From owner-freebsd-current  Mon Oct 16 11:49:02 1995
Return-Path: owner-current
Received: (from root@localhost)
          by freefall.freebsd.org (8.6.12/8.6.6) id LAA17111
          for current-outgoing; Mon, 16 Oct 1995 11:49:02 -0700
Received: from phaeton.artisoft.com (phaeton.Artisoft.COM [198.17.250.211])
          by freefall.freebsd.org (8.6.12/8.6.6) with ESMTP id LAA17106
          for <current@freebsd.org>; Mon, 16 Oct 1995 11:48:56 -0700
Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id LAA25077; Mon, 16 Oct 1995 11:44:02 -0700
From: Terry Lambert <terry@lambert.org>
Message-Id: <199510161844.LAA25077@phaeton.artisoft.com>
Subject: Re: lstat flushes namei cache entry
To: bde@zeta.org.au (Bruce Evans)
Date: Mon, 16 Oct 1995 11:44:01 -0700 (MST)
Cc: current@freebsd.org
In-Reply-To: <199510161138.VAA24539@godzilla.zeta.org.au> from "Bruce Evans" at Oct 16, 95 09:38:02 pm
X-Mailer: ELM [version 2.4 PL24]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Content-Length: 2576      
Sender: owner-current@freebsd.org
Precedence: bulk

> lstat() has been inefficient since symlinks were POSIXified.
> 
> Bruce
> 
> *** vfs_lookup.c~	Fri Aug 25 16:05:33 1995
> --- vfs_lookup.c	Mon Oct 16 21:06:30 1995
> ***************
> *** 267,272 ****
>   	wantparent = cnp->cn_flags & (LOCKPARENT | WANTPARENT);
>   	docache = (cnp->cn_flags & NOCACHE) ^ NOCACHE;
>   	if (cnp->cn_nameiop == DELETE ||
> ! 	    (wantparent && cnp->cn_nameiop != CREATE))
>   		docache = 0;
>   	rdonly = cnp->cn_flags & RDONLY;
> --- 280,297 ----
>   	wantparent = cnp->cn_flags & (LOCKPARENT | WANTPARENT);
>   	docache = (cnp->cn_flags & NOCACHE) ^ NOCACHE;
> + 	/*
> + 	 * XXX the following seems to be just to recover from not setting
> + 	 * NOCACHE for the DELETE cases (unlink, rmdir and the rename
> + 	 * source).  In BSD4.4lite[2], docache was also cleared for the
> + 	 * (wantparent && cnp->cn_nameiop == LOOKUP) case.  This case
> + 	 * seems to only occur for lstat and olstat, when it is wrong
> + 	 * to clear docache.  This case probably didn't occur before
> + 	 * BSD4.4lite.  LOCKPARENT was introduced for lstat to support
> + 	 * the new behaviour of symlinks (attributes inherited from the
> + 	 * parent.
> + 	 */
>   	if (cnp->cn_nameiop == DELETE ||
> ! 	    (wantparent && cnp->cn_nameiop != CREATE &&
> ! 	     cnp->cn_nameiop != LOOKUP))
>   		docache = 0;
>   	rdonly = cnp->cn_flags & RDONLY;


This will INCORRECTLY result in the contents of symlinks being seen as
successfully looked up inferior components and entered into the cache.

This is *BROKEN*.

Consider:

	cd /tmp
	mkdir foo
	cp /etc/ttys foo
	mkdir fee
	cd fee
	ln -s /tmp/foo/ttys xxx
	ls xxx
<xxx is found>
	ls foo
<ttys is found because foo was incorrectly entered into the cache
 inferior to fee's vnode>


The *correct* set of fixes for this is moving the cache lookup into
vfs_lookup.c instead of calling it per FS lookup mechanism.  If I can't
get a simple interface change into -current, it's unlikely that something
that large in scope would go in (I would need to have the ability to
disable it on a per FS type basis, etc.).

Integrate my other changes first, and my patches to nameifree() and
nfs_nameifree() to avoid duplicate free's, please.  Then we can go
off and consider moving the name cache.

Also run the "find' I asked for for the cn_pnbuf references *after* the
patches are applies so we can missed patches because of parallel
development on things like procfs and devfs.


					Thanks,
					Terry Lambert
					terry@lambert.org
---
Any opinions in this posting are my own and not those of my present
or previous employers.