]> granicus.if.org Git - curl/log
curl
10 years agotest2036: verify -O with no slash at all in the URL
Daniel Stenberg [Thu, 9 Oct 2014 14:50:26 +0000 (16:50 +0200)]
test2036: verify -O with no slash at all in the URL

Similar to test 76 but that test's URL has a slash just no file name
part.

10 years agoget_url_file_name: make no slash equal empty string
Daniel Stenberg [Thu, 9 Oct 2014 14:42:17 +0000 (16:42 +0200)]
get_url_file_name: make no slash equal empty string

10 years agoget_url_file_name: never return a NULL string *and* OK
Daniel Stenberg [Thu, 9 Oct 2014 14:37:11 +0000 (16:37 +0200)]
get_url_file_name: never return a NULL string *and* OK

Change 987a4a73 assumes that as it simplifies life in the calling
function.

Reported-by: Fabian Keil
10 years agoCmake: Build with GSSAPI (MIT or Heimdal)
Jakub Zakrzewski [Thu, 25 Sep 2014 13:03:00 +0000 (15:03 +0200)]
Cmake: Build with GSSAPI (MIT or Heimdal)

It tries hard to recognise SDK's on different platforms. On windows MIT
Kerberos installs SDK with other things and puts path into registry.
Heimdal have separate zip archive. On linux pkg-config is tried, then
krb5-config script and finally old-style libs and headers detection.

Command line args:
* CMAKE_USE_GSSAPI - enables GSSAPI detection
* GSS_ROOT_DIR - if set, should point to the root of GSSAPI installation
                 (the one with include and lib directories)

10 years agoCmake: Got rid of setup_curl_dependencies
Jakub Zakrzewski [Fri, 22 Aug 2014 15:02:59 +0000 (17:02 +0200)]
Cmake: Got rid of setup_curl_dependencies

There is no need for such function. Include_directories propagate by
themselves and having a function with one simple link statement makes
little sense.

10 years agoCmake: Avoid cycle directory dependencies.
Jakub Zakrzewski [Fri, 22 Aug 2014 14:59:30 +0000 (16:59 +0200)]
Cmake: Avoid cycle directory dependencies.

Because we prepended libraries to list, CMake had troubles resolving
link directory order as it detected some cycles. Appending to list ensures
that dependencies will preceed dependees.

10 years agoCmake: Fix library list provided to cURL tests.
Jakub Zakrzewski [Thu, 21 Aug 2014 11:37:07 +0000 (13:37 +0200)]
Cmake: Fix library list provided to cURL tests.

The list must be set after those nice CMake tests as we mess with
CMAKE_REQUIRED_LIBRARIES there.

10 years agoCmake: Check for OpenSSL before OpenLDAP.
Jakub Zakrzewski [Thu, 21 Aug 2014 11:15:59 +0000 (13:15 +0200)]
Cmake: Check for OpenSSL before OpenLDAP.

OpenLDAP might have been build with OpenSSL. Checking for OpenLDAP first
may result in undefined symbols. Of course, the found OpenSSL libraries
must also be linked whenever OpenLDAP is.

10 years agocurl_multi_fdset.3: improved the formatting slightly
Daniel Stenberg [Thu, 9 Oct 2014 11:41:13 +0000 (13:41 +0200)]
curl_multi_fdset.3: improved the formatting slightly

10 years agocurl_multi_fdset: explain the fd_set arguments
Daniel Stenberg [Thu, 9 Oct 2014 11:17:27 +0000 (13:17 +0200)]
curl_multi_fdset: explain the fd_set arguments

10 years agonss: do not fail if a CRL is already cached
Kamil Dudka [Wed, 8 Oct 2014 15:13:59 +0000 (17:13 +0200)]
nss: do not fail if a CRL is already cached

This fixes a copy-paste mistake from commit 2968f957.

10 years agoOS400: upgrade interface for pinned public key (no implementation yet)
Patrick Monnerat [Wed, 8 Oct 2014 13:47:04 +0000 (15:47 +0200)]
OS400: upgrade interface for pinned public key (no implementation yet)

10 years agoFormAdd: precaution against memdup() of NULL pointer
Daniel Stenberg [Wed, 8 Oct 2014 11:53:41 +0000 (13:53 +0200)]
FormAdd: precaution against memdup() of NULL pointer

Coverity CID 252518. This function is in general far too complicated for
its own good and really should be broken down into several smaller
funcitons instead - but I'm adding this protection here now since it
seems there's a risk the code flow can end up here and dereference a
NULL pointer.

10 years agooperate: avoid NULL dereference
Daniel Stenberg [Wed, 8 Oct 2014 11:18:55 +0000 (13:18 +0200)]
operate: avoid NULL dereference

Coverity CID 1241948. dumpeasysrc() would get called with
config->current set to NULL which could be dereferenced by a warnf()
call.

10 years agodo_sec_send: remove dead code
Daniel Stenberg [Wed, 8 Oct 2014 10:48:06 +0000 (12:48 +0200)]
do_sec_send: remove dead code

Coverity CID 1241951. The condition 'len >= 0' would always be true at
that point and thus not necessary to check for.

10 years agokrb5_encode: remove unused argument
Daniel Stenberg [Wed, 8 Oct 2014 10:25:07 +0000 (12:25 +0200)]
krb5_encode: remove unused argument

Coverity CID 1241957. Removed the unused argument. As this struct and
pointer now are used only for krb5, there's no need to keep unused
function arguments around.

10 years agooperate_do: skip superfluous check for NULL pointer
Daniel Stenberg [Wed, 8 Oct 2014 10:21:39 +0000 (12:21 +0200)]
operate_do: skip superfluous check for NULL pointer

Coverity CID 1243583. get_url_file_name() cannot fail and return a NULL
file name pointer so skip the check for that - it tricks coverity into
believing it can happen and it then warns later on when we use 'outfile'
without checking for NULL.

10 years agocurl_easy_getinfo.3: spell-fix
Daniel Stenberg [Tue, 7 Oct 2014 13:48:37 +0000 (15:48 +0200)]
curl_easy_getinfo.3: spell-fix

Reported-By: Luan Cestari
10 years agoGnuTLS: Implement public key pinning
moparisthebest [Wed, 1 Oct 2014 06:14:49 +0000 (02:14 -0400)]
GnuTLS: Implement public key pinning

10 years agoSSL: implement public key pinning
moparisthebest [Wed, 1 Oct 2014 02:31:17 +0000 (22:31 -0400)]
SSL: implement public key pinning

Option --pinnedpubkey takes a path to a public key in DER format and
only connect if it matches (currently only implemented with OpenSSL).

Provides CURLOPT_PINNEDPUBLICKEY for curl_easy_setopt().

Extract a public RSA key from a website like so:
openssl s_client -connect google.com:443 2>&1 < /dev/null | \
sed -n '/-----BEGIN/,/-----END/p' | openssl x509 -noout -pubkey \
| openssl rsa -pubin -outform DER > google.com.der

10 years agomulti_runsingle: fix possible memory leak
Daniel Stenberg [Tue, 7 Oct 2014 11:57:13 +0000 (13:57 +0200)]
multi_runsingle: fix possible memory leak

Coverity CID 1202837. 'newurl' can in fact be allocated even when
Curl_retry_request() returns failure so free it if need be.

10 years agoares::Curl_resolver_cancel: skip checking for NULL conn
Daniel Stenberg [Tue, 7 Oct 2014 11:33:43 +0000 (13:33 +0200)]
ares::Curl_resolver_cancel: skip checking for NULL conn

Coverity CID 1243581. 'conn' will never be NULL here, and if it would be
the subsequent statement would dereference it!

10 years agoparseconfig: skip a NULL check
Daniel Stenberg [Tue, 7 Oct 2014 08:29:06 +0000 (10:29 +0200)]
parseconfig: skip a NULL check

Coverity CID 1154198. This NULL check implies that the pointer _can_ be
NULL at this point, which it can't. Thus it is dead code. It tricks
static analyzers to warn about dereferencing the pointer since the code
seems to imply it can be NULL.

10 years agomulti-uv.c: call curl_multi_info_read() better
Waldek Kozba [Tue, 7 Oct 2014 07:59:59 +0000 (09:59 +0200)]
multi-uv.c: call curl_multi_info_read() better

Improves it for low-latency cases (like the communication with
localhost)

10 years agotool_go_sleep: use (void) to spell out we ignore the return value
Daniel Stenberg [Mon, 6 Oct 2014 06:53:12 +0000 (08:53 +0200)]
tool_go_sleep: use (void) to spell out we ignore the return value

Coverity CID 1222080.

10 years agossh_statemach_act: split out assignment from check
Daniel Stenberg [Mon, 6 Oct 2014 06:49:43 +0000 (08:49 +0200)]
ssh_statemach_act: split out assignment from check

just a minor code style thing to make the code clearer

10 years agocurl_schannel.c: Fixed possible memory or handle leak
Marc Hoersken [Sat, 4 Oct 2014 16:24:23 +0000 (18:24 +0200)]
curl_schannel.c: Fixed possible memory or handle leak

First try to fix possible memory leaks, in this case:
Only connssl->ctxt xor onnssl->cred being initialized.

10 years agogetparameter: remove dead code
Daniel Stenberg [Sat, 4 Oct 2014 14:14:39 +0000 (16:14 +0200)]
getparameter: remove dead code

Coverity CID 1061126. 'parse' will always be non-NULL here.

10 years agogetparameter: comment a switch FALLTHROUGH
Daniel Stenberg [Sat, 4 Oct 2014 14:12:06 +0000 (16:12 +0200)]
getparameter: comment a switch FALLTHROUGH

Coverity CID 1061118. Point out that it is on purpose.

10 years agochoose_mech: fix return code
Daniel Stenberg [Sat, 4 Oct 2014 13:37:42 +0000 (15:37 +0200)]
choose_mech: fix return code

Coverity CID 1241950. The pointer is never NULL but it might point to
NULL.

10 years agoCurl_sec_read_msg: spell out that we ignore return code
Daniel Stenberg [Sat, 4 Oct 2014 13:21:39 +0000 (15:21 +0200)]
Curl_sec_read_msg: spell out that we ignore return code

Coverity CID 1241947. Since if sscanf() fails, the previously set value
remains set.

10 years agononblock: call with (void) to show we ignore the return code
Daniel Stenberg [Sat, 4 Oct 2014 13:14:27 +0000 (15:14 +0200)]
nonblock: call with (void) to show we ignore the return code

Coverity pointed out several of these.

10 years agoparse_proxy: remove dead code.
Daniel Stenberg [Fri, 3 Oct 2014 21:51:19 +0000 (23:51 +0200)]
parse_proxy: remove dead code.

Coverity CID 982331.

10 years agoCurl_debug: document switch fallthroughs
Daniel Stenberg [Fri, 3 Oct 2014 21:49:39 +0000 (23:49 +0200)]
Curl_debug: document switch fallthroughs

10 years agocurl_multi_remove_handle: remove dead code
Daniel Stenberg [Fri, 3 Oct 2014 21:46:10 +0000 (23:46 +0200)]
curl_multi_remove_handle: remove dead code

Coverify CID 1157776. Removed a superfluous if() that always evaluated
true (and an else clause that never ran), and then re-indented the
function accordingly.

10 years agoCurl_pipeline_server_blacklisted: handle a NULL server name
Daniel Stenberg [Fri, 3 Oct 2014 21:40:57 +0000 (23:40 +0200)]
Curl_pipeline_server_blacklisted: handle a NULL server name

Coverity CID 1215284. The server name is extracted with
Curl_copy_header_value() and passed in to this function, and
copy_header_value can actually can fail and return NULL.

10 years agossh: comment "fallthrough" in switch statement
Daniel Stenberg [Fri, 3 Oct 2014 21:30:05 +0000 (23:30 +0200)]
ssh: comment "fallthrough" in switch statement

10 years agossh: improve key file search
Jeremy Lin [Tue, 16 Sep 2014 04:16:46 +0000 (21:16 -0700)]
ssh: improve key file search

For private keys, use the first match from: user-specified key file
(if provided), ~/.ssh/id_rsa, ~/.ssh/id_dsa, ./id_rsa, ./id_dsa

Note that the previous code only looked for id_dsa files. id_rsa is
now generally preferred, as it supports larger key sizes.

For public keys, use the user-specified key file, if provided.
Otherwise, try to extract the public key from the private key file.
This means that passing --pubkey is typically no longer required,
and makes the key-handling behavior more like OpenSSH.

10 years agoCURLOPT_HTTPHEADER.3: libcurl doesn't copy the whole list
Daniel Stenberg [Fri, 3 Oct 2014 11:35:40 +0000 (13:35 +0200)]
CURLOPT_HTTPHEADER.3: libcurl doesn't copy the whole list

10 years agodetect_proxy: fix possible single-byte memory leak
Daniel Stenberg [Thu, 2 Oct 2014 21:31:01 +0000 (23:31 +0200)]
detect_proxy: fix possible single-byte memory leak

Coverity CID 1202836. If the proxy environment variable returned an empty
string, it would be leaked. While an empty string is not really a proxy, other
logic in this function already allows a blank string to be returned so allow
that here to avoid the leak.

10 years agomulti_runsingle: fix memory leak
Daniel Stenberg [Thu, 2 Oct 2014 21:22:01 +0000 (23:22 +0200)]
multi_runsingle: fix memory leak

Coverity CID 1202837. There's a potential risk that 'newurl' gets
overwritten when it was already pointing to allocated memory.

10 years agopop3_perform_authentication: fix memory leak
Daniel Stenberg [Thu, 2 Oct 2014 21:07:06 +0000 (23:07 +0200)]
pop3_perform_authentication: fix memory leak

Coverity CID 1215287. There's a potential risk for a memory leak in
here, and moving the free call to be unconditional seems like a cheap
price to remove the risk.

10 years agoimap_perform_authentication: fix memory leak
Daniel Stenberg [Thu, 2 Oct 2014 21:01:45 +0000 (23:01 +0200)]
imap_perform_authentication: fix memory leak

Coverity CID 1215296. There's a potential risk for a memory leak in
here, and moving the free call to be unconditional seems like a cheap
price to remove the risk.

10 years agowait_or_timeout: return failure when Curl_poll() fails
Daniel Stenberg [Thu, 2 Oct 2014 20:52:23 +0000 (22:52 +0200)]
wait_or_timeout: return failure when Curl_poll() fails

Coverity detected this. CID 1241954. When Curl_poll() returns a negative value
'mcode' was uninitialized. Pretty harmless since this is debug code only and
would at worst cause an error to _not_ be returned...

10 years agocurl.1: mention quoting in the URL section
Daniel Stenberg [Wed, 1 Oct 2014 06:29:43 +0000 (08:29 +0200)]
curl.1: mention quoting in the URL section

and separate the example URLs with newlines

10 years agosmtp: Fixed intermittent "SSL3_WRITE_PENDING: bad write retry" error
Bill Nagel [Fri, 26 Sep 2014 18:55:01 +0000 (18:55 +0000)]
smtp: Fixed intermittent "SSL3_WRITE_PENDING: bad write retry" error

This patch fixes the "SSL3_WRITE_PENDING: bad write retry" error that
sometimes occurs when sending an email over SMTPS with OpenSSL. OpenSSL
appears to require the same pointer on a write that follows a retry
(CURLE_AGAIN) as discussed here:

http://stackoverflow.com/questions/2997218/why-am-i-getting-error1409f07fssl-routinesssl3-write-pending-bad-write-retr

10 years agoRELEASE-NOTES: synced with 53cbea22310f15
Daniel Stenberg [Tue, 30 Sep 2014 13:18:02 +0000 (15:18 +0200)]
RELEASE-NOTES: synced with 53cbea22310f15

10 years agofile: reject paths using embedded %00
Daniel Stenberg [Thu, 25 Sep 2014 11:44:24 +0000 (13:44 +0200)]
file: reject paths using embedded %00

Mostly because we use C strings and they end at a binary zero so we know
we can't open a file name using an embedded binary zero.

Reported-by: research@g0blin.co.uk
10 years agotest506: Fixed a couple of memory leaks in test
Dan Fandrich [Fri, 26 Sep 2014 04:41:23 +0000 (06:41 +0200)]
test506: Fixed a couple of memory leaks in test

10 years agoCURLOPT_COOKIELIST: Added "RELOAD" command
Yousuke Kimoto [Thu, 25 Sep 2014 14:25:52 +0000 (16:25 +0200)]
CURLOPT_COOKIELIST: Added "RELOAD" command

10 years agoCURLOPT_POSTREDIR.3: Added availability for CURL_REDIR_POST_303
Michael Wallner [Thu, 25 Sep 2014 13:11:04 +0000 (15:11 +0200)]
CURLOPT_POSTREDIR.3: Added availability for CURL_REDIR_POST_303

10 years agothreaded-resolver: revert Curl_expire_latest() switch
Daniel Stenberg [Tue, 23 Sep 2014 09:44:03 +0000 (11:44 +0200)]
threaded-resolver: revert Curl_expire_latest() switch

The switch to using Curl_expire_latest() in commit cacdc27f52b was a
mistake and was against the advice even mentioned in that commit. The
comparison in asyn-thread.c:Curl_resolver_is_resolved() makes
Curl_expire() the suitable function to use.

Bug: http://curl.haxx.se/bug/view.cgi?id=1426
Reported-By: graysky
10 years agolibcurl docs: improvements all over
Daniel Stenberg [Fri, 19 Sep 2014 13:07:58 +0000 (15:07 +0200)]
libcurl docs: improvements all over

10 years agobuild: Added WinIDN build configuration options
Steve Holme [Fri, 19 Sep 2014 11:43:10 +0000 (12:43 +0100)]
build: Added WinIDN build configuration options

Added initial support for WinIDN build configurations to the VC10+
project files.

10 years agotutorial: signals aren't used for the threaded resolver
Daniel Stenberg [Fri, 19 Sep 2014 10:54:19 +0000 (12:54 +0200)]
tutorial: signals aren't used for the threaded resolver

10 years agoFAQ: update the pronunciation section
Daniel Stenberg [Fri, 19 Sep 2014 08:01:45 +0000 (10:01 +0200)]
FAQ: update the pronunciation section

As we weren't using the correct phonetic description and doing it correctly
involves funny letters that I'm sure will cause problems for people in a text
document so I instead rephrased it and link to a WAV file with a person
actually saying 'curl'.

Reported-By: Dimitar Boevski
10 years agoCURLOPT_COOKIE*: added more cross-references
Daniel Stenberg [Thu, 18 Sep 2014 20:58:12 +0000 (22:58 +0200)]
CURLOPT_COOKIE*: added more cross-references

10 years agoBINDINGS: add node-libcurl
Daniel Stenberg [Thu, 18 Sep 2014 07:05:29 +0000 (09:05 +0200)]
BINDINGS: add node-libcurl

Reported-By: Jonathan Cardoso Machado
URL: http://curl.haxx.se/mail/lib-2014-09/0102.html

10 years agoREADME.http2: updated to reflect current status
Daniel Stenberg [Mon, 15 Sep 2014 21:13:35 +0000 (23:13 +0200)]
README.http2: updated to reflect current status

10 years agoformdata: removed unnecessary USE_SSLEAY use
Daniel Stenberg [Sat, 13 Sep 2014 17:09:56 +0000 (19:09 +0200)]
formdata: removed unnecessary USE_SSLEAY use

10 years agocurlssl: make tls backend symbols use curlssl in the name
Daniel Stenberg [Sat, 13 Sep 2014 13:31:12 +0000 (15:31 +0200)]
curlssl: make tls backend symbols use curlssl in the name

10 years agourl: let the backend decide CURLOPT_SSL_CTX_ support
Daniel Stenberg [Sat, 13 Sep 2014 13:28:08 +0000 (15:28 +0200)]
url: let the backend decide CURLOPT_SSL_CTX_ support

... to further remove specific TLS backend knowledge from url.c

10 years agovtls: have the backend tell if it supports CERTINFO
Daniel Stenberg [Sat, 13 Sep 2014 13:11:26 +0000 (15:11 +0200)]
vtls: have the backend tell if it supports CERTINFO

10 years agoconfigure: allow --with-ca-path with PolarSSL too
Catalin Patulea [Sat, 13 Sep 2014 05:58:04 +0000 (07:58 +0200)]
configure: allow --with-ca-path with PolarSSL too

Missed this in af45542c.

Signed-off-by: Catalin Patulea <cat@vv.carleton.ca>
10 years agoCURLOPT_CAPATH: return failure if set without backend support
Daniel Stenberg [Sat, 13 Sep 2014 12:56:27 +0000 (14:56 +0200)]
CURLOPT_CAPATH: return failure if set without backend support

10 years agohttp2: Fix busy loop when EOF is encountered
Tatsuhiro Tsujikawa [Sat, 13 Sep 2014 02:59:23 +0000 (11:59 +0900)]
http2: Fix busy loop when EOF is encountered

Previously we did not handle EOF from underlying transport socket and
wrongly just returned error code CURL_AGAIN from http2_recv, which
caused busy loop since socket has been closed.  This patch adds the
code to handle EOF situation and tells the upper layer that we got
EOF.

10 years agobuild: Added batch wrapper to checksrc.pl
Steve Holme [Sat, 13 Sep 2014 10:24:07 +0000 (11:24 +0100)]
build: Added batch wrapper to checksrc.pl

10 years agoRELEASE-NOTES: Synced with bd3df5ec6d
Steve Holme [Sat, 13 Sep 2014 09:23:36 +0000 (10:23 +0100)]
RELEASE-NOTES: Synced with bd3df5ec6d

10 years agosasl_sspi: Fixed Unicode build
Marcel Raad [Wed, 10 Sep 2014 12:32:56 +0000 (14:32 +0200)]
sasl_sspi: Fixed Unicode build

Bug: http://curl.haxx.se/bug/view.cgi?id=1422
Verified-by: Steve Holme
10 years agolibcurl-tutorial.3: fix GnuTLS link to thread-safety guidelines
Daniel Stenberg [Fri, 12 Sep 2014 19:02:12 +0000 (21:02 +0200)]
libcurl-tutorial.3: fix GnuTLS link to thread-safety guidelines

The former link was turned into a 404 at some point.

Reported-By: Askar Safin
10 years agocontributors.sh: split list of names at comma
Daniel Stenberg [Fri, 12 Sep 2014 13:12:06 +0000 (15:12 +0200)]
contributors.sh: split list of names at comma

... to support a list of names provided in a commit message.

10 years agontlm: Fixed HTTP proxy authentication when using Windows SSPI
Ulrich Telle [Fri, 12 Sep 2014 10:22:56 +0000 (12:22 +0200)]
ntlm: Fixed HTTP proxy authentication when using Windows SSPI

Removed ISC_REQ_* flags from calls to InitializeSecurityContext to fix
bug in NTLM handshake for HTTP proxy authentication.

NTLM handshake for HTTP proxy authentication failed with error
SEC_E_INVALID_TOKEN from InitializeSecurityContext for certain proxy
servers on generating the NTLM Type-3 message.

The flag ISC_REQ_CONFIDENTIALITY seems to cause the problem according
to the observations and suggestions made in a bug report for the
QT project (https://bugreports.qt-project.org/browse/QTBUG-17322).

Removing all the flags solved the problem.

Bug: http://curl.haxx.se/mail/lib-2014-08/0273.html
Reported-by: Ulrich Telle
Assisted-by: Steve Holme, Daniel Stenberg
10 years agonewlines: fix mixed newlines to LF-only
Ray Satiro [Fri, 12 Sep 2014 08:22:34 +0000 (10:22 +0200)]
newlines: fix mixed newlines to LF-only

I use the curl repo mainly on Windows with the typical Windows git
checkout which converts the LF line endings in the curl repo to CRLF
automatically on checkout. The automatic conversion is not done on files
in the repo with mixed line endings. I recently noticed some weird
output with projects/build-openssl.bat that I traced back to mixed line
endings, so I scanned the repo and there are files (excluding the
test data) that have mixed line endings.

I used this command below to do the scan. Unfortunately it's not as easy
as git grep, at least not on Windows. This gets the names of all the
files in the repo's HEAD, gets each of those files raw from HEAD, checks
for mixed line endings of both LF and CRLF, and prints the name if
mixed. I excluded path tests/data/test* because those can have mixed
line endings if I understand correctly.

for f in `git ls-tree --name-only --full-tree -r HEAD`;
do if [ -n "${f##tests/data/test*}" ];
    then git show "HEAD:$f" | \
        perl -0777 -ne 'exit 1 if /([^\r]\n.*\r\n)|(\r\n.*[^\r]\n)/';
    if [ $? -ne 0 ];
        then echo "$f";
    fi;
fi;
done

10 years agomk-ca-bundle.pl: converted tabs to spaces, deleted trailing spaces
Viktor Szakáts [Thu, 11 Sep 2014 14:26:26 +0000 (16:26 +0200)]
mk-ca-bundle.pl: converted tabs to spaces, deleted trailing spaces

10 years agoROADMAP: markdown eats underscores
Daniel Stenberg [Thu, 11 Sep 2014 08:56:20 +0000 (10:56 +0200)]
ROADMAP: markdown eats underscores

It interprets them as italic indictors unless we backtick the word.

10 years agoROADMAP: tiny formatting edit for nicer web output
Daniel Stenberg [Wed, 10 Sep 2014 22:15:12 +0000 (00:15 +0200)]
ROADMAP: tiny formatting edit for nicer web output

10 years agoROADMAP.md: Updated GSSAPI authentication following 7.38.0 additions
Steve Holme [Wed, 10 Sep 2014 21:39:49 +0000 (22:39 +0100)]
ROADMAP.md: Updated GSSAPI authentication following 7.38.0 additions

10 years agoINTERNALS: Added email and updated Kerberos details
Steve Holme [Wed, 10 Sep 2014 21:10:20 +0000 (22:10 +0100)]
INTERNALS: Added email and updated Kerberos details

10 years agoFEATURES: Updated Kerberos details
Steve Holme [Wed, 10 Sep 2014 21:02:38 +0000 (22:02 +0100)]
FEATURES: Updated Kerberos details

Added support for Kerberos 5 to the email protocols following the recent
additions in 7.38.0.

Removed Kerberos 4 as this has been gone for a while now.

10 years agoopenssl: build fix for versions < 0.9.8e
Paul Howarth [Wed, 10 Sep 2014 11:08:22 +0000 (13:08 +0200)]
openssl: build fix for versions < 0.9.8e

Bug: http://curl.haxx.se/mail/lib-2014-09/0064.html

10 years agomk-ca-bundle.pl: first, try downloading HTTPS with curl
Daniel Stenberg [Wed, 10 Sep 2014 10:14:13 +0000 (12:14 +0200)]
mk-ca-bundle.pl: first, try downloading HTTPS with curl

As a sort of step forward, this script will now first try to get the
data from the HTTPS URL using curl, and only if that fails it will
switch back to the HTTP transfer using perl's native LWP functionality.
To reduce the risk of this script being tricked.

Using HTTPS to get a cert bundle introduces a chicken-and-egg problem so
we can't really ever completely disable HTTP, but chances are that most
users already have a ca cert bundle that trusts the mozilla.org site
that this script downloads from.

A future version of this script will probably switch to require a
dedicated "insecure" command line option to allow downloading over HTTP
(or unverified HTTPS).

10 years agoLICENSE-MIXING: removed krb4 info
Daniel Stenberg [Wed, 10 Sep 2014 08:38:31 +0000 (10:38 +0200)]
LICENSE-MIXING: removed krb4 info

krb4 has been dropped since a while now

10 years agobump: on the 7.38.1-DEV train now!
Daniel Stenberg [Wed, 10 Sep 2014 08:15:30 +0000 (10:15 +0200)]
bump: on the 7.38.1-DEV train now!

10 years agoSSLCERTS: minor updates
Daniel Stenberg [Wed, 10 Sep 2014 08:13:04 +0000 (10:13 +0200)]
SSLCERTS: minor updates

Edited format to look better on the web, added a "it is about trust"
section.

10 years agodist: two cmake files are no more curl-7_38_0
Daniel Stenberg [Wed, 10 Sep 2014 06:06:44 +0000 (08:06 +0200)]
dist: two cmake files are no more

CMake/FindOpenSSL.cmake and FindZLIB.cmake are gone since 14aa8f0c117b

10 years agoRELEASE-NOTES: final update for 7.38.0
Daniel Stenberg [Tue, 9 Sep 2014 06:49:03 +0000 (08:49 +0200)]
RELEASE-NOTES: final update for 7.38.0

10 years agocookies: reject incoming cookies set for TLDs
Daniel Stenberg [Tue, 19 Aug 2014 19:11:20 +0000 (21:11 +0200)]
cookies: reject incoming cookies set for TLDs

Test 61 was modified to verify this.

CVE-2014-3620

Reported-by: Tim Ruehsen
URL: http://curl.haxx.se/docs/adv_20140910B.html

10 years agocookies: only use full host matches for hosts used as IP address
Tim Ruehsen [Tue, 19 Aug 2014 19:01:28 +0000 (21:01 +0200)]
cookies: only use full host matches for hosts used as IP address

By not detecting and rejecting domain names for partial literal IP
addresses properly when parsing received HTTP cookies, libcurl can be
fooled to both send cookies to wrong sites and to allow arbitrary sites
to set cookies for others.

CVE-2014-3613

Bug: http://curl.haxx.se/docs/adv_20140910A.html

10 years agoHISTORY: fix the 1998 title position
Daniel Stenberg [Tue, 9 Sep 2014 22:40:11 +0000 (00:40 +0200)]
HISTORY: fix the 1998 title position

10 years agoHISTORY: extended and now markdown
Daniel Stenberg [Tue, 9 Sep 2014 22:34:32 +0000 (00:34 +0200)]
HISTORY: extended and now markdown

10 years agoSSLCERTS: converted to markdown
Daniel Stenberg [Tue, 9 Sep 2014 21:46:58 +0000 (23:46 +0200)]
SSLCERTS: converted to markdown

Only minor edits to make it generate nice HTML output using markdown, as
this document serves both in source release tarballs as on the web site.

URL: http://curl.haxx.se/docs/sslcerts.html

10 years agoftp-wildcard.c: spell fix
Daniel Stenberg [Tue, 9 Sep 2014 09:10:04 +0000 (11:10 +0200)]
ftp-wildcard.c: spell fix

Reported-By: Frank Gevaerts
10 years agoRELEASE-NOTES: synced with 921a0c22a6f
Daniel Stenberg [Mon, 8 Sep 2014 08:26:43 +0000 (10:26 +0200)]
RELEASE-NOTES: synced with 921a0c22a6f

10 years agoTHANKS: synced with RELEASE-NOTES for 921a0c22a6f
Daniel Stenberg [Mon, 8 Sep 2014 08:26:32 +0000 (10:26 +0200)]
THANKS: synced with RELEASE-NOTES for 921a0c22a6f

10 years agopolarassl: avoid memset() when clearing the first byte is enough
Daniel Stenberg [Mon, 8 Sep 2014 08:11:34 +0000 (10:11 +0200)]
polarassl: avoid memset() when clearing the first byte is enough

10 years agopolarssl: support CURLOPT_CAPATH / --capath
Catalin Patulea [Sun, 7 Sep 2014 01:31:01 +0000 (03:31 +0200)]
polarssl: support CURLOPT_CAPATH / --capath

Signed-off-by: Catalin Patulea <cat@vv.carleton.ca>
10 years agoSECURITY: eh, make more sense!
Daniel Stenberg [Mon, 8 Sep 2014 08:00:18 +0000 (10:00 +0200)]
SECURITY: eh, make more sense!

10 years agoSECURITY: how to join the curl-security list
Daniel Stenberg [Mon, 8 Sep 2014 07:39:14 +0000 (09:39 +0200)]
SECURITY: how to join the curl-security list

10 years agoRELEASE-NOTES: fix the required nghttp2 version typo
Daniel Stenberg [Sun, 7 Sep 2014 22:12:18 +0000 (00:12 +0200)]
RELEASE-NOTES: fix the required nghttp2 version typo

10 years agoEnsure progress.size_dl/progress.size_ul are always >= 0
Brandon Casey [Fri, 29 Aug 2014 21:48:03 +0000 (23:48 +0200)]
Ensure progress.size_dl/progress.size_ul are always >= 0

Historically the default "unknown" value for progress.size_dl and
progress.size_ul has been zero, since these values are initialized
implicitly by the calloc that allocates the curl handle that these
variables are a part of.  Users of curl that install progress
callbacks may expect these values to always be >= 0.

Currently it is possible for progress.size_dl and progress.size_ul
to by set to a value of -1, if Curl_pgrsSetDownloadSize() or
Curl_pgrsSetUploadSize() are passed a "size" of -1 (which a few
places currently do, and a following patch will add more).  So
lets update Curl_pgrsSetDownloadSize() and Curl_pgrsSetUploadSize()
so they make sure that these variables always contain a value that
is >= 0.

Updates test579 and test599.

Signed-off-by: Brandon Casey <drafnel@gmail.com>