4.1.1 sees IBM 20Gb IDE disk geometry wrong.

4.1.1 sees IBM 20Gb IDE disk geometry wrong.

Post by Olivie » Sun, 15 Oct 2000 12:14:20



Subject: 4.1.1 sees IBM 30Gb IDE disk geometry wrong.
Date: Fri, 13 Oct 2000 08:38:22 +0000 (UTC)

Organization: AIT - CSIM program
Newsgroups: comp.unix.bsd.misc

Dear fellow FreeBSD'ers,

I've been pulling my hair out trying to install 4.1.1 on an IBM PC 300
system with two  IDE  disks:  one  10Gb  (master)  and  one  20Gb  IBM
DLTA-305020 (slave).

Everything works OK for  the  base  installation  on  the  10Gb  disk.
Trouble begins when I want to partition the 20Gb disk: FreeBSD sees it
as a 2Gb disk with geometry = 4096/16/63. I tried to use 'g' to  force
disk geometry in the partition editor, but it would revert back to the
geometry  it  has  detected and not let me create any partition bigger
that 2Gb anyway.

The  host  PC  is a brand new IBM PC model 300 (model number 70T). Its
BIOS  setup  gives  no  control  over  things  like disk geometry, LBA
mode...

The disk label says:

Capacity = 20.5 Gb, 40.188.960 blocks, geometry = 16383/16/63

What I have tried so far:

1) use Windows 98 boot disk: OK, can FDISK, FORMAT 20Gb partition

2) boot Linux Slackware 7.0 disks:  OK,  can  fdisk  and  mke2fs  20Gb
partition

3) think and think and think and browse though  FreeBSD  archives:  it
seems  that  other people have seen similar problems, mostly with 3.x,
and that  the  miracle  cure  was  to  force  LBA  mode  using  "flags
0xb0ffb0ff"  on  the  wdc0 config line. Tried to find a similar option
for the ata driver: does not exist or is not documented anywhere, even
in the source code (at least not visible to me)

4) try to overwrite the first blocks of the disk with zeroes in case
some bad partitition table in MBR confuses FreeBSD: no way

5) boot 3.5.1 kern+mfsroot disks forcing the flags mentioned in (3)
above at boot time: OK! can see the 20Gb, partition, newfs...

6) in despair, built a 4.1.1 kernel with ata driver disabled and using
old wd driver with above flags: OK! Disk is now seen with  geometry  =
2501/255/63 (same as Linux sees it) with LBA mode.

Is this bug in the new ata driver ? the PC's BIOS ? both ? is there  a
workaround for this using the new ata driver or am I stuck  using  the
old wdc driver ?

Thanks in advance for any help,
_Alain_

cscoms.net>
--
Alain FAUCONNET
Sr. System Administrator
CS Internet Co. Ltd. (Shin Corp) - Thailand

 
 
 

4.1.1 sees IBM 20Gb IDE disk geometry wrong.

Post by Tim » Tue, 17 Oct 2000 04:00:00



writes
>Subject: 4.1.1 sees IBM 30Gb IDE disk geometry wrong.
>Date: Fri, 13 Oct 2000 08:38:22 +0000 (UTC)

>Organization: AIT - CSIM program
>Newsgroups: comp.unix.bsd.misc

>Dear fellow FreeBSD'ers,

>I've been pulling my hair out trying to install 4.1.1 on an IBM PC 300
>system with two  IDE  disks:  one  10Gb  (master)  and  one  20Gb  IBM
>DLTA-305020 (slave).

>Everything works OK for  the  base  installation  on  the  10Gb  disk.
>Trouble begins when I want to partition the 20Gb disk: FreeBSD sees it
>as a 2Gb disk with geometry = 4096/16/63. I tried to use 'g' to  force
>disk geometry in the partition editor, but it would revert back to the
>geometry  it  has  detected and not let me create any partition bigger
>that 2Gb anyway.

[snip]

I had something similar. Win95, WinNT and RH6 could see all of the disk.
It turns out there are 4 jumper pairs on the HDD unit. The usual 3 for
disk master/slave/detected and the fourth was to only use the first
4,096 cyls of the disk. FreeBSD 4 was looking at the jumper, but the
other systems did not.
--
Tim K75RT, T595, MAG 19038, TWO#21 & YGL#3

Due to the limitations of current email, the lip movements may be
out of synchronisation as you move your finger under the text while reading.

http://www.race24.com/          Live updates from 24 hour endurance races.
                                World Superbike and GP results every week.

 
 
 

4.1.1 sees IBM 20Gb IDE disk geometry wrong.

Post by Alain Fauconn » Fri, 20 Oct 2000 04:00:00




>writes
>>Subject: 4.1.1 sees IBM 30Gb IDE disk geometry wrong.
>>Date: Fri, 13 Oct 2000 08:38:22 +0000 (UTC)

>>Organization: AIT - CSIM program
>>Newsgroups: comp.unix.bsd.misc

>>Dear fellow FreeBSD'ers,

>>I've been pulling my hair out trying to install 4.1.1 on an IBM PC 300
>>system with two  IDE  disks:  one  10Gb  (master)  and  one  20Gb  IBM
>>DLTA-305020 (slave).

>>Everything works OK for  the  base  installation  on  the  10Gb  disk.
>>Trouble begins when I want to partition the 20Gb disk: FreeBSD sees it
>>as a 2Gb disk with geometry = 4096/16/63. I tried to use 'g' to  force
>>disk geometry in the partition editor, but it would revert back to the
>>geometry  it  has  detected and not let me create any partition bigger
>>that 2Gb anyway.
>[snip]
>I had something similar. Win95, WinNT and RH6 could see all of the disk.
>It turns out there are 4 jumper pairs on the HDD unit. The usual 3 for
>disk master/slave/detected and the fourth was to only use the first
>4,096 cyls of the disk. FreeBSD 4 was looking at the jumper, but the
>other systems did not.

You were perfectly right, Tim.

After carefully looking at the disk, I have found out that  it  has  a
jumper labeled "2Gb clip". I have removed and now FreeBSD happily uses
the full 20Gb.

Thank you very much!

Yes... kind of obvious,  I  must  admit,  but  if  Windows  and  (more
suprisingly) Linux had not disregarded that wrong  geometry,  I  would
certainly  have  looked at the disk itself instead of trying to find a
bug in FreeBSD's drivers :-)

As  for why FreeBSD does it... well, the kernel developers must have a
good reason for that, I guess !

Greets,
_Alain_
--

cscoms.net> two lines)

--
Alain FAUCONNET
Sr. System Administrator
CS Internet Co. Ltd. (Shin Corp) - Thailand

 
 
 

4.1.1 sees IBM 20Gb IDE disk geometry wrong.

Post by Drew Laws » Fri, 20 Oct 2000 04:00:00





>>I had something similar. Win95, WinNT and RH6 could see all of the disk.
>>It turns out there are 4 jumper pairs on the HDD unit. The usual 3 for
>>disk master/slave/detected and the fourth was to only use the first
>>4,096 cyls of the disk. FreeBSD 4 was looking at the jumper, but the
>>other systems did not.

>You were perfectly right, Tim.

>After carefully looking at the disk, I have found out that  it  has  a
>jumper labeled "2Gb clip". I have removed and now FreeBSD happily uses
>the full 20Gb.

>Thank you very much!

>Yes... kind of obvious,  I  must  admit,  but  if  Windows  and  (more
>suprisingly) Linux had not disregarded that wrong  geometry,  I  would
>certainly  have  looked at the disk itself instead of trying to find a
>bug in FreeBSD's drivers :-)

>As  for why FreeBSD does it... well, the kernel developers must have a
>good reason for that, I guess !

In this case, it sounds as if FreeBSD was doing it right, and the
others were doing it wrong.

Often there are two ways to go in an implimentation.  One is to
follow the perceived market preference.  The other is to follow
the spec.  Microsoft is legendary for not following specs.  I can't
imagine why Linux was ignoring the jumper.  Maybe the developer
wasn't aware that it existed.

--
|Drew Lawson           | Mrs. Tweedy!                       |

|http://www.furrfu.com |                                    |

 
 
 

4.1.1 sees IBM 20Gb IDE disk geometry wrong.

Post by Tim » Sat, 21 Oct 2000 04:00:00



n.cscoms.TO_BE_REMOVED.A_DETRUIRE.com> writes


>>writes
>>>Subject: 4.1.1 sees IBM 30Gb IDE disk geometry wrong.
>>>Date: Fri, 13 Oct 2000 08:38:22 +0000 (UTC)

>>>Organization: AIT - CSIM program
>>>Newsgroups: comp.unix.bsd.misc
[snip]

>You were perfectly right, Tim.

>After carefully looking at the disk, I have found out that  it  has  a
>jumper labeled "2Gb clip". I have removed and now FreeBSD happily uses
>the full 20Gb.

>Thank you very much!

[snip]

Excellent. Glad to be of help after the frustration I had.
--
Tim K75RT, T595, MAG 19038, TWO#21 & YGL#3

Due to the limitations of current email, the lip movements may be
out of synchronisation as you move your finger under the text while reading.

http://www.race24.com/          Live updates from 24 hour endurance races.
                                World Superbike and GP results every week.

 
 
 

4.1.1 sees IBM 20Gb IDE disk geometry wrong.

Post by Charlie Sors » Fri, 27 Oct 2000 04:00:00



[...]
= >It turns out there are 4 jumper pairs on the HDD unit. The usual 3 for
= >disk master/slave/detected and the fourth was to only use the first
= >4,096 cyls of the disk. FreeBSD 4 was looking at the jumper, but the
= >other systems did not.
[...]
= After carefully looking at the disk, I have found out that  it  has  a
= jumper labeled "2Gb clip". I have removed and now FreeBSD happily uses
= the full 20Gb.

Just curious:  What is the purpose of the jumper?  In other words,
why would anyone want to make a 20GB disk look like a 2GB disk?
What am I missing?

[...]
= Yes... kind of obvious,  I  must  admit,  but  if  Windows  and  (more
= suprisingly) Linux had not disregarded that wrong  geometry,  I  would
= certainly  have  looked at the disk itself instead of trying to find a
= bug in FreeBSD's drivers :-)
=
= As  for why FreeBSD does it... well, the kernel developers must have a
= good reason for that, I guess !
=
= Greets,
= _Alain_
= --

= cscoms.net> two lines)
=
= --
= Alain FAUCONNET
= Sr. System Administrator
= CS Internet Co. Ltd. (Shin Corp) - Thailand

--
NB:  While I welcome *reasonable* and *relevant* e-mail responses, any
e-mail with a *long* "To: list" or any mailing list may be treated as
spam and reported to the sender's ISP.

--
Best regards,

Charlie "Older than dirt" Sorsby     Edgewood,  NM     "I'm the NRA!"

 
 
 

1. 2.2.x kernel sees wrong IDE hard drive geometry

Hi!  I have a Quantum Fireball 6.4 GB hard drive with default geometry
of 3128,255,63 (I believe) which gives it a size of 1.6GB.  According to
the drive doc, if I change the BIOS settings to manually define the
geometry to 13328,15,63 the drive will then have a size of 6.4GB.  I
have manually defined the drive accordingly, but when my Linux kernel
boots, it still sees the default hard drive geometry.

I then tried passing kernel parameters in my /etc/lilo.conf, but it
doesn't seem to help.

Here is my current /etc/lilo.conf:


boot=/dev/hda
map=/boot/map
install=/boot/boot.b
# append="hdc=13328,15,63"
disk = /dev/hdc
   bios = 0x82
   sectors = 63
   heads = 15
   cylinders = 13328
prompt
timeout=50
image=/boot/vmlinuz-2.2.5-15
        label=linux
        root=/dev/hda6
        append="hdc=13328,15,63"
        read-only

What am I missing?  I'm running RedHat 6.0 w/ the default 2.2.5 kernel
and I tried building a 2.2.13 kernel (to a boot floppy) and that did the
same thing.

Any ideas would be greatly appreciated!

Peace....

Tom

Sent via Deja.com http://www.deja.com/
Before you buy.

2. Forcing a non-zero exit status.....

3. disk geometry of an IDE drive not/wrong detected

4. kppp-netscape

5. Solaris 7x86 only sees 8GB on 20GB IDE

6. xdm and RH5.1 problems

7. Disk Geometry on Second IDE Disk

8. Help: "Soft Link" an Oracle 8i instance?

9. 20GB IDE disk on Ultra 5 ??

10. LILO fails: Wrong disk geometry on SCSI-HD

11. IBM DFRSS2F: Can't get disk geometry for SCSI-HD

12. IBM ThinkPad 720 Disk Geometry Problem