And this isn't a regression, the update from 2.6.38-8.42 doesn't touch this area, seems the issue happens by "luck", some code shuffle or anything else (timing?) may be made the issue more likely to happen on newer kernel.
And this isn't a regression, the update from 2.6.38-8.42 doesn't touch this area, seems the issue happens by "luck", some code shuffle or anything else (timing?) may be made the issue more likely to happen on newer kernel.