[PATCHES] IFP: support views

[Pavel Březina]

Since the NSS and PAM parts of the views are in the review process and I 
don't expect any big changes there, I'm sending IFP part. It depends on 
Sumit's patches.

[Sumit Bose]

On Fri, Oct 17, 2014 at 02:09:11PM +0200, Pavel Březina wrote:
> Since the NSS and PAM parts of the views are in the review process and I
> don't expect any big changes there, I'm sending IFP part. It depends on
> Sumit's patches.
> Subject: [PATCH 1/5] Add sysdb_search_[user|group]_override_attrs_by_name

ACK
> From 693a2c4392424e95acef1e38647fc5cf67f50f71 Mon Sep 17 00:00:00 2001
> From: =?UTF-8?q?Pavel=20B=C5=99ezina?= 
> Date: Thu, 2 Oct 2014 16:13:53 +0200
> Subject: [PATCH 2/5] Add sysdb_get_user_attr_with_views

...
> +int sysdb_get_user_attr_with_views(TALLOC_CTX *mem_ctx,
>                               struct sss_domain_info *domain,
>                               const char *name,
>                               const char **attributes,
>                               struct ldb_result **_res)
>
> +{
> int ret;
> struct ldb_result *orig_obj = NULL;
> struct ldb_result *override_obj = NULL;
> struct ldb_message_element *el = NULL;
> const char **attrs = NULL;
> bool has_override_dn;
> TALLOC_CTX *tmp_ctx;
> int count;
> tmp_ctx = talloc_new(NULL);
> if (tmp_ctx == NULL) {
>    DEBUG(SSSDBG_OP_FAILURE, "talloc_new failed.\n");
>    return ENOMEM;
> }
> /* We need overrideDN for views, so append it if missing. */

I think you can safely only do this if DOM_HAS_VIEWS(domain).
> has_override_dn = false;
> for (count = 0; attributes[count] != NULL; count++) {
>    if (strcmp(attributes[count], SYSDB_OVERRIDE_DN) == 0) {
>        has_override_dn = true;
>        break;
>    }
> }
> if (has_override_dn) {
>    /* Use original attributes. */
>    attrs = attributes;
> } else {
>    /* Copy original attributes and add overrideDN. */
>    attrs = talloc_zero_array(tmp_ctx, const char *, count + 2);
>    if (attrs == NULL) {
>        ret = ENOMEM;
>        goto done;
>    }
>    for (count = 0; attributes[count] != NULL; count++) {
>        attrs[count] = attributes[count];
>    }
>    attrs[count] = SYSDB_OVERRIDE_DN;
> }
> Subject: [PATCH 3/5] IFP: support views

ACK

bye,
Sumit

[Pavel Březina]

On 10/20/2014 01:31 PM, Sumit Bose wrote:
> On Fri, Oct 17, 2014 at 02:09:11PM +0200, Pavel Březina wrote:
> > Since the NSS and PAM parts of the views are in the review process and I
> > don't expect any big changes there, I'm sending IFP part. It depends on
> > Sumit's patches.
> > Subject: [PATCH 1/5] Add sysdb_search_[user|group]_override_attrs_by_name
>
> ACK
> > From 693a2c4392424e95acef1e38647fc5cf67f50f71 Mon Sep 17 00:00:00 2001
> > From: =?UTF-8?q?Pavel=20B=C5=99ezina?= 
> > Date: Thu, 2 Oct 2014 16:13:53 +0200
> > Subject: [PATCH 2/5] Add sysdb_get_user_attr_with_views
>
> ...
> > +int sysdb_get_user_attr_with_views(TALLOC_CTX *mem_ctx,
> >                               struct sss_domain_info *domain,
> >                               const char *name,
> >                               const char **attributes,
> >                               struct ldb_result **_res)
> >
> > +{
> > int ret;
> > struct ldb_result *orig_obj = NULL;
> > struct ldb_result *override_obj = NULL;
> > struct ldb_message_element *el = NULL;
> > const char **attrs = NULL;
> > bool has_override_dn;
> > TALLOC_CTX *tmp_ctx;
> > int count;
> > tmp_ctx = talloc_new(NULL);
> > if (tmp_ctx == NULL) {
> >    DEBUG(SSSDBG_OP_FAILURE, "talloc_new failed.\n");
> >    return ENOMEM;
> > }
> > /* We need overrideDN for views, so append it if missing. */
>
> I think you can safely only do this if DOM_HAS_VIEWS(domain).
> > has_override_dn = false;
> > for (count = 0; attributes[count] != NULL; count++) {
> >    if (strcmp(attributes[count], SYSDB_OVERRIDE_DN) == 0) {
> >        has_override_dn = true;
> >        break;
> >    }
> > }
> > if (has_override_dn) {
> >    /* Use original attributes. */
> >    attrs = attributes;
> > } else {
> >    /* Copy original attributes and add overrideDN. */
> >    attrs = talloc_zero_array(tmp_ctx, const char *, count + 2);
> >    if (attrs == NULL) {
> >        ret = ENOMEM;
> >        goto done;
> >    }
> >    for (count = 0; attributes[count] != NULL; count++) {
> >        attrs[count] = attributes[count];
> >    }
> >    attrs[count] = SYSDB_OVERRIDE_DN;
> > }
> > Subject: [PATCH 3/5] IFP: support views
>
> ACK
>
> bye,
> Sumit

That makes sense. New patches are attached.

[Sumit Bose]

On Mon, Oct 20, 2014 at 03:13:18PM +0200, Pavel Březina wrote:
> On 10/20/2014 01:31 PM, Sumit Bose wrote:
> > On Fri, Oct 17, 2014 at 02:09:11PM +0200, Pavel Březina wrote:
> > > Since the NSS and PAM parts of the views are in the review process and I
> > > don't expect any big changes there, I'm sending IFP part. It depends on
> > > Sumit's patches.
> > > Subject: [PATCH 1/5] Add sysdb_search_[user|group]_override_attrs_by_name
> >
> > ACK
> > > From 693a2c4392424e95acef1e38647fc5cf67f50f71 Mon Sep 17 00:00:00 2001
> > > From: =?UTF-8?q?Pavel=20B=C5=99ezina?= 
> > > Date: Thu, 2 Oct 2014 16:13:53 +0200
> > > Subject: [PATCH 2/5] Add sysdb_get_user_attr_with_views
> >
> > ...
> > > +int sysdb_get_user_attr_with_views(TALLOC_CTX *mem_ctx,
> > >                               struct sss_domain_info *domain,
> > >                               const char *name,
> > >                               const char **attributes,
> > >                               struct ldb_result **_res)
> > >
> > > +{
> > > int ret;
> > > struct ldb_result *orig_obj = NULL;
> > > struct ldb_result *override_obj = NULL;
> > > struct ldb_message_element *el = NULL;
> > > const char **attrs = NULL;
> > > bool has_override_dn;
> > > TALLOC_CTX *tmp_ctx;
> > > int count;
> > > tmp_ctx = talloc_new(NULL);
> > > if (tmp_ctx == NULL) {
> > >    DEBUG(SSSDBG_OP_FAILURE, "talloc_new failed.\n");
> > >    return ENOMEM;
> > > }
> > > /* We need overrideDN for views, so append it if missing. */
> >
> > I think you can safely only do this if DOM_HAS_VIEWS(domain).
> > > has_override_dn = false;
> > > for (count = 0; attributes[count] != NULL; count++) {
> > >    if (strcmp(attributes[count], SYSDB_OVERRIDE_DN) == 0) {
> > >        has_override_dn = true;
> > >        break;
> > >    }
> > > }
> > > if (has_override_dn) {
> > >    /* Use original attributes. */
> > >    attrs = attributes;
> > > } else {
> > >    /* Copy original attributes and add overrideDN. */
> > >    attrs = talloc_zero_array(tmp_ctx, const char *, count + 2);
> > >    if (attrs == NULL) {
> > >        ret = ENOMEM;
> > >        goto done;
> > >    }
> > >    for (count = 0; attributes[count] != NULL; count++) {
> > >        attrs[count] = attributes[count];
> > >    }
> > >    attrs[count] = SYSDB_OVERRIDE_DN;
> > > }
> > > Subject: [PATCH 3/5] IFP: support views
> >
> > ACK
> >
> > bye,
> > Sumit
>
> That makes sense. New patches are attached.

ACK

bye,
Sumit

[Jakub Hrozek]

On Mon, Oct 20, 2014 at 04:06:11PM +0200, Sumit Bose wrote:
> ACK
>
> bye,
> Sumit

Pushed to master:
* 1c00b0c92189433e1fbb7154fb8df71c7788245e
* cc5f7592e4d81f3a7336da20fc681b7e52c103b4
* 727d46f4dace666c809310b3f685eef387023f65