Messages in this thread | | | Subject | Re: [PATCH 2/3] printk, lockdep: Remove lockdep_off() usage | From | Peter Zijlstra <> | Date | Fri, 10 Jun 2011 15:23:55 +0200 |
| |
On Thu, 2011-06-09 at 15:06 +0200, Peter Zijlstra wrote: > Remove the lockdep_off() usage from printk(). Also add a > debug_locks_off() call to zap_locks() since that'll mess up the lock > state in a royal way anyway. Further switch to local_irq_ ops so that > the irq state is properly tracked (raw_local_irq_* isn't tracked by > lockdep, causing confusion). Also drop superfluous preempt_disable(), > disabling IRQs already avoids scheduling. >
OK, so while trying to decipher the irq status of printk() I found commit 1efc5da3 which explains why we've got lockdep_off() in there, there still is a recursion problem with trace_hardirqs_*.
/me goes poke at that..
| |