Re: [PATCH 1/3] change clone_flags type to u64

["Daniel Hokka Zakrisson" <daniel@hozac.com>]

Serge E. Hallyn wrote:
> Quoting Daniel Hokka Zakrisson (daniel@hozac.com):
>> Serge E. Hallyn wrote:
>> > Quoting Andi Kleen (andi@firstfloor.org):
>> >> > I guess that was a development rationale.
>> >>
>> >> But what rationale? It just doesn't make much sense to me.
>> >>
>> >> > Most of the namespaces are in
>> >> > use in the container projects like openvz, vserver and probably
>> others
>> >> > and we needed a way to activate the code.
>> >>
>> >> You could just have added it to feature groups over time.
>> >>
>> >> >
>> >> > Not perfect I agree.
>> >> >
>> >> > > With your current strategy are you sure that even 64bit will
>> >> > > be enough in the end? For me it rather looks like you'll
>> >> > > go through those quickly too as more and more of the kernel
>> >> > > is namespaced.
>> >> >
>> >> > well, we're reaching the end. I hope ! devpts is in progress and
>> >> > mq is just waiting for a clone flag.
>> >>
>> >> Are you sure?
>> >
>> > Well for one thing we can take a somewhat different approach to new
>> > clone flags.  I.e. we could extend CLONE_NEWIPC to do mq instead of
>> > introducing a new clone flag.  The name doesn't have 'sysv' in it,
>> > and globbing all ipc resources together makes some amount of sense.
>> > Similarly has hpa+eric pointed out earlier, suka could use
>> > CLONE_NEWDEV for ptys.  If we have net, pid, ipc, devices, that's a
>> > pretty reasonable split imo.  Perhaps we tie user to devices and get
>> > rid of CLONE_NEWUSER which I suspect noone is using atm (since only
>> > Dave has run into the CONFIG_USER_SCHED problem).  Or not.  We could
>> > roll uts into net, and give CLONE_NEWUTS a deprecation period.
>>
>> Please don't. Then we'd need to re-add it in Linux-VServer to support
>> guests where network namespaces aren't used...
>
> So these are networked vservers with a different hostname?  Just
> curious, what would be a typical use for these?

Layer 3 isolation will continue to be the default for Linux-VServer.

> Anyway then I guess we won't :)  Do you have other suggestions for
> ns clone flags which ought to be combined?  Do the rest of what I
> listed make sense to you?  (If not, then I guess I'll step out of the
> way and let you and Andi fight it out :)

I think putting mq under CLONE_NEWIPC makes sense, as well as using
CLONE_NEWDEV for the ptys. If CLONE_NEWUSER is to be combined with
anything, I think it makes more sense to combine it with CLONE_NEWPID than
CLONE_NEWDEV.

> thanks,
> -serge
>

-- 
Daniel Hokka Zakrisson