commit 903f9db10f18f735e62ba447147b6c434b6af003 upstream.
syzbot is reporting kernel panic [1] triggered by memory allocation failure
at tty_ldisc_get() from tty_ldisc_init(). But since both tty_ldisc_get()
and caller of tty_ldisc_init() can cleanly handle errors, tty_ldisc_init()
does not need to call panic() when tty_ldisc_get() failed.
[1] https://syzkaller.appspot.com/bug?id=883431818e036ae6a9981156a64b821110f39187
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Reported-by: syzbot <syzkaller@googlegroups.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Jiri Slaby <jslaby@suse.com>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 28b0f8a6962a24ed21737578f3b1b07424635c9e upstream.
A tty is hung up by __tty_hangup() setting file->f_op to
hung_up_tty_fops, which is skipped on ttys whose write operation isn't
tty_write(). This means that, for example, /dev/console whose write
op is redirected_tty_write() is never actually marked hung up.
Because n_tty_read() uses the hung up status to decide whether to
abort the waiting readers, the lack of hung-up marking can lead to the
following scenario.
1. A session contains two processes. The leader and its child. The
child ignores SIGHUP.
2. The leader exits and starts disassociating from the controlling
terminal (/dev/console).
3. __tty_hangup() skips setting f_op to hung_up_tty_fops.
4. SIGHUP is delivered and ignored.
5. tty_ldisc_hangup() is invoked. It wakes up the waits which should
clear the read lockers of tty->ldisc_sem.
6. The reader wakes up but because tty_hung_up_p() is false, it
doesn't abort and goes back to sleep while read-holding
tty->ldisc_sem.
7. The leader progresses to tty_ldisc_lock() in tty_ldisc_hangup()
and is now stuck in D sleep indefinitely waiting for
tty->ldisc_sem.
The following is Alan's explanation on why some ttys aren't hung up.
http://lkml.kernel.org/r/20171101170908.6ad08580@alans-desktop
1. It broke the serial consoles because they would hang up and close
down the hardware. With tty_port that *should* be fixable properly
for any cases remaining.
2. The console layer was (and still is) completely broken and doens't
refcount properly. So if you turn on console hangups it breaks (as
indeed does freeing consoles and half a dozen other things).
As neither can be fixed quickly, this patch works around the problem
by introducing a new flag, TTY_HUPPING, which is used solely to tell
n_tty_read() that hang-up is in progress for the console and the
readers should be aborted regardless of the hung-up status of the
device.
The following is a sample hung task warning caused by this issue.
INFO: task agetty:2662 blocked for more than 120 seconds.
Not tainted 4.11.3-dbg-tty-lockup-02478-gfd6c7ee-dirty #28
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
0 2662 1 0x00000086
Call Trace:
__schedule+0x267/0x890
schedule+0x36/0x80
schedule_timeout+0x23c/0x2e0
ldsem_down_write+0xce/0x1f6
tty_ldisc_lock+0x16/0x30
tty_ldisc_hangup+0xb3/0x1b0
__tty_hangup+0x300/0x410
disassociate_ctty+0x6c/0x290
do_exit+0x7ef/0xb00
do_group_exit+0x3f/0xa0
get_signal+0x1b3/0x5d0
do_signal+0x28/0x660
exit_to_usermode_loop+0x46/0x86
do_syscall_64+0x9c/0xb0
entry_SYSCALL64_slow_path+0x25/0x25
The following is the repro. Run "$PROG /dev/console". The parent
process hangs in D state.
#include <sys/types.h>
#include <sys/stat.h>
#include <sys/wait.h>
#include <sys/ioctl.h>
#include <fcntl.h>
#include <unistd.h>
#include <stdio.h>
#include <stdlib.h>
#include <errno.h>
#include <signal.h>
#include <time.h>
#include <termios.h>
int main(int argc, char **argv)
{
struct sigaction sact = { .sa_handler = SIG_IGN };
struct timespec ts1s = { .tv_sec = 1 };
pid_t pid;
int fd;
if (argc < 2) {
fprintf(stderr, "test-hung-tty /dev/$TTY\n");
return 1;
}
/* fork a child to ensure that it isn't already the session leader */
pid = fork();
if (pid < 0) {
perror("fork");
return 1;
}
if (pid > 0) {
/* top parent, wait for everyone */
while (waitpid(-1, NULL, 0) >= 0)
;
if (errno != ECHILD)
perror("waitpid");
return 0;
}
/* new session, start a new session and set the controlling tty */
if (setsid() < 0) {
perror("setsid");
return 1;
}
fd = open(argv[1], O_RDWR);
if (fd < 0) {
perror("open");
return 1;
}
if (ioctl(fd, TIOCSCTTY, 1) < 0) {
perror("ioctl");
return 1;
}
/* fork a child, sleep a bit and exit */
pid = fork();
if (pid < 0) {
perror("fork");
return 1;
}
if (pid > 0) {
nanosleep(&ts1s, NULL);
printf("Session leader exiting\n");
exit(0);
}
/*
* The child ignores SIGHUP and keeps reading from the controlling
* tty. Because SIGHUP is ignored, the child doesn't get killed on
* parent exit and the bug in n_tty makes the read(2) block the
* parent's control terminal hangup attempt. The parent ends up in
* D sleep until the child is explicitly killed.
*/
sigaction(SIGHUP, &sact, NULL);
printf("Child reading tty\n");
while (1) {
char buf[1024];
if (read(fd, buf, sizeof(buf)) < 0) {
perror("read");
return 1;
}
}
return 0;
}
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Alan Cox <alan@llwyncelyn.cymru>
Cc: stable@vger.kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit 2b022ab7542df60021ab57854b3faaaf42552eaf ]
In case that CONFIG_SLUB_DEBUG is on and pty is used, races between
release_one_tty and flush_to_ldisc work threads may happen and lead
to use-after-free condition on tty->link->port. Because SLUB_DEBUG
is turned on, freed tty->link->port is filled with POISON_FREE value.
So far without SLUB_DEBUG, port was filled with zero and flush_to_ldisc
could return without a problem by checking if tty is NULL.
CPU 0 CPU 1
----- -----
release_tty pty_write
cancel_work_sync(tty) to = tty->link
tty_kref_put(tty->link) tty_schedule_flip(to->port)
<< workqueue >> ...
release_one_tty ...
pty_cleanup ...
kfree(tty->link->port) << workqueue >>
flush_to_ldisc
tty = READ_ONCE(port->itty)
tty is 0x6b6b6b6b6b6b6b6b
!!PANIC!! access tty->ldisc
Unable to handle kernel paging request at virtual address 6b6b6b6b6b6b6b93
pgd = ffffffc0eb1c3000
[6b6b6b6b6b6b6b93] *pgd=0000000000000000, *pud=0000000000000000
------------[ cut here ]------------
Kernel BUG at ffffff800851154c [verbose debug info unavailable]
Internal error: Oops - BUG: 96000004 [#1] PREEMPT SMP
CPU: 3 PID: 265 Comm: kworker/u8:9 Tainted: G W 3.18.31-g0a58eeb #1
Hardware name: Qualcomm Technologies, Inc. MSM 8996pro v1.1 + PMI8996 Carbide (DT)
Workqueue: events_unbound flush_to_ldisc
task: ffffffc0ed610ec0 ti: ffffffc0ed624000 task.ti: ffffffc0ed624000
PC is at ldsem_down_read_trylock+0x0/0x4c
LR is at tty_ldisc_ref+0x24/0x4c
pc : [<ffffff800851154c>] lr : [<ffffff800850f6c0>] pstate: 80400145
sp : ffffffc0ed627cd0
x29: ffffffc0ed627cd0 x28: 0000000000000000
x27: ffffff8009e05000 x26: ffffffc0d382cfa0
x25: 0000000000000000 x24: ffffff800a012f08
x23: 0000000000000000 x22: ffffffc0703fbc88
x21: 6b6b6b6b6b6b6b6b x20: 6b6b6b6b6b6b6b93
x19: 0000000000000000 x18: 0000000000000001
x17: 00e80000f80d6f53 x16: 0000000000000001
x15: 0000007f7d826fff x14: 00000000000000a0
x13: 0000000000000000 x12: 0000000000000109
x11: 0000000000000000 x10: 0000000000000000
x9 : ffffffc0ed624000 x8 : ffffffc0ed611580
x7 : 0000000000000000 x6 : ffffff800a42e000
x5 : 00000000000003fc x4 : 0000000003bd1201
x3 : 0000000000000001 x2 : 0000000000000001
x1 : ffffff800851004c x0 : 6b6b6b6b6b6b6b93
Signed-off-by: Sahara <keun-o.park@darkmatter.ae>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit e9036d0662360cd4c79578565ce422ed5872f301 upstream.
When the tty lock is interrupted on attempted re-open, 2 tty krefs
are still held. Drop extra kref before returning failure from
tty_lock_interruptible(), and drop lookup kref before returning
failure from tty_open().
Fixes: 0bfd464d3fdd ("tty: Wait interruptibly for tty lock on reopen")
Reported-by: Dmitry Vyukov <dvyukov@google.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Cc: Jiri Slaby <jslaby@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 5c17c861a357e9458001f021a7afa7aab9937439 upstream.
ioctl(TIOCGETD) retrieves the line discipline id directly from the
ldisc because the line discipline id (c_line) in termios is untrustworthy;
userspace may have set termios via ioctl(TCSETS*) without actually
changing the line discipline via ioctl(TIOCSETD).
However, directly accessing the current ldisc via tty->ldisc is
unsafe; the ldisc ptr dereferenced may be stale if the line discipline
is changing via ioctl(TIOCSETD) or hangup.
Wait for the line discipline reference (just like read() or write())
to retrieve the "current" line discipline id.
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 7f22f6c935cda600660e623a411fe380015d28d9 upstream.
A small window exists where a tty reopen will observe the tty
just prior to imminent teardown (tty->count == 0); in this case, open()
returns EIO to userspace.
Instead, retry the open after checking for signals and yielding;
this interruptible retry loop allows teardown to commence and initialize
a new tty on retry. Never retry the BSD master pty reopen; there is no
guarantee the pty pair teardown is imminent since the slave file
descriptors may remain open indefinitely.
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 0bfd464d3fdd5bb322f9cace4cc47f1796545cf7 upstream.
Allow a signal to interrupt the wait for a tty reopen; eg., if
the tty has starting final close and is waiting for the device to
drain.
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The correct lock order is atomic_write_lock => termios_rwsem, as
established by tty_write() => n_tty_write().
Fixes: c274f6ef1c ("tty: Hold termios_rwsem for tcflow(TCIxxx)")
Reported-and-Tested-by: Dmitry Vyukov <dvyukov@google.com>
Cc: <stable@vger.kernel.org> # v3.18+
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Introduce API functions to restart and cancel tty buffer work, rather
than manipulate buffer work directly.
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
tty_write_message() allows the caller to directly write to a specific
tty. Since the line discipline is bypassed for the direct write,
nothing prevents the tty from being torn down after the tty count is
checked.
Hold the tty lock for the duration of the direct write.
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
tiocspgrp() is the ioctl handler for TIOCSPGRP, which runs in
non-atomic context; use spin_lock/unlock_irq (since interrupt state
is on).
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The job_control() check in n_tty_read() has nearly identical purpose
and results as tty_check_change(). Both functions' purpose is to
determine if the current task's pgrp is the foreground pgrp for the tty,
and if not, to signal the current pgrp.
Introduce __tty_check_change() which takes the signal to send
and performs the shared operations for job control() and
tty_check_change().
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This is mostly a hardening fix, given that write-only access to other
users' ttys is usually only given through setgid tty executables.
Signed-off-by: Jann Horn <jann@thejh.net>
Cc: stable@vger.kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit a3a10ce342 ("Avoid usb reset crashes by making tty_io cdevs truly
dynamic") which mixes using cdev_alloc() and cdev_init() is problematic.
Subsequent call to cdev_init() after cdev_alloc() sets kobj release method
from cdev_dynamic_release() to cdev_default_release() and thus makes it
impossible to free allocated cdev.
This patch also consolidates error path of cdev_add() as cdev can also leak
here if things went wrong.
Signed-off-by: Leon Yu <chianglungyu@gmail.com>
Fixes: a3a10ce342 ("Avoid usb reset crashes by making tty_io cdevs truly dynamic")
Acked-by: Richard Watts <rrw@kynesim.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Introduce tty_debug() macro to output uniform debug information for
tty core debug messages (function name and tty name).
Note: printk(KERN_DEBUG) is retained here over pr_debug() since
messages can be enabled in non-DEBUG builds.
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Output the function name, tty name, and invariant failure (if applicable).
Add the tty count to the tty_open() message. Fix the disassociate_ctty()
message, which printed the NULL pointer and the wrong message.
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
task_pgrp requires an rcu or tasklist lock to be obtained if the returned pid
is to be dereferenced, which kill_pgrp does. Obtain an RCU lock for the
duration of use.
Signed-off-by: Patrick Donnelly <batrick@batbytes.com>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
tty_name no longer uses the buf parameter, so remove it along with all
the 64 byte stack buffers that used to be passed in.
Mostly generated by the coccinelle script
@depends on patch@
identifier buf;
constant C;
expression tty;
@@
- char buf[C];
<+...
- tty_name(tty, buf)
+ tty_name(tty)
...+>
allmodconfig compiles, so I'm fairly confident the stack buffers
weren't used for other purposes as well.
Signed-off-by: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Acked-by: Jesper Nilsson <jesper.nilsson@axis.com>
Acked-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
All users of tty_name pass the return value (the provided buffer) to
some printf-like function. We can thus avoid the strcpy and, more
importantly, later remove the buf parameter completely, eliminating
the need for some 64 byte stack buffers.
Signed-off-by: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
All users of tty_name pass the result directly to a printf-like
function. This means we can actually let tty_name return the literal
"NULL tty" or tty->name directly, avoiding the strcpy and a lot of
medium-sized stack buffers. In preparation for that, make the return
type const char*.
While at it, we can also constify the tty parameter.
Signed-off-by: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We only care if anything other than the lower 3 bits of the tty has
changed, so just check that way, which makes it a bit faster, and more
obvious what is going on. Also, document this for future developers to
understand why we did this.
Reported-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org
Instead of manual calls of device_create_file() and
device_remove_file(), pass the static attribute groups using
device_create_with_groups().
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This problem was taken care of three times already in
* b0de59b573 (TTY: do not update
atime/mtime on read/write),
* 37b7f3c765 (TTY: fix atime/mtime
regression), and
* b0b885657b (tty: fix up atime/mtime
mess, take three)
But it still misses one point. As John Paul correctly points out, we
do not care about setting date. If somebody ever changes wall
time backwards (by mistake for example), tty timestamps are never
updated until the original wall time passes.
So check the absolute difference of times and if it large than "8
seconds or so", always update the time. That means we will update
immediatelly when changing time. Ergo, CAP_SYS_TIME can foul the
check, but it was always that way.
Thanks John for serving me this so nicely debugged.
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Reported-by: John Paul Perry <john_paul.perry@alcatel-lucent.com>
Cc: <stable@vger.kernel.org> # all, as b0b885657 was backported
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Exclusive mode ttys (TTY_EXCLUSIVE) do not allow further reopens;
fail the condition before associating the file pointer and calling
the driver open() method.
Prevents DTR programming when the tty is already in exclusive mode.
Reported-by: Shreyas Bethur <shreyas.bethur@ni.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Acked-by: Shreyas Bethur <shreyas.bethur@ni.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The functions put_device() and tty_kref_put() test whether their argument
is NULL and then return immediately.
Thus the test around the call is not needed.
This issue was detected by using the Coccinelle software.
Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When somebody calls TIOCSSERIAL ioctl with serial flags to set one of
* ASYNC_SESSION_LOCKOUT
* ASYNC_PGRP_LOCKOUT
* ASYNC_CALLOUT_NOHUP
* ASYNC_AUTOPROBE
nothing happens. We actually ignore the flags for over a decade at
least (I checked 2.6.0).
So start yelling at users who use those flags, that they shouldn't.
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Cc: Peter Hurley <peter@hurleysoftware.com>
Cc: Alan Cox <gnomes@lxorguk.ukuu.org.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The use of older function ptr calling style, (*fn)(), makes static
analysis more error-prone; replace with modern fn() style.
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
tty_hung_up_p() is equivalent to the open-coded test in tty_open().
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We need the fixes in drivers/tty/tty_io.c that were done in there for
future patches in this branch.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
tty_ldisc_flush() first clears the line discipline input buffer,
then clears the tty flip buffers. However, this allows for existing
data in the tty flip buffers to be added after the ldisc input
buffer has been cleared, but before the flip buffers have been cleared.
Add an optional ldisc parameter to tty_buffer_flush() to allow
tty_ldisc_flush() to pass the ldisc to clear.
NB: Initially, the plan was to do this automatically in
tty_buffer_flush(). However, an audit of the behavior of existing
line disciplines showed that performing a ldisc buffer flush on
ioctl(TCFLSH) was not always the outcome. For example, some line
disciplines have flush_buffer() methods but not ioctl() methods,
so a ->flush_buffer() command would be unexpected.
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When releasing the master pty, the slave pty also needs to be locked
to prevent concurrent tty count changes for the slave pty and to
ensure that only one parallel master and slave release observe the
final close, and proceed to destruct the pty pair. Conversely, when
releasing the slave pty, locking the master pty is not necessary
(since the master's state can be inferred by the slave tty count).
Introduce tty_lock_slave()/tty_unlock_slave() which acquires/releases
the tty lock of the slave pty. Remove tty_lock_pair()/tty_unlock_pair().
Dropping the tty_lock is no longer required to re-establish a stable
lock order.
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The local o_tty variable in tty_release() is now accessed only
when closing the pty master.
Set o_tty to slave pty when closing pty master, otherwise NULL;
use o_tty != NULL as replacement for pty_master.
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Passing the 'other' tty to tty_release_checks() only makes sense
for a pty pair; make o_tty scope local instead.
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Passing the 'other' tty to tty_ldisc_release() only makes sense
for a pty pair; make o_tty function local instead.
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Perform work flush for both ends of a pty pair within tty_flush_works(),
rather than calling twice.
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When the slave side closes and its tty count is 0, the pty
pair can be destroyed; the master side must have already
closed for the slave side tty count to be 0. Thus, only the
pty master close must check if the slave side has closed by
checking the slave tty count.
Remove the pre-computed closing flags and check the actual count(s).
Regular ttys are unaffected by this change.
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Holding the tty_lock() is necessary to prevent concurrent changes
to the tty count that may cause it to differ from the open file
list count. The tty_lock() is already held at all call sites.
NB: Note that the check for the pty master tty count is safe because
the slave's tty_lock() is held while decrementing the pty master
tty count.
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Releasing the tty locks while waiting for the tty wait queues to
be empty is no longer necessary nor desirable. Prior to
"tty: Don't take tty_mutex for tty count changes", dropping the
tty locks was necessary to reestablish the correct lock order between
tty_mutex and the tty locks. Dropping the global tty_mutex was necessary;
otherwise new ttys could not have been opened while waiting.
However, without needing the global tty_mutex held, the tty locks for
the releasing tty can now be held through the sleep. The sanity check
is for abnormal conditions caused by kernel bugs, not for recoverable
errors caused by misbehaving userspace; dropping the tty locks only
allows the tty state to get more sideways.
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Holding tty_mutex is no longer required to serialize changes to
the tty_count or to prevent concurrent opens of closing ttys;
tty_lock() is sufficient.
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Now that re-open is not permitted for a legacy BSD pty master,
using TTY_CLOSING to indicate when a tty can be torn-down is
no longer necessary.
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Holding tty_mutex for a tty re-open is no longer necessary since
"tty: Clarify re-open behavior of master ptys". Because the
slave tty count is no longer accessed by tty_reopen(), holding
tty_mutex to prevent concurrent final tty_release() of the slave
pty is not required.
As with "tty: Re-open /dev/tty without tty_mutex", holding a
tty kref until the tty_lock is acquired is sufficient to ensure
the tty has not been freed, which, in turn, is sufficient to
ensure the tty_lock can be safely acquired and the tty count
can be safely retrieved. A non-zero tty count with the tty lock
held guarantees that release_tty() has not run and cannot
run concurrently with tty_reopen().
Change tty_driver_lookup_tty() to acquire the tty kref, which
allows the tty_mutex to be dropped before acquiring the tty lock.
Dropping the tty_mutex before attempting the tty_lock allows
other ttys to be opened and released, without needing this
tty_reopen() to complete.
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Opening /dev/tty (ie., the controlling tty for the current task)
is always a re-open of the underlying tty. Because holding the
tty_lock is sufficient for safely re-opening a tty, and because
having a tty kref is sufficient for safely acquiring the tty_lock [1],
tty_open_current_tty() does not require holding tty_mutex.
Repurpose tty_open_current_tty() to perform the re-open itself and
refactor tty_open().
[1] Analysis of safely re-opening the current tty w/o tty_mutex
get_current_tty() gets a tty kref from the already kref'ed tty value of
current->signal->tty while holding the sighand lock for the current
task. This guarantees that the tty pointer returned from
get_current_tty() points to a tty which remains referenceable
while holding the kref.
Although release_tty() may run concurrently, and thus the driver
reference may be removed, release_one_tty() cannot have run, and
won't while holding the tty kref.
This, in turn, guarantees the tty_lock() can safely be acquired
(since tty->magic and tty->legacy_mutex are still a valid dereferences).
The tty_lock() also gets a tty kref to prevent the tty_unlock() from
dereferencing a released tty. Thus, the kref returned from
get_current_tty() can be released.
Lastly, the first operation of tty_reopen() is to check the tty count.
If non-zero, this ensures release_tty() is not running concurrently,
and the driver references have not been removed.
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Although perhaps not obvious, the TTY_CLOSING bit is set when the
tty count has been decremented to 0 (which occurs while holding
tty_lock). The only other case when tty count is 0 during a re-open
is when a legacy BSD pty master has been opened in parallel but
after the pty slave, which is unsupported and returns an error.
Thus !tty->count contains the complete set of degenerate conditions
under which a tty open fails.
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Re-opening master ptys is not allowed. Once opened and for the remaining
lifetime of the master pty, its tty count is 1. If its tty count has
dropped to 0, then the master pty was closed and TTY_CLOSING was set,
and destruction may begin imminently.
Besides the normal case of a legacy BSD pty master being re-opened
(which always returns -EIO), this code is only reachable in 2 degenerate
cases:
1. The pty master is the controlling terminal (this is possible through
the TIOCSCTTY ioctl). pty masters are not designed to be controlling
terminals and it's an oversight that tiocsctty() ever let that happen.
The attempted open of /dev/tty will always fail. No known program does
this.
2. The legacy BSD pty slave was opened first. The slave open will fail
in pty_open() and tty_release() will commence. But before tty_release()
claims the tty_mutex, there is a very small window where a parallel
master open might succeed. In a test of racing legacy BSD slave and
master parallel opens, where:
slave open attempts: 10000 success:4527 failure:5473
master open attempts: 11728 success:5789 failure:5939
only 8 master open attempts would have succeeded reaching this code and
successfully opened the master pty. This case is not possible with
SysV ptys.
Always return -EIO if a master pty is re-opened or the slave is opened
first and the master opened in parallel (for legacy BSD ptys).
Furthermore, now that changing the slave's count is not required,
the tty_lock is sufficient for preventing concurrent changes to the
tty being re-opened (or failing re-opening).
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Now that tty_ldisc_hangup() does not drop the tty lock, it is no
longer possible to observe TTY_HUPPING while holding the tty lock
on another cpu.
Remove TTY_HUPPING bit definition.
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The session and foreground process group pid references will be
non-NULL if tiocsctty() is stealing the controlling tty from another
session (ie., arg == 1 in tiocsctty()).
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Setting the controlling terminal for a session occurs with either
the first open of a non-pty master tty or with ioctl(TIOCSCTTY).
Since only the session leader can set the controlling terminal for
a session (and the session leader cannot change), it is not
necessary to prevent a process from attempting to set different
ttys as the controlling terminal concurrently.
So it's only necessary to prevent the same tty from becoming the
controlling terminal for different session leaders. The tty_lock()
is sufficient to prevent concurrent proc_set_tty() for the same
tty.
Remove the tty_mutex lock region; add tty_lock() to tiocsctty().
While this may appear to allow a race condition between opening
the controlling tty via tty_open_current_tty() and stealing the
controlling tty via ioctl(TIOCSCTTY, 1), that race condition already
existed. Even if the tty_mutex prevented stealing the controlling tty
while tty_open_current_tty() returned the original controlling tty,
it cannot prevent stealing the controlling tty before tty_open() returns.
Thus, tty_open() could already return a no-longer-controlling tty when
opening /dev/tty.
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
tiocspgrp() is the lone caller of session_of_pgrp(); relocate and
limit to file scope.
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
