Next Protocol Negotiation is a TLS extension that was created and used
for agreeing to use the SPDY protocol (the precursor to HTTP/2) for
HTTPS. In the early days of HTTP/2, before the spec was finalized and
shipped, the protocol could be enabled using this extension with some
servers.
curl supports the NPN extension with some TLS backends since then, with
a command line option `--npn` and in libcurl with
`CURLOPT_SSL_ENABLE_NPN`.
HTTP/2 proper is made to use the ALPN (Application-Layer Protocol
Negotiation) extension and the NPN extension has no purposes
anymore. The HTTP/2 spec was published in May 2015.
Today, use of NPN in the wild should be extremely rare and most likely
totally extinct. Chrome removed NPN support in Chrome 51, shipped in
June 2016. Removed in Firefox 53, April 2017.
Closes#9307
- parallel transfers: only alloc and keep errorbuffers in memory for
actual "live" transfers and not for the ones in the pending queue
- serial transfers: reuse the same fixed buffer for all transfers, not
allocated at all.
Closes#9394
When doing a huge amount of parallel transfers, we must not add them to
the per_transfer list frivolously since they all use memory after all.
This was previous done without really considering millions or billions
of transfers. Massive parallelism would use a lot of memory for no good
purpose.
The queue is now limited to twice the paralleism number.
This makes the 'Qd' value in the parallel progress meter mostly useless
for users, but works for now for us as a debug display.
Reported-by: justchen1369 on github
Fixes#8933Closes#9389
Amiga specific code which put the URL in the file comment was perhaps
accidentally removed in b889408500 having
originally been added in 5c215bdbdf.
Reworked to fit the code changes and added it back in.
Reported-by: Michael Trebilcock
Originally-added-by: Chris Young
Closes#9258
... as replacements for deprecated CURLOPT_PROTOCOLS and
CURLOPT_REDIR_PROTOCOLS as these new ones do not risk running into the
32 bit limit the old ones are facing.
CURLINFO_PROTCOOL is now deprecated.
The curl tool is updated to use the new options.
Added test 1597 to verify the libcurl protocol parser.
Closes#8992
Add licensing and copyright information for all files in this repository. This
either happens in the file itself as a comment header or in the file
`.reuse/dep5`.
This commit also adds a Github workflow to check pull requests and adapts
copyright.pl to the changes.
Closes#8869
When --remove-on-error is used with --no-clobber, it might have an
updated file name to remove.
Bug: https://curl.se/docs/CVE-2022-27778.html
CVE-2022-27778
Reported-by: Harry Sintonen
Closes#8824
This loop was using the number of bytes read from the file as condition
to keep reading.
From Linux's fread(3) man page:
> On success, fread() and fwrite() return the number of items read or
> written. This number equals the number of bytes transferred only when
> size is 1. If an error occurs, or the end of the file is reached, the
> return value is a short item count (or zero).
>
> The file position indicator for the stream is advanced by the number
> of bytes successfully read or written.
>
> fread() does not distinguish between end-of-file and error, and
> callers must use feof(3) and ferror(3) to determine which occurred.
This means that nread!=0 doesn't make much sense as an end condition for
the loop: nread==0 doesn't necessarily mean that EOF has been reached or
an error has occured (but that is usually the case) and nread!=0 doesn't
necessarily mean that EOF has not been reached or that no read errors
have occured. feof(3) and ferror(3) should be uses when using fread(3).
Currently curl has to performs an extra fread(3) call to get a return
value equal to 0 to stop looping.
This usually "works" (even though nread==0 shouldn't be interpreted as
EOF) if stdin is a pipe because EOF usually marks the "real" end of the
stream, so the extra fread(3) call will return immediately and the extra
read syscall won't be noticeable:
bash-5.1$ strace -e read curl -s -F file=@- 0x0.st <<< a 2>&1 |
> tail -n 5
read(0, "a\n", 4096) = 2
read(0, "", 4096) = 0
read(0, "", 4096) = 0
http://0x0.st/oRs.txt
+++ exited with 0 +++
bash-5.1$
But this doesn't work if curl is reading from stdin, stdin is a
terminal, and the EOF is being emulated using a shell with ^D. Two
consecutive ^D will be required in this case to actually make curl stop
reading:
bash-5.1$ curl -F file=@- 0x0.st
a
^D^D
http://0x0.st/oRs.txt
bash-5.1$
A possible workaround to this issue is to use a program that handles EOF
correctly to indirectly send data to curl's stdin:
bash-5.1$ cat - | curl -F file=@- 0x0.st
a
^D
http://0x0.st/oRs.txt
bash-5.1$
This patch makes curl handle EOF properly when using fread(3) in
file2memory() so that the workaround is not necessary.
Since curl was previously ignoring read errors caused by this fread(3),
ferror(3) is also used in the condition of the loop: read errors and EOF
will have the same meaning; this is done to somewhat preserve the old
behaviour instead of making the command fail when a read error occurs.
Closes#8701
As one implies PUT and the other POST, both cannot be used
simultaneously.
Add test 378 to verify.
Reported-by: Boris Verkhovskiy
Fixes#8704Closes#8715
If a transfer returns an error, using this option makes curl remove the
leftover downloded (partial) local file before exiting.
Added test 376 to verify
Closes#8503
Trying to use a proxy when libcurl was built with proxy support disabled
should make curl error out properly.
Remove knowledge of disabled features from the tool code and instead
make it properly respond to what libcurl returns. Update all tests to
properly require the necessary features to be present/absent so that the
test suite can still be run even with libcurl builds with disabled
features.
Ref: https://curl.se/mail/archive-2022-03/0013.htmlCloses#8565
Unless muted (with -s) When doing globbing, curl would output mime-like
separators between the separate transfers. This is not documented
anywhere, surprises users and clobbers the output. Gone now.
Updated test 18 and 1235
Reported-by: jonny112 on github
Bug: https://github.com/curl/curl/discussions/8257Closes#8278
The homedir() function is now renamed into findfile() and iterates over
all the environment variables trying to access the file in question
until it finds it. Last resort is then getpwuid() if
available. Previously it would first try to find a home directory and if
that was set, insist on checking only that directory for the file. This
now returns the full file name it finds.
The Windows specific checks are now done differently too and in this
order:
1 - %USERPROFILE%
2 - %APPDATA%
3 - %USERPROFILE%\\Application Data
The windows order is modified to match how the Windows 10 ssh tool works
when it searches for .ssh/known_hosts.
Reported-by: jeffrson on github
Co-authored-by: Jay Satiro
Fixes#8033Closes#8035
Until now, form field and file names where escaped using the
backslash-escaping algorithm defined for multipart mails. This commit
replaces this with the percent-escaping method for URLs.
As this may introduce incompatibilities with server-side applications, a
new libcurl option CURLOPT_MIME_OPTIONS with bitmask
CURLMIMEOPT_FORMESCAPE is introduced to revert to legacy use of
backslash-escaping. This is controlled by new cli tool option
--form-escape.
New tests and documentation are provided for this feature.
Reported by: Ryan Sleevi
Fixes#7789Closes#7805
When failing to create the output file for saving an etag, only fail
that particular single transfer and allow others to follow.
In a serial transfer setup, if no transfer at all is done due to them
all being skipped because of this error, curl will output an error
message and return exit code 26.
Added test 369 and 370 to verify.
Reported-by: Earnestly on github
Ref: #7942Closes#7945
If Retry-After: specifies a period that is longer than what fits within
--retry-max-time, then stop retrying immediately.
Added test 366 to verify.
Reported-by: Kari Pahula
Fixes#7675Closes#7676
Due to how this currently works internally, it needs a working initial
file name to store contents in, so it may still fail even with -J is
used (and thus accepting a name from content-disposition:) if the file
name part of the URL isn't "good enough".
Fixes#7628Closes#7635
- Abort via progress callback to fail early during parallel transfers.
When a critical error occurs during a transfer (eg --fail-early
constraint) then other running transfers will be aborted via progress
callback and finish with error CURLE_ABORTED_BY_CALLBACK (42). In this
case, the callback error does not become the most recent error and a
custom error message is used for those transfers:
curld --fail --fail-early --parallel
https://httpbin.org/status/404https://httpbin.org/delay/10
curl: (22) The requested URL returned error: 404
curl: (42) Transfer aborted due to critical error in another transfer
> echo %ERRORLEVEL%
22
Fixes https://github.com/curl/curl/issues/6939
Closes https://github.com/curl/curl/pull/6984
Warning: this will make existing curl command lines that use metalink to
stop working.
Reasons for removal:
1. We've found several security problems and issues involving the
metalink support in curl. The issues are not detailed here. When
working on those, it become apparent to the team that several of the
problems are due to the system design, metalink library API and what
the metalink RFC says. They are very hard to fix on the curl side
only.
2. The metalink usage with curl was only very briefly documented and was
not following the "normal" curl usage pattern in several ways, making
it surprising and non-intuitive which could lead to further security
issues.
3. The metalink library was last updated 6 years ago and wasn't so
active the years before that either. An unmaintained library means
there's a security problem waiting to happen. This is probably reason
enough.
4. Metalink requires an XML parsing library, which is complex code (even
the smaller alternatives) and to this day often gets security
updates.
5. Metalink is not a widely used curl feature. In the 2020 curl user
survey, only 1.4% of the responders said that they'd are using it. In
2021 that number was 1.2%. Searching the web also show very few
traces of it being used, even with other tools.
6. The torrent format and associated technology clearly won for
downloading large files from multiple sources in parallel.
Cloes #7176
- Disable auto credentials by default. This is a breaking change
for clients that are using it, wittingly or not.
- New libcurl ssl option value CURLSSLOPT_AUTO_CLIENT_CERT tells libcurl
to automatically locate and use a client certificate for
authentication, when requested by the server.
- New curl tool options --ssl-auto-client-cert and
--proxy-ssl-auto-client-cert map to CURLSSLOPT_AUTO_CLIENT_CERT.
This option is only supported for Schannel (the native Windows SSL
library). Prior to this change Schannel would, with no notification to
the client, attempt to locate a client certificate and send it to the
server, when requested by the server. Since the server can request any
certificate that supports client authentication in the OS certificate
store it could be a privacy violation and unexpected.
Fixes https://github.com/curl/curl/issues/2262
Reported-by: Jeroen Ooms
Assisted-by: Wes Hinsley
Assisted-by: Rich FitzJohn
Ref: https://curl.se/mail/lib-2021-02/0066.html
Reported-by: Morten Minde Neergaard
Closes https://github.com/curl/curl/pull/6673
- Save a parallel transfer's result code only when it fails and the
transfer is not being retried.
Prior to this change the result code was always set which meant that a
failed result could be erroneously discarded if a different transfer
later had a successful result (CURLE_OK).
Before:
> curl --fail -Z https://httpbin.org/status/404https://httpbin.org/delay/10
> echo %ERRORLEVEL%
0
After:
> curl --fail -Z https://httpbin.org/status/404https://httpbin.org/delay/10
> echo %ERRORLEVEL%
22
Closes #xxxx
The --create-file-mode code logic accepted the value but never actually
passed it on to libcurl!
Follow-up to a7696c7343 (shipped in 7.75.0)
Reported-by: Johannes Lesr
Fixes#6657Closes#6666
Previously only a single -b cookie parameter was supported with the last
one winning. This adds support for supplying multiple -b params to have
them serialized semicolon separated. Both cookiefiles and cookies can be
entered multiple times.
Closes#6649
Reviewed-by: Daniel Stenberg <daniel@haxx.se>
- New libcurl options CURLOPT_DOH_SSL_VERIFYHOST,
CURLOPT_DOH_SSL_VERIFYPEER and CURLOPT_DOH_SSL_VERIFYSTATUS do the
same as their respective counterparts.
- New curl tool options --doh-insecure and --doh-cert-status do the same
as their respective counterparts.
Prior to this change DOH SSL certificate verification settings for
verifyhost and verifypeer were supposed to be inherited respectively
from CURLOPT_SSL_VERIFYHOST and CURLOPT_SSL_VERIFYPEER, but due to a bug
were not. As a result DOH verification remained at the default, ie
enabled, and it was not possible to disable. This commit changes
behavior so that the DOH verification settings are independent and not
inherited.
Ref: https://github.com/curl/curl/pull/4579#issuecomment-554723676
Fixes https://github.com/curl/curl/issues/4578
Closes https://github.com/curl/curl/pull/6597
- Deduplicate the logic used by write-out and write-out json.
Rather than have separate writeLong, writeString, etc, logic for
each of write-out and write-out json instead have respective shared
functions that can output either format and a 'use_json' parameter to
indicate whether it is json that is output.
This will make it easier to maintain. Rather than have to go through
two sets of logic now we only have to go through one.
- Support write-out %{errormsg} and %{exitcode} in json.
- Clarify in the doc that %{exitcode} is the exit code of the transfer.
Prior to this change it just said "The numerical exitcode" which
implies it's the exit code of the tool, and it's not necessarily that.
Closes https://github.com/curl/curl/pull/6544
- Add support services without region and service prefixes in
the URL endpoint (ex. Min.IO, GCP, Yandex Cloud, Mail.Ru Cloud Solutions, etc)
by providing region and service parameters via aws-sigv4 option.
- Add [:region[:service]] suffix to aws-sigv4 option;
- Fix memory allocation errors.
- Refactor memory management.
- Use Curl_http_method instead() STRING_CUSTOMREQUEST.
- Refactor canonical headers generating.
- Remove repeated sha256_to_hex() usage.
- Add some docs fixes.
- Add some codestyle fixes.
- Add overloaded strndup() for debug - curl_dbg_strndup().
- Update tests.
Closes#6524
- 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.
Ref: https://curl.se/docs/manpage.html#-S
Bug: https://curl.se/mail/archive-2020-12/0017.html
Reported-by: Hongyi Zhao
Closes https://github.com/curl/curl/pull/6402
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.
Closes#6395
The match of the "-q" option (short for "--disable") should:
a) allow concatenation with other single-letters; and
b) be case-sensitive, lest confusing with "-Q" ("--quote")
Closes#6364
In particular, these ones can help a user to create its own error
message when one or transfers fail.
writeout: add 'onerror', 'url', 'urlnum', 'exitcode', 'errormsg'
onerror - lets a user only show the rest on non-zero exit codes
url - the input URL used for this transfer
urlnum - the numerical URL counter (0 indexed) for this transfer
exitcode - the numerical exit code for the transfer
errormsg - obvious
Reported-by: Earnestly on github
Fixes#6199Closes#6207
- enable in the build (configure)
- header parsing
- host name lookup
- unit tests for the above
- CI build
- CURL_VERSION_HSTS bit
- curl_version_info support
- curl -V output
- curl-config --features
- CURLOPT_HSTS_CTRL
- man page for CURLOPT_HSTS_CTRL
- curl --hsts (sets CURLOPT_HSTS_CTRL and works with --libcurl)
- man page for --hsts
- save cache to disk
- load cache from disk
- CURLOPT_HSTS
- man page for CURLOPT_HSTS
- added docs/HSTS.md
- fixed --version docs
- adjusted curl_easy_duphandle
Closes#5896
This was inadvertently dropped from the code when the parallel support
was added.
Regression since b88940850 (7.66.0)
Reviewed-by: Jay Satiro
Closes#6155
Previously, it would pass on a define (int) which could make libcurl
read junk as a value - which prevented the CURLOPT_IPRESOLVE option to
"take". This could then make test 2100 do two DoH requests instead of
one!
Fixes#6042Closes#6043
The previous sleep for retries would block all other concurrent
transfers. Starting now, the retry will instead be properly marked to
not get restarted until after the delay time but other transfers can
still continue in the mean time.
Closes#5917
Background:
148534d added CURLSSLOPT_NATIVE_CA to use the Windows OS certificate
store in libcurl w/ OpenSSL on Windows. CURLSSLOPT_NATIVE_CA overrides
CURLOPT_CAINFO if both are set. The curl tool will fall back to
CURLSSLOPT_NATIVE_CA if it could not find a certificate bundle to set
via CURLOPT_CAINFO.
Problem:
libcurl may be built with hardcoded paths to a certificate bundle or
directory, and if CURLSSLOPT_NATIVE_CA is used then those paths are
ignored.
Solution:
A solution is still being discussed but since there's an impending
release this commit removes using CURLSSLOPT_NATIVE_CA in the curl tool.
Ref: https://github.com/curl/curl/issues/5585
curl would previously show "curl: Saved to filename 'name from header'"
if -J was used and a name was picked from the Content-Disposition
header. That output could interfer with other stdout output, such as -w.
This commit removes that output line.
Bug: https://curl.haxx.se/mail/archive-2020-05/0044.html
Reported-by: Коваленко Анатолий Викторович
Closes#5435
This change introduces a generic way to provide binary data in setopt
options, called BLOBs.
This change introduces these new setopts:
CURLOPT_ISSUERCERT_BLOB, CURLOPT_PROXY_SSLCERT_BLOB,
CURLOPT_PROXY_SSLKEY_BLOB, CURLOPT_SSLCERT_BLOB and CURLOPT_SSLKEY_BLOB.
Reviewed-by: Daniel Stenberg
Closes#5357
- Stick to a single unified way to use structs
- Make checksrc complain on 'typedef struct {'
- Allow them in tests, public headers and examples
- Let MD4_CTX, MD5_CTX, and SHA256_CTX typedefs remain as they actually
typedef different types/structs depending on build conditions.
Closes#5338
- use `wmain` instead of `main` when `_UNICODE` is defined [0]
- define `argv_item_t` as `wchar_t *` in this case
- use the curl_multibyte gear to convert the command-line arguments to
UTF-8
This makes it possible to pass parameters with characters outside of
the current locale on Windows, which is required for some tests, e.g.
the IDN tests. Out of the box, this currently only works with the
Visual Studio project files, which default to Unicode, and winbuild
with the `ENABLE_UNICODE` option.
[0] https://devblogs.microsoft.com/oldnewthing/?p=40643
Ref: https://github.com/curl/curl/issues/3747
Closes https://github.com/curl/curl/pull/3784
This change inverse the order of processing for the --etag-compare and
--etag-save option to process first --etag-compare. This in turn allows
to use the same file name to compare and save an etag.
The original behavior of not failing if the etag file does not exists is
conserved.
Fixes#5179Closes#5180
- Implement new option CURLSSLOPT_REVOKE_BEST_EFFORT and
--ssl-revoke-best-effort to allow a "best effort" revocation check.
A best effort revocation check ignores errors that the revocation check
was unable to take place. The reasoning is described in detail below and
discussed further in the PR.
---
When running e.g. with Fiddler, the schannel backend fails with an
unhelpful error message:
Unknown error (0x80092012) - The revocation function was unable
to check revocation for the certificate.
Sadly, many enterprise users who are stuck behind MITM proxies suffer
the very same problem.
This has been discussed in plenty of issues:
https://github.com/curl/curl/issues/3727,
https://github.com/curl/curl/issues/264, for example.
In the latter, a Microsoft Edge developer even made the case that the
common behavior is to ignore issues when a certificate has no recorded
distribution point for revocation lists, or when the server is offline.
This is also known as "best effort" strategy and addresses the Fiddler
issue.
Unfortunately, this strategy was not chosen as the default for schannel
(and is therefore a backend-specific behavior: OpenSSL seems to happily
ignore the offline servers and missing distribution points).
To maintain backward-compatibility, we therefore add a new flag
(`CURLSSLOPT_REVOKE_BEST_EFFORT`) and a new option
(`--ssl-revoke-best-effort`) to select the new behavior.
Due to the many related issues Git for Windows and GitHub Desktop, the
plan is to make this behavior the default in these software packages.
The test 2070 was added to verify this behavior, adapted from 310.
Based-on-work-by: georgeok <giorgos.n.oikonomou@gmail.com>
Co-authored-by: Markus Olsson <j.markus.olsson@gmail.com>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Closes https://github.com/curl/curl/pull/4981
Trying to return early from the function if no new transfers were added
would break the "morep" argument and cause issues. This could lead to
zero content "transfers" (within quotes since they would never be
started) when parallel-max was reduced.
Reported-by: Gavin Wong
Analyzed-by: Jay Satiro
Fixes#4937Closes#5112
Make the tool check for alt-svc support at run-time and return error
accordingly if not present when the option is used.
Reported-by: Harry Sintonen
Closes#4878
As it was just unnecessary duplicated information already stored in the
'per_transfer' struct and that's around mostly anyway.
The duplicated pointer caused problems when the code flow was aborted
before the dupe was filled in and could cause a NULL pointer access.
Reported-by: Brian Carpenter
Fixes#4807Closes#4810
