Comment 79 for bug 585765

Igor Wojnicki wrote "Take a lookat StuartN's post: http://ubuntuforums.org/showpost.php?p=9468596&postcount=487
This might be a hint what causes the freezes."

Switching the root filesystem to writeback mode (instead of the default ordered mode) will work with EXT3 or EXT4. It is not a solution, but it is possibly a method of gathering further data. In my case, I have a system that is functioning (almost) perfectly with kernel 2.6.34 and freezing with kernel 2.6.32-22. The freeze now affects a single process, all the inodes attached to that process and subsequently all processes depending on those inodes - yes, the system does fully freeze after some time, but I have no difficulty accessing logs after recognizing that a process has frozen. (See http://www.iol.ie/~stuartneilson/Bootup_fsck.html)

Following the first message, "EXT4-fs warning (device sda4): dx_probe: dx entry: limit != root limit" (which would equally apply to EXT3-fs), leads to a commit 3d82abae9523c33d4a16fdfdfd2bdde316d7b56a with the title "dir_index: error out instead of BUG on corrupt dx dirs" - http://mirror.celinuxforum.org/gitstat//commit-detail.php?commit=3d82abae9523c33d4a16fdfdfd2bdde316d7b56a - http://lkml.indiana.edu/hypermail/linux/kernel/0709.3/0208.html

I have not located the version where this is merged into the Ubuntu kernel, or indeed if it is directly relevant. I am going to test Paul.Drover's suggested test for affected systems from http://ubuntuforums.org/showpost.php?p=9471373&postcount=496