[333] in arla-drinkers
Re: get_unused_fd()
daemon@ATHENA.MIT.EDU (Alexander Viro)
Wed Oct 21 07:01:24 1998
From owner-arla-drinkers@stacken.kth.se Wed Oct 21 11:01:22 1998
Return-Path: <owner-arla-drinkers@stacken.kth.se>
Delivered-To: arla-drinkers-mtg@bloom-picayune.mit.edu
Received: (qmail 310 invoked from network); 21 Oct 1998 11:01:21 -0000
Received: from unknown (HELO sundance.stacken.kth.se) (130.237.234.41)
by bloom-picayune.mit.edu with SMTP; 21 Oct 1998 11:01:20 -0000
Received: (from majordom@localhost)
by sundance.stacken.kth.se (8.8.8/8.8.8) id MAA05860
for arla-drinkers-list; Wed, 21 Oct 1998 12:55:14 +0200 (MET DST)
Received: from math.psu.edu (root@leibniz.math.psu.edu [146.186.130.2])
by sundance.stacken.kth.se (8.8.8/8.8.8) with ESMTP id MAA05854;
Wed, 21 Oct 1998 12:55:06 +0200 (MET DST)
Received: from albert.math.psu.edu (albert.math.psu.edu [146.186.130.220]) by math.psu.edu (8.8.5/8.7.3) with ESMTP id GAA18420; Wed, 21 Oct 1998 06:54:52 -0400 (EDT)
Received: from localhost (viro@localhost) by albert.math.psu.edu (8.8.8/8.6.9) with SMTP id OAA00965; Wed, 21 Oct 1998 14:54:39 -0400 (EDT)
Date: Wed, 21 Oct 1998 14:54:38 -0400 (EDT)
From: Alexander Viro <viro@math.psu.edu>
To: Magnus Ahltorp <map@stacken.kth.se>
cc: Linus Torvalds <torvalds@transmeta.com>, linux-kernel@vger.rutgers.edu,
arla-drinkers@stacken.kth.se
Subject: Re: get_unused_fd()
In-Reply-To: <lv1vhleak5z.fsf@sundance.stacken.kth.se>
Message-ID: <Pine.SOL.3.95.981021144332.942D-100000@albert.math.psu.edu>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Sender: owner-arla-drinkers@stacken.kth.se
Precedence: bulk
On 21 Oct 1998, Magnus Ahltorp wrote:
> > > Does anyone have religious problems with this patch?
> >
> > I'd like to know who and why needs it..
>
> I sent a message to the list some weeks ago (included below) and since
> no one has commented on that, I supposed that there were no
> objections.
>
> The thing is, it's very hard to do some serious file system modules
> hacking without these functions. Since the Linux module system is
> built the way it is, symbol names have to, as you know, be explicitly
> exported.
It would be nice if you'ld show how _exactly_ you are going to use
get_empty_filp(). There are several reasons to split inuse_filp into
several lists (for superblock; for tty; for protocol family). That would
simplify dquot.c and some nasty stuff in tty_io.c. Right now I'm
experimenting with such patch and if everything will go OK I'm going to
submit it to Linus in a day or two. We got waaay too many leaks/dangling
pointers/etc. in the code that deals with struct file's. I'm trying to go
through this stuff and massage it into something more coherent. And IMHO
get_empty_filp() belongs to guts of that stuff.
Al
--
"You're one of those condescending Unix computer users!"
"Here's a nickel, kid. Get yourself a better computer" - Dilbert.
