List view - Thread view

Subject: kernel/5431: ccd/ccdconfing rejects 0 interleave

From: djv

Date: 2007-03-30 06:57:12

>Number:         5431
>Category:       kernel
>Synopsis:       coding error in sys/dev/ccd.c falsely rejects interleave 0
>Confidential:   yes
>Severity:       critical
>Priority:       medium
>Responsible:    bugs
>State:          open
>Quarter:        
>Keywords:       
>Date-Required:
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Fri Mar 30 05:10:01 GMT 2007
>Closed-Date:
>Last-Modified:
>Originator:     Woodchuck
>Release:        OpenBSD 4.0
>Organization:
net
>Environment:
	
	System      : OpenBSD 4.0
	Architecture: OpenBSD.i386
	Machine     : i386
>Description:
	Attempting to configure a ccd with interleave 0
	(i.e. concatenation) with ccdconfig returns EINVAL
	from the ioctl call in ccdconfig.  This EINVAL is
	being generated in the kernel sys/dev/ccd.c from
	a sanity check on the ileave parameter, returning
	EINVAL if ileave How-To-Repeat:
	ccdconfig ccd0 0 0 /dev/sd0d /dev/sd1d
>Fix:


>Release-Note:
>Audit-Trail:
>Unformatted:

From: Woodchuck

Date: 2007-03-30 07:26:45

On Fri, 30 Mar 2007, <email removed> wrote:

> >Number:         5431
> >Category:       kernel
> >Fix:

This got snipped from the sendbug or I forgot to put it in:

[<email removed> dev 0:25]# diff -u ccd.c.old ccd.c
--- ccd.c.old   Fri Mar 30 00:04:46 2007
+++ ccd.c       Fri Mar 30 00:05:10 2007
@@ -1152,7 +1152,7 @@
                        return (EBUSY);
 
                if (ccio->ccio_ndisks == 0 || ccio->ccio_ndisks > INT_MAX ||
-                   ccio->ccio_ileave ccio_ileave < 0)
                        return (EINVAL);
 
                if ((error = ccdlock(cs)) != 0)

Dave

From: Thordur I. Bjornsson

Date: 2007-03-30 13:55:01

The following reply was made to PR kernel/5431; it has been noted by GNATS.

From: "Thordur I. Bjornsson" 
To: <email removed>
Cc: <email removed>
Subject: Re: kernel/5431: ccd/ccdconfing rejects 0 interleave
Date: Fri, 30 Mar 2007 11:37:06 +0000

 <email removed> wrote on Fri 30.Mar'07 at  0:57:12 -0400
 
 > >Number:         5431
 > >Category:       kernel
 > >Synopsis:       coding error in sys/dev/ccd.c falsely rejects interleave 0
 > >Confidential:   yes
 > >Severity:       critical
 > >Priority:       medium
 > >Responsible:    bugs
 > >State:          open
 > >Quarter:        
 > >Keywords:       
 > >Date-Required:
 > >Class:          sw-bug
 > >Submitter-Id:   net
 > >Arrival-Date:   Fri Mar 30 05:10:01 GMT 2007
 > >Closed-Date:
 > >Last-Modified:
 > >Originator:     Woodchuck
 > >Release:        OpenBSD 4.0
 > >Organization:
 > net
 > >Environment:
 > 	
 > 	System      : OpenBSD 4.0
 > 	Architecture: OpenBSD.i386
 > 	Machine     : i386
 > >Description:
 > 	Attempting to configure a ccd with interleave 0
 > 	(i.e. concatenation) with ccdconfig returns EINVAL
 > 	from the ioctl call in ccdconfig.  This EINVAL is
 > 	being generated in the kernel sys/dev/ccd.c from
 > 	a sanity check on the ileave parameter, returning
 > 	EINVAL if ileave  	valid value.  This bug entered with CVS revision
 > 	1.65 of ccd.c.
 > >How-To-Repeat:
 > 	ccdconfig ccd0 0 0 /dev/sd0d /dev/sd1d
 > >Fix:
 Yes indeed, the manual page says:
  To serially concatenate partitions, specify an interleave factor of 0.
 Mirroring configurations require an even number of components.
 
 This should fix it.
 
 Index: ccd.c
 ===================================================================
 RCS file: /cvs/src/sys/dev/ccd.c,v
 retrieving revision 1.68
 diff -u -p -r1.68 ccd.c
 --- ccd.c	29 Nov 2006 15:03:42 -0000	1.68
 +++ ccd.c	30 Mar 2007 11:34:48 -0000
 @@ -1151,7 +1151,7 @@ ccdioctl(dev_t dev, u_long cmd, caddr_t 
  			return (EBUSY);
  
  		if (ccio->ccio_ndisks == 0 || ccio->ccio_ndisks > INT_MAX ||
 -		    ccio->ccio_ileave ccio_ileave < 0)
  			return (EINVAL);
  
  		if ((error = ccdlock(cs)) != 0)

From: deraadt

Date: 2007-03-30 17:19:18

Synopsis: coding error in sys/dev/ccd.c falsely rejects interleave 0

State-Changed-From-To: open->closed
State-Changed-By: deraadt
State-Changed-When: Fri Mar 30 09:19:09 MDT 2007
State-Changed-Why: 
applied your patch (which matched mine)