[CRIU] Re: [PATCH cr 06/12] cr-dump: dump pstree after tasks

Andrew Vagin avagin at parallels.com
Mon Jun 4 06:13:52 EDT 2012


On Mon, Jun 04, 2012 at 01:11:53PM +0400, Pavel Emelyanov wrote:
> On 05/31/2012 04:58 PM, Andrey Vagin wrote:
> > 
> > because a pid in a target pid ns is got from parasite.
> > 
> > Signed-off-by: Andrey Vagin <avagin at openvz.org>
> > ---
> >  cr-dump.c |   54 ++++++++++++++++++++++++++++++++----------------------
> >  1 files changed, 32 insertions(+), 22 deletions(-)
> > 
> 
> > @@ -1584,13 +1584,13 @@ static int dump_one_task(struct pstree_item *item)
> >  	if (ret < 0)
> >  		goto err;
> >  
> > -	if (item->state == TASK_DEAD)
> > -		return dump_one_zombie(item, &pps_buf);
> > +	if (item->state == TASK_DEAD) {
> > +		/* FIXME don't support zombie in pid name space*/
> > +		item->pid.pid = item->pid.real_pid;
> >  
> > -	ret = -1;
> > -	cr_fdset = cr_task_fdset_open(item->pid.pid, O_DUMP);
> > -	if (!cr_fdset)
> > -		goto err;
> > +		BUG_ON(!list_empty(&item->children));
> > +		return dump_one_zombie(item, &pps_buf);
> > +	}
> >  
> >  	ret = collect_mappings(pid, &vma_area_list);
> >  	if (ret) {
> 
> I don't quite understand this hunk. Either it should be a separate patch or a better comment.

Sorry, I forgot about that and have fixed it in [CRIU] [PATCH 13/12] dump: fail if a zombie is in PIDNS


More information about the CRIU mailing list