'unannotated irqs-on' lockdep warning

On Thu, Jan 30, 2014 at 03:31:46PM +0100, Christian Gmeiner wrote:

[   19.859234] CPU: 0 PID: 1848 Comm: mkdir Not tainted 3.12.4 #44
[   19.865190] [<c0013900>] (unwind_backtrace+0x0/0xe0) from
[<c00113b8>] (show_stack+0x10/0x14)
[   19.873739] [<c00113b8>] (show_stack+0x10/0x14) from [<c044e040>]
(dump_stack+0x64/0xa4)
[   19.881851] [<c044e040>] (dump_stack+0x64/0xa4) from [<c0022718>]
(warn_slowpath_common+0x64/0x84)
[   19.890828] [<c0022718>] (warn_slowpath_common+0x64/0x84) from
[<c00227b8>] (warn_slowpath_fmt+0x2c/0x3c)
[   19.900413] [<c00227b8>] (warn_slowpath_fmt+0x2c/0x3c) from
[<c0076c84>] (check_flags.part.26+0xb4/0x1e4)
[   19.910001] [<c0076c84>] (check_flags.part.26+0xb4/0x1e4) from
[<c0079654>] (lock_release+0x3c/0x100)
[   19.919243] [<c0079654>] (lock_release+0x3c/0x100) from
[<c00485b4>] (lg_local_unlock+0x18/0x6c)
[   19.928055] [<c00485b4>] (lg_local_unlock+0x18/0x6c) from
[<c012a2cc>] (free_fs_struct+0x18/0x30)
[   19.936947] [<c012a2cc>] (free_fs_struct+0x18/0x30) from
[<c0024e24>] (do_exit+0x2ac/0x3f0)
[   19.945316] [<c0024e24>] (do_exit+0x2ac/0x3f0) from [<c002501c>]
(do_group_exit+0x88/0xb4)
[   19.953596] [<c002501c>] (do_group_exit+0x88/0xb4) from
[<c0025058>] (__wake_up_parent+0x0/0x18)
[   19.962391] ---[ end trace 98a70b5cdc7b49fe ]---
[   19.967017] possible reason: unannotated irqs-on.
[   19.971729] irq event stamp: 2910
[   19.975050] hardirqs last  enabled at (2909): [<c044a160>]
__slab_free+0x1c0/0x390
[   19.982661] hardirqs last disabled at (2910): [<c0456d14>]
__dabt_svc+0x34/0x60

So, I wonder how we got from __dabt_svc to __wake_up_parent.  It looks
like the unwinder has failed to do a proper job of unwinding, which
makes this undebuggable.

Can you rebuild in ARM mode with frame pointers enabled please?

-- 
FTTC broadband for 0.8mile line: 5.8Mbps down 500kbps up.  Estimation
in database were 13.1 to 19Mbit for a good line, about 7.5+ for a bad.
Estimate before purchase was "up to 13.2Mbit".