Bug 619063 - Desktop pauses while the kernel swapping-out to the (encrypted) swap partition
Summary: Desktop pauses while the kernel swapping-out to the (encrypted) swap partition
Keywords:
Status: CLOSED UPSTREAM
Alias: None
Product: Fedora
Classification: Fedora
Component: kernel
Version: 13
Hardware: x86_64
OS: Linux
low
low
Target Milestone: ---
Assignee: Kernel Maintainer List
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2010-07-28 13:40 UTC by Roland Pallai
Modified: 2011-03-30 23:02 UTC (History)
7 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed: 2011-03-30 23:02:20 UTC
Type: ---
Embargoed:


Attachments (Terms of Use)

Description Roland Pallai 2010-07-28 13:40:00 UTC
Description of problem:
On high order memory allocations kswapd evicts a lot of pages to the encrypted swap sometimes what causes long desktop pauses (mouse cursor, music playing stops). The system completely unusable until the hard disk has been flushed everything to the swap. The pause can be really long, up to 1 minute.

(The high order allocation because of cifs, see bug#617795)

I've vm.swappiness=0 and vm.min_free_kbytes=67584 setting in /etc/sysctl.conf. Swap is on LVM top of dm-crypt top of a high-performance SSD disk.

I'd like to hear flawless music playing under the swapping-out too.


Version-Release number of selected component (if applicable):
2.6.33.6-147.fc13.x86_64


How reproducible:
Make swap, fill the memory with usual memory-hungry desktop applications like gimp, firefox, netbeans, openoffice, make the free memory fragmented with updatedb and then try to allocate some high order pages (eg.: modprobe cifs CIFSMaxBufSize=128000). The desktop will freeze for a while.

If I remember correctly this issue doesn't depend on the encryption.


Additional info:

dstat log is at https://bugzilla.redhat.com/attachment.cgi?id=434098

cryptsetup status:
 /dev/mapper/luks-6eb56688-2689-49f6-8de4-6b30a0fb7cf2 is active:
  cipher:  aes-cbc-essiv:sha256
  keysize: 128 bits
  device:  /dev/sda2
  [...]

Comment 1 Roland Pallai 2011-03-30 23:02:20 UTC
I think this problem has been solved by upstream in kernel 2.6.38 as memory compaction replaced lumpy reclaim.

http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=3e7d344970673c5334cf7b5bb27c8c0942b06126


Note You need to log in before you can comment on or make changes to this bug.