25 Apr
2005
25 Apr
'05
10:45 a.m.
From Jules Richardson, 25. April 2005 16:25
One thing that'd be useful would be a patched version of dd that could
also retry reading bad blocks n times up to some command-line supplied
maximum; quite often I find for marginal data (such as flakey floppies
and hard disks) an initial read might work but one or more subsequent
reads work find. Far as I know standard dd won't retry reading data at
all.
I think sdd does all this.
See ftp://ftp.berlios.de/pub/sdd
regards,
chris
25 Apr
25 Apr
11:06 a.m.
On 25 Apr 2005, Christian Groessler wrote:
Thanks for all the feedback, everyone!
I checked out sdd and it does not mention anything about errors.
Nonetheless, I'll take everyone's input and get a Linux system running
(something I've been meaning to do anyway). If need be, I'll hack dd to
make it work in a sane manner with errors, and will let everyone know
about it at that time.
In the meantime I've written a file extractor that handles the sector
headers embedded in the Anadisk images and it seems to work pretty well.
It was just more complicated than it needed to be.
--
Sellam Ismail Vintage Computer Festival
------------------------------------------------------------------------------
International Man of Intrigue and Danger http://www.vintage.org
[ Old computing resources for business || Buy/Sell/Trade Vintage Computers ]
[ and academia at www.VintageTech.com || at http://marketplace.vintage.org ]
From Jules
Richardson, 25. April 2005 16:25
One thing that'd be useful would be a patched version of dd that could
also retry reading bad blocks n times up to some command-line supplied
maximum; quite often I find for marginal data (such as flakey floppies
and hard disks) an initial read might work but one or more subsequent
reads work find. Far as I know standard dd won't retry reading data at
all.
I think sdd does all this.
See ftp://ftp.berlios.de/pub/sdd 
11:50 a.m.
>>>> "Vintage" == Vintage
Computer Festival <vcf at siconic.com> writes:
> > From Jules Richardson, 25. April 2005 16:25
> >
> > One thing that'd be useful would be a patched version of dd that
> could > also retry reading bad blocks n times up to some
> command-line supplied > maximum; quite often I find for marginal
> data (such as flakey floppies > and hard disks) an initial read
> might work but one or more subsequent > reads work find. Far as I
> know standard dd won't retry reading data at > all.
That would make sense. The normal assumption is that drivers do
retry. The driver might have a way to adjust how hard it retries, but
certainly no competent driver would give up immediately on a disk I/O
error.
paul
6:31 p.m.
Paul wrote:
That would make sense. The normal assumption is that
drivers do
retry.
Not any more. SCSI and ATA drives are supposed to do that on their
own, so if they return an error to the host, that's after they've
retried multiple times.
It may be possible to tell the drives to use different retry counts,
or to hack the drivers on the host to do retries, or hack the
application program to do retries.
26 Apr
26 Apr
9:26 a.m.
>>>> "Eric" == Eric Smith
<eric at brouhaha.com> writes:
Eric> Paul wrote:
> That would make sense. The normal assumption is
that drivers do
> retry.
Eric> Not any more. SCSI and ATA drives are supposed to do that on
Eric> their own, so if they return an error to the host, that's after
Eric> they've retried multiple times.
Ok, my point really was that retrying *above* the driver layer isn't
likely to be useful.
Meanwhile, the disk drivers I know do include error recovery with
retries after a disk returns an error. Second guessing disk drives
turns out to be useful.
paul
7023
days inactive
7024
days old
4 comments
4 participants
participants (4)
-
cpg＠aladdin.de -
eric＠brouhaha.com -
pkoning＠equallogic.com -
vcf＠siconic.com