linux-nfs.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* rpc.gssd and proxiable/forwardable tickets.
@ 2010-10-29  8:40 Marc Schlinger
  2010-11-01 23:47 ` Kevin Coffman
  0 siblings, 1 reply; 2+ messages in thread
From: Marc Schlinger @ 2010-10-29  8:40 UTC (permalink / raw)
  To: linux-nfs


Hello,

I'm using WebAuth to authenticate my user and provide them a mean to
join their NFSv4 files through a web page.
I'd  like to have the kerberos credentials used by the web server, but I
didn't managed to impersonate the kerberos user with nfsv4 in a webauth
protected page.
When I try to list the an nfs directory  from the webpage I've got this
error from rpc.gssd:

CC file '/tmp/krb5cc_500' is expired or corrupt

My distribution is Fedora 12 and i'm using nfs-utils 1.2.1.

WebAuth is configured to ask the client a forwardable ticket for
nfs/<mynfsserver>@<myrealm>. In my application's code I can see the
ticket and even do a klist with it. The output looks like this:

$ klist
Ticket cache: FILE:/tmp/krb5cc_500
Default principal: marc@<myrealm>

Valid starting     Expires            Service principal
10/28/10 20:15:17  10/29/10 20:15:15  nfs/<mynfsserver>@<myrealm>
     Flags: FAT

So my application never gets the krbtgt tickets. Considering security, I
believe this is a good point.

I must confess that I didn't manage to follow rpc.gssd process with gdb
or with ltrace.
So until I'm able to trace gssd execution all things that follows are
pure suppositions.


While trying to find a valid credential_cache gssd calls a function in
utils/krb5_utils.c, "check_for_tgt", that does this loop:

     while (!found&&   (ret = krb5_cc_next_cred(context, ccache,&cur,
&creds)) == 0) {
         if (creds.server->length == 2&&
                 data_is_equal(creds.server->realm, principal->realm)&&
                 creds.server->data[0].length == 6&&
->              memcmp(creds.server->data[0].data, "krbtgt", 6) == 0&&
                 data_is_equal(creds.server->data[1], principal->realm)&&
                 creds.times.endtime>   time(NULL))
             found = 1;
         krb5_free_cred_contents(context,&creds);
     }


What I understand is that without a krbtgt entry, a credential cache
will be considered invalid.

Is there some reasons for this?
For what I've understand about kerberos protocol, a proxiable or
forwardable service ticket is sufficient to communicate with the nfs
server. But I may be wrong.


Thanks for your help.

Marc Schlinger





^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: rpc.gssd and proxiable/forwardable tickets.
  2010-10-29  8:40 rpc.gssd and proxiable/forwardable tickets Marc Schlinger
@ 2010-11-01 23:47 ` Kevin Coffman
  0 siblings, 0 replies; 2+ messages in thread
From: Kevin Coffman @ 2010-11-01 23:47 UTC (permalink / raw)
  To: Marc Schlinger; +Cc: linux-nfs

Hello Marc,
This sounds like a bug.  This should be considered a valid credentials
cache (without a TGT).  I don't have the cycles to attempt a fix, nor
am I sure what the correct fix would be.  I hope someone else does!

K.C.

On Fri, Oct 29, 2010 at 4:40 AM, Marc Schlinger
<marc.schlinger@agorabox.org> wrote:
>
> Hello,
>
> I'm using WebAuth to authenticate my user and provide them a mean to
> join their NFSv4 files through a web page.
> I'd  like to have the kerberos credentials used by the web server, but I
> didn't managed to impersonate the kerberos user with nfsv4 in a webauth
> protected page.
> When I try to list the an nfs directory  from the webpage I've got this
> error from rpc.gssd:
>
> CC file '/tmp/krb5cc_500' is expired or corrupt
>
> My distribution is Fedora 12 and i'm using nfs-utils 1.2.1.
>
> WebAuth is configured to ask the client a forwardable ticket for
> nfs/<mynfsserver>@<myrealm>. In my application's code I can see the
> ticket and even do a klist with it. The output looks like this:
>
> $ klist
> Ticket cache: FILE:/tmp/krb5cc_500
> Default principal: marc@<myrealm>
>
> Valid starting     Expires            Service principal
> 10/28/10 20:15:17  10/29/10 20:15:15  nfs/<mynfsserver>@<myrealm>
>    Flags: FAT
>
> So my application never gets the krbtgt tickets. Considering security, I
> believe this is a good point.
>
> I must confess that I didn't manage to follow rpc.gssd process with gdb
> or with ltrace.
> So until I'm able to trace gssd execution all things that follows are
> pure suppositions.
>
>
> While trying to find a valid credential_cache gssd calls a function in
> utils/krb5_utils.c, "check_for_tgt", that does this loop:
>
>    while (!found&&   (ret = krb5_cc_next_cred(context, ccache,&cur,
> &creds)) == 0) {
>        if (creds.server->length == 2&&
>                data_is_equal(creds.server->realm, principal->realm)&&
>                creds.server->data[0].length == 6&&
> ->              memcmp(creds.server->data[0].data, "krbtgt", 6) == 0&&
>                data_is_equal(creds.server->data[1], principal->realm)&&
>                creds.times.endtime>   time(NULL))
>            found = 1;
>        krb5_free_cred_contents(context,&creds);
>    }
>
>
> What I understand is that without a krbtgt entry, a credential cache
> will be considered invalid.
>
> Is there some reasons for this?
> For what I've understand about kerberos protocol, a proxiable or
> forwardable service ticket is sufficient to communicate with the nfs
> server. But I may be wrong.
>
>
> Thanks for your help.
>
> Marc Schlinger
>
>
>
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
>

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2010-11-01 23:47 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2010-10-29  8:40 rpc.gssd and proxiable/forwardable tickets Marc Schlinger
2010-11-01 23:47 ` Kevin Coffman

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).