All of lore.kernel.org
 help / color / mirror / Atom feed
* FIPS mode in the kernel break CIFS mounts?
@ 2013-02-28  7:08 Suresh Jayaraman
       [not found] ` <512F027E.2030902-Et1tbQHTxzrQT0dZR+AlfA@public.gmane.org>
  0 siblings, 1 reply; 4+ messages in thread
From: Suresh Jayaraman @ 2013-02-28  7:08 UTC (permalink / raw)
  To: linux-cifs

Hi all,

I was told that (haven't checked myself, yet) that booting in FIPS
(Federal Information Processing Standard) mode by adding "fips=1" to
grub.conf breaks CIFS mounts by returning -ENOENT.

I thought I'd ask if this is a known problem before I spend more time on
this.


Thanks

-- 
Suresh Jayaraman

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: FIPS mode in the kernel break CIFS mounts?
       [not found] ` <512F027E.2030902-Et1tbQHTxzrQT0dZR+AlfA@public.gmane.org>
@ 2013-02-28 15:02   ` Jeff Layton
       [not found]     ` <20130228070245.6152ab77-4QP7MXygkU+dMjc06nkz3ljfA9RmPOcC@public.gmane.org>
  0 siblings, 1 reply; 4+ messages in thread
From: Jeff Layton @ 2013-02-28 15:02 UTC (permalink / raw)
  To: Suresh Jayaraman; +Cc: linux-cifs

On Thu, 28 Feb 2013 12:38:46 +0530
Suresh Jayaraman <sjayaraman-Et1tbQHTxzrQT0dZR+AlfA@public.gmane.org> wrote:

> Hi all,
> 
> I was told that (haven't checked myself, yet) that booting in FIPS
> (Federal Information Processing Standard) mode by adding "fips=1" to
> grub.conf breaks CIFS mounts by returning -ENOENT.
> 
> I thought I'd ask if this is a known problem before I spend more time on
> this.
> 
> 
> Thanks
> 

Yep, known problem. We try to load several crypto routines during the
mount process, and that fails in FIPS mode since they aren't allowed.
The fix is to only load those routines once we know that we actually
need them and only fail in those cases. There are no patches for that
yet, AFAIK.

Note that enabling FIPS mode will almost certainly break both NTLMv1
and NTLMv2 auth, since those require md4 and md5. It might be nice to
update the manpage with what sec= options actually work after enabling
FIPS mode once the main problem is fixed.

-- 
Jeff Layton <jlayton-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: FIPS mode in the kernel break CIFS mounts?
       [not found]     ` <20130228070245.6152ab77-4QP7MXygkU+dMjc06nkz3ljfA9RmPOcC@public.gmane.org>
@ 2013-02-28 15:10       ` Steve French
       [not found]         ` <CAH2r5mttfG6o7Nnz+uh7yW6jh8S4TCg2A43DGvgt36cBQ_57dA-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
  0 siblings, 1 reply; 4+ messages in thread
From: Steve French @ 2013-02-28 15:10 UTC (permalink / raw)
  To: Jeff Layton; +Cc: Suresh Jayaraman, linux-cifs

On Thu, Feb 28, 2013 at 9:02 AM, Jeff Layton <jlayton-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote:
> On Thu, 28 Feb 2013 12:38:46 +0530
> Suresh Jayaraman <sjayaraman-Et1tbQHTxzrQT0dZR+AlfA@public.gmane.org> wrote:
>
>> Hi all,
>>
>> I was told that (haven't checked myself, yet) that booting in FIPS
>> (Federal Information Processing Standard) mode by adding "fips=1" to
>> grub.conf breaks CIFS mounts by returning -ENOENT.
>>
>> I thought I'd ask if this is a known problem before I spend more time on
>> this.
>>
>>
>> Thanks
>>
>
> Yep, known problem. We try to load several crypto routines during the
> mount process, and that fails in FIPS mode since they aren't allowed.
> The fix is to only load those routines once we know that we actually
> need them and only fail in those cases. There are no patches for that
> yet, AFAIK.
>
> Note that enabling FIPS mode will almost certainly break both NTLMv1
> and NTLMv2 auth, since those require md4 and md5. It might be nice to
> update the manpage with what sec= options actually work after enabling
> FIPS mode once the main problem is fixed.

Any good reference to this?  Would be good to track this via a bug report.
Although presumably we would be fine with SMB2/SMB3 and krb5,
it wasn't immediately obvious to me why MD5 would not be allowed since
I see it in various lists of FIPS algorithm certifications but not for
RHEL 6.2 which listed the following:

-FIPS-approved algorithms: AES (Certs. #1968, #1969, #1970, #1971 and
#1972); Triple-DES (Certs. #1278 and #1279); SHS (Certs. #1725 and
#1726); HMAC (Certs. #1187, #1188, #1199 and #1200); RNG (Certs.
#1033, #1034, #1035, #1036 and #1037); DSA (Certs #628, #629, #634 and
#635)


-- 
Thanks,

Steve

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: FIPS mode in the kernel break CIFS mounts?
       [not found]         ` <CAH2r5mttfG6o7Nnz+uh7yW6jh8S4TCg2A43DGvgt36cBQ_57dA-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
@ 2013-02-28 16:25           ` Jeff Layton
  0 siblings, 0 replies; 4+ messages in thread
From: Jeff Layton @ 2013-02-28 16:25 UTC (permalink / raw)
  To: Steve French; +Cc: Suresh Jayaraman, linux-cifs

On Thu, 28 Feb 2013 09:10:28 -0600
Steve French <smfrench-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:

> On Thu, Feb 28, 2013 at 9:02 AM, Jeff Layton <jlayton-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote:
> > On Thu, 28 Feb 2013 12:38:46 +0530
> > Suresh Jayaraman <sjayaraman-Et1tbQHTxzrQT0dZR+AlfA@public.gmane.org> wrote:
> >
> >> Hi all,
> >>
> >> I was told that (haven't checked myself, yet) that booting in FIPS
> >> (Federal Information Processing Standard) mode by adding "fips=1" to
> >> grub.conf breaks CIFS mounts by returning -ENOENT.
> >>
> >> I thought I'd ask if this is a known problem before I spend more time on
> >> this.
> >>
> >>
> >> Thanks
> >>
> >
> > Yep, known problem. We try to load several crypto routines during the
> > mount process, and that fails in FIPS mode since they aren't allowed.
> > The fix is to only load those routines once we know that we actually
> > need them and only fail in those cases. There are no patches for that
> > yet, AFAIK.
> >
> > Note that enabling FIPS mode will almost certainly break both NTLMv1
> > and NTLMv2 auth, since those require md4 and md5. It might be nice to
> > update the manpage with what sec= options actually work after enabling
> > FIPS mode once the main problem is fixed.
> 
> Any good reference to this?  Would be good to track this via a bug report.
> Although presumably we would be fine with SMB2/SMB3 and krb5,
> it wasn't immediately obvious to me why MD5 would not be allowed since
> I see it in various lists of FIPS algorithm certifications but not for
> RHEL 6.2 which listed the following:
> 
> -FIPS-approved algorithms: AES (Certs. #1968, #1969, #1970, #1971 and
> #1972); Triple-DES (Certs. #1278 and #1279); SHS (Certs. #1725 and
> #1726); HMAC (Certs. #1187, #1188, #1199 and #1200); RNG (Certs.
> #1033, #1034, #1035, #1036 and #1037); DSA (Certs #628, #629, #634 and
> #635)
> 
> 

I'm pretty sure md5 is proscribed by FIPS, though all the info I have
on this is second hand. The fact that md5 was blacklisted was one of
the reasons I avoided using it for the nfsd reboot recovery work that
was recently merged.

-- 
Jeff Layton <jlayton-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>

^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2013-02-28 16:25 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2013-02-28  7:08 FIPS mode in the kernel break CIFS mounts? Suresh Jayaraman
     [not found] ` <512F027E.2030902-Et1tbQHTxzrQT0dZR+AlfA@public.gmane.org>
2013-02-28 15:02   ` Jeff Layton
     [not found]     ` <20130228070245.6152ab77-4QP7MXygkU+dMjc06nkz3ljfA9RmPOcC@public.gmane.org>
2013-02-28 15:10       ` Steve French
     [not found]         ` <CAH2r5mttfG6o7Nnz+uh7yW6jh8S4TCg2A43DGvgt36cBQ_57dA-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2013-02-28 16:25           ` Jeff Layton

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.