> > > A patch to fix the problem and reorganize the code a bit for greater > > > readability is below. I'd appreciate it if people could try it out on > > > various locking litmus tests in our archives. > > > > Thanks for the quick solution, Alan. The results from our archives look > > good. > > Here's a much smaller patch, suitable for the -stable kernels. It fixes > the bug without doing the larger code reorganization (which will go into > a separate patch). Can you test this one? Testing in progress..., first results are good. (+1 on splitting the patches) Andrea