Messages in this thread | | | Date | Mon, 6 Oct 2008 06:52:25 -0400 | From | Alan Cox <> | Subject | Re: [PATCH 26/76] tty: Add a kref count |
| |
On Mon, Oct 06, 2008 at 12:20:37PM +0200, Louis Rilling wrote: > > I'm a bit puzzled by this 'tty_kref_put(p->signal->tty)'. AFAICS, this is a noop > since we are not in the CLONE_THREAD case and hence p->signal->tty == NULL.
When I first did it I was seeing non NULL values, but since then I've added the initialisation
> So, is it to make the code look more consistent? If so, is it worth adding extra > code and cycles for this (I doubt that gcc is able to optimize this away)? What > kind of future changes should this code protect against?
I'll take a look, you may well be right, and if so it can become a WARN_ON() instead.
| |