A Dimarts 17 Agost 2010 11:18:12, vàreu escriure:
Hello Merten,
thanks for the idea, but the problem means more than a performance problem, it
is a system hangup,
COW is the aufs read-write, in this case is mounted on disk to create a
persistent storage that saves the modifications for a Live system.
explaining two different cases:
# mount to tmpfs
mount -t tmps tmpfs /tmp
mkdir /tmp/rw /tmp/aufs
mount -t aufs -o br=/tmp/rw:${HOME} none /tmp/aufs
# an initramfs mounts aufs to a persistent directory on disk, and initializes
a Live system:
mkdir /lnet/cowroot /lnet/cow /root
mount /dev/sda3 /lnet/cowroot
mount -o bind,rw /lnet/cowroot/LneTPersCOW /lnet/cow
mount -t aufs -o br=/lnet/cow:${HOME} none /root
...
init /root
I believe that the problem is some incompatibility between the hard disk
filesystem and the aufs write routines, as write locking-unlocking,
for example taking some code (that does not show where is the problem, only I
have taken some code from aufs):
di_write_unlock(root);
err = test_dentry_busy(root, bindex, sigen, verbose);
err = test_inode_busy(root->d_sb, bindex, sigen, verbose);
di_write_lock_child(root); /* aufs_write_lock() calls ..._child() */
Jordi Pujol
Live never ending Tale
GNU/Linux Live forever!
http://livenet.selfip.com
------------------------------------------------------------------------------
This SF.net email is sponsored by
Make an app they can't live without
Enter the BlackBerry Developer Challenge
http://p.sf.net/sfu/RIM-dev2dev
