Before Bride of Frank | Boycott Citgo | 9/11 White Van Camera Crew | Hudson County Politics | Hudson County Facts | Outlaw Motorcycle Gangs | New Jersey Mafia | Hal Turner, FBI Informant
Removing Viruses and Spyware | Reinstalling Windows XP, Vista, or 7 | Reset Windows Passwords | Windows Blue Screen of Death | Computer Noise | Don't Trust External Hard Drives!
Internet Advertising SEO - Search Engine Optimization - Search Engine Marketing - SEM iTunes Beatles Billboards Anthony Olszewski Intellectual Predator Email this Page

CLASSIC PC | YOU DO SUCCEED | DOWNLOAD | FAQ LIBRARY
LINKS | LINUX | MAIN MENU | PS2 | REFERENCE | USENET


Data Loss on QUANTUM FIREBALL TM Series EIDE Harddrives with Windows for Workgroups V3.11: fixed by Firmware Rev. A6B.2DNR

Copyright (C) 1997 by Erik van Straten. All (registered) trademarks are recognized.
Last major change: 04/16/1997 by EvS. Last file modification:

CONTENTS

Return to COMPUTERCRAFT main menu

LATEST NEWS: GET THE PATCH FROM INTERNET!

Firmware Revision A6B.2DNR (available as a patch) fixes the bug, and is available for download from http://support.quantum.com since March 31, 1997!
On Jan 23, 1997, Quantum tech. support UK informed me that they had been able to reproduce the problem described below. The reason why it took so long before Quantum aknowledged this problem, is that it is hard to reproduce, especially when the PC is configured as described in the initial report I submitted to Quantum. In addition, apparently only a few people complained about this issue while huge amounts of disks have been sold. Because of this it is fully understandable that this issue did not get a high priority status for Quantum, considering that initially they really tried to reproduce the problem described below but did not succeed. Therefore I appreciate the fact that Quantum kept communicating with me, sent me a drive for testing and kept providing me with patches they thought might help. So although there is a problem with these drives, it has been handled in very a professional way (other companies might have sent their lawyers instead of a drive).
Buying hardware (and software) from companies that deny that their products (may) have bugs, and refuse to offer fixes, should be avoided in my personal opinion; a service minded organization is an absolute requirement. Fortunately, more and more companies are sharing information about problems with their products with the public. These companies include Western Digital, Intel, Microsoft and Quantum. This list is far from complete; before you buy, have a look on a companies' www and/or ftp site to see if they have such information online. If not, there always is a chance that that company sells bug-free products; the choice is yours!
Thanks to the persistence of some Quantum employees (see the acknowledgements below) and my stubborness (I was -and am- 100% convinced about this problem) Quantum continued testing, and finally was able to reproduce it and acknowledge it's existence. On Jan. 28, 1997, Quantum sent me a firmware patch which I tested that same night, and found that it indeed fixes the bug! (the tests that confirm this are described in the research history). I informed Quantum that the fix works, and meanwhile they continued their testing program to make sure that the fix doesn't introduce other issues. After finishing that, the patch was released to the public on Feb. 7, 1997, however, as far as I know, only on their UK BBS. Unfortunately, at that time not yet on any Internet site!
However, at the beginning of April, a number of people pointed out that Quantum has now confirmed the problem I describe here on one of their websites. Further, you can download the patch from there, and don't have to struggle with calling one of Quantum's BBSes. Note that you cannot find information about this issue on Quantum's regular website, www.quantum.com, but rather on the website support.quantum.com (this is te main main page, for the patch program see Aplication Notes - WfW311).
I do understand that you may be angry with Quantum if you've lost data because of the bug. However, I still suggest that, should you need to contact Quantum in any way, you tell Quantum that you like the fact that they publish and fix problems like these. Please help educate hardware and software manufacturers!
NOTE: The firmware that fixes the problem is Rev. A6B.2DNR. The "NR" part seems to be some Quantum-internal extension; after patching the drive it will report Rev. A6B.2D00 to programs that execute an "identify drive" command. Also note that Quantum may release later versions; this page may very soon be out of date! For further information, see "How to obtain and apply the patch".
Quantum assured me that the problem described here only occurs in the Fireball TM EIDE drives, not in the SCSI drives or non-TM Fireball EIDE drives.

Back to Top

INTRODUCTION

Users of all Quantum Fireball TM series EIDE drives manufactured before April 1, 1997, with Firmware Revisions from A6B.1F00 upto (but not including) A6B.2DNR, may experience significant data loss when they use an optimized Windows for Workgroups V3.11. After quitting WfW, back in DOS, a DIR command may not show all entries in C:\ you had before (from now on, I will refer to this as "the problem"). However, at that time only DOS is confused, nothing is really deleted yet; but data loss will occur when you do anything that causes DOS to write the erroneous rootdirectory contents to the disk. For example, should you create or remove a subdirectory, edit AUTOEXEC.BAT or even use a tool like NCD (Norton Change Directory), then DOS will "update" the actual rootdirectory sectors on your drive with wrong data. Afterwards, files and subdirectories will have disappeared, and CHKDSK or SCANDISK will report a large number of lost clusters. Depending on a number of parameters, you may or may not run into this problem, but it potentially exists for the drives with the firmware revisions I have tested.
I must note that the error does not always show up. Depending on certain parameters I'll specify later, it may show up every time, sometimes, or never at all. Of course I did a lot of research to determine that this error is due to these Quantum drives, and not some other problem. Tests were conducted in very different PC's, and the error was found to be reproducable, but with these drives only. On Western Digital WDAC22100H drives (with exactly the same number of cylinders, heads and sectors in the BIOS setup) configured with an identical software setup I was NOT able to reproduce these problems. Please check out the research history for more details.
Further, some people have reported data loss while using other operating systems, however, I have no proof of this. I was only able to reproduce the problems with WfW3.11 and MS-DOS 6.22. I did a few tests with MS Windows NT Workstation V3.51 (Service Pack 5 installed), and Windows95 OSR2 (FAT32 installed) and was not able to reproduce any problems. However, possibly problems with these harddrives (with mentioned old firmware) may arise under other operating systems, including MS Windows 95, NT4.x, Linux, OS/2, and others. Even if current versions of them do not invoke the problem, possibly future versions of them may do just that. In addition, adding new drivers like bus-master drivers for some motherboards may invoke this problem under other conditions. But again, I was not able to reproduce any problems under other OS'es than WfW311 and DOS622.
I would like to stress that -as far as I know- this problem only involves EIDE Fireball TM-series drives, not the older Fireball, any of the SCIS drives, or the (planned?) Fireball ST-series drives! Please also read my disclaimer below.


Back to Top


WHAT THE PROBLEM LOOKS LIKE

With my test configuration, a DIR /A in C:\ should result in the correct output as shown here:
(Compare this to the erroneous DIR listing)

     Volume in drive C is DISK1_VOL1
     Volume Serial Number is D96C-4DDB
     Directory of C:\

    IO       SYS        40,774 05-31-94   6:22a
    MSDOS    SYS        38,138 05-31-94   6:22a
    DOS          <DIR>         11-05-96   1:21a
    COMMAND  COM        54,645 05-31-94   6:22a
    WINA20   386         9,349 05-31-94   6:22a
    CONFIG   OLD            71 10-09-96   9:47p
    AUTOEXEC OLD            78 10-09-96   9:47p
    WINDOWS      <DIR>         11-05-96   1:22a
    CONFIG   SYS           119 10-09-96  10:01p
    AUTOEXEC BAT            99 11-05-96  12:47a
    386SPART PAR     2,113,536 11-05-96   1:47p
    1            <DIR>         11-05-96   1:24a
    2            <DIR>         11-05-96   1:24a
    3            <DIR>         11-05-96   1:24a
    4            <DIR>         11-05-96   1:24a
    5            <DIR>         11-05-96   1:24a
    6            <DIR>         11-05-96   1:24a
    7            <DIR>         11-05-96   1:24a
    8            <DIR>         11-05-96   1:24a
    CTATBUS4 DOC           968 06-30-95   3:18p
    CTATBUS4 EXE        29,712 06-30-95   3:06p
    !README! TXT        23,075 10-11-96   2:18p
           22 file(s)      2,310,564 bytes
                       1,043,709,952 bytes free

But sometimes, right after quitting WfW, DIR /A looks like this:
(Compare this to the correct DIR listing)

     Volume in drive C is DISK1_VOL1
     Volume Serial Number is D96C-4DDB
     Directory of C:\

    IO       SYS        40,774 05-31-94   6:22a
    MSDOS    SYS        38,138 05-31-94   6:22a
    DOS          <DIR>         11-05-96   1:21a
    COMMAND  COM        54,645 05-31-94   6:22a
    WINA20   386         9,349 05-31-94   6:22a
    CONFIG   OLD            71 10-09-96   9:47p
    AUTOEXEC OLD            78 10-09-96   9:47p
    WINDOWS      <DIR>         11-05-96   1:22a
    CONFIG   SYS           119 10-09-96  10:01p
    AUTOEXEC BAT            99 11-05-96  12:47a
    386SPART PAR     2,113,536 11-05-96   1:49p
    1            <DIR>         11-05-96   1:24a
    2            <DIR>         11-05-96   1:24a
    3            <DIR>         11-05-96   1:24a
    4            <DIR>         11-05-96   1:24a
           15 file(s)      2,256,809 bytes
                       1,043,709,952 bytes free

Note the missing entries, but the same number of free bytes on the drive! Also note that I have created dummy subdirectories numbered from 1 to 8 to show the problem. The program CTATBUS4 (described on my tools page) has nothing to do with the problem, it is on the disk but not running. With the configuration above, I do only get the erroneous results if a number of conditions are met, as described below, and only after quitting WfW. The erroneous listing is caused by the fact that DOS has only evaluated entries from the first rootdirectory sector. DOS 6.22 (and WfW311) use 32 bytes per entry, and each sector consists of 512 bytes, thus can store 16 entries (note: DOS7, Windows95 and Windows NT use a variable amount of entries to store long filenames in). DOS reports 15 rather than 16 entries because one entry is used to store the volume label in, and is not listed as a file-entry.
In another test setup, I found that if WINDOWS would have been invisible now (because it's entry is in another rootdir sector) you wouldn't be able to start it again! The same applies to any program or file you're trying to access that is hidden now, including contents of subdirectories currently hidden. However, I was able to restart Windows under the conditions listed above without error messages about a missing or corrupted swapfile.Apparently the error above is reset before Windows opens it's swapfile (which is usually hidden, DIR only lists it here because of the /A parameter).
Note: in all cases I saw the error like this, simpy running CHKDSK was sufficient to let DOS re-read the rootdirectory sectors and have correct DIR output afterwards. If you do not want to (or are unable to) apply the patch, a simple solution to prevent data loss is to move COMMAND.COM's directory entry out of the first directory sector as described below, or otherwise, immediately reboot your PC if you see such a truncated rootdirectory listing.

Back to Top

AFFECTED DRIVES

All Quantum Fireball TM EIDE drives manufactured before April 1, 1997 are likely to be affected by this problem. This includes the following drives (I think these are all the TM EIDE drives Quantum has sold, but the list may not be complete, and some of them may have not been sold at all):

Common Firmware Revisions for these drives are A6B.1F00 and A6B.1T00 when determined with programs that display the results of an identify-drive command (see my tools page for some examples of such programs); actually the full revision will have the two zeroes replaced by alphanumeric characters and can be determined only by running the patch-program (as far as I know). Quantum states that Fireball TM series SCSI drives are not affected, nor are the older Fireball's (without an extension like TM or ST), Sirocco, Bigfoot and planned Fireball ST series drives.

Back to Top


HOW TO PREVENT DATA LOSS WITHOUT APPLYING THE PATCH


Back to Top


ERROR CONDITIONS KNOWN BY ME, AS OF APRIL 16, 1997

I was able to create the error conditions by installing only DOS, WfW and the ONTRACKW.386 driver on a clean drive, but also on more useful configurations with typical applications (text-processing, internet support etc.) installed. Like me, you will ONLY be able to reproduce the problems with these drives when all of the following conditions are met:

Back to Top


LIKELY CAUSE OF THE PROBLEM

When you quit WfW311, the 32BDA driver is disabled. In addition, the 32BFA system is reset. This causes you to always end up in a rootdirectory after quitting WfW with 32BFA enabled, regardless of the directory you were in when you started windows. Then DOS will be forced to read the rootdirectory in order to reload COMMAND.COM. Under the error conditions above, it only gets non-zero values in the first (out of typically 32) rootdirectory sector! And a rootdir entry that starts with a zero-byte is considered unused by DOS. I also tested what happens when all standard 512 rootdirectory entries are filled (I even tested a C: partition with 1024 entries, filled with about 600 entries). In all cases, only entries out of the first rootdirectory sector showed up.
Note that it is perfectly possible that the Fireball TM EIDE drives are not causing the problem. If the problems are caused by blockmode changes (which I'm pretty sure about), then it may be hard to blame anyone, since behaviour after these switches may be badly documented in the ATA-specs (let me know if I'm wrong). You can probably blame Microsoft, the 32BDA driver manufacturer of the BIOS manufacturer as well. Anyway, I want to be able use these drives for running DOS and WfW without data loss.

Back to Top


CONCLUSION

The four Quantum Fireball TM2100A (and TM2110A) drives with firmware rev. A6B.1F00 I obtained on October 1, 1996, were unusable for running WfW311. My local supplier has replaced them by Western Digital WDAC22100H drives, on which I was not able to reproduce the problem (as on any other drive tested, mostly WD). At first, Quantum was not able to reproduce the problem I described in the first report I sent them, and after some time, they thought that they had fixed the bug, and sent me a drive for testing. It has firmware rev. A6B.2400, and unfortunately still has the same bug in it. I did a lot more tests with this drive, and wrote some more reports which descibe more parameters that affect this problem, and have also sent them to Quantum.
Meanwhile, Quantum has reproduced the bug and created a fix, and released it to the public. All sizes Quantum Fireball TM EIDE drives suffer from the problem described above (not the SCSI drives!). In another page I have described how to obtain and apply the patch, which will replace (patch) the firmware in your drive by a new version.


ACKNOWLEDGEMENTS

First of all, I would like to thank Rob Karlas, manager of the CEM (central electronics store for the Tech. Univ. Delft) for allowing me to replace the buggy drives I had by new WD drives, without doubting my findings.
Secondly, Phil Worsdale, James Doherty (both from Quantum UK) and Ralf-Peter Rohbeck (Quantum Germany) have been very helfpul and responsive to my emails. Unfortunately for them, initially there was not a lot they could do for me, since I had already replaced the drives by WD's and therefore couldn't do any of the tests they suggested. Phil was so kind to supply me with a drive for testing purposes, and forwarded my reports to the Quantum tech. people in Milpitas, USA.
I also reported the problem in an early stage to Ontrack, which was answered by Bill Foss. He mailed me a few suggestions, but at that time I no longer had the drives. Further he infomed me that they had also done some testing, but were not able to reproduce the problem. However, they did not have a FireBall TM2100A drive available, and conducted the tests on another Quantum drive (I don't know which one). I hope that Ontrack accepts my apologies for my conclusion in my first report, that suggested that possibly ONTRACKW.386 causes the problems, which is not true! (since the problem can also be reproduced with other 32BDA drivers, including the WFW311 builtin device=*wdctrl ).
Further I'd like to thank the collegues who kept waiting patiently for their PC's, and my room-mate Michiel van der Stam and email-mate Pybe Faber for listening ears and a lot of suggestions, and my wife Paula and my son Jeroen for not seeing much of me for some time.
I'd also like to thank all people who responded in newsgroups or via email, in particular Chris Unsicker, who reported identitcal problems with a number of new Gateway2000 PC's in some Usenet newsgroups, and Marnix Timmermans, for in-depth discussions on this subject. To all people not mentioned by name, espceially those who complemented me for this page (and the ones referenced here), thank you!
And finally I would like to thank Anthony Olszewski for publishing this on his webserver, and for maintaining this interesting site, www.computercraft.com.
Thanks again!

Back to Top


DISCLAIMER

As far as I know, only Quantum Fireball TM-series EIDE drives are involved, not the older Fireball's (the older Fireball's are very different drives; for example, they spin at 5400RPM while the TM drives do 4500RPM), and I do not expect this problem in the Fireball ST drives Quantum plans to release soon.
The patched firmware described, Rev. A6B.2DNR, works fine for me. As of April 15, 1997, I have used the TM2110A drive a lot for various experiments (like finding out the difference between Ontrack Diskmanager and Microhouse EzDrive) and I have never had any problems related with this drive and firmware. However, if it causes any problems for you, don't blame me!
I publised this information only to help people to prevent data loss, to get an idea how big the problem really is and to keep some pressure on Quantum; I am not publishing this to tell that you shouldn't be buying Quantum drives! There simply are no perfect harddrives, all brands have their problems now and then, and although it took Quantum some time to acknowledge this problem, finally they did come up with a solution, and made it available via the Internet. Should someone suggest that Quantum did not test these drives thoroughly enough before putting them to market, please consider that the problem described here is very complicated, and if you don't know where to look, may be hard to reproduce, if at all. Further, I still have no proof that the Quantum Fireball TM-series drives are actually causing the problem. As far as I know this the problem is caused by the access-mode switch that occurs when Windows 32bit disk access is terminated and you return to DOS, and possibly there is no reset-drive command sent to the drive. The drive might comply with the ATA specifications, but unfortunately the defacto standard is defined by Microsoft. This makes it hard to say who is to blame. However, I have not heard of similar problems reported on any other drive.
Further I'd like to state that there's no profit for me here; I've been using overtime hours to do the research (and writing), just because I like to solve problems and because I have a lot more hardware (and software) available for testing than typical users do. I am not related to Quantum in any way, apart from the fact that they've sent me a TM2110A drive for testing purposes (they've allowed me to keep it) and sent me a number of patches; further I have had an extensive and very pleasant email communication with some of Quantum's tech. support people (see the acknowledgements above).
I am not related to www.computercraft.com, other than that Anthony Olszewski picked a post of me from a usenet newsgroup, and then I offered to write updated and improved pages about this subject, which you're reading now. See this as my contribution to the Internet society. I'd like to dedicate these pages to all freeware programmers!

Back to Top


HOW TO CONTACT ME

NOTE: because I consider the problem solved since a patch is available, I will no longer be collecting user experiences related to this matter, and will only I answer related or unrelated emails when I have time to do so.
If you have any problems with obtaining and/or applying the patch, please contact Quantum. You can find the nearest support office on Quantum's technical support page or see my page "How to obtain and apply the patch" for more information.
Also, please, do NOT email me telling me that I should not be using WfW any longer. In fact, I also use NT4. But due to the fact that a lot of money (new software/hardware) and time (training users, installing new hard/software) is involved in switching to new operating systems for a department, most PC's where I work will continue to run WfW311 for some time (and believe me, a lot of companies have this same policy).
If you have general questions about harddrives, diskmanagers, formatting, possible bugs and fixes etc., please ask those questions in one of the Usenet Newsgroups; for PC related storage devices this would be comp.sys.ibm.pc.hardware.storage. If you don't have a newsreader, most www browsers will accept the following URL: news://your.news.server/comp.sys.ibm.pc.hardware.storage where you must of course replace your.news.server by your favourite host. Before posting messages (asking questions) in any newsgroup, please read the newsgroup's FAQ to make sure that you are not asking a Frequently Asked Question; see my tools page for some pointers to FAQ's. When I have time, I will be reading some of the postings in the PC storage group, but lots of other, often much more experienced people do as well, and they often come up with answers I am not aware of. This includes people (guru's actually) from companies like Quantum, Ontrack, Tapedisk corp, Datarescue and Lindsay Computer Systems (to name a few that come to my mind). So check it out!
Actually, the only reason why you could send me email is because you have important remarks about this page, for example corrections and/or additional information. I have a dedicated email account where you can reach me: qs@cpo.tn.tudelft.nl (qs still for Quantum Survey). I may or may not update the page. Please do not send me advertisements of other junk, and don't expect an immediate reply!

Back to Top


Copyright (C) 1997 by Erik van Straten. I grant Computercraft the non-exclusive right to publish this page. All (registered) trademarks are recognized.

Erik van Straten
System manager/programmer
Charged Particle Optics Research Group
Technical University Delft, Applied Physics dept.
Lorentzweg 1
2628CJ Delft
The Netherlands

Back to Top

Contact
COMPUTERCRAFT

Return To The COMPUTERCRAFT Main Menu

Asbury Park
Bruce Springsteen's Jersey Shore Rock Haven!

Featured Link
Great Domain Names For Sale!
Very brandable Domains for Sale -- The GET NJ family of Sites, managed by Anthony Olszewski, features tens of thousands of Pages Online at dozens of active domains, many with a New Jersey focus. Other advertising opportunities including enterprise and exclusive placements exist at a wide range of Web Sites. Your ad can appear at one Page or at many, many thousands of Pages simultaneously! A large slice of the domains have been Online for more than five years, some for over ten! In addition to advertising, many great Domains are available for purchase or license

Text Link Advertising Program
Business name, Web Site Link and a brief description or motto – runs for one month in the Page (or Pages) of your choice.

Hudson County Politics
From Frank Hague to Robert Janiszewski, in this New Jersey county, political corruption is a tradition. Former NJ Governor Brendan Byrne wants to be buried here so he can stay active in Democratic politics! You'll find lots about Senator Robert Menendez, too.

GRAVE ROBBER Jersey City Computer Repair
297 Griffith Street, Jersey City, NJ - 201-798-2292 - In the Heights just off of Kennedy Blvd. - Very close to Journal Square and Union City, just five minutes away from Hoboken, Downtown Jersey City, Newport, the Waterfront, Secaucus, North Bergen and Weehawken - Tech support for The Jersey City Mayor's Office during the administration of Bret Schundler - PC repair - Tivos, too!, upgrade, hardware install, software install, data recovery, spyware removal, virus removal, replace hard drive, replace motherboard, data recovered from notebook computers, recover lost XP passwords, password recovery

The Statue of Liberty, Ellis Island, and The Central Railroad Terminal
Visit Liberty State Park!

Stealth Jet at the World Trade Center During OpSail 2000
For All Sorts of Unique New York City Information, Visit GET NY!