Mark Brown wrote:
On Mon, Nov 10, 2008 at 10:11:24PM +0300, Dmitry Baryshkov wrote:
AC97 irq, I see a unending storm of irq requests. I've come with the following patch (against 2.6.27, if the idea is OK, I'll rebase against 2.6.28-rc).
That's going to cause the pxa25x to use the full timeout for the reset since it uses the interrupt to detect completion. This should just introduce a delay (which we shouldn't do) and may break something. I'd expect that simply switching pxa25x to polling GSR for completion like everything else does would deal with it but testing would be nice if anyone has a board...
I'll check on tosa (pxa255). However I didn't notice any additional delay during startup.
Otherwise it looks OK - we're not using the interrupt for anything else until after we've reset things so it's a sensible solution.