github
/
putty
зеркало из https://github.com/github/putty.git
1
0
Форкнуть 0
Код Задачи Пакеты Проекты Релизы Вики Активность
putty/ssh1connection.c

1214 строки
40 KiB
C
Исходник Обычный вид История

Move most of ssh.c out into separate source files. I've tried to separate out as many individually coherent changes from this work as I could into their own commits, but here's where I run out and have to commit the rest of this major refactoring as a big-bang change. Most of ssh.c is now no longer in ssh.c: all five of the main coroutines that handle layers of the SSH-1 and SSH-2 protocols now each have their own source file to live in, and a lot of the supporting functions have moved into the appropriate one of those too. The new abstraction is a vtable called 'PacketProtocolLayer', which has an input and output packet queue. Each layer's main coroutine is invoked from the method ssh_ppl_process_queue(), which is usually (though not exclusively) triggered automatically when things are pushed on the input queue. In SSH-2, the base layer is the transport protocol, and it contains a pair of subsidiary queues by which it passes some of its packets to the higher SSH-2 layers - first userauth and then connection, which are peers at the same level, with the former abdicating in favour of the latter at the appropriate moment. SSH-1 is simpler: the whole login phase of the protocol (crypto setup and authentication) is all in one module, and since SSH-1 has no repeat key exchange, that setup layer abdicates in favour of the connection phase when it's done. ssh.c itself is now about a tenth of its old size (which all by itself is cause for celebration!). Its main job is to set up all the layers, hook them up to each other and to the BPP, and to funnel data back and forth between that collection of modules and external things such as the network and the terminal. Once it's set up a collection of packet protocol layers, it communicates with them partly by calling methods of the base layer (and if that's ssh2transport then it will delegate some functionality to the corresponding methods of its higher layer), and partly by talking directly to the connection layer no matter where it is in the stack by means of the separate ConnectionLayer vtable which I introduced in commit 8001dd4cb, and to which I've now added quite a few extra methods replacing services that used to be internal function calls within ssh.c. (One effect of this is that the SSH-1 and SSH-2 channel storage is now no longer shared - there are distinct struct types ssh1_channel and ssh2_channel. That means a bit more code duplication, but on the plus side, a lot fewer confusing conditionals in the middle of half-shared functions, and less risk of a piece of SSH-1 escaping into SSH-2 or vice versa, which I remember has happened at least once in the past.) The bulk of this commit introduces the five new source files, their common header sshppl.h and some shared supporting routines in sshcommon.c, and rewrites nearly all of ssh.c itself. But it also includes a couple of other changes that I couldn't separate easily enough: Firstly, there's a new handling for socket EOF, in which ssh.c sets an 'input_eof' flag in the BPP, and that responds by checking a flag that tells it whether to report the EOF as an error or not. (This is the main reason for those new BPP_READ / BPP_WAITFOR macros - they can check the EOF flag every time the coroutine is resumed.) Secondly, the error reporting itself is changed around again. I'd expected to put some data fields in the public PacketProtocolLayer structure that it could set to report errors in the same way as the BPPs have been doing, but in the end, I decided propagating all those data fields around was a pain and that even the BPPs shouldn't have been doing it that way. So I've reverted to a system where everything calls back to functions in ssh.c itself to report any connection- ending condition. But there's a new family of those functions, categorising the possible such conditions by semantics, and each one has a different set of detailed effects (e.g. how rudely to close the network connection, what exit status should be passed back to the whole application, whether to send a disconnect message and/or display a GUI error box). I don't expect this to be immediately perfect: of course, the code has been through a big upheaval, new bugs are expected, and I haven't been able to do a full job of testing (e.g. I haven't tested every auth or kex method). But I've checked that it _basically_ works - both SSH protocols, all the different kinds of forwarding channel, more than one auth method, Windows and Linux, connection sharing - and I think it's now at the point where the easiest way to find further bugs is to let it out into the wild and see what users can spot.
2018-09-24 20:28:16 +03:00
/*
* Packet protocol layer for the SSH-1 'connection protocol', i.e.
* everything after authentication finishes.
*/
#include <assert.h>
#include "putty.h"
#include "ssh.h"
#include "sshbpp.h"
#include "sshppl.h"
#include "sshchan.h"
#include "sshcr.h"
struct ssh1_channel;
struct outstanding_succfail;
struct ssh1_connection_state {
int crState;
Get rid of lots of implicit pointer types. All the main backend structures - Ssh, Telnet, Pty, Serial etc - now describe structure types themselves rather than pointers to them. The same goes for the codebase-wide trait types Socket and Plug, and the supporting types SockAddr and Pinger. All those things that were typedefed as pointers are older types; the newer ones have the explicit * at the point of use, because that's what I now seem to be preferring. But whichever one of those is better, inconsistently using a mixture of the two styles is worse, so let's make everything consistent. A few types are still implicitly pointers, such as Bignum and some of the GSSAPI types; generally this is either because they have to be void *, or because they're typedefed differently on different platforms and aren't always pointers at all. Can't be helped. But I've got rid of the main ones, at least.
2018-10-04 21:10:23 +03:00
Ssh *ssh;
Move most of ssh.c out into separate source files. I've tried to separate out as many individually coherent changes from this work as I could into their own commits, but here's where I run out and have to commit the rest of this major refactoring as a big-bang change. Most of ssh.c is now no longer in ssh.c: all five of the main coroutines that handle layers of the SSH-1 and SSH-2 protocols now each have their own source file to live in, and a lot of the supporting functions have moved into the appropriate one of those too. The new abstraction is a vtable called 'PacketProtocolLayer', which has an input and output packet queue. Each layer's main coroutine is invoked from the method ssh_ppl_process_queue(), which is usually (though not exclusively) triggered automatically when things are pushed on the input queue. In SSH-2, the base layer is the transport protocol, and it contains a pair of subsidiary queues by which it passes some of its packets to the higher SSH-2 layers - first userauth and then connection, which are peers at the same level, with the former abdicating in favour of the latter at the appropriate moment. SSH-1 is simpler: the whole login phase of the protocol (crypto setup and authentication) is all in one module, and since SSH-1 has no repeat key exchange, that setup layer abdicates in favour of the connection phase when it's done. ssh.c itself is now about a tenth of its old size (which all by itself is cause for celebration!). Its main job is to set up all the layers, hook them up to each other and to the BPP, and to funnel data back and forth between that collection of modules and external things such as the network and the terminal. Once it's set up a collection of packet protocol layers, it communicates with them partly by calling methods of the base layer (and if that's ssh2transport then it will delegate some functionality to the corresponding methods of its higher layer), and partly by talking directly to the connection layer no matter where it is in the stack by means of the separate ConnectionLayer vtable which I introduced in commit 8001dd4cb, and to which I've now added quite a few extra methods replacing services that used to be internal function calls within ssh.c. (One effect of this is that the SSH-1 and SSH-2 channel storage is now no longer shared - there are distinct struct types ssh1_channel and ssh2_channel. That means a bit more code duplication, but on the plus side, a lot fewer confusing conditionals in the middle of half-shared functions, and less risk of a piece of SSH-1 escaping into SSH-2 or vice versa, which I remember has happened at least once in the past.) The bulk of this commit introduces the five new source files, their common header sshppl.h and some shared supporting routines in sshcommon.c, and rewrites nearly all of ssh.c itself. But it also includes a couple of other changes that I couldn't separate easily enough: Firstly, there's a new handling for socket EOF, in which ssh.c sets an 'input_eof' flag in the BPP, and that responds by checking a flag that tells it whether to report the EOF as an error or not. (This is the main reason for those new BPP_READ / BPP_WAITFOR macros - they can check the EOF flag every time the coroutine is resumed.) Secondly, the error reporting itself is changed around again. I'd expected to put some data fields in the public PacketProtocolLayer structure that it could set to report errors in the same way as the BPPs have been doing, but in the end, I decided propagating all those data fields around was a pain and that even the BPPs shouldn't have been doing it that way. So I've reverted to a system where everything calls back to functions in ssh.c itself to report any connection- ending condition. But there's a new family of those functions, categorising the possible such conditions by semantics, and each one has a different set of detailed effects (e.g. how rudely to close the network connection, what exit status should be passed back to the whole application, whether to send a disconnect message and/or display a GUI error box). I don't expect this to be immediately perfect: of course, the code has been through a big upheaval, new bugs are expected, and I haven't been able to do a full job of testing (e.g. I haven't tested every auth or kex method). But I've checked that it _basically_ works - both SSH protocols, all the different kinds of forwarding channel, more than one auth method, Windows and Linux, connection sharing - and I think it's now at the point where the easiest way to find further bugs is to let it out into the wild and see what users can spot.
2018-09-24 20:28:16 +03:00
Conf *conf;
int local_protoflags;
tree234 *channels; /* indexed by local id */
int got_pty;
int echoedit;
int ospeed, ispeed;
int stdout_throttling;
int session_ready;
int session_eof_pending, session_eof_sent, session_terminated;
int term_width, term_height, term_width_orig, term_height_orig;
int X11_fwd_enabled;
struct X11Display *x11disp;
struct X11FakeAuth *x11auth;
tree234 *x11authtree;
int agent_fwd_enabled;
tree234 *rportfwds;
PortFwdManager *portfwdmgr;
int portfwdmgr_configured;
int finished_setup;
/*
* These store the list of requests that we're waiting for
* SSH_SMSG_{SUCCESS,FAILURE} replies to. (Those messages don't
* come with any indication of what they're in response to, so we
* have to keep track of the queue ourselves.)
*/
struct outstanding_succfail *succfail_head, *succfail_tail;
ConnectionLayer cl;
PacketProtocolLayer ppl;
};
static int ssh1_rportfwd_cmp(void *av, void *bv)
{
struct ssh_rportfwd *a = (struct ssh_rportfwd *) av;
struct ssh_rportfwd *b = (struct ssh_rportfwd *) bv;
int i;
if ( (i = strcmp(a->dhost, b->dhost)) != 0)
return i < 0 ? -1 : +1;
if (a->dport > b->dport)
return +1;
if (a->dport < b->dport)
return -1;
return 0;
}
static void ssh1_connection_free(PacketProtocolLayer *);
static void ssh1_connection_process_queue(PacketProtocolLayer *);
static void ssh1_connection_special_cmd(PacketProtocolLayer *ppl,
SessionSpecialCode code, int arg);
static int ssh1_connection_want_user_input(PacketProtocolLayer *ppl);
static void ssh1_connection_got_user_input(PacketProtocolLayer *ppl);
static void ssh1_connection_reconfigure(PacketProtocolLayer *ppl, Conf *conf);
static const struct PacketProtocolLayerVtable ssh1_connection_vtable = {
ssh1_connection_free,
ssh1_connection_process_queue,
ssh1_common_get_specials,
ssh1_connection_special_cmd,
ssh1_connection_want_user_input,
ssh1_connection_got_user_input,
ssh1_connection_reconfigure,
NULL /* no layer names in SSH-1 */,
};
static struct ssh_rportfwd *ssh1_rportfwd_alloc(
ConnectionLayer *cl,
const char *shost, int sport, const char *dhost, int dport,
int addressfamily, const char *log_description, PortFwdRecord *pfr,
ssh_sharing_connstate *share_ctx);
static void ssh1_rportfwd_remove(
ConnectionLayer *cl, struct ssh_rportfwd *rpf);
static SshChannel *ssh1_lportfwd_open(
ConnectionLayer *cl, const char *hostname, int port,
const char *org, Channel *chan);
static int ssh1_agent_forwarding_permitted(ConnectionLayer *cl);
static void ssh1_terminal_size(ConnectionLayer *cl, int width, int height);
static void ssh1_stdout_unthrottle(ConnectionLayer *cl, int bufsize);
static int ssh1_stdin_backlog(ConnectionLayer *cl);
static void ssh1_throttle_all_channels(ConnectionLayer *cl, int throttled);
static int ssh1_ldisc_option(ConnectionLayer *cl, int option);
static const struct ConnectionLayerVtable ssh1_connlayer_vtable = {
ssh1_rportfwd_alloc,
ssh1_rportfwd_remove,
ssh1_lportfwd_open,
NULL /* add_sharing_x11_display */,
NULL /* remove_sharing_x11_display */,
NULL /* send_packet_from_downstream */,
NULL /* alloc_sharing_channel */,
NULL /* delete_sharing_channel */,
NULL /* sharing_queue_global_request */,
sshshare: notify cl when last downstream goes away. The check_termination function in ssh2connection is supposed to be called whenever it's possible that we've run out of (a) channels, and (b) sharing downstreams. I've been calling it on every channel close, but apparently completely forgot to add a callback from sshshare.c that also arranges to call it when we run out of downstreams.
2018-09-28 22:52:36 +03:00
NULL /* sharing_no_more_downstreams */,
Move most of ssh.c out into separate source files. I've tried to separate out as many individually coherent changes from this work as I could into their own commits, but here's where I run out and have to commit the rest of this major refactoring as a big-bang change. Most of ssh.c is now no longer in ssh.c: all five of the main coroutines that handle layers of the SSH-1 and SSH-2 protocols now each have their own source file to live in, and a lot of the supporting functions have moved into the appropriate one of those too. The new abstraction is a vtable called 'PacketProtocolLayer', which has an input and output packet queue. Each layer's main coroutine is invoked from the method ssh_ppl_process_queue(), which is usually (though not exclusively) triggered automatically when things are pushed on the input queue. In SSH-2, the base layer is the transport protocol, and it contains a pair of subsidiary queues by which it passes some of its packets to the higher SSH-2 layers - first userauth and then connection, which are peers at the same level, with the former abdicating in favour of the latter at the appropriate moment. SSH-1 is simpler: the whole login phase of the protocol (crypto setup and authentication) is all in one module, and since SSH-1 has no repeat key exchange, that setup layer abdicates in favour of the connection phase when it's done. ssh.c itself is now about a tenth of its old size (which all by itself is cause for celebration!). Its main job is to set up all the layers, hook them up to each other and to the BPP, and to funnel data back and forth between that collection of modules and external things such as the network and the terminal. Once it's set up a collection of packet protocol layers, it communicates with them partly by calling methods of the base layer (and if that's ssh2transport then it will delegate some functionality to the corresponding methods of its higher layer), and partly by talking directly to the connection layer no matter where it is in the stack by means of the separate ConnectionLayer vtable which I introduced in commit 8001dd4cb, and to which I've now added quite a few extra methods replacing services that used to be internal function calls within ssh.c. (One effect of this is that the SSH-1 and SSH-2 channel storage is now no longer shared - there are distinct struct types ssh1_channel and ssh2_channel. That means a bit more code duplication, but on the plus side, a lot fewer confusing conditionals in the middle of half-shared functions, and less risk of a piece of SSH-1 escaping into SSH-2 or vice versa, which I remember has happened at least once in the past.) The bulk of this commit introduces the five new source files, their common header sshppl.h and some shared supporting routines in sshcommon.c, and rewrites nearly all of ssh.c itself. But it also includes a couple of other changes that I couldn't separate easily enough: Firstly, there's a new handling for socket EOF, in which ssh.c sets an 'input_eof' flag in the BPP, and that responds by checking a flag that tells it whether to report the EOF as an error or not. (This is the main reason for those new BPP_READ / BPP_WAITFOR macros - they can check the EOF flag every time the coroutine is resumed.) Secondly, the error reporting itself is changed around again. I'd expected to put some data fields in the public PacketProtocolLayer structure that it could set to report errors in the same way as the BPPs have been doing, but in the end, I decided propagating all those data fields around was a pain and that even the BPPs shouldn't have been doing it that way. So I've reverted to a system where everything calls back to functions in ssh.c itself to report any connection- ending condition. But there's a new family of those functions, categorising the possible such conditions by semantics, and each one has a different set of detailed effects (e.g. how rudely to close the network connection, what exit status should be passed back to the whole application, whether to send a disconnect message and/or display a GUI error box). I don't expect this to be immediately perfect: of course, the code has been through a big upheaval, new bugs are expected, and I haven't been able to do a full job of testing (e.g. I haven't tested every auth or kex method). But I've checked that it _basically_ works - both SSH protocols, all the different kinds of forwarding channel, more than one auth method, Windows and Linux, connection sharing - and I think it's now at the point where the easiest way to find further bugs is to let it out into the wild and see what users can spot.
2018-09-24 20:28:16 +03:00
ssh1_agent_forwarding_permitted,
ssh1_terminal_size,
ssh1_stdout_unthrottle,
ssh1_stdin_backlog,
ssh1_throttle_all_channels,
ssh1_ldisc_option,
};
struct ssh1_channel {
struct ssh1_connection_state *connlayer;
unsigned remoteid, localid;
int type;
/* True if we opened this channel but server hasn't confirmed. */
int halfopen;
/* Bitmap of whether we've sent/received CHANNEL_CLOSE and
* CHANNEL_CLOSE_CONFIRMATION. */
#define CLOSES_SENT_CLOSE 1
#define CLOSES_SENT_CLOSECONF 2
#define CLOSES_RCVD_CLOSE 4
#define CLOSES_RCVD_CLOSECONF 8
int closes;
/*
* This flag indicates that an EOF is pending on the outgoing side
* of the channel: that is, wherever we're getting the data for
* this channel has sent us some data followed by EOF. We can't
* actually send the EOF until we've finished sending the data, so
* we set this flag instead to remind us to do so once our buffer
* is clear.
*/
int pending_eof;
/*
* True if this channel is causing the underlying connection to be
* throttled.
*/
int throttling_conn;
/*
* True if we currently have backed-up data on the direction of
* this channel pointing out of the SSH connection, and therefore
* would prefer the 'Channel' implementation not to read further
* local input if possible.
*/
int throttled_by_backlog;
Channel *chan; /* handle the client side of this channel, if not */
SshChannel sc; /* entry point for chan to talk back to */
};
static int ssh1channel_write(SshChannel *c, const void *buf, int len);
static void ssh1channel_write_eof(SshChannel *c);
static void ssh1channel_unclean_close(SshChannel *c, const char *err);
static void ssh1channel_unthrottle(SshChannel *c, int bufsize);
static Conf *ssh1channel_get_conf(SshChannel *c);
static const struct SshChannelVtable ssh1channel_vtable = {
ssh1channel_write,
ssh1channel_write_eof,
ssh1channel_unclean_close,
ssh1channel_unthrottle,
ssh1channel_get_conf,
NULL /* window_override_removed is only used by SSH-2 sharing */,
NULL /* x11_sharing_handover, likewise */,
};
static void ssh1_channel_init(struct ssh1_channel *c);
static void ssh1_channel_try_eof(struct ssh1_channel *c);
static void ssh1_channel_close_local(struct ssh1_channel *c,
const char *reason);
static void ssh1_channel_destroy(struct ssh1_channel *c);
static void ssh1_channel_check_close(struct ssh1_channel *c);
static int ssh1_check_termination(struct ssh1_connection_state *s);
typedef void (*sf_handler_fn_t)(struct ssh1_connection_state *s,
PktIn *pktin, void *ctx);
struct outstanding_succfail {
sf_handler_fn_t handler;
void *ctx;
struct outstanding_succfail *next;
};
static void ssh1_queue_succfail_handler(
struct ssh1_connection_state *s, sf_handler_fn_t handler, void *ctx)
{
struct outstanding_succfail *osf =
snew(struct outstanding_succfail);
osf->handler = handler;
osf->ctx = ctx;
if (s->succfail_tail)
s->succfail_tail->next = osf;
else
s->succfail_head = osf;
s->succfail_tail = osf;
}
static int ssh1_channelcmp(void *av, void *bv)
{
const struct ssh1_channel *a = (const struct ssh1_channel *) av;
const struct ssh1_channel *b = (const struct ssh1_channel *) bv;
if (a->localid < b->localid)
return -1;
if (a->localid > b->localid)
return +1;
return 0;
}
static int ssh1_channelfind(void *av, void *bv)
{
const unsigned *a = (const unsigned *) av;
const struct ssh1_channel *b = (const struct ssh1_channel *) bv;
if (*a < b->localid)
return -1;
if (*a > b->localid)
return +1;
return 0;
}
static void ssh1_channel_free(struct ssh1_channel *c)
{
if (c->chan)
chan_free(c->chan);
sfree(c);
}
PacketProtocolLayer *ssh1_connection_new(
Get rid of lots of implicit pointer types. All the main backend structures - Ssh, Telnet, Pty, Serial etc - now describe structure types themselves rather than pointers to them. The same goes for the codebase-wide trait types Socket and Plug, and the supporting types SockAddr and Pinger. All those things that were typedefed as pointers are older types; the newer ones have the explicit * at the point of use, because that's what I now seem to be preferring. But whichever one of those is better, inconsistently using a mixture of the two styles is worse, so let's make everything consistent. A few types are still implicitly pointers, such as Bignum and some of the GSSAPI types; generally this is either because they have to be void *, or because they're typedefed differently on different platforms and aren't always pointers at all. Can't be helped. But I've got rid of the main ones, at least.
2018-10-04 21:10:23 +03:00
Ssh *ssh, Conf *conf, ConnectionLayer **cl_out)
Move most of ssh.c out into separate source files. I've tried to separate out as many individually coherent changes from this work as I could into their own commits, but here's where I run out and have to commit the rest of this major refactoring as a big-bang change. Most of ssh.c is now no longer in ssh.c: all five of the main coroutines that handle layers of the SSH-1 and SSH-2 protocols now each have their own source file to live in, and a lot of the supporting functions have moved into the appropriate one of those too. The new abstraction is a vtable called 'PacketProtocolLayer', which has an input and output packet queue. Each layer's main coroutine is invoked from the method ssh_ppl_process_queue(), which is usually (though not exclusively) triggered automatically when things are pushed on the input queue. In SSH-2, the base layer is the transport protocol, and it contains a pair of subsidiary queues by which it passes some of its packets to the higher SSH-2 layers - first userauth and then connection, which are peers at the same level, with the former abdicating in favour of the latter at the appropriate moment. SSH-1 is simpler: the whole login phase of the protocol (crypto setup and authentication) is all in one module, and since SSH-1 has no repeat key exchange, that setup layer abdicates in favour of the connection phase when it's done. ssh.c itself is now about a tenth of its old size (which all by itself is cause for celebration!). Its main job is to set up all the layers, hook them up to each other and to the BPP, and to funnel data back and forth between that collection of modules and external things such as the network and the terminal. Once it's set up a collection of packet protocol layers, it communicates with them partly by calling methods of the base layer (and if that's ssh2transport then it will delegate some functionality to the corresponding methods of its higher layer), and partly by talking directly to the connection layer no matter where it is in the stack by means of the separate ConnectionLayer vtable which I introduced in commit 8001dd4cb, and to which I've now added quite a few extra methods replacing services that used to be internal function calls within ssh.c. (One effect of this is that the SSH-1 and SSH-2 channel storage is now no longer shared - there are distinct struct types ssh1_channel and ssh2_channel. That means a bit more code duplication, but on the plus side, a lot fewer confusing conditionals in the middle of half-shared functions, and less risk of a piece of SSH-1 escaping into SSH-2 or vice versa, which I remember has happened at least once in the past.) The bulk of this commit introduces the five new source files, their common header sshppl.h and some shared supporting routines in sshcommon.c, and rewrites nearly all of ssh.c itself. But it also includes a couple of other changes that I couldn't separate easily enough: Firstly, there's a new handling for socket EOF, in which ssh.c sets an 'input_eof' flag in the BPP, and that responds by checking a flag that tells it whether to report the EOF as an error or not. (This is the main reason for those new BPP_READ / BPP_WAITFOR macros - they can check the EOF flag every time the coroutine is resumed.) Secondly, the error reporting itself is changed around again. I'd expected to put some data fields in the public PacketProtocolLayer structure that it could set to report errors in the same way as the BPPs have been doing, but in the end, I decided propagating all those data fields around was a pain and that even the BPPs shouldn't have been doing it that way. So I've reverted to a system where everything calls back to functions in ssh.c itself to report any connection- ending condition. But there's a new family of those functions, categorising the possible such conditions by semantics, and each one has a different set of detailed effects (e.g. how rudely to close the network connection, what exit status should be passed back to the whole application, whether to send a disconnect message and/or display a GUI error box). I don't expect this to be immediately perfect: of course, the code has been through a big upheaval, new bugs are expected, and I haven't been able to do a full job of testing (e.g. I haven't tested every auth or kex method). But I've checked that it _basically_ works - both SSH protocols, all the different kinds of forwarding channel, more than one auth method, Windows and Linux, connection sharing - and I think it's now at the point where the easiest way to find further bugs is to let it out into the wild and see what users can spot.
2018-09-24 20:28:16 +03:00
{
struct ssh1_connection_state *s = snew(struct ssh1_connection_state);
memset(s, 0, sizeof(*s));
s->ppl.vt = &ssh1_connection_vtable;
s->conf = conf_copy(conf);
s->channels = newtree234(ssh1_channelcmp);
s->x11authtree = newtree234(x11_authcmp);
/* Need to get the frontend for s->cl now, because we won't be
* helpfully notified when a copy is written into s->ppl by our
* owner. */
s->cl.vt = &ssh1_connlayer_vtable;
s->cl.frontend = ssh_get_frontend(ssh);
s->portfwdmgr = portfwdmgr_new(&s->cl);
s->rportfwds = newtree234(ssh1_rportfwd_cmp);
*cl_out = &s->cl;
return &s->ppl;
}
static void ssh1_connection_free(PacketProtocolLayer *ppl)
{
struct ssh1_connection_state *s =
Rename FROMFIELD to 'container_of'. Ian Jackson points out that the Linux kernel has a macro of this name with the same purpose, and suggests that it's a good idea to use the same name as they do, so that at least some people reading one code base might recognise it from the other. I never really thought very hard about what order FROMFIELD's parameters should go in, and therefore I'm pleasantly surprised to find that my order agrees with the kernel's, so I don't have to permute every call site as part of making this change :-)
2018-10-06 01:49:08 +03:00
container_of(ppl, struct ssh1_connection_state, ppl);
Move most of ssh.c out into separate source files. I've tried to separate out as many individually coherent changes from this work as I could into their own commits, but here's where I run out and have to commit the rest of this major refactoring as a big-bang change. Most of ssh.c is now no longer in ssh.c: all five of the main coroutines that handle layers of the SSH-1 and SSH-2 protocols now each have their own source file to live in, and a lot of the supporting functions have moved into the appropriate one of those too. The new abstraction is a vtable called 'PacketProtocolLayer', which has an input and output packet queue. Each layer's main coroutine is invoked from the method ssh_ppl_process_queue(), which is usually (though not exclusively) triggered automatically when things are pushed on the input queue. In SSH-2, the base layer is the transport protocol, and it contains a pair of subsidiary queues by which it passes some of its packets to the higher SSH-2 layers - first userauth and then connection, which are peers at the same level, with the former abdicating in favour of the latter at the appropriate moment. SSH-1 is simpler: the whole login phase of the protocol (crypto setup and authentication) is all in one module, and since SSH-1 has no repeat key exchange, that setup layer abdicates in favour of the connection phase when it's done. ssh.c itself is now about a tenth of its old size (which all by itself is cause for celebration!). Its main job is to set up all the layers, hook them up to each other and to the BPP, and to funnel data back and forth between that collection of modules and external things such as the network and the terminal. Once it's set up a collection of packet protocol layers, it communicates with them partly by calling methods of the base layer (and if that's ssh2transport then it will delegate some functionality to the corresponding methods of its higher layer), and partly by talking directly to the connection layer no matter where it is in the stack by means of the separate ConnectionLayer vtable which I introduced in commit 8001dd4cb, and to which I've now added quite a few extra methods replacing services that used to be internal function calls within ssh.c. (One effect of this is that the SSH-1 and SSH-2 channel storage is now no longer shared - there are distinct struct types ssh1_channel and ssh2_channel. That means a bit more code duplication, but on the plus side, a lot fewer confusing conditionals in the middle of half-shared functions, and less risk of a piece of SSH-1 escaping into SSH-2 or vice versa, which I remember has happened at least once in the past.) The bulk of this commit introduces the five new source files, their common header sshppl.h and some shared supporting routines in sshcommon.c, and rewrites nearly all of ssh.c itself. But it also includes a couple of other changes that I couldn't separate easily enough: Firstly, there's a new handling for socket EOF, in which ssh.c sets an 'input_eof' flag in the BPP, and that responds by checking a flag that tells it whether to report the EOF as an error or not. (This is the main reason for those new BPP_READ / BPP_WAITFOR macros - they can check the EOF flag every time the coroutine is resumed.) Secondly, the error reporting itself is changed around again. I'd expected to put some data fields in the public PacketProtocolLayer structure that it could set to report errors in the same way as the BPPs have been doing, but in the end, I decided propagating all those data fields around was a pain and that even the BPPs shouldn't have been doing it that way. So I've reverted to a system where everything calls back to functions in ssh.c itself to report any connection- ending condition. But there's a new family of those functions, categorising the possible such conditions by semantics, and each one has a different set of detailed effects (e.g. how rudely to close the network connection, what exit status should be passed back to the whole application, whether to send a disconnect message and/or display a GUI error box). I don't expect this to be immediately perfect: of course, the code has been through a big upheaval, new bugs are expected, and I haven't been able to do a full job of testing (e.g. I haven't tested every auth or kex method). But I've checked that it _basically_ works - both SSH protocols, all the different kinds of forwarding channel, more than one auth method, Windows and Linux, connection sharing - and I think it's now at the point where the easiest way to find further bugs is to let it out into the wild and see what users can spot.
2018-09-24 20:28:16 +03:00
struct X11FakeAuth *auth;
struct ssh1_channel *c;
struct ssh_rportfwd *rpf;
conf_free(s->conf);
while ((c = delpos234(s->channels, 0)) != NULL)
ssh1_channel_free(c);
freetree234(s->channels);
if (s->x11disp)
x11_free_display(s->x11disp);
while ((auth = delpos234(s->x11authtree, 0)) != NULL)
x11_free_fake_auth(auth);
freetree234(s->x11authtree);
while ((rpf = delpos234(s->rportfwds, 0)) != NULL)
free_rportfwd(rpf);
freetree234(s->rportfwds);
portfwdmgr_free(s->portfwdmgr);
sfree(s);
}
void ssh1_connection_set_local_protoflags(PacketProtocolLayer *ppl, int flags)
{
assert(ppl->vt == &ssh1_connection_vtable);
struct ssh1_connection_state *s =
Rename FROMFIELD to 'container_of'. Ian Jackson points out that the Linux kernel has a macro of this name with the same purpose, and suggests that it's a good idea to use the same name as they do, so that at least some people reading one code base might recognise it from the other. I never really thought very hard about what order FROMFIELD's parameters should go in, and therefore I'm pleasantly surprised to find that my order agrees with the kernel's, so I don't have to permute every call site as part of making this change :-)
2018-10-06 01:49:08 +03:00
container_of(ppl, struct ssh1_connection_state, ppl);
Move most of ssh.c out into separate source files. I've tried to separate out as many individually coherent changes from this work as I could into their own commits, but here's where I run out and have to commit the rest of this major refactoring as a big-bang change. Most of ssh.c is now no longer in ssh.c: all five of the main coroutines that handle layers of the SSH-1 and SSH-2 protocols now each have their own source file to live in, and a lot of the supporting functions have moved into the appropriate one of those too. The new abstraction is a vtable called 'PacketProtocolLayer', which has an input and output packet queue. Each layer's main coroutine is invoked from the method ssh_ppl_process_queue(), which is usually (though not exclusively) triggered automatically when things are pushed on the input queue. In SSH-2, the base layer is the transport protocol, and it contains a pair of subsidiary queues by which it passes some of its packets to the higher SSH-2 layers - first userauth and then connection, which are peers at the same level, with the former abdicating in favour of the latter at the appropriate moment. SSH-1 is simpler: the whole login phase of the protocol (crypto setup and authentication) is all in one module, and since SSH-1 has no repeat key exchange, that setup layer abdicates in favour of the connection phase when it's done. ssh.c itself is now about a tenth of its old size (which all by itself is cause for celebration!). Its main job is to set up all the layers, hook them up to each other and to the BPP, and to funnel data back and forth between that collection of modules and external things such as the network and the terminal. Once it's set up a collection of packet protocol layers, it communicates with them partly by calling methods of the base layer (and if that's ssh2transport then it will delegate some functionality to the corresponding methods of its higher layer), and partly by talking directly to the connection layer no matter where it is in the stack by means of the separate ConnectionLayer vtable which I introduced in commit 8001dd4cb, and to which I've now added quite a few extra methods replacing services that used to be internal function calls within ssh.c. (One effect of this is that the SSH-1 and SSH-2 channel storage is now no longer shared - there are distinct struct types ssh1_channel and ssh2_channel. That means a bit more code duplication, but on the plus side, a lot fewer confusing conditionals in the middle of half-shared functions, and less risk of a piece of SSH-1 escaping into SSH-2 or vice versa, which I remember has happened at least once in the past.) The bulk of this commit introduces the five new source files, their common header sshppl.h and some shared supporting routines in sshcommon.c, and rewrites nearly all of ssh.c itself. But it also includes a couple of other changes that I couldn't separate easily enough: Firstly, there's a new handling for socket EOF, in which ssh.c sets an 'input_eof' flag in the BPP, and that responds by checking a flag that tells it whether to report the EOF as an error or not. (This is the main reason for those new BPP_READ / BPP_WAITFOR macros - they can check the EOF flag every time the coroutine is resumed.) Secondly, the error reporting itself is changed around again. I'd expected to put some data fields in the public PacketProtocolLayer structure that it could set to report errors in the same way as the BPPs have been doing, but in the end, I decided propagating all those data fields around was a pain and that even the BPPs shouldn't have been doing it that way. So I've reverted to a system where everything calls back to functions in ssh.c itself to report any connection- ending condition. But there's a new family of those functions, categorising the possible such conditions by semantics, and each one has a different set of detailed effects (e.g. how rudely to close the network connection, what exit status should be passed back to the whole application, whether to send a disconnect message and/or display a GUI error box). I don't expect this to be immediately perfect: of course, the code has been through a big upheaval, new bugs are expected, and I haven't been able to do a full job of testing (e.g. I haven't tested every auth or kex method). But I've checked that it _basically_ works - both SSH protocols, all the different kinds of forwarding channel, more than one auth method, Windows and Linux, connection sharing - and I think it's now at the point where the easiest way to find further bugs is to let it out into the wild and see what users can spot.
2018-09-24 20:28:16 +03:00
s->local_protoflags = flags;
}
static int ssh1_connection_filter_queue(struct ssh1_connection_state *s)
{
PktIn *pktin;
PktOut *pktout;
ptrlen data, host;
struct ssh1_channel *c;
unsigned localid, remid;
int port, expect_halfopen;
struct ssh_rportfwd pf, *pfp;
PacketProtocolLayer *ppl = &s->ppl; /* for ppl_logevent */
/* Cross-reference to ssh1login.c to handle the common packets
* between login and connection: DISCONNECT, DEBUG and IGNORE. */
extern int ssh1_common_filter_queue(PacketProtocolLayer *ppl);
while (1) {
if (ssh1_common_filter_queue(&s->ppl))
return TRUE;
if ((pktin = pq_peek(s->ppl.in_pq)) == NULL)
return FALSE;
switch (pktin->type) {
case SSH1_SMSG_SUCCESS:
case SSH1_SMSG_FAILURE:
if (!s->finished_setup) {
/* During initial setup, these messages are not
* filtered out, but go back to the main coroutine. */
return FALSE;
}
if (!s->succfail_head) {
ssh_remote_error(s->ppl.ssh,
"Received %s with no outstanding request",
ssh1_pkt_type(pktin->type));
return TRUE;
}
s->succfail_head->handler(s, pktin, s->succfail_head->ctx);
{
struct outstanding_succfail *tmp = s->succfail_head;
s->succfail_head = s->succfail_head->next;
sfree(tmp);
}
pq_pop(s->ppl.in_pq);
break;
case SSH1_SMSG_X11_OPEN:
remid = get_uint32(pktin);
/* Refuse if X11 forwarding is disabled. */
if (!s->X11_fwd_enabled) {
pktout = ssh_bpp_new_pktout(
s->ppl.bpp, SSH1_MSG_CHANNEL_OPEN_FAILURE);
put_uint32(pktout, remid);
pq_push(s->ppl.out_pq, pktout);
ppl_logevent(("Rejected X11 connect request"));
} else {
c = snew(struct ssh1_channel);
c->connlayer = s;
ssh1_channel_init(c);
c->remoteid = remid;
c->chan = x11_new_channel(s->x11authtree, &c->sc,
NULL, -1, FALSE);
c->remoteid = remid;
c->halfopen = FALSE;
pktout = ssh_bpp_new_pktout(
s->ppl.bpp, SSH1_MSG_CHANNEL_OPEN_CONFIRMATION);
put_uint32(pktout, c->remoteid);
put_uint32(pktout, c->localid);
pq_push(s->ppl.out_pq, pktout);
ppl_logevent(("Opened X11 forward channel"));
}
pq_pop(s->ppl.in_pq);
break;
case SSH1_SMSG_AGENT_OPEN:
remid = get_uint32(pktin);
/* Refuse if agent forwarding is disabled. */
if (!s->agent_fwd_enabled) {
pktout = ssh_bpp_new_pktout(
s->ppl.bpp, SSH1_MSG_CHANNEL_OPEN_FAILURE);
put_uint32(pktout, remid);
pq_push(s->ppl.out_pq, pktout);
} else {
c = snew(struct ssh1_channel);
c->connlayer = s;
ssh1_channel_init(c);
c->remoteid = remid;
c->chan = agentf_new(&c->sc);
c->halfopen = FALSE;
pktout = ssh_bpp_new_pktout(
s->ppl.bpp, SSH1_MSG_CHANNEL_OPEN_CONFIRMATION);
put_uint32(pktout, c->remoteid);
put_uint32(pktout, c->localid);
pq_push(s->ppl.out_pq, pktout);
}
pq_pop(s->ppl.in_pq);
break;
case SSH1_MSG_PORT_OPEN:
remid = get_uint32(pktin);
host = get_string(pktin);
port = toint(get_uint32(pktin));
pf.dhost = mkstr(host);
pf.dport = port;
pfp = find234(s->rportfwds, &pf, NULL);
if (!pfp) {
ppl_logevent(("Rejected remote port open request for %s:%d",
pf.dhost, port));
pktout = ssh_bpp_new_pktout(
s->ppl.bpp, SSH1_MSG_CHANNEL_OPEN_FAILURE);
put_uint32(pktout, remid);
pq_push(s->ppl.out_pq, pktout);
} else {
char *err;
c = snew(struct ssh1_channel);
c->connlayer = s;
ppl_logevent(("Received remote port open request for %s:%d",
pf.dhost, port));
err = portfwdmgr_connect(
s->portfwdmgr, &c->chan, pf.dhost, port,
&c->sc, pfp->addressfamily);
if (err) {
ppl_logevent(("Port open failed: %s", err));
sfree(err);
ssh1_channel_free(c);
pktout = ssh_bpp_new_pktout(
s->ppl.bpp, SSH1_MSG_CHANNEL_OPEN_FAILURE);
put_uint32(pktout, remid);
pq_push(s->ppl.out_pq, pktout);
} else {
ssh1_channel_init(c);
c->remoteid = remid;
c->halfopen = FALSE;
pktout = ssh_bpp_new_pktout(
s->ppl.bpp, SSH1_MSG_CHANNEL_OPEN_CONFIRMATION);
put_uint32(pktout, c->remoteid);
put_uint32(pktout, c->localid);
pq_push(s->ppl.out_pq, pktout);
ppl_logevent(("Forwarded port opened successfully"));
}
}
sfree(pf.dhost);
pq_pop(s->ppl.in_pq);
break;
case SSH1_MSG_CHANNEL_DATA:
case SSH1_MSG_CHANNEL_OPEN_CONFIRMATION:
case SSH1_MSG_CHANNEL_OPEN_FAILURE:
case SSH1_MSG_CHANNEL_CLOSE:
case SSH1_MSG_CHANNEL_CLOSE_CONFIRMATION:
/*
* Common preliminary code for all the messages from the
* server that cite one of our channel ids: look up that
* channel id, check it exists, and if it's for a sharing
* downstream, pass it on.
*/
localid = get_uint32(pktin);
c = find234(s->channels, &localid, ssh1_channelfind);
expect_halfopen = (
pktin->type == SSH1_MSG_CHANNEL_OPEN_CONFIRMATION ||
pktin->type == SSH1_MSG_CHANNEL_OPEN_FAILURE);
if (!c || c->halfopen != expect_halfopen) {
ssh_remote_error(
s->ppl.ssh, "Received %s for %s channel %u",
ssh1_pkt_type(pktin->type),
!c ? "nonexistent" : c->halfopen ? "half-open" : "open",
localid);
return TRUE;
}
switch (pktin->type) {
case SSH1_MSG_CHANNEL_OPEN_CONFIRMATION:
assert(c->halfopen);
c->remoteid = get_uint32(pktin);
c->halfopen = FALSE;
c->throttling_conn = FALSE;
chan_open_confirmation(c->chan);
/*
* Now that the channel is fully open, it's possible
* in principle to immediately close it. Check whether
* it wants us to!
*
* This can occur if a local socket error occurred
* between us sending out CHANNEL_OPEN and receiving
* OPEN_CONFIRMATION. If that happens, all we can do
* is immediately initiate close proceedings now that
* we know the server's id to put in the close
* message. We'll have handled that in this code by
* having already turned c->chan into a zombie, so its
* want_close method (which ssh1_channel_check_close
* will consult) will already be returning TRUE.
*/
ssh1_channel_check_close(c);
if (c->pending_eof)
ssh1_channel_try_eof(c); /* in case we had a pending EOF */
break;
case SSH1_MSG_CHANNEL_OPEN_FAILURE:
assert(c->halfopen);
chan_open_failed(c->chan, NULL);
chan_free(c->chan);
del234(s->channels, c);
ssh1_channel_free(c);
break;
case SSH1_MSG_CHANNEL_DATA:
data = get_string(pktin);
if (!get_err(pktin)) {
int bufsize = chan_send(
c->chan, FALSE, data.ptr, data.len);
if (!c->throttling_conn && bufsize > SSH1_BUFFER_LIMIT) {
c->throttling_conn = TRUE;
ssh_throttle_conn(s->ppl.ssh, +1);
}
}
break;
case SSH1_MSG_CHANNEL_CLOSE:
if (!(c->closes & CLOSES_RCVD_CLOSE)) {
c->closes |= CLOSES_RCVD_CLOSE;
chan_send_eof(c->chan);
ssh1_channel_check_close(c);
}
break;
case SSH1_MSG_CHANNEL_CLOSE_CONFIRMATION:
if (!(c->closes & CLOSES_RCVD_CLOSECONF)) {
if (!(c->closes & CLOSES_SENT_CLOSE)) {
ssh_remote_error(
s->ppl.ssh,
"Received CHANNEL_CLOSE_CONFIRMATION for channel"
" %u for which we never sent CHANNEL_CLOSE\n",
c->localid);
return TRUE;
}
c->closes |= CLOSES_RCVD_CLOSECONF;
ssh1_channel_check_close(c);
}
break;
}
pq_pop(s->ppl.in_pq);
break;
case SSH1_SMSG_STDOUT_DATA:
case SSH1_SMSG_STDERR_DATA:
data = get_string(pktin);
if (!get_err(pktin)) {
int bufsize = from_backend(
s->ppl.frontend, pktin->type == SSH1_SMSG_STDERR_DATA,
data.ptr, data.len);
if (!s->stdout_throttling && bufsize > SSH1_BUFFER_LIMIT) {
s->stdout_throttling = 1;
ssh_throttle_conn(s->ppl.ssh, +1);
}
}
pq_pop(s->ppl.in_pq);
break;
case SSH1_SMSG_EXIT_STATUS:
{
int exitcode = get_uint32(pktin);
ppl_logevent(("Server sent command exit status %d", exitcode));
ssh_got_exitcode(s->ppl.ssh, exitcode);
s->session_terminated = TRUE;
if (ssh1_check_termination(s))
return TRUE;
}
pq_pop(s->ppl.in_pq);
break;
default:
return FALSE;
}
}
}
static PktIn *ssh1_connection_pop(struct ssh1_connection_state *s)
{
ssh1_connection_filter_queue(s);
return pq_pop(s->ppl.in_pq);
}
static void ssh1_connection_process_queue(PacketProtocolLayer *ppl)
{
struct ssh1_connection_state *s =
Rename FROMFIELD to 'container_of'. Ian Jackson points out that the Linux kernel has a macro of this name with the same purpose, and suggests that it's a good idea to use the same name as they do, so that at least some people reading one code base might recognise it from the other. I never really thought very hard about what order FROMFIELD's parameters should go in, and therefore I'm pleasantly surprised to find that my order agrees with the kernel's, so I don't have to permute every call site as part of making this change :-)
2018-10-06 01:49:08 +03:00
container_of(ppl, struct ssh1_connection_state, ppl);
Move most of ssh.c out into separate source files. I've tried to separate out as many individually coherent changes from this work as I could into their own commits, but here's where I run out and have to commit the rest of this major refactoring as a big-bang change. Most of ssh.c is now no longer in ssh.c: all five of the main coroutines that handle layers of the SSH-1 and SSH-2 protocols now each have their own source file to live in, and a lot of the supporting functions have moved into the appropriate one of those too. The new abstraction is a vtable called 'PacketProtocolLayer', which has an input and output packet queue. Each layer's main coroutine is invoked from the method ssh_ppl_process_queue(), which is usually (though not exclusively) triggered automatically when things are pushed on the input queue. In SSH-2, the base layer is the transport protocol, and it contains a pair of subsidiary queues by which it passes some of its packets to the higher SSH-2 layers - first userauth and then connection, which are peers at the same level, with the former abdicating in favour of the latter at the appropriate moment. SSH-1 is simpler: the whole login phase of the protocol (crypto setup and authentication) is all in one module, and since SSH-1 has no repeat key exchange, that setup layer abdicates in favour of the connection phase when it's done. ssh.c itself is now about a tenth of its old size (which all by itself is cause for celebration!). Its main job is to set up all the layers, hook them up to each other and to the BPP, and to funnel data back and forth between that collection of modules and external things such as the network and the terminal. Once it's set up a collection of packet protocol layers, it communicates with them partly by calling methods of the base layer (and if that's ssh2transport then it will delegate some functionality to the corresponding methods of its higher layer), and partly by talking directly to the connection layer no matter where it is in the stack by means of the separate ConnectionLayer vtable which I introduced in commit 8001dd4cb, and to which I've now added quite a few extra methods replacing services that used to be internal function calls within ssh.c. (One effect of this is that the SSH-1 and SSH-2 channel storage is now no longer shared - there are distinct struct types ssh1_channel and ssh2_channel. That means a bit more code duplication, but on the plus side, a lot fewer confusing conditionals in the middle of half-shared functions, and less risk of a piece of SSH-1 escaping into SSH-2 or vice versa, which I remember has happened at least once in the past.) The bulk of this commit introduces the five new source files, their common header sshppl.h and some shared supporting routines in sshcommon.c, and rewrites nearly all of ssh.c itself. But it also includes a couple of other changes that I couldn't separate easily enough: Firstly, there's a new handling for socket EOF, in which ssh.c sets an 'input_eof' flag in the BPP, and that responds by checking a flag that tells it whether to report the EOF as an error or not. (This is the main reason for those new BPP_READ / BPP_WAITFOR macros - they can check the EOF flag every time the coroutine is resumed.) Secondly, the error reporting itself is changed around again. I'd expected to put some data fields in the public PacketProtocolLayer structure that it could set to report errors in the same way as the BPPs have been doing, but in the end, I decided propagating all those data fields around was a pain and that even the BPPs shouldn't have been doing it that way. So I've reverted to a system where everything calls back to functions in ssh.c itself to report any connection- ending condition. But there's a new family of those functions, categorising the possible such conditions by semantics, and each one has a different set of detailed effects (e.g. how rudely to close the network connection, what exit status should be passed back to the whole application, whether to send a disconnect message and/or display a GUI error box). I don't expect this to be immediately perfect: of course, the code has been through a big upheaval, new bugs are expected, and I haven't been able to do a full job of testing (e.g. I haven't tested every auth or kex method). But I've checked that it _basically_ works - both SSH protocols, all the different kinds of forwarding channel, more than one auth method, Windows and Linux, connection sharing - and I think it's now at the point where the easiest way to find further bugs is to let it out into the wild and see what users can spot.
2018-09-24 20:28:16 +03:00
PktIn *pktin;
PktOut *pktout;
if (ssh1_connection_filter_queue(s)) /* no matter why we were called */
return;
crBegin(s->crState);
if (ssh_agent_forwarding_permitted(&s->cl)) {
ppl_logevent(("Requesting agent forwarding"));
pktout = ssh_bpp_new_pktout(s->ppl.bpp,
SSH1_CMSG_AGENT_REQUEST_FORWARDING);
pq_push(s->ppl.out_pq, pktout);
crMaybeWaitUntilV((pktin = ssh1_connection_pop(s)) != NULL);
if (pktin->type == SSH1_SMSG_SUCCESS) {
ppl_logevent(("Agent forwarding enabled"));
s->agent_fwd_enabled = TRUE;
} else if (pktin->type == SSH1_SMSG_FAILURE) {
ppl_logevent(("Agent forwarding refused"));
} else {
ssh_proto_error(s->ppl.ssh, "Unexpected packet received"
" in response to agent forwarding request, "
"type %d (%s)", pktin->type,
ssh1_pkt_type(pktin->type));
return;
}
}
if (conf_get_int(s->conf, CONF_x11_forward)) {
s->x11disp =
x11_setup_display(conf_get_str(s->conf, CONF_x11_display),
s->conf);
if (!s->x11disp) {
/* FIXME: return an error message from x11_setup_display */
ppl_logevent(("X11 forwarding not enabled: unable to"
" initialise X display"));
} else {
s->x11auth = x11_invent_fake_auth
(s->x11authtree, conf_get_int(s->conf, CONF_x11_auth));
s->x11auth->disp = s->x11disp;
ppl_logevent(("Requesting X11 forwarding"));
pktout = ssh_bpp_new_pktout(
s->ppl.bpp, SSH1_CMSG_X11_REQUEST_FORWARDING);
put_stringz(pktout, s->x11auth->protoname);
put_stringz(pktout, s->x11auth->datastring);
if (s->local_protoflags & SSH1_PROTOFLAG_SCREEN_NUMBER)
put_uint32(pktout, s->x11disp->screennum);
pq_push(s->ppl.out_pq, pktout);
crMaybeWaitUntilV((pktin = ssh1_connection_pop(s)) != NULL);
if (pktin->type == SSH1_SMSG_SUCCESS) {
ppl_logevent(("X11 forwarding enabled"));
s->X11_fwd_enabled = TRUE;
} else if (pktin->type == SSH1_SMSG_FAILURE) {
ppl_logevent(("X11 forwarding refused"));
} else {
ssh_proto_error(s->ppl.ssh, "Unexpected packet received"
" in response to X11 forwarding request, "
"type %d (%s)", pktin->type,
ssh1_pkt_type(pktin->type));
return;
}
}
}
portfwdmgr_config(s->portfwdmgr, s->conf);
s->portfwdmgr_configured = TRUE;
if (!conf_get_int(s->conf, CONF_nopty)) {
/* Unpick the terminal-speed string. */
/* XXX perhaps we should allow no speeds to be sent. */
s->ospeed = 38400; s->ispeed = 38400; /* last-resort defaults */
sscanf(conf_get_str(s->conf, CONF_termspeed), "%d,%d",
&s->ospeed, &s->ispeed);
/* Send the pty request. */
pktout = ssh_bpp_new_pktout(s->ppl.bpp, SSH1_CMSG_REQUEST_PTY);
put_stringz(pktout, conf_get_str(s->conf, CONF_termtype));
put_uint32(pktout, s->term_height);
put_uint32(pktout, s->term_width);
s->term_width_orig = s->term_width;
s->term_height_orig = s->term_height;
put_uint32(pktout, 0); /* width in pixels */
put_uint32(pktout, 0); /* height in pixels */
write_ttymodes_to_packet_from_conf(
BinarySink_UPCAST(pktout), s->ppl.frontend, s->conf,
1, s->ospeed, s->ispeed);
pq_push(s->ppl.out_pq, pktout);
crMaybeWaitUntilV((pktin = ssh1_connection_pop(s)) != NULL);
if (pktin->type == SSH1_SMSG_SUCCESS) {
ppl_logevent(("Allocated pty (ospeed %dbps, ispeed %dbps)",
s->ospeed, s->ispeed));
s->got_pty = TRUE;
} else if (pktin->type == SSH1_SMSG_FAILURE) {
ppl_printf(("Server refused to allocate pty\r\n"));
s->echoedit = TRUE;
} else {
ssh_proto_error(s->ppl.ssh, "Unexpected packet received"
" in response to pty request, "
"type %d (%s)", pktin->type,
ssh1_pkt_type(pktin->type));
crStopV;
}
} else {
s->echoedit = TRUE;
}
/*
* Start the shell or command.
*
* Special case: if the first-choice command is an SSH-2
* subsystem (hence not usable here) and the second choice
* exists, we fall straight back to that.
*/
{
char *cmd = conf_get_str(s->conf, CONF_remote_cmd);
if (conf_get_int(s->conf, CONF_ssh_subsys) &&
conf_get_str(s->conf, CONF_remote_cmd2)) {
cmd = conf_get_str(s->conf, CONF_remote_cmd2);
ssh_got_fallback_cmd(s->ppl.ssh);
}
if (*cmd) {
pktout = ssh_bpp_new_pktout(s->ppl.bpp, SSH1_CMSG_EXEC_CMD);
put_stringz(pktout, cmd);
} else {
pktout = ssh_bpp_new_pktout(s->ppl.bpp, SSH1_CMSG_EXEC_SHELL);
}
pq_push(s->ppl.out_pq, pktout);
ppl_logevent(("Started session"));
}
s->session_ready = TRUE;
Fix minor mishandling of session typeahead. When the connection layer is ready to receive user input, it sets the flag causing ssh_ppl_want_user_input to return true. But one thing it _didn't_ do was to check whether the user input bufchain already had some data in it because the user had typed ahead of the session setup, and send that input immediately if so. Now it does.
2018-09-25 10:55:54 +03:00
ssh_ppl_got_user_input(&s->ppl); /* in case any input is already queued */
Move most of ssh.c out into separate source files. I've tried to separate out as many individually coherent changes from this work as I could into their own commits, but here's where I run out and have to commit the rest of this major refactoring as a big-bang change. Most of ssh.c is now no longer in ssh.c: all five of the main coroutines that handle layers of the SSH-1 and SSH-2 protocols now each have their own source file to live in, and a lot of the supporting functions have moved into the appropriate one of those too. The new abstraction is a vtable called 'PacketProtocolLayer', which has an input and output packet queue. Each layer's main coroutine is invoked from the method ssh_ppl_process_queue(), which is usually (though not exclusively) triggered automatically when things are pushed on the input queue. In SSH-2, the base layer is the transport protocol, and it contains a pair of subsidiary queues by which it passes some of its packets to the higher SSH-2 layers - first userauth and then connection, which are peers at the same level, with the former abdicating in favour of the latter at the appropriate moment. SSH-1 is simpler: the whole login phase of the protocol (crypto setup and authentication) is all in one module, and since SSH-1 has no repeat key exchange, that setup layer abdicates in favour of the connection phase when it's done. ssh.c itself is now about a tenth of its old size (which all by itself is cause for celebration!). Its main job is to set up all the layers, hook them up to each other and to the BPP, and to funnel data back and forth between that collection of modules and external things such as the network and the terminal. Once it's set up a collection of packet protocol layers, it communicates with them partly by calling methods of the base layer (and if that's ssh2transport then it will delegate some functionality to the corresponding methods of its higher layer), and partly by talking directly to the connection layer no matter where it is in the stack by means of the separate ConnectionLayer vtable which I introduced in commit 8001dd4cb, and to which I've now added quite a few extra methods replacing services that used to be internal function calls within ssh.c. (One effect of this is that the SSH-1 and SSH-2 channel storage is now no longer shared - there are distinct struct types ssh1_channel and ssh2_channel. That means a bit more code duplication, but on the plus side, a lot fewer confusing conditionals in the middle of half-shared functions, and less risk of a piece of SSH-1 escaping into SSH-2 or vice versa, which I remember has happened at least once in the past.) The bulk of this commit introduces the five new source files, their common header sshppl.h and some shared supporting routines in sshcommon.c, and rewrites nearly all of ssh.c itself. But it also includes a couple of other changes that I couldn't separate easily enough: Firstly, there's a new handling for socket EOF, in which ssh.c sets an 'input_eof' flag in the BPP, and that responds by checking a flag that tells it whether to report the EOF as an error or not. (This is the main reason for those new BPP_READ / BPP_WAITFOR macros - they can check the EOF flag every time the coroutine is resumed.) Secondly, the error reporting itself is changed around again. I'd expected to put some data fields in the public PacketProtocolLayer structure that it could set to report errors in the same way as the BPPs have been doing, but in the end, I decided propagating all those data fields around was a pain and that even the BPPs shouldn't have been doing it that way. So I've reverted to a system where everything calls back to functions in ssh.c itself to report any connection- ending condition. But there's a new family of those functions, categorising the possible such conditions by semantics, and each one has a different set of detailed effects (e.g. how rudely to close the network connection, what exit status should be passed back to the whole application, whether to send a disconnect message and/or display a GUI error box). I don't expect this to be immediately perfect: of course, the code has been through a big upheaval, new bugs are expected, and I haven't been able to do a full job of testing (e.g. I haven't tested every auth or kex method). But I've checked that it _basically_ works - both SSH protocols, all the different kinds of forwarding channel, more than one auth method, Windows and Linux, connection sharing - and I think it's now at the point where the easiest way to find further bugs is to let it out into the wild and see what users can spot.
2018-09-24 20:28:16 +03:00
/* If an EOF or a window-size change arrived before we were ready
* to handle either one, handle them now. */
if (s->session_eof_pending) {
ssh_ppl_special_cmd(&s->ppl, SS_EOF, 0);
s->session_eof_pending = FALSE;
}
if (s->term_width_orig != s->term_width ||
s->term_height_orig != s->term_height)
ssh_terminal_size(&s->cl, s->term_width, s->term_height);
ssh_ldisc_update(s->ppl.ssh);
s->finished_setup = TRUE;
while (1) {
/*
* By this point, most incoming packets are already being
* handled by filter_queue, and we need only pay attention to
* the unusual ones.
*/
if ((pktin = ssh1_connection_pop(s)) != NULL) {
ssh_proto_error(s->ppl.ssh, "Unexpected packet received, "
"type %d (%s)", pktin->type,
ssh1_pkt_type(pktin->type));
return;
}
while (bufchain_size(s->ppl.user_input) > 0) {
void *data;
int len;
bufchain_prefix(s->ppl.user_input, &data, &len);
if (len > 512)
len = 512;
pktout = ssh_bpp_new_pktout(s->ppl.bpp, SSH1_CMSG_STDIN_DATA);
put_string(pktout, data, len);
pq_push(s->ppl.out_pq, pktout);
bufchain_consume(s->ppl.user_input, len);
}
crReturnV;
}
crFinishV;
}
static void ssh1_channel_check_close(struct ssh1_channel *c)
{
struct ssh1_connection_state *s = c->connlayer;
PktOut *pktout;
if (c->halfopen) {
/*
* If we've sent out our own CHANNEL_OPEN but not yet seen
* either OPEN_CONFIRMATION or OPEN_FAILURE in response, then
* it's too early to be sending close messages of any kind.
*/
return;
}
if ((!((CLOSES_SENT_CLOSE | CLOSES_RCVD_CLOSE) & ~c->closes) ||
chan_want_close(c->chan, (c->closes & CLOSES_SENT_CLOSE),
(c->closes & CLOSES_RCVD_CLOSE))) &&
!(c->closes & CLOSES_SENT_CLOSECONF)) {
/*
* We have both sent and received CLOSE (or the channel type
* doesn't need us to), which means the channel is in final
* wind-up. Send CLOSE and/or CLOSE_CONFIRMATION, whichever we
* haven't sent yet.
*/
if (!(c->closes & CLOSES_SENT_CLOSE)) {
pktout = ssh_bpp_new_pktout(s->ppl.bpp, SSH1_MSG_CHANNEL_CLOSE);
put_uint32(pktout, c->remoteid);
pq_push(s->ppl.out_pq, pktout);
c->closes |= CLOSES_SENT_CLOSE;
}
if (c->closes & CLOSES_RCVD_CLOSE) {
pktout = ssh_bpp_new_pktout(
s->ppl.bpp, SSH1_MSG_CHANNEL_CLOSE_CONFIRMATION);
put_uint32(pktout, c->remoteid);
pq_push(s->ppl.out_pq, pktout);
c->closes |= CLOSES_SENT_CLOSECONF;
}
}
if (!((CLOSES_SENT_CLOSECONF | CLOSES_RCVD_CLOSECONF) & ~c->closes)) {
/*
* We have both sent and received CLOSE_CONFIRMATION, which
* means we're completely done with the channel.
*/
ssh1_channel_destroy(c);
}
}
static void ssh1_channel_try_eof(struct ssh1_channel *c)
{
struct ssh1_connection_state *s = c->connlayer;
PktOut *pktout;
assert(c->pending_eof); /* precondition for calling us */
if (c->halfopen)
return; /* can't close: not even opened yet */
c->pending_eof = FALSE; /* we're about to send it */
pktout = ssh_bpp_new_pktout(s->ppl.bpp, SSH1_MSG_CHANNEL_CLOSE);
put_uint32(pktout, c->remoteid);
pq_push(s->ppl.out_pq, pktout);
c->closes |= CLOSES_SENT_CLOSE;
ssh1_channel_check_close(c);
}
/*
* Close any local socket and free any local resources associated with
* a channel. This converts the channel into a zombie.
*/
static void ssh1_channel_close_local(struct ssh1_channel *c,
const char *reason)
{
struct ssh1_connection_state *s = c->connlayer;
PacketProtocolLayer *ppl = &s->ppl; /* for ppl_logevent */
const char *msg = chan_log_close_msg(c->chan);
if (msg != NULL)
ppl_logevent(("%s%s%s", msg, reason ? " " : "", reason ? reason : ""));
chan_free(c->chan);
c->chan = zombiechan_new();
}
static void ssh1_check_termination_callback(void *vctx)
{
struct ssh1_connection_state *s = (struct ssh1_connection_state *)vctx;
ssh1_check_termination(s);
}
static void ssh1_channel_destroy(struct ssh1_channel *c)
{
struct ssh1_connection_state *s = c->connlayer;
ssh1_channel_close_local(c, NULL);
del234(s->channels, c);
ssh1_channel_free(c);
/*
* If that was the last channel left open, we might need to
* terminate. But we'll be a bit cautious, by doing that in a
* toplevel callback, just in case anything on the current call
* stack objects to this entire PPL being freed.
*/
queue_toplevel_callback(ssh1_check_termination_callback, s);
}
static int ssh1_check_termination(struct ssh1_connection_state *s)
{
/*
* Decide whether we should terminate the SSH connection now.
* Called after a channel goes away, or when the main session
* returns SSH1_SMSG_EXIT_STATUS; we terminate when none of either
* is left.
*/
if (s->session_terminated && count234(s->channels) == 0) {
PktOut *pktout = ssh_bpp_new_pktout(
s->ppl.bpp, SSH1_CMSG_EXIT_CONFIRMATION);
pq_push(s->ppl.out_pq, pktout);
ssh_user_close(s->ppl.ssh, "Session finished");
return TRUE;
}
return FALSE;
}
/*
* Set up most of a new ssh1_channel. Leaves chan untouched (since it
* will sometimes have been filled in before calling this).
*/
static void ssh1_channel_init(struct ssh1_channel *c)
{
struct ssh1_connection_state *s = c->connlayer;
c->closes = 0;
c->pending_eof = FALSE;
c->throttling_conn = FALSE;
c->sc.vt = &ssh1channel_vtable;
c->localid = alloc_channel_id(s->channels, struct ssh1_channel);
add234(s->channels, c);
}
static Conf *ssh1channel_get_conf(SshChannel *sc)
{
Rename FROMFIELD to 'container_of'. Ian Jackson points out that the Linux kernel has a macro of this name with the same purpose, and suggests that it's a good idea to use the same name as they do, so that at least some people reading one code base might recognise it from the other. I never really thought very hard about what order FROMFIELD's parameters should go in, and therefore I'm pleasantly surprised to find that my order agrees with the kernel's, so I don't have to permute every call site as part of making this change :-)
2018-10-06 01:49:08 +03:00
struct ssh1_channel *c = container_of(sc, struct ssh1_channel, sc);
Move most of ssh.c out into separate source files. I've tried to separate out as many individually coherent changes from this work as I could into their own commits, but here's where I run out and have to commit the rest of this major refactoring as a big-bang change. Most of ssh.c is now no longer in ssh.c: all five of the main coroutines that handle layers of the SSH-1 and SSH-2 protocols now each have their own source file to live in, and a lot of the supporting functions have moved into the appropriate one of those too. The new abstraction is a vtable called 'PacketProtocolLayer', which has an input and output packet queue. Each layer's main coroutine is invoked from the method ssh_ppl_process_queue(), which is usually (though not exclusively) triggered automatically when things are pushed on the input queue. In SSH-2, the base layer is the transport protocol, and it contains a pair of subsidiary queues by which it passes some of its packets to the higher SSH-2 layers - first userauth and then connection, which are peers at the same level, with the former abdicating in favour of the latter at the appropriate moment. SSH-1 is simpler: the whole login phase of the protocol (crypto setup and authentication) is all in one module, and since SSH-1 has no repeat key exchange, that setup layer abdicates in favour of the connection phase when it's done. ssh.c itself is now about a tenth of its old size (which all by itself is cause for celebration!). Its main job is to set up all the layers, hook them up to each other and to the BPP, and to funnel data back and forth between that collection of modules and external things such as the network and the terminal. Once it's set up a collection of packet protocol layers, it communicates with them partly by calling methods of the base layer (and if that's ssh2transport then it will delegate some functionality to the corresponding methods of its higher layer), and partly by talking directly to the connection layer no matter where it is in the stack by means of the separate ConnectionLayer vtable which I introduced in commit 8001dd4cb, and to which I've now added quite a few extra methods replacing services that used to be internal function calls within ssh.c. (One effect of this is that the SSH-1 and SSH-2 channel storage is now no longer shared - there are distinct struct types ssh1_channel and ssh2_channel. That means a bit more code duplication, but on the plus side, a lot fewer confusing conditionals in the middle of half-shared functions, and less risk of a piece of SSH-1 escaping into SSH-2 or vice versa, which I remember has happened at least once in the past.) The bulk of this commit introduces the five new source files, their common header sshppl.h and some shared supporting routines in sshcommon.c, and rewrites nearly all of ssh.c itself. But it also includes a couple of other changes that I couldn't separate easily enough: Firstly, there's a new handling for socket EOF, in which ssh.c sets an 'input_eof' flag in the BPP, and that responds by checking a flag that tells it whether to report the EOF as an error or not. (This is the main reason for those new BPP_READ / BPP_WAITFOR macros - they can check the EOF flag every time the coroutine is resumed.) Secondly, the error reporting itself is changed around again. I'd expected to put some data fields in the public PacketProtocolLayer structure that it could set to report errors in the same way as the BPPs have been doing, but in the end, I decided propagating all those data fields around was a pain and that even the BPPs shouldn't have been doing it that way. So I've reverted to a system where everything calls back to functions in ssh.c itself to report any connection- ending condition. But there's a new family of those functions, categorising the possible such conditions by semantics, and each one has a different set of detailed effects (e.g. how rudely to close the network connection, what exit status should be passed back to the whole application, whether to send a disconnect message and/or display a GUI error box). I don't expect this to be immediately perfect: of course, the code has been through a big upheaval, new bugs are expected, and I haven't been able to do a full job of testing (e.g. I haven't tested every auth or kex method). But I've checked that it _basically_ works - both SSH protocols, all the different kinds of forwarding channel, more than one auth method, Windows and Linux, connection sharing - and I think it's now at the point where the easiest way to find further bugs is to let it out into the wild and see what users can spot.
2018-09-24 20:28:16 +03:00
struct ssh1_connection_state *s = c->connlayer;
return s->conf;
}
static void ssh1channel_write_eof(SshChannel *sc)
{
Rename FROMFIELD to 'container_of'. Ian Jackson points out that the Linux kernel has a macro of this name with the same purpose, and suggests that it's a good idea to use the same name as they do, so that at least some people reading one code base might recognise it from the other. I never really thought very hard about what order FROMFIELD's parameters should go in, and therefore I'm pleasantly surprised to find that my order agrees with the kernel's, so I don't have to permute every call site as part of making this change :-)
2018-10-06 01:49:08 +03:00
struct ssh1_channel *c = container_of(sc, struct ssh1_channel, sc);
Move most of ssh.c out into separate source files. I've tried to separate out as many individually coherent changes from this work as I could into their own commits, but here's where I run out and have to commit the rest of this major refactoring as a big-bang change. Most of ssh.c is now no longer in ssh.c: all five of the main coroutines that handle layers of the SSH-1 and SSH-2 protocols now each have their own source file to live in, and a lot of the supporting functions have moved into the appropriate one of those too. The new abstraction is a vtable called 'PacketProtocolLayer', which has an input and output packet queue. Each layer's main coroutine is invoked from the method ssh_ppl_process_queue(), which is usually (though not exclusively) triggered automatically when things are pushed on the input queue. In SSH-2, the base layer is the transport protocol, and it contains a pair of subsidiary queues by which it passes some of its packets to the higher SSH-2 layers - first userauth and then connection, which are peers at the same level, with the former abdicating in favour of the latter at the appropriate moment. SSH-1 is simpler: the whole login phase of the protocol (crypto setup and authentication) is all in one module, and since SSH-1 has no repeat key exchange, that setup layer abdicates in favour of the connection phase when it's done. ssh.c itself is now about a tenth of its old size (which all by itself is cause for celebration!). Its main job is to set up all the layers, hook them up to each other and to the BPP, and to funnel data back and forth between that collection of modules and external things such as the network and the terminal. Once it's set up a collection of packet protocol layers, it communicates with them partly by calling methods of the base layer (and if that's ssh2transport then it will delegate some functionality to the corresponding methods of its higher layer), and partly by talking directly to the connection layer no matter where it is in the stack by means of the separate ConnectionLayer vtable which I introduced in commit 8001dd4cb, and to which I've now added quite a few extra methods replacing services that used to be internal function calls within ssh.c. (One effect of this is that the SSH-1 and SSH-2 channel storage is now no longer shared - there are distinct struct types ssh1_channel and ssh2_channel. That means a bit more code duplication, but on the plus side, a lot fewer confusing conditionals in the middle of half-shared functions, and less risk of a piece of SSH-1 escaping into SSH-2 or vice versa, which I remember has happened at least once in the past.) The bulk of this commit introduces the five new source files, their common header sshppl.h and some shared supporting routines in sshcommon.c, and rewrites nearly all of ssh.c itself. But it also includes a couple of other changes that I couldn't separate easily enough: Firstly, there's a new handling for socket EOF, in which ssh.c sets an 'input_eof' flag in the BPP, and that responds by checking a flag that tells it whether to report the EOF as an error or not. (This is the main reason for those new BPP_READ / BPP_WAITFOR macros - they can check the EOF flag every time the coroutine is resumed.) Secondly, the error reporting itself is changed around again. I'd expected to put some data fields in the public PacketProtocolLayer structure that it could set to report errors in the same way as the BPPs have been doing, but in the end, I decided propagating all those data fields around was a pain and that even the BPPs shouldn't have been doing it that way. So I've reverted to a system where everything calls back to functions in ssh.c itself to report any connection- ending condition. But there's a new family of those functions, categorising the possible such conditions by semantics, and each one has a different set of detailed effects (e.g. how rudely to close the network connection, what exit status should be passed back to the whole application, whether to send a disconnect message and/or display a GUI error box). I don't expect this to be immediately perfect: of course, the code has been through a big upheaval, new bugs are expected, and I haven't been able to do a full job of testing (e.g. I haven't tested every auth or kex method). But I've checked that it _basically_ works - both SSH protocols, all the different kinds of forwarding channel, more than one auth method, Windows and Linux, connection sharing - and I think it's now at the point where the easiest way to find further bugs is to let it out into the wild and see what users can spot.
2018-09-24 20:28:16 +03:00
if (c->closes & CLOSES_SENT_CLOSE)
return;
c->pending_eof = TRUE;
ssh1_channel_try_eof(c);
}
static void ssh1channel_unclean_close(SshChannel *sc, const char *err)
{
Rename FROMFIELD to 'container_of'. Ian Jackson points out that the Linux kernel has a macro of this name with the same purpose, and suggests that it's a good idea to use the same name as they do, so that at least some people reading one code base might recognise it from the other. I never really thought very hard about what order FROMFIELD's parameters should go in, and therefore I'm pleasantly surprised to find that my order agrees with the kernel's, so I don't have to permute every call site as part of making this change :-)
2018-10-06 01:49:08 +03:00
struct ssh1_channel *c = container_of(sc, struct ssh1_channel, sc);
Move most of ssh.c out into separate source files. I've tried to separate out as many individually coherent changes from this work as I could into their own commits, but here's where I run out and have to commit the rest of this major refactoring as a big-bang change. Most of ssh.c is now no longer in ssh.c: all five of the main coroutines that handle layers of the SSH-1 and SSH-2 protocols now each have their own source file to live in, and a lot of the supporting functions have moved into the appropriate one of those too. The new abstraction is a vtable called 'PacketProtocolLayer', which has an input and output packet queue. Each layer's main coroutine is invoked from the method ssh_ppl_process_queue(), which is usually (though not exclusively) triggered automatically when things are pushed on the input queue. In SSH-2, the base layer is the transport protocol, and it contains a pair of subsidiary queues by which it passes some of its packets to the higher SSH-2 layers - first userauth and then connection, which are peers at the same level, with the former abdicating in favour of the latter at the appropriate moment. SSH-1 is simpler: the whole login phase of the protocol (crypto setup and authentication) is all in one module, and since SSH-1 has no repeat key exchange, that setup layer abdicates in favour of the connection phase when it's done. ssh.c itself is now about a tenth of its old size (which all by itself is cause for celebration!). Its main job is to set up all the layers, hook them up to each other and to the BPP, and to funnel data back and forth between that collection of modules and external things such as the network and the terminal. Once it's set up a collection of packet protocol layers, it communicates with them partly by calling methods of the base layer (and if that's ssh2transport then it will delegate some functionality to the corresponding methods of its higher layer), and partly by talking directly to the connection layer no matter where it is in the stack by means of the separate ConnectionLayer vtable which I introduced in commit 8001dd4cb, and to which I've now added quite a few extra methods replacing services that used to be internal function calls within ssh.c. (One effect of this is that the SSH-1 and SSH-2 channel storage is now no longer shared - there are distinct struct types ssh1_channel and ssh2_channel. That means a bit more code duplication, but on the plus side, a lot fewer confusing conditionals in the middle of half-shared functions, and less risk of a piece of SSH-1 escaping into SSH-2 or vice versa, which I remember has happened at least once in the past.) The bulk of this commit introduces the five new source files, their common header sshppl.h and some shared supporting routines in sshcommon.c, and rewrites nearly all of ssh.c itself. But it also includes a couple of other changes that I couldn't separate easily enough: Firstly, there's a new handling for socket EOF, in which ssh.c sets an 'input_eof' flag in the BPP, and that responds by checking a flag that tells it whether to report the EOF as an error or not. (This is the main reason for those new BPP_READ / BPP_WAITFOR macros - they can check the EOF flag every time the coroutine is resumed.) Secondly, the error reporting itself is changed around again. I'd expected to put some data fields in the public PacketProtocolLayer structure that it could set to report errors in the same way as the BPPs have been doing, but in the end, I decided propagating all those data fields around was a pain and that even the BPPs shouldn't have been doing it that way. So I've reverted to a system where everything calls back to functions in ssh.c itself to report any connection- ending condition. But there's a new family of those functions, categorising the possible such conditions by semantics, and each one has a different set of detailed effects (e.g. how rudely to close the network connection, what exit status should be passed back to the whole application, whether to send a disconnect message and/or display a GUI error box). I don't expect this to be immediately perfect: of course, the code has been through a big upheaval, new bugs are expected, and I haven't been able to do a full job of testing (e.g. I haven't tested every auth or kex method). But I've checked that it _basically_ works - both SSH protocols, all the different kinds of forwarding channel, more than one auth method, Windows and Linux, connection sharing - and I think it's now at the point where the easiest way to find further bugs is to let it out into the wild and see what users can spot.
2018-09-24 20:28:16 +03:00
char *reason;
reason = dupprintf("due to local error: %s", err);
ssh1_channel_close_local(c, reason);
sfree(reason);
c->pending_eof = FALSE; /* this will confuse a zombie channel */
ssh1_channel_check_close(c);
}
static void ssh1channel_unthrottle(SshChannel *sc, int bufsize)
{
Rename FROMFIELD to 'container_of'. Ian Jackson points out that the Linux kernel has a macro of this name with the same purpose, and suggests that it's a good idea to use the same name as they do, so that at least some people reading one code base might recognise it from the other. I never really thought very hard about what order FROMFIELD's parameters should go in, and therefore I'm pleasantly surprised to find that my order agrees with the kernel's, so I don't have to permute every call site as part of making this change :-)
2018-10-06 01:49:08 +03:00
struct ssh1_channel *c = container_of(sc, struct ssh1_channel, sc);
Move most of ssh.c out into separate source files. I've tried to separate out as many individually coherent changes from this work as I could into their own commits, but here's where I run out and have to commit the rest of this major refactoring as a big-bang change. Most of ssh.c is now no longer in ssh.c: all five of the main coroutines that handle layers of the SSH-1 and SSH-2 protocols now each have their own source file to live in, and a lot of the supporting functions have moved into the appropriate one of those too. The new abstraction is a vtable called 'PacketProtocolLayer', which has an input and output packet queue. Each layer's main coroutine is invoked from the method ssh_ppl_process_queue(), which is usually (though not exclusively) triggered automatically when things are pushed on the input queue. In SSH-2, the base layer is the transport protocol, and it contains a pair of subsidiary queues by which it passes some of its packets to the higher SSH-2 layers - first userauth and then connection, which are peers at the same level, with the former abdicating in favour of the latter at the appropriate moment. SSH-1 is simpler: the whole login phase of the protocol (crypto setup and authentication) is all in one module, and since SSH-1 has no repeat key exchange, that setup layer abdicates in favour of the connection phase when it's done. ssh.c itself is now about a tenth of its old size (which all by itself is cause for celebration!). Its main job is to set up all the layers, hook them up to each other and to the BPP, and to funnel data back and forth between that collection of modules and external things such as the network and the terminal. Once it's set up a collection of packet protocol layers, it communicates with them partly by calling methods of the base layer (and if that's ssh2transport then it will delegate some functionality to the corresponding methods of its higher layer), and partly by talking directly to the connection layer no matter where it is in the stack by means of the separate ConnectionLayer vtable which I introduced in commit 8001dd4cb, and to which I've now added quite a few extra methods replacing services that used to be internal function calls within ssh.c. (One effect of this is that the SSH-1 and SSH-2 channel storage is now no longer shared - there are distinct struct types ssh1_channel and ssh2_channel. That means a bit more code duplication, but on the plus side, a lot fewer confusing conditionals in the middle of half-shared functions, and less risk of a piece of SSH-1 escaping into SSH-2 or vice versa, which I remember has happened at least once in the past.) The bulk of this commit introduces the five new source files, their common header sshppl.h and some shared supporting routines in sshcommon.c, and rewrites nearly all of ssh.c itself. But it also includes a couple of other changes that I couldn't separate easily enough: Firstly, there's a new handling for socket EOF, in which ssh.c sets an 'input_eof' flag in the BPP, and that responds by checking a flag that tells it whether to report the EOF as an error or not. (This is the main reason for those new BPP_READ / BPP_WAITFOR macros - they can check the EOF flag every time the coroutine is resumed.) Secondly, the error reporting itself is changed around again. I'd expected to put some data fields in the public PacketProtocolLayer structure that it could set to report errors in the same way as the BPPs have been doing, but in the end, I decided propagating all those data fields around was a pain and that even the BPPs shouldn't have been doing it that way. So I've reverted to a system where everything calls back to functions in ssh.c itself to report any connection- ending condition. But there's a new family of those functions, categorising the possible such conditions by semantics, and each one has a different set of detailed effects (e.g. how rudely to close the network connection, what exit status should be passed back to the whole application, whether to send a disconnect message and/or display a GUI error box). I don't expect this to be immediately perfect: of course, the code has been through a big upheaval, new bugs are expected, and I haven't been able to do a full job of testing (e.g. I haven't tested every auth or kex method). But I've checked that it _basically_ works - both SSH protocols, all the different kinds of forwarding channel, more than one auth method, Windows and Linux, connection sharing - and I think it's now at the point where the easiest way to find further bugs is to let it out into the wild and see what users can spot.
2018-09-24 20:28:16 +03:00
struct ssh1_connection_state *s = c->connlayer;
if (c->throttling_conn && bufsize <= SSH1_BUFFER_LIMIT) {
c->throttling_conn = 0;
ssh_throttle_conn(s->ppl.ssh, -1);
}
}
static int ssh1channel_write(SshChannel *sc, const void *buf, int len)
{
Rename FROMFIELD to 'container_of'. Ian Jackson points out that the Linux kernel has a macro of this name with the same purpose, and suggests that it's a good idea to use the same name as they do, so that at least some people reading one code base might recognise it from the other. I never really thought very hard about what order FROMFIELD's parameters should go in, and therefore I'm pleasantly surprised to find that my order agrees with the kernel's, so I don't have to permute every call site as part of making this change :-)
2018-10-06 01:49:08 +03:00
struct ssh1_channel *c = container_of(sc, struct ssh1_channel, sc);
Move most of ssh.c out into separate source files. I've tried to separate out as many individually coherent changes from this work as I could into their own commits, but here's where I run out and have to commit the rest of this major refactoring as a big-bang change. Most of ssh.c is now no longer in ssh.c: all five of the main coroutines that handle layers of the SSH-1 and SSH-2 protocols now each have their own source file to live in, and a lot of the supporting functions have moved into the appropriate one of those too. The new abstraction is a vtable called 'PacketProtocolLayer', which has an input and output packet queue. Each layer's main coroutine is invoked from the method ssh_ppl_process_queue(), which is usually (though not exclusively) triggered automatically when things are pushed on the input queue. In SSH-2, the base layer is the transport protocol, and it contains a pair of subsidiary queues by which it passes some of its packets to the higher SSH-2 layers - first userauth and then connection, which are peers at the same level, with the former abdicating in favour of the latter at the appropriate moment. SSH-1 is simpler: the whole login phase of the protocol (crypto setup and authentication) is all in one module, and since SSH-1 has no repeat key exchange, that setup layer abdicates in favour of the connection phase when it's done. ssh.c itself is now about a tenth of its old size (which all by itself is cause for celebration!). Its main job is to set up all the layers, hook them up to each other and to the BPP, and to funnel data back and forth between that collection of modules and external things such as the network and the terminal. Once it's set up a collection of packet protocol layers, it communicates with them partly by calling methods of the base layer (and if that's ssh2transport then it will delegate some functionality to the corresponding methods of its higher layer), and partly by talking directly to the connection layer no matter where it is in the stack by means of the separate ConnectionLayer vtable which I introduced in commit 8001dd4cb, and to which I've now added quite a few extra methods replacing services that used to be internal function calls within ssh.c. (One effect of this is that the SSH-1 and SSH-2 channel storage is now no longer shared - there are distinct struct types ssh1_channel and ssh2_channel. That means a bit more code duplication, but on the plus side, a lot fewer confusing conditionals in the middle of half-shared functions, and less risk of a piece of SSH-1 escaping into SSH-2 or vice versa, which I remember has happened at least once in the past.) The bulk of this commit introduces the five new source files, their common header sshppl.h and some shared supporting routines in sshcommon.c, and rewrites nearly all of ssh.c itself. But it also includes a couple of other changes that I couldn't separate easily enough: Firstly, there's a new handling for socket EOF, in which ssh.c sets an 'input_eof' flag in the BPP, and that responds by checking a flag that tells it whether to report the EOF as an error or not. (This is the main reason for those new BPP_READ / BPP_WAITFOR macros - they can check the EOF flag every time the coroutine is resumed.) Secondly, the error reporting itself is changed around again. I'd expected to put some data fields in the public PacketProtocolLayer structure that it could set to report errors in the same way as the BPPs have been doing, but in the end, I decided propagating all those data fields around was a pain and that even the BPPs shouldn't have been doing it that way. So I've reverted to a system where everything calls back to functions in ssh.c itself to report any connection- ending condition. But there's a new family of those functions, categorising the possible such conditions by semantics, and each one has a different set of detailed effects (e.g. how rudely to close the network connection, what exit status should be passed back to the whole application, whether to send a disconnect message and/or display a GUI error box). I don't expect this to be immediately perfect: of course, the code has been through a big upheaval, new bugs are expected, and I haven't been able to do a full job of testing (e.g. I haven't tested every auth or kex method). But I've checked that it _basically_ works - both SSH protocols, all the different kinds of forwarding channel, more than one auth method, Windows and Linux, connection sharing - and I think it's now at the point where the easiest way to find further bugs is to let it out into the wild and see what users can spot.
2018-09-24 20:28:16 +03:00
struct ssh1_connection_state *s = c->connlayer;
assert(!(c->closes & CLOSES_SENT_CLOSE));
PktOut *pktout = ssh_bpp_new_pktout(s->ppl.bpp, SSH1_MSG_CHANNEL_DATA);
put_uint32(pktout, c->remoteid);
put_string(pktout, buf, len);
pq_push(s->ppl.out_pq, pktout);
/*
* In SSH-1 we can return 0 here - implying that channels are
* never individually throttled - because the only circumstance
* that can cause throttling will be the whole SSH connection
* backing up, in which case _everything_ will be throttled as a
* whole.
*/
return 0;
}
static SshChannel *ssh1_lportfwd_open(
ConnectionLayer *cl, const char *hostname, int port,
const char *org, Channel *chan)
{
struct ssh1_connection_state *s =
Rename FROMFIELD to 'container_of'. Ian Jackson points out that the Linux kernel has a macro of this name with the same purpose, and suggests that it's a good idea to use the same name as they do, so that at least some people reading one code base might recognise it from the other. I never really thought very hard about what order FROMFIELD's parameters should go in, and therefore I'm pleasantly surprised to find that my order agrees with the kernel's, so I don't have to permute every call site as part of making this change :-)
2018-10-06 01:49:08 +03:00
container_of(cl, struct ssh1_connection_state, cl);
Move most of ssh.c out into separate source files. I've tried to separate out as many individually coherent changes from this work as I could into their own commits, but here's where I run out and have to commit the rest of this major refactoring as a big-bang change. Most of ssh.c is now no longer in ssh.c: all five of the main coroutines that handle layers of the SSH-1 and SSH-2 protocols now each have their own source file to live in, and a lot of the supporting functions have moved into the appropriate one of those too. The new abstraction is a vtable called 'PacketProtocolLayer', which has an input and output packet queue. Each layer's main coroutine is invoked from the method ssh_ppl_process_queue(), which is usually (though not exclusively) triggered automatically when things are pushed on the input queue. In SSH-2, the base layer is the transport protocol, and it contains a pair of subsidiary queues by which it passes some of its packets to the higher SSH-2 layers - first userauth and then connection, which are peers at the same level, with the former abdicating in favour of the latter at the appropriate moment. SSH-1 is simpler: the whole login phase of the protocol (crypto setup and authentication) is all in one module, and since SSH-1 has no repeat key exchange, that setup layer abdicates in favour of the connection phase when it's done. ssh.c itself is now about a tenth of its old size (which all by itself is cause for celebration!). Its main job is to set up all the layers, hook them up to each other and to the BPP, and to funnel data back and forth between that collection of modules and external things such as the network and the terminal. Once it's set up a collection of packet protocol layers, it communicates with them partly by calling methods of the base layer (and if that's ssh2transport then it will delegate some functionality to the corresponding methods of its higher layer), and partly by talking directly to the connection layer no matter where it is in the stack by means of the separate ConnectionLayer vtable which I introduced in commit 8001dd4cb, and to which I've now added quite a few extra methods replacing services that used to be internal function calls within ssh.c. (One effect of this is that the SSH-1 and SSH-2 channel storage is now no longer shared - there are distinct struct types ssh1_channel and ssh2_channel. That means a bit more code duplication, but on the plus side, a lot fewer confusing conditionals in the middle of half-shared functions, and less risk of a piece of SSH-1 escaping into SSH-2 or vice versa, which I remember has happened at least once in the past.) The bulk of this commit introduces the five new source files, their common header sshppl.h and some shared supporting routines in sshcommon.c, and rewrites nearly all of ssh.c itself. But it also includes a couple of other changes that I couldn't separate easily enough: Firstly, there's a new handling for socket EOF, in which ssh.c sets an 'input_eof' flag in the BPP, and that responds by checking a flag that tells it whether to report the EOF as an error or not. (This is the main reason for those new BPP_READ / BPP_WAITFOR macros - they can check the EOF flag every time the coroutine is resumed.) Secondly, the error reporting itself is changed around again. I'd expected to put some data fields in the public PacketProtocolLayer structure that it could set to report errors in the same way as the BPPs have been doing, but in the end, I decided propagating all those data fields around was a pain and that even the BPPs shouldn't have been doing it that way. So I've reverted to a system where everything calls back to functions in ssh.c itself to report any connection- ending condition. But there's a new family of those functions, categorising the possible such conditions by semantics, and each one has a different set of detailed effects (e.g. how rudely to close the network connection, what exit status should be passed back to the whole application, whether to send a disconnect message and/or display a GUI error box). I don't expect this to be immediately perfect: of course, the code has been through a big upheaval, new bugs are expected, and I haven't been able to do a full job of testing (e.g. I haven't tested every auth or kex method). But I've checked that it _basically_ works - both SSH protocols, all the different kinds of forwarding channel, more than one auth method, Windows and Linux, connection sharing - and I think it's now at the point where the easiest way to find further bugs is to let it out into the wild and see what users can spot.
2018-09-24 20:28:16 +03:00
PacketProtocolLayer *ppl = &s->ppl; /* for ppl_logevent */
struct ssh1_channel *c = snew(struct ssh1_channel);
PktOut *pktout;
c->connlayer = s;
ssh1_channel_init(c);
c->halfopen = TRUE;
c->chan = chan;
ppl_logevent(("Opening connection to %s:%d for %s", hostname, port, org));
pktout = ssh_bpp_new_pktout(s->ppl.bpp, SSH1_MSG_PORT_OPEN);
put_uint32(pktout, c->localid);
put_stringz(pktout, hostname);
put_uint32(pktout, port);
/* originator string would go here, but we didn't specify
* SSH_PROTOFLAG_HOST_IN_FWD_OPEN */
pq_push(s->ppl.out_pq, pktout);
return &c->sc;
}
static void ssh1_rportfwd_response(struct ssh1_connection_state *s,
PktIn *pktin, void *ctx)
{
PacketProtocolLayer *ppl = &s->ppl; /* for ppl_logevent */
struct ssh_rportfwd *rpf = (struct ssh_rportfwd *)ctx;
if (pktin->type == SSH1_SMSG_SUCCESS) {
ppl_logevent(("Remote port forwarding from %s enabled",
rpf->log_description));
} else {
ppl_logevent(("Remote port forwarding from %s refused",
rpf->log_description));
struct ssh_rportfwd *realpf = del234(s->rportfwds, rpf);
assert(realpf == rpf);
portfwdmgr_close(s->portfwdmgr, rpf->pfr);
free_rportfwd(rpf);
}
}
static struct ssh_rportfwd *ssh1_rportfwd_alloc(
ConnectionLayer *cl,
const char *shost, int sport, const char *dhost, int dport,
int addressfamily, const char *log_description, PortFwdRecord *pfr,
ssh_sharing_connstate *share_ctx)
{
struct ssh1_connection_state *s =
Rename FROMFIELD to 'container_of'. Ian Jackson points out that the Linux kernel has a macro of this name with the same purpose, and suggests that it's a good idea to use the same name as they do, so that at least some people reading one code base might recognise it from the other. I never really thought very hard about what order FROMFIELD's parameters should go in, and therefore I'm pleasantly surprised to find that my order agrees with the kernel's, so I don't have to permute every call site as part of making this change :-)
2018-10-06 01:49:08 +03:00
container_of(cl, struct ssh1_connection_state, cl);
Move most of ssh.c out into separate source files. I've tried to separate out as many individually coherent changes from this work as I could into their own commits, but here's where I run out and have to commit the rest of this major refactoring as a big-bang change. Most of ssh.c is now no longer in ssh.c: all five of the main coroutines that handle layers of the SSH-1 and SSH-2 protocols now each have their own source file to live in, and a lot of the supporting functions have moved into the appropriate one of those too. The new abstraction is a vtable called 'PacketProtocolLayer', which has an input and output packet queue. Each layer's main coroutine is invoked from the method ssh_ppl_process_queue(), which is usually (though not exclusively) triggered automatically when things are pushed on the input queue. In SSH-2, the base layer is the transport protocol, and it contains a pair of subsidiary queues by which it passes some of its packets to the higher SSH-2 layers - first userauth and then connection, which are peers at the same level, with the former abdicating in favour of the latter at the appropriate moment. SSH-1 is simpler: the whole login phase of the protocol (crypto setup and authentication) is all in one module, and since SSH-1 has no repeat key exchange, that setup layer abdicates in favour of the connection phase when it's done. ssh.c itself is now about a tenth of its old size (which all by itself is cause for celebration!). Its main job is to set up all the layers, hook them up to each other and to the BPP, and to funnel data back and forth between that collection of modules and external things such as the network and the terminal. Once it's set up a collection of packet protocol layers, it communicates with them partly by calling methods of the base layer (and if that's ssh2transport then it will delegate some functionality to the corresponding methods of its higher layer), and partly by talking directly to the connection layer no matter where it is in the stack by means of the separate ConnectionLayer vtable which I introduced in commit 8001dd4cb, and to which I've now added quite a few extra methods replacing services that used to be internal function calls within ssh.c. (One effect of this is that the SSH-1 and SSH-2 channel storage is now no longer shared - there are distinct struct types ssh1_channel and ssh2_channel. That means a bit more code duplication, but on the plus side, a lot fewer confusing conditionals in the middle of half-shared functions, and less risk of a piece of SSH-1 escaping into SSH-2 or vice versa, which I remember has happened at least once in the past.) The bulk of this commit introduces the five new source files, their common header sshppl.h and some shared supporting routines in sshcommon.c, and rewrites nearly all of ssh.c itself. But it also includes a couple of other changes that I couldn't separate easily enough: Firstly, there's a new handling for socket EOF, in which ssh.c sets an 'input_eof' flag in the BPP, and that responds by checking a flag that tells it whether to report the EOF as an error or not. (This is the main reason for those new BPP_READ / BPP_WAITFOR macros - they can check the EOF flag every time the coroutine is resumed.) Secondly, the error reporting itself is changed around again. I'd expected to put some data fields in the public PacketProtocolLayer structure that it could set to report errors in the same way as the BPPs have been doing, but in the end, I decided propagating all those data fields around was a pain and that even the BPPs shouldn't have been doing it that way. So I've reverted to a system where everything calls back to functions in ssh.c itself to report any connection- ending condition. But there's a new family of those functions, categorising the possible such conditions by semantics, and each one has a different set of detailed effects (e.g. how rudely to close the network connection, what exit status should be passed back to the whole application, whether to send a disconnect message and/or display a GUI error box). I don't expect this to be immediately perfect: of course, the code has been through a big upheaval, new bugs are expected, and I haven't been able to do a full job of testing (e.g. I haven't tested every auth or kex method). But I've checked that it _basically_ works - both SSH protocols, all the different kinds of forwarding channel, more than one auth method, Windows and Linux, connection sharing - and I think it's now at the point where the easiest way to find further bugs is to let it out into the wild and see what users can spot.
2018-09-24 20:28:16 +03:00
struct ssh_rportfwd *rpf = snew(struct ssh_rportfwd);
rpf->shost = dupstr(shost);
rpf->sport = sport;
rpf->dhost = dupstr(dhost);
rpf->dport = dport;
rpf->addressfamily = addressfamily;
rpf->log_description = dupstr(log_description);
rpf->pfr = pfr;
if (add234(s->rportfwds, rpf) != rpf) {
free_rportfwd(rpf);
return NULL;
}
PktOut *pktout = ssh_bpp_new_pktout(
s->ppl.bpp, SSH1_CMSG_PORT_FORWARD_REQUEST);
put_uint32(pktout, rpf->sport);
put_stringz(pktout, rpf->dhost);
put_uint32(pktout, rpf->dport);
pq_push(s->ppl.out_pq, pktout);
ssh1_queue_succfail_handler(s, ssh1_rportfwd_response, rpf);
return rpf;
}
static void ssh1_rportfwd_remove(ConnectionLayer *cl, struct ssh_rportfwd *rpf)
{
/*
* We cannot cancel listening ports on the server side in SSH-1!
* There's no message to support it.
*/
}
static int ssh1_agent_forwarding_permitted(ConnectionLayer *cl)
{
struct ssh1_connection_state *s =
Rename FROMFIELD to 'container_of'. Ian Jackson points out that the Linux kernel has a macro of this name with the same purpose, and suggests that it's a good idea to use the same name as they do, so that at least some people reading one code base might recognise it from the other. I never really thought very hard about what order FROMFIELD's parameters should go in, and therefore I'm pleasantly surprised to find that my order agrees with the kernel's, so I don't have to permute every call site as part of making this change :-)
2018-10-06 01:49:08 +03:00
container_of(cl, struct ssh1_connection_state, cl);
Move most of ssh.c out into separate source files. I've tried to separate out as many individually coherent changes from this work as I could into their own commits, but here's where I run out and have to commit the rest of this major refactoring as a big-bang change. Most of ssh.c is now no longer in ssh.c: all five of the main coroutines that handle layers of the SSH-1 and SSH-2 protocols now each have their own source file to live in, and a lot of the supporting functions have moved into the appropriate one of those too. The new abstraction is a vtable called 'PacketProtocolLayer', which has an input and output packet queue. Each layer's main coroutine is invoked from the method ssh_ppl_process_queue(), which is usually (though not exclusively) triggered automatically when things are pushed on the input queue. In SSH-2, the base layer is the transport protocol, and it contains a pair of subsidiary queues by which it passes some of its packets to the higher SSH-2 layers - first userauth and then connection, which are peers at the same level, with the former abdicating in favour of the latter at the appropriate moment. SSH-1 is simpler: the whole login phase of the protocol (crypto setup and authentication) is all in one module, and since SSH-1 has no repeat key exchange, that setup layer abdicates in favour of the connection phase when it's done. ssh.c itself is now about a tenth of its old size (which all by itself is cause for celebration!). Its main job is to set up all the layers, hook them up to each other and to the BPP, and to funnel data back and forth between that collection of modules and external things such as the network and the terminal. Once it's set up a collection of packet protocol layers, it communicates with them partly by calling methods of the base layer (and if that's ssh2transport then it will delegate some functionality to the corresponding methods of its higher layer), and partly by talking directly to the connection layer no matter where it is in the stack by means of the separate ConnectionLayer vtable which I introduced in commit 8001dd4cb, and to which I've now added quite a few extra methods replacing services that used to be internal function calls within ssh.c. (One effect of this is that the SSH-1 and SSH-2 channel storage is now no longer shared - there are distinct struct types ssh1_channel and ssh2_channel. That means a bit more code duplication, but on the plus side, a lot fewer confusing conditionals in the middle of half-shared functions, and less risk of a piece of SSH-1 escaping into SSH-2 or vice versa, which I remember has happened at least once in the past.) The bulk of this commit introduces the five new source files, their common header sshppl.h and some shared supporting routines in sshcommon.c, and rewrites nearly all of ssh.c itself. But it also includes a couple of other changes that I couldn't separate easily enough: Firstly, there's a new handling for socket EOF, in which ssh.c sets an 'input_eof' flag in the BPP, and that responds by checking a flag that tells it whether to report the EOF as an error or not. (This is the main reason for those new BPP_READ / BPP_WAITFOR macros - they can check the EOF flag every time the coroutine is resumed.) Secondly, the error reporting itself is changed around again. I'd expected to put some data fields in the public PacketProtocolLayer structure that it could set to report errors in the same way as the BPPs have been doing, but in the end, I decided propagating all those data fields around was a pain and that even the BPPs shouldn't have been doing it that way. So I've reverted to a system where everything calls back to functions in ssh.c itself to report any connection- ending condition. But there's a new family of those functions, categorising the possible such conditions by semantics, and each one has a different set of detailed effects (e.g. how rudely to close the network connection, what exit status should be passed back to the whole application, whether to send a disconnect message and/or display a GUI error box). I don't expect this to be immediately perfect: of course, the code has been through a big upheaval, new bugs are expected, and I haven't been able to do a full job of testing (e.g. I haven't tested every auth or kex method). But I've checked that it _basically_ works - both SSH protocols, all the different kinds of forwarding channel, more than one auth method, Windows and Linux, connection sharing - and I think it's now at the point where the easiest way to find further bugs is to let it out into the wild and see what users can spot.
2018-09-24 20:28:16 +03:00
return conf_get_int(s->conf, CONF_agentfwd) && agent_exists();
}
static void ssh1_connection_special_cmd(PacketProtocolLayer *ppl,
SessionSpecialCode code, int arg)
{
struct ssh1_connection_state *s =
Rename FROMFIELD to 'container_of'. Ian Jackson points out that the Linux kernel has a macro of this name with the same purpose, and suggests that it's a good idea to use the same name as they do, so that at least some people reading one code base might recognise it from the other. I never really thought very hard about what order FROMFIELD's parameters should go in, and therefore I'm pleasantly surprised to find that my order agrees with the kernel's, so I don't have to permute every call site as part of making this change :-)
2018-10-06 01:49:08 +03:00
container_of(ppl, struct ssh1_connection_state, ppl);
Move most of ssh.c out into separate source files. I've tried to separate out as many individually coherent changes from this work as I could into their own commits, but here's where I run out and have to commit the rest of this major refactoring as a big-bang change. Most of ssh.c is now no longer in ssh.c: all five of the main coroutines that handle layers of the SSH-1 and SSH-2 protocols now each have their own source file to live in, and a lot of the supporting functions have moved into the appropriate one of those too. The new abstraction is a vtable called 'PacketProtocolLayer', which has an input and output packet queue. Each layer's main coroutine is invoked from the method ssh_ppl_process_queue(), which is usually (though not exclusively) triggered automatically when things are pushed on the input queue. In SSH-2, the base layer is the transport protocol, and it contains a pair of subsidiary queues by which it passes some of its packets to the higher SSH-2 layers - first userauth and then connection, which are peers at the same level, with the former abdicating in favour of the latter at the appropriate moment. SSH-1 is simpler: the whole login phase of the protocol (crypto setup and authentication) is all in one module, and since SSH-1 has no repeat key exchange, that setup layer abdicates in favour of the connection phase when it's done. ssh.c itself is now about a tenth of its old size (which all by itself is cause for celebration!). Its main job is to set up all the layers, hook them up to each other and to the BPP, and to funnel data back and forth between that collection of modules and external things such as the network and the terminal. Once it's set up a collection of packet protocol layers, it communicates with them partly by calling methods of the base layer (and if that's ssh2transport then it will delegate some functionality to the corresponding methods of its higher layer), and partly by talking directly to the connection layer no matter where it is in the stack by means of the separate ConnectionLayer vtable which I introduced in commit 8001dd4cb, and to which I've now added quite a few extra methods replacing services that used to be internal function calls within ssh.c. (One effect of this is that the SSH-1 and SSH-2 channel storage is now no longer shared - there are distinct struct types ssh1_channel and ssh2_channel. That means a bit more code duplication, but on the plus side, a lot fewer confusing conditionals in the middle of half-shared functions, and less risk of a piece of SSH-1 escaping into SSH-2 or vice versa, which I remember has happened at least once in the past.) The bulk of this commit introduces the five new source files, their common header sshppl.h and some shared supporting routines in sshcommon.c, and rewrites nearly all of ssh.c itself. But it also includes a couple of other changes that I couldn't separate easily enough: Firstly, there's a new handling for socket EOF, in which ssh.c sets an 'input_eof' flag in the BPP, and that responds by checking a flag that tells it whether to report the EOF as an error or not. (This is the main reason for those new BPP_READ / BPP_WAITFOR macros - they can check the EOF flag every time the coroutine is resumed.) Secondly, the error reporting itself is changed around again. I'd expected to put some data fields in the public PacketProtocolLayer structure that it could set to report errors in the same way as the BPPs have been doing, but in the end, I decided propagating all those data fields around was a pain and that even the BPPs shouldn't have been doing it that way. So I've reverted to a system where everything calls back to functions in ssh.c itself to report any connection- ending condition. But there's a new family of those functions, categorising the possible such conditions by semantics, and each one has a different set of detailed effects (e.g. how rudely to close the network connection, what exit status should be passed back to the whole application, whether to send a disconnect message and/or display a GUI error box). I don't expect this to be immediately perfect: of course, the code has been through a big upheaval, new bugs are expected, and I haven't been able to do a full job of testing (e.g. I haven't tested every auth or kex method). But I've checked that it _basically_ works - both SSH protocols, all the different kinds of forwarding channel, more than one auth method, Windows and Linux, connection sharing - and I think it's now at the point where the easiest way to find further bugs is to let it out into the wild and see what users can spot.
2018-09-24 20:28:16 +03:00
PktOut *pktout;
if (code == SS_PING || code == SS_NOP) {
if (!(s->ppl.remote_bugs & BUG_CHOKES_ON_SSH1_IGNORE)) {
pktout = ssh_bpp_new_pktout(s->ppl.bpp, SSH1_MSG_IGNORE);
put_stringz(pktout, "");
pq_push(s->ppl.out_pq, pktout);
}
} else if (code == SS_EOF) {
if (!s->session_ready) {
/*
* Buffer the EOF to send as soon as the main session is
* fully set up.
*/
s->session_eof_pending = TRUE;
} else if (!s->session_eof_sent) {
pktout = ssh_bpp_new_pktout(s->ppl.bpp, SSH1_CMSG_EOF);
pq_push(s->ppl.out_pq, pktout);
ppl_logevent(("Sent EOF message"));
s->session_eof_sent = TRUE;
}
}
}
static void ssh1_terminal_size(ConnectionLayer *cl, int width, int height)
{
struct ssh1_connection_state *s =
Rename FROMFIELD to 'container_of'. Ian Jackson points out that the Linux kernel has a macro of this name with the same purpose, and suggests that it's a good idea to use the same name as they do, so that at least some people reading one code base might recognise it from the other. I never really thought very hard about what order FROMFIELD's parameters should go in, and therefore I'm pleasantly surprised to find that my order agrees with the kernel's, so I don't have to permute every call site as part of making this change :-)
2018-10-06 01:49:08 +03:00
container_of(cl, struct ssh1_connection_state, cl);
Move most of ssh.c out into separate source files. I've tried to separate out as many individually coherent changes from this work as I could into their own commits, but here's where I run out and have to commit the rest of this major refactoring as a big-bang change. Most of ssh.c is now no longer in ssh.c: all five of the main coroutines that handle layers of the SSH-1 and SSH-2 protocols now each have their own source file to live in, and a lot of the supporting functions have moved into the appropriate one of those too. The new abstraction is a vtable called 'PacketProtocolLayer', which has an input and output packet queue. Each layer's main coroutine is invoked from the method ssh_ppl_process_queue(), which is usually (though not exclusively) triggered automatically when things are pushed on the input queue. In SSH-2, the base layer is the transport protocol, and it contains a pair of subsidiary queues by which it passes some of its packets to the higher SSH-2 layers - first userauth and then connection, which are peers at the same level, with the former abdicating in favour of the latter at the appropriate moment. SSH-1 is simpler: the whole login phase of the protocol (crypto setup and authentication) is all in one module, and since SSH-1 has no repeat key exchange, that setup layer abdicates in favour of the connection phase when it's done. ssh.c itself is now about a tenth of its old size (which all by itself is cause for celebration!). Its main job is to set up all the layers, hook them up to each other and to the BPP, and to funnel data back and forth between that collection of modules and external things such as the network and the terminal. Once it's set up a collection of packet protocol layers, it communicates with them partly by calling methods of the base layer (and if that's ssh2transport then it will delegate some functionality to the corresponding methods of its higher layer), and partly by talking directly to the connection layer no matter where it is in the stack by means of the separate ConnectionLayer vtable which I introduced in commit 8001dd4cb, and to which I've now added quite a few extra methods replacing services that used to be internal function calls within ssh.c. (One effect of this is that the SSH-1 and SSH-2 channel storage is now no longer shared - there are distinct struct types ssh1_channel and ssh2_channel. That means a bit more code duplication, but on the plus side, a lot fewer confusing conditionals in the middle of half-shared functions, and less risk of a piece of SSH-1 escaping into SSH-2 or vice versa, which I remember has happened at least once in the past.) The bulk of this commit introduces the five new source files, their common header sshppl.h and some shared supporting routines in sshcommon.c, and rewrites nearly all of ssh.c itself. But it also includes a couple of other changes that I couldn't separate easily enough: Firstly, there's a new handling for socket EOF, in which ssh.c sets an 'input_eof' flag in the BPP, and that responds by checking a flag that tells it whether to report the EOF as an error or not. (This is the main reason for those new BPP_READ / BPP_WAITFOR macros - they can check the EOF flag every time the coroutine is resumed.) Secondly, the error reporting itself is changed around again. I'd expected to put some data fields in the public PacketProtocolLayer structure that it could set to report errors in the same way as the BPPs have been doing, but in the end, I decided propagating all those data fields around was a pain and that even the BPPs shouldn't have been doing it that way. So I've reverted to a system where everything calls back to functions in ssh.c itself to report any connection- ending condition. But there's a new family of those functions, categorising the possible such conditions by semantics, and each one has a different set of detailed effects (e.g. how rudely to close the network connection, what exit status should be passed back to the whole application, whether to send a disconnect message and/or display a GUI error box). I don't expect this to be immediately perfect: of course, the code has been through a big upheaval, new bugs are expected, and I haven't been able to do a full job of testing (e.g. I haven't tested every auth or kex method). But I've checked that it _basically_ works - both SSH protocols, all the different kinds of forwarding channel, more than one auth method, Windows and Linux, connection sharing - and I think it's now at the point where the easiest way to find further bugs is to let it out into the wild and see what users can spot.
2018-09-24 20:28:16 +03:00
s->term_width = width;
s->term_height = height;
if (s->session_ready) {
PktOut *pktout = ssh_bpp_new_pktout(s->ppl.bpp, SSH1_CMSG_WINDOW_SIZE);
put_uint32(pktout, s->term_height);
put_uint32(pktout, s->term_width);
put_uint32(pktout, 0);
put_uint32(pktout, 0);
pq_push(s->ppl.out_pq, pktout);
}
}
static void ssh1_stdout_unthrottle(ConnectionLayer *cl, int bufsize)
{
struct ssh1_connection_state *s =
Rename FROMFIELD to 'container_of'. Ian Jackson points out that the Linux kernel has a macro of this name with the same purpose, and suggests that it's a good idea to use the same name as they do, so that at least some people reading one code base might recognise it from the other. I never really thought very hard about what order FROMFIELD's parameters should go in, and therefore I'm pleasantly surprised to find that my order agrees with the kernel's, so I don't have to permute every call site as part of making this change :-)
2018-10-06 01:49:08 +03:00
container_of(cl, struct ssh1_connection_state, cl);
Move most of ssh.c out into separate source files. I've tried to separate out as many individually coherent changes from this work as I could into their own commits, but here's where I run out and have to commit the rest of this major refactoring as a big-bang change. Most of ssh.c is now no longer in ssh.c: all five of the main coroutines that handle layers of the SSH-1 and SSH-2 protocols now each have their own source file to live in, and a lot of the supporting functions have moved into the appropriate one of those too. The new abstraction is a vtable called 'PacketProtocolLayer', which has an input and output packet queue. Each layer's main coroutine is invoked from the method ssh_ppl_process_queue(), which is usually (though not exclusively) triggered automatically when things are pushed on the input queue. In SSH-2, the base layer is the transport protocol, and it contains a pair of subsidiary queues by which it passes some of its packets to the higher SSH-2 layers - first userauth and then connection, which are peers at the same level, with the former abdicating in favour of the latter at the appropriate moment. SSH-1 is simpler: the whole login phase of the protocol (crypto setup and authentication) is all in one module, and since SSH-1 has no repeat key exchange, that setup layer abdicates in favour of the connection phase when it's done. ssh.c itself is now about a tenth of its old size (which all by itself is cause for celebration!). Its main job is to set up all the layers, hook them up to each other and to the BPP, and to funnel data back and forth between that collection of modules and external things such as the network and the terminal. Once it's set up a collection of packet protocol layers, it communicates with them partly by calling methods of the base layer (and if that's ssh2transport then it will delegate some functionality to the corresponding methods of its higher layer), and partly by talking directly to the connection layer no matter where it is in the stack by means of the separate ConnectionLayer vtable which I introduced in commit 8001dd4cb, and to which I've now added quite a few extra methods replacing services that used to be internal function calls within ssh.c. (One effect of this is that the SSH-1 and SSH-2 channel storage is now no longer shared - there are distinct struct types ssh1_channel and ssh2_channel. That means a bit more code duplication, but on the plus side, a lot fewer confusing conditionals in the middle of half-shared functions, and less risk of a piece of SSH-1 escaping into SSH-2 or vice versa, which I remember has happened at least once in the past.) The bulk of this commit introduces the five new source files, their common header sshppl.h and some shared supporting routines in sshcommon.c, and rewrites nearly all of ssh.c itself. But it also includes a couple of other changes that I couldn't separate easily enough: Firstly, there's a new handling for socket EOF, in which ssh.c sets an 'input_eof' flag in the BPP, and that responds by checking a flag that tells it whether to report the EOF as an error or not. (This is the main reason for those new BPP_READ / BPP_WAITFOR macros - they can check the EOF flag every time the coroutine is resumed.) Secondly, the error reporting itself is changed around again. I'd expected to put some data fields in the public PacketProtocolLayer structure that it could set to report errors in the same way as the BPPs have been doing, but in the end, I decided propagating all those data fields around was a pain and that even the BPPs shouldn't have been doing it that way. So I've reverted to a system where everything calls back to functions in ssh.c itself to report any connection- ending condition. But there's a new family of those functions, categorising the possible such conditions by semantics, and each one has a different set of detailed effects (e.g. how rudely to close the network connection, what exit status should be passed back to the whole application, whether to send a disconnect message and/or display a GUI error box). I don't expect this to be immediately perfect: of course, the code has been through a big upheaval, new bugs are expected, and I haven't been able to do a full job of testing (e.g. I haven't tested every auth or kex method). But I've checked that it _basically_ works - both SSH protocols, all the different kinds of forwarding channel, more than one auth method, Windows and Linux, connection sharing - and I think it's now at the point where the easiest way to find further bugs is to let it out into the wild and see what users can spot.
2018-09-24 20:28:16 +03:00
if (s->stdout_throttling && bufsize < SSH1_BUFFER_LIMIT) {
s->stdout_throttling = 0;
ssh_throttle_conn(s->ppl.ssh, -1);
}
}
static int ssh1_stdin_backlog(ConnectionLayer *cl)
{
return 0;
}
static void ssh1_throttle_all_channels(ConnectionLayer *cl, int throttled)
{
struct ssh1_connection_state *s =
Rename FROMFIELD to 'container_of'. Ian Jackson points out that the Linux kernel has a macro of this name with the same purpose, and suggests that it's a good idea to use the same name as they do, so that at least some people reading one code base might recognise it from the other. I never really thought very hard about what order FROMFIELD's parameters should go in, and therefore I'm pleasantly surprised to find that my order agrees with the kernel's, so I don't have to permute every call site as part of making this change :-)
2018-10-06 01:49:08 +03:00
container_of(cl, struct ssh1_connection_state, cl);
Move most of ssh.c out into separate source files. I've tried to separate out as many individually coherent changes from this work as I could into their own commits, but here's where I run out and have to commit the rest of this major refactoring as a big-bang change. Most of ssh.c is now no longer in ssh.c: all five of the main coroutines that handle layers of the SSH-1 and SSH-2 protocols now each have their own source file to live in, and a lot of the supporting functions have moved into the appropriate one of those too. The new abstraction is a vtable called 'PacketProtocolLayer', which has an input and output packet queue. Each layer's main coroutine is invoked from the method ssh_ppl_process_queue(), which is usually (though not exclusively) triggered automatically when things are pushed on the input queue. In SSH-2, the base layer is the transport protocol, and it contains a pair of subsidiary queues by which it passes some of its packets to the higher SSH-2 layers - first userauth and then connection, which are peers at the same level, with the former abdicating in favour of the latter at the appropriate moment. SSH-1 is simpler: the whole login phase of the protocol (crypto setup and authentication) is all in one module, and since SSH-1 has no repeat key exchange, that setup layer abdicates in favour of the connection phase when it's done. ssh.c itself is now about a tenth of its old size (which all by itself is cause for celebration!). Its main job is to set up all the layers, hook them up to each other and to the BPP, and to funnel data back and forth between that collection of modules and external things such as the network and the terminal. Once it's set up a collection of packet protocol layers, it communicates with them partly by calling methods of the base layer (and if that's ssh2transport then it will delegate some functionality to the corresponding methods of its higher layer), and partly by talking directly to the connection layer no matter where it is in the stack by means of the separate ConnectionLayer vtable which I introduced in commit 8001dd4cb, and to which I've now added quite a few extra methods replacing services that used to be internal function calls within ssh.c. (One effect of this is that the SSH-1 and SSH-2 channel storage is now no longer shared - there are distinct struct types ssh1_channel and ssh2_channel. That means a bit more code duplication, but on the plus side, a lot fewer confusing conditionals in the middle of half-shared functions, and less risk of a piece of SSH-1 escaping into SSH-2 or vice versa, which I remember has happened at least once in the past.) The bulk of this commit introduces the five new source files, their common header sshppl.h and some shared supporting routines in sshcommon.c, and rewrites nearly all of ssh.c itself. But it also includes a couple of other changes that I couldn't separate easily enough: Firstly, there's a new handling for socket EOF, in which ssh.c sets an 'input_eof' flag in the BPP, and that responds by checking a flag that tells it whether to report the EOF as an error or not. (This is the main reason for those new BPP_READ / BPP_WAITFOR macros - they can check the EOF flag every time the coroutine is resumed.) Secondly, the error reporting itself is changed around again. I'd expected to put some data fields in the public PacketProtocolLayer structure that it could set to report errors in the same way as the BPPs have been doing, but in the end, I decided propagating all those data fields around was a pain and that even the BPPs shouldn't have been doing it that way. So I've reverted to a system where everything calls back to functions in ssh.c itself to report any connection- ending condition. But there's a new family of those functions, categorising the possible such conditions by semantics, and each one has a different set of detailed effects (e.g. how rudely to close the network connection, what exit status should be passed back to the whole application, whether to send a disconnect message and/or display a GUI error box). I don't expect this to be immediately perfect: of course, the code has been through a big upheaval, new bugs are expected, and I haven't been able to do a full job of testing (e.g. I haven't tested every auth or kex method). But I've checked that it _basically_ works - both SSH protocols, all the different kinds of forwarding channel, more than one auth method, Windows and Linux, connection sharing - and I think it's now at the point where the easiest way to find further bugs is to let it out into the wild and see what users can spot.
2018-09-24 20:28:16 +03:00
struct ssh1_channel *c;
int i;
for (i = 0; NULL != (c = index234(s->channels, i)); i++)
chan_set_input_wanted(c->chan, !throttled);
}
static int ssh1_ldisc_option(ConnectionLayer *cl, int option)
{
struct ssh1_connection_state *s =
Rename FROMFIELD to 'container_of'. Ian Jackson points out that the Linux kernel has a macro of this name with the same purpose, and suggests that it's a good idea to use the same name as they do, so that at least some people reading one code base might recognise it from the other. I never really thought very hard about what order FROMFIELD's parameters should go in, and therefore I'm pleasantly surprised to find that my order agrees with the kernel's, so I don't have to permute every call site as part of making this change :-)
2018-10-06 01:49:08 +03:00
container_of(cl, struct ssh1_connection_state, cl);
Move most of ssh.c out into separate source files. I've tried to separate out as many individually coherent changes from this work as I could into their own commits, but here's where I run out and have to commit the rest of this major refactoring as a big-bang change. Most of ssh.c is now no longer in ssh.c: all five of the main coroutines that handle layers of the SSH-1 and SSH-2 protocols now each have their own source file to live in, and a lot of the supporting functions have moved into the appropriate one of those too. The new abstraction is a vtable called 'PacketProtocolLayer', which has an input and output packet queue. Each layer's main coroutine is invoked from the method ssh_ppl_process_queue(), which is usually (though not exclusively) triggered automatically when things are pushed on the input queue. In SSH-2, the base layer is the transport protocol, and it contains a pair of subsidiary queues by which it passes some of its packets to the higher SSH-2 layers - first userauth and then connection, which are peers at the same level, with the former abdicating in favour of the latter at the appropriate moment. SSH-1 is simpler: the whole login phase of the protocol (crypto setup and authentication) is all in one module, and since SSH-1 has no repeat key exchange, that setup layer abdicates in favour of the connection phase when it's done. ssh.c itself is now about a tenth of its old size (which all by itself is cause for celebration!). Its main job is to set up all the layers, hook them up to each other and to the BPP, and to funnel data back and forth between that collection of modules and external things such as the network and the terminal. Once it's set up a collection of packet protocol layers, it communicates with them partly by calling methods of the base layer (and if that's ssh2transport then it will delegate some functionality to the corresponding methods of its higher layer), and partly by talking directly to the connection layer no matter where it is in the stack by means of the separate ConnectionLayer vtable which I introduced in commit 8001dd4cb, and to which I've now added quite a few extra methods replacing services that used to be internal function calls within ssh.c. (One effect of this is that the SSH-1 and SSH-2 channel storage is now no longer shared - there are distinct struct types ssh1_channel and ssh2_channel. That means a bit more code duplication, but on the plus side, a lot fewer confusing conditionals in the middle of half-shared functions, and less risk of a piece of SSH-1 escaping into SSH-2 or vice versa, which I remember has happened at least once in the past.) The bulk of this commit introduces the five new source files, their common header sshppl.h and some shared supporting routines in sshcommon.c, and rewrites nearly all of ssh.c itself. But it also includes a couple of other changes that I couldn't separate easily enough: Firstly, there's a new handling for socket EOF, in which ssh.c sets an 'input_eof' flag in the BPP, and that responds by checking a flag that tells it whether to report the EOF as an error or not. (This is the main reason for those new BPP_READ / BPP_WAITFOR macros - they can check the EOF flag every time the coroutine is resumed.) Secondly, the error reporting itself is changed around again. I'd expected to put some data fields in the public PacketProtocolLayer structure that it could set to report errors in the same way as the BPPs have been doing, but in the end, I decided propagating all those data fields around was a pain and that even the BPPs shouldn't have been doing it that way. So I've reverted to a system where everything calls back to functions in ssh.c itself to report any connection- ending condition. But there's a new family of those functions, categorising the possible such conditions by semantics, and each one has a different set of detailed effects (e.g. how rudely to close the network connection, what exit status should be passed back to the whole application, whether to send a disconnect message and/or display a GUI error box). I don't expect this to be immediately perfect: of course, the code has been through a big upheaval, new bugs are expected, and I haven't been able to do a full job of testing (e.g. I haven't tested every auth or kex method). But I've checked that it _basically_ works - both SSH protocols, all the different kinds of forwarding channel, more than one auth method, Windows and Linux, connection sharing - and I think it's now at the point where the easiest way to find further bugs is to let it out into the wild and see what users can spot.
2018-09-24 20:28:16 +03:00
/* We always return the same value for LD_ECHO and LD_EDIT */
return s->echoedit;
}
static int ssh1_connection_want_user_input(PacketProtocolLayer *ppl)
{
struct ssh1_connection_state *s =
Rename FROMFIELD to 'container_of'. Ian Jackson points out that the Linux kernel has a macro of this name with the same purpose, and suggests that it's a good idea to use the same name as they do, so that at least some people reading one code base might recognise it from the other. I never really thought very hard about what order FROMFIELD's parameters should go in, and therefore I'm pleasantly surprised to find that my order agrees with the kernel's, so I don't have to permute every call site as part of making this change :-)
2018-10-06 01:49:08 +03:00
container_of(ppl, struct ssh1_connection_state, ppl);
Move most of ssh.c out into separate source files. I've tried to separate out as many individually coherent changes from this work as I could into their own commits, but here's where I run out and have to commit the rest of this major refactoring as a big-bang change. Most of ssh.c is now no longer in ssh.c: all five of the main coroutines that handle layers of the SSH-1 and SSH-2 protocols now each have their own source file to live in, and a lot of the supporting functions have moved into the appropriate one of those too. The new abstraction is a vtable called 'PacketProtocolLayer', which has an input and output packet queue. Each layer's main coroutine is invoked from the method ssh_ppl_process_queue(), which is usually (though not exclusively) triggered automatically when things are pushed on the input queue. In SSH-2, the base layer is the transport protocol, and it contains a pair of subsidiary queues by which it passes some of its packets to the higher SSH-2 layers - first userauth and then connection, which are peers at the same level, with the former abdicating in favour of the latter at the appropriate moment. SSH-1 is simpler: the whole login phase of the protocol (crypto setup and authentication) is all in one module, and since SSH-1 has no repeat key exchange, that setup layer abdicates in favour of the connection phase when it's done. ssh.c itself is now about a tenth of its old size (which all by itself is cause for celebration!). Its main job is to set up all the layers, hook them up to each other and to the BPP, and to funnel data back and forth between that collection of modules and external things such as the network and the terminal. Once it's set up a collection of packet protocol layers, it communicates with them partly by calling methods of the base layer (and if that's ssh2transport then it will delegate some functionality to the corresponding methods of its higher layer), and partly by talking directly to the connection layer no matter where it is in the stack by means of the separate ConnectionLayer vtable which I introduced in commit 8001dd4cb, and to which I've now added quite a few extra methods replacing services that used to be internal function calls within ssh.c. (One effect of this is that the SSH-1 and SSH-2 channel storage is now no longer shared - there are distinct struct types ssh1_channel and ssh2_channel. That means a bit more code duplication, but on the plus side, a lot fewer confusing conditionals in the middle of half-shared functions, and less risk of a piece of SSH-1 escaping into SSH-2 or vice versa, which I remember has happened at least once in the past.) The bulk of this commit introduces the five new source files, their common header sshppl.h and some shared supporting routines in sshcommon.c, and rewrites nearly all of ssh.c itself. But it also includes a couple of other changes that I couldn't separate easily enough: Firstly, there's a new handling for socket EOF, in which ssh.c sets an 'input_eof' flag in the BPP, and that responds by checking a flag that tells it whether to report the EOF as an error or not. (This is the main reason for those new BPP_READ / BPP_WAITFOR macros - they can check the EOF flag every time the coroutine is resumed.) Secondly, the error reporting itself is changed around again. I'd expected to put some data fields in the public PacketProtocolLayer structure that it could set to report errors in the same way as the BPPs have been doing, but in the end, I decided propagating all those data fields around was a pain and that even the BPPs shouldn't have been doing it that way. So I've reverted to a system where everything calls back to functions in ssh.c itself to report any connection- ending condition. But there's a new family of those functions, categorising the possible such conditions by semantics, and each one has a different set of detailed effects (e.g. how rudely to close the network connection, what exit status should be passed back to the whole application, whether to send a disconnect message and/or display a GUI error box). I don't expect this to be immediately perfect: of course, the code has been through a big upheaval, new bugs are expected, and I haven't been able to do a full job of testing (e.g. I haven't tested every auth or kex method). But I've checked that it _basically_ works - both SSH protocols, all the different kinds of forwarding channel, more than one auth method, Windows and Linux, connection sharing - and I think it's now at the point where the easiest way to find further bugs is to let it out into the wild and see what users can spot.
2018-09-24 20:28:16 +03:00
return s->session_ready && !s->session_eof_sent;
}
static void ssh1_connection_got_user_input(PacketProtocolLayer *ppl)
{
struct ssh1_connection_state *s =
Rename FROMFIELD to 'container_of'. Ian Jackson points out that the Linux kernel has a macro of this name with the same purpose, and suggests that it's a good idea to use the same name as they do, so that at least some people reading one code base might recognise it from the other. I never really thought very hard about what order FROMFIELD's parameters should go in, and therefore I'm pleasantly surprised to find that my order agrees with the kernel's, so I don't have to permute every call site as part of making this change :-)
2018-10-06 01:49:08 +03:00
container_of(ppl, struct ssh1_connection_state, ppl);
Move most of ssh.c out into separate source files. I've tried to separate out as many individually coherent changes from this work as I could into their own commits, but here's where I run out and have to commit the rest of this major refactoring as a big-bang change. Most of ssh.c is now no longer in ssh.c: all five of the main coroutines that handle layers of the SSH-1 and SSH-2 protocols now each have their own source file to live in, and a lot of the supporting functions have moved into the appropriate one of those too. The new abstraction is a vtable called 'PacketProtocolLayer', which has an input and output packet queue. Each layer's main coroutine is invoked from the method ssh_ppl_process_queue(), which is usually (though not exclusively) triggered automatically when things are pushed on the input queue. In SSH-2, the base layer is the transport protocol, and it contains a pair of subsidiary queues by which it passes some of its packets to the higher SSH-2 layers - first userauth and then connection, which are peers at the same level, with the former abdicating in favour of the latter at the appropriate moment. SSH-1 is simpler: the whole login phase of the protocol (crypto setup and authentication) is all in one module, and since SSH-1 has no repeat key exchange, that setup layer abdicates in favour of the connection phase when it's done. ssh.c itself is now about a tenth of its old size (which all by itself is cause for celebration!). Its main job is to set up all the layers, hook them up to each other and to the BPP, and to funnel data back and forth between that collection of modules and external things such as the network and the terminal. Once it's set up a collection of packet protocol layers, it communicates with them partly by calling methods of the base layer (and if that's ssh2transport then it will delegate some functionality to the corresponding methods of its higher layer), and partly by talking directly to the connection layer no matter where it is in the stack by means of the separate ConnectionLayer vtable which I introduced in commit 8001dd4cb, and to which I've now added quite a few extra methods replacing services that used to be internal function calls within ssh.c. (One effect of this is that the SSH-1 and SSH-2 channel storage is now no longer shared - there are distinct struct types ssh1_channel and ssh2_channel. That means a bit more code duplication, but on the plus side, a lot fewer confusing conditionals in the middle of half-shared functions, and less risk of a piece of SSH-1 escaping into SSH-2 or vice versa, which I remember has happened at least once in the past.) The bulk of this commit introduces the five new source files, their common header sshppl.h and some shared supporting routines in sshcommon.c, and rewrites nearly all of ssh.c itself. But it also includes a couple of other changes that I couldn't separate easily enough: Firstly, there's a new handling for socket EOF, in which ssh.c sets an 'input_eof' flag in the BPP, and that responds by checking a flag that tells it whether to report the EOF as an error or not. (This is the main reason for those new BPP_READ / BPP_WAITFOR macros - they can check the EOF flag every time the coroutine is resumed.) Secondly, the error reporting itself is changed around again. I'd expected to put some data fields in the public PacketProtocolLayer structure that it could set to report errors in the same way as the BPPs have been doing, but in the end, I decided propagating all those data fields around was a pain and that even the BPPs shouldn't have been doing it that way. So I've reverted to a system where everything calls back to functions in ssh.c itself to report any connection- ending condition. But there's a new family of those functions, categorising the possible such conditions by semantics, and each one has a different set of detailed effects (e.g. how rudely to close the network connection, what exit status should be passed back to the whole application, whether to send a disconnect message and/or display a GUI error box). I don't expect this to be immediately perfect: of course, the code has been through a big upheaval, new bugs are expected, and I haven't been able to do a full job of testing (e.g. I haven't tested every auth or kex method). But I've checked that it _basically_ works - both SSH protocols, all the different kinds of forwarding channel, more than one auth method, Windows and Linux, connection sharing - and I think it's now at the point where the easiest way to find further bugs is to let it out into the wild and see what users can spot.
2018-09-24 20:28:16 +03:00
if (s->session_ready && !s->session_eof_sent)
queue_idempotent_callback(&s->ppl.ic_process_queue);
}
static void ssh1_connection_reconfigure(PacketProtocolLayer *ppl, Conf *conf)
{
struct ssh1_connection_state *s =
Rename FROMFIELD to 'container_of'. Ian Jackson points out that the Linux kernel has a macro of this name with the same purpose, and suggests that it's a good idea to use the same name as they do, so that at least some people reading one code base might recognise it from the other. I never really thought very hard about what order FROMFIELD's parameters should go in, and therefore I'm pleasantly surprised to find that my order agrees with the kernel's, so I don't have to permute every call site as part of making this change :-)
2018-10-06 01:49:08 +03:00
container_of(ppl, struct ssh1_connection_state, ppl);
Move most of ssh.c out into separate source files. I've tried to separate out as many individually coherent changes from this work as I could into their own commits, but here's where I run out and have to commit the rest of this major refactoring as a big-bang change. Most of ssh.c is now no longer in ssh.c: all five of the main coroutines that handle layers of the SSH-1 and SSH-2 protocols now each have their own source file to live in, and a lot of the supporting functions have moved into the appropriate one of those too. The new abstraction is a vtable called 'PacketProtocolLayer', which has an input and output packet queue. Each layer's main coroutine is invoked from the method ssh_ppl_process_queue(), which is usually (though not exclusively) triggered automatically when things are pushed on the input queue. In SSH-2, the base layer is the transport protocol, and it contains a pair of subsidiary queues by which it passes some of its packets to the higher SSH-2 layers - first userauth and then connection, which are peers at the same level, with the former abdicating in favour of the latter at the appropriate moment. SSH-1 is simpler: the whole login phase of the protocol (crypto setup and authentication) is all in one module, and since SSH-1 has no repeat key exchange, that setup layer abdicates in favour of the connection phase when it's done. ssh.c itself is now about a tenth of its old size (which all by itself is cause for celebration!). Its main job is to set up all the layers, hook them up to each other and to the BPP, and to funnel data back and forth between that collection of modules and external things such as the network and the terminal. Once it's set up a collection of packet protocol layers, it communicates with them partly by calling methods of the base layer (and if that's ssh2transport then it will delegate some functionality to the corresponding methods of its higher layer), and partly by talking directly to the connection layer no matter where it is in the stack by means of the separate ConnectionLayer vtable which I introduced in commit 8001dd4cb, and to which I've now added quite a few extra methods replacing services that used to be internal function calls within ssh.c. (One effect of this is that the SSH-1 and SSH-2 channel storage is now no longer shared - there are distinct struct types ssh1_channel and ssh2_channel. That means a bit more code duplication, but on the plus side, a lot fewer confusing conditionals in the middle of half-shared functions, and less risk of a piece of SSH-1 escaping into SSH-2 or vice versa, which I remember has happened at least once in the past.) The bulk of this commit introduces the five new source files, their common header sshppl.h and some shared supporting routines in sshcommon.c, and rewrites nearly all of ssh.c itself. But it also includes a couple of other changes that I couldn't separate easily enough: Firstly, there's a new handling for socket EOF, in which ssh.c sets an 'input_eof' flag in the BPP, and that responds by checking a flag that tells it whether to report the EOF as an error or not. (This is the main reason for those new BPP_READ / BPP_WAITFOR macros - they can check the EOF flag every time the coroutine is resumed.) Secondly, the error reporting itself is changed around again. I'd expected to put some data fields in the public PacketProtocolLayer structure that it could set to report errors in the same way as the BPPs have been doing, but in the end, I decided propagating all those data fields around was a pain and that even the BPPs shouldn't have been doing it that way. So I've reverted to a system where everything calls back to functions in ssh.c itself to report any connection- ending condition. But there's a new family of those functions, categorising the possible such conditions by semantics, and each one has a different set of detailed effects (e.g. how rudely to close the network connection, what exit status should be passed back to the whole application, whether to send a disconnect message and/or display a GUI error box). I don't expect this to be immediately perfect: of course, the code has been through a big upheaval, new bugs are expected, and I haven't been able to do a full job of testing (e.g. I haven't tested every auth or kex method). But I've checked that it _basically_ works - both SSH protocols, all the different kinds of forwarding channel, more than one auth method, Windows and Linux, connection sharing - and I think it's now at the point where the easiest way to find further bugs is to let it out into the wild and see what users can spot.
2018-09-24 20:28:16 +03:00
conf_free(s->conf);
s->conf = conf_copy(conf);
if (s->portfwdmgr_configured)
portfwdmgr_config(s->portfwdmgr, s->conf);
}