BLKBSZSET still not working on 2.4.18 ?

BLKBSZSET still not working on 2.4.18 ?

[Ludovic Drolez]

Hi !

I'm trying to backup a partition on an IDE drive which has an odd number 
of sectors (204939). With a stock open/read you cannot access the last 
sector, and that why I tried the BLKBSZSET ioctl to set the basic read 
block size to 512 bytes. I verified the writen value with BLKBSZGET 
ioctl, but I still cannot read this last sector !

I've tried also this:
- FreeBSD : works but it don't want to port all my software to it
- raw devices: can't read the last sector:

# raw  /dev/raw1 /dev/hda2
# dd if=/dev/raw1 of=C bs=512
dd: reading `/dev/raw1': No such device or address
204938+0 records in
204938+0 records out


What can it do ? Wait for a patch in 2.5.xxx ?

TIA,

    Ludovic Drolez.

Re: BLKBSZSET still not working on 2.4.18 ?

[Andries Brouwer]

On Fri, Jan 10, 2003 at 04:32:51PM +0100, Ludovic Drolez wrote:

> I'm trying to backup a partition on an IDE drive which has an odd number 
> of sectors (204939). With a stock open/read you cannot access the last 
> sector
> 
> What can I do ? Wait for a patch in 2.5.xxx ?

Hmm - I recall fixing this both for 2.4 and 2.5.

If that patch is not part of current 2.4, then probably this should be
regarded as a known deficiency of 2.4. If I were to maintain a stable
2.4 I would not accept such changes.

You can test 2.5. If it is wrong there I must submit a patch.

Andries

Re: BLKBSZSET still not working on 2.4.18 ?

[Alan Cox]

On Fri, 2003-01-10 at 15:32, Ludovic Drolez wrote:
> I'm trying to backup a partition on an IDE drive which has an odd number 
> of sectors (204939). With a stock open/read you cannot access the last 
> sector, and that why I tried the BLKBSZSET ioctl to set the basic read 
> block size to 512 bytes. I verified the writen value with BLKBSZGET 
> ioctl, but I still cannot read this last sector !

Its a known 2.4 limitation. The last odd sector isnt normally used by
anything so it has never been a big issue (except with EFI partition
data). There is a patch to allow the last sector to be recovered but
its quite ugly so never went mainstream.

Re: BLKBSZSET still not working on 2.4.18 ?

[John Bradford]

> > I'm trying to backup a partition on an IDE drive which has an odd number 
> > of sectors (204939). With a stock open/read you cannot access the last 
> > sector, and that why I tried the BLKBSZSET ioctl to set the basic read 
> > block size to 512 bytes. I verified the writen value with BLKBSZGET 
> > ioctl, but I still cannot read this last sector !
> 
> Its a known 2.4 limitation. The last odd sector isnt normally used by
> anything so it has never been a big issue (except with EFI partition
> data).

Didn't some really obscure IBM drives use it for something internally,
and shortly after everybody else had to stop using it incase they
overwrote the custom data at the end of an IBM disk, or am I thinking
of something else?

John.

Re: BLKBSZSET still not working on 2.4.18 ?

[Andries Brouwer]

On Fri, Jan 10, 2003 at 06:14:51PM +0000, Alan Cox wrote:

> On Fri, 2003-01-10 at 17:08, John Bradford wrote:
> > Didn't some really obscure IBM drives use it for something internally,
> > and shortly after everybody else had to stop using it incase they
> > overwrote the custom data at the end of an IBM disk, or am I thinking
> > of something else?

> Something else - EFI uses the last sector for partitioning as one example.
> Drives do have protected private areas but they are shielded from normal
> use for obvious reasons

There is also a much older matter. In the distant past IBM used the
last cylinder for testing. That means that many FDISK versions and many
BIOSes subtract one from the number of available cylinders.
There are several ways to ask the BIOS for the size of a disk,
and some of these calls may invoke others, and then subtract one.
I have seen disks that had lost three cylinders that way.

Andries

Re: BLKBSZSET still not working on 2.4.18 ?

[Alan Cox]

On Fri, 2003-01-10 at 17:08, John Bradford wrote:
> Didn't some really obscure IBM drives use it for something internally,
> and shortly after everybody else had to stop using it incase they
> overwrote the custom data at the end of an IBM disk, or am I thinking
> of something else?

Something else - EFI uses the last sector for partitioning as one example.
Drives do have protected private areas but they are shielded from normal
use for obvious reasons

Re: BLKBSZSET still not working on 2.4.18 ?

[Ludovic Drolez]

Andries Brouwer wrote:
> You can test 2.5. If it is wrong there I must submit a patch.
> Andries

It seems to work perfectly with 2.5.56 even without using the BLKBSZSET 
ioctl.
But, I'm still reluctant to use an unstable kernel for backuping a 
partition. But maybe the 2.5 is stable enough to read blocks from 
IDE/SCSI drives, and send them over NFSv2 / IPv4 ?

Cheers,

-- 
Ludovic DROLEZ                                       Free&ALter Soft
152, rue de Grigy - Technopole Metz 2000                  57070 METZ
tel : 03 87 75 55 21                            fax : 03 87 75 19 26

Re: BLKBSZSET still not working on 2.4.18 ?

[Andries Brouwer]

On Mon, Jan 13, 2003 at 12:28:36PM +0100, Ludovic Drolez wrote:

> >You can test 2.5. If it is wrong there I must submit a patch.

> It seems to work perfectly with 2.5.56

Excellent.

> But, I'm still reluctant to use an unstable kernel for backuping a 
> partition. But maybe the 2.5 is stable enough to read blocks from 
> IDE/SCSI drives, and send them over NFSv2 / IPv4 ?

Stable / unstable is not a binary decision. Some things are broken
in 2.4. Many things are broken in 2.5. Still I use 2.5.recent
on a daily basis without meeting bad problems (if I refrain from
inserting and removing USB devices).
Try, and if it works for you, fine.  But no guarantees given.

Andries