Hi Alex,
I stumbled over this lockdep warning yesterday while testing my VT-d
changes. It looks like one code path is taking the locks:
group->device_lock
driver_lock
pci_bus_sem
while another path is taking
pci_bus_sem
group->device_lock
which could lead to a deadlock. I attach the full warning, can you
please have a look?
[ 252.892008] ======================================================
[ 252.892008] [ INFO: possible circular locking dependency detected ]
[ 252.892008] 4.2.0-rc3+ #16 Not tainted
[ 252.892008] -------------------------------------------------------
[ 252.892008] qemu-system-x86/4986 is trying to acquire lock:
[ 252.892008] (&group->device_lock){+.+.+.}, at: [<ffffffffa0569da4>] vfio_group_get_device+0x24/0xb0 [vfio]
[ 252.892008]
but task is already holding lock:
[ 252.892008] (pci_bus_sem){++++.+}, at: [<ffffffff813ee47e>] pci_walk_bus+0x2e/0xa0
[ 252.892008]
which lock already depends on the new lock.
[ 252.892008]
the existing dependency chain (in reverse order) is:
[ 252.892008]
-> #2 (pci_bus_sem){++++.+}:
[ 252.892008] [<ffffffff810c9152>] __lock_acquire+0xca2/0x1550
[ 252.892008] [<ffffffff810c9b6f>] lock_acquire+0xdf/0x2c0
[ 252.892008] [<ffffffff81710c4c>] down_read+0x4c/0xa0
[ 252.892008] [<ffffffff813ee47e>] pci_walk_bus+0x2e/0xa0
[ 252.892008] [<ffffffffa0657dcb>] vfio_pci_release+0x18b/0x3c0 [vfio_pci]
[ 252.892008] [<ffffffffa056a100>] vfio_device_fops_release+0x20/0x40 [vfio]
[ 252.892008] [<ffffffff81211460>] __fput+0xf0/0x200
[ 252.892008] [<ffffffff812115be>] ____fput+0xe/0x10
[ 252.892008] [<ffffffff8108f5fd>] task_work_run+0x8d/0xc0
[ 252.892008] [<ffffffff8106d35c>] do_exit+0x32c/0xc30
[ 252.892008] [<ffffffff8106f10c>] do_group_exit+0x4c/0xc0
[ 252.892008] [<ffffffff8107d678>] get_signal+0x328/0x9e0
[ 252.892008] [<ffffffff81003458>] do_signal+0x28/0x9e0
[ 252.892008] [<ffffffff81003e75>] do_notify_resume+0x65/0x80
[ 252.892008] [<ffffffff81713e2e>] int_signal+0x12/0x17
[ 252.892008]
-> #1 (driver_lock){+.+.+.}:
[ 252.892008] [<ffffffff810c9152>] __lock_acquire+0xca2/0x1550
[ 252.892008] [<ffffffff810c9b6f>] lock_acquire+0xdf/0x2c0
[ 252.892008] [<ffffffff8170e71b>] mutex_lock_nested+0x6b/0x420
[ 252.892008] [<ffffffffa06576b8>] vfio_pci_open+0x38/0x270 [vfio_pci]
[ 252.892008] [<ffffffffa056ab17>] vfio_group_fops_unl_ioctl+0x267/0x460 [vfio]
[ 252.892008] [<ffffffff8122454d>] do_vfs_ioctl+0x30d/0x580
[ 252.892008] [<ffffffff81224839>] SyS_ioctl+0x79/0x90
[ 252.892008] [<ffffffff81713c32>] entry_SYSCALL_64_fastpath+0x16/0x7a
[ 252.892008]
-> #2 (&group->device_lock){+.+.+.}:
[ 252.892008] [<ffffffff810c6c6c>] check_prevs_add+0x8fc/0x900
[ 252.892008] [<ffffffff810c9152>] __lock_acquire+0xca2/0x1550
[ 252.892008] [<ffffffff810c9b6f>] lock_acquire+0xdf/0x2c0
[ 252.892008] [<ffffffff8170e71b>] mutex_lock_nested+0x6b/0x420
[ 252.892008] [<ffffffffa0569da4>] vfio_group_get_device+0x24/0xb0 [vfio]
[ 252.892008] [<ffffffffa056a165>] vfio_device_get_from_dev+0x45/0xa0 [vfio]
[ 252.892008] [<ffffffffa065762c>] vfio_pci_get_devs+0x2c/0x80 [vfio_pci]
[ 252.892008] [<ffffffffa065706d>] vfio_pci_walk_wrapper+0x5d/0x70 [vfio_pci]
[ 252.892008] [<ffffffff813ee4c5>] pci_walk_bus+0x75/0xa0
[ 252.892008] [<ffffffffa0657e93>] vfio_pci_release+0x253/0x3c0 [vfio_pci]
[ 252.892008] [<ffffffffa056a100>] vfio_device_fops_release+0x20/0x40 [vfio]
[ 252.892008] [<ffffffff81211460>] __fput+0xf0/0x200
[ 252.892008] [<ffffffff812115be>] ____fput+0xe/0x10
[ 252.892008] [<ffffffff8108f5fd>] task_work_run+0x8d/0xc0
[ 252.892008] [<ffffffff8106d35c>] do_exit+0x32c/0xc30
[ 252.892008] [<ffffffff8106f10c>] do_group_exit+0x4c/0xc0
[ 252.892008] [<ffffffff8107d678>] get_signal+0x328/0x9e0
[ 252.892008] [<ffffffff81003458>] do_signal+0x28/0x9e0
[ 252.892008] [<ffffffff81003e75>] do_notify_resume+0x65/0x80
[ 252.892008] [<ffffffff81713e2e>] int_signal+0x12/0x17
[ 252.892008]
other info that might help us debug this:
[ 252.892008] Chain exists of:
&group->device_lock --> driver_lock --> pci_bus_sem
[ 252.892008] Possible unsafe locking scenario:
[ 252.892008] CPU0 CPU1
[ 252.892008] ---- ----
[ 252.892008] lock(pci_bus_sem);
[ 252.892008] lock(driver_lock);
[ 252.892008] lock(pci_bus_sem);
[ 252.892008] lock(&group->device_lock);
[ 252.892008]
*** DEADLOCK ***
[ 252.892008] 2 locks held by qemu-system-x86/4986:
[ 252.892008] #0: (driver_lock){+.+.+.}, at: [<ffffffffa0657c67>] vfio_pci_release+0x27/0x3c0 [vfio_pci]
[ 252.892008] #1: (pci_bus_sem){++++.+}, at: [<ffffffff813ee47e>] pci_walk_bus+0x2e/0xa0
[ 252.892008]
stack backtrace:
[ 252.892008] CPU: 5 PID: 4986 Comm: qemu-system-x86 Not tainted 4.2.0-rc3+ #16
[ 252.892008] Hardware name: Dell Inc. Precision T3610/09M8Y8, BIOS A06 02/28/2014
[ 252.892008] ffffffff828e8d10 ffff88042af47818 ffffffff8170a128 0000000000000000
[ 252.892008] ffffffff828e54e0 ffff88042af47868 ffffffff81704b1d ffff880374b41180
[ 252.892008] ffff88042af478a8 ffff88042af47868 0000000000000001 ffff880374b41a20
[ 252.892008] Call Trace:
[ 252.892008] [<ffffffff8170a128>] dump_stack+0x4c/0x6e
[ 252.892008] [<ffffffff81704b1d>] print_circular_bug+0x202/0x213
[ 252.892008] [<ffffffff810c6c6c>] check_prevs_add+0x8fc/0x900
[ 252.892008] [<ffffffff8100dac9>] ? sched_clock+0x9/0x10
[ 252.892008] [<ffffffff810c9152>] __lock_acquire+0xca2/0x1550
[ 252.892008] [<ffffffff810c9b6f>] lock_acquire+0xdf/0x2c0
[ 252.892008] [<ffffffffa0569da4>] ? vfio_group_get_device+0x24/0xb0 [vfio]
[ 252.892008] [<ffffffff8170e71b>] mutex_lock_nested+0x6b/0x420
[ 252.892008] [<ffffffffa0569da4>] ? vfio_group_get_device+0x24/0xb0 [vfio]
[ 252.892008] [<ffffffffa0569da4>] ? vfio_group_get_device+0x24/0xb0 [vfio]
[ 252.892008] [<ffffffff810c72fd>] ? trace_hardirqs_on+0xd/0x10
[ 252.892008] [<ffffffffa0569da4>] vfio_group_get_device+0x24/0xb0 [vfio]
[ 252.892008] [<ffffffffa056a165>] vfio_device_get_from_dev+0x45/0xa0 [vfio]
[ 252.892008] [<ffffffffa0657010>] ? vfio_pci_count_devs+0x10/0x10 [vfio_pci]
[ 252.892008] [<ffffffffa065762c>] vfio_pci_get_devs+0x2c/0x80 [vfio_pci]
[ 252.892008] [<ffffffffa0657010>] ? vfio_pci_count_devs+0x10/0x10 [vfio_pci]
[ 252.892008] [<ffffffffa065706d>] vfio_pci_walk_wrapper+0x5d/0x70 [vfio_pci]
[ 252.892008] [<ffffffff813ee4c5>] pci_walk_bus+0x75/0xa0
[ 252.892008] [<ffffffffa0657e93>] vfio_pci_release+0x253/0x3c0 [vfio_pci]
[ 252.892008] [<ffffffffa0657600>] ? vfio_pci_rw+0x60/0x60 [vfio_pci]
[ 252.892008] [<ffffffffa056a100>] vfio_device_fops_release+0x20/0x40 [vfio]
[ 252.892008] [<ffffffff81211460>] __fput+0xf0/0x200
[ 252.892008] [<ffffffff812115be>] ____fput+0xe/0x10
[ 252.892008] [<ffffffff8108f5fd>] task_work_run+0x8d/0xc0
[ 252.892008] [<ffffffff8106d35c>] do_exit+0x32c/0xc30
[ 252.892008] [<ffffffff81713110>] ? _raw_spin_unlock_irq+0x30/0x60
[ 252.892008] [<ffffffff8106f10c>] do_group_exit+0x4c/0xc0
[ 252.892008] [<ffffffff8107d678>] get_signal+0x328/0x9e0
[ 252.892008] [<ffffffff81003458>] do_signal+0x28/0x9e0
[ 252.892008] [<ffffffff8110cba1>] ? do_futex+0xd1/0x500
[ 252.892008] [<ffffffff811ca67d>] ? __might_fault+0x4d/0xa0
[ 252.892008] [<ffffffff81713ddb>] ? int_very_careful+0x5/0x46
[ 252.892008] [<ffffffff81003e75>] do_notify_resume+0x65/0x80
[ 252.892008] [<ffffffff81713e2e>] int_signal+0x12/0x17
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
