iommu lockdep trace on 3.0.0

(it says 2.6.40, but it's 3.0.0 in all but name)

Hit this while booting.

Dave

[  706.848926] =======================================================
[  706.857477] [ INFO: possible circular locking dependency detected ]
[  706.863738] 2.6.40-4.fc15.x86_64.debug #1
[  706.867746] -------------------------------------------------------
[  706.874058] modprobe/1702 is trying to acquire lock:
[  706.879047]  (&(&iommu->lock)->rlock){......}, at: [<ffffffff8128d308>] domain_remove_one_dev_info+0x1c1/0x20b
[  706.889143] 
[  706.889144] but task is already holding lock:
[  706.895029]  (device_domain_lock){..-...}, at: [<ffffffff8128d279>] domain_remove_one_dev_info+0x132/0x20b
[  706.904795] 
[  706.904796] which lock already depends on the new lock.
[  706.904798] 
[  706.913062] 
[  706.913063] the existing dependency chain (in reverse order) is:
[  706.920608] 
[  706.920609] -> #1 (device_domain_lock){..-...}:
[  706.926744]        [<ffffffff8108ab96>] lock_acquire+0xbf/0x103
[  706.932684]        [<ffffffff814f6c77>] _raw_spin_lock_irqsave+0x45/0x7f
[  706.939408]        [<ffffffff8128c16b>] domain_context_mapping_one+0x2b7/0x49a
[  706.946695]        [<ffffffff8128d6c8>] domain_context_mapping+0x3d/0xe5
[  706.953437]        [<ffffffff814f2097>] iommu_prepare_identity_map+0x18f/0x1ae
[  706.960718]        [<ffffffff81d74d18>] intel_iommu_init+0x7f0/0xa9d
[  706.967132]        [<ffffffff81d4c0d0>] pci_iommu_init+0x29/0x54
[  706.973166]        [<ffffffff81002099>] do_one_initcall+0x7f/0x13a
[  706.979372]        [<ffffffff81d44c98>] kernel_init+0xdf/0x159
[  706.985266]        [<ffffffff814ff0a4>] kernel_thread_helper+0x4/0x10
[  707.054277] 
[  707.054278] -> #0 (&(&iommu->lock)->rlock){......}:
[  707.061381]        [<ffffffff8108a42c>] __lock_acquire+0xa1a/0xcf7
[  707.067580]        [<ffffffff8108ab96>] lock_acquire+0xbf/0x103
[  707.074883]        [<ffffffff814f6c77>] _raw_spin_lock_irqsave+0x45/0x7f
[  707.081583]        [<ffffffff8128d308>] domain_remove_one_dev_info+0x1c1/0x20b
[  707.090185]        [<ffffffff8128d9aa>] device_notifier+0x54/0x7e
[  707.097489]        [<ffffffff814fa9a4>] notifier_call_chain+0x84/0xbb
[  707.103938]        [<ffffffff8107b16c>] __blocking_notifier_call_chain+0x67/0x84
[  707.111380]        [<ffffffff8107b19d>] blocking_notifier_call_chain+0x14/0x16
[  707.119981]        [<ffffffff81326492>] __device_release_driver+0xcd/0xd2
[  707.127814]        [<ffffffff81326b4f>] driver_detach+0x99/0xc2
[  707.133769]        [<ffffffff8132630f>] bus_remove_driver+0xba/0xdf
[  707.140200]        [<ffffffff813271f5>] driver_unregister+0x6a/0x75
[  707.146468]        [<ffffffff81279592>] pci_unregister_driver+0x44/0x8d
[  707.163869]        [<ffffffff81094039>] sys_delete_module+0x1dd/0x251
[  707.170319]        [<ffffffff814fdf02>] system_call_fastpath+0x16/0x1b
[  707.177535] 
[  707.177536] other info that might help us debug this:
[  707.177537] 
[  707.185549]  Possible unsafe locking scenario:
[  707.185550] 
[  707.191498]        CPU0                    CPU1
[  707.196043]        ----                    ----
[  707.200574]   lock(device_domain_lock);
[  707.204505]                                lock(&(&iommu->lock)->rlock);
[  707.211266]                                lock(device_domain_lock);
[  707.217679]   lock(&(&iommu->lock)->rlock);
[  707.221933] 
[  707.221934]  *** DEADLOCK ***
[  707.221935] 
[  707.227943] 4 locks held by modprobe/1702:
[  707.232066]  #0:  (&__lockdep_no_validate__){+.+.+.}, at: [<ffffffff81326b30>] driver_detach+0x7a/0xc2
[  707.241488]  #1:  (&__lockdep_no_validate__){+.+.+.}, at: [<ffffffff81326b3e>] driver_detach+0x88/0xc2
[  707.250961]  #2:  (&(&priv->bus_notifier)->rwsem){.+.+.+}, at: [<ffffffff8107b154>] __blocking_notifier_call_chain+0x4f/0x84
[  707.262303]  #3:  (device_domain_lock){..-...}, at: [<ffffffff8128d279>] domain_remove_one_dev_info+0x132/0x20b
[  707.272494] 
[  707.272495] stack backtrace:
[  707.276915] Pid: 1702, comm: modprobe Not tainted 2.6.40-4.fc15.x86_64.debug #1
[  707.284208] Call Trace:
[  707.286697]  [<ffffffff814ed313>] print_circular_bug+0x1f8/0x209
[  707.292692]  [<ffffffff8108a42c>] __lock_acquire+0xa1a/0xcf7
[  707.298363]  [<ffffffff8128d308>] ? domain_remove_one_dev_info+0x1c1/0x20b
[  707.305226]  [<ffffffff8108ab96>] lock_acquire+0xbf/0x103
[  707.310619]  [<ffffffff8128d308>] ? domain_remove_one_dev_info+0x1c1/0x20b
[  707.324019]  [<ffffffff814f6c77>] _raw_spin_lock_irqsave+0x45/0x7f
[  707.330219]  [<ffffffff8128d308>] ? domain_remove_one_dev_info+0x1c1/0x20b
[  707.337082]  [<ffffffff810874ea>] ? trace_hardirqs_off+0xd/0xf
[  707.342933]  [<ffffffff8128d308>] domain_remove_one_dev_info+0x1c1/0x20b
[  707.349624]  [<ffffffff8128d9aa>] device_notifier+0x54/0x7e
[  707.355216]  [<ffffffff814fa9a4>] notifier_call_chain+0x84/0xbb
[  707.361129]  [<ffffffff8107b16c>] __blocking_notifier_call_chain+0x67/0x84
[  707.368021]  [<ffffffff8107b19d>] blocking_notifier_call_chain+0x14/0x16
[  707.374710]  [<ffffffff81326492>] __device_release_driver+0xcd/0xd2
[  707.380995]  [<ffffffff81326b4f>] driver_detach+0x99/0xc2
[  707.386390]  [<ffffffff8132630f>] bus_remove_driver+0xba/0xdf
[  707.392163]  [<ffffffff813271f5>] driver_unregister+0x6a/0x75
[  707.397904]  [<ffffffff81279592>] pci_unregister_driver+0x44/0x8d
[  707.410102]  [<ffffffff81094039>] sys_delete_module+0x1dd/0x251
[  707.416014]  [<ffffffff814f7739>] ? retint_swapgs+0x13/0x1b
[  707.421582]  [<ffffffff810aff1d>] ? audit_syscall_entry+0x11c/0x148
[  707.427840]  [<ffffffff8126089e>] ? trace_hardirqs_on_thunk+0x3a/0x3f
[  707.434297]  [<ffffffff814fdf02>] system_call_fastpath+0x16/0x1b

Looks like the same thing I hit. I've sent a patch that fixes it for me a couple of times:

https://lists.linux-foundation.org/pipermail/iommu/2011-July/002784.html