19 Sep
2007
19 Sep
'07
7:22 a.m.
New thread for the the final version of the patches for my fixes to the ad1848 auto-calibration loop. I made a couple suggested changes. These should be ok to apply, and should be independent of the other as yet un-applied patches that have been posted by Rene and Krzysztof.
I'm not an expert on this chip by any means, but from looking at the code I think there are few things that could be fixed.
The reg_lock spinlock isn't acquired from the irq handler, as the handler doesn't use any of the indirect registers. I think this means that is isn't necessary to use the irq disabling versions of the spin_lock functions with reg_lock.
It does look like there is a different SMP race condition wrt the irq handler.