January 2nd, 2010

balancing enable/disable_irq_wake

On zaurus, I started to get ugly warnings about unbalanced enable/disable_irq_wake on cca 32-rc2. First, I thought warning is absolutely stupid: if IRQ is not wakeup capable (which seemed to be happening on zaurus), there's no need to WARN_ON() for disable after (unsuccessful) enable.

But now it seems that low-level zaurus code is pretty strange (enabling wakeups in groups when hw has no such limitation), and that the warning was actually very useful...