SELinux Archive on lore.kernel.org
 help / color / Atom feed
From: James Carter <jwcart2@gmail.com>
To: Dominick Grift <dominick.grift@defensec.nl>
Cc: SElinux list <selinux@vger.kernel.org>
Subject: Re: [PATCH v3] secilc/docs: document expandtypeattribute
Date: Fri, 31 Jul 2020 16:22:44 -0400
Message-ID: <CAP+JOzSgp1Xgv29v__SAjZthonqRw=ZmYJ_3=MZm5FX8EOjDBw@mail.gmail.com> (raw)
In-Reply-To: <ypjltuxnla05.fsf@defensec.nl>

On Fri, Jul 31, 2020 at 4:12 PM Dominick Grift
<dominick.grift@defensec.nl> wrote:
>
> James Carter <jwcart2@gmail.com> writes:
>
> > On Fri, Jul 31, 2020 at 3:26 AM Dominick Grift
> > <dominick.grift@defensec.nl> wrote:
> >>
> >> James Carter <jwcart2@gmail.com> writes:
> >>
> >> > On Thu, Jul 30, 2020 at 9:14 AM Dominick Grift
> >> > <dominick.grift@defensec.nl> wrote:
> >> >>
> >> >> This was added for Androids Treble in 2017.
> >> >>
> >> >> I was unsure whether this belongs in type_statements or in conditional_statements.
> >> >>
> >> >
> >> > I think that it fits best with the type statements as you have it.
> >> >
> >> >> Signed-off-by: Dominick Grift <dominick.grift@defensec.nl>
> >> >> ---
> >> >> v2: overriden is overridden
> >> >> v3: add link to README.md
> >> >>
> >> >>  secilc/docs/README.md              |  1 +
> >> >>  secilc/docs/cil_type_statements.md | 38 ++++++++++++++++++++++++++++++
> >> >>  2 files changed, 39 insertions(+)
> >> >>
> >> >> diff --git a/secilc/docs/README.md b/secilc/docs/README.md
> >> >> index 3f1838e6..efab2a71 100644
> >> >> --- a/secilc/docs/README.md
> >> >> +++ b/secilc/docs/README.md
> >> >> @@ -126,6 +126,7 @@ CIL (Common Intermediate Language)
> >> >>    * [typealiasactual](cil_type_statements.md#typealiasactual)
> >> >>    * [typeattribute](cil_type_statements.md#typeattribute)
> >> >>    * [typeattributeset](cil_type_statements.md#typeattributeset)
> >> >> +  * [expandtypeattribute](cil_type_statements.md#expandtypeattribute)
> >> >>    * [typebounds](cil_type_statements.md#typebounds)
> >> >>    * [typechange](cil_type_statements.md#typechange)
> >> >>    * [typemember](cil_type_statements.md#typemember)
> >> >> diff --git a/secilc/docs/cil_type_statements.md b/secilc/docs/cil_type_statements.md
> >> >> index f9dd3a76..f819b3c6 100644
> >> >> --- a/secilc/docs/cil_type_statements.md
> >> >> +++ b/secilc/docs/cil_type_statements.md
> >> >> @@ -213,6 +213,44 @@ This example is equivalent to `{ domain -kernel.process -ueventd.process -init.p
> >> >>          )
> >> >>      )
> >> >>
> >> >> +expandtypeattribute
> >> >> +-------------------
> >> >> +
> >> >> +Allows expansion compiler defaults for one or more previously declared [`typeattribute`](cil_type_statements.md#typeattribute) identifiers to be overridden.
> >> >
> >> > The wording confused me at first.
> >> > I think "Overrides the compiler defaults for the expansion of one ...
> >> > identifiers." would be clearer.
> >> >
> >> >> +
> >> >> +**Statement definition:**
> >> >> +
> >> >> +    (expandtypeattribute typeattribute_id true|false)
> >> >> +
> >> >> +**Where:**
> >> >> +
> >> >> +<table>
> >> >> +<colgroup>
> >> >> +<col width="25%" />
> >> >> +<col width="75%" />
> >> >> +</colgroup>
> >> >> +<tbody>
> >> >> +<tr class="odd">
> >> >> +<td align="left"><p><code>expandtypeattribute</code></p></td>
> >> >> +<td align="left"><p>The <code>expandtypeattribute</code> keyword.</p></td>
> >> >> +</tr>
> >> >> +<tr class="even">
> >> >> +<td align="left"><p><code>typeattribute_id</code></p></td>
> >> >> +<td align="left"><p>One or more previously declared <code>typeattribute</code> identifiers.</p></td>
> >> >> +</tr>
> >> >> +<tr class="odd">
> >> >> +<td align="left"><p><code>true | false</code></p></td>
> >> >> +<td align="left"><p>Either true or false.</p></td>
> >> >> +</tr>
> >> >> +</tbody>
> >> >> +</table>
> >> >> +
> >> >> +**Example:**
> >> >> +
> >> >> +This example will use the expandtypeattribute statement to forcibly expand a previously declared `domain` type attribute.
> >> >> +
> >> >> +    (expandtypeattribute domain true)
> >> >> +
> >> >
> >> > It would be nice to have another example that shows a list of type
> >> > attributes, so there is an example of that syntax as well.
> >>
> >> I was looking into an example but turns out that either i am
> >> misunderstanding this functionality or that it does not work as advertised:
> >>
> >> Example:
> >>
> >> 1. compiler defaults to expand typeattributes with less
> >> than four members
> >> 2. override pets and dogs expansion in policy
> >>
> >> # cat > mytest.cil <<EOF
> >> (sid mysid)
> >> (sidorder (mysid))
> >>
> >> (class myclass (mypermission))
> >> (classorder (unordered myclass))
> >>
> >> (type blue)
> >> (type green)
> >> (type red)
> >> (typeattribute colors)
> >> (typeattributeset colors (blue green red))
> >>
> >> (type dog)
> >> (type cat)
> >> (type goldfish)
> >> (typeattribute pets)
> >> (typeattributeset pets (dog cat goldfish))
> >>
> >> (dontaudit pets colors (myclass (mypermission)))
> >>
> >> (expandtypeattribute (colors pets) false)
> >> EOF
> >>
> >> # secilc -v -X 4 mytest.cil
> >> # sesearch policy.32 --dontaudit
> >>
> >
> > The key is that expandtypeattribute works differently for the default
> > behavior than when you ask for attributes to be expanded.
> >
> > The default rules for removing attributes are complex. In general,
> > attributes automatically created when converting a binary module to
> > CIL; attributes that are not used in allow and dontaudit rules; and
> > attributes that have only one type assigned are all removed. This
> > means that the attribute rule and the typeattributeset rules for the
> > attribute are removed and all rules using the attribute are expanded.
> > Using expandtypeattribute in this case will keep the attribute
> > declaration, the typeattributeset rules, and any allow rules.
> >
> > When you use "-X SIZE" all allow rules with attributes that contain
> > less than SIZE types will be expanded and the attribute removed. Using
> > expandtypeattribute in this case will cause the attribute declaration
> > and the typeattributeset rules to be retained, but the allow rules
> > will still be expanded.
> >
> > I just verified that everything does work like that.
> >
> > It is a bit of a mess, but it is working as intended.
> >
> > I probably need to document how the default expansion works. ;)
>
>
> So that makes the "false" scenario practically irrelevant? when would
> one ever use "false" instead of "true"?
>

That is actually the most valuable option. It is the way to ensure
that an attribute definition is not removed. It is being used, for
example, by Google to keep attributes that are used in the neverallow
rules in their compatibility test suite from being expanded.

Jim

> >
> > Thanks,
> > Jim
> >
> >
> >> >
> >> > Thanks for doing this.
> >> > Jim
> >> >
> >> >>  typebounds
> >> >>  ----------
> >> >>
> >> >> --
> >> >> 2.28.0.rc1
> >> >>
> >>
> >> --
> >> gpg --locate-keys dominick.grift@defensec.nl
> >> Key fingerprint = FCD2 3660 5D6B 9D27 7FC6  E0FF DA7E 521F 10F6 4098
> >> https://sks-keyservers.net/pks/lookup?op=get&search=0xDA7E521F10F64098
> >> Dominick Grift
>
> --
> gpg --locate-keys dominick.grift@defensec.nl
> Key fingerprint = FCD2 3660 5D6B 9D27 7FC6  E0FF DA7E 521F 10F6 4098
> https://sks-keyservers.net/pks/lookup?op=get&search=0xDA7E521F10F64098
> Dominick Grift

  reply index

Thread overview: 19+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-07-30  9:23 [PATCH] " Dominick Grift
2020-07-30 11:45 ` [PATCH v2] " Dominick Grift
2020-07-30 13:11   ` [PATCH v3] " Dominick Grift
2020-07-30 22:22     ` James Carter
2020-07-31  7:26       ` Dominick Grift
2020-07-31 19:50         ` James Carter
2020-07-31 20:12           ` Dominick Grift
2020-07-31 20:22             ` James Carter [this message]
2020-07-31 21:07               ` Dominick Grift
2020-08-02 12:34       ` [PATCH v4] " Dominick Grift
2020-08-03 20:56         ` James Carter
2020-08-04  7:18           ` Dominick Grift
2020-08-04 14:45             ` James Carter
2020-08-04 15:48               ` Dominick Grift
2020-08-04 20:23                 ` James Carter
2020-08-04 20:29                   ` Dominick Grift
2020-08-05 19:23                     ` James Carter
2020-08-05 19:48                       ` [PATCH v5] " Dominick Grift
2020-08-05 20:22                         ` James Carter

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to='CAP+JOzSgp1Xgv29v__SAjZthonqRw=ZmYJ_3=MZm5FX8EOjDBw@mail.gmail.com' \
    --to=jwcart2@gmail.com \
    --cc=dominick.grift@defensec.nl \
    --cc=selinux@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

SELinux Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/selinux/0 selinux/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 selinux selinux/ https://lore.kernel.org/selinux \
		selinux@vger.kernel.org
	public-inbox-index selinux

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/org.kernel.vger.selinux


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git