Bad Blocks mapping and Low Level Format on HDs >4GB

Martin Bertelsmann

unread,

Mar 12, 1998, 3:00:00 AM3/12/98

to

Hi,

does anybody know how to successfully low-level format a drive bigger
than 4GB and successfully map out bad blocks (by hand as the drive
doesnt do this automatically)? It seems like HDtoolBox etc. dont work
there...

Ciao

Martin

Etienne Vogt

unread,

Mar 12, 1998, 3:00:00 AM3/12/98

to

In article <3507E1...@student.uni-ulm.de> Martin Bertelsmann <Goetz-Martin...@student.uni-ulm.de> writes:
> does anybody know how to successfully low-level format a drive bigger
> than 4GB and successfully map out bad blocks (by hand as the drive
> doesnt do this automatically)? It seems like HDtoolBox etc. dont work
> there...

Yes, just don't do it.

I'm serious. Modern drives don't need to be low level formatted. That's
done once and for all at the factory. If you try it, it will do nothing
at best, or totally kill your drive at worst.
Drives that need low level format are very old ESDI or ST-506 drives. I
have never seen such a drive with a >4Gb capacity. ;-)

If the drive has developped bad blocks, then you should return it to the
manufacturer if it is still under warranty (most drives have a 5 year
warranty now and drives >4Gb are likely not that old).

--
Etienne Vogt (Etienne.Vogt @ obspm.fr)
PhD in astrophysics, Meudon Observatory, France.
Amiga programmer.

Georges Heinesch

unread,

Apr 26, 1998, 3:00:00 AM4/26/98

to

In a message of 12-Mar-98 14:58:39, Etienne Vogt wrote:

>> does anybody know how to successfully low-level format a drive
>> bigger than 4GB and successfully map out bad blocks (by hand as the
>> drive doesnt do this automatically)? It seems like HDtoolBox etc.
>> dont work there...

> Yes, just don't do it.

> I'm serious. Modern drives don't need to be low level formatted.
> That's done once and for all at the factory. If you try it, it will
> do nothing at best, or totally kill your drive at worst. Drives that
> need low level format are very old ESDI or ST-506 drives. I have
> never seen such a drive with a >4Gb capacity. ;-)

> If the drive has developped bad blocks, then you should return it to
> the manufacturer if it is still under warranty (most drives have a 5
> year warranty now and drives >4Gb are likely not that old).

Not quite correct. For example the Quantum Atlas XP34550 4.5 GB needs
to be low-level-formatted if the drive is written above tze 4 GB limit
with the original (not V43.x) scsi.device and/or FFS since the RDB is
overwritten. Without a low-level-format, tzhe drive cannot be
restored.

--
Cu Georges Heinesch, Luxembourg
geo...@ibm.net - geo...@geocities.com
http://www.geocities.com/yosemite/2480
PGP 2.6.3i / 5.0i public key on request and on public servers

... her name was Lola, ...

Gene Heskett

unread,

Apr 26, 1998, 3:00:00 AM4/26/98

to

Gene Heskett sends Greetings to Georges Heinesch;

>In a message of 12-Mar-98 14:58:39, Etienne Vogt wrote:

>>> does anybody know how to successfully low-level format a drive
>>> bigger than 4GB and successfully map out bad blocks (by hand as the
>>> drive doesnt do this automatically)? It seems like HDtoolBox etc.
>>> dont work there...

>> Yes, just don't do it.

>> I'm serious. Modern drives don't need to be low level formatted.
>> That's done once and for all at the factory. If you try it, it will
>> do nothing at best, or totally kill your drive at worst. Drives that
>> need low level format are very old ESDI or ST-506 drives. I have
>> never seen such a drive with a >4Gb capacity. ;-)

>> If the drive has developped bad blocks, then you should return it to
>> the manufacturer if it is still under warranty (most drives have a 5
>> year warranty now and drives >4Gb are likely not that old).

>Not quite correct. For example the Quantum Atlas XP34550 4.5 GB needs
>to be low-level-formatted if the drive is written above tze 4 GB limit
>with the original (not V43.x) scsi.device and/or FFS since the RDB is
>overwritten. Without a low-level-format, tzhe drive cannot be
>restored.

Not quite correct, Georges. If, when the drive was originally prepared,
a 'mountlist' or whatever the partitioning tool calls it, was saved off
to a floppy, along with a copy of that tool, then the drive can be
restored simply by reinstalling *that* mountlist with *that tool*.

But, then you go on safari to figure out which of the software chain is
broken. If it broke when doing a write that went over that 4 gig size I
quoted earlier, then the filesystem needs updateing. If it broke while
writing a file that straddled any power of 2 below that maximum partition
size in the earlier message, then the 'whatever.device' in the cards own
eproms is probably breaking. The AlfaData's I've had were broken at the
1 gig mark even with the latest roms, but the oktopussy replacement
fixes that I think. The rev 7 A2091 roms die at the 2 gig mark, the
gururom replacement goes forever I think. On the IDE front, the buddha
goes forever, but will apparently have problems with the replacement
filesystems, but on that, my hand isn't anywhere near a bible. Yet...
But two of them have bitten me now, refusing to recognize a replacement
filesystem unless it was installed with the newest HDInstTool.

Cheers, Gene
--
Gene Heskett, CET, UHK |Amiga A2k Zeus040 50 megs fast/2 megs chip
Ch. Eng. @ WDTV-5 |A2091,GuruRom,1g Seagate,CDROM,Multiface III
<gene_h...@iolinc.net> or |Buddha + 4 gig WDC drive, 525 meg tape
<gene_h...@westvirginia.net>|Stylus Pro, EnPrint, Picasso-II, 17" vga
--

Joanne Dow

unread,

Apr 27, 1998, 3:00:00 AM4/27/98

to

Gene Heskett wrote:

Actually, Gene, he is correct. I have identified an amazing bug in the
generic Quantum SCSI code that seems to appear on ALL their current
models. If you write to block zero for more than a single block the drive
becomes effectively catatonic and MUST be reformatted at a low level.
I found this attempting to duplicate some 2gig drives that are part of some
"AudioBox" (www.show-control.com) hardware we are addressing for
something we are developing. You can write multiple blocks anywhere
else on the drive but not block zero. I consider this to be a *MAJOR*
defect in the Quantum drives. And it forces me to reverse my stance
on low level format utilities - with *GREAT* reluctance. (I have been
known to use "impolite" and "unladilike" language before. I think I may
have melted part of that poor disk drive the way I addressed it. It had
a "wilted look" when I got done. And it still persisted in its bad behavior.)

(It DOES leave me tempted to declare Quantum drives to be pieces of
uttermost trash, though. If anyone at Quantum cares to email me with a
valid reason for this stupid behavior I will listen. But I will take a LOT of
convincing considering the "luck" we have had with Quantum things
around Wizardess Castle.)

{^_^} Joanne Dow, jd...@bix.com
(Who discovered this one and confirmed it two weeks ago.)

pro...@frii.com

unread,

Apr 27, 1998, 3:00:00 AM4/27/98

to

X-Newsreader: AWebNews

--
On Mon, 27 Apr 1998 09:53:35 +0000, Joanne Dow <jd...@bix.com> wrote ...

..... I'm tempted to declare Quantum drives pieces of trash........

So what drives do you reccommend? I went to Quantums after I had a couple of
Seagate drives die prematurely.

Thanks,

___Ron

Georges Heinesch

unread,

Apr 27, 1998, 3:00:00 AM4/27/98

to

Can you tell me a good tool to use for saving and restoring the RDB?

> But, then you go on safari to figure out which of the software chain
> is broken. If it broke when doing a write that went over that 4 gig
> size I quoted earlier, then the filesystem needs updateing.

Are you sure the FFS needs an update? ;)

I just tested the scsi.device 43.x with the FFS 40.1, and it *seems*
that this worked perfectly. I could write abov ethe 4 GB limit without
overwriting the first partition. To me it seems that the scsi.device
is the deciding factor when it comes to the 4 GB limit.

> If it
> broke while writing a file that straddled any power of 2 below that
> maximum partition size in the earlier message, then the
> 'whatever.device' in the cards own eproms is probably breaking.

While talking about max. partition sizes, what is the max. size of a
partition. Is semms to depend on the FFS and the bytes/block size.
Heinz Wrobel says in the README of his FFS 43.19, that the limit is
2,25 GB. However many people say itäs 2 GB. What is it now? When can I
expect an error while using a partition bigger than 2 GB?

> But two of them have bitten me now, refusing to
> recognize a replacement filesystem unless it was installed with the
> newest HDInstTool.

Do you mean this one here?

hdinst.lha 86.680
http://ftp.wustl.edu/pub/aminet/disk/misc/hdinst.lha

TIA

--
Cu Georges Heinesch, Luxembourg
geo...@ibm.net - geo...@geocities.com
http://www.geocities.com/yosemite/2480
PGP 2.6.3i / 5.0i public key on request and on public servers

... Hakuna Matata ...

Greg Tallent

unread,

Apr 27, 1998, 3:00:00 AM4/27/98

to

That's hard to say. I went to Seagate drives after my 2 Quantum Fireball
1.08Gig drives failed within a month of each other and they were just over a
year old.

>Thanks,

>___Ron

<TSB>
Greg Tallent |Amiga2000 GVP 040/33mhz/3.1 26 megs,2 gig/OpalVision |
gwt @ gte.net |Supra 28.8k, Lightwave 3.5, SCSI-Zip, Picasso II 1Meg|

All bribes cheerfully accepted.

Gene Heskett

unread,

Apr 27, 1998, 3:00:00 AM4/27/98

to

Gene Heskett sends Greetings to Georges Heinesch;

>In a message of 27-Apr-98 02:30:47, Gene Heskett wrote:

aminet/disk/misc/hdinst.lha

Also, Joanne Dow, part of the rdprep team, should be releasing a new
rdprepx shortly. It has been tweeked abit since the version in the
rdp391.lha archive, and should replace those parts of rdp391.lha as
required. Rdprepx itself is/was at revision 3.101 as of my testing 2
days ago now.

>> But, then you go on safari to figure out which of the software chain
>> is broken. If it broke when doing a write that went over that 4 gig
>> size I quoted earlier, then the filesystem needs updateing.

>Are you sure the FFS needs an update? ;)

>I just tested the scsi.device 43.x with the FFS 40.1, and it *seems*

>that this worked perfectly. I could write above the 4 GB limit without

>overwriting the first partition. To me it seems that the scsi.device
>is the deciding factor when it comes to the 4 GB limit.

I think its supposed to be the final limit, but Heinz and Ralph would
need to confirm that I think. Thats so I don't come across as an
'expert' since they are supposed to be more than 50 miles/80 klicks from
home, _and_ carrying a briefcase. I'm home, and the briefcase is in the
closet. :)

>> If it
>> broke while writing a file that straddled any power of 2 below that
>> maximum partition size in the earlier message, then the
>> 'whatever.device' in the cards own eproms is probably breaking.

>While talking about max. partition sizes, what is the max. size of a
>partition. Is semms to depend on the FFS and the bytes/block size.
>Heinz Wrobel says in the README of his FFS 43.19, that the limit is
>2,25 GB. However many people say itäs 2 GB. What is it now? When can I
>expect an error while using a partition bigger than 2 GB?

The maximum partition size is that which 0x7FFFFFFF translates to, which
is the figures I had forwarded earlier. I have no idea where the 2.25
gig figure came from unless it was calculated using an early pentium. :)

Most of us do use the 2 gig phrase as the limitation, but that is in
fact taking a rather severe liberty in rounding the numbers down to be
on the 'safe' side. I'm sorry for the confusion that may have created.

>Do you mean this one here?

>hdinst.lha 86.680
>http://ftp.wustl.edu/pub/aminet/disk/misc/hdinst.lha

Yes.

Joanne Dow

unread,

Apr 28, 1998, 3:00:00 AM4/28/98

to

Georges Heinesch wrote:

> >>Not quite correct. For example the Quantum Atlas XP34550 4.5 GB
> >>needs to be low-level-formatted if the drive is written above tze 4
> >>GB limit with the original (not V43.x) scsi.device and/or FFS since
> >>the RDB is overwritten. Without a low-level-format, tzhe drive
> >>cannot be restored.
>
> > Not quite correct, Georges. If, when the drive was originally
> > prepared, a 'mountlist' or whatever the partitioning tool calls it,
> > was saved off to a floppy, along with a copy of that tool, then the
> > drive can be restored simply by reinstalling *that* mountlist with
> > *that tool*.

(aside - Georges is correct. Multiple block writes to block zero on
aQuantum drive will require a reformat to make the drive useable again!)

> Can you tell me a good tool to use for saving and restoring the RDB?

<http://home.earthlink.net/~jdow/RDTool.html> includes a pointer to the
very latest RDPrepX. For unit 4 on an A3000 SCSI bus the command
"rdprepx -dscsi.device -q3 -t -u4 -sdf0:004.list" will save the RDBs and
"rdprepx -dscsi.device -q3 -t -u4 -r004.list" (from df0:) will restore
the
RDBs. This was one of the very first addon features I put into the
original
Microbotics RDPrep tool done my Kodiak. It is still backwards compatible
with Kodiak's code; but, I have rewritten, incrementally, everyting in
RDPrepX
so I feel save calling it my own code now. {^_-}

> > But, then you go on safari to figure out which of the software chain
> > is broken. If it broke when doing a write that went over that 4 gig
> > size I quoted earlier, then the filesystem needs updateing.
>
> Are you sure the FFS needs an update? ;)

Yes - 40.1 (or 40.4 which some developers have) is only good out to
4Gigs.
43.19 is the projected update. It is not suitable for use as an update
yet as it
still, Heinz's protestations not withstanding, does have the timeout in
it. It is
just set a long ways off. (Sooner than 2008, though....)

> I just tested the scsi.device 43.x with the FFS 40.1, and it *seems*

> that this worked perfectly. I could write abov ethe 4 GB limit without

> overwriting the first partition. To me it seems that the scsi.device
> is the deciding factor when it comes to the 4 GB limit.

Yup - that is the upgrade proposed.

> > If it
> > broke while writing a file that straddled any power of 2 below that
> > maximum partition size in the earlier message, then the
> > 'whatever.device' in the cards own eproms is probably breaking.

Nope - 'tis not the way the internals of scsi.device or anything else
work, quite. In fact if a scsi device actually used signed longs and
tried
to address block -N the scsi chain would report an error and no access
would occur. (It'd be a block just a whole long way out there.... {^_-})

> While talking about max. partition sizes, what is the max. size of a
> partition. Is semms to depend on the FFS and the bytes/block size.
> Heinz Wrobel says in the README of his FFS 43.19, that the limit is
> 2,25 GB. However many people say itäs 2 GB. What is it now? When can I
> expect an error while using a partition bigger than 2 GB?

Partitions are guaranteed safe to 2^31 bytes. Evidence indicates that
they
are safe to a full 2^32 bytes even on 40.1 ffs. Anyone who builds such a
partition has very carefully earned whatever problems he or she should
subsequently encounter. (Revalidating that partition, especially if
heavily
used, would be a harrowing experience to watch as the disk churned along
into the night. {^_-} Unless there is a compelling need I recommend many
250-500Meg partitions. Your stomache lining will last longer if you do
things that lead to revalidations.)

{^_^} Joanne Dow, jd...@bix.com

Warren Block

unread,

Apr 29, 1998, 3:00:00 AM4/29/98

to

pro...@frii.com wrote:
> On Mon, 27 Apr 1998 09:53:35 +0000, Joanne Dow <jd...@bix.com> wrote ...
> ..... I'm tempted to declare Quantum drives pieces of trash........
>
> So what drives do you reccommend? I went to Quantums after I had a couple of
> Seagate drives die prematurely.

Conner is gone, swallowed whole by Seagate. Western Digital drives used
to be really bad, maybe still are. Maxtor drives had a periodic
oscillation between good and bad, Micropolis collapsed and disappeared.
Fuji, IBM, DEC? I've seen problem reports with all.

Maybe this explains why prices are so low...

If you can, buy local at a place where you can return the drive if
necessary. That's not possible for many people. Around here, SCSI drives
are regarded as very esoteric, and the stores only have IDE.

--
Warren Block * Rapid City, South Dakota * USA
http://extra.newsguy.com/~wblock

Georges Heinesch

unread,

Apr 30, 1998, 3:00:00 AM4/30/98

to

In a message of 28-Apr-98 03:10:48, Gene Heskett wrote:

[...]

>>> If it
>>> broke while writing a file that straddled any power of 2 below
>>> that maximum partition size in the earlier message, then the
>>> 'whatever.device' in the cards own eproms is probably breaking.

>>While talking about max. partition sizes, what is the max. size of a

>>partition. Is semms to depend on the FFS and the bytes/block size.
>>Heinz Wrobel says in the README of his FFS 43.19, that the limit is
>>2,25 GB. However many people say itäs 2 GB. What is it now? When can
>>I expect an error while using a partition bigger than 2 GB?

> The maximum partition size is that which 0x7FFFFFFF translates to,

> which is the figures I had forwarded earlier. I have no idea where
> the 2.25 gig figure came from unless it was calculated using an
> early pentium. :)

Well, reading the docs again, I saw that the 2.25 GB limit is for the
*file* size (not the partition size) with FFS 40.x.

> Most of us do use the 2 gig phrase as the limitation, but that is in
> fact taking a rather severe liberty in rounding the numbers down to
> be on the 'safe' side. I'm sorry for the confusion that may have
> created.

No problem ;)

I just (quick) formatted the entire drive (4.5 GB as one big partition
and filled it with 173 25 MB files until it was full! All files were
intact. I used the scsi.device 43.23 and FFS 43.19. This leads me to
believe that not the 2 GB and even the 4 GBis a limit. It seems to be
higher than that.

Could it be that it depends on how many files you save on the disk?

--
Cu Georges Heinesch, Luxembourg
geo...@ibm.net - geo...@geocities.com
http://www.geocities.com/yosemite/2480
PGP 2.6.3i / 5.0i public key on request and on public servers

... Only happy at FL 370, Mach 0.74

Georges Heinesch

unread,

May 10, 1998, 3:00:00 AM5/10/98

to

In a message of 28-Apr-98 04:31:24, Joanne Dow wrote:

[...]

>> Can you tell me a good tool to use for saving and restoring the
>> RDB?

> <http://home.earthlink.net/~jdow/RDTool.html> includes a pointer to
> the very latest RDPrepX. For unit 4 on an A3000 SCSI bus the command
> "rdprepx -dscsi.device -q3 -t -u4 -sdf0:004.list" will save the RDBs
> and "rdprepx -dscsi.device -q3 -t -u4 -r004.list" (from df0:) will
> restore the RDBs. This was one of the very first addon features I
> put into the original Microbotics RDPrep tool done my Kodiak. It is
> still backwards compatible with Kodiak's code; but, I have
> rewritten, incrementally, everyting in RDPrepX so I feel save
> calling it my own code now. {^_-}

rdprepx locks my A3000 immediately! The SCSI LED remains on and the
SCSI is locked completely. I use the 43.23 scsi.device from Heinz
Wrobel.

[...]

--
Cu Georges Heinesch, Luxembourg
geo...@ibm.net - geo...@geocities.com
http://www.geocities.com/yosemite/2480
PGP 2.6.3i / 5.0i public key on request and on public servers

... like a virgin ...