Opinions about MAXIMUM_PARTS crippling in util-linux-2.11r-miscfixes.patch

[DAVID BALAZIC]

James Olin Oden <joden at malachi.lee.k12.nc.us> wrote :

> > On Mon, 8 Sep 2003, Wojtek Pilorz wrote:
> >
> > > Would something so simple as to check whether the name specified matches
> > > .*/hd[a-l] and only then allow up to 60 partitions would be acceptable for
> > > inclusion in RHL?
> >
> > There are too many exceptions for the rule to be useful.
> >
> > I know it's a pain for the three people who have a need for more than 16
> You could add one other person that would be interested in this feature (-;.
> 
> Seriously, what would be an acceptable check for this.  Is there a for
> sure way of looking at something in /proc, or using an ioctl to verify
> that a drive is an ide drive?  If so, I or the person originally emailing
> would be happy to code such a patch.

I think that is the user does not know he is partitioning an IDE drive,
then he should partition at all. When you make above 16 partitions, you usually
have the neccessary amount of IQ to know, IMHO :-)

As Alan said, the limit is in the allocation of device minor numbers. Currently
IDE/ATA(PI) units have 64 minors ( that is 63 minors wasted on CD drives ), while
SCSI devices have about 16 or so. There are also only 32 IDE device files created
by default on Red Hat Linux ( at least it was like that last time I checked ), so
you must create them by hand if you want to use them.
There are several projects in the work to remove or at least lift these kernel limits,
like devfs, dynamic device number allocatios and so on ( don't ask me for details,
I don't know them ).

About the fdisk 16 partition limit ( also see my note below ). I run into this
stupid (IMHO) limit myself. I think the raply from redhat was "SCSI has a 16
partition limit and it would confuse people, if they could make more partition
and then not see them.".

On the other hand parted/disk-druid has no such limitation.
So what is the logic (or should I say unlogic) behind this ?

Is there a reason for a 16 partition limit ?
If yes, why is the default partition tool not respecting this limit ?
If no, why is fdisk crippled ?


That is strange, when I request some logical, meanigfull and user-friendly feature,
I get the "UNIX gives you enough rope...", but when I ask for "more rope" they give me
"oh, some users might be confused...".

I guess it's just a clever looking way of making excuses.


Kind regards,
David Balazic