[CRIU] [PATCH] criu: Add exec-cmd option (v2)

[Andrew Vagin]

On Fri, Mar 21, 2014 at 10:07:01AM +0200, Deyan Doychev wrote:
> On 03/21/2014 09:32 AM, Andrew Vagin wrote:
> > On Fri, Mar 21, 2014 at 08:40:41AM +0400, Pavel Emelyanov wrote:
> >> > On 03/20/2014 08:40 PM, Christopher Covington wrote:
> >>> > > Hi Deyan,
> >>> > > 
> >>> > > On 03/20/2014 12:24 PM, Deyan Doychev wrote:
> >>>> > >> From: Deyan Doychev <deyandoichev at gmail.com>
> >>>> > >>
> >>>> > >> The --exec-cmd option specifies a command that will be execvp()-ed on successful
> >>>> > >> restore. This way the command specified here will become the parent process of
> >>>> > >> the restored process tree.
> >>>> > >>
> >>>> > >> When this option is specified criu will fork to become a daemon before it starts
> >>>> > >> restoring the processes. It also implies the -d option so waiting for the
> >>>> > >> restored processes to finish is responsibility of the command specified here.
> >>>> > >>
> >>>> > >> This option will be used when restoring LinuX Containers and it seems helpful
> >>>> > >> for perf or other use cases when restored processes must be supervised by a
> >>>> > >> parent.
> >>>> > >>
> >>>> > >> Two directions were researched in order to integrate CRIU and LXC:
> >>>> > >>
> >>>> > >> 1. We tell to CRIU, that after restoring container is should execve()
> >>>> > >>    lxc properly explaining to it that there's a new container hanging
> >>>> > >>    around.
> >>>> > >>
> >>>> > >> 2. We make LXC set himself as child subreaper, then fork() criu and ask
> >>>> > >>    it to detach (-d) from restore container afterwards. Being a subreaper,
> >>>> > >>    it should get the container's init into his child list after it.
> >>>> > >>
> >>>> > >> The main reason for choosing the first option is that the second one can't work
> >>>> > >> with the RPC service. If we call restore via the service then criu service will
> >>>> > >> be the top-most task in the hierarchy and will not be able to reparent the
> >>>> > >> restore trees to any other task in the system. Calling execve from service
> >>>> > >> worker sub-task (and daemonizing it) should solve this.
> >>> > > 
> >>> > > Looks good overall. On the subject of file descriptor sharing, I wonder if all
> >>> > > of the CRIU open, dup, etc. calls should also be changed to use the O_CLOEXEC
> >>> > > flag?
> >> > 
> >> > Presumably they should. There're quite a few of them, I'm pretty sure
> >> > that the main criu task only uses some service fds. By the way, what do
> >> > you think about redirecting the logfd into 1 and 2 before exec()-ing?
> > We can do this if a criu is executed as a sevice. If we execute
> > 'criu restore' directly, we need to leave original descriptors.
> >
> 
> We could do it when --restore-detached or --daemon are specified. This
> way we will always redirect logfd into 1 and 2 when we leave the exec-ed
> command running in background no matter if restore was started on
> command line with -d or as a service.

Yes, you are right.

> 
>