NetBSD Problem Report #51275

From dholland@netbsd.org  Sun Jun 26 07:47:46 2016
Return-Path: <dholland@netbsd.org>
Received: from mail.netbsd.org (mail.netbsd.org [199.233.217.200])
	(using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits))
	(Client CN "mail.netbsd.org", Issuer "Postmaster NetBSD.org" (verified OK))
	by mollari.NetBSD.org (Postfix) with ESMTPS id 16F267A0EB
	for <gnats-bugs@gnats.NetBSD.org>; Sun, 26 Jun 2016 07:47:46 +0000 (UTC)
Message-Id: <20160626074745.90D8C85EA0@mail.netbsd.org>
Date: Sun, 26 Jun 2016 07:47:45 +0000 (UTC)
From: dholland@NetBSD.org
Reply-To: dholland@NetBSD.org
To: gnats-bugs@NetBSD.org
Subject: syncing a full lfs wedges
X-Send-Pr-Version: 3.95

>Number:         51275
>Category:       kern
>Synopsis:       syncing a full lfs wedges
>Confidential:   no
>Severity:       critical
>Priority:       medium
>Responsible:    dholland
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sun Jun 26 07:50:00 +0000 2016
>Last-Modified:  Wed Aug 03 17:45:00 +0000 2016
>Originator:     David A. Holland
>Release:        NetBSD 7.99.31 (20160620)
>Organization:
>Environment:
System: NetBSD amnesiac 7.99.31 NetBSD 7.99.31 (GENERIC) #14: Sun Jun 26 03:11:19 EDT 2016  dholland@valkyrie:/y/trees/netbsd/lfs2/sys/arch/amd64/compile/obj.amd64/GENERIC amd64
Architecture: x86_64
Machine: amd64
>Description:


	% dd if=/dev/zero of=image bs=102400 count=1000
	% newfs_lfs ./image
	% qemu-system-x86_64 -nographic -m 64 -hda rootdisk -hdb image
	(rootdisk is from anita install)
	[...]
	# mount -t lfs /dev/wd1b /mnt
	# cd /mnt
	# dd if=/dev/zero of=blef bs=64k
	dd: blef: No space left on device
	1228+0 records in
	1227+0 records out
	80412672 bytes transferred in 2.261 secs (35565091 bytes/sec)
	# sync
	# <hangs>

If you don't sync, things seem to be fine (at least for a while); if
the volume isn't full, syncing isn't lethal.

If you do sync you get a prompt back and there's 1-2 seconds before it
wedges.

>How-To-Repeat:

>Fix:

>Release-Note:

>Audit-Trail:

Responsible-Changed-From-To: kern-bug-people->dholland
Responsible-Changed-By: dholland@NetBSD.org
Responsible-Changed-When: Sun, 26 Jun 2016 17:17:46 +0000
Responsible-Changed-Why:
lf
s


From: coypu@SDF.ORG
To: gnats-bugs@netbsd.org
Cc: 
Subject: Re: kern/51275
Date: Wed, 3 Aug 2016 17:42:41 +0000

 Note that this isn't "LFS wedges", but "LFS wedges forever, and no
 writes will ever succeed"

 sync is irrelevant here.

 I've noted that the cleaner died when doing the big write (using crash
 and then ps/l |grep lfs, I didn't see the cleaner)

 After reboot:

 Cleaner trace wedges:
 crash> bt/a fffffe8132d42b60
 trace: pid 167 lid 1 at 0xfffffe80412888a0
 sleepq_block() at sleepq_block+0x97
 mtsleep() at mtsleep+0x102
 lfs_reserveavail.isra.2() at lfs_reserveavail.isra.2+0x173
 lfs_reserve() at lfs_reserve+0x140
 lfs_truncate() at lfs_truncate+0x162
 ulfs_inactive() at ulfs_inactive+0x11b
 VOP_INACTIVE() at VOP_INACTIVE+0x55
 vrelel() at vrelel+0x259
 lfs_bmapv() at lfs_bmapv+0x2a8
 lfs_fcntl() at lfs_fcntl+0xbcd
 VOP_FCNTL() at VOP_FCNTL+0x59
 sys_fcntl() at sys_fcntl+0x375
 syscall() at syscall+0x15b
 --- syscall (number 92) ---
 79468ce3dcba:

 All things hang there.

>Unformatted:

Home	PR Database Search