$Id: WARNINGS,v 1.33 2006/05/19 16:33:33 chrfranke Exp $

The most recent version of this file can be found here:
http://smartmontools.cvs.sourceforge.net/smartmontools/sm5/WARNINGS?view=markup

The following are reports of serious problems (eg system lockup) which
were due to smartmontools.  There are DARWIN, LINUX, FREEBSD, SOLARIS
and WINDOWS sections below.


LINUX
-----

You may also wish to search the linux-kernel mailing list for problem
reports concerning smartmontools.  Here is the URL:
http://groups.google.com/groups?as_q=smartmontools&safe=images&ie=UTF-8&oe=UTF-8&as_ugroup=linux.kernel&lr=&num=100&hl=en

SYSTEM:   Any system with USB ports and USB storage devices
PROBLEM:  Using smartd/smartctl on USB "SCSI" storage devices can cause kernel hang
REPORTER: see link below
LINK:     https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=107615
NOTE:     USB storage devices are handled as SCSI devices by the kernel. But many of these
          devices do not comply with SCSI specs, and can cause the kernel to hang.
          Avoid using smartd/smartctl on these devices (they don't do SMART anyway).
          In particular, the use of smartd DEVICESCAN in /etc/smartd.conf can cause
          these devices (typically represented by /dev/sda or /dev/sdb) to hang, and
          the kernel to lock up.
FIXED:    This problem should be fixed in smartmontools-5.25 and greater.


SYSTEM:   Intel 875WP1-E motherboard with SATA drives on motherboard's SATA ports
PROBLEM:  smartd makes NTP time drift
REPORTER: nohez@cmie.com
LINK:     http://groups.google.de/groups?hl=en&lr=&ie=UTF-8&oe=UTF-8&safe=off&selm=Pine.LNX.4.33.0310111545530.1047-100000%40venus.cmie.ernet.in.lucky.linux.kernel
NOTE:     When using SATA disks, linux kernel k_smp-2.4.21-108 (SMP because
          of hyper-threading) and xntp-4.1.1-177, the server time went
          out of sync with system time.  Problem goes away when SATA
          disks removed.


SYSTEM:   Dell servers using AACRAID (SCSI)
PROBLEM:  Locked up, needed to be rebooted
REPORTER: drew@eastvan.bc.ca
LINK:     http://sourceforge.net/mailarchive/forum.php?thread_id=1311313&forum_id=12495


SYSTEM:   Box with Promise 20265 IDE-controller (pdc202xx-driver) and > 2.4.18 kernel with ide-taskfile support
PROBLEM:  Smartctl locks system solid when used on /dev/hd[ef].
REPORTER: Georg Acher <acher@in.tum.de>
LINK:     http://sourceforge.net/mailarchive/forum.php?thread_id=1457979&forum_id=12495
NOTE:     Lockup doesn't happen with 2.4.18 kernel, and doesn't affect /dev/hd[a-d]
          This appears to be a problem with the pdc202xx-driver and has been reported
          to the pdcx maintainers.  If you enable the Promise-BIOS (ATA100-BIOS) then
          everything will work fine.  But if you disable it, then the machine will hang.


SYSTEM:   Box with Promise 20262 IDE-controller
PROBLEM:  Smartctl locks system solid
REPORTER: Ben Low <ben@bdlow.net>
LINK:     http://sourceforge.net/mailarchive/message.php?msg_id=5074201
NOTE:     Similar to previous report: Promise Ultra66 2-port card (20262) which, with
          linux 2.4.20, suffers from the lockups reported above.  But it was
          impossible to enable the Promiste BIOS.  A kernel patch is referenced
          to fix the problem.


SYSTEM:   Promise 20265 IDE-controller
PROBLEM:  Smartctl locks system solid when used on CDROM/DVD device
REPORTER: see link below
LINK:     http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=208964
NOTE:     Problem seems to affect kernel 2.4.21 only.


SYSTEM:   Promise IDE-controllers and perhaps others also
PROBLEM:  System freezes under heavy load, perhaps when running SMART commands
REPORTER: Mario 'BitKoenig' Holbe Mario.Holbe@RZ.TU-Ilmenau.DE
LINK:     http://groups.google.de/groups?hl=en&lr=&ie=UTF-8&oe=UTF-8&selm=1wUXW-2FA-9%40gated-at.bofh.it
NOTE:     Before freezing, SYSLOG shows the following message(s)
          kernel: hdf: dma timer expiry: dma status == 0xXX
       	  where XX is two hexidecimal digits. This may be a kernel bug
          or an underlying hardware problem.  It's not clear if
          smartmontools plays a role in provoking this problem.  FINAL
          NOTE: Problem was COMPLETELY resolved by replacing the power
          supply.  See URL above, entry on May 29, 2004 by Holbe.  Other
          things to try are exchanging cables, and cleaning PCI slots.

FREEBSD
-------

[No problem reports yet.]


SOLARIS
-------

[No problem reports yet.]


CYGWIN and WINDOWS
------------------

SYSTEM:   Any Windows NT4, 2000 or XP system.
PROBLEM:  Use of undocumented system calls for IDE/ATA read log
          (smartctl -l, --log, -a, --all) may affect system stability.
REPORTER: Christian Franke <smartmontools-support@lists.sourceforge.net>
NOTE:     The IOCTL call SMART_RCV_DRIVE_DATA does not support 
          ATA_SMART_READ_LOG_SECTOR on NT4/2000/XP. The Win32 
          implementation of smartctl/smartd uses undocumented
          and possibly buggy system calls for this purpose:
          NT4: IOCTL_SCSI_PASS_THROUGH with undocumented pseudo SCSI
          command SCSIOP_ATA_PASSTHROUGH (0xCC).
          2000/XP: Undocumented IOCTL_IDE_PASS_THROUGH.


DARWIN
------

SYSTEM:   Any system before Tiger
PROBLEM:  Can't switch off SMART, can't switch off auto-save, can't run
          short tests.
REPORTER: Geoff Keating <geoffk@geoffk.org>
NOTE:     There's a bug in the system library: when you ask it to
          do any of these things, it does the inverse (switches on,
          runs extended tests).  Radar 3727283.

SYSTEM:	  All known systems
PROBLEM:  When drive is asleep, SMART commands fail
REPORTER: Geoff Keating <geoffk@geoffk.org>
NOTE:	  You can prevent the drive from sleeping by saying
	  pmset -a disksleep 0
	  or by unchecking the 'Put the hard disk(s) to sleep when possible'
	  checkbox in the Energy Saver preferences.  Radar 4094403.