2006-02-08 23:19:59

by Christoph Lameter

Subject: Re: + fork-allow-init-to-become-a-session-leader.patch added to -mm tree

On Wed, 8 Feb 2006 [email protected] wrote:

> + if (unlikely(p->pid == 1)) {
> + p->signal->tty = NULL;
> + p->signal->leader = 1;
> + p->signal->pgrp = 1;
> + p->signal->session = 1;

Would it not be better to do these special settings for init from
init itself?

2006-02-09 21:32:18

by Eric W. Biederman

Subject: Re: + fork-allow-init-to-become-a-session-leader.patch added to -mm tree

Christoph Lameter <[email protected]> writes:

> On Wed, 8 Feb 2006 [email protected] wrote:
>
>> + if (unlikely(p->pid == 1)) {
>> + p->signal->tty = NULL;
>> + p->signal->leader = 1;
>> + p->signal->pgrp = 1;
>> + p->signal->session = 1;
>
> Would it not be better to do these special settings for init from
> init itself?

So if you mean from within the kernel context that is doable,
so long as we only have one process with pid == 1.

Although we may be able fix the kernel limitations in setsid().

For multiple pid == 1 case I don't know of another place in
the kernel I can possibly put this. I can't put it later
because there is no place later, and I can't put it earlier
or else my efforts would just get stomped.

So I did it this way so I don't have to come back and change it to
this next week. I'd love to have a better place to put this
code.

Eric