[parted-devel] Problem with cylinder size on SD cards, and other problems

[Marcus Brown]

Hi Leslie,

leslie.polzer at gmx.net wrote:
> Hello Marcus,
> 
> On Thu, Oct 12, 2006 at 09:36:41AM +1100, Marcus Brown wrote:
> 
>> Recently I realised (while testing smaller cards) that the cylinder
>> size was defaulting to 8225kB when creating and new label + partition.
>> Of course, this is way too large for a 16MB card ;)
> Indeed.  But why do you need the cylinder size?

TBH, I'm not really sure :)
However, the original intention had been to be able to create partitions
using parted with "unit cyl" so that the partitions were aligned to
cylinder boundaries.

I also discovered by trial and error, that some of my dodgy scripts, like
converting logical-to-primary only worked with "unit cyl", although I
can't seem to prove that anymore :( (even with the old version of RR).
(see below)

In addition, creating a label+partition using any unit sets the unusual
cylinder size, which only seems to cause problems elsewhere later
(these cards need to be capable of being modified by non-parted
partitioning programmes, and non-Linux OSes).

> 
>> The way I've got around this is by using fdisk to create the label and
>> partition instead, which usually results in a cylinder size of 32KB.
>>
>> It also takes a lot longer to execute :(.
>>
>> I've also had to start working in sectors, rather than cylinders,
>> which is a bit more painful.
> I guess.  The libparted-based GNU fdisk (currently beta, ready for
> release soon) has, like its util-linux counterpart, options to set the
> CHS values.  parted will get this in the mid-term future, too (I invite
> you to create a patch in any case).

err... yes ... C :))
I'd like to think I was 'multi-lingual' but my proficiency in C is not
good, and I certainly wouldn't know where to start until I know what the
problem is and the history behind it. This may, of course, change in the
near future ;)

> 
>> Convert logical to primary -------------------------- Initially, I had
>> a function (which worked) which was written because cards formatted by
>> PocketPC contained an extended partition with a single logical fat32.
>> It worked by performing a dummy resize of partition 5, removing both
>> table entries, and creating a new primary with the same start/end as
>> the previous logical. For some reason this no longer works, (although
>> I can't tell you when/why it stopped;) any suggestions on how to do
>> this properly?
> It's hard to say without knowing exactly about the disk layout and the
> steps you performed.  Give us more information and we might be able to
> help.

I've gone to a bit of effort to attempt to prove this theory to myself,
and you, however, I've run out of time atm, so I though it best to post
my results so far. See attached notes.

> 
>   I wish you all the best for your project and hope that we can resolve
> these issues with you.

Thanks :)

> 
>   Leslie
> 
>> ps. hope this is the right list :)
> Yes, anything remotely Parted goes here :)
> 
> 

Ah, good. I thought my questions may have been more 'user' oriented
rather than 'devel'.



Marcus.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 20061013 logical-to-primary tests on 256MB SD.txt.gz
Type: application/x-gzip
Size: 1579 bytes
Desc: not available
Url : http://lists.alioth.debian.org/pipermail/parted-devel/attachments/20061013/f56309ee/20061013logical-to-primarytestson256MBSD.txt.bin
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 20061013 logical-to-primary tests on 512MB SD.txt.gz
Type: application/x-gzip
Size: 2295 bytes
Desc: not available
Url : http://lists.alioth.debian.org/pipermail/parted-devel/attachments/20061013/f56309ee/20061013logical-to-primarytestson512MBSD.txt.bin