[CRIU] [PATCH 14/14] mount: don't open the same root twise

Andrew Vagin avagin at parallels.com
Fri Apr 18 02:21:41 PDT 2014


On Fri, Apr 18, 2014 at 12:42:34PM +0400, Pavel Emelyanov wrote:
> On 04/17/2014 11:24 PM, Andrey Vagin wrote:
> > Signed-off-by: Andrey Vagin <avagin at openvz.org>
> > ---
> >  mount.c | 7 +++++++
> >  1 file changed, 7 insertions(+)
> > 
> > diff --git a/mount.c b/mount.c
> > index 92c6a6c..72ec81d 100644
> > --- a/mount.c
> > +++ b/mount.c
> > @@ -1789,10 +1789,15 @@ out:
> >  
> >  int mntns_collect_root(pid_t pid)
> >  {
> > +	static int mntns_root_pid = -1;
> > +
> >  	int fd, pfd;
> >  	int ret;
> >  	char path[PATH_MAX + 1];
> >  
> > +	if (mntns_root_pid == pid) /* The required root is already opened */
> > +		return get_service_fd(ROOT_FD_OFF);
> > +
> >  	close_service_fd(ROOT_FD_OFF);
> >  
> >  	if (!(root_ns_mask & CLONE_NEWNS)) {
> > @@ -1841,6 +1846,8 @@ int mntns_collect_root(pid_t pid)
> >  
> >  set_root:
> >  	ret = install_service_fd(ROOT_FD_OFF, fd);
> > +	if (ret >= 0)
> > +		mntns_root_pid = pid;
> 
> If we're going to get mntns root fd always, I see no sense in service
> FD. It's like open_proc_fd thing.

It's service fd to avoid conflicts with restored descriptors.

> 
> >  	close(fd);
> >  	return ret;
> >  }
> > 
> 
> 


More information about the CRIU mailing list