2009/11/10 Axel Dörfler <axeld@xxxxxxxxxxxxxxxx>: > Is that already with the scheduler_affine changes I made or on a single > CPU machine? > Also, the spinlock latency debugging only checks the spinlocks - > currently not even the time one needed to wait to get it. And you might > hold more than one spinlock without reenabling interrupts, too. It's with the changes on a dual core Athlon 64. > Another thing we could do is measure the time each interrupt handler > took, and panic if that ever takes too long with interrupts turned off. Might be interesting. /Fredrik Holmqvist, TQH