Re: Problem with kvm on -tip

From: Avi Kivity
Date: Tue Apr 14 2009 - 04:21:10 EST


Peter Zijlstra wrote:
On Sat, 2009-04-11 at 15:08 +0300, Avi Kivity wrote:
[ 3293.134688] BUG: MAX_LOCK_DEPTH too low!
Looks like a genuine issue, need to increase MAX_LOCK_DEPTH. Andrea?

[ 3293.134704] turning off the locking correctness validator.
[ 3293.134718] Pid: 5117, comm: kvm Not tainted 2.6.30-rc1-tip-01420-g58e70a8
#18
[ 3293.134727] Call Trace:
[ 3293.134749] [<ffffffff802805f6>] __lock_acquire+0x4c6/0xbf0
[ 3293.134764] [<ffffffff80280e2e>] lock_acquire+0x10e/0x160
[ 3293.134780] [<ffffffff802f3760>] ? mm_take_all_locks+0x110/0x150
[ 3293.134798] [<ffffffff80580c3b>] _spin_lock_nest_lock+0x3b/0x50
[ 3293.134811] [<ffffffff802f3760>] ? mm_take_all_locks+0x110/0x150
[ 3293.134823] [<ffffffff802f3760>] mm_take_all_locks+0x110/0x150
[ 3293.134838] [<ffffffff803093af>] do_mmu_notifier_register+0xdf/0x1f0
[ 3293.134852] [<ffffffff803094f3>] mmu_notifier_register+0x13/0x20
[ 3293.134899] [<ffffffffa02edede>] kvm_dev_ioctl+0x1ae/0x360 [kvm]
[ 3293.134914] [<ffffffff80327a16>] vfs_ioctl+0x36/0xb0
[ 3293.134927] [<ffffffff80327b22>] do_vfs_ioctl+0x92/0x5c0
[ 3293.134942] [<ffffffff80273d9b>] ? up_read+0x2b/0x40
[ 3293.134955] [<ffffffff8032809f>] sys_ioctl+0x4f/0x80
[ 3293.134971] [<ffffffff8020c1f2>] system_call_fastpath+0x16/0x1b request

Thing is, its grabbing all vma locks, and we have a lock depth limit of
48. Now when we started this, the claim was that kvm would only need
this when the process was very fresh and would thus not yet have many
vma, ergo we should never run into this limit.

Has that changed?

Hasn't changed; this is on VM creation.

--
I have a truly marvellous patch that fixes the bug which this
signature is too narrow to contain.

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