From owner-freebsd-hackers  Thu May 10 20:20:49 2001
Delivered-To: freebsd-hackers@freebsd.org
Received: from bazooka.unixfreak.org (bazooka.unixfreak.org [63.198.170.138])
	by hub.freebsd.org (Postfix) with ESMTP id 9272C37B423
	for <hackers@freebsd.org>; Thu, 10 May 2001 20:20:41 -0700 (PDT)
	(envelope-from dima@unixfreak.org)
Received: from spike.unixfreak.org (spike [63.198.170.139])
	by bazooka.unixfreak.org (Postfix) with ESMTP
	id F3FB43E28; Thu, 10 May 2001 20:20:40 -0700 (PDT)
To: Poul-Henning Kamp <phk@critter.freebsd.dk>
Cc: hackers@freebsd.org
Subject: Re: Who's cleaning up after disk_clone? 
In-Reply-To: <75404.989482620@critter>; from phk@critter.freebsd.dk on "Thu, 10 May 2001 10:17:00 +0200"
Date: Thu, 10 May 2001 20:20:40 -0700
From: Dima Dorfman <dima@unixfreak.org>
Message-Id: <20010511032041.F3FB43E28@bazooka.unixfreak.org>
Sender: owner-freebsd-hackers@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG

Poul-Henning Kamp <phk@critter.freebsd.dk> writes:
> In message <20010510074028.847EE3E28@bazooka.unixfreak.org>, Dima Dorfman writes:
> >	- what should be destroying /dev/md0c?
> 
> The disk-minilayer.
> 
> We need some to keep track of cloned dev_t's so we can nuke them
> at various strategic points, havn't gotten to that yet.

Well, that sounds pretty easy to do.  Patch attached.  It solves this
problem quite effectively.  At this point it only nukes the clones
when disk_destroy is called; I don't know if there are other places
this should be done.

Comments?

					Dima Dorfman
					dima@unixfreak.org


Index: sys/conf.h
===================================================================
RCS file: /st/src/FreeBSD/src/sys/sys/conf.h,v
retrieving revision 1.127
diff -u -r1.127 conf.h
--- sys/conf.h	2001/05/01 08:13:17	1.127
+++ sys/conf.h	2001/05/11 03:17:14
@@ -86,6 +86,7 @@
 			int __sid_bsize_best; /* optimal block size */
 		} __si_disk;
 	} __si_u;
+	LIST_ENTRY(specinfo)	si_clist;
 };
 
 #define si_tty		__si_u.__si_tty.__sit_tty
Index: sys/disk.h
===================================================================
RCS file: /st/src/FreeBSD/src/sys/sys/disk.h,v
retrieving revision 1.17
diff -u -r1.17 disk.h
--- sys/disk.h	2000/06/15 20:30:53	1.17
+++ sys/disk.h	2001/05/11 03:17:14
@@ -30,6 +30,7 @@
 	dev_t			d_dev;
 	struct diskslices	*d_slice;
 	struct disklabel	d_label;
+	LIST_HEAD(, specinfo)	d_clones;
 	LIST_ENTRY(disk)	d_list;
 };
 
Index: kern/subr_disk.c
===================================================================
RCS file: /st/src/FreeBSD/src/sys/kern/subr_disk.c,v
retrieving revision 1.36
diff -u -r1.36 subr_disk.c
--- kern/subr_disk.c	2001/05/06 21:55:22	1.36
+++ kern/subr_disk.c	2001/05/11 03:17:14
@@ -86,6 +86,7 @@
 
 		*dev = make_dev(pdev->si_devsw, dkmakeminor(u, s, p), 
 		    UID_ROOT, GID_OPERATOR, 0640, name);
+		LIST_INSERT_HEAD(&dp->d_clones, *dev, si_clist);
 		return;
 	}
 }
@@ -129,6 +130,7 @@
 	dp->d_dev = dev;
 	dp->d_dsflags = flags;
 	dp->d_devsw = cdevsw;
+	LIST_INIT(&dp->d_clones);
 	LIST_INSERT_HEAD(&disklist, dp, d_list);
 	if (!once) {
 		EVENTHANDLER_REGISTER(dev_clone, disk_clone, 0, 1000);
@@ -174,7 +176,14 @@
 void
 disk_destroy(dev_t dev)
 {
+	dev_t pdev;
+
 	LIST_REMOVE(dev->si_disk, d_list);
+	while (!LIST_EMPTY(&dev->si_disk->d_clones)) {
+		pdev = LIST_FIRST(&dev->si_disk->d_clones);
+		LIST_REMOVE(pdev, si_clist);
+		destroy_dev(pdev);
+	}
 	bzero(dev->si_disk, sizeof(*dev->si_disk));
     	dev->si_disk = NULL;
 	destroy_dev(dev);

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