Data corruption bug in this kernel and 3.6.2

Note: if you are using ext4 with 3.6.2+, lightly-modified filesystems are at risk of severe corruption if you unmount them before the journal has a chance to wrap (at least, if you do that twice). See <http://lkml.indiana.edu/hypermail/linux/kernel/1210.2/047...>, which may fix it (I haven't tested it, but will tomorrow).

Like every other Linux user for so very many reasons, I owe tytso a beer if I see him. (I suspect that so many people owe him beers by now that alcohol poisoning is a major risk.)