[CRIU] [PATCH 1/2] restore: don't pass cgroup nsroot= option to mount()

[Tycho Andersen]

On Thu, Mar 24, 2016 at 10:11:45AM +0300, Pavel Emelyanov wrote:
> On 03/24/2016 12:41 AM, Tycho Andersen wrote:
> > https://lkml.org/lkml/2016/3/21/777 introduces a new cgroup mount option
> > called nsroot, but doesn't allow you to pass it to mount() to actually
> > mount it. Let's chop this option off so that we don't get errors from
> > mount().
> 
> What if this option comes in the middle of the options string?

Right now it's not, but I suppose it could be if the kernel adds more
cgroup mount options. I'll update the patch to handle this.

Tycho

> > There may be better places to chop this off (e.g. perhaps on dump),
> > although I think it is useful to have it in there for debugging purposes,
> > and only costs us a few extra bytes to keep it in the images since we're
> > going to have other cgroup options anyways.
> > 
> > Signed-off-by: Tycho Andersen <tycho.andersen at canonical.com>
> > ---
> >  criu/mount.c | 12 ++++++++++++
> >  1 file changed, 12 insertions(+)
> > 
> > diff --git a/criu/mount.c b/criu/mount.c
> > index a425828..be79999 100644
> > --- a/criu/mount.c
> > +++ b/criu/mount.c
> > @@ -2866,6 +2866,18 @@ static int collect_mnt_from_image(struct mount_info **pms, struct ns_id *nsid)
> >  		/* FIXME: abort unsupported early */
> >  		pm->fstype = decode_fstype(me->fstype, me->fsname);
> >  
> > +		if (pm->fstype->code == FSTYPE__CGROUP) {
> > +			char *pos;
> > +
> > +			pos = strstr(pm->options, "nsroot=");
> > +			if (pos > pm->options && *(pos-1) == ',')
> > +				pos--;
> > +			if (pos) {
> > +				pr_info("trimming nsroot option %s from cgroup mount\n", pos);
> > +				*pos = 0;
> > +			}
> > +		}
> > +
> >  		if (get_mp_root(me, pm))
> >  			goto err;
> >  
> > 
>