From owner-freebsd-current  Sun Oct 17 19:55:12 1999
Delivered-To: freebsd-current@freebsd.org
Received: from gizmo.internode.com.au (gizmo.internode.com.au [192.83.231.115])
	by hub.freebsd.org (Postfix) with ESMTP
	id 1533114D49; Sun, 17 Oct 1999 19:55:00 -0700 (PDT)
	(envelope-from newton@gizmo.internode.com.au)
Received: (from newton@localhost)
	by gizmo.internode.com.au (8.9.3/8.9.3) id MAA73538;
	Mon, 18 Oct 1999 12:23:59 +0930 (CST)
	(envelope-from newton)
From: Mark Newton <newton@internode.com.au>
Message-Id: <199910180253.MAA73538@gizmo.internode.com.au>
Subject: Re: duplicate filenames under Linux emulation
To: ak@freenet.co.uk (Alex)
Date: Mon, 18 Oct 1999 12:23:58 +0930 (CST)
Cc: kurt@boolean.net, freebsd-emulation@FreeBSD.ORG,
	freebsd-current@FreeBSD.ORG
In-Reply-To: <380A9278.638D7FC2@freenet.co.uk> from "Alex" at Oct 18, 99 03:22:32 am
X-Mailer: ELM [version 2.4 PL25]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-freebsd-current@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG

Alex wrote:

 >  444964 -rw-r--r--   1  ak      users      102042 Oct 17 20:23
 > bxutils-c.c
 >  444964 -rw-r--r--   1  ak      users      102042 Oct 17 20:23
 > bxutils-c.c

For what it's worth, I have this bug in the to-do list for the 
SVR4 emulator too.  It's caused by the fact that dirent structures
for BSD and SysVR4 are of different sizes, so svr4_getdents()
(and svr4_getdents64(), its 64-bit counterpart) fetches a buffer
full of BSD dirent structures from disk (usually more than the number
which have actually been requested) with VOP_READDIR(), rewrites them
into another buffer full of svr4_dirent structures, and returns the
contents of the second buffer to userspace.  Next time getdents() is
called the emulated version continues from where it left off in the
first buffer until it has been exhaused, at which point it issues another 
VOP_READDIR() call to refresh it.

It looks like the duplicates are the entries which appear on the 
boundaries of the buffer first buffer (i.e.: the last directory entry
in a buffer will be re-read into the beginning of the buffer when
VOP_READDIR() is called again).

I haven't had time to examine it in detail, but I suspect that 
incrementing an offset parameter for VOP_READDIR() aio request will
fix this bug.

If it does, let me know and I'll fold the result into svr4_getdents()
as well :-)

    - mark

----
Mark Newton                               Email:  newton@internode.com.au (W)
Network Engineer                          Email:  newton@atdot.dotat.org  (H)
Internode Systems Pty Ltd                 Desk:   +61-8-82232999
"Network Man" - Anagram of "Mark Newton"  Mobile: +61-416-202-223


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message