Oddity with CGROUPS and 32-bit machine

[depontius]

My temporary machine is a 32-bit Atom. Aside from the general slowness there is something extra-slow about it. Normally I boot or shutdown and walk away, it's so slow. Recently I stayed nearby, and saw a bunch of kernel oopses fly by. They all appear to be related to CGROUPS. I remember back when CGROUPS first appeared they made a machine really sluggish, and were one of the first things I removed from the kernel config. Then at some point someone came out with a userspace "speedy-200" that made more proper use of CGROUPS and made the machine act better, and shortly after that SCHED_AUTOGROUP came out, and I've been running that way ever since.

So seeing CGROUP in those oopses prompted me to remove them from the kernel config - maybe something is wrong with CGROUPS and SCHED_AUTOGROUP on this early Atom processor. There's another factor at work - a memory leak. Within a week or so of uptime, something chews almost all of my 2G RAM and drives me into swap. So it's entirely possible that the sluggishness is the memory leak, not the CGROUPS. Anyway, I've been running a CGROUP-free kernel "merely Atom-slow" since after work, and we'll see how that holds up.

Just wondering if anyone else has seen anything like this.

My own machine has been down for the last 6 months or so. The replacement was built last week, but so far I haven't gotten the motherboard to recognize my EFI partition. It's learning mode time, and there are tons of EFI HowTo's, all slightly different. I need to boil something down and discover just what "looks wrong" about mine. Now that I type this, I realize that I can take my SystemRescueCD over to my sister-in-law's and boot it on their new Windows machine, which is presumably Win-8.1, and look at a "valid" ESP partition directly.
_________________
.sigs waste space and bandwidth

[Pearlseattle]

[depontius]