Messages in this thread | | | Subject | Re: 2.6.37-git5 -- INFO: possible irq lock inversion dependency detected | From | Johannes Berg <> | Date | Wed, 12 Jan 2011 10:26:53 +0100 |
| |
On Tue, 2011-01-11 at 15:10 -0500, Miles Lane wrote: > [ INFO: possible irq lock inversion dependency detected ] > 2.6.37-git5 #1 > --------------------------------------------------------- > dbus-daemon/1512 just changed the state of lock: > (&(&list->lock)->rlock#7){+.-...}, at: [<ffffffffa01d4ecc>] > ieee80211_rx_handlers+0x36/0x1ae3 [mac80211] > but this lock was taken by another, HARDIRQ-safe lock in the past: > (&(&priv->lock)->rlock){-.-...} > > and interrupts could create inverse lock ordering between them.
Thanks Miles. I'm not sure I fully understand this trace, but I think you can probably easily reproduce it. Would you be willing to test with the patch on this email? I think it's related:
http://thread.gmane.org/gmane.linux.kernel.wireless.general/62842/focus=62848
johannes
| |