This request was evaluated by Red Hat Product Management for inclusion
in a Red Hat Enterprise Linux maintenance release. Product Management has 
requested further review of this request by Red Hat Engineering, for potential
inclusion in a Red Hat Enterprise Linux Update release for currently deployed 
products. This request is not yet committed for inclusion in an Update release.

I've started going over the fscache code in detail for the backport and have noticed some significant bugs in its implementation -- enough that I'm not comfortable turning this support on in 6.1. I'm going to push this bug out to 6.2, and work with Suresh to clean the code up.

I think we'll probably still want to take most of the fscache patches for 6.1, but plan to leave the Kconfig option disabled. Luckily most of the code gets compiled out to a no-op when it's disabled. That'll allow us to more easily merge bugfixes that depend on the changes in those patches without turning on support for it until it's ready.

I'll plan to pull in most of those patches in the set for bug 651865, and then leave this bug open to pull in the fixes and enable the kconfig option.

This request was evaluated by Red Hat Product Management for inclusion
in a Red Hat Enterprise Linux maintenance release. Product Management has 
requested further review of this request by Red Hat Engineering, for potential
inclusion in a Red Hat Enterprise Linux Update release for currently deployed 
products. This request is not yet committed for inclusion in an Update release.

Testing with a backport of the upstream code and then later the upstream code exposed at least one trivially reproducible oops when fsc is used. Based on that, I'm pulling the plug on this feature in RHEL6. Upstream discussion of this particular problem is here:

    http://article.gmane.org/gmane.linux.kernel.cifs/3466

We can reevaluate this feature for RHEL7 if there is demand for it.

Actually...there are some patches we need to take in in order to make sure that fscache is clearly disabled. Right now, if you specify -o fsc, the option is just ignored. We need to make sure that if someone specifies that, that the mount errors out properly. So, I'll convert this over to a bug to ensure that those patches go in.

How to test:

Attempt to mount a server with the '-o fsc' option. For instance:

    # mount //cifs-server.devel.redhat.com/export /mnt/cifs -o fsc

...assuming that the mount works without the fsc option, it will also succeed on RHEL6.1 with that option even though CONFIG_CIFS_FSCACHE is disabled. With the patches for this bug, you'll get a EINVAL error on the mount attempt and there will be a log message like this in dmesg:

    CIFS VFS: FS-Cache support needs CONFIG_CIFS_FSCACHE kernel config option set

Patch(es) available on kernel-2.6.32-160.el6

The bug is verified, details is list below:


[root@hp-ml370g5-01 statd]# uname -a
Linux hp-ml370g5-01.rhts.eng.bos.redhat.com 2.6.32-178.el6.x86_64 #1 SMP Wed Aug 3 10:43:13 EDT 2011 x86_64 x86_64 x86_64 GNU/Linux
[root@hp-ml370g5-01 statd]# rpm -q cifs-utils
cifs-utils-4.8.1-5.el6.x86_64
[root@hp-ml370g5-01 statd]# mount.cifs //dell-pesc1420-01.rhts.eng.bos.redhat.com/test /mnt/test -o user=root,password=redhat,fsc
mount error(22): Invalid argument
Refer to the mount.cifs(8) manual page (e.g. man mount.cifs)
[root@hp-ml370g5-01 statd]# dmesg | tail
**snip**
CIFS VFS: FS-Cache support needs CONFIG_CIFS_FSCACHE kernel config option set
CIFS VFS: cifs_mount failed w/return code = -22

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

http://rhn.redhat.com/errata/RHSA-2011-1530.html