Commit Graph

156 Commits

Author SHA1 Message Date
nicm
c913fb99b6 Tidy the code that works out the socket path, and just use the full path
in the global socket_path rather than copying it.
2015-11-24 22:27:22 +00:00
nicm
bef217b241 Switch a fprintf to a fatal, and wrap some long lines. 2015-11-24 22:04:36 +00:00
nicm
577c0e3e5a Use __unused rather than rolling our own. 2015-11-18 14:27:44 +00:00
nicm
a582b62287 Accidentally turned off pledge, turn it back on. 2015-11-15 14:32:48 +00:00
nicm
c56b81a2ce Push stdout and stderr to clients more aggressively, and add an event to
continue if the send fails.
2015-11-14 09:41:06 +00:00
nicm
b0a99e85b6 Don't shift version out of peerid, it is needed later. 2015-10-31 13:43:38 +00:00
nicm
01defc9f49 Because pledge(2) does not allow us to pass directory file descriptors
around, we can't use file descriptors for the working directory because
we will be unable to pass it to a privileged process to tell it where to
read or write files or spawn children. So move tmux back to using
strings for the current working directory. We try to check it exists
with access() when it is set but ultimately fall back to ~ if it fails
at time of use (or / if that fails too).
2015-10-31 08:13:58 +00:00
nicm
bf9c933cae Like options, move the environ struct into environ.c. 2015-10-28 09:51:55 +00:00
nicm
44657bf932 Move struct options into options.c. 2015-10-27 15:58:42 +00:00
nicm
07b0ea03c3 Break the common process set up, event loop and imsg dispatch code
between server and client out into a separate internal API. This will
make it easier to add another process.
2015-10-27 13:23:24 +00:00
nicm
174a2ad731 Pass current directory as a string rather than a file descriptor because
pledge doesn't let us pass directory file descriptors.
2015-10-18 20:42:42 +00:00
nicm
9c601ebde8 Add pledge "stdio unix sendfd proc exec tty" to tmux client process,
"sendfd" is dropped after first message from the server.
2015-10-17 18:30:43 +00:00
guenther
241fd72f75 Userspace doesn't need to use SUN_LEN(): connect() and bind() must accept
sizeof(struct sockaddr_un), so do the simple, portable thing

ok beck@ deraadt@
2015-10-11 00:26:23 +00:00
nicm
2a62917444 Don't leak fd and path on failure. 2015-09-24 12:06:20 +00:00
nicm
67ee995cc1 No need to keep global options around for client which doesn't use them. 2015-09-09 12:09:21 +00:00
nicm
c6e9160c67 Login shell can be a client flag, and move the exec code into client.c. 2015-08-30 22:40:25 +00:00
nicm
dd92b6e83d Event base does not need to be global. 2015-08-30 22:19:07 +00:00
nicm
4e637b1b61 Ignore environment variables that are too long to send to the server. 2015-07-13 18:10:26 +00:00
nicm
29c29e7717 Add a format for client PID (client_pid) and server PID (pid). Diff for
client_pid from Thomas Adam.
2015-06-14 10:07:44 +00:00
nicm
6b2129696f Move the nested check from client to server and compare the client tty
name to all the pane pty names instead of comparing socket paths. This
means that "new -d" will work without unsetting $TMUX.
2015-06-04 23:27:51 +00:00
nicm
a5a873dccc Set up signal handler earlier so that we don't get zombies, reported by
sobrado@.
2015-04-24 20:58:44 +00:00
nicm
7a72eff4a4 Simplify error messages when socket connect fails, suggested by "Karthik K". 2015-04-21 22:21:41 +00:00
nicm
02df86079b Fix some format specifier nits, from Ben Boeckel. 2015-03-31 17:45:10 +00:00
nicm
900f6fc17e Tidy up some includes. 2014-10-20 23:27:14 +00:00
nicm
045d0c3b9f Call waitpid on SIGCHLD even if client not attached, it is possible (on
very slow platforms) for the first daemon() child to still be running
when client_attached is set so we end up with a zombie. From J Raynor.
2014-10-01 23:23:19 +00:00
nicm
4e956d545a Various minor style and spacing nits. 2014-09-01 21:50:18 +00:00
nicm
8e4ae12b4d lockf is entirely useless and it was a mistake to change to it, go back
to using flock which actually works sensibly. Also always retry the lock
to fix a potential race, and add some extra logging.
2014-07-21 10:52:48 +00:00
krw
2b67907176 An EOF is a good reason to close a connection.
ok nicm@
2014-07-13 20:51:08 +00:00
nicm
5b2c8156d5 fcntl.h is still needed here. 2014-04-29 22:31:22 +00:00
nicm
bce952777a Remove some unnecessary includes and fix a typo. 2014-04-17 11:38:35 +00:00
nicm
b185449d07 Fix a memory/fd leak reported by Tiago Cunha. 2014-01-09 21:20:45 +00:00
nicm
adc1f21eae Three small changes from Tiago Cunha:
- Check for truncation when copying path.
- Don't need to use a temporary buffer in screen_set_title.
- Include strerror in output when connecting to server fails.
2014-01-09 14:05:55 +00:00
benno
7624800ddc from nicm: : handle msgbuf_write() returning EAGAIN 2013-11-13 20:43:36 +00:00
nicm
99e3cbc526 Use format_get_command() and some spacing tweaks. 2013-10-10 12:35:30 +00:00
nicm
c1ccefc62d We accidentally haven't been using $TMUX to work out the session for a
while and in fact it is less useful that using the client ttyname. So
don't bother and don't pass it from the client. If we need it in future
it is in c->environ.
2013-10-10 12:29:35 +00:00
nicm
7936ce3874 Show session name in detached message. Requested by somebody a few
months ago who didn't bother testing it. But it works for me anyway.
2013-10-10 12:28:08 +00:00
nicm
b8b85fbb0c Don't look at string[length - 1] if length == 0. 2013-10-10 12:27:38 +00:00
nicm
282c5f9644 Alter how tmux handles the working directory to internally use file
descriptors rather than strings.

- Each session still has a current working directory.

- New sessions still get their working directory from the client that
  created them or its attached session if any.

- New windows are created by default in the session working directory.

- The -c flag to new, neww, splitw allows the working directory to be
  overridden.

- The -c flag to attach let's the session working directory be changed.

- The default-path option has been removed.

To get the equivalent to default-path '.', do:

        bind c neww -c $PWD

To get the equivalent of default-path '~', do:

        bind c neww -c ~

This also changes the client identify protocol to be a set of messages rather
than one as well as some other changes that should make it easier to make
backwards-compatible protocol changes in future.
2013-10-10 12:26:34 +00:00
nicm
10c38436aa Similarly for MSG_COMMAND - allow full imsg limit not arbitrary 2048. 2013-10-10 12:13:56 +00:00
nicm
6c093010e0 Remove CMD_SENDENVIRON. 2013-10-10 12:09:34 +00:00
Nicholas Marriott
d89b35e682 Use lockf which is more portable than flock, from Dagobert Michelsen. 2013-04-22 08:42:19 +00:00
Nicholas Marriott
88b92df849 We ignore SIGWINCH until ready, so send a MSG_RESIZE immediately when
becoming ready.
2013-03-25 11:38:57 +00:00
Nicholas Marriott
6fee3e9e4b Rename session idx to session id throughout and add $ prefix to targets
to use it, extended from a diff from George Nachman.
2013-03-25 10:11:45 +00:00
Nicholas Marriott
62db3c8efe Send DSC 1000p at the beginning of a -CC client's lifetime and ST and
the end, from George Nachman.
2013-03-25 10:03:24 +00:00
Nicholas Marriott
20636d956d Add a command queue to standardize and simplify commands that call other
commands and allow a command to block execution of subsequent
commands. This allows run-shell and if-shell to be synchronous which has
been much requested.

Each client has a default command queue and commands are consumed one at
a time from it. A command may suspend execution from the queue by
returning CMD_RETURN_WAIT and then resume it by calling cmd_continue() -
for example run-shell does this from the callback that is fired after
the job is freed.

When the command queue becomes empty, command clients are automatically
exited (unless attaching). A callback is also fired - this is used for
nested commands in, for example, if-shell which can block execution of
the client's cmdq until a new cmdq becomes empty.

Also merge all the old error/info/print functions together and lose the
old curclient/cmdclient distinction - a cmdq is bound to one client (or
none if in the configuration file), this is a command client if
c->session is NULL otherwise an attached client.
2013-03-24 09:54:10 +00:00
Nicholas Marriott
8d59b189cc No more lint means no more ARGSUSED. 2013-03-22 10:31:22 +00:00
Nicholas Marriott
2e43372d0b If stdin in the client is enable immediately, tmux will eat anything
sent to stdin before it is needed, which can be inconvenient (eg pasting
commands). Instead, start with stdin disabled and reuse MSG_STDIN from
server->client to mean that stdin should be enabled. Based on a diff
from Chris Johnsen.
2012-09-03 08:48:57 +00:00
Nicholas Marriott
84a4007b06 Change a log to fprintf that was missed last time around, from Tiago Cunha. 2012-08-27 21:35:11 +00:00
Nicholas Marriott
df912e3540 xfree is not particularly helpful, remove it. From Thomas Adam. 2012-07-10 11:53:01 +00:00
Nicholas Marriott
0f00c3b47a Actually write all the data to stdout/stderr. 2012-06-18 13:34:56 +00:00
Nicholas Marriott
2942eca895 Add a skeleton mode to tmux (called "control mode") that let's tmux
commands be sent and output received on stdout. This can be used to
integrate with other terminal emulators and should allow some other
things to be made simpler later. More to come so doesn't do much yet and
deliberately not documented.
2012-06-18 13:16:42 +00:00
Nicholas Marriott
196710e2d3 Simplify logging and just fprintf(stderr, ...) for early errors. 2012-05-25 08:28:10 +00:00
Nicholas Marriott
7a4679a17f Instead of passing stdin/stdout/stderr file descriptors over imsg and
handling them in the server, handle them in the client and pass buffers
over imsg. This is much tidier for some upcoming changes and the
performance hit isn't critical.

The tty fd is still passed to the server as before.

This bumps the tmux protocol version so new clients and old servers are
incompatible.
2012-05-21 18:27:42 +00:00
Nicholas Marriott
94795eb48e Add missing prototype. 2012-04-23 23:25:02 +00:00
Nicholas Marriott
d3e432069b Use an enum for client exit reasons, from George Nachman. 2012-04-23 23:18:43 +00:00
Nicholas Marriott
0edaa34bfd Add sys/file.h. 2012-03-24 11:23:19 +00:00
Nicholas Marriott
bf9e7a1c68 Move MSG_IDENTIFY to the last sent by the client, this will be needed by
control clients and irrelevant for others.
2012-03-19 08:42:06 +00:00
Nicholas Marriott
ac9ebc29a2 Use a lock file and flock() to serialize server start, avoids problems
when running a bunch of tmux from cron at the same time. Based on a diff
from Tim Ruehsen.
2012-03-09 09:57:40 +00:00
Nicholas Marriott
7954126dcd Add a -P option to detach to HUP the client's parent process (usually
causing it to exit as well).
2011-03-03 08:51:47 +00:00
Nicholas Marriott
69cb1f830e Move all calls to fcntl(...O_NONBLOCK) into a function and clear the
flag on the stdio file descriptors before closing them (fixes things
like "tmux ls && cat").
2011-01-08 01:52:36 +00:00
Nicholas Marriott
248fb14f08 Merge the before and after attach client code into one in client.c
(instead of two in tmux.c and client.c).
2010-10-18 20:00:02 +00:00
Nicholas Marriott
f56b4ec2ff Trying to set FD_CLOEXEC on every fd is a lost cause, just use
closefrom() before exec.
2010-10-16 08:31:55 +00:00
Nicholas Marriott
fc9f08235b Can't call event_del() without event_set() first - so call event_set()
when setting up the client.
2010-08-23 17:36:32 +00:00
Nicholas Marriott
5f5104e782 MSG_EXIT can now have a return code in the message, so check for that
size as well. Stops the client fatal()ing on exit.
2010-08-22 16:09:49 +00:00
Nicholas Marriott
76bbdeb586 Send all three of stdin, stdout, stderr from the client to the server, so that
commands can directly make use of them. This means that load-buffer and
save-buffer can have "-" as the file to read from stdin or write to stdout.

This is a protocol version bump so the tmux server will need to be restarted
after upgrade (or an older client used).
2010-06-28 22:10:42 +00:00
Nicholas Marriott
510ec3fb9e Fix problems with window sizing seen by Raghavendra D Prabhu when
starting tmux from .xinitrc.

One of the very few things the server relies on the client for now is to
pass through a message on SIGWINCH, but there is a condition where
potentially a SIGWINCH may be lost during the transition from unattached
(main.c) to attached (client.c). So trigger a size change immediately
after the client installs its SIGWINCH handler.

Also, when the terminal is resized, reset the scroll region and cursor
position. Previously, we were clearing our saved idea of these, but in
fact some terminals do not reset them on resize, so this caused problems
during redraw.

While here make a resize to the same size not cause a redraw and rename
the tmux.out output log file to include the tmux PID.
2010-06-05 16:47:11 +00:00
Nicholas Marriott
36c0871c23 This ioctl(TIOCGWINSZ) call is no longer necessary, the result is never
used and the server now does it later on the tty fd directly.
2010-06-05 16:29:45 +00:00
Joel Sing
72d1afa169 Catch SIGHUP and terminate if running as a client. This prevents clients
from being left hanging around when, for example, a SSH session is
disconnected.

ok nicm@
2010-05-12 15:05:39 +00:00
Nicholas Marriott
c4a2fdf15b Put this back in with the initialisation in the right order. 2010-05-04 17:28:16 +00:00
Nicholas Marriott
af5e0bd15a Revert last change, it appears to be broken somehow. 2010-05-04 08:48:06 +00:00
Nicholas Marriott
ec1d37b1b2 Make signal handler setup/teardown two common functions instead of six,
and reset SIGCHLD after fork to fix problems with some shells. From
Romain Francois.
2010-05-03 16:06:32 +00:00
Nicholas Marriott
15a64b805e Massive spaces->tabs and trailing whitespace cleanup, hopefully for the last
time now I've configured emacs to make them displayed in really annoying
colours...
2009-12-03 22:50:09 +00:00
Nicholas Marriott
4ca857e0e9 Remove a couple of unused arguments where possible, and add /* ARGSUSED */ to
the rest to reduce lint output.
2009-11-26 21:37:13 +00:00
Nicholas Marriott
a18a374d51 imsg_read returns ssize_t not int, pointed out by lint via deraadt. 2009-11-13 18:07:52 +00:00
Nicholas Marriott
9f47c6083a Don't return 1 unless there was actually a problem (signal/lost server) rather
than for all events (normal exit/detach/etc).
2009-11-10 15:47:48 +00:00
Nicholas Marriott
862fe15c32 Move some common code into a function. 2009-11-04 22:57:49 +00:00
Nicholas Marriott
abf3a5d50e Initial changes to move tmux to libevent.
This moves the client-side loops are pretty much fully over to event-based only
(tmux.c and client.c) but server-side (server.c and friends) treats libevent as
a sort of clever poll, waking up after every event to run various things.

Moving the server stuff over to bufferevents and timers and so on will come
later.
2009-11-04 20:50:11 +00:00
Nicholas Marriott
2a585dc4ed Leftover unused variable :-/. 2009-11-02 13:42:25 +00:00
Nicholas Marriott
42fd44f1db There isn't much point in doing lstat before connect so instead just do connect
and handle ENOENT from it which is a little tidier.
2009-11-02 13:41:25 +00:00
Nicholas Marriott
a8b1379ccb Clear signal flags /before/ taking action and continue afterwards to reduce
chance of dropping signals. Pointed out by deraadt@.
2009-10-26 21:38:18 +00:00
Nicholas Marriott
284b94662a Tidy identify message send into a separate function. 2009-10-21 21:11:55 +00:00
Nicholas Marriott
90ad041fa5 Client tidying: get rid of client_ctx struct in favour of two variables in
client.c, and move the functions in client-fn.c into other files.
2009-10-21 20:11:47 +00:00
Nicholas Marriott
0907ca1931 Do this in a better way - print messages when exiting with nonzero.
Also remove the login shell information from server-info, only the client
should care about it.
2009-10-13 13:15:26 +00:00
Nicholas Marriott
760e39e405 Don't print exit messages when used as a login shell, requested by martynas@ a
while back.
2009-10-13 13:11:06 +00:00
Nicholas Marriott
9200a0be7a Support -c like sh(1) to execute a command, useful when tmux is a login
shell. Suggested by halex@.

This includes another protocol version increase (the last for now) so again
restart the tmux server before upgrading.
2009-09-23 12:03:30 +00:00
Nicholas Marriott
b01dcd7971 Remove the internal tmux locking and instead detach each client and run the
command specified by a new option "lock-command" (by default "lock -np") in
each client.

This means each terminal has to be unlocked individually but simplifies the
code and allows the system password to be used to unlock.

Note that the set-password command is gone, so it will need to be removed from
configuration files, and the -U command line flag has been removed.

This is the third protocol version change so again it is best to stop the tmux
server before upgrading.
2009-09-23 06:18:47 +00:00
Nicholas Marriott
962fa20b36 Trim some code by moving the ioctl(TIOCGWINSZ) after SIGWINCH from the client
into the server.

This is another (the second of four) protocol version changes coming this
morning, so again the server should be killed before upgrading.
2009-09-23 06:12:58 +00:00
Nicholas Marriott
64caf59e84 Don't attempt to open() the tty path, rely on the client sending its stdin fd
with imsg and fatal if it doesn't, then set the FD_CLOEXEC flag in tty_init
instead of tty_open to prevent them leaking into child processes if any are
created between the two calls.

This bumps the protocol version, so the tmux server should be killed before
upgrading.
2009-09-23 06:05:02 +00:00
Nicholas Marriott
273f1b385c Regularise some fatal messages. 2009-09-20 14:58:12 +00:00
Nicholas Marriott
751a2fa915 Fix a race condition when asking a client to take over the terminal (switching
to a different poll loop):

If a MSG_READY was followed very quickly by a MSG_EXIT (for example if doing
"tmux new 'exit'"), both messages could be read as part of the same imsg_read
in the first client poll loop. The MSG_READY would then cause a switch to the
second client loop, which would immediately call poll(2) again, causing the
client to hang forever waiting for an exit message that it already had.

Change to call imsg_get to process any existing messages before polling.
2009-09-02 23:49:25 +00:00
Nicholas Marriott
459abafcea That was the wrong fix. MSG_ERROR should set the error and the client should
use the error and exit on MSG_EXIT (it was being handled in the default
case). Undo the last change, move the errstr check into the MSG_EXIT case, and
add a comment.
2009-09-02 20:15:49 +00:00
Nicholas Marriott
7a4bac82d7 Set exittype for error exit as well as the error string. 2009-09-02 20:00:10 +00:00
Nicholas Marriott
85e8b70625 imsg closes the fd after sending, so dup() STDIN_FILENO before passing it to
the parent, otherwise TIOCGWINSZ will fail when the window is resized (that
could actually be moved into the server but this is more future-proof and
avoids breaking the protocol).
2009-08-12 06:04:28 +00:00
Nicholas Marriott
4310282a4c Have the client pass its stdin fd to the server when identifying itself and
have the server use that rather than reopening the tty. If the fd isn't given,
use the old behaviour (so no need for a version change).

This allows tmux to be used as the shell, so also change so that when working
out the command to execute if default-command is empty (the default), tmux will
try not execute itself.
2009-08-11 21:28:11 +00:00
Nicholas Marriott
f0635717b3 Switch tmux to use imsg. This is the last major change to make the
client-server protocol more resilient and make the protocol versioning work
properly. In future, the only things requiring a protocol version bump will be
changes in the message structs, and (when both client and server have this
change) mixing different versions should nicely report an error message.

As a side effect this also makes the code tidier, fixes a problem with the way
errors reported during server startup were handled, and supports fd passing
(which will be used in future).

Looked over by eric@, thanks.

Please note that mixing a client with this change with an older server or vice
versa may cause tmux to crash or hang - tmux should be completely exited before
upgrading.
2009-08-11 17:18:35 +00:00
Nicholas Marriott
6491274f60 Infrastructure and commands to manage the environment for processes started
within tmux.

There is a global environment, copied from the external environment when the
server is started and each sesssion has an (initially empty) session
environment which overrides it.

New commands set-environment and show-environment manipulate or display the
environments.

A new session option, update-environment, is a space-separated list of
variables which are updated from the external environment into the session
environment every time a new session is created - the default is DISPLAY.
2009-08-08 21:52:43 +00:00
Nicholas Marriott
e985629440 Tidy function a little by using a temporary variable. 2009-08-08 21:18:23 +00:00
Nicholas Marriott
5f13bb0c3a There aren't many client message types or code to handle them so get rid of the
lookup table and use a switch, merge the tiny handler functions into it, and
move the whole lot to client.c.

Also change client_msg_dispatch to consume as many messages as possible and
move the call to it to the right place so it checks for signals afterwards.

Prompted by suggestions from eric@.
2009-07-30 16:32:12 +00:00
Nicholas Marriott
479d614884 Tell the server when the client gets SIGTERM so it can clean up the terminal
properly, rather than just exiting.
2009-07-30 16:16:19 +00:00
Nicholas Marriott
34a82e7629 Make all messages sent between the client and server fixed size.
This is the first of two changes to make the protocol more resilient and less
sensitive to other changes in the code, particularly with commands. The client
now packs argv into a buffer and sends it to the server for parsing, rather
than doing it itself and sending the parsed command data.

As a side-effect this also removes a lot of now-unused command marshalling
code.

Mixing a server without this change and a client with or vice versa will cause
tmux to hang or crash, please ensure that tmux is entirely killed before
upgrading.
2009-07-26 12:58:44 +00:00