Jens, the elapsed time for `mkefs2 /dev/hdc5' has just gone from nine
seconds to sixty, with the anticipatory scheduler (at least). It is due to
your recent merge.
This is the enty enth time that block changes have been slammed into
mainline without having had any exposure to -mm testers or even to me
personally. And it's the second time obvious regressions have gone in as a
result. I have a New Year's resolution for you ;)