Re: [patch] generic-pidhash-2.5.36-D4, BK-curr

[Linus Torvalds]

On Thu, 19 Sep 2002, Ingo Molnar wrote:
> 
> the attached patch is a significantly cleaned up version of the generic
> pidhash patch, against BK-curr.

Hmm.. I think I like it. One big improvement is just the renaming of
"struct idtag" to "struct pid", which makes it look a lot less abstract to
me. Maybe that's just me.

However:

>  	read_lock(&tasklist_lock);
> - 	for_each_process(p) {
> -		if ((tty->session > 0) && (p->session == tty->session) &&
> -		    p->leader) {
> -			send_sig(SIGHUP,p,1);
> -			send_sig(SIGCONT,p,1);
> +	if (tty->session > 0)
> +		for_each_task_pid(tty->session, PIDTYPE_SID, p, l, pid) {
> +			if (p->tty == tty)
> +				p->tty = NULL;
> +			if (!p->leader)
> +				continue;
> +			send_sig(SIGHUP, p, 1);
> +			send_sig(SIGCONT, p, 1);
>  			if (tty->pgrp > 0)
>  				p->tty_old_pgrp = tty->pgrp;
>  		}
> -		if (p->tty == tty)
> -			p->tty = NULL;
> -	}

This looks a bit wrong. In particular, the old code used to set "p->tty" 
to NULL if it matched any process, while the new code only does that for 
processes in the right session. Hmm?

Can you double-check some of these things?

		Linus