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#3178Closes#3180
- 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
Fixes https://github.com/curl/curl/issues/2969
Closes https://github.com/curl/curl/pull/3012
- 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
Closes https://github.com/curl/curl/issues/2980
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#2724Closes#2791
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
Commit 38203f1585 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
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#2691Closes#2694
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#2641Closes#2644
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
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
... instead of previous separate struct fields, to make it easier to
extend and change individual backends without having to modify them all.
closes#2547
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
This reverts commit dc85437736.
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
When peer verification is disabled, calling
SSL_CTX_load_verify_locations is not necessary. Only call it when
verification is enabled to save resources and increase performance.
Closes#2290
Follow-up to 84fcaa2e7. libressl does not have the API even if it says it is
late OpenSSL version...
Fixes#2246Closes#2247
Reported-by: jungle-boogie on github
Prior to this change SSLKEYLOGFILE used line buffering on WIN32 just
like it does for other platforms. However, the Windows CRT does not
actually support line buffering (_IOLBF) and will use full buffering
(_IOFBF) instead. We can't use full buffering because multiple processes
may be writing to the file and that could lead to corruption, and since
full buffering is the only buffering available this commit disables
buffering for Windows SSLKEYLOGFILE entirely (_IONBF).
Ref: https://github.com/curl/curl/pull/1346#issuecomment-350530901
- Allow proxy_ssl to be checked for pending data even when connssl does
not yet have an SSL handle.
This change is for posterity. Currently there doesn't seem to be a code
path that will cause a pending data check when proxyssl could have
pending data and the connssl handle doesn't yet exist [1].
[1]: Recall that an https proxy connection starts out in connssl but if
the destination is also https then the proxy SSL backend data is moved
from connssl to proxyssl, which means connssl handle is temporarily
empty until an SSL handle for the destination can be created.
Ref: https://github.com/curl/curl/commit/f4a6238#commitcomment-24396542
Closes https://github.com/curl/curl/pull/1916
commit d3ab7c5a21 broke the boringssl build since it doesn't have
RSA_flags(), so we disable that code block for boringssl builds.
Reported-by: W. Mark Kubacki
Fixes#2117
... since the 'tv' stood for timeval and this function does not return a
timeval struct anymore.
Also, cleaned up the Curl_timediff*() functions to avoid typecasts and
clean up the descriptive comments.
Closes#2011
... to cater for systems with unsigned time_t variables.
- Renamed the functions to curlx_timediff and Curl_timediff_us.
- Added overflow protection for both of them in either direction for
both 32 bit and 64 bit time_ts
- Reprefixed the curlx_time functions to use Curl_*
Reported-by: Peter Piekarski
Fixes#2004Closes#2005
Those were temporary things we'd add and remove for our own convenience
long ago. The last few stayed around for too long as an oversight but
have since been removed. These days we have a running
BORINGSSL_API_VERSION counter which is bumped when we find it
convenient, but 2015-11-19 was quite some time ago, so just check
OPENSSL_IS_BORINGSSL.
Closes#1979
In some cases the RSA key does not support verifying it because it's
located on a smart card, an engine wants to hide it, ...
Check the flags on the key before trying to verify it.
OpenSSL does the same thing internally; see ssl/ssl_rsa.c
Closes#1904
lib/vtls/openssl.c uses OpenSSL APIs from BUF_MEM and BIO APIs. Include
their headers directly rather than relying on other OpenSSL headers
including things.
Closes https://github.com/curl/curl/pull/1891
Another mistake in my manual fixups of the largely mechanical
search-and-replace ("connssl->" -> "BACKEND->"), just like the previous
commit concerning HTTPS proxies (and hence not caught during my
earlier testing).
Fixes#1855Closes#1871
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
In d65e6cc4f (vtls: prepare the SSL backends for encapsulated private
data, 2017-06-21), this developer prepared for a separation of the
private data of the SSL backends from the general connection data.
This conversion was partially automated (search-and-replace) and
partially manual (e.g. proxy_ssl's backend data).
Sadly, there was a crucial error in the manual part, where the wrong
handle was used: rather than connecting ssl[sockindex]' BIO to the
proxy_ssl[sockindex]', we reconnected proxy_ssl[sockindex]. The reason
was an incorrect location to paste "BACKEND->"... d'oh.
Reported by Jay Satiro in https://github.com/curl/curl/issues/1855.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
This is an adaptation of 2 of Peter Wu's SSLKEYLOGFILE implementations.
The first one, written for old OpenSSL versions:
https://git.lekensteyn.nl/peter/wireshark-notes/tree/src/sslkeylog.c
The second one, written for BoringSSL and new OpenSSL versions:
https://github.com/curl/curl/pull/1346
Note the first one is GPL licensed but the author gave permission to
waive that license for libcurl.
As of right now this feature is disabled by default, and does not have
a configure option to enable it. To enable this feature define
ENABLE_SSLKEYLOGFILE when building libcurl and set environment
variable SSLKEYLOGFILE to a pathname that will receive the keys.
And in Wireshark change your preferences to point to that key file:
Edit > Preferences > Protocols > SSL > Master-Secret
Co-authored-by: Peter Wu
Ref: https://github.com/curl/curl/pull/1030
Ref: https://github.com/curl/curl/pull/1346
Closes https://github.com/curl/curl/pull/1866
Up2date versions of OpenSSL maintain the default reasonably secure
without breaking compatibility, so it is better not to override the
default by curl. Suggested at https://bugzilla.redhat.com/1483972Closes#1846
There is information about the compiled-in SSL backends that is really
no concern of any code other than the SSL backend itself, such as which
function (if any) implements SHA-256 summing.
And there is information that is really interesting to the user, such as
the name, or the curl_sslbackend value.
Let's factor out the latter into a publicly visible struct. This
information will be used in the upcoming API to set the SSL backend
globally.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
When building software for the masses, it is sometimes not possible to
decide for all users which SSL backend is appropriate.
Git for Windows, for example, uses cURL to perform clones, fetches and
pushes via HTTPS, and some users strongly prefer OpenSSL, while other
users really need to use Secure Channel because it offers
enterprise-ready tools to manage credentials via Windows' Credential
Store.
The current Git for Windows versions use the ugly work-around of
building libcurl once with OpenSSL support and once with Secure Channel
support, and switching out the binaries in the installer depending on
the user's choice.
Needless to say, this is a super ugly workaround that actually only
works in some cases: Git for Windows also comes in a portable form, and
in a form intended for third-party applications requiring Git
functionality, in which cases this "swap out libcurl-4.dll" simply is
not an option.
Therefore, the Git for Windows project has a vested interest in teaching
cURL to make the SSL backend a *runtime* option.
This patch makes that possible.
By running ./configure with multiple --with-<backend> options, cURL will
be built with multiple backends.
For the moment, the backend can be configured using the environment
variable CURL_SSL_BACKEND (valid values are e.g. "openssl" and
"schannel").
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
So far, all of the SSL backends' private data has been declared as
part of the ssl_connect_data struct, in one big #if .. #elif .. #endif
block.
This can only work as long as the SSL backend is a compile-time option,
something we want to change in the next commits.
Therefore, let's encapsulate the exact data needed by each SSL backend
into a private struct, and let's avoid bleeding any SSL backend-specific
information into urldata.h. This is also necessary to allow multiple SSL
backends to be compiled in at the same time, as e.g. OpenSSL's and
CyaSSL's headers cannot be included in the same .c file.
To avoid too many malloc() calls, we simply append the private structs
to the connectdata struct in allocate_conn().
This requires us to take extra care of alignment issues: struct fields
often need to be aligned on certain boundaries e.g. 32-bit values need to
be stored at addresses that divide evenly by 4 (= 32 bit / 8
bit-per-byte).
We do that by assuming that no SSL backend's private data contains any
fields that need to be aligned on boundaries larger than `long long`
(typically 64-bit) would need. Under this assumption, we simply add a
dummy field of type `long long` to the `struct connectdata` struct. This
field will never be accessed but acts as a placeholder for the four
instances of ssl_backend_data instead. the size of each ssl_backend_data
struct is stored in the SSL backend-specific metadata, to allow
allocate_conn() to know how much extra space to allocate, and how to
initialize the ssl[sockindex]->backend and proxy_ssl[sockindex]->backend
pointers.
This would appear to be a little complicated at first, but is really
necessary to encapsulate the private data of each SSL backend correctly.
And we need to encapsulate thusly if we ever want to allow selecting
CyaSSL and OpenSSL at runtime, as their headers cannot be included within
the same .c file (there are just too many conflicting definitions and
declarations for that).
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
At the moment, cURL's SSL backend needs to be configured at build time.
As such, it is totally okay for them to hard-code their backend-specific
data in the ssl_connect_data struct.
In preparation for making the SSL backend a runtime option, let's make
the access of said private data a bit more abstract so that it can be
adjusted later in an easy manner.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
In the ongoing endeavor to abstract out all SSL backend-specific
functionality, this is the next step: Instead of hard-coding how the
different SSL backends access their internal data in getinfo.c, let's
implement backend-specific functions to do that task.
This will also allow for switching SSL backends as a runtime option.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
These functions are all available via the Curl_ssl struct now, no need
to declare them separately anymore.
As the global declarations are removed, the corresponding function
definitions are marked as file-local. The only two exceptions here are
Curl_mbedtls_shutdown() and Curl_polarssl_shutdown(): only the
declarations were removed, there are no function definitions to mark
file-local.
Please note that Curl_nss_force_init() is *still* declared globally, as
the only SSL backend-specific function, because it was introduced
specifically for the use case where cURL was compiled with
`--without-ssl --with-nss`. For details, see f3b77e561 (http_ntlm: add
support for NSS, 2010-06-27).
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
The entire idea of introducing the Curl_ssl struct to describe SSL
backends is to prepare for choosing the SSL backend at runtime.
To that end, convert all the #ifdef have_curlssl_* style conditionals
to use bit flags instead.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
The SHA-256 checksumming is also an SSL backend-specific function.
Let's include it in the struct declaring the functionality of SSL
backends.
In contrast to MD5, there is no fall-back code. To indicate this, the
respective entries are NULL for those backends that offer no support for
SHA-256 checksumming.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
The MD5 summing is also an SSL backend-specific function. So let's
include it, offering the previous fall-back code as a separate function
now: Curl_none_md5sum(). To allow for that, the signature had to be
changed so that an error could be returned from the implementation
(Curl_none_md5sum() can run out of memory).
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
This is the first step to unify the SSL backend handling. Now all the
SSL backend-specific functionality is accessed via a global instance of
the Curl_ssl struct.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
The idea of introducing the Curl_ssl struct was to unify how the SSL
backends are declared and called. To this end, we now provide an
instance of the Curl_ssl struct for each and every SSL backend.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
... to make all libcurl internals able to use the same data types for
the struct members. The timeval struct differs subtly on several
platforms so it makes it cumbersome to use everywhere.
Ref: #1652Closes#1693
ERR_error_string with NULL parameter is not thread-safe. The library
writes the string into some static buffer. Two threads doing this at
once may clobber each other and run into problems. Switch to
ERR_error_string_n which avoids this problem and is explicitly
bounds-checked.
Also clean up some remnants of OpenSSL 0.9.5 around here. A number of
comments (fixed buffer size, explaining that ERR_error_string_n was
added in a particular version) date to when ossl_strerror tried to
support pre-ERR_error_string_n OpenSSLs.
Closes#1424
- If SSL_get_error is called but no extended error detail is available
then show that SSL_ERROR_* as a string.
Prior to this change there was some inconsistency in that case: the
SSL_ERROR_* code may or may not have been shown, or may have been shown
as unknown even if it was known.
Ref: https://github.com/curl/curl/issues/1300
Closes https://github.com/curl/curl/pull/1348
This commit introduces the CURL_SSLVERSION_MAX_* constants as well as
the --tls-max option of the curl tool.
Closes https://github.com/curl/curl/pull/1166
Mark intended fallthroughs with /* FALLTHROUGH */ so that gcc will know
it's expected and won't warn on [-Wimplicit-fallthrough=].
Closes https://github.com/curl/curl/pull/1297
SSL_CTX_add_extra_chain_cert takes ownership of the given certificate
while, despite the similar name, SSL_CTX_add_client_CA does not. Thus
it's best to call SSL_CTX_add_client_CA before
SSL_CTX_add_extra_chain_cert, while the code still has ownership of the
argument.
Closes https://github.com/curl/curl/pull/1236
ERR_PACK is an internal detail of OpenSSL. Also, when using it, a
function name must be specified which is overly specific: the test will
break whenever OpenSSL internally change things so that a different
function creates the error.
Closes#1157
* HTTPS proxies:
An HTTPS proxy receives all transactions over an SSL/TLS connection.
Once a secure connection with the proxy is established, the user agent
uses the proxy as usual, including sending CONNECT requests to instruct
the proxy to establish a [usually secure] TCP tunnel with an origin
server. HTTPS proxies protect nearly all aspects of user-proxy
communications as opposed to HTTP proxies that receive all requests
(including CONNECT requests) in vulnerable clear text.
With HTTPS proxies, it is possible to have two concurrent _nested_
SSL/TLS sessions: the "outer" one between the user agent and the proxy
and the "inner" one between the user agent and the origin server
(through the proxy). This change adds supports for such nested sessions
as well.
A secure connection with a proxy requires its own set of the usual SSL
options (their actual descriptions differ and need polishing, see TODO):
--proxy-cacert FILE CA certificate to verify peer against
--proxy-capath DIR CA directory to verify peer against
--proxy-cert CERT[:PASSWD] Client certificate file and password
--proxy-cert-type TYPE Certificate file type (DER/PEM/ENG)
--proxy-ciphers LIST SSL ciphers to use
--proxy-crlfile FILE Get a CRL list in PEM format from the file
--proxy-insecure Allow connections to proxies with bad certs
--proxy-key KEY Private key file name
--proxy-key-type TYPE Private key file type (DER/PEM/ENG)
--proxy-pass PASS Pass phrase for the private key
--proxy-ssl-allow-beast Allow security flaw to improve interop
--proxy-sslv2 Use SSLv2
--proxy-sslv3 Use SSLv3
--proxy-tlsv1 Use TLSv1
--proxy-tlsuser USER TLS username
--proxy-tlspassword STRING TLS password
--proxy-tlsauthtype STRING TLS authentication type (default SRP)
All --proxy-foo options are independent from their --foo counterparts,
except --proxy-crlfile which defaults to --crlfile and --proxy-capath
which defaults to --capath.
Curl now also supports %{proxy_ssl_verify_result} --write-out variable,
similar to the existing %{ssl_verify_result} variable.
Supported backends: OpenSSL, GnuTLS, and NSS.
* A SOCKS proxy + HTTP/HTTPS proxy combination:
If both --socks* and --proxy options are given, Curl first connects to
the SOCKS proxy and then connects (through SOCKS) to the HTTP or HTTPS
proxy.
TODO: Update documentation for the new APIs and --proxy-* options.
Look for "Added in 7.XXX" marks.
- Fix GnuTLS code for CURL_SSLVERSION_TLSv1_2 that broke when the
TLS 1.3 support was added in 6ad3add.
- Homogenize across code for all backends the error message when TLS 1.3
is not available to "<backend>: TLS 1.3 is not yet supported".
- Return an error when a user-specified ssl version is unrecognized.
---
Prior to this change our code for some of the backends used the
'default' label in the switch statement (ie ver unrecognized) for
ssl.version and treated it the same as CURL_SSLVERSION_DEFAULT.
Bug: https://curl.haxx.se/mail/lib-2016-11/0048.html
Reported-by: Kamil Dudka
We're mostly saying just "curl" in lower case these days so here's a big
cleanup to adapt to this reality. A few instances are left as the
project could still formally be considered called cURL.
... to make it less likely that we forget that the function actually
does case insentive compares. Also replaced several invokes of the
function with a plain strcmp when case sensitivity is not an issue (like
comparing with "-").
Curl_select_ready() was the former API that was replaced with
Curl_select_check() a while back and the former arg setup was provided
with a define (in order to leave existing code unmodified).
Now we instead offer SOCKET_READABLE and SOCKET_WRITABLE for the most
common shortcuts where only one socket is checked. They're also more
visibly macros.
LibreSSL defines `OPENSSL_VERSION_NUMBER` as `0x20000000L` for all
versions returning `LibreSSL/2.0.0` for any LibreSSL version.
This change provides a local OpenSSL_version_num function replacement
returning LIBRESSL_VERSION_NUMBER instead.
Closes#1029
The OpenSSL function CRYTPO_cleanup_all_ex_data() cannot be called
multiple times without crashing - and other libs might call it! We
basically cannot call it without risking a crash. The function is a
no-op since OpenSSL 1.1.0.
Not calling this function only risks a small memory leak with OpenSSL <
1.1.0.
Bug: https://curl.haxx.se/mail/lib-2016-09/0045.html
Reported-by: Todd Short
OpenSSL 1.0.1 and 1.0.2 build an error queue that is stored per-thread
so we need to clean it when easy handles are freed, in case the thread
will be killed in which the easy handle was used. All OpenSSL code in
libcurl should extract the error in association with the error already
so clearing this queue here should be harmless at worst.
Fixes#964
... by partially reverting f975f06033. The allocation could be made by
OpenSSL so the free must be made with OPENSSL_free() to avoid problems.
Reported-by: Harold Stuart
Fixes#1005
CURLINFO_SSL_VERIFYRESULT does not get the certificate verification
result when SSL_connect fails because of a certificate verification
error.
This fix saves the result of SSL_get_verify_result so that it is
returned by CURLINFO_SSL_VERIFYRESULT.
Closes https://github.com/curl/curl/pull/995
Undo change introduced in d4643d6 which caused iPAddress match to be
ignored if dNSName was present but did not match.
Also, if iPAddress is present but does not match, and dNSName is not
present, fail as no-match. Prior to this change in such a case the CN
would be checked for a match.
Bug: https://github.com/curl/curl/issues/959
Reported-by: wmsch@users.noreply.github.com
Prior to this change we called Curl_ssl_getsessionid and
Curl_ssl_addsessionid regardless of whether session ID reusing was
enabled. According to comments that is in case session ID reuse was
disabled but then later enabled.
The old way was not intuitive and probably not something users expected.
When a user disables session ID caching I'd guess they don't expect the
session ID to be cached anyway in case the caching is later enabled.
Sessionid cache management is inseparable from managing individual
session lifetimes. E.g. for reference-counted sessions (like those in
SChannel and OpenSSL engines) every session addition and removal
should be accompanied with refcount increment and decrement
respectively. Failing to do so synchronously leads to a race condition
that causes symptoms like use-after-free and memory corruption.
This commit:
- makes existing session cache locking explicit, thus allowing
individual engines to manage lock's scope.
- fixes OpenSSL and SChannel engines by putting refcount management
inside this lock's scope in relevant places.
- adds these explicit locking calls to other engines that use
sessionid cache to accommodate for this change. Note, however,
that it is unknown whether any of these engines could also have
this race.
Bug: https://github.com/curl/curl/issues/815Fixes#815Closes#847
... introduced in curl-7_48_0-293-g2968c83:
Error: COMPILER_WARNING:
lib/vtls/openssl.c: scope_hint: In function ‘Curl_ossl_check_cxn’
lib/vtls/openssl.c:767:15: warning: conversion to ‘int’ from ‘ssize_t’
may alter its value [-Wconversion]
- In the case of recv error, limit returning 'connection still in place'
to EINPROGRESS, EAGAIN and EWOULDBLOCK.
This is an improvement on the parent commit which changed the openssl
connection check to use recv MSG_PEEK instead of SSL_peek.
Ref: https://github.com/curl/curl/commit/856baf5#comments
Calling SSL_peek can cause bytes to be read from the raw socket which in
turn can upset the select machinery that determines whether there's data
available on the socket.
Since Curl_ossl_check_cxn only tries to determine whether the socket is
alive and doesn't actually need to see the bytes SSL_peek seems like
the wrong function to call.
We're able to occasionally reproduce a connect timeout due to this
bug. What happens is that Curl doesn't know to call SSL_connect again
after the peek happens since data is buffered in the SSL buffer and thus
select won't fire for this socket.
Closes#795
Only protocols that actually have a protocol registered for ALPN and NPN
should try to get that negotiated in the TLS handshake. That is only
HTTPS (well, http/1.1 and http/2) right now. Previously ALPN and NPN
would wrongly be used in all handshakes if libcurl was built with it
enabled.
Reported-by: Jay Satiro
Fixes#789
