From mboxrd@z Thu Jan  1 00:00:00 1970
Message-ID: <534D6117.4020506@tycho.nsa.gov>
Date: Tue, 15 Apr 2014 12:40:55 -0400
From: Stephen Smalley <sds@tycho.nsa.gov>
MIME-Version: 1.0
To: Richard Haines <richard_c_haines@btinternet.com>, selinux@tycho.nsa.gov,
 Eric Paris <eparis@redhat.com>, Daniel J Walsh <dwalsh@redhat.com>
Subject: Re: [PATCH] libsepol: Skip duplicate filename_trans rules in
 state->out policy.
References: <1397579252-1378-1-git-send-email-richard_c_haines@btinternet.com>
In-Reply-To: <1397579252-1378-1-git-send-email-richard_c_haines@btinternet.com>
Content-Type: text/plain; charset=ISO-8859-1
List-Id: "Security-Enhanced Linux \(SELinux\) mailing list"
 <selinux.tycho.nsa.gov>
List-Post: <mailto:selinux@tycho.nsa.gov>
List-Help: <mailto:selinux-request@tycho.nsa.gov?subject=help>

On 04/15/2014 12:27 PM, Richard Haines wrote:
> The current detection of duplicate rules does not cover the state->out
> policy and therefore will duplicate filename transition rules if already
> present.
> 
> Signed-off-by: Richard Haines <richard_c_haines@btinternet.com>
> ---
>  libsepol/src/expand.c | 14 ++++++++++++++
>  1 file changed, 14 insertions(+)
> 
> diff --git a/libsepol/src/expand.c b/libsepol/src/expand.c
> index acb6906..e908fdb 100644
> --- a/libsepol/src/expand.c
> +++ b/libsepol/src/expand.c
> @@ -1534,6 +1534,20 @@ static int expand_filename_trans(expand_state_t *state, filename_trans_rule_t *r
>  				if (cur_trans)
>  					continue;
>  
> +				/* Now check if duplicate rule in state->out policy */
> +				cur_trans = state->out->filename_trans;
> +
> +				while (cur_trans) {
> +					if (cur_trans->stype == (i + 1) &&
> +					    cur_trans->ttype == (j + 1) &&
> +					    cur_trans->tclass == cur_rule->tclass &&
> +					    !strcmp(cur_trans->name, cur_rule->name))
> +						break;
> +					cur_trans = cur_trans->next;
> +				}
> +				if (cur_trans)
> +					continue;
> +
>  				new_trans = malloc(sizeof(*new_trans));
>  				if (!new_trans) {
>  					ERR(state->handle, "Out of memory!");

Isn't this effectively a revert of:

commit a29f6820c52b60b9028298cde9962dd140bbf9ea
Author: Adam Tkac <atkac@redhat.com>
Date:   Fri May 25 17:55:08 2012 +0200

    libsepol: filename_trans: use some better sorting to compare and merge

The kernel switched to using a hashtab for filename_trans rules in
commit 2463c26d50adc282d19317013ba0ff473823ca47
Author: Eric Paris <eparis@redhat.com>
Date:   Thu Apr 28 15:11:21 2011 -0400

    SELinux: put name based create rules in a hashtable

Is there a reason we don't do this in libsepol too?