[CRIU] Re: Can't restore a process, which has been dumped in a second time

[Cyrill Gorcunov]

On Fri, Aug 24, 2012 at 12:14:01PM +0400, Andrey Vagin wrote:
> The problem is that the restorer can't mmap vma with zero size.
> I found that this vma is a part of stack from a first "crtools restore".
> 
> We already tried to fix such problem by df81acafa711811208442059e193f239c8f98c8d
> """
> restorer: Calculate self-stack size with guard pages
> 
> Otherwise we might not unmap the whole stack leaving
> a guard page in restoree memory map. Not a big deal
> but anyway.
> 
> Signed-off-by: Cyrill Gorcunov <gorcunov at openvz.org>
> Signed-off-by: Pavel Emelyanov <xemul at parallels.com>
> """
> 
> We care about own stack, but don't care about a previous stacks
> (e.g:before clone(...)).

Could you please elaborate, you've done checkpoint, then restored
a program and then checkpointed it again?

> I think we should find a way to detect the flag MAP_GROWSDOWN... Do
> you have any ideas?

I would though about putting a flag in /proc/smaps maybe?

	Cyrill