Re: Labeled NFS [v5]

[Casey Schaufler <casey@schaufler-ca.com>]

On 11/29/2012 4:46 PM, David Quigley wrote:
> On 11/29/2012 19:34, Casey Schaufler wrote:
>> On 11/29/2012 4:07 PM, David Quigley wrote:
>>> On 11/29/2012 17:28, Casey Schaufler wrote:
>>>> On 11/28/2012 6:08 PM, Casey Schaufler wrote:
>>>>> On 11/28/2012 5:14 PM, Dave Quigley wrote:
>>>>>> On 11/28/2012 1:57 PM, Casey Schaufler wrote:
>>>>>>> On 11/20/2012 7:28 PM, Dave Quigley wrote:
>>>>>>>> On 11/20/2012 9:52 PM, Casey Schaufler wrote:
>>>>>>>>> On 11/20/2012 4:37 PM, Dave Quigley wrote:
>>>>>>>>>> ...
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> Or I could just give you this link and you should be good to
>>>>>>>>>> go ;)
>>>>>>>>>>
>>>>>>>>>> http://www.selinuxproject.org/~dpquigl/nfs-utils-rpms/
>>>>>>>>>>
>>>>>>>>>> I haven't tried it but it should work. If it doesn't let me
>>>>>>>>>> know and
>>>>>>>>>> i'll try to fix it on my end. I'd imagine you might need to yum
>>>>>>>>>> remove
>>>>>>>>>> nfs-utils first before adding this new one or you could also
>>>>>>>>>> try an
>>>>>>>>>> rpm with the upgrade flag for this instead. Good luck.
>>>>>>> ...
>>>>>>>
>>>>>>>
>>>>>>> I've tried on Fedora17 and Ubuntu12.04, and I'm getting the
>>>>>>> attached stack trace on mount. After mounting I'm getting
>>>>>>> denials when I should, but also when I shouldn't.
>>>>>>>
>>>>>>> I've tried tracking down the issue, but there's a lot going on
>>>>>>> that I don't find obvious. I added a dentry_init hook just for
>>>>>>> grins, but it's not getting called.
>>>>>>>
>>>>>>> .
>>>>>>>
>>>>>>>
>>>>>> Any chance of you throwing a kickstart file my way that's configured
>>>>>> with SMACK so I can use it for a test box (both server and
>>>>>> client)? I
>>>>>> can have the guys working with me test for SMACK as well if you
>>>>>> provide an appropriate test harness and image for testing.
>>>>> I've attached the .config from my Fedora17 machine. Who knows, maybe
>>>>> I got something wrong there. I get the error doing the test on the
>>>>> loopback interface (mount -t nfs4 localhist:/ /mnt).
>>>>
>>>> I've done some instrumentation and security_ismaclabel() is getting
>>>> called with "selinux", but never "SMACK64". I would guess that
>>>> somewhere
>>>> in the tools you're telling the kernel to expect "selinux". Where is
>>>> that, so that I can tell it to try "SMACK64" instead?
>>>>
>>>>
>>>>
>>>> -- 
>>>> This message was distributed to subscribers of the selinux mailing
>>>> list.
>>>> If you no longer wish to subscribe, send mail to
>>>> majordomo@tycho.nsa.gov with
>>>> the words "unsubscribe selinux" without quotes as the message.
>>>
>>>
>>> What tools do you use in SMACK to see the labels?
>>
>> attr -S -g SMACK64 <path>
>
> ok so that seems to work for SELinux as well. Never knew about that.
> I'd always just rip the xattr out of the inode with getfattr.
>
>
>>
>>> Do you just use getxattr? If so can you try calling that and seeing
>>> what happens? I'm concerned that you aren't getting any attribute
>>> information on that file.
>>
>> I would think that were it not for the case that access is denied
>> and I get an audit record for nfsd that reports a subject label of "_"
>> (which is correct for nfsd but not the process attempting access) and
>> an object label of "WhooHoo", which is correct. The server side
>> looks like it might be working right, given the information that it
>> has.
>>
>
> Ok so this is the problem. nfsd is a kernel thread I believe. In
> SELinux land it has the type kernel_t which is all powerful. We don't
> have client label transport yet (That requires RPCSECGSSv3). Is there
> a way you can have that kernel thread running as a type that has
> access to everything?

That would be having CAP_MAC_ADMIN and CAP_MAC_OVERRIDE in Smackese.
Looking at /proc/<pid-of-nfsd>/status we see CapEff of fff...fff which
is to say, all capabilities.
 

> I think that is the current problem. Which makes perfect sense. If
> your kernel threads don't get started with max privilege then the
> server would be denied access on all of the file attributes and
> wouldn't be able to ship it over the wire properly.

OK. I haven't had to do anything with kernel threads so far.
Where is NFS setting these up? Poking around fs/nfsd looks like
the place, but I haven't seen anything there that makes it look
like they would be running without capabilities. Clearly, that's
what I'm seeing. It looks as if the credential of nfsd does not
match what /proc reports. Bother.


> I'm not sure what you need to do but you'll probably have to work this
> out. We have a usage mode in the IETF spec which has a non-mac
> enforcing server which still support object labeling. In the SELinux
> case it works for us since kernel_t can access anything. Ideally when
> RPCSECGSSv3 is finished and merged we'll be able to choose whether to
> use the label of the process on the client side or kernel_t for the
> server if its not available.
>
>>> Do you have a disto that I can use that has full smack integration and
>>> is easy to setup?
>>
>> There's no full integration, but Ubuntu is easy to set up because they
>> compile in all the LSMs.
>> Set "security=smack" on the boot line in grub.cfg and reboot.
>>
>> All processes and files will get the floor ("_") label unless you change
>> one. You can change
>> a file label with:
>>      # attr -S -s SMACK64 WhooHoo path
>> or execute at a different label with:
>>     # (echo WhooHoo > /proc/self/attr/current ; command)
>>
>
> I'm not out of here until really late tonight so getting an Ubuntu VM
> setup probably won't happen until sometime next week when everything
> calms down. However I think we isolated the problem above. If I'm
> correct this is strictly a smack labeling problem. I don't know if you
> need to put some code into smack to init kernel threads with a more
> powerful label or not so I'll leave it up to you on how to address this.
>
>
> Dave
>