On Sat, 8 Dec 2012, Mel Gorman wrote:

Commit 8852aac2 (workqueue: mod_delayed_work_on() shouldn't queue timer on
0 delay) is causing the following boot failure for me. Found by bisection
but no further analysis unfortunately until I get back home properly.
I've added the relevant maintainers for megasas and SCSI in case it's
somehow specific to that driver.

This should be fixed by commit c1d390d8e612. The Megaraid SAS driver was 
doing insane things, using a work-struct for delayed work, and casting 
pointers around. It had happened to work for all the wrong reasons before, 
the mod_delayed_work_on() changes just exposed how crazy that crap was.

So please test current -git. If it's still broken, we'll need to know, but 
I'm assuming this is the already-known-and-fixed bug.

             Linus