On Wed, Jul 18, 2007 at 12:53:08AM -0700, Rajouri Jammu wrote: > Thanks for the quick response. > > Re Lookup intents, yes, I did see a kernel panic with intent of OPEN. > > The following work around seems to have resolved the problem. Can > you please help me figure out if this is OK to do or not?
This issue has been given a hard look over the last several months by various members of the kernel community and by the Unionfs guys. You probably want to get familiar with those discussions before you go too far forward with a proposed fix. Here was the first false start: http://lkml.org/lkml/2007/3/9/95 http://lkml.org/lkml/2007/3/9/238 So the lookup_one_len_nd() approach won't work. Jeff Sipek tells me that Andrew Morton has patches, possibly going into 2.6.23, that introduce vfs_path_lookup(). So far, this allows file create on NFS. http://lkml.org/lkml/2007/5/23/483 Really, the problem has to do with how NFS handles things *after* the lookup. http://www.fsl.cs.sunysb.edu/~jsipek/kernel/how_to_do_lookups_the_right_way.txt In short, the solution is going to have to touch code outside of eCryptfs. It might be necessary to split open_namei() (fs/namei.c) to really do this right. > In ecryptfs_lookup(), instead of calling lookup_one_len(), what if we > directly call the lower layer lookup so that the flags in nd also are > passed down. > > lower_dentry = d_lookup(lower_dir_dentry, &dentry->d_name); > if (!lower_dentry) { > > temp_dentry = d_alloc(lower_dir_dentry, &dentry->d_name); > if (temp_dentry) { > lower_dentry = > lower_dir_dentry->d_inode->i_op->lookup(lower_dir_dentry->d_inode, > > temp_dentry, nd); > if (lower_dentry) > dput(temp_dentry); > else > lower_dentry = temp_dentry; > } > rc = -ENOMEM; > d_drop(dentry); > goto out; > } > > > On 7/17/07, Michael Halcrow <[EMAIL PROTECTED]> wrote: > > On Tue, Jul 17, 2007 at 04:30:41PM -0700, Rajouri Jammu wrote: > > > I'm interested in running ecryptfs on networked file systems. > > > > > > I noticed the following comment in the sources. > > > > > > > /** > > > > * This is a temporary hack to prevent inadvertent mounting on lower > > > > * filesystems that are known to be currently incompatible with > > > > * eCryptfs. This mainly includes networked filesystems. > > > > */ > > > > > > Can someone please let me know why ecryptfs is not compatible with > > > NFS? > > > > Here is a short laundry list: > > > > - Lookup intents have been problematic with stacked filesystems on > > NFS. > > - writepages() NULL ptr deref, which might have a fix: > > - http://www.mail-archive.com/[EMAIL PROTECTED]/msg06097.html > > - Unlinks keep sillyrenamed files around until eCryptfs is > > unmounted. I have no idea why. > > > > Mike > > .___________________________________________________________________. > > Michael A. Halcrow > > Security Software Engineer, IBM Linux Technology Center > > GnuPG Fingerprint: 419C 5B1E 948A FA73 A54C 20F5 DB40 8531 6DCA 8769 > > > > If this is flashing then you're a winner! > > > > -----BEGIN PGP SIGNATURE----- > > Version: GnuPG v1.4.1 (GNU/Linux) > > > > iQEVAwUBRp1qU9tAhTFtyodpAQM2Bwf/UknPunsBANZLHTXB6aTNo955/BQ8tb1E > > WxHQj+VF8QWrfaYdg0tgdYhaIvagYNpCq5HMDWQHr4FNGpgQNu8uqQ3fsu/pIMWK > > AGtlRUFCNzT9sJME9Xiigp2qu7WZ67vjenrYYErfT9rmeiF/FW6T1qd155yYQdmu > > hh277TIybVDlHO7jzfgJkJr94CaQr9smubUhcQkm2AkPIiyvOwN/LGHlxvboS2l/ > > mJQgZHZdZ/v1J0sJq2TpPdj9svKCYEdRodZzWMMRgFNrxjnSZ2lrqm+g8qUxmyda > > OnJjde4oJ0Vb5ER54E5ErMWnK47tvwhZhBw/J4XS+ZULs1MAZ5JfCQ== > > =L03Z > > -----END PGP SIGNATURE----- > > > > > > ------------------------------------------------------------------------- > This SF.net email is sponsored by DB2 Express > Download DB2 Express C - the FREE version of DB2 express and take > control of your XML. No limits. Just data. Click to get it now. > http://sourceforge.net/powerbar/db2/ > _______________________________________________ > eCryptfs-devel mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/ecryptfs-devel
pgpabpsxfAUQT.pgp
Description: PGP signature
------------------------------------------------------------------------- This SF.net email is sponsored by DB2 Express Download DB2 Express C - the FREE version of DB2 express and take control of your XML. No limits. Just data. Click to get it now. http://sourceforge.net/powerbar/db2/
_______________________________________________ eCryptfs-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/ecryptfs-devel
