curl/lib/vtls/wolfssl.c

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

1430 lines
43 KiB
C
Raw Normal View History

/***************************************************************************
* _ _ ____ _
* Project ___| | | | _ \| |
* / __| | | | |_) | |
* | (__| |_| | _ <| |___
* \___|\___/|_| \_\_____|
*
* Copyright (C) Daniel Stenberg, <daniel@haxx.se>, et al.
*
* This software is licensed as described in the file COPYING, which
* you should have received as part of this distribution. The terms
2020-11-04 21:02:01 +08:00
* are also available at https://curl.se/docs/copyright.html.
*
* You may opt to use, copy, modify, merge, publish, distribute and/or sell
* copies of the Software, and permit persons to whom the Software is
* furnished to do so, under the terms of the COPYING file.
*
* This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
* KIND, either express or implied.
*
* SPDX-License-Identifier: curl
*
***************************************************************************/
/*
* Source file for all wolfSSL specific code for the TLS/SSL layer. No code
* but vtls.c should ever call or use these functions.
*
*/
build: fix circular header inclusion with other packages This commit renames lib/setup.h to lib/curl_setup.h and renames lib/setup_once.h to lib/curl_setup_once.h. Removes the need and usage of a header inclusion guard foreign to libcurl. [1] Removes the need and presence of an alarming notice we carried in old setup_once.h [2] ---------------------------------------- 1 - lib/setup_once.h used __SETUP_ONCE_H macro as header inclusion guard up to commit ec691ca3 which changed this to HEADER_CURL_SETUP_ONCE_H, this single inclusion guard is enough to ensure that inclusion of lib/setup_once.h done from lib/setup.h is only done once. Additionally lib/setup.h has always used __SETUP_ONCE_H macro to protect inclusion of setup_once.h even after commit ec691ca3, this was to avoid a circular header inclusion triggered when building a c-ares enabled version with c-ares sources available which also has a setup_once.h header. Commit ec691ca3 exposes the real nature of __SETUP_ONCE_H usage in lib/setup.h, it is a header inclusion guard foreign to libcurl belonging to c-ares's setup_once.h The renaming this commit does, fixes the circular header inclusion, and as such removes the need and usage of a header inclusion guard foreign to libcurl. Macro __SETUP_ONCE_H no longer used in libcurl. 2 - Due to the circular interdependency of old lib/setup_once.h and the c-ares setup_once.h header, old file lib/setup_once.h has carried back from 2006 up to now days an alarming and prominent notice about the need of keeping libcurl's and c-ares's setup_once.h in sync. Given that this commit fixes the circular interdependency, the need and presence of mentioned notice is removed. All mentioned interdependencies come back from now old days when the c-ares project lived inside a curl subdirectory. This commit removes last traces of such fact.
2013-01-07 02:06:49 +08:00
#include "curl_setup.h"
#ifdef USE_WOLFSSL
#define WOLFSSL_OPTIONS_IGNORE_SYS
#include <wolfssl/version.h>
#include <wolfssl/options.h>
/* To determine what functions are available we rely on one or both of:
- the user's options.h generated by wolfSSL
- the symbols detected by curl's configure
Since they are markedly different from one another, and one or the other may
not be available, we do some checking below to bring things in sync. */
/* HAVE_ALPN is wolfSSL's build time symbol for enabling ALPN in options.h. */
#ifndef HAVE_ALPN
#ifdef HAVE_WOLFSSL_USEALPN
#define HAVE_ALPN
#endif
#endif
2011-08-24 20:00:42 +08:00
#include <limits.h>
#include "urldata.h"
#include "sendf.h"
#include "inet_pton.h"
2013-12-25 18:20:39 +08:00
#include "vtls.h"
#include "vtls_int.h"
#include "keylog.h"
#include "parsedate.h"
#include "connect.h" /* for the connect timeout */
#include "select.h"
#include "strcase.h"
#include "x509asn1.h"
#include "curl_printf.h"
#include "multiif.h"
#include <wolfssl/openssl/ssl.h>
#include <wolfssl/ssl.h>
#include <wolfssl/error-ssl.h>
#include "wolfssl.h"
/* The last #include files should be: */
#include "curl_memory.h"
#include "memdebug.h"
/* KEEP_PEER_CERT is a product of the presence of build time symbol
OPENSSL_EXTRA without NO_CERTS, depending on the version. KEEP_PEER_CERT is
in wolfSSL's settings.h, and the latter two are build time symbols in
options.h. */
#ifndef KEEP_PEER_CERT
#if defined(HAVE_WOLFSSL_GET_PEER_CERTIFICATE) || \
(defined(OPENSSL_EXTRA) && !defined(NO_CERTS))
#define KEEP_PEER_CERT
#endif
#endif
#if defined(HAVE_WOLFSSL_FULL_BIO) && HAVE_WOLFSSL_FULL_BIO
#define USE_BIO_CHAIN
#else
#undef USE_BIO_CHAIN
#endif
cmake: add support for "unity" builds Aka "jumbo" or "amalgamation" builds. It means to compile all sources per target as a single C source. This is experimental. You can enable it by passing `-DCMAKE_UNITY_BUILD=ON` to cmake. It requires CMake 3.16 or newer. It makes builds (much) faster, allows for better optimizations and tends to promote less ambiguous code. Also add a new AppVeyor CI job and convert an existing one to use "unity" mode (one MSVC, one MinGW), and enable it for one macOS CI job. Fix related issues: - add missing include guard to `easy_lock.h`. - rename static variables and functions (and a macro) with names reused across sources, or shadowed by local variables. - add an `#undef` after use. - add a missing `#undef` before use. - move internal definitions from `ftp.h` to `ftp.c`. - `curl_memory.h` fixes to make it work when included repeatedly. - stop building/linking curlx bits twice for a static-mode curl tool. These caused doubly defined symbols in unity builds. - silence missing extern declarations compiler warning for ` _CRT_glob`. - fix extern declarations for `tool_freq` and `tool_isVistaOrGreater`. - fix colliding static symbols in debug mode: `debugtime()` and `statename`. - rename `ssl_backend_data` structure to unique names for each TLS-backend, along with the `ssl_connect_data` struct member referencing them. This required adding casts for each access. - add workaround for missing `[P]UNICODE_STRING` types in certain Windows builds when compiling `lib/ldap.c`. To support "unity" builds, we had to enable `SCHANNEL_USE_BLACKLISTS` for Schannel (a Windows `schannel.h` option) _globally_. This caused an indirect inclusion of Windows `schannel.h` from `ldap.c` via `winldap.h` to have it enabled as well. This requires `[P]UNICODE_STRING` types, which is apperantly not defined automatically (as seen with both MSVS and mingw-w64). This patch includes `<subauth.h>` to fix it. Ref: https://github.com/curl/curl/runs/13987772013 Ref: https://dev.azure.com/daniel0244/curl/_build/results?buildId=15827&view=logs&jobId=2c9f582d-e278-56b6-4354-f38a4d851906&j=2c9f582d-e278-56b6-4354-f38a4d851906&t=90509b00-34fa-5a81-35d7-5ed9569d331c - tweak unity builds to compile `lib/memdebug.c` separately in memory trace builds to avoid PP confusion. - force-disable unity for test programs. - do not compile and link libcurl sources to libtests _twice_ when libcurl is built in static mode. KNOWN ISSUES: - running tests with unity builds may fail in cases. - some build configurations/env may not compile in unity mode. E.g.: https://ci.appveyor.com/project/curlorg/curl/builds/47230972/job/51wfesgnfuauwl8q#L250 Ref: https://github.com/libssh2/libssh2/issues/1034 Ref: https://cmake.org/cmake/help/latest/prop_tgt/UNITY_BUILD.html Ref: https://en.wikipedia.org/wiki/Unity_build Closes #11095
2023-05-09 18:10:40 +08:00
struct wolfssl_ssl_backend_data {
WOLFSSL_CTX *ctx;
WOLFSSL *handle;
CURLcode io_result; /* result of last BIO cfilter operation */
vtls: encapsulate SSL backend-specific data 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>
2017-07-29 04:09:35 +08:00
};
#ifdef OPENSSL_EXTRA
/*
* Availability note:
* The TLS 1.3 secret callback (wolfSSL_set_tls13_secret_cb) was added in
* WolfSSL 4.4.0, but requires the -DHAVE_SECRET_CALLBACK build option. If that
* option is not set, then TLS 1.3 will not be logged.
* For TLS 1.2 and before, we use wolfSSL_get_keys().
* SSL_get_client_random and wolfSSL_get_keys require OPENSSL_EXTRA
* (--enable-opensslextra or --enable-all).
*/
#if defined(HAVE_SECRET_CALLBACK) && defined(WOLFSSL_TLS13)
static int
wolfssl_tls13_secret_callback(SSL *ssl, int id, const unsigned char *secret,
int secretSz, void *ctx)
{
const char *label;
unsigned char client_random[SSL3_RANDOM_SIZE];
(void)ctx;
if(!ssl || !Curl_tls_keylog_enabled()) {
return 0;
}
switch(id) {
case CLIENT_EARLY_TRAFFIC_SECRET:
label = "CLIENT_EARLY_TRAFFIC_SECRET";
break;
case CLIENT_HANDSHAKE_TRAFFIC_SECRET:
label = "CLIENT_HANDSHAKE_TRAFFIC_SECRET";
break;
case SERVER_HANDSHAKE_TRAFFIC_SECRET:
label = "SERVER_HANDSHAKE_TRAFFIC_SECRET";
break;
case CLIENT_TRAFFIC_SECRET:
label = "CLIENT_TRAFFIC_SECRET_0";
break;
case SERVER_TRAFFIC_SECRET:
label = "SERVER_TRAFFIC_SECRET_0";
break;
case EARLY_EXPORTER_SECRET:
label = "EARLY_EXPORTER_SECRET";
break;
case EXPORTER_SECRET:
label = "EXPORTER_SECRET";
break;
default:
return 0;
}
if(SSL_get_client_random(ssl, client_random, SSL3_RANDOM_SIZE) == 0) {
/* Should never happen as wolfSSL_KeepArrays() was called before. */
return 0;
}
Curl_tls_keylog_write(label, client_random, secret, secretSz);
return 0;
}
#endif /* defined(HAVE_SECRET_CALLBACK) && defined(WOLFSSL_TLS13) */
static void
wolfssl_log_tls12_secret(SSL *ssl)
{
unsigned char *ms, *sr, *cr;
unsigned int msLen, srLen, crLen, i, x = 0;
#if LIBWOLFSSL_VERSION_HEX >= 0x0300d000 /* >= 3.13.0 */
/* wolfSSL_GetVersion is available since 3.13, we use it instead of
* SSL_version since the latter relies on OPENSSL_ALL (--enable-opensslall or
* --enable-all). Failing to perform this check could result in an unusable
* key log line when TLS 1.3 is actually negotiated. */
switch(wolfSSL_GetVersion(ssl)) {
case WOLFSSL_SSLV3:
case WOLFSSL_TLSV1:
case WOLFSSL_TLSV1_1:
case WOLFSSL_TLSV1_2:
break;
default:
/* TLS 1.3 does not use this mechanism, the "master secret" returned below
* is not directly usable. */
return;
}
#endif
if(wolfSSL_get_keys(ssl, &ms, &msLen, &sr, &srLen, &cr, &crLen) !=
SSL_SUCCESS) {
return;
}
/* Check for a missing master secret and skip logging. That can happen if
* curl rejects the server certificate and aborts the handshake.
*/
for(i = 0; i < msLen; i++) {
x |= ms[i];
}
if(x == 0) {
return;
}
Curl_tls_keylog_write("CLIENT_RANDOM", cr, ms, msLen);
}
#endif /* OPENSSL_EXTRA */
static int do_file_type(const char *type)
{
if(!type || !type[0])
return SSL_FILETYPE_PEM;
if(strcasecompare(type, "PEM"))
return SSL_FILETYPE_PEM;
if(strcasecompare(type, "DER"))
return SSL_FILETYPE_ASN1;
return -1;
}
#ifdef HAVE_LIBOQS
struct group_name_map {
const word16 group;
const char *name;
};
static const struct group_name_map gnm[] = {
{ WOLFSSL_KYBER_LEVEL1, "KYBER_LEVEL1" },
{ WOLFSSL_KYBER_LEVEL3, "KYBER_LEVEL3" },
{ WOLFSSL_KYBER_LEVEL5, "KYBER_LEVEL5" },
{ WOLFSSL_P256_KYBER_LEVEL1, "P256_KYBER_LEVEL1" },
{ WOLFSSL_P384_KYBER_LEVEL3, "P384_KYBER_LEVEL3" },
{ WOLFSSL_P521_KYBER_LEVEL5, "P521_KYBER_LEVEL5" },
{ 0, NULL }
};
#endif
#ifdef USE_BIO_CHAIN
static int wolfssl_bio_cf_create(WOLFSSL_BIO *bio)
{
wolfSSL_BIO_set_shutdown(bio, 1);
wolfSSL_BIO_set_data(bio, NULL);
return 1;
}
static int wolfssl_bio_cf_destroy(WOLFSSL_BIO *bio)
{
if(!bio)
return 0;
return 1;
}
static long wolfssl_bio_cf_ctrl(WOLFSSL_BIO *bio, int cmd, long num, void *ptr)
{
struct Curl_cfilter *cf = BIO_get_data(bio);
long ret = 1;
(void)cf;
(void)ptr;
switch(cmd) {
case BIO_CTRL_GET_CLOSE:
ret = (long)wolfSSL_BIO_get_shutdown(bio);
break;
case BIO_CTRL_SET_CLOSE:
wolfSSL_BIO_set_shutdown(bio, (int)num);
break;
case BIO_CTRL_FLUSH:
/* we do no delayed writes, but if we ever would, this
* needs to trigger it. */
ret = 1;
break;
case BIO_CTRL_DUP:
ret = 1;
break;
#ifdef BIO_CTRL_EOF
case BIO_CTRL_EOF:
/* EOF has been reached on input? */
return (!cf->next || !cf->next->connected);
#endif
default:
ret = 0;
break;
}
return ret;
}
static int wolfssl_bio_cf_out_write(WOLFSSL_BIO *bio,
const char *buf, int blen)
{
struct Curl_cfilter *cf = wolfSSL_BIO_get_data(bio);
struct ssl_connect_data *connssl = cf->ctx;
cmake: add support for "unity" builds Aka "jumbo" or "amalgamation" builds. It means to compile all sources per target as a single C source. This is experimental. You can enable it by passing `-DCMAKE_UNITY_BUILD=ON` to cmake. It requires CMake 3.16 or newer. It makes builds (much) faster, allows for better optimizations and tends to promote less ambiguous code. Also add a new AppVeyor CI job and convert an existing one to use "unity" mode (one MSVC, one MinGW), and enable it for one macOS CI job. Fix related issues: - add missing include guard to `easy_lock.h`. - rename static variables and functions (and a macro) with names reused across sources, or shadowed by local variables. - add an `#undef` after use. - add a missing `#undef` before use. - move internal definitions from `ftp.h` to `ftp.c`. - `curl_memory.h` fixes to make it work when included repeatedly. - stop building/linking curlx bits twice for a static-mode curl tool. These caused doubly defined symbols in unity builds. - silence missing extern declarations compiler warning for ` _CRT_glob`. - fix extern declarations for `tool_freq` and `tool_isVistaOrGreater`. - fix colliding static symbols in debug mode: `debugtime()` and `statename`. - rename `ssl_backend_data` structure to unique names for each TLS-backend, along with the `ssl_connect_data` struct member referencing them. This required adding casts for each access. - add workaround for missing `[P]UNICODE_STRING` types in certain Windows builds when compiling `lib/ldap.c`. To support "unity" builds, we had to enable `SCHANNEL_USE_BLACKLISTS` for Schannel (a Windows `schannel.h` option) _globally_. This caused an indirect inclusion of Windows `schannel.h` from `ldap.c` via `winldap.h` to have it enabled as well. This requires `[P]UNICODE_STRING` types, which is apperantly not defined automatically (as seen with both MSVS and mingw-w64). This patch includes `<subauth.h>` to fix it. Ref: https://github.com/curl/curl/runs/13987772013 Ref: https://dev.azure.com/daniel0244/curl/_build/results?buildId=15827&view=logs&jobId=2c9f582d-e278-56b6-4354-f38a4d851906&j=2c9f582d-e278-56b6-4354-f38a4d851906&t=90509b00-34fa-5a81-35d7-5ed9569d331c - tweak unity builds to compile `lib/memdebug.c` separately in memory trace builds to avoid PP confusion. - force-disable unity for test programs. - do not compile and link libcurl sources to libtests _twice_ when libcurl is built in static mode. KNOWN ISSUES: - running tests with unity builds may fail in cases. - some build configurations/env may not compile in unity mode. E.g.: https://ci.appveyor.com/project/curlorg/curl/builds/47230972/job/51wfesgnfuauwl8q#L250 Ref: https://github.com/libssh2/libssh2/issues/1034 Ref: https://cmake.org/cmake/help/latest/prop_tgt/UNITY_BUILD.html Ref: https://en.wikipedia.org/wiki/Unity_build Closes #11095
2023-05-09 18:10:40 +08:00
struct wolfssl_ssl_backend_data *backend =
(struct wolfssl_ssl_backend_data *)connssl->backend;
struct Curl_easy *data = CF_DATA_CURRENT(cf);
ssize_t nwritten;
CURLcode result = CURLE_OK;
DEBUGASSERT(data);
nwritten = Curl_conn_cf_send(cf->next, data, buf, blen, &result);
cmake: add support for "unity" builds Aka "jumbo" or "amalgamation" builds. It means to compile all sources per target as a single C source. This is experimental. You can enable it by passing `-DCMAKE_UNITY_BUILD=ON` to cmake. It requires CMake 3.16 or newer. It makes builds (much) faster, allows for better optimizations and tends to promote less ambiguous code. Also add a new AppVeyor CI job and convert an existing one to use "unity" mode (one MSVC, one MinGW), and enable it for one macOS CI job. Fix related issues: - add missing include guard to `easy_lock.h`. - rename static variables and functions (and a macro) with names reused across sources, or shadowed by local variables. - add an `#undef` after use. - add a missing `#undef` before use. - move internal definitions from `ftp.h` to `ftp.c`. - `curl_memory.h` fixes to make it work when included repeatedly. - stop building/linking curlx bits twice for a static-mode curl tool. These caused doubly defined symbols in unity builds. - silence missing extern declarations compiler warning for ` _CRT_glob`. - fix extern declarations for `tool_freq` and `tool_isVistaOrGreater`. - fix colliding static symbols in debug mode: `debugtime()` and `statename`. - rename `ssl_backend_data` structure to unique names for each TLS-backend, along with the `ssl_connect_data` struct member referencing them. This required adding casts for each access. - add workaround for missing `[P]UNICODE_STRING` types in certain Windows builds when compiling `lib/ldap.c`. To support "unity" builds, we had to enable `SCHANNEL_USE_BLACKLISTS` for Schannel (a Windows `schannel.h` option) _globally_. This caused an indirect inclusion of Windows `schannel.h` from `ldap.c` via `winldap.h` to have it enabled as well. This requires `[P]UNICODE_STRING` types, which is apperantly not defined automatically (as seen with both MSVS and mingw-w64). This patch includes `<subauth.h>` to fix it. Ref: https://github.com/curl/curl/runs/13987772013 Ref: https://dev.azure.com/daniel0244/curl/_build/results?buildId=15827&view=logs&jobId=2c9f582d-e278-56b6-4354-f38a4d851906&j=2c9f582d-e278-56b6-4354-f38a4d851906&t=90509b00-34fa-5a81-35d7-5ed9569d331c - tweak unity builds to compile `lib/memdebug.c` separately in memory trace builds to avoid PP confusion. - force-disable unity for test programs. - do not compile and link libcurl sources to libtests _twice_ when libcurl is built in static mode. KNOWN ISSUES: - running tests with unity builds may fail in cases. - some build configurations/env may not compile in unity mode. E.g.: https://ci.appveyor.com/project/curlorg/curl/builds/47230972/job/51wfesgnfuauwl8q#L250 Ref: https://github.com/libssh2/libssh2/issues/1034 Ref: https://cmake.org/cmake/help/latest/prop_tgt/UNITY_BUILD.html Ref: https://en.wikipedia.org/wiki/Unity_build Closes #11095
2023-05-09 18:10:40 +08:00
backend->io_result = result;
CURL_TRC_CF(data, cf, "bio_write(len=%d) -> %zd, %d",
blen, nwritten, result);
wolfSSL_BIO_clear_retry_flags(bio);
if(nwritten < 0 && CURLE_AGAIN == result)
BIO_set_retry_write(bio);
return (int)nwritten;
}
static int wolfssl_bio_cf_in_read(WOLFSSL_BIO *bio, char *buf, int blen)
{
struct Curl_cfilter *cf = wolfSSL_BIO_get_data(bio);
struct ssl_connect_data *connssl = cf->ctx;
cmake: add support for "unity" builds Aka "jumbo" or "amalgamation" builds. It means to compile all sources per target as a single C source. This is experimental. You can enable it by passing `-DCMAKE_UNITY_BUILD=ON` to cmake. It requires CMake 3.16 or newer. It makes builds (much) faster, allows for better optimizations and tends to promote less ambiguous code. Also add a new AppVeyor CI job and convert an existing one to use "unity" mode (one MSVC, one MinGW), and enable it for one macOS CI job. Fix related issues: - add missing include guard to `easy_lock.h`. - rename static variables and functions (and a macro) with names reused across sources, or shadowed by local variables. - add an `#undef` after use. - add a missing `#undef` before use. - move internal definitions from `ftp.h` to `ftp.c`. - `curl_memory.h` fixes to make it work when included repeatedly. - stop building/linking curlx bits twice for a static-mode curl tool. These caused doubly defined symbols in unity builds. - silence missing extern declarations compiler warning for ` _CRT_glob`. - fix extern declarations for `tool_freq` and `tool_isVistaOrGreater`. - fix colliding static symbols in debug mode: `debugtime()` and `statename`. - rename `ssl_backend_data` structure to unique names for each TLS-backend, along with the `ssl_connect_data` struct member referencing them. This required adding casts for each access. - add workaround for missing `[P]UNICODE_STRING` types in certain Windows builds when compiling `lib/ldap.c`. To support "unity" builds, we had to enable `SCHANNEL_USE_BLACKLISTS` for Schannel (a Windows `schannel.h` option) _globally_. This caused an indirect inclusion of Windows `schannel.h` from `ldap.c` via `winldap.h` to have it enabled as well. This requires `[P]UNICODE_STRING` types, which is apperantly not defined automatically (as seen with both MSVS and mingw-w64). This patch includes `<subauth.h>` to fix it. Ref: https://github.com/curl/curl/runs/13987772013 Ref: https://dev.azure.com/daniel0244/curl/_build/results?buildId=15827&view=logs&jobId=2c9f582d-e278-56b6-4354-f38a4d851906&j=2c9f582d-e278-56b6-4354-f38a4d851906&t=90509b00-34fa-5a81-35d7-5ed9569d331c - tweak unity builds to compile `lib/memdebug.c` separately in memory trace builds to avoid PP confusion. - force-disable unity for test programs. - do not compile and link libcurl sources to libtests _twice_ when libcurl is built in static mode. KNOWN ISSUES: - running tests with unity builds may fail in cases. - some build configurations/env may not compile in unity mode. E.g.: https://ci.appveyor.com/project/curlorg/curl/builds/47230972/job/51wfesgnfuauwl8q#L250 Ref: https://github.com/libssh2/libssh2/issues/1034 Ref: https://cmake.org/cmake/help/latest/prop_tgt/UNITY_BUILD.html Ref: https://en.wikipedia.org/wiki/Unity_build Closes #11095
2023-05-09 18:10:40 +08:00
struct wolfssl_ssl_backend_data *backend =
(struct wolfssl_ssl_backend_data *)connssl->backend;
struct Curl_easy *data = CF_DATA_CURRENT(cf);
ssize_t nread;
CURLcode result = CURLE_OK;
DEBUGASSERT(data);
/* OpenSSL catches this case, so should we. */
if(!buf)
return 0;
nread = Curl_conn_cf_recv(cf->next, data, buf, blen, &result);
cmake: add support for "unity" builds Aka "jumbo" or "amalgamation" builds. It means to compile all sources per target as a single C source. This is experimental. You can enable it by passing `-DCMAKE_UNITY_BUILD=ON` to cmake. It requires CMake 3.16 or newer. It makes builds (much) faster, allows for better optimizations and tends to promote less ambiguous code. Also add a new AppVeyor CI job and convert an existing one to use "unity" mode (one MSVC, one MinGW), and enable it for one macOS CI job. Fix related issues: - add missing include guard to `easy_lock.h`. - rename static variables and functions (and a macro) with names reused across sources, or shadowed by local variables. - add an `#undef` after use. - add a missing `#undef` before use. - move internal definitions from `ftp.h` to `ftp.c`. - `curl_memory.h` fixes to make it work when included repeatedly. - stop building/linking curlx bits twice for a static-mode curl tool. These caused doubly defined symbols in unity builds. - silence missing extern declarations compiler warning for ` _CRT_glob`. - fix extern declarations for `tool_freq` and `tool_isVistaOrGreater`. - fix colliding static symbols in debug mode: `debugtime()` and `statename`. - rename `ssl_backend_data` structure to unique names for each TLS-backend, along with the `ssl_connect_data` struct member referencing them. This required adding casts for each access. - add workaround for missing `[P]UNICODE_STRING` types in certain Windows builds when compiling `lib/ldap.c`. To support "unity" builds, we had to enable `SCHANNEL_USE_BLACKLISTS` for Schannel (a Windows `schannel.h` option) _globally_. This caused an indirect inclusion of Windows `schannel.h` from `ldap.c` via `winldap.h` to have it enabled as well. This requires `[P]UNICODE_STRING` types, which is apperantly not defined automatically (as seen with both MSVS and mingw-w64). This patch includes `<subauth.h>` to fix it. Ref: https://github.com/curl/curl/runs/13987772013 Ref: https://dev.azure.com/daniel0244/curl/_build/results?buildId=15827&view=logs&jobId=2c9f582d-e278-56b6-4354-f38a4d851906&j=2c9f582d-e278-56b6-4354-f38a4d851906&t=90509b00-34fa-5a81-35d7-5ed9569d331c - tweak unity builds to compile `lib/memdebug.c` separately in memory trace builds to avoid PP confusion. - force-disable unity for test programs. - do not compile and link libcurl sources to libtests _twice_ when libcurl is built in static mode. KNOWN ISSUES: - running tests with unity builds may fail in cases. - some build configurations/env may not compile in unity mode. E.g.: https://ci.appveyor.com/project/curlorg/curl/builds/47230972/job/51wfesgnfuauwl8q#L250 Ref: https://github.com/libssh2/libssh2/issues/1034 Ref: https://cmake.org/cmake/help/latest/prop_tgt/UNITY_BUILD.html Ref: https://en.wikipedia.org/wiki/Unity_build Closes #11095
2023-05-09 18:10:40 +08:00
backend->io_result = result;
CURL_TRC_CF(data, cf, "bio_read(len=%d) -> %zd, %d", blen, nread, result);
wolfSSL_BIO_clear_retry_flags(bio);
if(nread < 0 && CURLE_AGAIN == result)
BIO_set_retry_read(bio);
else if(nread == 0)
connssl->peer_closed = TRUE;
return (int)nread;
}
static WOLFSSL_BIO_METHOD *wolfssl_bio_cf_method = NULL;
static void wolfssl_bio_cf_init_methods(void)
{
wolfssl_bio_cf_method = wolfSSL_BIO_meth_new(BIO_TYPE_MEM, "wolfSSL CF BIO");
wolfSSL_BIO_meth_set_write(wolfssl_bio_cf_method, &wolfssl_bio_cf_out_write);
wolfSSL_BIO_meth_set_read(wolfssl_bio_cf_method, &wolfssl_bio_cf_in_read);
wolfSSL_BIO_meth_set_ctrl(wolfssl_bio_cf_method, &wolfssl_bio_cf_ctrl);
wolfSSL_BIO_meth_set_create(wolfssl_bio_cf_method, &wolfssl_bio_cf_create);
wolfSSL_BIO_meth_set_destroy(wolfssl_bio_cf_method, &wolfssl_bio_cf_destroy);
}
static void wolfssl_bio_cf_free_methods(void)
{
wolfSSL_BIO_meth_free(wolfssl_bio_cf_method);
}
#else /* USE_BIO_CHAIN */
#define wolfssl_bio_cf_init_methods() Curl_nop_stmt
#define wolfssl_bio_cf_free_methods() Curl_nop_stmt
#endif /* !USE_BIO_CHAIN */
/*
* This function loads all the client/CA certificates and CRLs. Setup the TLS
* layer and do all necessary magic.
*/
static CURLcode
wolfssl_connect_step1(struct Curl_cfilter *cf, struct Curl_easy *data)
{
char *ciphers, *curves;
struct ssl_connect_data *connssl = cf->ctx;
cmake: add support for "unity" builds Aka "jumbo" or "amalgamation" builds. It means to compile all sources per target as a single C source. This is experimental. You can enable it by passing `-DCMAKE_UNITY_BUILD=ON` to cmake. It requires CMake 3.16 or newer. It makes builds (much) faster, allows for better optimizations and tends to promote less ambiguous code. Also add a new AppVeyor CI job and convert an existing one to use "unity" mode (one MSVC, one MinGW), and enable it for one macOS CI job. Fix related issues: - add missing include guard to `easy_lock.h`. - rename static variables and functions (and a macro) with names reused across sources, or shadowed by local variables. - add an `#undef` after use. - add a missing `#undef` before use. - move internal definitions from `ftp.h` to `ftp.c`. - `curl_memory.h` fixes to make it work when included repeatedly. - stop building/linking curlx bits twice for a static-mode curl tool. These caused doubly defined symbols in unity builds. - silence missing extern declarations compiler warning for ` _CRT_glob`. - fix extern declarations for `tool_freq` and `tool_isVistaOrGreater`. - fix colliding static symbols in debug mode: `debugtime()` and `statename`. - rename `ssl_backend_data` structure to unique names for each TLS-backend, along with the `ssl_connect_data` struct member referencing them. This required adding casts for each access. - add workaround for missing `[P]UNICODE_STRING` types in certain Windows builds when compiling `lib/ldap.c`. To support "unity" builds, we had to enable `SCHANNEL_USE_BLACKLISTS` for Schannel (a Windows `schannel.h` option) _globally_. This caused an indirect inclusion of Windows `schannel.h` from `ldap.c` via `winldap.h` to have it enabled as well. This requires `[P]UNICODE_STRING` types, which is apperantly not defined automatically (as seen with both MSVS and mingw-w64). This patch includes `<subauth.h>` to fix it. Ref: https://github.com/curl/curl/runs/13987772013 Ref: https://dev.azure.com/daniel0244/curl/_build/results?buildId=15827&view=logs&jobId=2c9f582d-e278-56b6-4354-f38a4d851906&j=2c9f582d-e278-56b6-4354-f38a4d851906&t=90509b00-34fa-5a81-35d7-5ed9569d331c - tweak unity builds to compile `lib/memdebug.c` separately in memory trace builds to avoid PP confusion. - force-disable unity for test programs. - do not compile and link libcurl sources to libtests _twice_ when libcurl is built in static mode. KNOWN ISSUES: - running tests with unity builds may fail in cases. - some build configurations/env may not compile in unity mode. E.g.: https://ci.appveyor.com/project/curlorg/curl/builds/47230972/job/51wfesgnfuauwl8q#L250 Ref: https://github.com/libssh2/libssh2/issues/1034 Ref: https://cmake.org/cmake/help/latest/prop_tgt/UNITY_BUILD.html Ref: https://en.wikipedia.org/wiki/Unity_build Closes #11095
2023-05-09 18:10:40 +08:00
struct wolfssl_ssl_backend_data *backend =
(struct wolfssl_ssl_backend_data *)connssl->backend;
struct ssl_primary_config *conn_config = Curl_ssl_cf_get_primary_config(cf);
const struct curl_blob *ca_info_blob = conn_config->ca_info_blob;
const struct ssl_config_data *ssl_config = Curl_ssl_cf_get_config(cf, data);
const char * const ssl_cafile =
/* CURLOPT_CAINFO_BLOB overrides CURLOPT_CAINFO */
(ca_info_blob ? NULL : conn_config->CAfile);
const char * const ssl_capath = conn_config->CApath;
WOLFSSL_METHOD* req_method = NULL;
#ifdef HAVE_LIBOQS
word16 oqsAlg = 0;
size_t idx = 0;
#endif
#ifdef HAVE_SNI
bool sni = FALSE;
#define use_sni(x) sni = (x)
#else
#define use_sni(x) Curl_nop_stmt
#endif
bool imported_native_ca = false;
bool imported_ca_info_blob = false;
DEBUGASSERT(backend);
if(connssl->state == ssl_connection_complete)
return CURLE_OK;
if(conn_config->version_max != CURL_SSLVERSION_MAX_NONE) {
failf(data, "wolfSSL does not support to set maximum SSL/TLS version");
return CURLE_SSL_CONNECT_ERROR;
}
/* check to see if we've been told to use an explicit SSL/TLS version */
switch(conn_config->version) {
case CURL_SSLVERSION_DEFAULT:
case CURL_SSLVERSION_TLSv1:
#if LIBWOLFSSL_VERSION_HEX >= 0x03003000 /* >= 3.3.0 */
/* minimum protocol version is set later after the CTX object is created */
req_method = SSLv23_client_method();
#else
infof(data, "wolfSSL <3.3.0 cannot be configured to use TLS 1.0-1.2, "
"TLS 1.0 is used exclusively");
req_method = TLSv1_client_method();
#endif
use_sni(TRUE);
break;
case CURL_SSLVERSION_TLSv1_0:
#if defined(WOLFSSL_ALLOW_TLSV10) && !defined(NO_OLD_TLS)
req_method = TLSv1_client_method();
use_sni(TRUE);
break;
#else
failf(data, "wolfSSL does not support TLS 1.0");
return CURLE_NOT_BUILT_IN;
#endif
case CURL_SSLVERSION_TLSv1_1:
#ifndef NO_OLD_TLS
req_method = TLSv1_1_client_method();
use_sni(TRUE);
#else
failf(data, "wolfSSL does not support TLS 1.1");
return CURLE_NOT_BUILT_IN;
#endif
break;
case CURL_SSLVERSION_TLSv1_2:
#ifndef WOLFSSL_NO_TLS12
req_method = TLSv1_2_client_method();
use_sni(TRUE);
#else
failf(data, "wolfSSL does not support TLS 1.2");
return CURLE_NOT_BUILT_IN;
#endif
break;
case CURL_SSLVERSION_TLSv1_3:
2018-03-02 02:10:36 +08:00
#ifdef WOLFSSL_TLS13
req_method = wolfTLSv1_3_client_method();
use_sni(TRUE);
break;
#else
failf(data, "wolfSSL: TLS 1.3 is not yet supported");
return CURLE_SSL_CONNECT_ERROR;
2018-03-02 02:10:36 +08:00
#endif
default:
failf(data, "Unrecognized parameter passed via CURLOPT_SSLVERSION");
return CURLE_SSL_CONNECT_ERROR;
}
if(!req_method) {
failf(data, "SSL: couldn't create a method");
return CURLE_OUT_OF_MEMORY;
}
if(backend->ctx)
wolfSSL_CTX_free(backend->ctx);
backend->ctx = wolfSSL_CTX_new(req_method);
if(!backend->ctx) {
failf(data, "SSL: couldn't create a context");
return CURLE_OUT_OF_MEMORY;
}
switch(conn_config->version) {
case CURL_SSLVERSION_DEFAULT:
case CURL_SSLVERSION_TLSv1:
#if LIBWOLFSSL_VERSION_HEX > 0x03004006 /* > 3.4.6 */
/* Versions 3.3.0 to 3.4.6 we know the minimum protocol version is
* whatever minimum version of TLS was built in and at least TLS 1.0. For
* later library versions that could change (eg TLS 1.0 built in but
* defaults to TLS 1.1) so we have this short circuit evaluation to find
* the minimum supported TLS version.
*/
if((wolfSSL_CTX_SetMinVersion(backend->ctx, WOLFSSL_TLSV1) != 1) &&
(wolfSSL_CTX_SetMinVersion(backend->ctx, WOLFSSL_TLSV1_1) != 1) &&
(wolfSSL_CTX_SetMinVersion(backend->ctx, WOLFSSL_TLSV1_2) != 1)
2018-03-02 02:10:36 +08:00
#ifdef WOLFSSL_TLS13
&& (wolfSSL_CTX_SetMinVersion(backend->ctx, WOLFSSL_TLSV1_3) != 1)
2018-03-02 02:10:36 +08:00
#endif
) {
failf(data, "SSL: couldn't set the minimum protocol version");
return CURLE_SSL_CONNECT_ERROR;
}
#endif
FALLTHROUGH();
default:
break;
}
ciphers = conn_config->cipher_list;
2017-01-07 06:00:45 +08:00
if(ciphers) {
if(!SSL_CTX_set_cipher_list(backend->ctx, ciphers)) {
2017-01-07 06:00:45 +08:00
failf(data, "failed setting cipher list: %s", ciphers);
return CURLE_SSL_CIPHER;
}
infof(data, "Cipher selection: %s", ciphers);
2017-01-07 06:00:45 +08:00
}
curves = conn_config->curves;
if(curves) {
#ifdef HAVE_LIBOQS
for(idx = 0; gnm[idx].name != NULL; idx++) {
if(strncmp(curves, gnm[idx].name, strlen(gnm[idx].name)) == 0) {
oqsAlg = gnm[idx].group;
break;
}
}
if(oqsAlg == 0)
#endif
{
if(!SSL_CTX_set1_curves_list(backend->ctx, curves)) {
failf(data, "failed setting curves list: '%s'", curves);
return CURLE_SSL_CIPHER;
}
}
}
#if !defined(NO_FILESYSTEM) && defined(WOLFSSL_SYS_CA_CERTS)
/* load native CA certificates */
if(ssl_config->native_ca_store) {
if(wolfSSL_CTX_load_system_CA_certs(backend->ctx) != WOLFSSL_SUCCESS) {
infof(data, "error importing native CA store, continuing anyway");
}
else {
imported_native_ca = true;
infof(data, "successfully imported native CA store");
}
}
#endif /* !NO_FILESYSTEM */
/* load certificate blob */
if(ca_info_blob) {
if(wolfSSL_CTX_load_verify_buffer(backend->ctx, ca_info_blob->data,
ca_info_blob->len,
SSL_FILETYPE_PEM) != SSL_SUCCESS) {
if(imported_native_ca) {
infof(data, "error importing CA certificate blob, continuing anyway");
}
else {
failf(data, "error importing CA certificate blob");
return CURLE_SSL_CACERT_BADFILE;
}
}
else {
imported_ca_info_blob = true;
infof(data, "successfully imported CA certificate blob");
}
}
#ifndef NO_FILESYSTEM
/* load trusted cacert from file if not blob */
if(ssl_cafile || ssl_capath) {
int rc =
wolfSSL_CTX_load_verify_locations_ex(backend->ctx,
ssl_cafile,
ssl_capath,
WOLFSSL_LOAD_FLAG_IGNORE_ERR);
if(SSL_SUCCESS != rc) {
if(conn_config->verifypeer && !imported_ca_info_blob &&
!imported_native_ca) {
/* Fail if we insist on successfully verifying the server. */
failf(data, "error setting certificate verify locations:"
" CAfile: %s CApath: %s",
ssl_cafile ? ssl_cafile : "none",
ssl_capath ? ssl_capath : "none");
return CURLE_SSL_CACERT_BADFILE;
}
else {
/* Just continue with a warning if no strict certificate
verification is required. */
infof(data, "error setting certificate verify locations,"
" continuing anyway:");
}
}
else {
/* Everything is fine. */
infof(data, "successfully set certificate verify locations:");
}
infof(data, " CAfile: %s", ssl_cafile ? ssl_cafile : "none");
infof(data, " CApath: %s", ssl_capath ? ssl_capath : "none");
}
/* Load the client certificate, and private key */
if(ssl_config->primary.clientcert && ssl_config->key) {
int file_type = do_file_type(ssl_config->cert_type);
if(file_type == WOLFSSL_FILETYPE_PEM) {
if(wolfSSL_CTX_use_certificate_chain_file(backend->ctx,
ssl_config->primary.clientcert)
!= 1) {
failf(data, "unable to use client certificate");
return CURLE_SSL_CONNECT_ERROR;
}
}
else if(file_type == WOLFSSL_FILETYPE_ASN1) {
if(wolfSSL_CTX_use_certificate_file(backend->ctx,
ssl_config->primary.clientcert,
file_type) != 1) {
failf(data, "unable to use client certificate");
return CURLE_SSL_CONNECT_ERROR;
}
}
else {
failf(data, "unknown cert type");
return CURLE_BAD_FUNCTION_ARGUMENT;
}
file_type = do_file_type(ssl_config->key_type);
if(wolfSSL_CTX_use_PrivateKey_file(backend->ctx, ssl_config->key,
file_type) != 1) {
failf(data, "unable to set private key");
return CURLE_SSL_CONNECT_ERROR;
}
}
#endif /* !NO_FILESYSTEM */
/* SSL always tries to verify the peer, this only says whether it should
* fail to connect if the verification fails, or if it should continue
* anyway. In the latter case the result of the verification is checked with
* SSL_get_verify_result() below. */
wolfSSL_CTX_set_verify(backend->ctx,
conn_config->verifypeer?SSL_VERIFY_PEER:
SSL_VERIFY_NONE, NULL);
#ifdef HAVE_SNI
if(sni && connssl->peer.sni) {
size_t sni_len = strlen(connssl->peer.sni);
if((sni_len < USHRT_MAX)) {
if(wolfSSL_CTX_UseSNI(backend->ctx, WOLFSSL_SNI_HOST_NAME,
connssl->peer.sni,
(unsigned short)sni_len) != 1) {
failf(data, "Failed to set SNI");
return CURLE_SSL_CONNECT_ERROR;
}
}
}
#endif
/* give application a chance to interfere with SSL set up. */
if(data->set.ssl.fsslctx) {
CURLcode result = (*data->set.ssl.fsslctx)(data, backend->ctx,
data->set.ssl.fsslctxp);
if(result) {
failf(data, "error signaled by ssl ctx callback");
return result;
}
}
#ifdef NO_FILESYSTEM
else if(conn_config->verifypeer) {
failf(data, "SSL: Certificates can't be loaded because wolfSSL was built"
" with \"no filesystem\". Either disable peer verification"
" (insecure) or if you are building an application with libcurl you"
" can load certificates via CURLOPT_SSL_CTX_FUNCTION.");
return CURLE_SSL_CONNECT_ERROR;
}
#endif
/* Let's make an SSL structure */
if(backend->handle)
wolfSSL_free(backend->handle);
backend->handle = wolfSSL_new(backend->ctx);
if(!backend->handle) {
failf(data, "SSL: couldn't create a handle");
return CURLE_OUT_OF_MEMORY;
}
#ifdef HAVE_LIBOQS
if(oqsAlg) {
if(wolfSSL_UseKeyShare(backend->handle, oqsAlg) != WOLFSSL_SUCCESS) {
failf(data, "unable to use oqs KEM");
}
}
#endif
2016-03-29 06:18:09 +08:00
#ifdef HAVE_ALPN
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
if(connssl->alpn) {
struct alpn_proto_buf proto;
CURLcode result;
2016-03-29 06:18:09 +08:00
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
result = Curl_alpn_to_proto_str(&proto, connssl->alpn);
if(result ||
wolfSSL_UseALPN(backend->handle, (char *)proto.data, proto.len,
2016-03-29 06:18:09 +08:00
WOLFSSL_ALPN_CONTINUE_ON_MISMATCH) != SSL_SUCCESS) {
failf(data, "SSL: failed setting ALPN protocols");
return CURLE_SSL_CONNECT_ERROR;
}
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
infof(data, VTLS_INFOF_ALPN_OFFER_1STR, proto.data);
2016-03-29 06:18:09 +08:00
}
#endif /* HAVE_ALPN */
#ifdef OPENSSL_EXTRA
if(Curl_tls_keylog_enabled()) {
/* Ensure the Client Random is preserved. */
wolfSSL_KeepArrays(backend->handle);
#if defined(HAVE_SECRET_CALLBACK) && defined(WOLFSSL_TLS13)
wolfSSL_set_tls13_secret_cb(backend->handle,
wolfssl_tls13_secret_callback, NULL);
#endif
}
#endif /* OPENSSL_EXTRA */
#ifdef HAVE_SECURE_RENEGOTIATION
if(wolfSSL_UseSecureRenegotiation(backend->handle) != SSL_SUCCESS) {
failf(data, "SSL: failed setting secure renegotiation");
return CURLE_SSL_CONNECT_ERROR;
}
#endif /* HAVE_SECURE_RENEGOTIATION */
/* Check if there's a cached ID we can/should use here! */
if(ssl_config->primary.sessionid) {
void *ssl_sessionid = NULL;
Curl_ssl_sessionid_lock(data);
if(!Curl_ssl_getsessionid(cf, data, &connssl->peer,
&ssl_sessionid, NULL)) {
/* we got a session id, use it! */
if(!SSL_set_session(backend->handle, ssl_sessionid)) {
Curl_ssl_delsessionid(data, ssl_sessionid);
infof(data, "Can't use session ID, going on without");
}
else
infof(data, "SSL reusing session ID");
}
Curl_ssl_sessionid_unlock(data);
}
#ifdef USE_BIO_CHAIN
{
WOLFSSL_BIO *bio;
bio = BIO_new(wolfssl_bio_cf_method);
if(!bio)
return CURLE_OUT_OF_MEMORY;
wolfSSL_BIO_set_data(bio, cf);
wolfSSL_set_bio(backend->handle, bio, bio);
}
#else /* USE_BIO_CHAIN */
/* pass the raw socket into the SSL layer */
if(!wolfSSL_set_fd(backend->handle,
(int)Curl_conn_cf_get_socket(cf, data))) {
failf(data, "SSL: SSL_set_fd failed");
return CURLE_SSL_CONNECT_ERROR;
}
#endif /* !USE_BIO_CHAIN */
connssl->connecting_state = ssl_connect_2;
return CURLE_OK;
}
static CURLcode
wolfssl_connect_step2(struct Curl_cfilter *cf, struct Curl_easy *data)
{
int ret = -1;
struct ssl_connect_data *connssl = cf->ctx;
cmake: add support for "unity" builds Aka "jumbo" or "amalgamation" builds. It means to compile all sources per target as a single C source. This is experimental. You can enable it by passing `-DCMAKE_UNITY_BUILD=ON` to cmake. It requires CMake 3.16 or newer. It makes builds (much) faster, allows for better optimizations and tends to promote less ambiguous code. Also add a new AppVeyor CI job and convert an existing one to use "unity" mode (one MSVC, one MinGW), and enable it for one macOS CI job. Fix related issues: - add missing include guard to `easy_lock.h`. - rename static variables and functions (and a macro) with names reused across sources, or shadowed by local variables. - add an `#undef` after use. - add a missing `#undef` before use. - move internal definitions from `ftp.h` to `ftp.c`. - `curl_memory.h` fixes to make it work when included repeatedly. - stop building/linking curlx bits twice for a static-mode curl tool. These caused doubly defined symbols in unity builds. - silence missing extern declarations compiler warning for ` _CRT_glob`. - fix extern declarations for `tool_freq` and `tool_isVistaOrGreater`. - fix colliding static symbols in debug mode: `debugtime()` and `statename`. - rename `ssl_backend_data` structure to unique names for each TLS-backend, along with the `ssl_connect_data` struct member referencing them. This required adding casts for each access. - add workaround for missing `[P]UNICODE_STRING` types in certain Windows builds when compiling `lib/ldap.c`. To support "unity" builds, we had to enable `SCHANNEL_USE_BLACKLISTS` for Schannel (a Windows `schannel.h` option) _globally_. This caused an indirect inclusion of Windows `schannel.h` from `ldap.c` via `winldap.h` to have it enabled as well. This requires `[P]UNICODE_STRING` types, which is apperantly not defined automatically (as seen with both MSVS and mingw-w64). This patch includes `<subauth.h>` to fix it. Ref: https://github.com/curl/curl/runs/13987772013 Ref: https://dev.azure.com/daniel0244/curl/_build/results?buildId=15827&view=logs&jobId=2c9f582d-e278-56b6-4354-f38a4d851906&j=2c9f582d-e278-56b6-4354-f38a4d851906&t=90509b00-34fa-5a81-35d7-5ed9569d331c - tweak unity builds to compile `lib/memdebug.c` separately in memory trace builds to avoid PP confusion. - force-disable unity for test programs. - do not compile and link libcurl sources to libtests _twice_ when libcurl is built in static mode. KNOWN ISSUES: - running tests with unity builds may fail in cases. - some build configurations/env may not compile in unity mode. E.g.: https://ci.appveyor.com/project/curlorg/curl/builds/47230972/job/51wfesgnfuauwl8q#L250 Ref: https://github.com/libssh2/libssh2/issues/1034 Ref: https://cmake.org/cmake/help/latest/prop_tgt/UNITY_BUILD.html Ref: https://en.wikipedia.org/wiki/Unity_build Closes #11095
2023-05-09 18:10:40 +08:00
struct wolfssl_ssl_backend_data *backend =
(struct wolfssl_ssl_backend_data *)connssl->backend;
struct ssl_primary_config *conn_config = Curl_ssl_cf_get_primary_config(cf);
#ifndef CURL_DISABLE_PROXY
const char * const pinnedpubkey = Curl_ssl_cf_is_proxy(cf)?
data->set.str[STRING_SSL_PINNEDPUBLICKEY_PROXY]:
data->set.str[STRING_SSL_PINNEDPUBLICKEY];
#else
const char * const pinnedpubkey = data->set.str[STRING_SSL_PINNEDPUBLICKEY];
#endif
DEBUGASSERT(backend);
wolfSSL_ERR_clear_error();
SSL: Several SSL-backend related fixes axTLS: This will make the axTLS backend perform the RFC2818 checks, honoring the VERIFYHOST setting similar to the OpenSSL backend. Generic for OpenSSL and axTLS: Move the hostcheck and cert_hostcheck functions from the lib/ssluse.c files to make them genericly available for both the OpenSSL, axTLS and other SSL backends. They are now in the new lib/hostcheck.c file. CyaSSL: CyaSSL now also has the RFC2818 checks enabled by default. There is a limitation that the verifyhost can not be enabled exclusively on the Subject CN field comparison. This SSL backend will thus behave like the NSS and the GnuTLS (meaning: RFC2818 ok, or bust). In other words: setting verifyhost to 0 or 1 will disable the Subject Alt Names checks too. Schannel: Updated the schannel information messages: Split the IP address usage message from the verifyhost setting and changed the message about disabling SNI (Server Name Indication, used in HTTP virtual hosting) into a message stating that the Subject Alternative Names checks are being disabled when verifyhost is set to 0 or 1. As a side effect of switching off the RFC2818 related servername checks with SCH_CRED_NO_SERVERNAME_CHECK (http://msdn.microsoft.com/en-us/library/aa923430.aspx) the SNI feature is being disabled. This effect is not documented in MSDN, but Wireshark output clearly shows the effect (details on the libcurl maillist). PolarSSL: Fix the prototype change in PolarSSL of ssl_set_session() and the move of the peer_cert from the ssl_context to the ssl_session. Found this change in the PolarSSL SVN between r1316 and r1317 where the POLARSSL_VERSION_NUMBER was at 0x01010100. But to accommodate the Ubuntu PolarSSL version 1.1.4 the check is to discriminate between lower then PolarSSL version 1.2.0 and 1.2.0 and higher. Note: The PolarSSL SVN trunk jumped from version 1.1.1 to 1.2.0. Generic: All the SSL backends are fixed and checked to work with the ssl.verifyhost as a boolean, which is an internal API change.
2012-11-03 09:06:51 +08:00
/* Enable RFC2818 checks */
if(conn_config->verifyhost) {
char *snihost = connssl->peer.sni?
connssl->peer.sni : connssl->peer.hostname;
if(wolfSSL_check_domain_name(backend->handle, snihost) == SSL_FAILURE)
return CURLE_SSL_CONNECT_ERROR;
SSL: Several SSL-backend related fixes axTLS: This will make the axTLS backend perform the RFC2818 checks, honoring the VERIFYHOST setting similar to the OpenSSL backend. Generic for OpenSSL and axTLS: Move the hostcheck and cert_hostcheck functions from the lib/ssluse.c files to make them genericly available for both the OpenSSL, axTLS and other SSL backends. They are now in the new lib/hostcheck.c file. CyaSSL: CyaSSL now also has the RFC2818 checks enabled by default. There is a limitation that the verifyhost can not be enabled exclusively on the Subject CN field comparison. This SSL backend will thus behave like the NSS and the GnuTLS (meaning: RFC2818 ok, or bust). In other words: setting verifyhost to 0 or 1 will disable the Subject Alt Names checks too. Schannel: Updated the schannel information messages: Split the IP address usage message from the verifyhost setting and changed the message about disabling SNI (Server Name Indication, used in HTTP virtual hosting) into a message stating that the Subject Alternative Names checks are being disabled when verifyhost is set to 0 or 1. As a side effect of switching off the RFC2818 related servername checks with SCH_CRED_NO_SERVERNAME_CHECK (http://msdn.microsoft.com/en-us/library/aa923430.aspx) the SNI feature is being disabled. This effect is not documented in MSDN, but Wireshark output clearly shows the effect (details on the libcurl maillist). PolarSSL: Fix the prototype change in PolarSSL of ssl_set_session() and the move of the peer_cert from the ssl_context to the ssl_session. Found this change in the PolarSSL SVN between r1316 and r1317 where the POLARSSL_VERSION_NUMBER was at 0x01010100. But to accommodate the Ubuntu PolarSSL version 1.1.4 the check is to discriminate between lower then PolarSSL version 1.2.0 and 1.2.0 and higher. Note: The PolarSSL SVN trunk jumped from version 1.1.1 to 1.2.0. Generic: All the SSL backends are fixed and checked to work with the ssl.verifyhost as a boolean, which is an internal API change.
2012-11-03 09:06:51 +08:00
}
ret = wolfSSL_connect(backend->handle);
#ifdef OPENSSL_EXTRA
if(Curl_tls_keylog_enabled()) {
/* If key logging is enabled, wait for the handshake to complete and then
* proceed with logging secrets (for TLS 1.2 or older).
*
* During the handshake (ret==-1), wolfSSL_want_read() is true as it waits
* for the server response. At that point the master secret is not yet
* available, so we must not try to read it.
* To log the secret on completion with a handshake failure, detect
* completion via the observation that there is nothing to read or write.
* Note that OpenSSL SSL_want_read() is always true here. If wolfSSL ever
* changes, the worst case is that no key is logged on error.
*/
if(ret == SSL_SUCCESS ||
(!wolfSSL_want_read(backend->handle) &&
!wolfSSL_want_write(backend->handle))) {
wolfssl_log_tls12_secret(backend->handle);
/* Client Random and master secrets are no longer needed, erase these.
* Ignored while the handshake is still in progress. */
wolfSSL_FreeArrays(backend->handle);
}
}
#endif /* OPENSSL_EXTRA */
if(ret != 1) {
char error_buffer[WOLFSSL_MAX_ERROR_SZ];
int detail = wolfSSL_get_error(backend->handle, ret);
if(SSL_ERROR_WANT_READ == detail) {
connssl->connecting_state = ssl_connect_2_reading;
return CURLE_OK;
}
SSL: Several SSL-backend related fixes axTLS: This will make the axTLS backend perform the RFC2818 checks, honoring the VERIFYHOST setting similar to the OpenSSL backend. Generic for OpenSSL and axTLS: Move the hostcheck and cert_hostcheck functions from the lib/ssluse.c files to make them genericly available for both the OpenSSL, axTLS and other SSL backends. They are now in the new lib/hostcheck.c file. CyaSSL: CyaSSL now also has the RFC2818 checks enabled by default. There is a limitation that the verifyhost can not be enabled exclusively on the Subject CN field comparison. This SSL backend will thus behave like the NSS and the GnuTLS (meaning: RFC2818 ok, or bust). In other words: setting verifyhost to 0 or 1 will disable the Subject Alt Names checks too. Schannel: Updated the schannel information messages: Split the IP address usage message from the verifyhost setting and changed the message about disabling SNI (Server Name Indication, used in HTTP virtual hosting) into a message stating that the Subject Alternative Names checks are being disabled when verifyhost is set to 0 or 1. As a side effect of switching off the RFC2818 related servername checks with SCH_CRED_NO_SERVERNAME_CHECK (http://msdn.microsoft.com/en-us/library/aa923430.aspx) the SNI feature is being disabled. This effect is not documented in MSDN, but Wireshark output clearly shows the effect (details on the libcurl maillist). PolarSSL: Fix the prototype change in PolarSSL of ssl_set_session() and the move of the peer_cert from the ssl_context to the ssl_session. Found this change in the PolarSSL SVN between r1316 and r1317 where the POLARSSL_VERSION_NUMBER was at 0x01010100. But to accommodate the Ubuntu PolarSSL version 1.1.4 the check is to discriminate between lower then PolarSSL version 1.2.0 and 1.2.0 and higher. Note: The PolarSSL SVN trunk jumped from version 1.1.1 to 1.2.0. Generic: All the SSL backends are fixed and checked to work with the ssl.verifyhost as a boolean, which is an internal API change.
2012-11-03 09:06:51 +08:00
else if(SSL_ERROR_WANT_WRITE == detail) {
connssl->connecting_state = ssl_connect_2_writing;
return CURLE_OK;
}
SSL: Several SSL-backend related fixes axTLS: This will make the axTLS backend perform the RFC2818 checks, honoring the VERIFYHOST setting similar to the OpenSSL backend. Generic for OpenSSL and axTLS: Move the hostcheck and cert_hostcheck functions from the lib/ssluse.c files to make them genericly available for both the OpenSSL, axTLS and other SSL backends. They are now in the new lib/hostcheck.c file. CyaSSL: CyaSSL now also has the RFC2818 checks enabled by default. There is a limitation that the verifyhost can not be enabled exclusively on the Subject CN field comparison. This SSL backend will thus behave like the NSS and the GnuTLS (meaning: RFC2818 ok, or bust). In other words: setting verifyhost to 0 or 1 will disable the Subject Alt Names checks too. Schannel: Updated the schannel information messages: Split the IP address usage message from the verifyhost setting and changed the message about disabling SNI (Server Name Indication, used in HTTP virtual hosting) into a message stating that the Subject Alternative Names checks are being disabled when verifyhost is set to 0 or 1. As a side effect of switching off the RFC2818 related servername checks with SCH_CRED_NO_SERVERNAME_CHECK (http://msdn.microsoft.com/en-us/library/aa923430.aspx) the SNI feature is being disabled. This effect is not documented in MSDN, but Wireshark output clearly shows the effect (details on the libcurl maillist). PolarSSL: Fix the prototype change in PolarSSL of ssl_set_session() and the move of the peer_cert from the ssl_context to the ssl_session. Found this change in the PolarSSL SVN between r1316 and r1317 where the POLARSSL_VERSION_NUMBER was at 0x01010100. But to accommodate the Ubuntu PolarSSL version 1.1.4 the check is to discriminate between lower then PolarSSL version 1.2.0 and 1.2.0 and higher. Note: The PolarSSL SVN trunk jumped from version 1.1.1 to 1.2.0. Generic: All the SSL backends are fixed and checked to work with the ssl.verifyhost as a boolean, which is an internal API change.
2012-11-03 09:06:51 +08:00
/* There is no easy way to override only the CN matching.
* This will enable the override of both mismatching SubjectAltNames
* as also mismatching CN fields */
else if(DOMAIN_NAME_MISMATCH == detail) {
#if 1
failf(data, " subject alt name(s) or common name do not match \"%s\"",
connssl->peer.dispname);
SSL: Several SSL-backend related fixes axTLS: This will make the axTLS backend perform the RFC2818 checks, honoring the VERIFYHOST setting similar to the OpenSSL backend. Generic for OpenSSL and axTLS: Move the hostcheck and cert_hostcheck functions from the lib/ssluse.c files to make them genericly available for both the OpenSSL, axTLS and other SSL backends. They are now in the new lib/hostcheck.c file. CyaSSL: CyaSSL now also has the RFC2818 checks enabled by default. There is a limitation that the verifyhost can not be enabled exclusively on the Subject CN field comparison. This SSL backend will thus behave like the NSS and the GnuTLS (meaning: RFC2818 ok, or bust). In other words: setting verifyhost to 0 or 1 will disable the Subject Alt Names checks too. Schannel: Updated the schannel information messages: Split the IP address usage message from the verifyhost setting and changed the message about disabling SNI (Server Name Indication, used in HTTP virtual hosting) into a message stating that the Subject Alternative Names checks are being disabled when verifyhost is set to 0 or 1. As a side effect of switching off the RFC2818 related servername checks with SCH_CRED_NO_SERVERNAME_CHECK (http://msdn.microsoft.com/en-us/library/aa923430.aspx) the SNI feature is being disabled. This effect is not documented in MSDN, but Wireshark output clearly shows the effect (details on the libcurl maillist). PolarSSL: Fix the prototype change in PolarSSL of ssl_set_session() and the move of the peer_cert from the ssl_context to the ssl_session. Found this change in the PolarSSL SVN between r1316 and r1317 where the POLARSSL_VERSION_NUMBER was at 0x01010100. But to accommodate the Ubuntu PolarSSL version 1.1.4 the check is to discriminate between lower then PolarSSL version 1.2.0 and 1.2.0 and higher. Note: The PolarSSL SVN trunk jumped from version 1.1.1 to 1.2.0. Generic: All the SSL backends are fixed and checked to work with the ssl.verifyhost as a boolean, which is an internal API change.
2012-11-03 09:06:51 +08:00
return CURLE_PEER_FAILED_VERIFICATION;
#else
/* When the wolfssl_check_domain_name() is used and you desire to
* continue on a DOMAIN_NAME_MISMATCH, i.e. 'ssl_config.verifyhost
* == 0', CyaSSL version 2.4.0 will fail with an INCOMPLETE_DATA
* error. The only way to do this is currently to switch the
* Wolfssl_check_domain_name() in and out based on the
* 'ssl_config.verifyhost' value. */
if(conn_config->verifyhost) {
SSL: Several SSL-backend related fixes axTLS: This will make the axTLS backend perform the RFC2818 checks, honoring the VERIFYHOST setting similar to the OpenSSL backend. Generic for OpenSSL and axTLS: Move the hostcheck and cert_hostcheck functions from the lib/ssluse.c files to make them genericly available for both the OpenSSL, axTLS and other SSL backends. They are now in the new lib/hostcheck.c file. CyaSSL: CyaSSL now also has the RFC2818 checks enabled by default. There is a limitation that the verifyhost can not be enabled exclusively on the Subject CN field comparison. This SSL backend will thus behave like the NSS and the GnuTLS (meaning: RFC2818 ok, or bust). In other words: setting verifyhost to 0 or 1 will disable the Subject Alt Names checks too. Schannel: Updated the schannel information messages: Split the IP address usage message from the verifyhost setting and changed the message about disabling SNI (Server Name Indication, used in HTTP virtual hosting) into a message stating that the Subject Alternative Names checks are being disabled when verifyhost is set to 0 or 1. As a side effect of switching off the RFC2818 related servername checks with SCH_CRED_NO_SERVERNAME_CHECK (http://msdn.microsoft.com/en-us/library/aa923430.aspx) the SNI feature is being disabled. This effect is not documented in MSDN, but Wireshark output clearly shows the effect (details on the libcurl maillist). PolarSSL: Fix the prototype change in PolarSSL of ssl_set_session() and the move of the peer_cert from the ssl_context to the ssl_session. Found this change in the PolarSSL SVN between r1316 and r1317 where the POLARSSL_VERSION_NUMBER was at 0x01010100. But to accommodate the Ubuntu PolarSSL version 1.1.4 the check is to discriminate between lower then PolarSSL version 1.2.0 and 1.2.0 and higher. Note: The PolarSSL SVN trunk jumped from version 1.1.1 to 1.2.0. Generic: All the SSL backends are fixed and checked to work with the ssl.verifyhost as a boolean, which is an internal API change.
2012-11-03 09:06:51 +08:00
failf(data,
" subject alt name(s) or common name do not match \"%s\"\n",
connssl->dispname);
SSL: Several SSL-backend related fixes axTLS: This will make the axTLS backend perform the RFC2818 checks, honoring the VERIFYHOST setting similar to the OpenSSL backend. Generic for OpenSSL and axTLS: Move the hostcheck and cert_hostcheck functions from the lib/ssluse.c files to make them genericly available for both the OpenSSL, axTLS and other SSL backends. They are now in the new lib/hostcheck.c file. CyaSSL: CyaSSL now also has the RFC2818 checks enabled by default. There is a limitation that the verifyhost can not be enabled exclusively on the Subject CN field comparison. This SSL backend will thus behave like the NSS and the GnuTLS (meaning: RFC2818 ok, or bust). In other words: setting verifyhost to 0 or 1 will disable the Subject Alt Names checks too. Schannel: Updated the schannel information messages: Split the IP address usage message from the verifyhost setting and changed the message about disabling SNI (Server Name Indication, used in HTTP virtual hosting) into a message stating that the Subject Alternative Names checks are being disabled when verifyhost is set to 0 or 1. As a side effect of switching off the RFC2818 related servername checks with SCH_CRED_NO_SERVERNAME_CHECK (http://msdn.microsoft.com/en-us/library/aa923430.aspx) the SNI feature is being disabled. This effect is not documented in MSDN, but Wireshark output clearly shows the effect (details on the libcurl maillist). PolarSSL: Fix the prototype change in PolarSSL of ssl_set_session() and the move of the peer_cert from the ssl_context to the ssl_session. Found this change in the PolarSSL SVN between r1316 and r1317 where the POLARSSL_VERSION_NUMBER was at 0x01010100. But to accommodate the Ubuntu PolarSSL version 1.1.4 the check is to discriminate between lower then PolarSSL version 1.2.0 and 1.2.0 and higher. Note: The PolarSSL SVN trunk jumped from version 1.1.1 to 1.2.0. Generic: All the SSL backends are fixed and checked to work with the ssl.verifyhost as a boolean, which is an internal API change.
2012-11-03 09:06:51 +08:00
return CURLE_PEER_FAILED_VERIFICATION;
}
else {
infof(data,
" subject alt name(s) and/or common name do not match \"%s\"",
connssl->dispname);
SSL: Several SSL-backend related fixes axTLS: This will make the axTLS backend perform the RFC2818 checks, honoring the VERIFYHOST setting similar to the OpenSSL backend. Generic for OpenSSL and axTLS: Move the hostcheck and cert_hostcheck functions from the lib/ssluse.c files to make them genericly available for both the OpenSSL, axTLS and other SSL backends. They are now in the new lib/hostcheck.c file. CyaSSL: CyaSSL now also has the RFC2818 checks enabled by default. There is a limitation that the verifyhost can not be enabled exclusively on the Subject CN field comparison. This SSL backend will thus behave like the NSS and the GnuTLS (meaning: RFC2818 ok, or bust). In other words: setting verifyhost to 0 or 1 will disable the Subject Alt Names checks too. Schannel: Updated the schannel information messages: Split the IP address usage message from the verifyhost setting and changed the message about disabling SNI (Server Name Indication, used in HTTP virtual hosting) into a message stating that the Subject Alternative Names checks are being disabled when verifyhost is set to 0 or 1. As a side effect of switching off the RFC2818 related servername checks with SCH_CRED_NO_SERVERNAME_CHECK (http://msdn.microsoft.com/en-us/library/aa923430.aspx) the SNI feature is being disabled. This effect is not documented in MSDN, but Wireshark output clearly shows the effect (details on the libcurl maillist). PolarSSL: Fix the prototype change in PolarSSL of ssl_set_session() and the move of the peer_cert from the ssl_context to the ssl_session. Found this change in the PolarSSL SVN between r1316 and r1317 where the POLARSSL_VERSION_NUMBER was at 0x01010100. But to accommodate the Ubuntu PolarSSL version 1.1.4 the check is to discriminate between lower then PolarSSL version 1.2.0 and 1.2.0 and higher. Note: The PolarSSL SVN trunk jumped from version 1.1.1 to 1.2.0. Generic: All the SSL backends are fixed and checked to work with the ssl.verifyhost as a boolean, which is an internal API change.
2012-11-03 09:06:51 +08:00
return CURLE_OK;
}
#endif
}
#if LIBWOLFSSL_VERSION_HEX >= 0x02007000 /* 2.7.0 */
else if(ASN_NO_SIGNER_E == detail) {
if(conn_config->verifypeer) {
failf(data, " CA signer not available for verification");
return CURLE_SSL_CACERT_BADFILE;
}
else {
/* Just continue with a warning if no strict certificate
verification is required. */
infof(data, "CA signer not available for verification, "
"continuing anyway");
}
}
#endif
else if(backend->io_result == CURLE_AGAIN) {
return CURLE_OK;
}
SSL: Several SSL-backend related fixes axTLS: This will make the axTLS backend perform the RFC2818 checks, honoring the VERIFYHOST setting similar to the OpenSSL backend. Generic for OpenSSL and axTLS: Move the hostcheck and cert_hostcheck functions from the lib/ssluse.c files to make them genericly available for both the OpenSSL, axTLS and other SSL backends. They are now in the new lib/hostcheck.c file. CyaSSL: CyaSSL now also has the RFC2818 checks enabled by default. There is a limitation that the verifyhost can not be enabled exclusively on the Subject CN field comparison. This SSL backend will thus behave like the NSS and the GnuTLS (meaning: RFC2818 ok, or bust). In other words: setting verifyhost to 0 or 1 will disable the Subject Alt Names checks too. Schannel: Updated the schannel information messages: Split the IP address usage message from the verifyhost setting and changed the message about disabling SNI (Server Name Indication, used in HTTP virtual hosting) into a message stating that the Subject Alternative Names checks are being disabled when verifyhost is set to 0 or 1. As a side effect of switching off the RFC2818 related servername checks with SCH_CRED_NO_SERVERNAME_CHECK (http://msdn.microsoft.com/en-us/library/aa923430.aspx) the SNI feature is being disabled. This effect is not documented in MSDN, but Wireshark output clearly shows the effect (details on the libcurl maillist). PolarSSL: Fix the prototype change in PolarSSL of ssl_set_session() and the move of the peer_cert from the ssl_context to the ssl_session. Found this change in the PolarSSL SVN between r1316 and r1317 where the POLARSSL_VERSION_NUMBER was at 0x01010100. But to accommodate the Ubuntu PolarSSL version 1.1.4 the check is to discriminate between lower then PolarSSL version 1.2.0 and 1.2.0 and higher. Note: The PolarSSL SVN trunk jumped from version 1.1.1 to 1.2.0. Generic: All the SSL backends are fixed and checked to work with the ssl.verifyhost as a boolean, which is an internal API change.
2012-11-03 09:06:51 +08:00
else {
failf(data, "SSL_connect failed with error %d: %s", detail,
wolfSSL_ERR_error_string(detail, error_buffer));
SSL: Several SSL-backend related fixes axTLS: This will make the axTLS backend perform the RFC2818 checks, honoring the VERIFYHOST setting similar to the OpenSSL backend. Generic for OpenSSL and axTLS: Move the hostcheck and cert_hostcheck functions from the lib/ssluse.c files to make them genericly available for both the OpenSSL, axTLS and other SSL backends. They are now in the new lib/hostcheck.c file. CyaSSL: CyaSSL now also has the RFC2818 checks enabled by default. There is a limitation that the verifyhost can not be enabled exclusively on the Subject CN field comparison. This SSL backend will thus behave like the NSS and the GnuTLS (meaning: RFC2818 ok, or bust). In other words: setting verifyhost to 0 or 1 will disable the Subject Alt Names checks too. Schannel: Updated the schannel information messages: Split the IP address usage message from the verifyhost setting and changed the message about disabling SNI (Server Name Indication, used in HTTP virtual hosting) into a message stating that the Subject Alternative Names checks are being disabled when verifyhost is set to 0 or 1. As a side effect of switching off the RFC2818 related servername checks with SCH_CRED_NO_SERVERNAME_CHECK (http://msdn.microsoft.com/en-us/library/aa923430.aspx) the SNI feature is being disabled. This effect is not documented in MSDN, but Wireshark output clearly shows the effect (details on the libcurl maillist). PolarSSL: Fix the prototype change in PolarSSL of ssl_set_session() and the move of the peer_cert from the ssl_context to the ssl_session. Found this change in the PolarSSL SVN between r1316 and r1317 where the POLARSSL_VERSION_NUMBER was at 0x01010100. But to accommodate the Ubuntu PolarSSL version 1.1.4 the check is to discriminate between lower then PolarSSL version 1.2.0 and 1.2.0 and higher. Note: The PolarSSL SVN trunk jumped from version 1.1.1 to 1.2.0. Generic: All the SSL backends are fixed and checked to work with the ssl.verifyhost as a boolean, which is an internal API change.
2012-11-03 09:06:51 +08:00
return CURLE_SSL_CONNECT_ERROR;
}
}
if(pinnedpubkey) {
2016-03-29 06:18:09 +08:00
#ifdef KEEP_PEER_CERT
X509 *x509;
const char *x509_der;
int x509_der_len;
struct Curl_X509certificate x509_parsed;
struct Curl_asn1Element *pubkey;
CURLcode result;
x509 = wolfSSL_get_peer_certificate(backend->handle);
if(!x509) {
failf(data, "SSL: failed retrieving server certificate");
return CURLE_SSL_PINNEDPUBKEYNOTMATCH;
}
x509_der = (const char *)wolfSSL_X509_get_der(x509, &x509_der_len);
if(!x509_der) {
failf(data, "SSL: failed retrieving ASN.1 server certificate");
return CURLE_SSL_PINNEDPUBKEYNOTMATCH;
}
memset(&x509_parsed, 0, sizeof(x509_parsed));
if(Curl_parseX509(&x509_parsed, x509_der, x509_der + x509_der_len))
return CURLE_SSL_PINNEDPUBKEYNOTMATCH;
pubkey = &x509_parsed.subjectPublicKeyInfo;
if(!pubkey->header || pubkey->end <= pubkey->header) {
failf(data, "SSL: failed retrieving public key from server certificate");
return CURLE_SSL_PINNEDPUBKEYNOTMATCH;
}
result = Curl_pin_peer_pubkey(data,
pinnedpubkey,
(const unsigned char *)pubkey->header,
(size_t)(pubkey->end - pubkey->header));
wolfSSL_FreeX509(x509);
if(result) {
failf(data, "SSL: public key does not match pinned public key");
return result;
}
#else
failf(data, "Library lacks pinning support built-in");
return CURLE_NOT_BUILT_IN;
#endif
}
2016-03-29 06:18:09 +08:00
#ifdef HAVE_ALPN
if(connssl->alpn) {
2016-03-29 06:18:09 +08:00
int rc;
char *protocol = NULL;
unsigned short protocol_len = 0;
rc = wolfSSL_ALPN_GetProtocol(backend->handle, &protocol, &protocol_len);
2016-03-29 06:18:09 +08:00
if(rc == SSL_SUCCESS) {
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
Curl_alpn_set_negotiated(cf, data, (const unsigned char *)protocol,
protocol_len);
2016-03-29 06:18:09 +08:00
}
else if(rc == SSL_ALPN_NOT_FOUND)
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
Curl_alpn_set_negotiated(cf, data, NULL, 0);
2016-03-29 06:18:09 +08:00
else {
failf(data, "ALPN, failure getting protocol, error %d", rc);
return CURLE_SSL_CONNECT_ERROR;
}
}
#endif /* HAVE_ALPN */
connssl->connecting_state = ssl_connect_3;
#if (LIBWOLFSSL_VERSION_HEX >= 0x03009010)
infof(data, "SSL connection using %s / %s",
wolfSSL_get_version(backend->handle),
wolfSSL_get_cipher_name(backend->handle));
#else
infof(data, "SSL connected");
#endif
return CURLE_OK;
}
static CURLcode
wolfssl_connect_step3(struct Curl_cfilter *cf, struct Curl_easy *data)
{
CURLcode result = CURLE_OK;
struct ssl_connect_data *connssl = cf->ctx;
cmake: add support for "unity" builds Aka "jumbo" or "amalgamation" builds. It means to compile all sources per target as a single C source. This is experimental. You can enable it by passing `-DCMAKE_UNITY_BUILD=ON` to cmake. It requires CMake 3.16 or newer. It makes builds (much) faster, allows for better optimizations and tends to promote less ambiguous code. Also add a new AppVeyor CI job and convert an existing one to use "unity" mode (one MSVC, one MinGW), and enable it for one macOS CI job. Fix related issues: - add missing include guard to `easy_lock.h`. - rename static variables and functions (and a macro) with names reused across sources, or shadowed by local variables. - add an `#undef` after use. - add a missing `#undef` before use. - move internal definitions from `ftp.h` to `ftp.c`. - `curl_memory.h` fixes to make it work when included repeatedly. - stop building/linking curlx bits twice for a static-mode curl tool. These caused doubly defined symbols in unity builds. - silence missing extern declarations compiler warning for ` _CRT_glob`. - fix extern declarations for `tool_freq` and `tool_isVistaOrGreater`. - fix colliding static symbols in debug mode: `debugtime()` and `statename`. - rename `ssl_backend_data` structure to unique names for each TLS-backend, along with the `ssl_connect_data` struct member referencing them. This required adding casts for each access. - add workaround for missing `[P]UNICODE_STRING` types in certain Windows builds when compiling `lib/ldap.c`. To support "unity" builds, we had to enable `SCHANNEL_USE_BLACKLISTS` for Schannel (a Windows `schannel.h` option) _globally_. This caused an indirect inclusion of Windows `schannel.h` from `ldap.c` via `winldap.h` to have it enabled as well. This requires `[P]UNICODE_STRING` types, which is apperantly not defined automatically (as seen with both MSVS and mingw-w64). This patch includes `<subauth.h>` to fix it. Ref: https://github.com/curl/curl/runs/13987772013 Ref: https://dev.azure.com/daniel0244/curl/_build/results?buildId=15827&view=logs&jobId=2c9f582d-e278-56b6-4354-f38a4d851906&j=2c9f582d-e278-56b6-4354-f38a4d851906&t=90509b00-34fa-5a81-35d7-5ed9569d331c - tweak unity builds to compile `lib/memdebug.c` separately in memory trace builds to avoid PP confusion. - force-disable unity for test programs. - do not compile and link libcurl sources to libtests _twice_ when libcurl is built in static mode. KNOWN ISSUES: - running tests with unity builds may fail in cases. - some build configurations/env may not compile in unity mode. E.g.: https://ci.appveyor.com/project/curlorg/curl/builds/47230972/job/51wfesgnfuauwl8q#L250 Ref: https://github.com/libssh2/libssh2/issues/1034 Ref: https://cmake.org/cmake/help/latest/prop_tgt/UNITY_BUILD.html Ref: https://en.wikipedia.org/wiki/Unity_build Closes #11095
2023-05-09 18:10:40 +08:00
struct wolfssl_ssl_backend_data *backend =
(struct wolfssl_ssl_backend_data *)connssl->backend;
const struct ssl_config_data *ssl_config = Curl_ssl_cf_get_config(cf, data);
DEBUGASSERT(ssl_connect_3 == connssl->connecting_state);
DEBUGASSERT(backend);
if(ssl_config->primary.sessionid) {
bool incache;
bool added = FALSE;
void *old_ssl_sessionid = NULL;
/* wolfSSL_get1_session allocates memory that has to be freed. */
WOLFSSL_SESSION *our_ssl_sessionid = wolfSSL_get1_session(backend->handle);
if(our_ssl_sessionid) {
Curl_ssl_sessionid_lock(data);
incache = !(Curl_ssl_getsessionid(cf, data, &connssl->peer,
&old_ssl_sessionid, NULL));
if(incache) {
if(old_ssl_sessionid != our_ssl_sessionid) {
infof(data, "old SSL session ID is stale, removing");
Curl_ssl_delsessionid(data, old_ssl_sessionid);
incache = FALSE;
}
}
if(!incache) {
result = Curl_ssl_addsessionid(cf, data, &connssl->peer,
our_ssl_sessionid, 0, NULL);
if(result) {
Curl_ssl_sessionid_unlock(data);
wolfSSL_SESSION_free(our_ssl_sessionid);
failf(data, "failed to store ssl session");
return result;
}
else {
added = TRUE;
}
}
Curl_ssl_sessionid_unlock(data);
if(!added) {
/* If the session info wasn't added to the cache, free our copy. */
wolfSSL_SESSION_free(our_ssl_sessionid);
}
}
}
connssl->connecting_state = ssl_connect_done;
return result;
}
static ssize_t wolfssl_send(struct Curl_cfilter *cf,
struct Curl_easy *data,
const void *mem,
size_t len,
CURLcode *curlcode)
{
struct ssl_connect_data *connssl = cf->ctx;
cmake: add support for "unity" builds Aka "jumbo" or "amalgamation" builds. It means to compile all sources per target as a single C source. This is experimental. You can enable it by passing `-DCMAKE_UNITY_BUILD=ON` to cmake. It requires CMake 3.16 or newer. It makes builds (much) faster, allows for better optimizations and tends to promote less ambiguous code. Also add a new AppVeyor CI job and convert an existing one to use "unity" mode (one MSVC, one MinGW), and enable it for one macOS CI job. Fix related issues: - add missing include guard to `easy_lock.h`. - rename static variables and functions (and a macro) with names reused across sources, or shadowed by local variables. - add an `#undef` after use. - add a missing `#undef` before use. - move internal definitions from `ftp.h` to `ftp.c`. - `curl_memory.h` fixes to make it work when included repeatedly. - stop building/linking curlx bits twice for a static-mode curl tool. These caused doubly defined symbols in unity builds. - silence missing extern declarations compiler warning for ` _CRT_glob`. - fix extern declarations for `tool_freq` and `tool_isVistaOrGreater`. - fix colliding static symbols in debug mode: `debugtime()` and `statename`. - rename `ssl_backend_data` structure to unique names for each TLS-backend, along with the `ssl_connect_data` struct member referencing them. This required adding casts for each access. - add workaround for missing `[P]UNICODE_STRING` types in certain Windows builds when compiling `lib/ldap.c`. To support "unity" builds, we had to enable `SCHANNEL_USE_BLACKLISTS` for Schannel (a Windows `schannel.h` option) _globally_. This caused an indirect inclusion of Windows `schannel.h` from `ldap.c` via `winldap.h` to have it enabled as well. This requires `[P]UNICODE_STRING` types, which is apperantly not defined automatically (as seen with both MSVS and mingw-w64). This patch includes `<subauth.h>` to fix it. Ref: https://github.com/curl/curl/runs/13987772013 Ref: https://dev.azure.com/daniel0244/curl/_build/results?buildId=15827&view=logs&jobId=2c9f582d-e278-56b6-4354-f38a4d851906&j=2c9f582d-e278-56b6-4354-f38a4d851906&t=90509b00-34fa-5a81-35d7-5ed9569d331c - tweak unity builds to compile `lib/memdebug.c` separately in memory trace builds to avoid PP confusion. - force-disable unity for test programs. - do not compile and link libcurl sources to libtests _twice_ when libcurl is built in static mode. KNOWN ISSUES: - running tests with unity builds may fail in cases. - some build configurations/env may not compile in unity mode. E.g.: https://ci.appveyor.com/project/curlorg/curl/builds/47230972/job/51wfesgnfuauwl8q#L250 Ref: https://github.com/libssh2/libssh2/issues/1034 Ref: https://cmake.org/cmake/help/latest/prop_tgt/UNITY_BUILD.html Ref: https://en.wikipedia.org/wiki/Unity_build Closes #11095
2023-05-09 18:10:40 +08:00
struct wolfssl_ssl_backend_data *backend =
(struct wolfssl_ssl_backend_data *)connssl->backend;
char error_buffer[WOLFSSL_MAX_ERROR_SZ];
int memlen = (len > (size_t)INT_MAX) ? INT_MAX : (int)len;
int rc;
DEBUGASSERT(backend);
wolfSSL_ERR_clear_error();
rc = wolfSSL_write(backend->handle, mem, memlen);
if(rc <= 0) {
int err = wolfSSL_get_error(backend->handle, rc);
switch(err) {
case SSL_ERROR_WANT_READ:
case SSL_ERROR_WANT_WRITE:
/* there's data pending, re-invoke SSL_write() */
CURL_TRC_CF(data, cf, "wolfssl_send(len=%zu) -> AGAIN", len);
*curlcode = CURLE_AGAIN;
return -1;
default:
if(backend->io_result == CURLE_AGAIN) {
CURL_TRC_CF(data, cf, "wolfssl_send(len=%zu) -> AGAIN", len);
*curlcode = CURLE_AGAIN;
return -1;
}
CURL_TRC_CF(data, cf, "wolfssl_send(len=%zu) -> %d, %d", len, rc, err);
failf(data, "SSL write: %s, errno %d",
wolfSSL_ERR_error_string(err, error_buffer),
SOCKERRNO);
*curlcode = CURLE_SEND_ERROR;
return -1;
}
}
CURL_TRC_CF(data, cf, "wolfssl_send(len=%zu) -> %d", len, rc);
return rc;
}
static void wolfssl_close(struct Curl_cfilter *cf, struct Curl_easy *data)
{
struct ssl_connect_data *connssl = cf->ctx;
cmake: add support for "unity" builds Aka "jumbo" or "amalgamation" builds. It means to compile all sources per target as a single C source. This is experimental. You can enable it by passing `-DCMAKE_UNITY_BUILD=ON` to cmake. It requires CMake 3.16 or newer. It makes builds (much) faster, allows for better optimizations and tends to promote less ambiguous code. Also add a new AppVeyor CI job and convert an existing one to use "unity" mode (one MSVC, one MinGW), and enable it for one macOS CI job. Fix related issues: - add missing include guard to `easy_lock.h`. - rename static variables and functions (and a macro) with names reused across sources, or shadowed by local variables. - add an `#undef` after use. - add a missing `#undef` before use. - move internal definitions from `ftp.h` to `ftp.c`. - `curl_memory.h` fixes to make it work when included repeatedly. - stop building/linking curlx bits twice for a static-mode curl tool. These caused doubly defined symbols in unity builds. - silence missing extern declarations compiler warning for ` _CRT_glob`. - fix extern declarations for `tool_freq` and `tool_isVistaOrGreater`. - fix colliding static symbols in debug mode: `debugtime()` and `statename`. - rename `ssl_backend_data` structure to unique names for each TLS-backend, along with the `ssl_connect_data` struct member referencing them. This required adding casts for each access. - add workaround for missing `[P]UNICODE_STRING` types in certain Windows builds when compiling `lib/ldap.c`. To support "unity" builds, we had to enable `SCHANNEL_USE_BLACKLISTS` for Schannel (a Windows `schannel.h` option) _globally_. This caused an indirect inclusion of Windows `schannel.h` from `ldap.c` via `winldap.h` to have it enabled as well. This requires `[P]UNICODE_STRING` types, which is apperantly not defined automatically (as seen with both MSVS and mingw-w64). This patch includes `<subauth.h>` to fix it. Ref: https://github.com/curl/curl/runs/13987772013 Ref: https://dev.azure.com/daniel0244/curl/_build/results?buildId=15827&view=logs&jobId=2c9f582d-e278-56b6-4354-f38a4d851906&j=2c9f582d-e278-56b6-4354-f38a4d851906&t=90509b00-34fa-5a81-35d7-5ed9569d331c - tweak unity builds to compile `lib/memdebug.c` separately in memory trace builds to avoid PP confusion. - force-disable unity for test programs. - do not compile and link libcurl sources to libtests _twice_ when libcurl is built in static mode. KNOWN ISSUES: - running tests with unity builds may fail in cases. - some build configurations/env may not compile in unity mode. E.g.: https://ci.appveyor.com/project/curlorg/curl/builds/47230972/job/51wfesgnfuauwl8q#L250 Ref: https://github.com/libssh2/libssh2/issues/1034 Ref: https://cmake.org/cmake/help/latest/prop_tgt/UNITY_BUILD.html Ref: https://en.wikipedia.org/wiki/Unity_build Closes #11095
2023-05-09 18:10:40 +08:00
struct wolfssl_ssl_backend_data *backend =
(struct wolfssl_ssl_backend_data *)connssl->backend;
(void) data;
DEBUGASSERT(backend);
if(backend->handle) {
char buf[32];
/* Maybe the server has already sent a close notify alert.
Read it to avoid an RST on the TCP connection. */
(void)wolfSSL_read(backend->handle, buf, (int)sizeof(buf));
if(!connssl->peer_closed)
(void)wolfSSL_shutdown(backend->handle);
wolfSSL_free(backend->handle);
backend->handle = NULL;
}
if(backend->ctx) {
wolfSSL_CTX_free(backend->ctx);
backend->ctx = NULL;
}
}
static ssize_t wolfssl_recv(struct Curl_cfilter *cf,
struct Curl_easy *data,
char *buf, size_t blen,
CURLcode *curlcode)
{
struct ssl_connect_data *connssl = cf->ctx;
cmake: add support for "unity" builds Aka "jumbo" or "amalgamation" builds. It means to compile all sources per target as a single C source. This is experimental. You can enable it by passing `-DCMAKE_UNITY_BUILD=ON` to cmake. It requires CMake 3.16 or newer. It makes builds (much) faster, allows for better optimizations and tends to promote less ambiguous code. Also add a new AppVeyor CI job and convert an existing one to use "unity" mode (one MSVC, one MinGW), and enable it for one macOS CI job. Fix related issues: - add missing include guard to `easy_lock.h`. - rename static variables and functions (and a macro) with names reused across sources, or shadowed by local variables. - add an `#undef` after use. - add a missing `#undef` before use. - move internal definitions from `ftp.h` to `ftp.c`. - `curl_memory.h` fixes to make it work when included repeatedly. - stop building/linking curlx bits twice for a static-mode curl tool. These caused doubly defined symbols in unity builds. - silence missing extern declarations compiler warning for ` _CRT_glob`. - fix extern declarations for `tool_freq` and `tool_isVistaOrGreater`. - fix colliding static symbols in debug mode: `debugtime()` and `statename`. - rename `ssl_backend_data` structure to unique names for each TLS-backend, along with the `ssl_connect_data` struct member referencing them. This required adding casts for each access. - add workaround for missing `[P]UNICODE_STRING` types in certain Windows builds when compiling `lib/ldap.c`. To support "unity" builds, we had to enable `SCHANNEL_USE_BLACKLISTS` for Schannel (a Windows `schannel.h` option) _globally_. This caused an indirect inclusion of Windows `schannel.h` from `ldap.c` via `winldap.h` to have it enabled as well. This requires `[P]UNICODE_STRING` types, which is apperantly not defined automatically (as seen with both MSVS and mingw-w64). This patch includes `<subauth.h>` to fix it. Ref: https://github.com/curl/curl/runs/13987772013 Ref: https://dev.azure.com/daniel0244/curl/_build/results?buildId=15827&view=logs&jobId=2c9f582d-e278-56b6-4354-f38a4d851906&j=2c9f582d-e278-56b6-4354-f38a4d851906&t=90509b00-34fa-5a81-35d7-5ed9569d331c - tweak unity builds to compile `lib/memdebug.c` separately in memory trace builds to avoid PP confusion. - force-disable unity for test programs. - do not compile and link libcurl sources to libtests _twice_ when libcurl is built in static mode. KNOWN ISSUES: - running tests with unity builds may fail in cases. - some build configurations/env may not compile in unity mode. E.g.: https://ci.appveyor.com/project/curlorg/curl/builds/47230972/job/51wfesgnfuauwl8q#L250 Ref: https://github.com/libssh2/libssh2/issues/1034 Ref: https://cmake.org/cmake/help/latest/prop_tgt/UNITY_BUILD.html Ref: https://en.wikipedia.org/wiki/Unity_build Closes #11095
2023-05-09 18:10:40 +08:00
struct wolfssl_ssl_backend_data *backend =
(struct wolfssl_ssl_backend_data *)connssl->backend;
char error_buffer[WOLFSSL_MAX_ERROR_SZ];
int buffsize = (blen > (size_t)INT_MAX) ? INT_MAX : (int)blen;
int nread;
DEBUGASSERT(backend);
wolfSSL_ERR_clear_error();
*curlcode = CURLE_OK;
nread = wolfSSL_read(backend->handle, buf, buffsize);
if(nread <= 0) {
int err = wolfSSL_get_error(backend->handle, nread);
switch(err) {
case SSL_ERROR_ZERO_RETURN: /* no more data */
CURL_TRC_CF(data, cf, "wolfssl_recv(len=%zu) -> CLOSED", blen);
*curlcode = CURLE_OK;
return 0;
case SSL_ERROR_NONE:
case SSL_ERROR_WANT_READ:
case SSL_ERROR_WANT_WRITE:
/* there's data pending, re-invoke wolfSSL_read() */
CURL_TRC_CF(data, cf, "wolfssl_recv(len=%zu) -> AGAIN", blen);
*curlcode = CURLE_AGAIN;
return -1;
default:
if(backend->io_result == CURLE_AGAIN) {
CURL_TRC_CF(data, cf, "wolfssl_recv(len=%zu) -> AGAIN", blen);
*curlcode = CURLE_AGAIN;
return -1;
}
failf(data, "SSL read: %s, errno %d",
wolfSSL_ERR_error_string(err, error_buffer), SOCKERRNO);
*curlcode = CURLE_RECV_ERROR;
return -1;
}
}
CURL_TRC_CF(data, cf, "wolfssl_recv(len=%zu) -> %d", blen, nread);
return nread;
}
static void wolfssl_session_free(void *ptr)
{
wolfSSL_SESSION_free(ptr);
}
static size_t wolfssl_version(char *buffer, size_t size)
{
#if LIBWOLFSSL_VERSION_HEX >= 0x03006000
return msnprintf(buffer, size, "wolfSSL/%s", wolfSSL_lib_version());
#elif defined(WOLFSSL_VERSION)
return msnprintf(buffer, size, "wolfSSL/%s", WOLFSSL_VERSION);
#endif
}
static int wolfssl_init(void)
{
int ret;
#ifdef OPENSSL_EXTRA
Curl_tls_keylog_open();
#endif
ret = (wolfSSL_Init() == SSL_SUCCESS);
wolfssl_bio_cf_init_methods();
return ret;
}
static void wolfssl_cleanup(void)
{
wolfssl_bio_cf_free_methods();
wolfSSL_Cleanup();
#ifdef OPENSSL_EXTRA
Curl_tls_keylog_close();
#endif
}
static bool wolfssl_data_pending(struct Curl_cfilter *cf,
const struct Curl_easy *data)
{
struct ssl_connect_data *ctx = cf->ctx;
cmake: add support for "unity" builds Aka "jumbo" or "amalgamation" builds. It means to compile all sources per target as a single C source. This is experimental. You can enable it by passing `-DCMAKE_UNITY_BUILD=ON` to cmake. It requires CMake 3.16 or newer. It makes builds (much) faster, allows for better optimizations and tends to promote less ambiguous code. Also add a new AppVeyor CI job and convert an existing one to use "unity" mode (one MSVC, one MinGW), and enable it for one macOS CI job. Fix related issues: - add missing include guard to `easy_lock.h`. - rename static variables and functions (and a macro) with names reused across sources, or shadowed by local variables. - add an `#undef` after use. - add a missing `#undef` before use. - move internal definitions from `ftp.h` to `ftp.c`. - `curl_memory.h` fixes to make it work when included repeatedly. - stop building/linking curlx bits twice for a static-mode curl tool. These caused doubly defined symbols in unity builds. - silence missing extern declarations compiler warning for ` _CRT_glob`. - fix extern declarations for `tool_freq` and `tool_isVistaOrGreater`. - fix colliding static symbols in debug mode: `debugtime()` and `statename`. - rename `ssl_backend_data` structure to unique names for each TLS-backend, along with the `ssl_connect_data` struct member referencing them. This required adding casts for each access. - add workaround for missing `[P]UNICODE_STRING` types in certain Windows builds when compiling `lib/ldap.c`. To support "unity" builds, we had to enable `SCHANNEL_USE_BLACKLISTS` for Schannel (a Windows `schannel.h` option) _globally_. This caused an indirect inclusion of Windows `schannel.h` from `ldap.c` via `winldap.h` to have it enabled as well. This requires `[P]UNICODE_STRING` types, which is apperantly not defined automatically (as seen with both MSVS and mingw-w64). This patch includes `<subauth.h>` to fix it. Ref: https://github.com/curl/curl/runs/13987772013 Ref: https://dev.azure.com/daniel0244/curl/_build/results?buildId=15827&view=logs&jobId=2c9f582d-e278-56b6-4354-f38a4d851906&j=2c9f582d-e278-56b6-4354-f38a4d851906&t=90509b00-34fa-5a81-35d7-5ed9569d331c - tweak unity builds to compile `lib/memdebug.c` separately in memory trace builds to avoid PP confusion. - force-disable unity for test programs. - do not compile and link libcurl sources to libtests _twice_ when libcurl is built in static mode. KNOWN ISSUES: - running tests with unity builds may fail in cases. - some build configurations/env may not compile in unity mode. E.g.: https://ci.appveyor.com/project/curlorg/curl/builds/47230972/job/51wfesgnfuauwl8q#L250 Ref: https://github.com/libssh2/libssh2/issues/1034 Ref: https://cmake.org/cmake/help/latest/prop_tgt/UNITY_BUILD.html Ref: https://en.wikipedia.org/wiki/Unity_build Closes #11095
2023-05-09 18:10:40 +08:00
struct wolfssl_ssl_backend_data *backend;
(void)data;
DEBUGASSERT(ctx && ctx->backend);
cmake: add support for "unity" builds Aka "jumbo" or "amalgamation" builds. It means to compile all sources per target as a single C source. This is experimental. You can enable it by passing `-DCMAKE_UNITY_BUILD=ON` to cmake. It requires CMake 3.16 or newer. It makes builds (much) faster, allows for better optimizations and tends to promote less ambiguous code. Also add a new AppVeyor CI job and convert an existing one to use "unity" mode (one MSVC, one MinGW), and enable it for one macOS CI job. Fix related issues: - add missing include guard to `easy_lock.h`. - rename static variables and functions (and a macro) with names reused across sources, or shadowed by local variables. - add an `#undef` after use. - add a missing `#undef` before use. - move internal definitions from `ftp.h` to `ftp.c`. - `curl_memory.h` fixes to make it work when included repeatedly. - stop building/linking curlx bits twice for a static-mode curl tool. These caused doubly defined symbols in unity builds. - silence missing extern declarations compiler warning for ` _CRT_glob`. - fix extern declarations for `tool_freq` and `tool_isVistaOrGreater`. - fix colliding static symbols in debug mode: `debugtime()` and `statename`. - rename `ssl_backend_data` structure to unique names for each TLS-backend, along with the `ssl_connect_data` struct member referencing them. This required adding casts for each access. - add workaround for missing `[P]UNICODE_STRING` types in certain Windows builds when compiling `lib/ldap.c`. To support "unity" builds, we had to enable `SCHANNEL_USE_BLACKLISTS` for Schannel (a Windows `schannel.h` option) _globally_. This caused an indirect inclusion of Windows `schannel.h` from `ldap.c` via `winldap.h` to have it enabled as well. This requires `[P]UNICODE_STRING` types, which is apperantly not defined automatically (as seen with both MSVS and mingw-w64). This patch includes `<subauth.h>` to fix it. Ref: https://github.com/curl/curl/runs/13987772013 Ref: https://dev.azure.com/daniel0244/curl/_build/results?buildId=15827&view=logs&jobId=2c9f582d-e278-56b6-4354-f38a4d851906&j=2c9f582d-e278-56b6-4354-f38a4d851906&t=90509b00-34fa-5a81-35d7-5ed9569d331c - tweak unity builds to compile `lib/memdebug.c` separately in memory trace builds to avoid PP confusion. - force-disable unity for test programs. - do not compile and link libcurl sources to libtests _twice_ when libcurl is built in static mode. KNOWN ISSUES: - running tests with unity builds may fail in cases. - some build configurations/env may not compile in unity mode. E.g.: https://ci.appveyor.com/project/curlorg/curl/builds/47230972/job/51wfesgnfuauwl8q#L250 Ref: https://github.com/libssh2/libssh2/issues/1034 Ref: https://cmake.org/cmake/help/latest/prop_tgt/UNITY_BUILD.html Ref: https://en.wikipedia.org/wiki/Unity_build Closes #11095
2023-05-09 18:10:40 +08:00
backend = (struct wolfssl_ssl_backend_data *)ctx->backend;
if(backend->handle) /* SSL is in use */
return (0 != wolfSSL_pending(backend->handle)) ? TRUE : FALSE;
else
return FALSE;
}
/*
* This function is called to shut down the SSL layer but keep the
* socket open (CCC - Clear Command Channel)
*/
static int wolfssl_shutdown(struct Curl_cfilter *cf,
struct Curl_easy *data)
{
struct ssl_connect_data *ctx = cf->ctx;
cmake: add support for "unity" builds Aka "jumbo" or "amalgamation" builds. It means to compile all sources per target as a single C source. This is experimental. You can enable it by passing `-DCMAKE_UNITY_BUILD=ON` to cmake. It requires CMake 3.16 or newer. It makes builds (much) faster, allows for better optimizations and tends to promote less ambiguous code. Also add a new AppVeyor CI job and convert an existing one to use "unity" mode (one MSVC, one MinGW), and enable it for one macOS CI job. Fix related issues: - add missing include guard to `easy_lock.h`. - rename static variables and functions (and a macro) with names reused across sources, or shadowed by local variables. - add an `#undef` after use. - add a missing `#undef` before use. - move internal definitions from `ftp.h` to `ftp.c`. - `curl_memory.h` fixes to make it work when included repeatedly. - stop building/linking curlx bits twice for a static-mode curl tool. These caused doubly defined symbols in unity builds. - silence missing extern declarations compiler warning for ` _CRT_glob`. - fix extern declarations for `tool_freq` and `tool_isVistaOrGreater`. - fix colliding static symbols in debug mode: `debugtime()` and `statename`. - rename `ssl_backend_data` structure to unique names for each TLS-backend, along with the `ssl_connect_data` struct member referencing them. This required adding casts for each access. - add workaround for missing `[P]UNICODE_STRING` types in certain Windows builds when compiling `lib/ldap.c`. To support "unity" builds, we had to enable `SCHANNEL_USE_BLACKLISTS` for Schannel (a Windows `schannel.h` option) _globally_. This caused an indirect inclusion of Windows `schannel.h` from `ldap.c` via `winldap.h` to have it enabled as well. This requires `[P]UNICODE_STRING` types, which is apperantly not defined automatically (as seen with both MSVS and mingw-w64). This patch includes `<subauth.h>` to fix it. Ref: https://github.com/curl/curl/runs/13987772013 Ref: https://dev.azure.com/daniel0244/curl/_build/results?buildId=15827&view=logs&jobId=2c9f582d-e278-56b6-4354-f38a4d851906&j=2c9f582d-e278-56b6-4354-f38a4d851906&t=90509b00-34fa-5a81-35d7-5ed9569d331c - tweak unity builds to compile `lib/memdebug.c` separately in memory trace builds to avoid PP confusion. - force-disable unity for test programs. - do not compile and link libcurl sources to libtests _twice_ when libcurl is built in static mode. KNOWN ISSUES: - running tests with unity builds may fail in cases. - some build configurations/env may not compile in unity mode. E.g.: https://ci.appveyor.com/project/curlorg/curl/builds/47230972/job/51wfesgnfuauwl8q#L250 Ref: https://github.com/libssh2/libssh2/issues/1034 Ref: https://cmake.org/cmake/help/latest/prop_tgt/UNITY_BUILD.html Ref: https://en.wikipedia.org/wiki/Unity_build Closes #11095
2023-05-09 18:10:40 +08:00
struct wolfssl_ssl_backend_data *backend;
int retval = 0;
(void)data;
DEBUGASSERT(ctx && ctx->backend);
cmake: add support for "unity" builds Aka "jumbo" or "amalgamation" builds. It means to compile all sources per target as a single C source. This is experimental. You can enable it by passing `-DCMAKE_UNITY_BUILD=ON` to cmake. It requires CMake 3.16 or newer. It makes builds (much) faster, allows for better optimizations and tends to promote less ambiguous code. Also add a new AppVeyor CI job and convert an existing one to use "unity" mode (one MSVC, one MinGW), and enable it for one macOS CI job. Fix related issues: - add missing include guard to `easy_lock.h`. - rename static variables and functions (and a macro) with names reused across sources, or shadowed by local variables. - add an `#undef` after use. - add a missing `#undef` before use. - move internal definitions from `ftp.h` to `ftp.c`. - `curl_memory.h` fixes to make it work when included repeatedly. - stop building/linking curlx bits twice for a static-mode curl tool. These caused doubly defined symbols in unity builds. - silence missing extern declarations compiler warning for ` _CRT_glob`. - fix extern declarations for `tool_freq` and `tool_isVistaOrGreater`. - fix colliding static symbols in debug mode: `debugtime()` and `statename`. - rename `ssl_backend_data` structure to unique names for each TLS-backend, along with the `ssl_connect_data` struct member referencing them. This required adding casts for each access. - add workaround for missing `[P]UNICODE_STRING` types in certain Windows builds when compiling `lib/ldap.c`. To support "unity" builds, we had to enable `SCHANNEL_USE_BLACKLISTS` for Schannel (a Windows `schannel.h` option) _globally_. This caused an indirect inclusion of Windows `schannel.h` from `ldap.c` via `winldap.h` to have it enabled as well. This requires `[P]UNICODE_STRING` types, which is apperantly not defined automatically (as seen with both MSVS and mingw-w64). This patch includes `<subauth.h>` to fix it. Ref: https://github.com/curl/curl/runs/13987772013 Ref: https://dev.azure.com/daniel0244/curl/_build/results?buildId=15827&view=logs&jobId=2c9f582d-e278-56b6-4354-f38a4d851906&j=2c9f582d-e278-56b6-4354-f38a4d851906&t=90509b00-34fa-5a81-35d7-5ed9569d331c - tweak unity builds to compile `lib/memdebug.c` separately in memory trace builds to avoid PP confusion. - force-disable unity for test programs. - do not compile and link libcurl sources to libtests _twice_ when libcurl is built in static mode. KNOWN ISSUES: - running tests with unity builds may fail in cases. - some build configurations/env may not compile in unity mode. E.g.: https://ci.appveyor.com/project/curlorg/curl/builds/47230972/job/51wfesgnfuauwl8q#L250 Ref: https://github.com/libssh2/libssh2/issues/1034 Ref: https://cmake.org/cmake/help/latest/prop_tgt/UNITY_BUILD.html Ref: https://en.wikipedia.org/wiki/Unity_build Closes #11095
2023-05-09 18:10:40 +08:00
backend = (struct wolfssl_ssl_backend_data *)ctx->backend;
if(backend->handle) {
wolfSSL_ERR_clear_error();
wolfSSL_free(backend->handle);
cmake: add support for "unity" builds Aka "jumbo" or "amalgamation" builds. It means to compile all sources per target as a single C source. This is experimental. You can enable it by passing `-DCMAKE_UNITY_BUILD=ON` to cmake. It requires CMake 3.16 or newer. It makes builds (much) faster, allows for better optimizations and tends to promote less ambiguous code. Also add a new AppVeyor CI job and convert an existing one to use "unity" mode (one MSVC, one MinGW), and enable it for one macOS CI job. Fix related issues: - add missing include guard to `easy_lock.h`. - rename static variables and functions (and a macro) with names reused across sources, or shadowed by local variables. - add an `#undef` after use. - add a missing `#undef` before use. - move internal definitions from `ftp.h` to `ftp.c`. - `curl_memory.h` fixes to make it work when included repeatedly. - stop building/linking curlx bits twice for a static-mode curl tool. These caused doubly defined symbols in unity builds. - silence missing extern declarations compiler warning for ` _CRT_glob`. - fix extern declarations for `tool_freq` and `tool_isVistaOrGreater`. - fix colliding static symbols in debug mode: `debugtime()` and `statename`. - rename `ssl_backend_data` structure to unique names for each TLS-backend, along with the `ssl_connect_data` struct member referencing them. This required adding casts for each access. - add workaround for missing `[P]UNICODE_STRING` types in certain Windows builds when compiling `lib/ldap.c`. To support "unity" builds, we had to enable `SCHANNEL_USE_BLACKLISTS` for Schannel (a Windows `schannel.h` option) _globally_. This caused an indirect inclusion of Windows `schannel.h` from `ldap.c` via `winldap.h` to have it enabled as well. This requires `[P]UNICODE_STRING` types, which is apperantly not defined automatically (as seen with both MSVS and mingw-w64). This patch includes `<subauth.h>` to fix it. Ref: https://github.com/curl/curl/runs/13987772013 Ref: https://dev.azure.com/daniel0244/curl/_build/results?buildId=15827&view=logs&jobId=2c9f582d-e278-56b6-4354-f38a4d851906&j=2c9f582d-e278-56b6-4354-f38a4d851906&t=90509b00-34fa-5a81-35d7-5ed9569d331c - tweak unity builds to compile `lib/memdebug.c` separately in memory trace builds to avoid PP confusion. - force-disable unity for test programs. - do not compile and link libcurl sources to libtests _twice_ when libcurl is built in static mode. KNOWN ISSUES: - running tests with unity builds may fail in cases. - some build configurations/env may not compile in unity mode. E.g.: https://ci.appveyor.com/project/curlorg/curl/builds/47230972/job/51wfesgnfuauwl8q#L250 Ref: https://github.com/libssh2/libssh2/issues/1034 Ref: https://cmake.org/cmake/help/latest/prop_tgt/UNITY_BUILD.html Ref: https://en.wikipedia.org/wiki/Unity_build Closes #11095
2023-05-09 18:10:40 +08:00
backend->handle = NULL;
}
return retval;
}
static CURLcode
wolfssl_connect_common(struct Curl_cfilter *cf,
struct Curl_easy *data,
bool nonblocking,
bool *done)
{
CURLcode result;
struct ssl_connect_data *connssl = cf->ctx;
connections: introduce http/3 happy eyeballs New cfilter HTTP-CONNECT for h3/h2/http1.1 eyeballing. - filter is installed when `--http3` in the tool is used (or the equivalent CURLOPT_ done in the library) - starts a QUIC/HTTP/3 connect right away. Should that not succeed after 100ms (subject to change), a parallel attempt is started for HTTP/2 and HTTP/1.1 via TCP - both attempts are subject to IPv6/IPv4 eyeballing, same as happens for other connections - tie timeout to the ip-version HAPPY_EYEBALLS_TIMEOUT - use a `soft` timeout at half the value. When the soft timeout expires, the HTTPS-CONNECT filter checks if the QUIC filter has received any data from the server. If not, it will start the HTTP/2 attempt. HTTP/3(ngtcp2) improvements. - setting call_data in all cfilter calls similar to http/2 and vtls filters for use in callback where no stream data is available. - returning CURLE_PARTIAL_FILE for prematurely terminated transfers - enabling pytest test_05 for h3 - shifting functionality to "connect" UDP sockets from ngtcp2 implementation into the udp socket cfilter. Because unconnected UDP sockets are weird. For example they error when adding to a pollset. HTTP/3(quiche) improvements. - fixed upload bug in quiche implementation, now passes 251 and pytest - error codes on stream RESET - improved debug logs - handling of DRAIN during connect - limiting pending event queue HTTP/2 cfilter improvements. - use LOG_CF macros for dynamic logging in debug build - fix CURLcode on RST streams to be CURLE_PARTIAL_FILE - enable pytest test_05 for h2 - fix upload pytests and improve parallel transfer performance. GOAWAY handling for ngtcp2/quiche - during connect, when the remote server refuses to accept new connections and closes immediately (so the local conn goes into DRAIN phase), the connection is torn down and a another attempt is made after a short grace period. This is the behaviour observed with nghttpx when we tell it to shut down gracefully. Tested in pytest test_03_02. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). - new tests/tests-httpd/scorecard.py for testing h3/h2 protocol implementation. Invoke: python3 tests/tests-httpd/scorecard.py --help for usage. Improvements on gathering connect statistics and socket access. - new CF_CTRL_CONN_REPORT_STATS cfilter control for having cfilters report connection statistics. This is triggered when the connection has completely connected. - new void Curl_pgrsTimeWas(..) method to report a timer update with a timestamp of when it happend. This allows for updating timers "later", e.g. a connect statistic after full connectivity has been reached. - in case of HTTP eyeballing, the previous changes will update statistics only from the filter chain that "won" the eyeballing. - new cfilter query CF_QUERY_SOCKET for retrieving the socket used by a filter chain. Added methods Curl_conn_cf_get_socket() and Curl_conn_get_socket() for convenient use of this query. - Change VTLS backend to query their sub-filters for the socket when checks during the handshake are made. HTTP/3 documentation on how https eyeballing works. TLS improvements - ALPN selection for SSL/SSL-PROXY filters in one vtls set of functions, replaces copy of logic in all tls backends. - standardized the infof logging of offered ALPNs - ALPN negotiated: have common function for all backends that sets alpn proprty and connection related things based on the negotiated protocol (or lack thereof). Scorecard with Caddy. - configure can be run with `--with-test-caddy=path` to specify which caddy to use for testing - tests/tests-httpd/scorecard.py now measures download speeds with caddy pytest improvements - adding Makfile to clean gen dir - adding nghttpx rundir creation on start - checking httpd version 2.4.55 for test_05 cases where it is needed. Skipping with message if too old. - catch exception when checking for caddy existance on system. Closes #10349
2023-02-02 00:13:12 +08:00
curl_socket_t sockfd = Curl_conn_cf_get_socket(cf, data);
int what;
/* check if the connection has already been established */
if(ssl_connection_complete == connssl->state) {
*done = TRUE;
return CURLE_OK;
}
if(ssl_connect_1 == connssl->connecting_state) {
/* Find out how much more time we're allowed */
const timediff_t timeout_ms = Curl_timeleft(data, NULL, TRUE);
if(timeout_ms < 0) {
/* no need to continue if time already is up */
failf(data, "SSL connection timeout");
return CURLE_OPERATION_TIMEDOUT;
}
result = wolfssl_connect_step1(cf, data);
if(result)
return result;
}
while(ssl_connect_2 == connssl->connecting_state ||
ssl_connect_2_reading == connssl->connecting_state ||
ssl_connect_2_writing == connssl->connecting_state) {
/* check allowed time left */
const timediff_t timeout_ms = Curl_timeleft(data, NULL, TRUE);
if(timeout_ms < 0) {
/* no need to continue if time already is up */
failf(data, "SSL connection timeout");
return CURLE_OPERATION_TIMEDOUT;
}
/* if ssl is expecting something, check if it's available. */
if(connssl->connecting_state == ssl_connect_2_reading
|| connssl->connecting_state == ssl_connect_2_writing) {
curl_socket_t writefd = ssl_connect_2_writing ==
connssl->connecting_state?sockfd:CURL_SOCKET_BAD;
curl_socket_t readfd = ssl_connect_2_reading ==
connssl->connecting_state?sockfd:CURL_SOCKET_BAD;
what = Curl_socket_check(readfd, CURL_SOCKET_BAD, writefd,
nonblocking?0:timeout_ms);
if(what < 0) {
/* fatal error */
failf(data, "select/poll on SSL socket, errno: %d", SOCKERRNO);
return CURLE_SSL_CONNECT_ERROR;
}
else if(0 == what) {
if(nonblocking) {
*done = FALSE;
return CURLE_OK;
}
else {
/* timeout */
failf(data, "SSL connection timeout");
return CURLE_OPERATION_TIMEDOUT;
}
}
/* socket is readable or writable */
}
/* Run transaction, and return to the caller if it failed or if
* this connection is part of a multi handle and this loop would
* execute again. This permits the owner of a multi handle to
* abort a connection attempt before step2 has completed while
* ensuring that a client using select() or epoll() will always
* have a valid fdset to wait on.
*/
result = wolfssl_connect_step2(cf, data);
if(result || (nonblocking &&
(ssl_connect_2 == connssl->connecting_state ||
ssl_connect_2_reading == connssl->connecting_state ||
ssl_connect_2_writing == connssl->connecting_state)))
return result;
} /* repeat step2 until all transactions are done. */
if(ssl_connect_3 == connssl->connecting_state) {
result = wolfssl_connect_step3(cf, data);
if(result)
return result;
}
if(ssl_connect_done == connssl->connecting_state) {
connssl->state = ssl_connection_complete;
*done = TRUE;
}
else
*done = FALSE;
/* Reset our connect state machine */
connssl->connecting_state = ssl_connect_1;
return CURLE_OK;
}
static CURLcode wolfssl_connect_nonblocking(struct Curl_cfilter *cf,
struct Curl_easy *data,
bool *done)
{
return wolfssl_connect_common(cf, data, TRUE, done);
}
static CURLcode wolfssl_connect(struct Curl_cfilter *cf,
struct Curl_easy *data)
{
CURLcode result;
bool done = FALSE;
result = wolfssl_connect_common(cf, data, FALSE, &done);
if(result)
return result;
DEBUGASSERT(done);
return CURLE_OK;
}
static CURLcode wolfssl_random(struct Curl_easy *data,
unsigned char *entropy, size_t length)
{
WC_RNG rng;
(void)data;
if(wc_InitRng(&rng))
return CURLE_FAILED_INIT;
if(length > UINT_MAX)
return CURLE_FAILED_INIT;
if(wc_RNG_GenerateBlock(&rng, entropy, (unsigned)length))
return CURLE_FAILED_INIT;
if(wc_FreeRng(&rng))
return CURLE_FAILED_INIT;
return CURLE_OK;
}
static CURLcode wolfssl_sha256sum(const unsigned char *tmp, /* input */
size_t tmplen,
unsigned char *sha256sum /* output */,
size_t unused)
2015-07-01 08:23:54 +08:00
{
wc_Sha256 SHA256pw;
2015-07-01 08:23:54 +08:00
(void)unused;
if(wc_InitSha256(&SHA256pw))
return CURLE_FAILED_INIT;
wc_Sha256Update(&SHA256pw, tmp, (word32)tmplen);
wc_Sha256Final(&SHA256pw, sha256sum);
return CURLE_OK;
2015-07-01 08:23:54 +08:00
}
static void *wolfssl_get_internals(struct ssl_connect_data *connssl,
CURLINFO info UNUSED_PARAM)
{
cmake: add support for "unity" builds Aka "jumbo" or "amalgamation" builds. It means to compile all sources per target as a single C source. This is experimental. You can enable it by passing `-DCMAKE_UNITY_BUILD=ON` to cmake. It requires CMake 3.16 or newer. It makes builds (much) faster, allows for better optimizations and tends to promote less ambiguous code. Also add a new AppVeyor CI job and convert an existing one to use "unity" mode (one MSVC, one MinGW), and enable it for one macOS CI job. Fix related issues: - add missing include guard to `easy_lock.h`. - rename static variables and functions (and a macro) with names reused across sources, or shadowed by local variables. - add an `#undef` after use. - add a missing `#undef` before use. - move internal definitions from `ftp.h` to `ftp.c`. - `curl_memory.h` fixes to make it work when included repeatedly. - stop building/linking curlx bits twice for a static-mode curl tool. These caused doubly defined symbols in unity builds. - silence missing extern declarations compiler warning for ` _CRT_glob`. - fix extern declarations for `tool_freq` and `tool_isVistaOrGreater`. - fix colliding static symbols in debug mode: `debugtime()` and `statename`. - rename `ssl_backend_data` structure to unique names for each TLS-backend, along with the `ssl_connect_data` struct member referencing them. This required adding casts for each access. - add workaround for missing `[P]UNICODE_STRING` types in certain Windows builds when compiling `lib/ldap.c`. To support "unity" builds, we had to enable `SCHANNEL_USE_BLACKLISTS` for Schannel (a Windows `schannel.h` option) _globally_. This caused an indirect inclusion of Windows `schannel.h` from `ldap.c` via `winldap.h` to have it enabled as well. This requires `[P]UNICODE_STRING` types, which is apperantly not defined automatically (as seen with both MSVS and mingw-w64). This patch includes `<subauth.h>` to fix it. Ref: https://github.com/curl/curl/runs/13987772013 Ref: https://dev.azure.com/daniel0244/curl/_build/results?buildId=15827&view=logs&jobId=2c9f582d-e278-56b6-4354-f38a4d851906&j=2c9f582d-e278-56b6-4354-f38a4d851906&t=90509b00-34fa-5a81-35d7-5ed9569d331c - tweak unity builds to compile `lib/memdebug.c` separately in memory trace builds to avoid PP confusion. - force-disable unity for test programs. - do not compile and link libcurl sources to libtests _twice_ when libcurl is built in static mode. KNOWN ISSUES: - running tests with unity builds may fail in cases. - some build configurations/env may not compile in unity mode. E.g.: https://ci.appveyor.com/project/curlorg/curl/builds/47230972/job/51wfesgnfuauwl8q#L250 Ref: https://github.com/libssh2/libssh2/issues/1034 Ref: https://cmake.org/cmake/help/latest/prop_tgt/UNITY_BUILD.html Ref: https://en.wikipedia.org/wiki/Unity_build Closes #11095
2023-05-09 18:10:40 +08:00
struct wolfssl_ssl_backend_data *backend =
(struct wolfssl_ssl_backend_data *)connssl->backend;
(void)info;
DEBUGASSERT(backend);
return backend->handle;
}
const struct Curl_ssl Curl_ssl_wolfssl = {
{ CURLSSLBACKEND_WOLFSSL, "WolfSSL" }, /* info */
#ifdef KEEP_PEER_CERT
SSLSUPP_PINNEDPUBKEY |
#endif
#ifdef USE_BIO_CHAIN
SSLSUPP_HTTPS_PROXY |
#endif
SSLSUPP_CA_PATH |
SSLSUPP_CAINFO_BLOB |
SSLSUPP_SSL_CTX,
cmake: add support for "unity" builds Aka "jumbo" or "amalgamation" builds. It means to compile all sources per target as a single C source. This is experimental. You can enable it by passing `-DCMAKE_UNITY_BUILD=ON` to cmake. It requires CMake 3.16 or newer. It makes builds (much) faster, allows for better optimizations and tends to promote less ambiguous code. Also add a new AppVeyor CI job and convert an existing one to use "unity" mode (one MSVC, one MinGW), and enable it for one macOS CI job. Fix related issues: - add missing include guard to `easy_lock.h`. - rename static variables and functions (and a macro) with names reused across sources, or shadowed by local variables. - add an `#undef` after use. - add a missing `#undef` before use. - move internal definitions from `ftp.h` to `ftp.c`. - `curl_memory.h` fixes to make it work when included repeatedly. - stop building/linking curlx bits twice for a static-mode curl tool. These caused doubly defined symbols in unity builds. - silence missing extern declarations compiler warning for ` _CRT_glob`. - fix extern declarations for `tool_freq` and `tool_isVistaOrGreater`. - fix colliding static symbols in debug mode: `debugtime()` and `statename`. - rename `ssl_backend_data` structure to unique names for each TLS-backend, along with the `ssl_connect_data` struct member referencing them. This required adding casts for each access. - add workaround for missing `[P]UNICODE_STRING` types in certain Windows builds when compiling `lib/ldap.c`. To support "unity" builds, we had to enable `SCHANNEL_USE_BLACKLISTS` for Schannel (a Windows `schannel.h` option) _globally_. This caused an indirect inclusion of Windows `schannel.h` from `ldap.c` via `winldap.h` to have it enabled as well. This requires `[P]UNICODE_STRING` types, which is apperantly not defined automatically (as seen with both MSVS and mingw-w64). This patch includes `<subauth.h>` to fix it. Ref: https://github.com/curl/curl/runs/13987772013 Ref: https://dev.azure.com/daniel0244/curl/_build/results?buildId=15827&view=logs&jobId=2c9f582d-e278-56b6-4354-f38a4d851906&j=2c9f582d-e278-56b6-4354-f38a4d851906&t=90509b00-34fa-5a81-35d7-5ed9569d331c - tweak unity builds to compile `lib/memdebug.c` separately in memory trace builds to avoid PP confusion. - force-disable unity for test programs. - do not compile and link libcurl sources to libtests _twice_ when libcurl is built in static mode. KNOWN ISSUES: - running tests with unity builds may fail in cases. - some build configurations/env may not compile in unity mode. E.g.: https://ci.appveyor.com/project/curlorg/curl/builds/47230972/job/51wfesgnfuauwl8q#L250 Ref: https://github.com/libssh2/libssh2/issues/1034 Ref: https://cmake.org/cmake/help/latest/prop_tgt/UNITY_BUILD.html Ref: https://en.wikipedia.org/wiki/Unity_build Closes #11095
2023-05-09 18:10:40 +08:00
sizeof(struct wolfssl_ssl_backend_data),
vtls: encapsulate SSL backend-specific data 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>
2017-07-29 04:09:35 +08:00
wolfssl_init, /* init */
wolfssl_cleanup, /* cleanup */
wolfssl_version, /* version */
Curl_none_check_cxn, /* check_cxn */
wolfssl_shutdown, /* shutdown */
wolfssl_data_pending, /* data_pending */
wolfssl_random, /* random */
Curl_none_cert_status_request, /* cert_status_request */
wolfssl_connect, /* connect */
wolfssl_connect_nonblocking, /* connect_nonblocking */
lib: introduce struct easy_poll_set for poll information Connection filter had a `get_select_socks()` method, inspired by the various `getsocks` functions involved during the lifetime of a transfer. These, depending on transfer state (CONNECT/DO/DONE/ etc.), return sockets to monitor and flag if this shall be done for POLLIN and/or POLLOUT. Due to this design, sockets and flags could only be added, not removed. This led to problems in filters like HTTP/2 where flow control prohibits the sending of data until the peer increases the flow window. The general transfer loop wants to write, adds POLLOUT, the socket is writeable but no data can be written. This leads to cpu busy loops. To prevent that, HTTP/2 did set the `SEND_HOLD` flag of such a blocked transfer, so the transfer loop cedes further attempts. This works if only one such filter is involved. If a HTTP/2 transfer goes through a HTTP/2 proxy, two filters are setting/clearing this flag and may step on each other's toes. Connection filters `get_select_socks()` is replaced by `adjust_pollset()`. They get passed a `struct easy_pollset` that keeps up to `MAX_SOCKSPEREASYHANDLE` sockets and their `POLLIN|POLLOUT` flags. This struct is initialized in `multi_getsock()` by calling the various `getsocks()` implementations based on transfer state, as before. After protocol handlers/transfer loop have set the sockets and flags they want, the `easy_pollset` is *always* passed to the filters. Filters "higher" in the chain are called first, starting at the first not-yet-connection one. Each filter may add sockets and/or change flags. When all flags are removed, the socket itself is removed from the pollset. Example: * transfer wants to send, adds POLLOUT * http/2 filter has a flow control block, removes POLLOUT and adds POLLIN (it is waiting on a WINDOW_UPDATE from the server) * TLS filter is connected and changes nothing * h2-proxy filter also has a flow control block on its tunnel stream, removes POLLOUT and adds POLLIN also. * socket filter is connected and changes nothing * The resulting pollset is then mixed together with all other transfers and their pollsets, just as before. Use of `SEND_HOLD` is no longer necessary in the filters. All filters are adapted for the changed method. The handling in `multi.c` has been adjusted, but its state handling the the protocol handlers' `getsocks` method are untouched. The most affected filters are http/2, ngtcp2, quiche and h2-proxy. TLS filters needed to be adjusted for the connecting handshake read/write handling. No noticeable difference in performance was detected in local scorecard runs. Closes #11833
2023-09-04 18:06:07 +08:00
Curl_ssl_adjust_pollset, /* adjust_pollset */
wolfssl_get_internals, /* get_internals */
wolfssl_close, /* close_one */
Curl_none_close_all, /* close_all */
wolfssl_session_free, /* session_free */
Curl_none_set_engine, /* set_engine */
Curl_none_set_engine_default, /* set_engine_default */
Curl_none_engines_list, /* engines_list */
Curl_none_false_start, /* false_start */
wolfssl_sha256sum, /* sha256sum */
NULL, /* associate_connection */
NULL, /* disassociate_connection */
NULL, /* free_multi_ssl_backend_data */
wolfssl_recv, /* recv decrypted data */
wolfssl_send, /* send data to encrypt */
};
#endif