Messages in this thread | | | Subject | Re: printk() vs tty_io | From | Peter Zijlstra <> | Date | Thu, 15 Dec 2011 11:55:17 +0100 |
| |
On Thu, 2011-12-15 at 10:32 +0000, Alan Cox wrote:
> For lockdep spews do they need to be spewed synchronously - clearly it's > good if they can be but its possible that they could be buffered ?
This wasn't about lockdep per-se, it all started with me trying to cleanup printk()'s abuse of lockdep_off()/lockdep_on. Which then grew into something a little bigger.
I'm fine with dropping the larger goal of making printk() work in all but NMI context and focus on the initial cleanup.
| |