Daniel Stenberg [Tue, 19 Jan 2021 14:57:24 +0000 (15:57 +0100)]
doh: allocate state struct on demand
... instead of having it static within the Curl_easy struct. This takes
away 1176 bytes (18%) from the Curl_easy struct that aren't used very
often and instead makes the code allocate it when needed.
Daniel Stenberg [Tue, 19 Jan 2021 14:30:59 +0000 (15:30 +0100)]
socks: use the download buffer instead
The SOCKS code now uses the generic download buffer for temporary
storage during the connection procedure, instead of having its own
private 600 byte buffer that adds to the connectdata struct size. This
works fine because this point the buffer is allocated but is not use for
download yet since the connection hasn't completed.
This reduces the connection struct size by 22% on a 64bit arch!
The SOCKS buffer needs to be at least 600 bytes, and the download buffer
is guaranteed to never be smaller than 1000 bytes.
Daniel Stenberg [Tue, 19 Jan 2021 07:23:52 +0000 (08:23 +0100)]
urldata: make magic be the first struct field
By making the `magic` identifier the same size and at the same place
within the structs (easy, multi, share), libcurl will be able to more
reliably detect and safely error out if an application passes in the
wrong handle to APIs. Easier to detect and less likely to cause crashes
if done.
Such mixups can't be detected at compile-time due to them being
typedefed void pointers - unless `CURL_STRICTER` is defined.
Daniel Stenberg [Mon, 18 Jan 2021 15:40:39 +0000 (16:40 +0100)]
libssh: avoid plain free() of libssh-memory
Since curl's own memory debugging system redefines free() calls to track
and fiddle with memory, it cannot be used on memory allocated by 3rd
party libraries.
Third party libraries SHOULD NOT require free() to release allocated
resources for this reason - and libs can use separate healp allocators
on some systems (like Windows) so free() doesn't necessarily work
anyway.
Filed as an issue with libssh: https://bugs.libssh.org/T268
Daniel Stenberg [Fri, 8 Jan 2021 16:58:15 +0000 (17:58 +0100)]
lib: pass in 'struct Curl_easy *' to most functions
... in most cases instead of 'struct connectdata *' but in some cases in
addition to.
- We mostly operate on transfers and not connections.
- We need the transfer handle to log, store data and more. Everything in
libcurl is driven by a transfer (the CURL * in the public API).
- This work clarifies and separates the transfers from the connections
better.
- We should avoid "conn->data". Since individual connections can be used
by many transfers when multiplexing, making sure that conn->data
points to the current and correct transfer at all times is difficult
and has been notoriously error-prone over the years. The goal is to
ultimately remove the conn->data pointer for this reason.
Daniel Stenberg [Fri, 15 Jan 2021 12:04:20 +0000 (13:04 +0100)]
test1918: verify curl_easy_option_by_name() and curl_easy_option_by_id()
... and as a practical side-effect, make sure that the
Curl_easyopts_check() function is asserted in debug builds, which we
want to detect mismatches between the options list in easyoptions.c and
the options in curl.h
Daniel Stenberg [Thu, 14 Jan 2021 13:24:07 +0000 (14:24 +0100)]
multi: set the PRETRANSFER time-stamp when we switch to PERFORM
... instead of at end of the DO state. This makes the timer more
accurate for the protocols that use the DOING state (such as FTP), and
simplifies how the function (now called init_perform) is called.
The timer will then include the entire procedure up to PERFORM -
including all instructions for getting the transfer started.
Jay Satiro [Wed, 13 Jan 2021 07:03:57 +0000 (02:03 -0500)]
gitattributes: Set batch files to CRLF line endings on checkout
If a batch file is run without CRLF line endings (ie LF-only) then
arbitrary behavior may occur. I consider that a bug in Windows, however
the effects can be serious enough (eg unintended code executed) that
we're fixing it in the repo by requiring CRLF line endings for batch
files on checkout.
Prior to this change the checked-out line endings of batch files were
dependent on a user's git preferences. On Windows it is common for git
users to have automatic CRLF conversion enabled (core.autocrlf true),
but those users that don't would run into this behavior.
For example a user has reported running the Visual Studio project
generator batch file (projects/generate.bat) and it looped forever.
Output showed that the Windows OS interpreter was occasionally jumping
to arbitrary points in the batch file and executing commands. This
resulted in unintended files being removed (a removal sequence called)
and looping forever.
Jay Satiro [Thu, 19 Nov 2020 21:49:24 +0000 (16:49 -0500)]
tool_doswin: Restore original console settings on CTRL signal
- Move Windows terminal init code from tool_main to tool_doswin.
- Restore the original console settings on CTRL+C and CTRL+BREAK.
Background: On Windows the curl tool changes the console settings to
enable virtual terminal processing (eg color output) if supported
(ie Win 10). The original settings are restored on exit but prior to
this change were not restored in the case of the CTRL signals.
Windows VT behavior varies depending on console/powershell/terminal;
refer to the discussion in #6226.
Emil Engler [Mon, 11 Jan 2021 19:27:58 +0000 (20:27 +0100)]
docs: fix line length bug in gen.pl
The script warns if the length of $opt and $desc is > 78. However, these
two variables are on totally separate lines so the check makes no sense.
Also the $bitmask field is totally forgotten. Currently this leads to
two warnings within `--resolve` and `--aws-sigv4`.
- During the end-of-headers response phase do not mark the tunnel
complete unless the response body was completely parsed/ignored.
Prior to this change if the entirety of a CONNECT response with chunked
encoding was not received by the time the final header was parsed then
the connection would be marked done prematurely, before all the chunked
data could be read in and ignored (since this is what we do with any
CONNECT response body) and the connection could not be used.
Daniel Stenberg [Tue, 5 Jan 2021 13:30:21 +0000 (14:30 +0100)]
pretransfer: setup the User-Agent header here
... and not in the connection setup, as for multiplexed transfers the
connection setup might be skipped and then the transfer would end up
without the set user-agent!
Reported-by: Flameborn on github Assisted-by: Andrey Gursky Assisted-by: Jay Satiro Assisted-by: Mike Gelfand
Fixes #6312
Closes #6417
Jay Satiro [Sat, 2 Jan 2021 08:40:24 +0000 (03:40 -0500)]
tool_operate: fix the suppression logic of some error messages
- Fix the failed truncation and failed writing body error messages to
not be shown unless error messages are shown. (ie the user has
specified -sS, or has not specified -s).
- Also prefix same error messages with "curl: ", for example:
curl: (23) Failed to truncate, exiting
Prior to this change the failed truncation error messages would be shown
if not -s, but did not account for -sS which should show.
Prior to this change the failed writing body error messages would be
shown always.
Jay Satiro [Tue, 29 Dec 2020 20:46:42 +0000 (15:46 -0500)]
wolfssl: Support wolfSSL builds missing TLS 1.1
The wolfSSL TLS library defines NO_OLD_TLS in some of their build
configurations and that causes the library to be built without TLS 1.1.
For example if MD5 is explicitly disabled when building wolfSSL then
that defines NO_OLD_TLS and the library is built without TLS 1.1 [1].
Prior to this change attempting to build curl with a wolfSSL that was
built with NO_OLD_TLS would cause a build link error undefined reference
to wolfTLSv1_client_method.
Daniel Stenberg [Fri, 1 Jan 2021 22:41:21 +0000 (23:41 +0100)]
httpauth: make multi-request auth work with custom port
When doing HTTP authentication and a port number set with CURLOPT_PORT,
the code would previously have the URL's port number override as if it
had been a redirect to an absolute URL.
Added test 1568 to verify.
Reported-by: UrsusArctos on github
Fixes #6397
Closes #6400
Emil Engler [Thu, 31 Dec 2020 09:11:49 +0000 (10:11 +0100)]
language: s/behaviour/behavior/g
We currently use both spellings the british "behaviour" and the american
"behavior". However "behavior" is more used in the project so I think
it's worth dropping the british name.
projects / thirdparty / curl.git / log
