NetBSD Problem Report #42091

From john@reva.xtremeunix.com  Fri Sep 18 21:57:24 2009
Return-Path: <john@reva.xtremeunix.com>
Received: from mail.netbsd.org (mail.netbsd.org [204.152.190.11])
	by www.NetBSD.org (Postfix) with ESMTP id 6A45863BC1D
	for <gnats-bugs@gnats.NetBSD.org>; Fri, 18 Sep 2009 21:57:24 +0000 (UTC)
Message-Id: <200909182156.n8ILuvCX025277@reva.xtremeunix.com>
Date: Fri, 18 Sep 2009 21:56:57 GMT
From: john@ziaspace.com
Reply-To: john@ziaspace.com
To: gnats-bugs@gnats.NetBSD.org
Subject: ld devices can't be set up with more than 2^31 sectors
X-Send-Pr-Version: 3.95

>Number:         42091
>Category:       port-amd64
>Synopsis:       ld devices can't be set up with more than 2^31 sectors
>Confidential:   no
>Severity:       serious
>Priority:       high
>Responsible:    port-amd64-maintainer
>State:          closed
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Fri Sep 18 22:00:00 +0000 2009
>Closed-Date:    Mon May 27 04:46:58 +0000 2019
>Last-Modified:  Mon May 27 04:46:58 +0000 2019
>Originator:     John Klos
>Release:        NetBSD 5.0_STABLE (tested with current, too)
>Organization:

>Environment:


System: NetBSD reva.xtremeunix.com 5.0_STABLE NetBSD 5.0_STABLE (REVA-$Revision: 5.323 $) #0: Tue Sep 8 06:27:49 UTC 2009 john@sage.klos.com:/usr/obj/sys/arch/amiga/compile/REVA amiga
Architecture: m68k
Machine: amiga
>Description:




>How-To-Repeat:


Set up some sort of hardware RAID:

siisata0 at pci6 dev 0 function 0: vendor 0x1095 product 0x3132 (RAID mass storage, revision 0x01)
ataraid0: found 1 RAID volume
ld0 at ataraid0 vendtype 3 unit 0: nVidia ATA RAID-1 array
ld0: 1397 GB, 182401 cyl, 255 head, 63 sec, 512 bytes/sect x 2930277120 sectors

Then try to install NetBSD or to create a disklabel partition greater than 
2^31 sectors:

partition> P
4 partitions:
#        size    offset     fstype [fsize bsize cpg/sgs]
 d: 2930277120         0     unused      0     0        # (Cyl.      0 - 182401*)
partition> a
Filesystem type [?] [unused]: 4.2BSD
Start offset ('x' to start after partition 'x') [0c, 0s, 0M]: 64
Partition size ('$' for all remaining) [0c, 0s, 0M]: 2913499840
 a: 2147483648        64     4.2BSD      0     0     0  # (Cyl.      0*- 133674*)

>Fix:


>Release-Note:

>Audit-Trail:
From: "Jonathan A. Kollasch" <jakllsch@kollasch.net>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: port-amd64/42091: ld devices can't be set up with more than
 2^31 sectors
Date: Wed, 18 Nov 2009 19:52:55 +0000

 This looks like it could be caused by disklabel(8)'s getnum() returning
 'int'.  This function seems to only be used in interactive (-i) mode,
 so maybe -e mode would work.

 In any case, the disklabel can only hold 32-bit sector counts and offsets.
 So, until 2TiB+ disks come about (IOW the next generation of disks),
 we're mostly fine.

State-Changed-From-To: open->closed
State-Changed-By: dsl@NetBSD.org
State-Changed-When: Sun, 22 Nov 2009 17:44:10 +0000
State-Changed-Why:
I think all the issues with disklabels between 2^31 and 2^32 sectors
have been fixed. If not there will be duplicate PRs elsewhere.
Disks larger than 2^32 sectors are a bigger problem.


From: "Jonathan A. Kollasch" <jakllsch@kollasch.net>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: port-amd64/42091: ld devices can't be set up with more than
 2^31 sectors
Date: Wed, 25 Nov 2009 02:13:36 +0000

 Changing the "rv = (int) ..." lines in interact.c:getnum() to "rv =
 (uint32_t) ..." fixes the problem on a (virtual) 1.5TB drive for me.
 But it looks like much of that file isn't perfectly unsigned-safe.

State-Changed-From-To: closed->analyzed
State-Changed-By: jakllsch@NetBSD.org
State-Changed-When: Wed, 25 Nov 2009 02:23:50 +0000
State-Changed-Why:
A fix is possible, and the problem hasn't gone away.


From: Stephen Borrill <netbsd@precedence.co.uk>
To: gnats-bugs@netbsd.org
Cc: 
Subject: Re: port-amd64/42091: ld devices can't be set up with more than 2^31
 sectors
Date: Wed, 25 Nov 2009 10:22:01 +0000 (GMT)

 This is only a problem in the interactive part of disklabel(8) 
 (as shown by the fact that jakllsch's patch is for interact.c). It's not a 
 problem with the disklabel itself (which was dsl's reason for closing).

 If you hand-edit a disklabel and restore with disklabel -R it works fine.

 I'd be tempted to use daddr_t throughout and then check against UINT32_MAX 
 when coming to write out the label (or at various other times when a 
 warning would be appropriate).

From: coypu@sdf.org
To: gnats-bugs@netbsd.org
Cc: 
Subject: Re: port-amd64/42091: ld devices can't be set up with more than 2^31
 sectors
Date: Mon, 27 May 2019 02:37:46 +0000

 Is this an issue with GPT instead of disklabel?

 I think we were abandoning disklabel in favour of other formats instead
 of extending it to 64bit.

State-Changed-From-To: analyzed->closed
State-Changed-By: mlelstv@NetBSD.org
State-Changed-When: Mon, 27 May 2019 04:46:58 +0000
State-Changed-Why:
This is about a limitation in the disklabel tool, it was fixed with PR bin/43354.


>Unformatted:
 When using drives larger than 1TB in a hardware mirror, partitions can't have more than 2^31 sectors. 

Home	PR Database Search