Mailing List Archive

From: M.Hunold@t-online.de (Michael Hunold)
I have encountered some really strange oddities with
my two-year-old FUJITSU MPB3032ATU 3,2Gb ide-drive.
Until yesterday I did not bother about the drive
at all -- it simply worked. It was partitioned
for use with W98 (1 GB) and Linux (2GB) and worked
fine with both.
Then I bought a new drive and though formatting
my old drive and making it use ext2fs would be a good
idea... 8-)
First of all, the kernel-messages and an "hdparm"
output:
Kernel? Which version?
> hdc: FUJITSU MPB3032ATU, ATA DISK drive
> hdc: FUJITSU MPB3032ATU, 2014MB w/0kB Cache, CHS=4092/16/63, (U)DMA
As you can see, the drive does not get recognised correctly.
The CHS should be more like 6704/15/63 (as printed on top of the
drive and entered as BIOS/CMOS values) to get 3,2Gb
-- with 4092/16/63 it only gets 2Gb.
This disk has jumpers.
The default is 6704/15/63 but you can jumper it to 4092/16/63,
as you probably did. The jumper settings are:
(default master)
1..
122
(clipped master)
1.2
1.2
Due to this, I cannot use (Linux)"fdisk" to partition the drive
correctly. It refuses to accept my "false" values for the
partition edges.
Not so pessimistic. You can (i) change the jumpers, or (ii) give the
kernel explicit boot parameters, or (iii) give fdisk explicit geometry
parameters.
Now comes the tricky part: when I use (win98)"fdisk", I can
make one big "fat32"-3,2Gb-partition, the program seems to take the
BIOS values and not the values the drive is perhaps reporting.
After formatting, it seems that everything is fine.
When I now start Linux the drive still does not get
recognised correctly but I can use it to the full 3,2GB. I
tried to write 32 100Mb files to it and succeeded.
How do you start Linux? (From Windows? From a cold machine?)
What are the kernel boot messages? (dmesg | grep hdc)
Btw: when I try to read any of the 12 files that lie behind
the 2Gb barrier, Win98 fails badly, e.g. completly crashes
the machine.
Ok, I now have the following situation: a drive that was
partitioned and formatted with Win98-tools fails with
Win98 but can be completely used with Linux (fat32 formatted),
even if Linux fails to correctly recognize it.
Any ideas of how to solve the problem?
As mentioned above, you can easily avoid problems, in several ways.
In order to understand all details of what happens, we would need
some more information.
(And when you give hdparm output, give both hdparm -i and hdparm -I;
in some kernel versions the kernel itself fiddles with the hdparm -i
report, so that it no longer represents what the disk said.)
Andries
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

Hello,
> I have encountered some really strange oddities with
> my two-year-old FUJITSU MPB3032ATU 3,2Gb ide-drive.
>
> Kernel? Which version?
Sorry for the inconvenience. I´m using 2.2.10.
> This disk has jumpers.
> The default is 6704/15/63 but you can jumper it to 4092/16/63,
> as you probably did. The jumper settings are:
Nope. I jumpered it to 1.., just like the documentation proposes.
122
> How do you start Linux? (From Windows? From a cold machine?)
I use ´loadlin´ and added a menu-entry to "config.sys".
But even if I use a so-called "rescue-disk" from my
distribution kit (2.2.10, too) and start with a cold
machine, the results are the same.
> What are the kernel boot messages? (dmesg | grep hdc)
ide1: BM-DMA at 0xf008-0xf00f, BIOS settings: hdc:pio, hdd:pio
hdc: FUJITSU MPB3032ATU, ATA DISK drive
hdc: FUJITSU MPB3032ATU, 2014MB w/0kB Cache, CHS=4092/16/63, (U)DMA
hdc: hdc1
> (And when you give hdparm output, give both hdparm -i and hdparm -I;
> in some kernel versions the kernel itself fiddles with the hdparm -i
> report, so that it no longer represents what the disk said.)
michael:/home/michael # hdparm -i /dev/hdc
/dev/hdc:
Model=FUJITSU MPB3032ATU, FwRev=2009, SerialNo=01138683
Config={ HardSect NotMFM HdSw>15uSec Fixed DTR>10Mbs RotSpdTol>.5% }
RawCHS=4092/16/63, TrkSize=0, SectSize=0, ECCbytes=4
BuffType=0(?), BuffSize=0kB, MaxMultSect=16, MultSect=off
DblWordIO=no, maxPIO=2(fast), DMA=yes, maxDMA=0(slow)
CurCHS=4092/16/63, CurSects=4124736, LBA=yes, LBAsects=4124736
tDMA={min:120,rec:120}, DMA modes: mword0 mword1 *mword2
IORDY=yes, tPIO={min:240,w/IORDY:120}, PIO modes: mode3 mode4
Ok, now have a look at this. I don´t know why the "-I"
parameter screws things up like this:
michael:/home/michael # hdparm -I /dev/hdc
/dev/hdc:
Model=UFIJST UPM3B30A2UT , FwRev=0290 ,
SerialNo=
Config={ HardSect NotMFM HdSw>15uSec Fixed DTR>10Mbs RotSpdTol>.5% }
RawCHS=4092/16/63, TrkSize=0, SectSize=0, ECCbytes=4
BuffType=0(?), BuffSize=0kB, MaxMultSect=16, MultSect=?0?
DblWordIO=no, maxPIO=2(fast), DMA=yes, maxDMA=0(slow)
CurCHS=4364/15/63, CurSects=4123980, LBA=yes, LBAsects=4124736
tDMA={min:120,rec:120}, DMA modes: mword0 mword1 *mword2
IORDY=yes, tPIO={min:240,w/IORDY:120}, PIO modes: mode3 mode4
> Not so pessimistic. You can (i) change the jumpers, or (ii) give the
> kernel explicit boot parameters, or (iii) give fdisk explicit geometry
> parameters.
Ok, I tried the "extra functionality menu" of "fdisk"
with no success.After changing the C/H/S to 6704/15/63,
syncing (and even rebooting) still the same results.
In fact: when "fdisk" has finished changing and syncing and I start it
again, it still says "4092/16/32". Because of this, it is
impossible to use "mke2fs"...
So I suppose that it is some hd-failure (perhaps the drive does not
report is geometry correctly) and so both Linux and Windows
fail. Of course, Linux is smarter than Windows and can use the
drive anyway... (at least if it is formatted with fat32)
Btw: I am using
michael:/home/michael # fdisk -v
fdisk v2.9t
> Andries
CU,
Michael.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

From M.Hunold@t-online.de Wed Oct 27 20:07:26 1999
Sorry for the inconvenience. I=B4m using 2.2.10.
Good. (And also Quoted Unreadable, I see. Not so good.)
> This disk has jumpers.
> The default is 6704/15/63 but you can jumper it to 4092/16/63,
> as you probably did. The jumper settings are:
Nope. I jumpered it to 1.., just like the documentation proposes.
122
> What are the kernel boot messages? (dmesg | grep hdc)
ide1: BM-DMA at 0xf008-0xf00f, BIOS settings: hdc:pio, hdd:pio
hdc: FUJITSU MPB3032ATU, ATA DISK drive
hdc: FUJITSU MPB3032ATU, 2014MB w/0kB Cache, CHS=3D4092/16/63, (U)DMA
hdc: hdc1
> (And when you give hdparm output, give both hdparm -i and hdparm -I;
> in some kernel versions the kernel itself fiddles with the hdparm -i
> report, so that it no longer represents what the disk said.)
michael:/home/michael # hdparm -i /dev/hdc
/dev/hdc:
Model=3DFUJITSU MPB3032ATU, FwRev=3D2009, SerialNo=3D01138683
RawCHS=3D4092/16/63, TrkSize=3D0, SectSize=3D0, ECCbytes=3D4
CurCHS=3D4092/16/63, CurSects=3D4124736, LBA=3Dyes, LBAsects=3D4124736
Ok, now have a look at this. I don=B4t know why the "-I" =
parameter screws things up like this:
michael:/home/michael # hdparm -I /dev/hdc
/dev/hdc:
Model=3DUFIJST UPM3B30A2UT , FwRev=3D0290 ,
RawCHS=3D4092/16/63, TrkSize=3D0, SectSize=3D0, ECCbytes=3D4
CurCHS=3D4364/15/63, CurSects=3D4123980, LBA=3Dyes, LBAsects=3D4124736
The same in both cases, except that the model is given in BigEndian order
and -i corrects for that but -I doesnt.
And by now it is completely clear what happens on the Linux side.
The drive reports LBAsects=4124736, so it says that its total capacity
is 2.11 GB, a standard clipped capacity. Most disks only clip the CHS info
and let the operating system know the truth via LBAcapacity, but this disk
cheats on all fronts so that Linux has no way of knowing total capacity.
So, you'll have to boot with boot parameter hdc=6704,15,63.
Now this is precisely the behaviour one would expect with jumpers
set to `clip'. I am surprised that this disk still does this with
jumpers in default position.
So, I understand the Linux side but not the disk side of things.
Andries
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

Hi,
The disk report 0kB Cache.
That is probably another reason why DMA can't be enabled since
there is no memory to DMA to/from - it can't DMA directly to disk.
/RogerL
> > What are the kernel boot messages? (dmesg | grep hdc)
>
> ide1: BM-DMA at 0xf008-0xf00f, BIOS settings: hdc:pio, hdd:pio
> hdc: FUJITSU MPB3032ATU, ATA DISK drive
> hdc: FUJITSU MPB3032ATU, 2014MB w/0kB Cache, CHS=4092/16/63, (U)DMA
> hdc: hdc1
>
--
The Internet interprets Windows as damage,
and routes around it.
Roger Larsson
Skellefteå
Sweden
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

> The disk report 0kB Cache.
>
> That is probably another reason why DMA can't be enabled since
> there is no memory to DMA to/from - it can't DMA directly to disk.
DMA isnt affected by cache. Also drives reporting 0K cache dont neccessarily
have no cache
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

On Wed, 27 Oct 1999, Roger Larsson wrote:
> Hi,
>
Hi!
>
> The disk report 0kB Cache.
Main too.
>
> That is probably another reason why DMA can't be enabled since
> there is no memory to DMA to/from - it can't DMA directly to disk.
>
But for me DMA works.
It's a fujitsu, ut I'm at work now and I haven't the model type.
Why it shows me 0kb cache?
> /RogerL
>
>
> > > What are the kernel boot messages? (dmesg | grep hdc)
> >
> > ide1: BM-DMA at 0xf008-0xf00f, BIOS settings: hdc:pio, hdd:pio
> > hdc: FUJITSU MPB3032ATU, ATA DISK drive
> > hdc: FUJITSU MPB3032ATU, 2014MB w/0kB Cache, CHS=4092/16/63, (U)DMA
> > hdc: hdc1
> >
>
> --
>
> The Internet interprets Windows as damage,
> and routes around it.
>
> Roger Larsson
> Skellefteå
> Sweden
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.rutgers.edu
> Please read the FAQ at http://www.tux.org/lkml/
>
---
Catalin(ux) BOIE
catab@deuroconsult.ro
http://www2.deuroconsult.ro/~catab
Fax: (561) 760-6212
ICQ: 45443500
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

Hello,
> michael:/home/michael # hdparm -I /dev/hdc
>
> /dev/hdc:
>
> Model=3DUFIJST UPM3B30A2UT , FwRev=3D0290 ,
> RawCHS=3D4092/16/63, TrkSize=3D0, SectSize=3D0, ECCbytes=3D4
> CurCHS=3D4364/15/63, CurSects=3D4123980, LBA=3Dyes, LBAsects=3D4124736
> The drive reports LBAsects=4124736, so it says that its total capacity
> is 2.11 GB, a standard clipped capacity. Most disks only clip the CHS info
> and let the operating system know the truth via LBAcapacity, but this disk
> cheats on all fronts so that Linux has no way of knowing total capacity.
>
> So, you'll have to boot with boot parameter hdc=6704,15,63.
Thanks for the hint with the boot parameter. Now I have finally
an ext2-fs with 3.2Gb on it...
Although it does not matter, I get still these boot-messages:
ide1: BM-DMA at 0xf008-0xf00f, BIOS settings: hdc:pio, hdd:pio
hdc: FUJITSU MPB3032ATU, ATA DISK drive
hdc: FUJITSU MPB3032ATU, 2014MB w/0kB Cache, CHS=6704/15/63, (U)DMA
Shouldn´t the ide-driver calculate the size of an hdd when its
geometry-values gets overriden by a boot paramter rather than
take the "false" values reported by the drive?
> Now this is precisely the behaviour one would expect with jumpers
> set to `clip'. I am surprised that this disk still does this with
> jumpers in default position.
> So, I understand the Linux side but not the disk side of things.
Ok, thanks for you help. I have already contacted the german
division of Fujitsu -- perhaps it is a really an hardware issue.
But even if, I don´t think they will admit it for a 2 year-old
drive...
> Andries
CU,
Michael.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

Hello,
> And by now it is completely clear what happens on the Linux side.
>
> The drive reports LBAsects=4124736, so it says that its total capacity
> is 2.11 GB, a standard clipped capacity. Most disks only clip the CHS info
> and let the operating system know the truth via LBAcapacity, but this disk
> cheats on all fronts so that Linux has no way of knowing total capacity.
> Now this is precisely the behaviour one would expect with jumpers
> set to `clip'. I am surprised that this disk still does this with
> jumpers in default position.
> So, I understand the Linux side but not the disk side of things.
The german support of Fujitsu sent me a diagnostic tool and this
states that my drive has an hardware-failure or at least does not
behave like it should.
They said that I can send the drive to their support and they will
fix it with no extra charge because it still has warranty.
After all we can see that Linux can even handle "evil" drives...
8-)
> Andries
CU,
Michael.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

Hello,
> And by now it is completely clear what happens on the Linux side.
>
> The drive reports LBAsects=4124736, so it says that its total capacity
> is 2.11 GB, a standard clipped capacity. Most disks only clip the CHS info
> and let the operating system know the truth via LBAcapacity, but this disk
> cheats on all fronts so that Linux has no way of knowing total capacity.
> Now this is precisely the behaviour one would expect with jumpers
> set to `clip'. I am surprised that this disk still does this with
> jumpers in default position.
> So, I understand the Linux side but not the disk side of things.
The german support of Fujitsu sent me a diagnostic tool and this
states that my drive has an hardware-failure or at least does not
behave like it should.
They said that I can send the drive to their support and they will
fix it with no extra charge because it still has warranty.
After all we can see that Linux can even handle "evil" drives...
8-)
> Andries
CU,
Michael.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/