curl/lib/krb5.c

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

903 lines
23 KiB
C
Raw Normal View History

/* GSSAPI/krb5 support for FTP - loosely based on old krb4.c
*
* Copyright (c) 1995, 1996, 1997, 1998, 1999 Kungliga Tekniska Högskolan
* (Royal Institute of Technology, Stockholm, Sweden).
* Copyright (C) Daniel Stenberg
* All rights reserved.
*
* SPDX-License-Identifier: BSD-3-Clause
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
*
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
*
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* 3. Neither the name of the Institute nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE INSTITUTE AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE INSTITUTE OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE. */
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"
#if defined(HAVE_GSSAPI) && !defined(CURL_DISABLE_FTP)
#ifdef HAVE_NETDB_H
#include <netdb.h>
#endif
#ifdef HAVE_ARPA_INET_H
#include <arpa/inet.h>
#endif
#include "urldata.h"
cf-socket: improvements in socket I/O handling - Curl_write_plain/Curl_read_plain have been eliminated. Last code use now uses Curl_conn_send/recv so that requests use conn->send/revc callbacks which defaults to cfilters use. - Curl_recv_plain/Curl_send_plain have been internalized in cf-socket.c. - USE_RECV_BEFORE_SEND_WORKAROUND (active on Windows) has been moved into cf-socket.c. The pre_recv buffer is held at the socket filter context. `postponed_data` structures have been removed from `connectdata`. - the hanger in HTTP/2 request handling was a result of read buffering on all sends and the multi handling is not prepared for this. The following happens: - multi preforms on a HTTP/2 easy handle - h2 reads and processes data - this leads to a send of h2 data - which receives and buffers before the send - h2 returns - multi selects on the socket, but no data arrives (its in the buffer already) the workaround now receives data in a loop as long as there is something in the buffer. The real fix would be for multi to change, so that `data_pending` is evaluated before deciding to wait on the socket. io_buffer, optional, in cf-socket.c, http/2 sets state.drain if lower filter have pending data. This io_buffer is only available/used when the -DUSE_RECV_BEFORE_SEND_WORKAROUND is active, e.g. on Windows configurations. It also maintains the original checks on protocol handler being HTTP and conn->send/recv not being replaced. The HTTP/2 (nghttp2) cfilter now sets data->state.drain when it finds out that the "lower" filter chain has still pending data at the end of its IO operation. This prevents the processing from becoming stalled. Closes #10280
2023-01-30 23:03:00 +08:00
#include "cfilters.h"
#include "cf-socket.h"
#include "curl_base64.h"
#include "ftp.h"
#include "curl_gssapi.h"
#include "sendf.h"
#include "curl_krb5.h"
#include "warnless.h"
#include "strcase.h"
#include "strdup.h"
/* The last 3 #include files should be in this order */
#include "curl_printf.h"
#include "curl_memory.h"
#include "memdebug.h"
static CURLcode ftpsend(struct Curl_easy *data, struct connectdata *conn,
const char *cmd)
{
ssize_t bytes_written;
#define SBUF_SIZE 1024
char s[SBUF_SIZE];
size_t write_len;
char *sptr = s;
CURLcode result = CURLE_OK;
#ifdef HAVE_GSSAPI
unsigned char data_sec = conn->data_prot;
#endif
if(!cmd)
return CURLE_BAD_FUNCTION_ARGUMENT;
write_len = strlen(cmd);
if(!write_len || write_len > (sizeof(s) -3))
return CURLE_BAD_FUNCTION_ARGUMENT;
memcpy(&s, cmd, write_len);
strcpy(&s[write_len], "\r\n"); /* append a trailing CRLF */
write_len += 2;
bytes_written = 0;
for(;;) {
#ifdef HAVE_GSSAPI
conn->data_prot = PROT_CMD;
#endif
result = Curl_nwrite(data, FIRSTSOCKET, sptr, write_len,
&bytes_written);
#ifdef HAVE_GSSAPI
DEBUGASSERT(data_sec > PROT_NONE && data_sec < PROT_LAST);
conn->data_prot = data_sec;
#endif
if(result)
break;
Curl_debug(data, CURLINFO_HEADER_OUT, sptr, (size_t)bytes_written);
if(bytes_written != (ssize_t)write_len) {
write_len -= bytes_written;
sptr += bytes_written;
}
else
break;
}
return result;
}
static int
krb5_init(void *app_data)
{
gss_ctx_id_t *context = app_data;
/* Make sure our context is initialized for krb5_end. */
*context = GSS_C_NO_CONTEXT;
return 0;
}
static int
krb5_check_prot(void *app_data, int level)
{
(void)app_data; /* unused */
if(level == PROT_CONFIDENTIAL)
return -1;
return 0;
}
static int
krb5_decode(void *app_data, void *buf, int len,
int level UNUSED_PARAM,
struct connectdata *conn UNUSED_PARAM)
{
gss_ctx_id_t *context = app_data;
OM_uint32 maj, min;
gss_buffer_desc enc, dec;
(void)level;
(void)conn;
enc.value = buf;
enc.length = len;
maj = gss_unwrap(&min, *context, &enc, &dec, NULL, NULL);
if(maj != GSS_S_COMPLETE)
return -1;
memcpy(buf, dec.value, dec.length);
len = curlx_uztosi(dec.length);
gss_release_buffer(&min, &dec);
return len;
}
static int
krb5_encode(void *app_data, const void *from, int length, int level, void **to)
{
gss_ctx_id_t *context = app_data;
gss_buffer_desc dec, enc;
OM_uint32 maj, min;
int state;
int len;
/* NOTE that the cast is safe, neither of the krb5, gnu gss and heimdal
* libraries modify the input buffer in gss_wrap()
*/
dec.value = (void *)from;
dec.length = length;
maj = gss_wrap(&min, *context,
level == PROT_PRIVATE,
GSS_C_QOP_DEFAULT,
&dec, &state, &enc);
if(maj != GSS_S_COMPLETE)
return -1;
/* malloc a new buffer, in case gss_release_buffer doesn't work as
expected */
*to = malloc(enc.length);
if(!*to)
return -1;
memcpy(*to, enc.value, enc.length);
len = curlx_uztosi(enc.length);
gss_release_buffer(&min, &enc);
return len;
}
static int
krb5_auth(void *app_data, struct Curl_easy *data, struct connectdata *conn)
{
int ret = AUTH_OK;
char *p;
const char *host = conn->host.name;
ssize_t nread;
curl_socklen_t l = sizeof(conn->local_addr);
CURLcode result;
const char *service = data->set.str[STRING_SERVICE_NAME] ?
data->set.str[STRING_SERVICE_NAME] :
"ftp";
const char *srv_host = "host";
gss_buffer_desc input_buffer, output_buffer, _gssresp, *gssresp;
OM_uint32 maj, min;
gss_name_t gssname;
gss_ctx_id_t *context = app_data;
struct gss_channel_bindings_struct chan;
size_t base64_sz = 0;
struct sockaddr_in *remote_addr =
(struct sockaddr_in *)(void *)&conn->remote_addr->sa_addr;
char *stringp;
if(getsockname(conn->sock[FIRSTSOCKET],
(struct sockaddr *)&conn->local_addr, &l) < 0)
perror("getsockname()");
chan.initiator_addrtype = GSS_C_AF_INET;
chan.initiator_address.length = l - 4;
chan.initiator_address.value = &conn->local_addr.sin_addr.s_addr;
chan.acceptor_addrtype = GSS_C_AF_INET;
chan.acceptor_address.length = l - 4;
chan.acceptor_address.value = &remote_addr->sin_addr.s_addr;
chan.application_data.length = 0;
chan.application_data.value = NULL;
/* this loop will execute twice (once for service, once for host) */
2011-09-03 22:06:10 +08:00
for(;;) {
/* this really shouldn't be repeated here, but can't help it */
if(service == srv_host) {
result = ftpsend(data, conn, "AUTH GSSAPI");
if(result)
return -2;
if(Curl_GetFTPResponse(data, &nread, NULL))
return -1;
if(data->state.buffer[0] != '3')
return -1;
}
stringp = aprintf("%s@%s", service, host);
if(!stringp)
return -2;
input_buffer.value = stringp;
input_buffer.length = strlen(stringp);
maj = gss_import_name(&min, &input_buffer, GSS_C_NT_HOSTBASED_SERVICE,
&gssname);
free(stringp);
if(maj != GSS_S_COMPLETE) {
gss_release_name(&min, &gssname);
if(service == srv_host) {
failf(data, "Error importing service name %s@%s", service, host);
return AUTH_ERROR;
}
service = srv_host;
continue;
}
/* We pass NULL as |output_name_type| to avoid a leak. */
gss_display_name(&min, gssname, &output_buffer, NULL);
2023-07-18 02:10:38 +08:00
infof(data, "Trying against %s", (char *)output_buffer.value);
gssresp = GSS_C_NO_BUFFER;
*context = GSS_C_NO_CONTEXT;
do {
/* Release the buffer at each iteration to avoid leaking: the first time
we are releasing the memory from gss_display_name. The last item is
taken care by a final gss_release_buffer. */
gss_release_buffer(&min, &output_buffer);
ret = AUTH_OK;
maj = Curl_gss_init_sec_context(data,
&min,
context,
gssname,
&Curl_krb5_mech_oid,
&chan,
gssresp,
&output_buffer,
TRUE,
NULL);
if(gssresp) {
free(_gssresp.value);
gssresp = NULL;
}
if(GSS_ERROR(maj)) {
infof(data, "Error creating security context");
ret = AUTH_ERROR;
break;
}
if(output_buffer.length) {
char *cmd;
result = Curl_base64_encode((char *)output_buffer.value,
2011-08-24 17:56:23 +08:00
output_buffer.length, &p, &base64_sz);
if(result) {
infof(data, "base64-encoding: %s", curl_easy_strerror(result));
ret = AUTH_ERROR;
break;
}
cmd = aprintf("ADAT %s", p);
if(cmd)
result = ftpsend(data, conn, cmd);
else
result = CURLE_OUT_OF_MEMORY;
free(p);
free(cmd);
if(result) {
ret = -2;
break;
}
if(Curl_GetFTPResponse(data, &nread, NULL)) {
ret = -1;
break;
}
if(data->state.buffer[0] != '2' && data->state.buffer[0] != '3') {
infof(data, "Server didn't accept auth data");
ret = AUTH_ERROR;
break;
}
_gssresp.value = NULL; /* make sure it is initialized */
p = data->state.buffer + 4;
p = strstr(p, "ADAT=");
if(p) {
result = Curl_base64_decode(p + 5,
(unsigned char **)&_gssresp.value,
&_gssresp.length);
if(result) {
failf(data, "base64-decoding: %s", curl_easy_strerror(result));
ret = AUTH_CONTINUE;
break;
}
}
gssresp = &_gssresp;
}
} while(maj == GSS_S_CONTINUE_NEEDED);
gss_release_name(&min, &gssname);
gss_release_buffer(&min, &output_buffer);
if(gssresp)
free(_gssresp.value);
if(ret == AUTH_OK || service == srv_host)
return ret;
service = srv_host;
}
return ret;
}
static void krb5_end(void *app_data)
{
OM_uint32 min;
gss_ctx_id_t *context = app_data;
if(*context != GSS_C_NO_CONTEXT) {
OM_uint32 maj = gss_delete_sec_context(&min, context, GSS_C_NO_BUFFER);
(void)maj;
DEBUGASSERT(maj == GSS_S_COMPLETE);
}
}
static const struct Curl_sec_client_mech Curl_krb5_client_mech = {
"GSSAPI",
sizeof(gss_ctx_id_t),
krb5_init,
krb5_auth,
krb5_end,
krb5_check_prot,
krb5_encode,
krb5_decode
};
static const struct {
unsigned char level;
const char *name;
} level_names[] = {
{ PROT_CLEAR, "clear" },
{ PROT_SAFE, "safe" },
{ PROT_CONFIDENTIAL, "confidential" },
{ PROT_PRIVATE, "private" }
};
static unsigned char name_to_level(const char *name)
{
int i;
for(i = 0; i < (int)sizeof(level_names)/(int)sizeof(level_names[0]); i++)
if(curl_strequal(name, level_names[i].name))
return level_names[i].level;
return PROT_NONE;
}
/* Convert a protocol |level| to its char representation.
We take an int to catch programming mistakes. */
static char level_to_char(int level)
{
switch(level) {
case PROT_CLEAR:
return 'C';
case PROT_SAFE:
return 'S';
case PROT_CONFIDENTIAL:
return 'E';
case PROT_PRIVATE:
return 'P';
case PROT_CMD:
/* Fall through */
default:
/* Those 2 cases should not be reached! */
break;
}
DEBUGASSERT(0);
/* Default to the most secure alternative. */
return 'P';
}
/* Send an FTP command defined by |message| and the optional arguments. The
function returns the ftp_code. If an error occurs, -1 is returned. */
static int ftp_send_command(struct Curl_easy *data, const char *message, ...)
{
int ftp_code;
ssize_t nread = 0;
va_list args;
char print_buffer[50];
va_start(args, message);
mvsnprintf(print_buffer, sizeof(print_buffer), message, args);
va_end(args);
if(ftpsend(data, data->conn, print_buffer)) {
ftp_code = -1;
}
else {
if(Curl_GetFTPResponse(data, &nread, &ftp_code))
ftp_code = -1;
}
(void)nread; /* Unused */
return ftp_code;
}
/* Read |len| from the socket |fd| and store it in |to|. Return a CURLcode
saying whether an error occurred or CURLE_OK if |len| was read. */
static CURLcode
cf-socket: improvements in socket I/O handling - Curl_write_plain/Curl_read_plain have been eliminated. Last code use now uses Curl_conn_send/recv so that requests use conn->send/revc callbacks which defaults to cfilters use. - Curl_recv_plain/Curl_send_plain have been internalized in cf-socket.c. - USE_RECV_BEFORE_SEND_WORKAROUND (active on Windows) has been moved into cf-socket.c. The pre_recv buffer is held at the socket filter context. `postponed_data` structures have been removed from `connectdata`. - the hanger in HTTP/2 request handling was a result of read buffering on all sends and the multi handling is not prepared for this. The following happens: - multi preforms on a HTTP/2 easy handle - h2 reads and processes data - this leads to a send of h2 data - which receives and buffers before the send - h2 returns - multi selects on the socket, but no data arrives (its in the buffer already) the workaround now receives data in a loop as long as there is something in the buffer. The real fix would be for multi to change, so that `data_pending` is evaluated before deciding to wait on the socket. io_buffer, optional, in cf-socket.c, http/2 sets state.drain if lower filter have pending data. This io_buffer is only available/used when the -DUSE_RECV_BEFORE_SEND_WORKAROUND is active, e.g. on Windows configurations. It also maintains the original checks on protocol handler being HTTP and conn->send/recv not being replaced. The HTTP/2 (nghttp2) cfilter now sets data->state.drain when it finds out that the "lower" filter chain has still pending data at the end of its IO operation. This prevents the processing from becoming stalled. Closes #10280
2023-01-30 23:03:00 +08:00
socket_read(struct Curl_easy *data, int sockindex, void *to, size_t len)
{
char *to_p = to;
CURLcode result;
ssize_t nread = 0;
while(len > 0) {
cf-socket: improvements in socket I/O handling - Curl_write_plain/Curl_read_plain have been eliminated. Last code use now uses Curl_conn_send/recv so that requests use conn->send/revc callbacks which defaults to cfilters use. - Curl_recv_plain/Curl_send_plain have been internalized in cf-socket.c. - USE_RECV_BEFORE_SEND_WORKAROUND (active on Windows) has been moved into cf-socket.c. The pre_recv buffer is held at the socket filter context. `postponed_data` structures have been removed from `connectdata`. - the hanger in HTTP/2 request handling was a result of read buffering on all sends and the multi handling is not prepared for this. The following happens: - multi preforms on a HTTP/2 easy handle - h2 reads and processes data - this leads to a send of h2 data - which receives and buffers before the send - h2 returns - multi selects on the socket, but no data arrives (its in the buffer already) the workaround now receives data in a loop as long as there is something in the buffer. The real fix would be for multi to change, so that `data_pending` is evaluated before deciding to wait on the socket. io_buffer, optional, in cf-socket.c, http/2 sets state.drain if lower filter have pending data. This io_buffer is only available/used when the -DUSE_RECV_BEFORE_SEND_WORKAROUND is active, e.g. on Windows configurations. It also maintains the original checks on protocol handler being HTTP and conn->send/recv not being replaced. The HTTP/2 (nghttp2) cfilter now sets data->state.drain when it finds out that the "lower" filter chain has still pending data at the end of its IO operation. This prevents the processing from becoming stalled. Closes #10280
2023-01-30 23:03:00 +08:00
nread = Curl_conn_recv(data, sockindex, to_p, len, &result);
if(nread > 0) {
len -= nread;
to_p += nread;
}
else {
if(result == CURLE_AGAIN)
continue;
return result;
}
}
return CURLE_OK;
}
/* Write |len| bytes from the buffer |to| to the socket |fd|. Return a
CURLcode saying whether an error occurred or CURLE_OK if |len| was
written. */
static CURLcode
cf-socket: improvements in socket I/O handling - Curl_write_plain/Curl_read_plain have been eliminated. Last code use now uses Curl_conn_send/recv so that requests use conn->send/revc callbacks which defaults to cfilters use. - Curl_recv_plain/Curl_send_plain have been internalized in cf-socket.c. - USE_RECV_BEFORE_SEND_WORKAROUND (active on Windows) has been moved into cf-socket.c. The pre_recv buffer is held at the socket filter context. `postponed_data` structures have been removed from `connectdata`. - the hanger in HTTP/2 request handling was a result of read buffering on all sends and the multi handling is not prepared for this. The following happens: - multi preforms on a HTTP/2 easy handle - h2 reads and processes data - this leads to a send of h2 data - which receives and buffers before the send - h2 returns - multi selects on the socket, but no data arrives (its in the buffer already) the workaround now receives data in a loop as long as there is something in the buffer. The real fix would be for multi to change, so that `data_pending` is evaluated before deciding to wait on the socket. io_buffer, optional, in cf-socket.c, http/2 sets state.drain if lower filter have pending data. This io_buffer is only available/used when the -DUSE_RECV_BEFORE_SEND_WORKAROUND is active, e.g. on Windows configurations. It also maintains the original checks on protocol handler being HTTP and conn->send/recv not being replaced. The HTTP/2 (nghttp2) cfilter now sets data->state.drain when it finds out that the "lower" filter chain has still pending data at the end of its IO operation. This prevents the processing from becoming stalled. Closes #10280
2023-01-30 23:03:00 +08:00
socket_write(struct Curl_easy *data, int sockindex, const void *to,
size_t len)
{
const char *to_p = to;
CURLcode result;
ssize_t written;
while(len > 0) {
cf-socket: improvements in socket I/O handling - Curl_write_plain/Curl_read_plain have been eliminated. Last code use now uses Curl_conn_send/recv so that requests use conn->send/revc callbacks which defaults to cfilters use. - Curl_recv_plain/Curl_send_plain have been internalized in cf-socket.c. - USE_RECV_BEFORE_SEND_WORKAROUND (active on Windows) has been moved into cf-socket.c. The pre_recv buffer is held at the socket filter context. `postponed_data` structures have been removed from `connectdata`. - the hanger in HTTP/2 request handling was a result of read buffering on all sends and the multi handling is not prepared for this. The following happens: - multi preforms on a HTTP/2 easy handle - h2 reads and processes data - this leads to a send of h2 data - which receives and buffers before the send - h2 returns - multi selects on the socket, but no data arrives (its in the buffer already) the workaround now receives data in a loop as long as there is something in the buffer. The real fix would be for multi to change, so that `data_pending` is evaluated before deciding to wait on the socket. io_buffer, optional, in cf-socket.c, http/2 sets state.drain if lower filter have pending data. This io_buffer is only available/used when the -DUSE_RECV_BEFORE_SEND_WORKAROUND is active, e.g. on Windows configurations. It also maintains the original checks on protocol handler being HTTP and conn->send/recv not being replaced. The HTTP/2 (nghttp2) cfilter now sets data->state.drain when it finds out that the "lower" filter chain has still pending data at the end of its IO operation. This prevents the processing from becoming stalled. Closes #10280
2023-01-30 23:03:00 +08:00
written = Curl_conn_send(data, sockindex, to_p, len, &result);
if(written > 0) {
len -= written;
to_p += written;
}
else {
if(result == CURLE_AGAIN)
continue;
return result;
}
}
return CURLE_OK;
}
cf-socket: improvements in socket I/O handling - Curl_write_plain/Curl_read_plain have been eliminated. Last code use now uses Curl_conn_send/recv so that requests use conn->send/revc callbacks which defaults to cfilters use. - Curl_recv_plain/Curl_send_plain have been internalized in cf-socket.c. - USE_RECV_BEFORE_SEND_WORKAROUND (active on Windows) has been moved into cf-socket.c. The pre_recv buffer is held at the socket filter context. `postponed_data` structures have been removed from `connectdata`. - the hanger in HTTP/2 request handling was a result of read buffering on all sends and the multi handling is not prepared for this. The following happens: - multi preforms on a HTTP/2 easy handle - h2 reads and processes data - this leads to a send of h2 data - which receives and buffers before the send - h2 returns - multi selects on the socket, but no data arrives (its in the buffer already) the workaround now receives data in a loop as long as there is something in the buffer. The real fix would be for multi to change, so that `data_pending` is evaluated before deciding to wait on the socket. io_buffer, optional, in cf-socket.c, http/2 sets state.drain if lower filter have pending data. This io_buffer is only available/used when the -DUSE_RECV_BEFORE_SEND_WORKAROUND is active, e.g. on Windows configurations. It also maintains the original checks on protocol handler being HTTP and conn->send/recv not being replaced. The HTTP/2 (nghttp2) cfilter now sets data->state.drain when it finds out that the "lower" filter chain has still pending data at the end of its IO operation. This prevents the processing from becoming stalled. Closes #10280
2023-01-30 23:03:00 +08:00
static CURLcode read_data(struct Curl_easy *data, int sockindex,
struct krb5buffer *buf)
{
sendf: change Curl_read_plain to wrap Curl_recv_plain (take 2) Prior to this change Curl_read_plain would attempt to read the socket directly. On Windows that's a problem because recv data may be cached by libcurl and that data is only drained using Curl_recv_plain. Rather than rewrite Curl_read_plain to handle cached recv data, I changed it to wrap Curl_recv_plain, in much the same way that Curl_write_plain already wraps Curl_send_plain. Curl_read_plain -> Curl_recv_plain Curl_write_plain -> Curl_send_plain This fixes a bug in the schannel backend where decryption of arbitrary TLS records fails because cached recv data is never drained. We send data (TLS records formed by Schannel) using Curl_write_plain, which calls Curl_send_plain, and that may do a recv-before-send ("pre-receive") to cache received data. The code calls Curl_read_plain to read data (TLS records from the server), which prior to this change did not call Curl_recv_plain and therefore cached recv data wasn't retrieved, resulting in malformed TLS records and decryption failure (SEC_E_DECRYPT_FAILURE). The bug has only been observed during Schannel TLS 1.3 handshakes. Refer to the issue and PR for more information. -- This is take 2 of the original fix. It preserves the original behavior of Curl_read_plain to write 0 to the bytes read parameter on error, since apparently some callers expect that (SOCKS tests were hanging). The original fix which landed in 12e1def5 and was later reverted in 18383fbf failed to work properly because it did not do that. Also, it changes Curl_write_plain the same way to complement Curl_read_plain, and it changes Curl_send_plain to return -1 instead of 0 on CURLE_AGAIN to complement Curl_recv_plain. Behavior on error with these changes: Curl_recv_plain returns -1 and *code receives error code. Curl_send_plain returns -1 and *code receives error code. Curl_read_plain returns error code and *n (bytes read) receives 0. Curl_write_plain returns error code and *written receives 0. -- Ref: https://github.com/curl/curl/issues/9431#issuecomment-1312420361 Assisted-by: Joel Depooter Reported-by: Egor Pugin Fixes https://github.com/curl/curl/issues/9431 Closes https://github.com/curl/curl/pull/9949
2022-11-14 16:30:30 +08:00
struct connectdata *conn = data->conn;
int len;
CURLcode result;
int nread;
cf-socket: improvements in socket I/O handling - Curl_write_plain/Curl_read_plain have been eliminated. Last code use now uses Curl_conn_send/recv so that requests use conn->send/revc callbacks which defaults to cfilters use. - Curl_recv_plain/Curl_send_plain have been internalized in cf-socket.c. - USE_RECV_BEFORE_SEND_WORKAROUND (active on Windows) has been moved into cf-socket.c. The pre_recv buffer is held at the socket filter context. `postponed_data` structures have been removed from `connectdata`. - the hanger in HTTP/2 request handling was a result of read buffering on all sends and the multi handling is not prepared for this. The following happens: - multi preforms on a HTTP/2 easy handle - h2 reads and processes data - this leads to a send of h2 data - which receives and buffers before the send - h2 returns - multi selects on the socket, but no data arrives (its in the buffer already) the workaround now receives data in a loop as long as there is something in the buffer. The real fix would be for multi to change, so that `data_pending` is evaluated before deciding to wait on the socket. io_buffer, optional, in cf-socket.c, http/2 sets state.drain if lower filter have pending data. This io_buffer is only available/used when the -DUSE_RECV_BEFORE_SEND_WORKAROUND is active, e.g. on Windows configurations. It also maintains the original checks on protocol handler being HTTP and conn->send/recv not being replaced. The HTTP/2 (nghttp2) cfilter now sets data->state.drain when it finds out that the "lower" filter chain has still pending data at the end of its IO operation. This prevents the processing from becoming stalled. Closes #10280
2023-01-30 23:03:00 +08:00
result = socket_read(data, sockindex, &len, sizeof(len));
if(result)
return result;
if(len) {
/* only realloc if there was a length */
len = ntohl(len);
if(len > CURL_MAX_INPUT_LENGTH)
len = 0;
else
buf->data = Curl_saferealloc(buf->data, len);
}
if(!len || !buf->data)
return CURLE_OUT_OF_MEMORY;
cf-socket: improvements in socket I/O handling - Curl_write_plain/Curl_read_plain have been eliminated. Last code use now uses Curl_conn_send/recv so that requests use conn->send/revc callbacks which defaults to cfilters use. - Curl_recv_plain/Curl_send_plain have been internalized in cf-socket.c. - USE_RECV_BEFORE_SEND_WORKAROUND (active on Windows) has been moved into cf-socket.c. The pre_recv buffer is held at the socket filter context. `postponed_data` structures have been removed from `connectdata`. - the hanger in HTTP/2 request handling was a result of read buffering on all sends and the multi handling is not prepared for this. The following happens: - multi preforms on a HTTP/2 easy handle - h2 reads and processes data - this leads to a send of h2 data - which receives and buffers before the send - h2 returns - multi selects on the socket, but no data arrives (its in the buffer already) the workaround now receives data in a loop as long as there is something in the buffer. The real fix would be for multi to change, so that `data_pending` is evaluated before deciding to wait on the socket. io_buffer, optional, in cf-socket.c, http/2 sets state.drain if lower filter have pending data. This io_buffer is only available/used when the -DUSE_RECV_BEFORE_SEND_WORKAROUND is active, e.g. on Windows configurations. It also maintains the original checks on protocol handler being HTTP and conn->send/recv not being replaced. The HTTP/2 (nghttp2) cfilter now sets data->state.drain when it finds out that the "lower" filter chain has still pending data at the end of its IO operation. This prevents the processing from becoming stalled. Closes #10280
2023-01-30 23:03:00 +08:00
result = socket_read(data, sockindex, buf->data, len);
if(result)
return result;
nread = conn->mech->decode(conn->app_data, buf->data, len,
conn->data_prot, conn);
if(nread < 0)
return CURLE_RECV_ERROR;
buf->size = (size_t)nread;
buf->index = 0;
return CURLE_OK;
}
static size_t
buffer_read(struct krb5buffer *buf, void *data, size_t len)
{
if(buf->size - buf->index < len)
len = buf->size - buf->index;
memcpy(data, (char *)buf->data + buf->index, len);
buf->index += len;
return len;
}
/* Matches Curl_recv signature */
static ssize_t sec_recv(struct Curl_easy *data, int sockindex,
char *buffer, size_t len, CURLcode *err)
{
size_t bytes_read;
size_t total_read = 0;
struct connectdata *conn = data->conn;
*err = CURLE_OK;
/* Handle clear text response. */
if(conn->sec_complete == 0 || conn->data_prot == PROT_CLEAR)
cf-socket: improvements in socket I/O handling - Curl_write_plain/Curl_read_plain have been eliminated. Last code use now uses Curl_conn_send/recv so that requests use conn->send/revc callbacks which defaults to cfilters use. - Curl_recv_plain/Curl_send_plain have been internalized in cf-socket.c. - USE_RECV_BEFORE_SEND_WORKAROUND (active on Windows) has been moved into cf-socket.c. The pre_recv buffer is held at the socket filter context. `postponed_data` structures have been removed from `connectdata`. - the hanger in HTTP/2 request handling was a result of read buffering on all sends and the multi handling is not prepared for this. The following happens: - multi preforms on a HTTP/2 easy handle - h2 reads and processes data - this leads to a send of h2 data - which receives and buffers before the send - h2 returns - multi selects on the socket, but no data arrives (its in the buffer already) the workaround now receives data in a loop as long as there is something in the buffer. The real fix would be for multi to change, so that `data_pending` is evaluated before deciding to wait on the socket. io_buffer, optional, in cf-socket.c, http/2 sets state.drain if lower filter have pending data. This io_buffer is only available/used when the -DUSE_RECV_BEFORE_SEND_WORKAROUND is active, e.g. on Windows configurations. It also maintains the original checks on protocol handler being HTTP and conn->send/recv not being replaced. The HTTP/2 (nghttp2) cfilter now sets data->state.drain when it finds out that the "lower" filter chain has still pending data at the end of its IO operation. This prevents the processing from becoming stalled. Closes #10280
2023-01-30 23:03:00 +08:00
return Curl_conn_recv(data, sockindex, buffer, len, err);
if(conn->in_buffer.eof_flag) {
conn->in_buffer.eof_flag = 0;
return 0;
}
bytes_read = buffer_read(&conn->in_buffer, buffer, len);
len -= bytes_read;
total_read += bytes_read;
buffer += bytes_read;
while(len > 0) {
cf-socket: improvements in socket I/O handling - Curl_write_plain/Curl_read_plain have been eliminated. Last code use now uses Curl_conn_send/recv so that requests use conn->send/revc callbacks which defaults to cfilters use. - Curl_recv_plain/Curl_send_plain have been internalized in cf-socket.c. - USE_RECV_BEFORE_SEND_WORKAROUND (active on Windows) has been moved into cf-socket.c. The pre_recv buffer is held at the socket filter context. `postponed_data` structures have been removed from `connectdata`. - the hanger in HTTP/2 request handling was a result of read buffering on all sends and the multi handling is not prepared for this. The following happens: - multi preforms on a HTTP/2 easy handle - h2 reads and processes data - this leads to a send of h2 data - which receives and buffers before the send - h2 returns - multi selects on the socket, but no data arrives (its in the buffer already) the workaround now receives data in a loop as long as there is something in the buffer. The real fix would be for multi to change, so that `data_pending` is evaluated before deciding to wait on the socket. io_buffer, optional, in cf-socket.c, http/2 sets state.drain if lower filter have pending data. This io_buffer is only available/used when the -DUSE_RECV_BEFORE_SEND_WORKAROUND is active, e.g. on Windows configurations. It also maintains the original checks on protocol handler being HTTP and conn->send/recv not being replaced. The HTTP/2 (nghttp2) cfilter now sets data->state.drain when it finds out that the "lower" filter chain has still pending data at the end of its IO operation. This prevents the processing from becoming stalled. Closes #10280
2023-01-30 23:03:00 +08:00
if(read_data(data, sockindex, &conn->in_buffer))
return -1;
if(conn->in_buffer.size == 0) {
if(bytes_read > 0)
conn->in_buffer.eof_flag = 1;
return bytes_read;
}
bytes_read = buffer_read(&conn->in_buffer, buffer, len);
len -= bytes_read;
total_read += bytes_read;
buffer += bytes_read;
}
return total_read;
}
/* Send |length| bytes from |from| to the |fd| socket taking care of encoding
and negotiating with the server. |from| can be NULL. */
static void do_sec_send(struct Curl_easy *data, struct connectdata *conn,
curl_socket_t fd, const char *from, int length)
{
int bytes, htonl_bytes; /* 32-bit integers for htonl */
char *buffer = NULL;
char *cmd_buffer;
size_t cmd_size = 0;
CURLcode error;
enum protection_level prot_level = conn->data_prot;
bool iscmd = (prot_level == PROT_CMD)?TRUE:FALSE;
DEBUGASSERT(prot_level > PROT_NONE && prot_level < PROT_LAST);
if(iscmd) {
if(!strncmp(from, "PASS ", 5) || !strncmp(from, "ACCT ", 5))
prot_level = PROT_PRIVATE;
else
prot_level = conn->command_prot;
}
bytes = conn->mech->encode(conn->app_data, from, length, prot_level,
(void **)&buffer);
if(!buffer || bytes <= 0)
return; /* error */
if(iscmd) {
error = Curl_base64_encode(buffer, curlx_sitouz(bytes),
&cmd_buffer, &cmd_size);
if(error) {
free(buffer);
return; /* error */
}
if(cmd_size > 0) {
static const char *enc = "ENC ";
static const char *mic = "MIC ";
if(prot_level == PROT_PRIVATE)
socket_write(data, fd, enc, 4);
else
socket_write(data, fd, mic, 4);
socket_write(data, fd, cmd_buffer, cmd_size);
socket_write(data, fd, "\r\n", 2);
infof(data, "Send: %s%s", prot_level == PROT_PRIVATE?enc:mic,
cmd_buffer);
free(cmd_buffer);
}
}
else {
htonl_bytes = htonl(bytes);
socket_write(data, fd, &htonl_bytes, sizeof(htonl_bytes));
socket_write(data, fd, buffer, curlx_sitouz(bytes));
}
free(buffer);
}
static ssize_t sec_write(struct Curl_easy *data, struct connectdata *conn,
curl_socket_t fd, const char *buffer, size_t length)
{
ssize_t tx = 0, len = conn->buffer_size;
if(len <= 0)
len = length;
while(length) {
if(length < (size_t)len)
len = length;
do_sec_send(data, conn, fd, buffer, curlx_sztosi(len));
length -= len;
buffer += len;
tx += len;
}
return tx;
}
/* Matches Curl_send signature */
static ssize_t sec_send(struct Curl_easy *data, int sockindex,
const void *buffer, size_t len, CURLcode *err)
{
struct connectdata *conn = data->conn;
curl_socket_t fd = conn->sock[sockindex];
*err = CURLE_OK;
return sec_write(data, conn, fd, buffer, len);
}
int Curl_sec_read_msg(struct Curl_easy *data, struct connectdata *conn,
char *buffer, enum protection_level level)
{
/* decoded_len should be size_t or ssize_t but conn->mech->decode returns an
int */
int decoded_len;
char *buf;
int ret_code = 0;
size_t decoded_sz = 0;
CURLcode error;
(void) data;
if(!conn->mech)
/* not initialized, return error */
return -1;
DEBUGASSERT(level > PROT_NONE && level < PROT_LAST);
error = Curl_base64_decode(buffer + 4, (unsigned char **)&buf, &decoded_sz);
if(error || decoded_sz == 0)
return -1;
if(decoded_sz > (size_t)INT_MAX) {
free(buf);
return -1;
}
decoded_len = curlx_uztosi(decoded_sz);
decoded_len = conn->mech->decode(conn->app_data, buf, decoded_len,
level, conn);
if(decoded_len <= 0) {
free(buf);
return -1;
}
{
buf[decoded_len] = '\n';
Curl_debug(data, CURLINFO_HEADER_IN, buf, decoded_len + 1);
}
buf[decoded_len] = '\0';
if(decoded_len <= 3)
/* suspiciously short */
return 0;
if(buf[3] != '-')
ret_code = atoi(buf);
if(buf[decoded_len - 1] == '\n')
buf[decoded_len - 1] = '\0';
strcpy(buffer, buf);
free(buf);
return ret_code;
}
static int sec_set_protection_level(struct Curl_easy *data)
{
int code;
struct connectdata *conn = data->conn;
unsigned char level = conn->request_data_prot;
DEBUGASSERT(level > PROT_NONE && level < PROT_LAST);
if(!conn->sec_complete) {
infof(data, "Trying to change the protection level after the"
" completion of the data exchange.");
return -1;
}
/* Bail out if we try to set up the same level */
if(conn->data_prot == level)
return 0;
if(level) {
char *pbsz;
unsigned int buffer_size = 1 << 20; /* 1048576 */
code = ftp_send_command(data, "PBSZ %u", buffer_size);
if(code < 0)
return -1;
if(code/100 != 2) {
failf(data, "Failed to set the protection's buffer size.");
return -1;
}
conn->buffer_size = buffer_size;
pbsz = strstr(data->state.buffer, "PBSZ=");
if(pbsz) {
/* stick to default value if the check fails */
if(!strncmp(pbsz, "PBSZ=", 5) && ISDIGIT(pbsz[5]))
buffer_size = atoi(&pbsz[5]);
if(buffer_size < conn->buffer_size)
conn->buffer_size = buffer_size;
}
}
/* Now try to negotiate the protection level. */
code = ftp_send_command(data, "PROT %c", level_to_char(level));
if(code < 0)
return -1;
if(code/100 != 2) {
failf(data, "Failed to set the protection level.");
return -1;
}
conn->data_prot = level;
if(level == PROT_PRIVATE)
conn->command_prot = level;
return 0;
}
int
Curl_sec_request_prot(struct connectdata *conn, const char *level)
{
unsigned char l = name_to_level(level);
if(l == PROT_NONE)
return -1;
DEBUGASSERT(l > PROT_NONE && l < PROT_LAST);
conn->request_data_prot = l;
return 0;
}
static CURLcode choose_mech(struct Curl_easy *data, struct connectdata *conn)
{
int ret;
void *tmp_allocation;
const struct Curl_sec_client_mech *mech = &Curl_krb5_client_mech;
tmp_allocation = realloc(conn->app_data, mech->size);
if(!tmp_allocation) {
failf(data, "Failed realloc of size %zu", mech->size);
mech = NULL;
return CURLE_OUT_OF_MEMORY;
}
conn->app_data = tmp_allocation;
if(mech->init) {
ret = mech->init(conn->app_data);
if(ret) {
infof(data, "Failed initialization for %s. Skipping it.",
mech->name);
return CURLE_FAILED_INIT;
}
}
infof(data, "Trying mechanism %s...", mech->name);
ret = ftp_send_command(data, "AUTH %s", mech->name);
if(ret < 0)
return CURLE_COULDNT_CONNECT;
if(ret/100 != 3) {
switch(ret) {
case 504:
infof(data, "Mechanism %s is not supported by the server (server "
"returned ftp code: 504).", mech->name);
break;
case 534:
infof(data, "Mechanism %s was rejected by the server (server returned "
"ftp code: 534).", mech->name);
break;
default:
if(ret/100 == 5) {
infof(data, "server does not support the security extensions");
return CURLE_USE_SSL_FAILED;
}
break;
}
return CURLE_LOGIN_DENIED;
}
/* Authenticate */
ret = mech->auth(conn->app_data, data, conn);
if(ret != AUTH_CONTINUE) {
if(ret != AUTH_OK) {
/* Mechanism has dumped the error to stderr, don't error here. */
return CURLE_USE_SSL_FAILED;
}
DEBUGASSERT(ret == AUTH_OK);
conn->mech = mech;
conn->sec_complete = 1;
conn->recv[FIRSTSOCKET] = sec_recv;
conn->send[FIRSTSOCKET] = sec_send;
conn->recv[SECONDARYSOCKET] = sec_recv;
conn->send[SECONDARYSOCKET] = sec_send;
conn->command_prot = PROT_SAFE;
/* Set the requested protection level */
/* BLOCKING */
(void)sec_set_protection_level(data);
}
return CURLE_OK;
}
CURLcode
Curl_sec_login(struct Curl_easy *data, struct connectdata *conn)
{
return choose_mech(data, conn);
}
void
Curl_sec_end(struct connectdata *conn)
{
if(conn->mech && conn->mech->end)
conn->mech->end(conn->app_data);
free(conn->app_data);
conn->app_data = NULL;
if(conn->in_buffer.data) {
free(conn->in_buffer.data);
conn->in_buffer.data = NULL;
conn->in_buffer.size = 0;
conn->in_buffer.index = 0;
conn->in_buffer.eof_flag = 0;
}
conn->sec_complete = 0;
conn->data_prot = PROT_CLEAR;
conn->mech = NULL;
}
#endif /* HAVE_GSSAPI && !CURL_DISABLE_FTP */