[PATCH v2 2/2] ioctl console test v2

[Eugene Syromiatnikov]

On Wed, Mar 18, 2020 at 09:55:20PM +0530, Bran S wrote:
> On Wed, 18 Mar 2020 at 02:49, Eugene Syromiatnikov <esyr at redhat.com> wrote:
> >
> > On Tue, Mar 17, 2020 at 09:52:48PM +0530, Bran S wrote:
> > > On Mon, 16 Mar 2020 at 15:56, Eugene Syromiatnikov <esyr at redhat.com> wrote:
> > > >
> > > > On Mon, Mar 16, 2020 at 03:37:07PM +0530, Bran S wrote:
> > > > > There can be any number of subreapers right ?
> > > >
> > > > In a system? Yes, but the processes are re-parented to the closest one
> > > > up in the process tree hierarchy, so the subreaper a process
> > > > is to be re-parented to is pretty much determined.
> > > >
> > >
> > > So we follow the hierarchy upwards and for each pid we check if it is
> > > a subreaper if yes then the ppid should match to it.
> > > Is that correct ? If not please point me in the direction so as to
> > > find the subreaper's process id.
> >
> > Well, I can only repeat the Dmitry's suggestion[1] to check out
> > the documentation for the respective prctl(2) options.
> >
> > [1] https://lists.strace.io/pipermail/strace-devel/2020-March/009500.html
> 
> I think I am missing something obvious here. PR_GET_CHILD_SUBREAPER
> tells if current process is a subreaper.
> ----
>     int ret;
>     unsigned long arg2;
>     ret = prctl(PR_GET_CHILD_SUBREAPER, &arg2);
> 
>     printf("ret: %d\n", ret);
>     printf("arg2: %d\n", arg2);
> ----
> How can it be used to confirm that the ppid process is a subreaper is
> unclear to me ? I will look through it all once again with fresh eyes.

Oh, you're right, I've overlooked that bit, my apologies.  In that case,
it seems that the easiest way is to have an additional helper program
that makes itself reparented and reports its PPID.

> On a sidenote: Is there a difference between subreaper and child
> subreaper ? I think they are similar terms.

Yes, the only subreaper in kernel is child subreaper, so far.