Sophie

Sophie

distrib > Mandriva > 9.1 > ppc > by-pkgid > 11aee6d3ccb87648f555e01c51133785 > files > 7

smartmontools-5.1-9.2mdk.ppc.rpm

CHANGELOG for smartmontools

$Id: CHANGELOG,v 1.103 2003/03/13 15:34:37 ballen4705 Exp $

Copyright (C) 2002-3 Bruce Allen <smartmontools-support@lists.sourceforge.net>

Home page of code is: http://smartmontools.sourceforge.net

This program is free software; you can redistribute it and/or modify it
under the terms of the GNU General Public License as published by the Free
Software Foundation; either version 2, or (at your option) any later
version.

You should have received a copy of the GNU General Public License (for
example COPYING); if not, write to the Free Software Foundation, Inc., 675
Mass Ave, Cambridge, MA 02139, USA.

This code was originally developed as a Senior Thesis by Michael Cornwell
at the Concurrent Systems Laboratory (now part of the Storage Systems
Research Center), Jack Baskin School of Engineering, University of
California, Santa Cruz. http://ssrc.soe.ucsc.edu/

Maintainers/Developers Key:
[BA] Bruce Allen            <smartmontools-support@lists.sourceforge.net>
[EB] Erik Inge Bolsø        <knan@mo.himolde.no>
[SB] Stanislav Brabec       <sbrabec@suse.cz>
[PC] Peter Cassidy          <pcassidy@mac.com>
[FM] Frederic L. W. Meunier <0@pervalidus.net>
[PW] Phil Williams          <phil@subbacultcha.demon.co.uk>
[DG] Douglas Gilbert        <dougg@torque.net>

NOTES FOR FUTURE RELEASES: see TODO file.

CURRENT RELEASE (see VERSION file in this directory):
 
  <ADDITIONS TO THE CHANGE LOG SHOULD BE ADDED HERE, PLEASE>

smartmontools-5.1-9
  
  [BA] smartctl: if HDIO_DRIVE_TASK ioctl() is not implemented (no
       kernel support, then try to assess drive health by examining
       Attribute values/thresholds directly.

  [BA] smartd/smartctl: added -v 200,writeerrorcount option/Directive
       for Fujitsu disks.

  [BA] smartd: Now send email if any of the SMART commands fails,
       or if open()ing the device fails.  This is often noted
       as a common disk failure mode.

  [BA] smartd/smartctl: Added -v N,raw8 -v N,raw16 and -v N,raw48
       Directives/Options for printing Raw Attributes in different
       Formats.

  [BA] smartd: Added -r ID and -R ID for reporting/tracking Raw
       values of Attributes.

  [BA] smartd/smartctl: Changed printing of spin-up-time attribute
       raw value to reflect current/average as per IBM standard.

  [BA] smartd/smartctl: Added -v 9,seconds option for disks which
       use Attribute 9 for power-on lifetime in seconds.

  [BA] smartctl: Added a warning message so that users of some IBM
       disks are warned to update their firmware.  Note: we may want
       to add a command-line flag to disable the warning messages.
       I have done this in a general way, using regexp, so that we
       can add warnings about any type of disk that we wish...

smartmontools-5.1-7

  [BA] smartd: Created a subdirectory examplescripts/ of source
       directory that contains executable scripts for the -M exec PATH
       Directive of smartd.

smartmontools-5.1-5

  [BA] smartd: DEVICESCAN in /etc/smartd.conf
       can now be followed by all the same Directives as a regular
       device name like /dev/hda takes.  This allows one to use
       (for example):
       DEVICESCAN -m root@yoyodyne.com
       in the /etc/smartd.conf file.

  [BA] smartd: Added -c (--checkonce) command-line option. This checks
       all devices once, then exits.  The exit status can be
       used to learn if devices were detected, and if smartd is
       functioning correctly. This is primarily for Distribution
       scripters.

  [BA] smartd: Implemented -M exec Directive for
       smartd.conf.  This makes it possible to run an
       arbitrary script or mailing program with the
       -m option.

  [PW] smartd: Modified -M Directive so that it can be given
       multiple times.  Added -M exec Directive.

smartmontools-5.1-4

  [BA] Fixed bug in smartctl pointed out by Pierre Gentile.
       -d scsi didn't work because tryata and tryscsi were
       reversed -- now works on /devfs SCSI devices.

  [BA] Fixed bug in smartctl pointed out by Gregory Goddard
       <ggoddard@ufl.edu>.  Manual says that bit 6 of return
       value turned on if errors found in smart error log.  But
       this wasn't implemented.

smartmontools-5.1-3

  [BA] Modified printing format for 9,minutes to read
       Xh+Ym not X h + Y m, so that fields are fixed width.

  [BA] Added Attribute 240 "head flying hours"

smartmontools-5.1.1

  [BA] As requested, local time/date now printed by smartctl -i

  [PW] Added "help" argument to -v for smartctl

  [PW] Added -D, --showdirectives option to smartd

  [DG] add '-l selftest' capability for SCSI devices (update smartctl.8)

  [BA] smartd,smartctl: added additional Attribute modification option
       -v 220,temp and -v 9,temp.

  [PW] Renamed smartd option -X to -d

START OF SMARTMONTOOLS 5.1 series

smartmontools-5.0.50

  [PW] Changed smartd.conf Directives -- see man page

  [BA/DG] Fixed uncommented comment in smartd.conf

  [DG] Correct 'Recommended start stop count' for SCSI devices

  [PW] Replaced smartd.conf directive -C with smartd option -i

  [PW] Changed options for smartctl -- see man page.

  [BA] Use strerror() to generate system call error messages.

  [BA] smartd: fflush() all open streams before fork().

  [BA] smartctl, smartd simplified internal handling of checksums
  for simpler porting and less code.

smartmontools-5.0.49

  [PW] smartd --debugmode changed to --debug

  [BA] smartd/smartctl added attribute 230 Head Amplitude from
  IBM DPTA-353750.

  [PW] Added list of proposed new options for smartctl to README.

  [PW] smartd: ParseOpts() now uses getopt_long() if HAVE_GETOPT_LONG is
  defined and uses getopt() otherwise.  This is controlled by CPPFLAGS in
  the Makefile.

  [BA] smartd: Fixed a couple of error messages done with perror()
  to redirect them as needed.

smartmontools-5.0.48

  [BA] smartctl: The -O option to enable an Immediate off-line test
  did not print out the correct time that the test would take to
  complete.  This is because the test timer is volatile and not
  fixed.  This has been fixed, and the smartctl.8 man page has been
  updated to explain how to track the Immediate offline test as it
  progresses, and to further emphasize the differences between the
  off-line immediate test and the self-tests.

  [BA] smartd/smartctl: Added new attribute (200) Multi_Zone_Error_Rate

  [BA] smartctl: modified so that arguments could have either a single -
  as in -ea or multiple ones as in -e -a.  Improved warning message for
  device not opened, and fixed error in redirection of error output of
  HD identity command.

  [PW] smartd: added support for long options.  All short options are still
  supported; see manpage for available long options.

  [BA] smartctl.  When raw Attribute value was 2^31 or larger, did
  not print correctly.

smartmontools-5.0.46

  [BA] smartd: added smartd.conf Directives -T and -s.  The -T Directive
  enables/disables Automatic Offline Testing.  The -s Directive
  enables/disables Attribute Autosave. Documentation and
  example configuration file updated to agree.

  [BA] smartd: user can make smartd check the disks at any time
  (ie, interrupt sleep) by sending signal SIGUSR1 to smartd.  This
  can be done for example with:
  kill -USR1 <pid>
  where <pid> is the process ID number of smartd.

  [EB] scsi: don't trust the data we receive from the drive too
  much. It very well might have errors (like zero response length).
  Seen on Megaraid logical drive, and verified in the driver source.

  [BA] smartd: added Directive -m for sending test email and
  for modifying email reminder behavior.  Updated manual, and sample
  configuration file to illustrate & explain this.

  [BA] smartd: increased size of a continued smartd.conf line to
  1023 characters.

  [BA] Simplified Directive parsers and improved warning/error
  messages.

smartmontools-5.0.45

  [EB] Fixed bug in smartd where testunitready logic inverted
  prevented functioning on scsi devices.
  The bug in question only affects smartd users with scsi devices.
  To see if your version of smartd has the testunitready() bug, do
     smartd -V
  If the version of the module smartd.c in a line like:
     Module: smartd.c      revision: 1.66   date: 2002/11/17
  has a revision greater than or equal to 1.30, and less than or equal to
  1.64, then your version of the code has this problem.
  This problem affected releases starting with RELEASE_5_0_16 up to and
  including RELEASE_5_0_43.

  [BA] Added testunitnotready to smartctl for symmetry with smartd.

  [SB] added Czech descriptions to .spec file
  [SB]  corrected comment in smartd.conf example

  [BA] Changed way that entries in the ATA error log are printed,
  to make it clearer which is the most recent error and
  which is the oldest one.

  NOTE: All changes made prior to this point were done by Bruce Allen
  [BA] although several of them had been suggested by earlier postings
  by Stanislav Brabec [SB].

smartmontools-5.0.43

  Changed Temperature_Centigrade to Temperature_Celsius.
  The term "Centigrade" ceased to exist in 1948.  (c.f
  http://www.bartleby.com/64/C004/016.html).

smartmontools-5.0.42

  Modified SCSI device check to also send warning emails if
  requested in directives file.

  Added a new smartd configuration file Directive: -M ADDRESS.
  This sends a single warning email to ADDRESS for failures or
  errors detected with the -c, -L, -l, or -f Directives.

smartmontools-5.0.38

  Modified perror() statements in atacmds.c so that printout for SMART
  commands errors is properly suppressed or queued depending upon users
  choices for error reporting modes.

  Added Italian descriptions to smartmontools.spec file.

  Started impementing send-mail-on-error for smartd; not yet enabled.

  Added -P (Permissive) Directive to smartd.conf file to allow SMART
  monitoring of pre-ATA-3 Rev 4 disks that have SMART but do not have
  a SMART capability bit.
 
  Removed charset encodings from smartmontools.spec file for non-English
  fields.

smartmontools-5.0.32

  Added manual page smartd.conf.5 for configuration file.

  smartctl: Missing ANSI prototype in failuretest(); fixed.

  smartctl: Checksum warnings now printed on stdout, or are silent,
  depending upon -q and -Q settings.

smartmontools-5.0.31

  Changed Makefile so that the -V option does not reflect file state
  before commit!

  smartctl: added new options -W, -U, and -P to control if and how the
  smartctl exits if an error is detected in either a SMART data
  structure checksum, or a SMART command returns an error.

  modified manual page to break options into slightly more logical
  categories.

  reformatted 'usage' message order to agree with man page ordering

  modified .spec file so that locale information now contains
  character set definition.   Changed pt_BR to pt since we do not use any
  aspect other than language.  See man setlocale.

smartmontools-5.0.30
  smartctl: added new options -n and -N to force device to be ATA or SCSI
  smartctl: no longer dies silently if device path does not start/dev/X
  smartctl: now handles arbitrary device paths

smartmontools-5.0.29
  Modified .spec file and Makefile to make them more compliant with
  the "right" way of doing things.

smartmontools-5.0.26
  Fixed typesetting error in man page smartd.8

  Removed redundant variable (harmless) from smartd.c

smartmontools-5.0.25

  Added a new directive for the configuration file.  If the word
  DEVICESCAN appears before any non-commented material in the
  configuration file, then the confi file will be ignored and the
  devices wil be scanned.

smartmontools-5.0.24

    Note: it has now been confirmed that the code modifications between
    5.0.23 and 5.0.24 have eliminated the GCC 3.2 problems.  Note that
    there is a GCC bug howerver, see #848 at
    http://gcc.gnu.org/cgi-bin/gnatsweb.pl?database=gcc&cmd=query

      Added new Directive for Configuration file:
-C <N> This sets the time in between disk checks to be <N>
      seconds apart.  Note that  although  you  can  give
       this Directive multiple times on different lines of
       the configuration file, only the final  value  that
       is  given  has  an  effect,  and applies to all the
       disks.  The default value of <N> is 1800  sec,  and
       the minimum allowed value is ten seconds.

    Problem wasn't the print format. F.L.W. Meunier <0@pervalidus.net>
    sent me a gcc 3.2 build and I ran it under a debugger.  The
    problem seems to be with passing the very large (2x512+4) byte
    data structures as arguments.  I never liked this anyway; it was
    inherited from smartsuite.  So I've changed all the heavyweight
    functions (ATA ones, anyone) to just passing pointers, not hideous
    kB size structures on the stack.  Hopefully this will now build OK
    under gcc 3.2 with any sensible compilation options.

smartmontools-5.0.23

    Because of reported problems with GCC 3.2 compile, I have gone
    thorough the code and explicitly changed all print format
    parameters to correspond EXACTLY to int unless they have to be
    promoted to long longs.  To quote from the glibc bible: [From
    GLIBC Manual: Since the prototype doesn't specify types for
    optional arguments, in a call to a variadic function the default
    argument promotions are performed on the optional argument
    values. This means the objects of type char or short int (whether
    signed or not) are promoted to either int or unsigned int, as
    appropriate.]

smartmontools-5.0.22

    smartd, smartctl now warn if they find an attribute whose ID
    number does not match between Data and Threshold structures.

    Fixed nasty bug which led to wrong number of arguments for a
    varargs statement, with attendent stack corruption.  Sheesh!
    Have added script to CVS attic to help find such nasties in the
    future.

smartmontools-5.0.21

    Eliminated some global variables out of header files and other
    minor cleanup of smartd.

smartmontools-5.0.20

    Did some revision of the man page for smartd and made the usage
    messages for Directives 100% consistent.

smartmontools-5.0-19

    smartd: prints warning message when it gets SIGHUP, saying that it
    is NOT re-reading the config file.

    smartctl: updated man page to say self-test commands -O,x,X,s,S,A
    appear to be supported in the code.  [I can't test these,  can anyone
    report?]

smartmontools-5.0-18

    smartctl: smartctl would previously print the LBA of a self-test
    if it completed, and the LBA was not 0 or 0xff...f However
    according to the specs this is not correct.  According to the
    specs, if the self-test completed without error then LBA is
    undefined.  This version fixes that.  LBA value only printed if
    self-test encountered an error.

smartmontools-5.0-17

    smartd has changed significantly. This is the first CVS checkin of
    code that extends the options available for smartd.  The following
    options can be placed into the /etc/smartd.conf file, and control the
    behavior of smartd.
    Configuration file Directives (following device name):
    -A     Device is an ATA device
    -S     Device is a SCSI device
    -c     Monitor SMART Health Status
    -l     Monitor SMART Error Log for changes
    -L     Monitor SMART Self-Test Log for new errors
    -f     Monitor for failure of any 'Usage' Attributes
    -p     Report changes in 'Prefailure' Attributes
    -u     Report changes in 'Usage' Attributes
    -t     Equivalent to -p and -u Directives
    -a     Equivalent to -c -l -L -f -t Directives
    -i ID  Ignore Attribute ID for -f Directive
    -I ID  Ignore Attribute ID for -p, -u or -t Directive
    #      Comment: text after a hash sign is ignored
    \      Line continuation character

    cleaned up functions used for printing CVS IDs.  Now use string
    library, as it should be.

    modified length of device name string in smartd internal structure
    to accomodate max length device name strings

    removed un-implemented (-e = Email notification) option from
    command line arg list.  We'll put it back on when implemeneted.

    smartd now logs serious (fatal) conditions in its operation at
    loglevel LOG_CRIT rather than LOG_INFO before exiting with error.

    smartd used to open a file descriptor for each SMART enabled
    device, and then keep it open the entire time smartd was running.
    This meant that some commands, like IOREADBLKPART did not work,
    since the fd to the device was open.  smartd now opens the device
    when it needs to read values, then closes it.  Also, if one time
    around it can't open the device, it simply prints a warning
    message but does not give up.  Have eliminated the .fd field from
    data structures -- no longer gets used.

    smartd now opens SCSI devices as well using O_RDONLY rather than
    O_RDWR.  If someone can no longer monitor a SCSI device that used
    to be readable, this may well be the reason why.

    smartd never checked if the number of ata or scsi devices detected
    was greater than the max number it could monitor.  Now it does.

smartmontools-5.0-16

    smartd on startup now looks in the configuration file /etc/smartd.conf for
    a list of devices which to include in its monitoring list.  See man page
    (man smartd) for syntax.

    smartd: close file descriptors of SCSI device if not SMART capable
    Closes ALL file descriptors after forking to daemon.

    added new temperature attribute (231, temperature)

    smartd: now open ATA disks using O_RDONLY

smartmontools-5.0-11

    smartd now prints the name of a failed or changed attribute
    into logfile, not just ID number

    Changed name of -p (print version) option to -V

    Minor change in philosophy: if a SMART command fails or the device
    appears incapable of a SMART command that the user has asked for,
    complain by printing an error message, but go ahead and try
    anyway.  Since unimplemented SMART commands should just return an
    error but not cause disk problems, this should't cause any
    difficulty.

    Added two new flags: q and Q.  q is quiet mode - only print: For
    the -l option, errors recorded in the SMART error log; For the -L
    option, errors recorded in the device self-test log; For the -c
    SMART "disk failing" status or device attributes (pre-failure or
    usage) which failed either now or in the past; For the -v option
    device attributes (pre-failure or usage) which failed either now
    or in the past.  Q is Very Quiet mode: Print no ouput.  The only
    way to learn about what was found is to use the exit status of
    smartctl.

    smartctl now returns sensible values (bitmask).  See smartctl.h
    for the values, and the man page for documentation.

    The SMART status check now uses the correct ATA call.  If failure
    is detected we search through attributes to list the failed ones.
    If the SMART status check shows GOOD, we then look to see if their
    are any usage attributes or prefail attributes have failed at any
    time.  If so we print them.

    Modified function that prints vendor attributes to say if the
    attribute has currently failed or has ever failed.

    -p option now prints out license info and CVS strings for all
    modules in the code, nicely formatted.

    Previous versions of this code (and Smartsuite) only generate
    SMART failure errors if the value of an attribute is below the
    threshold and the prefailure bit is set.  However the ATA Spec
    (ATA4 <=Rev 4) says that it is a SMART failure if the value of an
    attribute is LESS THAN OR EQUAL to the threshold and the
    prefailure bit is set.  This is now fixed in both smartctl and
    smartd.  Note that this is a troubled subject -- the original
    SFF 8035i specification defining SMART was inconsistent about
    this.  One section says that Attribute==Threshold is pass,
    and another section says it is fail.  However the ATA specs are
    consistent and say Attribute==Threshold is a fail.

    smartd did not print the correct value of any failing SMART attribute.  It
    printed the index in the attribute table, not the attribute
    ID. This is fixed.

    when starting self-tests in captive mode ioctl returns EIO because
    the drive has been busied out.  Detect this and don't return an eror
    in this case.  Check this this is correct (or how to fix it?)
 
    fixed possible error in how to determine ATA standard support
    for devices with no ATA minor revision number.

    device opened only in read-only not read-write mode.  Don't need R/W 
    access to get smart data. Check this with Andre.

    smartctl now handles all possible choices of "multiple options"
    gracefully.  It goes through the following phases of operation,
    in order: INFORMATION, ENABLE/DISABLE, DISPLAY DATA, RUN/ABORT TESTS.
    Documentation has bee updated to explain the different phases of
    operation.  Control flow through ataPrintMain()
    simplified.

    If reading device identity information fails, try seeing if the info
    can be accessed using a "DEVICE PACKET" command.  This way we can
    at least get device info.

    Modified Makefile to automatically tag CVS archive on issuance of
    a release

    Modified drive detection so minor device ID code showing ATA-3 rev
    0 (no SMART) is known to not be SMART capable.

    Now verify the checksum of the device ID data structure, and of the
    attributes threshold structure.  Before neither of these
    structures had their checksums verified.

    New behavior vis-a-vis checksums.  If they are wrong, we log
    warning messages to stdout, stderr, and syslog, but carry on
    anyway.  All functions now call a checksumwarning routine if the
    checksum doesn't vanish as it should.

    Changed Read Hard Disk Identity function to get fresh info from
    the disk on each call rather than to use the values that were read
    upon boot-up into the BIOS.  This is the biggest change in this
    release.  The ioctl(device, HDIO_GET_IDENTITY, buf ) call should
    be avoided in such code.  Note that if people get garbled strings
    for the model, serial no and firmware versions of their drives,
    then blame goes here (the BIOS does the byte swapping for you,
    apparently!)
    
    Function ataSmartSupport now looks at correct bits in drive
    identity structure to verify first that these bits are valid,
    before using them.
    
    Function ataIsSmartEnabled() written which uses the Drive ID state
    information to tell if SMART is enabled or not.  We'll carry this
    along for the moment without using it.

    Function ataDoesSmartWork() guaranteed to work if the device
    supports SMART.

    Replace some numbers by #define MACROS

    Wrote Function TestTime to return test time associated with each
    different type of test.

    Thinking of the future, have added a new function called
    ataSmartStatus2().  Eventually when I understand how to use the
    TASKFILE API and am sure that this works correctly, it will
    replace ataSmartStatus().  This queries the drive directly to
    see if the SMART status is OK, rather than comparing thresholds to
    attribute values ourselves. But I need to get some drives that fail
    their SMART status to check it.


smartmontools-5.0-10
    Removed extraneous space before printing in some error messages
    Fixed additional typos in documentation
    Fixed some character buffers that were too short for their contents.

smartmontools-5.0-9

    Put project home path into all source files near the top
    Corrected typos in the documentation
    Modified Makefile so that Mandrake Cooker won't increment version number
    (unless they happen to be working on my machine, which I doubt!)

smartmontools-5.0-8:

    For IBM disks whose raw temp data includes three temps. print all
    three

    print timestamps for error log to msec precision

    added -m option for Hitachi disks that store power on life in
    minutes

    added -L option for printing self-test error logs

    in -l option, now print power on lifetime, so that one can see
    when the error took place

    updated SMART structure definitions to ATA-5 spec

    added -p option

    added -f and -F options to enable/disable autosave threshold
    parameters

    changed argv parsing to use getops -- elminate buffer overflow
    vulnerability

    expanded and corrected documentation

    fixed problem with smartd.  It did not actually call
    ataSmartEnable()!  Since the argument was left out, the test
    always suceeded because it evaluated to a pointer to the function.

    smartd: closed open file descriptors if device does not support
    smart. Note: this still needs to be fixed for SCSI devices


smartmontools-5.0-0  STARTED with smartsuite-2.1-2