CCing linux-api and stable to give them a chance to confirm that
changing proc symlink content is OK.
On 15/10/2021 11:10, Christian Brauner wrote:
> On Wed, Oct 13, 2021 at 05:47:53PM +0200, Mickaël Salaün wrote:
>>
>> On 12/10/2021 23:09, Paul Moore wrote:
>>> On Tue, Oct 12, 2021 at 4:38 PM Ondrej Mosnacek <omosnace@xxxxxxxxxx> wrote:
>>>>
>>>> On Tue, Oct 12, 2021 at 8:12 PM Paul Moore <paul@xxxxxxxxxxxxxx> wrote:
>>>>> On Tue, Oct 12, 2021 at 6:38 AM Christian Brauner
>>>>> <christian.brauner@xxxxxxxxxx> wrote:
>>>>>> On Mon, Oct 11, 2021 at 04:38:55PM +0200, Mickaël Salaün wrote:
>>>>>>> On 11/10/2021 15:37, Christian Brauner wrote:
>>>>>>>> From: Christian Brauner <christian.brauner@xxxxxxxxxx>
>>>>>>>>
>>>>>>>> Make the name of the anon inode fd "[landlock-ruleset]" instead of
>>>>>>>> "landlock-ruleset". This is minor but most anon inode fds already
>>>>>>>> carry square brackets around their name:
>>>>>>>>
>>>>>>>> [eventfd]
>>>>>>>> [eventpoll]
>>>>>>>> [fanotify]
>>>>>>>> [fscontext]
>>>>>>>> [io_uring]
>>>>>>>> [pidfd]
>>>>>>>> [signalfd]
>>>>>>>> [timerfd]
>>>>>>>> [userfaultfd]
>>>>>>>>
>>>>>>>> For the sake of consistency lets do the same for the landlock-ruleset anon
>>>>>>>> inode fd that comes with landlock. We did the same in
>>>>>>>> 1cdc415f1083 ("uapi, fsopen: use square brackets around "fscontext" [ver #2]")
>>>>>>>> for the new mount api.
>>>>>>>
>>>>>>> Before creating "landlock-ruleset" FD, I looked at other anonymous FD
>>>>>>> and saw this kind of inconsistency. I don't get why we need to add extra
>>>>>>> characters to names, those brackets seem useless. If it should be part
>>>>>>
>>>>>> Past inconsistency shouldn't justify future inconsistency. If you have a
>>>>>> strong opinion about this for landlock I'm not going to push for it.
>>>>>> Exchanging more than 2-3 email about something like this seems too much.
>>>>>
>>>>> [NOTE: adding the SELinux list as well as Chris (SELinux refrence
>>>>> policy maintainer) and Petr (Fedora/RHEL SELinux)]
>>>>>
>>>>> Chris and Petr, do either of you currently have any policy that
>>>>> references the "landlock-ruleset" anonymous inode? In other words,
>>>>> would adding the brackets around the name cause you any problems?
>>>>
>>>> AFAIU, the anon_inode transitions (the only mechanism where the "file
>>>> name" would be exposed to the policy) are done only for inodes created
>>>> by anon_inode_getfd_secure(), which is currently only used by
>>>> userfaultfd. So you don't even need to ask that question; at this
>>>> point it should be safe to change any of the names except
>>>> "[userfaultfd]" as far as SELinux policy is concerned.
>>>
>>> There is also io_uring if you look at selinux/next.
>>>
>>> Regardless, thanks, I didn't check to see if landlock was using the
>>> new anon inode interface, since both Mickaël and Christian were
>>> concerned about breaking SELinux I had assumed they were using it :)
>>>
>>
>> Ok, thanks Paul and Ondrej.
>>
>> Such anonymous inode names seem to be only exposed to proc for now.
>> Let's change this name then. I think it make sense to backport this
>> patch down to 5.13 to fix all the inconsistencies.
>
> Thank you. I do appreciate the point about this being annoying that we
> have this inconsistency and it has bothered me too.
>
> Christian
>
