TTY/n_gsm: Removing the wrong tty_unlock/lock() in gsm_dlci_release()
authorChuansheng Liu <chuansheng.liu@intel.com>
Wed, 18 Dec 2013 05:30:11 +0000 (13:30 +0800)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Thu, 19 Dec 2013 00:58:28 +0000 (16:58 -0800)
commitbe7065725590dd8fbf5d6a614bcfaec4a860b998
tree0b174c4c46637a464980c487abd63d1d45e9a658
parent8be3da65d9222e1e63b7f9d001e872947588946f
TTY/n_gsm: Removing the wrong tty_unlock/lock() in gsm_dlci_release()

Commit 4d9b109060f690f5c835(tty: Prevent deadlock in n_gsm driver)
tried to close all the virtual ports synchronously before closing the
phycial ports, so the tty_vhangup() is used.

But the tty_unlock/lock() is wrong:
tty_release
 tty_ldisc_release
  tty_lock_pair(tty, o_tty)  < == Here the tty is for physical port
  tty_ldisc_kill
    gsmld_close
      gsm_cleanup_mux
        gsm_dlci_release
          tty = tty_port_tty_get(&dlci->port)
                            < == Here the tty(s) are for virtual port

They are different ttys, so before tty_vhangup(virtual tty), do not need
to call the tty_unlock(virtual tty) at all which causes unbalanced unlock
warning.

When enabling mutex debugging option, we will hit the below warning also:
[   99.276903] =====================================
[   99.282172] [ BUG: bad unlock balance detected! ]
[   99.287442] 3.10.20-261976-gaec5ba0 #44 Tainted: G           O
[   99.293972] -------------------------------------
[   99.299240] mmgr/152 is trying to release lock (&tty->legacy_mutex) at:
[   99.306693] [<c1b2dcad>] mutex_unlock+0xd/0x10
[   99.311669] but there are no more locks to release!
[   99.317131]
[   99.317131] other info that might help us debug this:
[   99.324440] 3 locks held by mmgr/152:
[   99.328542]  #0:  (&tty->legacy_mutex/1){......}, at: [<c1b30ab0>] tty_lock_nested+0x40/0x90
[   99.338116]  #1:  (&tty->ldisc_mutex){......}, at: [<c15dbd02>] tty_ldisc_kill+0x22/0xd0
[   99.347284]  #2:  (&gsm->mutex){......}, at: [<c15e3d83>] gsm_cleanup_mux+0x73/0x170
[   99.356060]
[   99.356060] stack backtrace:
[   99.360932] CPU: 0 PID: 152 Comm: mmgr Tainted: G           O 3.10.20-261976-gaec5ba0 #44
[   99.370086]  ef4a4de0 ef4a4de0 ef4c1d98 c1b27b91 ef4c1db8 c1292655 c1dd10f5 c1b2dcad
[   99.378921]  c1b2dcad ef4a4de0 ef4a528c ffffffff ef4c1dfc c12930dd 00000246 00000000
[   99.387754]  00000000 00000000 c15e1926 00000000 00000001 ddfa7530 00000003 c1b2dcad
[   99.396588] Call Trace:
[   99.399326]  [<c1b27b91>] dump_stack+0x16/0x18
[   99.404307]  [<c1292655>] print_unlock_imbalance_bug+0xe5/0xf0
[   99.410840]  [<c1b2dcad>] ? mutex_unlock+0xd/0x10
[   99.416110]  [<c1b2dcad>] ? mutex_unlock+0xd/0x10
[   99.421382]  [<c12930dd>] lock_release_non_nested+0x1cd/0x210
[   99.427818]  [<c15e1926>] ? gsm_destroy_network+0x36/0x130
[   99.433964]  [<c1b2dcad>] ? mutex_unlock+0xd/0x10
[   99.439235]  [<c12931a2>] lock_release+0x82/0x1c0
[   99.444505]  [<c1b2dcad>] ? mutex_unlock+0xd/0x10
[   99.449776]  [<c1b2dcad>] ? mutex_unlock+0xd/0x10
[   99.455047]  [<c1b2dc2f>] __mutex_unlock_slowpath+0x5f/0xd0
[   99.461288]  [<c1b2dcad>] mutex_unlock+0xd/0x10
[   99.466365]  [<c1b30bb1>] tty_unlock+0x21/0x50
[   99.471345]  [<c15e3dd1>] gsm_cleanup_mux+0xc1/0x170
[   99.476906]  [<c15e44d2>] gsmld_close+0x52/0x90
[   99.481983]  [<c15db905>] tty_ldisc_close.isra.1+0x35/0x50
[   99.488127]  [<c15dbd0c>] tty_ldisc_kill+0x2c/0xd0
[   99.493494]  [<c15dc7af>] tty_ldisc_release+0x2f/0x50
[   99.499152]  [<c15d572c>] tty_release+0x37c/0x4b0
[   99.504424]  [<c1b2dcad>] ? mutex_unlock+0xd/0x10
[   99.509695]  [<c1b2dcad>] ? mutex_unlock+0xd/0x10
[   99.514967]  [<c1372f6e>] ? eventpoll_release_file+0x7e/0x90
[   99.521307]  [<c1335849>] __fput+0xd9/0x200
[   99.525996]  [<c133597d>] ____fput+0xd/0x10
[   99.530685]  [<c125c731>] task_work_run+0x81/0xb0
[   99.535957]  [<c12019e9>] do_notify_resume+0x49/0x70
[   99.541520]  [<c1b30dc4>] work_notifysig+0x29/0x31
[   99.546897] ------------[ cut here ]------------

So here we can call tty_vhangup() directly which is for virtual port.

Reviewed-by: Chao Bi <chao.bi@intel.com>
Signed-off-by: Liu, Chuansheng <chuansheng.liu@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/tty/n_gsm.c