inconsistent lock state (tty? input?) - Kernel

This is a discussion on inconsistent lock state (tty? input?) - Kernel ; On current git (a447c0932445f92ce6f4c1bd020f62c5097a7842): [ 12.288256] Adding 1020112k swap on /dev/sda2. Priority:-1 extents:1 across:1020112k [ 33.469646] Marking TSC unstable due to cpufreq changes [ 37.320027] Clocksource tsc unstable (delta = -122876252 ns) [ 46.696880] [ 46.696883] ================================= [ 46.696899] [ ...

+ Reply to Thread
Results 1 to 3 of 3

Thread: inconsistent lock state (tty? input?)

  1. inconsistent lock state (tty? input?)

    On current git (a447c0932445f92ce6f4c1bd020f62c5097a7842):
    [ 12.288256] Adding 1020112k swap on /dev/sda2. Priority:-1 extents:1 across:1020112k
    [ 33.469646] Marking TSC unstable due to cpufreq changes
    [ 37.320027] Clocksource tsc unstable (delta = -122876252 ns)
    [ 46.696880]
    [ 46.696883] =================================
    [ 46.696899] [ INFO: inconsistent lock state ]
    [ 46.696904] 2.6.27-03556-gbc1d26d #15
    [ 46.696908] ---------------------------------
    [ 46.696912] inconsistent {hardirq-on-W} -> {in-hardirq-R} usage.
    [ 46.696917] swapper/0 [HC1[1]:SC0[0]:HE0:SE1] takes:
    [ 46.696921] (tasklist_lock){--?-}, at: [] send_sigio+0x3f/0xa1
    [ 46.696940] {hardirq-on-W} state was registered at:
    [ 46.696946] [] mark_held_locks+0x52/0x72
    [ 46.696955] [] trace_hardirqs_on_caller+0xf9/0x13a
    [ 46.696963] [] trace_hardirqs_on+0xd/0xf
    [ 46.696970] [] _spin_unlock_irq+0x30/0x57
    [ 46.696980] [] proc_clear_tty+0x4e/0x5b
    [ 46.696988] [] sys_setsid+0x86/0xa2
    [ 46.696998] [] system_call_fastpath+0x16/0x1b
    [ 46.697006] [] 0xffffffffffffffff
    [ 46.697015] irq event stamp: 406110
    [ 46.697019] hardirqs last enabled at (406109): [] trace_hardirqs_on+0xd/0xf
    [ 46.697027] hardirqs last disabled at (406110): [] trace_hardirqs_off_thunk+0x3a/0x3c
    [ 46.697036] softirqs last enabled at (406098): [] __do_softirq+0xbc/0xc4
    [ 46.697045] softirqs last disabled at (406065): [] call_softirq+0x1c/0x28
    [ 46.697053]
    [ 46.697054] other info that might help us debug this:
    [ 46.697059] 6 locks held by swapper/0:
    [ 46.697064] #0: (&serio->lock){++..}, at: [] serio_interrupt+0x2b/0x83
    [ 46.697080] #1: (&dev->event_lock){++..}, at: [] input_event+0x40/0x7a
    [ 46.697096] #2: (rcu_read_lock){..--}, at: [] input_pass_event+0x0/0xfa
    [ 46.697111] #3: (rcu_read_lock){..--}, at: [] mousedev_notify_readers+0x0/0x237
    [ 46.697127] #4: (fasync_lock){..+.}, at: [] kill_fasync+0x2c/0x4e
    [ 46.697142] #5: (&f->f_owner.lock){..+.}, at: [] send_sigio+0x24/0xa1
    [ 46.697155]
    [ 46.697156] stack backtrace:
    [ 46.697165] Pid: 0, comm: swapper Not tainted 2.6.27-03556-gbc1d26d #15
    [ 46.697171] Call Trace:
    [ 46.697175] [] print_usage_bug+0x1ad/0x1be
    [ 46.697188] [] mark_lock_irq+0x29b/0x681
    [ 46.697193] [] mark_lock+0x1f9/0x2d1
    [ 46.697198] [] mark_irqflags+0x2c/0xf2
    [ 46.697204] [] __lock_acquire+0x4e8/0x634
    [ 46.697210] [] lock_acquire+0x85/0xa9
    [ 46.697216] [] ? send_sigio+0x3f/0xa1
    [ 46.697222] [] _read_lock+0x3e/0x4a
    [ 46.697227] [] ? send_sigio+0x3f/0xa1
    [ 46.697233] [] send_sigio+0x3f/0xa1
    [ 46.697238] [] __kill_fasync+0x54/0x65
    [ 46.697244] [] kill_fasync+0x3a/0x4e
    [ 46.697249] [] mousedev_notify_readers+0x1b6/0x237
    [ 46.697256] [] ? mousedev_notify_readers+0x0/0x237
    [ 46.697261] [] mousedev_event+0x119/0x14e
    [ 46.697267] [] input_pass_event+0x94/0xfa
    [ 46.697274] [] ? input_pass_event+0x0/0xfa
    [ 46.697279] [] input_handle_event+0x25c/0x281
    [ 46.697286] [] input_event+0x62/0x7a
    [ 46.697293] [] psmouse_process_byte+0x257/0x267
    [ 46.697298] [] psmouse_handle_byte+0x17/0x10d
    [ 46.697305] [] psmouse_interrupt+0x277/0x285
    [ 46.697311] [] ? serio_interrupt+0x2b/0x83
    [ 46.697316] [] serio_interrupt+0x47/0x83
    [ 46.697322] [] i8042_interrupt+0x20b/0x223
    [ 46.697328] [] handle_IRQ_event+0x25/0x5a
    [ 46.697334] [] handle_edge_irq+0xdd/0x123
    [ 46.697340] [] do_IRQ+0xe4/0x14c
    [ 46.697345] [] ret_from_intr+0x0/0xf
    [ 46.697350] [] ? __atomic_notifier_call_chain+0x0/0xb7
    [ 46.697363] [] ? default_idle+0x32/0x4c
    [ 46.697369] [] ? default_idle+0x30/0x4c
    [ 46.697374] [] ? cpu_idle+0x4b/0x87
    [ 46.697494] [] ? rest_init+0x75/0x77

    Didn't happen on 4480f15b3306f43bbb0310d461142b4e897ca45b (Oct 12)
    and fd048088306656824958e7783ffcee27e241b361 (Oct 11).
    --
    To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
    the body of a message to majordomo@vger.kernel.org
    More majordomo info at http://vger.kernel.org/majordomo-info.html
    Please read the FAQ at http://www.tux.org/lkml/

  2. Re: inconsistent lock state (tty? input?)

    On Mon, 13 Oct 2008 20:37:43 +0200
    Marcin Slusarz wrote:

    > On current git (a447c0932445f92ce6f4c1bd020f62c5097a7842):

    \

    good spotting;

    the patch below should fix this:

    From 26c117549415c55585f46ad38cd58a5f84f5adc7 Mon Sep 17 00:00:00 2001
    From: Arjan van de Ven
    Date: Mon, 13 Oct 2008 14:56:39 -0400
    Subject: [PATCH] tty: make sure that proc_clear_tty stores the cpu flags

    proc_clear_tty() gets called with interrupts off (while holding the task list lock)
    from sys_setid. This means that it needs the _irqsave version of the locking primitives.

    Reported-by: Marcin Slusarz
    Signed-off-by: Arjan van de Ven
    ---
    drivers/char/tty_io.c | 5 +++--
    1 files changed, 3 insertions(+), 2 deletions(-)

    diff --git a/drivers/char/tty_io.c b/drivers/char/tty_io.c
    index e4dce87..4715961 100644
    --- a/drivers/char/tty_io.c
    +++ b/drivers/char/tty_io.c
    @@ -3595,9 +3595,10 @@ EXPORT_SYMBOL(tty_devnum);

    void proc_clear_tty(struct task_struct *p)
    {
    - spin_lock_irq(&p->sighand->siglock);
    + unsigned long flags;
    + spin_lock_irqsave(&p->sighand->siglock, flags);
    p->signal->tty = NULL;
    - spin_unlock_irq(&p->sighand->siglock);
    + spin_unlock_irqrestore(&p->sighand->siglock, flags);
    }

    /* Called under the sighand lock */
    --
    1.5.5.1



    --
    Arjan van de Ven Intel Open Source Technology Centre
    For development, discussion and tips for power savings,
    visit http://www.lesswatts.org
    --
    To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
    the body of a message to majordomo@vger.kernel.org
    More majordomo info at http://vger.kernel.org/majordomo-info.html
    Please read the FAQ at http://www.tux.org/lkml/

  3. Re: inconsistent lock state (tty? input?)

    On Mon, Oct 13, 2008 at 02:57:12PM -0400, Arjan van de Ven wrote:
    > On Mon, 13 Oct 2008 20:37:43 +0200
    > Marcin Slusarz wrote:
    >
    > > On current git (a447c0932445f92ce6f4c1bd020f62c5097a7842):

    > \
    >
    > good spotting;
    >
    > the patch below should fix this:
    >
    > From 26c117549415c55585f46ad38cd58a5f84f5adc7 Mon Sep 17 00:00:00 2001
    > From: Arjan van de Ven
    > Date: Mon, 13 Oct 2008 14:56:39 -0400
    > Subject: [PATCH] tty: make sure that proc_clear_tty stores the cpu flags
    >
    > proc_clear_tty() gets called with interrupts off (while holding the task list lock)
    > from sys_setid. This means that it needs the _irqsave version of the locking primitives.
    >
    > Reported-by: Marcin Slusarz
    > Signed-off-by: Arjan van de Ven
    > ---
    > drivers/char/tty_io.c | 5 +++--
    > 1 files changed, 3 insertions(+), 2 deletions(-)
    >
    > diff --git a/drivers/char/tty_io.c b/drivers/char/tty_io.c
    > index e4dce87..4715961 100644
    > --- a/drivers/char/tty_io.c
    > +++ b/drivers/char/tty_io.c
    > @@ -3595,9 +3595,10 @@ EXPORT_SYMBOL(tty_devnum);
    >
    > void proc_clear_tty(struct task_struct *p)
    > {
    > - spin_lock_irq(&p->sighand->siglock);
    > + unsigned long flags;
    > + spin_lock_irqsave(&p->sighand->siglock, flags);
    > p->signal->tty = NULL;
    > - spin_unlock_irq(&p->sighand->siglock);
    > + spin_unlock_irqrestore(&p->sighand->siglock, flags);
    > }
    >
    > /* Called under the sighand lock */
    > --


    It worked, thanks!

    But I had to manually apply this patch (this function looks different now)
    Updated patch below:
    ---
    From: Arjan van de Ven
    Subject: [PATCH] tty: make sure that proc_clear_tty stores the cpu flags

    proc_clear_tty() gets called with interrupts off (while holding the task list lock)
    from sys_setid. This means that it needs the _irqsave version of the locking primitives.

    Reported-and-tested-by: Marcin Slusarz
    Signed-off-by: Arjan van de Ven
    ---
    drivers/char/tty_io.c | 5 +++--
    1 files changed, 3 insertions(+), 2 deletions(-)

    diff --git a/drivers/char/tty_io.c b/drivers/char/tty_io.c
    index 7053d63..3f48d88 100644
    --- a/drivers/char/tty_io.c
    +++ b/drivers/char/tty_io.c
    @@ -3032,11 +3032,12 @@ EXPORT_SYMBOL(tty_devnum);

    void proc_clear_tty(struct task_struct *p)
    {
    + unsigned long flags;
    struct tty_struct *tty;
    - spin_lock_irq(&p->sighand->siglock);
    + spin_lock_irqsave(&p->sighand->siglock, flags);
    tty = p->signal->tty;
    p->signal->tty = NULL;
    - spin_unlock_irq(&p->sighand->siglock);
    + spin_unlock_irqrestore(&p->sighand->siglock, flags);
    tty_kref_put(tty);
    }

    --
    1.5.6.4

    --
    To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
    the body of a message to majordomo@vger.kernel.org
    More majordomo info at http://vger.kernel.org/majordomo-info.html
    Please read the FAQ at http://www.tux.org/lkml/

+ Reply to Thread