2004-06-22 09:30:24

by David Balazic

[permalink] [raw]
Subject: Disk copy, last sector problem

Hi!

cat /dev/hda > /dev/hdc

This would not copy the entire disk as expected, but miss the last sector if
the number of
sectors on hda is odd. ( I used "cat" becasue it has the simplest syntax,
"dd" and other behave the same ).
Has this been fixed recently ?
What about suppport of other sectors sizes, like 8kb ?

Regards,
David Bala?ic

----------------------------------------------------------------------------
-----------
http://noepatents.org/ Innovation, not litigation !
---
David Balazic mailto:[email protected]
HERMES Softlab http://www.hermes-softlab.com
Zagrebska cesta 104 Phone: +386 2 450 8851
SI-2000 Maribor
Slovenija
----------------------------------------------------------------------------
-----------
"Be excellent to each other." -
Bill S. Preston, Esq. & "Ted" Theodore Logan
----------------------------------------------------------------------------
-----------









2004-06-22 17:08:33

by Philippe Troin

[permalink] [raw]
Subject: Re: Disk copy, last sector problem

David Balazic <[email protected]> writes:

> Hi!
>
> cat /dev/hda > /dev/hdc
>
> This would not copy the entire disk as expected, but miss the last sector if
> the number of
> sectors on hda is odd. ( I used "cat" becasue it has the simplest syntax,
> "dd" and other behave the same ).
> Has this been fixed recently ?
> What about suppport of other sectors sizes, like 8kb ?

Have you tried setting the device block size to its sector size?

blockdev --setbsz $(blockdev --getss /dev/...) /dev/...

Phil.

2004-06-23 17:59:49

by Andries Brouwer

[permalink] [raw]
Subject: Re: Disk copy, last sector problem

On Tue, Jun 22, 2004 at 09:52:54AM -0700, Philippe Troin wrote:
> David Balazic <[email protected]> writes:
>
> > Hi!
> >
> > cat /dev/hda > /dev/hdc
> >
> > This would not copy the entire disk as expected, but miss the last sector if
> > the number of
> > sectors on hda is odd. ( I used "cat" becasue it has the simplest syntax,
> > "dd" and other behave the same ).
> > Has this been fixed recently ?
> > What about suppport of other sectors sizes, like 8kb ?
>
> Have you tried setting the device block size to its sector size?
>
> blockdev --setbsz $(blockdev --getss /dev/...) /dev/...

If I understand correctly David is not reporting a problem, but
vaguely recalls that there was a problem in this area long ago,
and asks whet the current status is.

(Yes, today things are better, but not perfect yet :-))

2004-06-24 07:34:00

by David Balazic

[permalink] [raw]
Subject: RE: Disk copy, last sector problem

> From: Andries Brouwer[SMTP:[email protected]]
>
> On Tue, Jun 22, 2004 at 09:52:54AM -0700, Philippe Troin wrote:
> > David Balazic <[email protected]> writes:
> >
> > > Hi!
> > >
> > > cat /dev/hda > /dev/hdc
> > >
> > > This would not copy the entire disk as expected, but miss the last
> sector if
> > > the number of
> > > sectors on hda is odd. ( I used "cat" becasue it has the simplest
> syntax,
> > > "dd" and other behave the same ).
> > > Has this been fixed recently ?
> > > What about suppport of other sectors sizes, like 8kb ?
> >
> > Have you tried setting the device block size to its sector size?
> >
> > blockdev --setbsz $(blockdev --getss /dev/...) /dev/...
>
> If I understand correctly David is not reporting a problem, but
> vaguely recalls that there was a problem in this area long ago,
> and asks whet the current status is.
>
> (Yes, today things are better, but not perfect yet :-))
>
So the copy will still miss the last sector ?
But the blockdev command is a working workaround ?
Are there any downsides of setting the block size to
512 bytes right at boot for all hard drives ?
What about 8kb sectors, do they work ?

Thanks for your time ;-)
David

2004-06-29 22:14:52

by Bill Davidsen

[permalink] [raw]
Subject: Re: Disk copy, last sector problem

David Balazic wrote:
>>From: Andries Brouwer[SMTP:[email protected]]
>>
>>On Tue, Jun 22, 2004 at 09:52:54AM -0700, Philippe Troin wrote:
>>
>>>David Balazic <[email protected]> writes:
>>>
>>>
>>>>Hi!
>>>>
>>>>cat /dev/hda > /dev/hdc
>>>>
>>>>This would not copy the entire disk as expected, but miss the last
>>
>>sector if
>>
>>>>the number of
>>>>sectors on hda is odd. ( I used "cat" becasue it has the simplest
>>
>>syntax,
>>
>>>>"dd" and other behave the same ).
>>>>Has this been fixed recently ?
>>>>What about suppport of other sectors sizes, like 8kb ?
>>>
>>>Have you tried setting the device block size to its sector size?
>>>
>>> blockdev --setbsz $(blockdev --getss /dev/...) /dev/...
>>
>>If I understand correctly David is not reporting a problem, but
>>vaguely recalls that there was a problem in this area long ago,
>>and asks whet the current status is.
>>
>>(Yes, today things are better, but not perfect yet :-))
>>
>
> So the copy will still miss the last sector ?
> But the blockdev command is a working workaround ?
> Are there any downsides of setting the block size to
> 512 bytes right at boot for all hard drives ?
> What about 8kb sectors, do they work ?

Given that the change from 1k to 4k, I suspect that going to 512 will
suck rocks off the bottom of the ocean in terms of performance. I don't
think that 8k works in 2.4, it didn't work in my 2.6, but the error was
"device busy" so it may not be related to anything but having no spare
drive.

--
-bill davidsen ([email protected])
"The secret to procrastination is to put things off until the
last possible moment - but no longer" -me

2004-06-29 22:20:00

by Al Viro

[permalink] [raw]
Subject: Re: Disk copy, last sector problem

On Tue, Jun 29, 2004 at 06:16:25PM -0400, Bill Davidsen wrote:
> >So the copy will still miss the last sector ?
> >But the blockdev command is a working workaround ?
> >Are there any downsides of setting the block size to
> >512 bytes right at boot for all hard drives ?
> >What about 8kb sectors, do they work ?
>
> Given that the change from 1k to 4k, I suspect that going to 512 will
> suck rocks off the bottom of the ocean in terms of performance. I don't
> think that 8k works in 2.4, it didn't work in my 2.6, but the error was
> "device busy" so it may not be related to anything but having no spare
> drive.

First of all, it *will* work without any "blockdev command" crap.
And no, performace is not a problem - we get requests immediately
merged and unlike regular files we don't spend time figuring out
where the hell next block is.