Problems with Promise IDE controller under 2.4.1
Ole Aamot writes:
We experience trouble with the Promise (PDC20265) IDE controller and seven 75GB IBM disks on a single CPU (Pentium-III) server.
Linux 2.4.1 fails to detect correct geometry for the four last disks (identified as hde, hdf, hdg, hdh).
But there is no indication of what the problems could be, or what he thinks the geometry should be (and why). I see nothing very wrong in the posted data.
Andries
[Don't forget: (i) geometry does not exist (ii) hdparm is just some user program.]
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/
Quoted 11 lines · On Wed, 31 Jan 2001 [email protected] wrote:
Ole Aamot writes:
We experience trouble with the Promise (PDC20265) IDE controller and seven 75GB IBM disks on a single CPU (Pentium-III) server.
Linux 2.4.1 fails to detect correct geometry for the four last disks (identified as hde, hdf, hdg, hdh).
But there is no indication of what the problems could be, or what he thinks the geometry should be (and why). I see nothing very wrong in the posted data.
We agree Andries, but the enduser wants to see stuff the same.
Cheers,
Quoted 6 lines
Andries
[Don't forget: (i) geometry does not exist (ii) hdparm is just some user program.]
Andre Hedrick
Linux ATA Development
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/
Quoted 5 lines · Andre Hedrick
Quoted 3 lines
But there is no indication of what the problems could be, or what he thinks the geometry should be (and why). I see nothing very wrong in the posted data.
We agree Andries, but the enduser wants to see stuff the same.
In my case, I have two identical Maxtor drives, but they reported different geometry. How could that be? Move the "virgin" drive to the motherboard IDE controller and suddenly the geometry is the same. Use fdisk and partition the disk, write it, and then move to the promise controller and the "correct" geometry was used (that is, it is now the same as when hooked up to the motherboard ide controller).
Why was it important to me? I'm doing RAID1 and it is really nice to have the same geometry so that the partition info is the same between
the two drives. Makes life easier.
Signature · 5 lines
-rupa - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected] Please read the FAQ at http://www.tux.org/lkml/
Quoted 16 lines · Rupa Schomaker
Quoted 5 lines · Andre Hedrick
Quoted 3 lines
But there is no indication of what the problems could be, or what he thinks the geometry should be (and why). I see nothing very wrong in the posted data.
We agree Andries, but the enduser wants to see stuff the same.
In my case, I have two identical Maxtor drives, but they reported different geometry. How could that be? Move the "virgin" drive to the motherboard IDE controller and suddenly the geometry is the same. Use fdisk and partition the disk, write it, and then move to the promise controller and the "correct" geometry was used (that is, it is now the same as when hooked up to the motherboard ide controller).
Quoted 3 lines
Why was it important to me? I'm doing RAID1 and it is really nice to have the same geometry so that the partition info is the same between
the two drives. Makes life easier.
Please read the above and pass the geometry to the kernel. Mother boards have to do a translation to use the drive completely.
Cheers,
Andre Hedrick
Linux ATA Development
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/
Quoted 18 lines · Andre Hedrick
Quoted 16 lines · Rupa Schomaker
Quoted 5 lines · Andre Hedrick
Quoted 3 lines
But there is no indication of what the problems could be, or what he thinks the geometry should be (and why). I see nothing very wrong in the posted data.
We agree Andries, but the enduser wants to see stuff the same.
In my case, I have two identical Maxtor drives, but they reported different geometry. How could that be? Move the "virgin" drive to the motherboard IDE controller and suddenly the geometry is the same. Use fdisk and partition the disk, write it, and then move to the promise controller and the "correct" geometry was used (that is, it is now the same as when hooked up to the motherboard ide controller).
Quoted 6 lines
Quoted 3 lines
Why was it important to me? I'm doing RAID1 and it is really nice to have the same geometry so that the partition info is the same between
the two drives. Makes life easier.Please read the above and pass the geometry to the kernel. Mother boards have to do a translation to use the drive completely.
Andre,
But now it doesn't matter. The drive was tainted (fdisk run while attached to the mainboard controller) and now that geometry is "stuck". <shrug> I was mostly explaining why it is nice to get the same geometry on two identical drives (RAID1 is easier for the human to deal with).
Signature · 5 lines
-rupa - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected] Please read the FAQ at http://www.tux.org/lkml/
Quoted 5 lines · Rupa Schomaker
But now it doesn't matter. The drive was tainted (fdisk run while attached to the mainboard controller) and now that geometry is "stuck". <shrug> I was mostly explaining why it is nice to get the same geometry on two identical drives (RAID1 is easier for the human to deal with).
It doesn't but I got out of the geometry business two+ years ago... Andries (Mr. FDisk) Brouwer is the man......
Cheers,
Andre Hedrick
Linux ATA Development
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/
Simple solution is to have kernel fall-back to LBA style translations instead of kernel "basic" translations. This would make it match the first two "BIOS" drives on most systems, and not really hurt anything in most cases.
Even better would be to add a stage in front of the fall-back, which queries the BIOS (from kernel startup code) for translation info on ALL drives.
Signature · 3 lines
Mark Lord Real-Time Remedies Inc. [email protected]
Quoted 16 lines · Rupa Schomaker
Quoted 5 lines · Andre Hedrick
Quoted 3 lines
But there is no indication of what the problems could be, or what he thinks the geometry should be (and why). I see nothing very wrong in the posted data.
We agree Andries, but the enduser wants to see stuff the same.
In my case, I have two identical Maxtor drives, but they reported different geometry. How could that be? Move the "virgin" drive to the motherboard IDE controller and suddenly the geometry is the same. Use fdisk and partition the disk, write it, and then move to the promise controller and the "correct" geometry was used (that is, it is now the same as when hooked up to the motherboard ide controller).
Quoted 6 lines
Why was it important to me? I'm doing RAID1 and it is really nice to have the same geometry so that the partition info is the same between
the two drives. Makes life easier.--
-rupa
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/
Quoted 8 lines · Mark Lord
Simple solution is to have kernel fall-back to LBA style translations instead of kernel "basic" translations. This would make it match the first two "BIOS" drives on most systems, and not really hurt anything in most cases.
Even better would be to add a stage in front of the fall-back, which queries the BIOS (from kernel startup code) for translation info on ALL drives.
We did that once Mark and people bitch as squealed like a stuck pig!
Also, Linus is still on the line and holding that linux has a policy of not policies.
Maybe a compile option could help...
Cheers,
Andre Hedrick
Linux ATA Development
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/
Quoted 5 lines · Andre Hedrick
Quoted 3 lines · Mark Lord
Even better would be to add a stage in front of the fall-back, which queries the BIOS (from kernel startup code) for translation info on ALL drives.
Maybe a compile option could help...
kernel parameter passed via lilo...
-Dan
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/
Okay......
Quoted 12 lines · Dan Hollis
Quoted 5 lines · Andre Hedrick
Quoted 3 lines · Mark Lord
Even better would be to add a stage in front of the fall-back, which queries the BIOS (from kernel startup code) for translation info on ALL drives.
Maybe a compile option could help...
kernel parameter passed via lilo...
(quoted diff) +0
-2-Dan -
Quoted 4 lines
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/
Andre Hedrick
Linux ATA Development
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/
[Rupa Schomaker]
Quoted 2 lines
In my case, I have two identical Maxtor drives, but they reported different geometry.
[...]
Quoted 3 lines
I'm doing RAID1 and it is really nice to have the same geometry so that the partition info is the same between the two drives. Makes life easier.
If that's what you needed, you could have used 'dd' to copy the partition table from one drive to the other. Easier than going in and re-cabling just to fool your BIOS.
Peter
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
Please read the FAQ at http://www.tux.org/lkml/