Re: [LSF/MM TOPIC] new mount(2) userspace api

[ebiederm@xmission.com (Eric W. Biederman)]

Amir Goldstein <amir73il@gmail.com> writes:

> On Fri, Jan 27, 2017 at 11:46 AM, Miklos Szeredi <miklos@szeredi.hu> wrote:
>> On Fri, Jan 13, 2017 at 3:20 PM, Miklos Szeredi <miklos@szeredi.hu> wrote:
>>> On Thu, Jan 12, 2017 at 6:56 PM, Omar Sandoval <osandov@osandov.com> wrote:
>>>> On Wed, Jan 11, 2017 at 11:37:03AM +0100, Miklos Szeredi wrote:
>>>>> General direction has been laid out by Al, but it would be good to
>>>>> discuss the details.
>>>>>
>>>>> I plan to work on this after overlayfs is "done".
>>>>>
>>>>> Thanks,
>>>>> Miklos
>>>>
>>>> Hi, Miklos,
>>>>
>>>> What's the background on this? Is there a mailing list thread?
>>>
>>> http://marc.info/?t=138677271200004&r=1&w=2
>>>
>>> https://lwn.net/Articles/609376/
>>
>> Found one more:
>>
>> https://lists.linuxfoundation.org/pipermail/ksummit-discuss/2014-August/001199.html
>
>
> Miklos,
>
> WRT the suggestion to use a netlink socket that was made in LSF, be aware that
> netlink sockets are per network-ns and mounts as you know are per mount-ns.
> That may be a problem when setting up a mount from non root network-ns
> or maybe not? - I am not sure.

Netlink TLV's may be used over an ordinary file descriptor and that
seems much more reasonable that netlink if the netlink style of
marshalling arguments is desired.

Eric