2014-02-26 13:46:56

by Lorenz Bucher

[permalink] [raw]
Subject: [PATCH] cifs: connect: added option krb5mi which may allow package signing

Solves an issue with dfs where servers using package signing are
mixed with servers which don't use package signing.

Signed-off-by: Lorenz Bucher <[email protected]>
---
fs/cifs/connect.c | 3 +++
1 file changed, 3 insertions(+)

diff --git a/fs/cifs/connect.c b/fs/cifs/connect.c
index 56c152d..9d06dd8 100644
--- a/fs/cifs/connect.c
+++ b/fs/cifs/connect.c
@@ -1104,6 +1104,9 @@ cifs_parse_mount_options(const char *mountdata, const char *devname,
if (!value || !*value) {
cERROR(1, "no security value specified");
continue;
+ } else if (strnicmp(value, "krb5mi", 6) == 0) {
+ vol->secFlg |= CIFSSEC_MAY_KRB5 |
+ CIFSSEC_MAY_SIGN;
} else if (strnicmp(value, "krb5i", 5) == 0) {
vol->secFlg |= CIFSSEC_MAY_KRB5 |
CIFSSEC_MUST_SIGN;
--
1.7.9.5


2014-02-26 20:26:44

by Jeffrey Layton

[permalink] [raw]
Subject: Re: [PATCH] cifs: connect: added option krb5mi which may allow package signing

On Wed, 26 Feb 2014 14:46:47 +0100
"Lorenz Bucher" <[email protected]> wrote:

> Solves an issue with dfs where servers using package signing are
> mixed with servers which don't use package signing.
>
> Signed-off-by: Lorenz Bucher <[email protected]>
> ---
> fs/cifs/connect.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/fs/cifs/connect.c b/fs/cifs/connect.c
> index 56c152d..9d06dd8 100644
> --- a/fs/cifs/connect.c
> +++ b/fs/cifs/connect.c
> @@ -1104,6 +1104,9 @@ cifs_parse_mount_options(const char *mountdata, const char *devname,
> if (!value || !*value) {
> cERROR(1, "no security value specified");
> continue;
> + } else if (strnicmp(value, "krb5mi", 6) == 0) {
> + vol->secFlg |= CIFSSEC_MAY_KRB5 |
> + CIFSSEC_MAY_SIGN;
> } else if (strnicmp(value, "krb5i", 5) == 0) {
> vol->secFlg |= CIFSSEC_MAY_KRB5 |
> CIFSSEC_MUST_SIGN;

This looks like it applies to a really old kernel. I think that problem
was what commit 0b7bc84000d71f3647ca33ab1bf5bd928535c846 was intended
to fix.

--
Jeff Layton <[email protected]>

2014-02-27 09:13:21

by Lorenz Bucher

[permalink] [raw]
Subject: Aw: Re: [PATCH] cifs: connect: added option krb5mi which may allow package signing


 
 

Gesendet: Mittwoch, 26. Februar 2014 um 21:26 Uhr
Von: "Jeff Layton" <[email protected]>
An: "Lorenz Bucher" <[email protected]>
Cc: [email protected], [email protected], [email protected], [email protected]
Betreff: Re: [PATCH] cifs: connect: added option krb5mi which may allow package signing
On Wed, 26 Feb 2014 14:46:47 +0100
"Lorenz Bucher" <[email protected]> wrote:

> Solves an issue with dfs where servers using package signing are
> mixed with servers which don't use package signing.
>
> Signed-off-by: Lorenz Bucher <[email protected]>
> ---
> fs/cifs/connect.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/fs/cifs/connect.c b/fs/cifs/connect.c
> index 56c152d..9d06dd8 100644
> --- a/fs/cifs/connect.c
> +++ b/fs/cifs/connect.c
> @@ -1104,6 +1104,9 @@ cifs_parse_mount_options(const char *mountdata, const char *devname,
> if (!value || !*value) {
> cERROR(1, "no security value specified");
> continue;
> + } else if (strnicmp(value, "krb5mi", 6) == 0) {
> + vol->secFlg |= CIFSSEC_MAY_KRB5 |
> + CIFSSEC_MAY_SIGN;
> } else if (strnicmp(value, "krb5i", 5) == 0) {
> vol->secFlg |= CIFSSEC_MAY_KRB5 |
> CIFSSEC_MUST_SIGN;

This looks like it applies to a really old kernel. I think that problem
was what commit 0b7bc84000d71f3647ca33ab1bf5bd928535c846 was intended
to fix.

--
Jeff Layton <[email protected]>

Its the current kernel from LTS Ubuntu 12.04
3.2.0-59-generic #90-Ubuntu SMP Tue Jan 7 22:43:51 UTC 2014

2014-02-27 15:03:30

by Jeffrey Layton

[permalink] [raw]
Subject: Re: [PATCH] cifs: connect: added option krb5mi which may allow package signing

On Thu, 27 Feb 2014 10:13:14 +0100
"Lorenz Bucher" <[email protected]> wrote:

>
>  
>  
>
> Gesendet: Mittwoch, 26. Februar 2014 um 21:26 Uhr
> Von: "Jeff Layton" <[email protected]>
> An: "Lorenz Bucher" <[email protected]>
> Cc: [email protected], [email protected], [email protected], [email protected]
> Betreff: Re: [PATCH] cifs: connect: added option krb5mi which may allow package signing
> On Wed, 26 Feb 2014 14:46:47 +0100
> "Lorenz Bucher" <[email protected]> wrote:
>
> > Solves an issue with dfs where servers using package signing are
> > mixed with servers which don't use package signing.
> >
> > Signed-off-by: Lorenz Bucher <[email protected]>
> > ---
> > fs/cifs/connect.c | 3 +++
> > 1 file changed, 3 insertions(+)
> >
> > diff --git a/fs/cifs/connect.c b/fs/cifs/connect.c
> > index 56c152d..9d06dd8 100644
> > --- a/fs/cifs/connect.c
> > +++ b/fs/cifs/connect.c
> > @@ -1104,6 +1104,9 @@ cifs_parse_mount_options(const char *mountdata, const char *devname,
> > if (!value || !*value) {
> > cERROR(1, "no security value specified");
> > continue;
> > + } else if (strnicmp(value, "krb5mi", 6) == 0) {
> > + vol->secFlg |= CIFSSEC_MAY_KRB5 |
> > + CIFSSEC_MAY_SIGN;
> > } else if (strnicmp(value, "krb5i", 5) == 0) {
> > vol->secFlg |= CIFSSEC_MAY_KRB5 |
> > CIFSSEC_MUST_SIGN;
>
> This looks like it applies to a really old kernel. I think that problem
> was what commit 0b7bc84000d71f3647ca33ab1bf5bd928535c846 was intended
> to fix.
>
> --
> Jeff Layton <[email protected]>
>
> Its the current kernel from LTS Ubuntu 12.04
> 3.2.0-59-generic #90-Ubuntu SMP Tue Jan 7 22:43:51 UTC 2014
>

Yep, that kernel is pretty old.

You're probably best off asking the ubuntu kernel maintainers to take
that patch (or asking them to backport 0b7bc84000d). It should
be reasonably safe but they'll have to make that call.

--
Jeff Layton <[email protected]>