[CRIU] Re: [PATCH cr 6/8] pidns: zombie processes are not supported yet

Pavel Emelyanov xemul at parallels.com
Wed May 16 08:01:14 EDT 2012


On 05/16/2012 03:58 PM, Andrew Vagin wrote:
> On Wed, May 16, 2012 at 03:49:11PM +0400, Pavel Emelyanov wrote:
>> On 05/16/2012 12:32 PM, Andrey Vagin wrote:
>>>
>>> Signed-off-by: Andrey Vagin <avagin at openvz.org>
>>> ---
>>>  cr-dump.c |   13 ++++++++++++-
>>>  1 files changed, 12 insertions(+), 1 deletions(-)
>>>
>>> diff --git a/cr-dump.c b/cr-dump.c
>>> index 25679d6..11f31a0 100644
>>> --- a/cr-dump.c
>>> +++ b/cr-dump.c
>>> @@ -1550,8 +1550,19 @@ static int dump_one_task(struct pstree_item *item)
>>>  	if (ret < 0)
>>>  		goto err;
>>>  
>>> -	if (item->state == TASK_DEAD)
>>> +	if (item->state == TASK_DEAD) {
>>> +		int i;
>>> +		/* FIXME don't support zombie in pid name space*/
>>
>> O_o So do we support them or not?
> 
> We don't support it in a separate pid namespace and we support it if
> it's restored in current pid namespace.

I don't see the

pr_err("Zombies in pid namespace are not supported);
return -1;

lines in this patch. Is this "not supported" reported by the BUG_ON?

>>
>>> +		item->pid.pid = item->pid.real_pid;
>>> +		for (i = 0; i < item->parent->nr_children; i++)
>>> +			if (item->parent->children[i].real_pid == item->pid.real_pid) {
>>> +				item->parent->children[i].pid = item->pid.pid;
>>> +				break;
>>> +			}
>>> +
>>> +		BUG_ON(i == item->parent->nr_children);
>>>  		return dump_one_zombie(item, &pps_buf);
>>> +	}
>>>  
>>>  	ret = collect_mappings(pid, &vma_area_list);
>>>  	if (ret) {
>>
> .
> 



More information about the CRIU mailing list