tmpfs leaks on image builds

[Robert Collins]

On jenkins, after a few builds that failed:
tmpfs 16G 431M 16G 3% /tmp/image.7iPUftnL
tmpfs 16G 253M 16G 2% /tmp/image.eYHo8AmR
tmpfs 16G 226M 16G 2% /tmp/image.Jb483haJ
tmpfs 16G 445M 16G 3% /tmp/image.T5UcRmsQ
tmpfs 16G 431M 16G 3% /tmp/image.FBbd76NK
tmpfs 16G 253M 16G 2% /tmp/image.PfYNJGqO
tmpfs 16G 226M 16G 2% /tmp/image.Ji2pG7Qe
tmpfs 16G 445M 16G 3% /tmp/image.iZkB74FO
tmpfs 16G 1.3G 15G 8% /tmp/image.XLfDtMAM
tmpfs 16G 1.3G 15G 8% /tmp/image.0FBkhw0H
tmpfs 16G 904M 15G 6% /tmp/image.cgAimA6H

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/28687

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/28687
Committed: http://github.com/stackforge/diskimage-builder/commit/20661e8d8070c034bbc962feeb15b72d710dee5d
Submitter: Jenkins
Branch: master

commit 20661e8d8070c034bbc962feeb15b72d710dee5d
Author: Clint Byrum <email address hidden>
Date: Thu May 9 09:34:53 2013 -0700

    Retry losetup -d for up to 10 seconds.

    When partition tables are added to loopback devices, this can set off a
    chain of udev hooks that may still be holding the loopback open. Failing
    to detach loopback devices was the reason we were seeing leaked tmpfs
    volumes.

    Fixes bug #1178091

    Change-Id: I836d6e2bbce824951dd4786e3ef28273ea18ee73