On Thu, Feb 28, 2013 at 11:43:37AM -0800, Greg KH wrote:
On Thu, Feb 28, 2013 at 11:36:58AM -0800, chrysippus@operamail.com wrote:
Can you reproduce this without the closed source kernel module you have loaded installed?
The taint comes from nvidia.ko.
Removing it from the equation, I continue to see the occassional, similar error message.
I'm not currently able to reproduce on demand; it's intermittent. This is, now, after a server reboot, as well.
It appears to be related to how much I'm taxing the server. At least, while running a large, parallel build on the box I saw a couple of those messages.
I'm cobbling up some scripted tests to flex memory, i/o, etc. to see what I'm able to figure out.
Regardless of taint or not, error or not, my question remains re: what, if any, values for vm.swappiness & vm.vfs_cache_pressure are recommended.
The default ones should be fine. You have something that is doing a lot of small kernel allocations so your free memory in that "pool" is running out of space, and the kernel is having to fall back on another one. It's not really a big deal, as the message says, but tracking down the root cause might be good to do, if at all possible.
Sorry, got that wrong, Jeff is right, this is a "big" allocation that is needed. The nvidia driver is known for taking up these memory blocks, so this should go away if you remove that driver. thanks, greg k-h -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org