Re: For review: pid_namespaces(7) man page (draft 2)

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

"Michael Kerrisk (man-pages)" <mtk.manpages@gmail.com> writes:

> Eric et al.,
>
> [CCing Li because of reboot(2) changes]
>
> I have (I think) addressed all previous comments in the current draft
> of the pid_namespaces(7) page. This is a final sanity check before I
> call this page complete (modulo any future kernel changes).

Baring the 3.9 reference to reboot(2) which should be 3.4 I don't see
any problems.

I think you have reached the point where if there are any problems left
I will just have to send you patches after you publish because my eyes
are glassing over and I am not going to see any more problems right now.

One of those patches I am going to have to send you is I am fixing the
case below for 3.10
>            clone(..., CLONE_VM, ...);
>            setns(fd, CLONE_NEWPID);      /* Fails */
>
>            clone(..., CLONE_VM, ...);
>            unshare(CLONE_NEWPID);        /* Fails */
>
It turns out that the check in fork/clone is sufficient.

The verbage about all of the namespaces is correct however since
unshare(CLONE_NEWPID) doesn't actually change the pid_namespace it
doesn't matter.

Eric