locale for the test
...as otherwise it might use a different decimal sign.
Bug: #2436
Reported-by: Oumph on github
Jay Satiro (31 Mar 2018)

- docs: fix CURLINFO_*_T examples use of CURL_FORMAT_CURL_OFF_T
- Put a percent sign before each CURL_FORMAT_CURL_OFF_T in printf.
For example "%" CURL_FORMAT_CURL_OFF_T becomes %lld or similar.
Reported-by: David L.
Sergei Nikulov (27 Mar 2018)

- [Michał Janiszewski brought this change]
cmake: Add advapi32 as explicit link library for win32

ARM targets need advapi32 explicitly.
Closes #2363
Daniel Stenberg (27 Mar 2018)

- TODO: connection cache sharing is now supporte

- travis: enable apt retry on fail
This is a workaround for an unsolved travis issue that is causing CI

instances to sporadically fail due to 'unable to connect' issues during
apt stage.
Ref: https://github.com/travis-ci/travis-ci/issues/8507
Ref: https://github.com/travis-ci/travis-ci/issues/9112#issuecomment-376305909
Michael Kaufmann (26 Mar 2018)

- runtests.pl: fix warning 'use of uninitialized value'
follow-up to a9a7b60
Closes #2428

- gitignore: ignore more generated files
- threaded resolver: track resolver time and set suitable timeout values
In order to make curl_multi_timeout() return suitable "sleep" times even

when there's no socket to wait for while the name is being resolved in a
helper thread.
It will increases the timeouts as time passes.
Closes #2419
- [Howard Chu brought this change]
openldap: fix for NULL return from ldap_get_attribute_ber()
Closes #2399
GitHub (22 Mar 2018)

- [Sergei Nikulov brought this change]
travis-ci: enable -Werror for CMake builds (#2418)
cmake: avoid warn-as-error during config checks (#2411)
- Move the CURL_WERROR option processing after the configuration checks

to avoid failures in case of warnings during the configuration checks.
This is a partial fix for #2358

timeval: remove compilation warning by casting (#2417)
This is fixes #2358

- http2: read pending frames (including GOAWAY) in connection-check
If a connection has received a GOAWAY frame while not being used, the
function now reads frames off the connection before trying to reuse it
to avoid reusing connections the server has told us not to use.
Reported-by: Alex Baines

Fixes #1967
Closes #2402
- [Bas van Schaik brought this change]
CI: add lgtm.yml for tweaking lgtm.com analysis
Closes #2414
- CURLINFO_SSL_VERIFYRESULT.3: fix the example, add some text
Reported-by: Michal Trybus
Fixes #2400
- TODO: expand ~/ in config files
Closes #2317
- cookie.d: mention that "-" as filename means stdin
Reported-by: Dongliang Mu
Fixes #2410
- CURLINFO_COOKIELIST.3: made the example not leak memory
Reported-by: Muz Dima
- vauth/cleartext: fix integer overflow check
Make the integer overflow check not rely on the undefined behavior that
a size_t wraps around on overflow.
Detected by lgtm.com
Closes #2408
- lib/curl_path.h: add #ifdef header guard
- vauth/ntlm.h: fix the #ifdef header guard

- examples/hiperfifo: checksrc compliance
- [Nikos Tsipinakis brought this change]
parsedate: support UT timezone
RFC822 section 5.2 mentions Universal Time, 'UT', to be synonymous with

GMT.
Closes #2401
- [Don brought this change]
cmake: add support for brotli
Currently CMake cannot detect Brotli support. This adds detection of the
libraries and associated header files. It also adds this to the
generated config.
Closes #2392
- [Chris Araman brought this change]
darwinssl: fix iOS build
Patrick Monnerat (18 Mar 2018)

- ILE/RPG binding: Add CURLOPT_HAPROXYPROTOCOL/Fix CURLOPT_DNS_SHUFFLE_ADDRESSES

- [Rick Deist brought this change]
resolve: add CURLOPT_DNS_SHUFFLE_ADDRESSES
This patch adds CURLOPT_DNS_SHUFFLE_ADDRESSES to explicitly request

shuffling of IP addresses returned for a hostname when there is more
than one. This is useful when the application knows that a round robin
approach is appropriate and is willing to accept the consequences of
potentially discarding some preference order returned by the system's
implementation.
Closes #1694
- add_handle/easy_perform: clear errorbuffer on start if set
To offer applications a more defined behavior, we clear the buffer as

early as possible.
Assisted-by: Jay Satiro
Fixes #2190
Closes #2377
- [Lawrence Matthews brought this change]
CURLOPT_HAPROXYPROTOCOL: support the HAProxy PROXY protocol
Add --haproxy-protocol for the command line tool

Closes #2162
- curl_version_info.3: fix ssl_version description
Reported-by: Vincas Razma

Fixes #2364
- multi: improved pending transfers handling => improved performance
When a transfer is requested to get done and it is put in the pending

queue when limited by number of connections, total or per-host, libcurl
would previously very aggressively retry *ALL* pending transfers to get
them transferring. That was very time consuming.
By reducing the aggressiveness in how pending are being retried, we

waste MUCH less time on putting transfers back into pending again.
Some test cases got a factor 30(!) speed improvement with this change.
Reported-by: Cyril B
Fixes #2369
Closes #2383
- pause: when changing pause state, update socket state
Especially unpausing a transfer might have to move the socket back to the
"currently used sockets" hash to get monitored. Otherwise it would never get
any more data and get stuck. Easily triggered with pausing using the
multi_socket API.
Reported-by: Philip Prindeville

Fixes #2393
Closes #2391
- [Philip Prindeville brought this change]
examples/hiperfifo.c: improved
* use member struct event’s instead of pointers to alloc’d struct

events
* simplify the cases for the mcode_or_die() function via macros;
* make multi_timer_cb() actually do what the block comment says it

should;
* accept a “stop” command on the FIFO to shut down the service;
* use cleaner notation for unused variables than the (void) hack;
* allow following redirections (304’s);
- rate-limit: use three second window to better handle high speeds
Due to very frequent updates of the rate limit "window", it could

attempt to rate limit within the same milliseconds and that then made
the calculations wrong, leading to it not behaving correctly on very
fast transfers.
This new logic updates the rate limit "window" to be no shorter than the
last three seconds and only updating the timestamps for this when
switching between the states TOOFAST/PERFORM.
Reported-by: 刘佩东
Fixes #2386
Closes #2388
- [luz.paz brought this change]
cleanup: misc typos in strings and comments
Found via `codespell`
Closes #2389
- RELEASE-NOTES: toward 7.60.0
- [Kobi Gurkan brought this change]
http2: fixes typo
Closes #2387
- user-agent.d:: mention --proxy-header as well
Bug: https://github.com/curl/curl/issues/2381
- transfer: make HTTP without headers count correct body size
This is what "HTTP/0.9" basically looks like.
Reported on IRC
Closes #2382
- test1208: marked flaky
It fails somewhere between every 3rd to 10th travis-CI run
- SECURITY-PROCESS: mention how we write/add advisories
- [dasimx brought this change]
FTP: fix typo in recursive callback detection for seeking
Fixes #2380
Version 7.59.0 (13 Mar 2018)

- release: 7.59.0
Kamil Dudka (13 Mar 2018)

- tests/.../spnego.py: fix identifier typo
Detected by Coverity Analysis:

Error: IDENTIFIER_TYPO:
curl-7.58.0/tests/python_dependencies/impacket/spnego.py:229: identifier_typo:
Using "SuportedMech" appears to be a typo:
* Identifier "SuportedMech" is only known to be referenced here, or in copies of
this code.
* Identifier "SupportedMech" is referenced elsewhere at least 4 times.
curl-7.58.0/tests/python_dependencies/impacket/smbserver.py:2651: identifier_use:
Example 1: Using identifier "SupportedMech".
curl-7.58.0/tests/python_dependencies/impacket/smbserver.py:2308: identifier_use:
Example 2: Using identifier "SupportedMech".
curl-7.58.0/tests/python_dependencies/impacket/spnego.py:252: identifier_use:
Example 3: Using identifier "SupportedMech" (2 total uses in this function).
curl-7.58.0/tests/python_dependencies/impacket/spnego.py:229: remediation: Should
identifier "SuportedMech" be replaced by "SupportedMech"?
Closes #2379

- CURLOPT_COOKIEFILE.3: "-" as file name means stdin
Reported-by: Aron Bergman

[ci skip]
- Revert "hostip: fix compiler warning: 'variable set but not used'"
This reverts commit a577059f92fc65bd6b81717f0737f897a5b34248.
The assignment really needs to be there or we risk working with an

uninitialized pointer.

- limit-rate: fix compiler warning
follow-up to 72a0f62
Viktor Szakats (12 Mar 2018)

- checksrc.pl: add -i and -m options
To sync it with changes made for the libssh2 project.

Also cleanup some whitespace.
- curl-openssl.m4: fix spelling [ci skip]
- FAQ: fix a broken URL [ci skip]

- http2: mark the connection for close on GOAWAY
... don't consider it an error!
Assisted-by: Jay Satiro

Reported-by: Łukasz Domeradzki
Fixes #2365
Closes #2375
- credits: Viktor prefers without accent

- openldap: white space changes, fixed up the copyright years
- openldap: check ldap_get_attribute_ber() results for NULL before using
CVE-2018-1000121
Bug: https://curl.haxx.se/docs/adv_2018-97a2.html
- FTP: reject path components with control codes
Refuse to operate when given path components featuring byte values lower
than 32.
Previously, inserting a %00 sequence early in the directory part when

using the 'singlecwd' ftp method could make curl write a zero byte
outside of the allocated buffer.
Test case 340 verifies.
CVE-2018-1000120
Reported-by: Duy Phan Thanh
Bug: https://curl.haxx.se/docs/adv_2018-9cd6.html
- readwrite: make sure excess reads don't go beyond buffer end
CVE-2018-1000122
Bug: https://curl.haxx.se/docs/adv_2018-b047.html
Detected by OSS-fuzz
- BUGS: updated link to security process
- limit-rate: kick in even before "limit" data has been received
... and make sure to avoid integer overflows with really large values.
Reported-by: 刘佩东
Fixes #2371
Closes #2373
- docs/SECURITY.md -> docs/SECURITY-PROCESS.md
- SECURITY.md: call it the security process

- Curl_range: fix FTP-only and FILE-only builds
follow-up to e04417d
- hostip: fix compiler warning: 'variable set but not used'

- HTTP: allow "header;" to replace an internal header with a blank one

Fixes #2357
Closes #2362
- http2: verbose output new MAX_CONCURRENT_STREAMS values

... as it is interesting for many users.
- SECURITY: distros' max embargo time is 14 days now
Patrick Monnerat (8 Mar 2018)

- curl tool: accept --compressed also if Brotli is enabled and zlib is not.

- THANKS + mailmap: remove duplicates, fixup full names
- [sergii.kavunenko brought this change]
WolfSSL: adding TLSv1.3
Closes #2349

Changes

Uploaded by

Copyright:

Available Formats

Changes

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Changes

Uploaded by

Copyright:

Available Formats

_ _ ____ _

Version 7.64.0 (6 Feb 2019)

Daniel Stenberg (6 Feb 2019)

- RELEASE-PROCEDURE: update the release calendar

- THANKS: 7.64.0 status

Daniel Gustafsson (5 Feb 2019)

Commit 7a09b52c98ac8d840a8a9907b1a1d9a9e684bcf5 introduced support

- [Etienne Simard brought this change]

CONTRIBUTE.md: Fix grammatical errors

Daniel Stenberg (4 Feb 2019)

docs: use $(INSTALL_DATA) to install man page

Jay Satiro (4 Feb 2019)

runtests.pl: Fix perl call to include srcdir

- Use explicit include opt for perl calls.

Prior to this change some scripts couldn't find their dependencies.

This patch fixes that oversight by making calls to perl sub-shells

Daniel Stenberg (4 Feb 2019)

smtp: avoid risk of buffer overflow in strtol

- ntlm: fix *_type3_message size check to avoid buffer overflow

- NTLM: fix size check condition for type2 received data

Marcel Raad (1 Feb 2019)

spnego_sspi: add support for channel binding

Attempt to add support for Secure Channel binding when negotiate

Daniel Stenberg (1 Feb 2019)

- schannel: stop calling it "winssl"

Stick to "Schannel" everywhere. The configure option --with-winssl is

- multi: set the EXPIRE_*TIMEOUT timers at TIMER_STARTSINGLE time

- [John Marshall brought this change]

doc: use meaningless port number in CURLOPT_LOCALPORT example

GitHub (29 Jan 2019)

Escape the '\'

A backslash should be escaped in Roff / Troff.

Jay Satiro (29 Jan 2019)

By default WinSSL selects and send a client certificate automatically,

Reported-by: Jeroen Ooms

Daniel Stenberg (28 Jan 2019)

sigpipe: if mbedTLS is used, ignore SIGPIPE

mbedTLS doesn't have a sigpipe management. If a write/read occurs when

Signed-off-by: Jeremie Rapin <j.rapin@overkiz.com>

- unit1653: make it survive torture tests

Jay Satiro (28 Jan 2019)

Since GetTickCount is only being used when GetTickCount64 isn't

Daniel Stenberg (26 Jan 2019)

The previously used ax_code_coverage.m4 is not license compatible and

Reported-by: William A. Rowe Jr

- [Felix Hädicke brought this change]

setopt: enable CURLOPT_SSH_KNOWNHOSTS and CURLOPT_SSH_KEYFUNCTION for libssh

CURLOPT_SSH_KNOWNHOSTS and CURLOPT_SSH_KEYFUNCTION are supported for

- [Felix Hädicke brought this change]

libssh: do not let libssh create socket

By default, libssh creates a new socket, instead of using the socket

Pass the socket created by curl to libssh using ssh_options_set() with

- [Archangel_SDY brought this change]

schannel: preserve original certificate path parameter

Daniel Gustafsson (20 Jan 2019)

There is no real gain in performing memcmp() comparisons on single

Daniel Stenberg (19 Jan 2019)

- [hhb brought this change]

configure: fix recv/send/select detection on Android

This reverts commit d4f25201fb7da03fc88f90d51101beb3d0026db9.

Follow-up to 2fa0d57e2e3. The xlc symbol is only defined there if a