[Devel] Re: [RFC][PATCH 4/5] utsname namespaces: sysctl hack

[Serge E. Hallyn]

Quoting Eric W. Biederman (ebiederm at xmission.com):
> "Serge E. Hallyn" <serue at us.ibm.com> writes:
> 
> > Quoting Eric W. Biederman (ebiederm at xmission.com):
> >> Kirill Korotaev <dev at sw.ru> writes:
> >> 
> >> > Serge,
> >> >
> >> > can we do nothing with sysctls at this moment, instead of commiting hacks?
> >> 
> >> Except that we modify a static table changing the uts behaviour in
> >> proc_doutsstring isn't all that bad.
> >> 
> >> I'm just about to start on something more comprehensive, in
> >> the sysctl case.
> >
> > So assuming that I take out the switch(), leaving that for a better
> > solution by Eric (or Dave, or whoever),
> >
> > Is it time to ask for the utsname namespace patch to be tried out
> > in -mm?
> 
> Can we please suggest a syscall interface?

We can, but I was hoping that would be a separate patch, separate
discussion.

Are you asking for a new syscall, specifically to unshare utsname()?  Or
for discussion over whether we want to do
	one syscall per namespace
	extend CLONE_NEWns flags
	use unshare
	use namespacefs

-serge