Messages in this thread | | | Date | Wed, 16 Jan 2008 17:12:32 +0000 | From | "Bryn M. Reeves" <> | Subject | Re: [PATCH 04/27] kill filp_open() |
| |
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Dave Hansen wrote: > On Wed, 2008-01-16 at 00:52 -0800, Andrew Morton wrote: >> On Thu, 01 Nov 2007 16:08:31 -0700 Dave Hansen <haveblue@us.ibm.com> wrote: >>> Replace all callers with open_namei() directly, and move the >>> nameidata stack allocation into open_namei(). >> ftp://ftp.kernel.org/pub/linux/kernel/people/agk/patches/2.6/editing/dm-loop.patch >> is using filp_open() and hence doesn't work very well. >> >> A shall revert dm-loop.patch and run away. > > This one's pretty easy, thank goodness. Just replace filp_open() with > open_namei(): > > /filp_open(/open_namei(AT_FDCWD, / > > BTW, why do we need this on top of the existing loopback driver? Can > they really share no code?
The current dm-loop patch was really written to demonstrate the value of block-mapping as an alternative to doing file backed I/O via the pagecache.
Personally, I would like to see a single "loopback block devices" driver in the kernel with loop.c and dm-loop as alternate interfaces to it.
There seems to be some demand for this, e.g.:
http://bugzilla.kernel.org/show_bug.cgi?id=5333
And numerous similar posts to dm-devel (although I think obsoleting may be going a bit far to start with!).
Regards, Bryn. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (GNU/Linux) Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org
iD8DBQFHjjsA6YSQoMYUY94RAv+fAJ90mbqDQ/17nd3j9OAIXXc8Y4SRbQCfR75I KtDxHXDnYmCmBlw/lOzGmms= =A+Vb -----END PGP SIGNATURE-----
| |