Subject: Re: wedges vs. not-quite-wedges, was > 1T filesystems, disklabels,
To: Jonathan Stone <jonathan@DSG.Stanford.EDU>
From: Bill Studenmund <wrstuden@netbsd.org>
List: tech-kern
Date: 12/19/2002 17:08:33
On Thu, 19 Dec 2002, Jonathan Stone wrote:

> In message <Pine.NEB.4.33.0212191553170.9004-100000@vespasia.home-net.internetc
> onnect.net>Bill Studenmund writes
>
> >So what exactly is wrong with an LVM? Have you used one?
>
>
> Yes, I have. From early AIX to Veritas.  There's no call to be
> insulting; What you propose is a subset of an LVM, and LVMs dont
> really solve the issues I've brought up.

I'm sorry, I am NOT proposing a subset of an LVM.

I am proposing extending our partitioning methodology to support
partitions coming from different partitioning schemes.

While I'd love for us to have an LVM, and I think we should, I think the
logical volumes of an LVM are different beasties from partitions, and as
such neither the partitioning stuff I'm describing nor wedges will do for
them.

i.e. either we have a full-blooded LVM running the disk, or we don't. :-)

Sorry if I was insulting.

> As I've said, several times now: you and I are focused on *different
> problems*; your solutions seem, from my perspective, to make my
> problems signficantly *worse*.
>
> I havent yet seen any substantive answer to the following scenarios:

I think they might have crossed in private mail.

> > *  Suppose I have a CF with an MS-dos filesystem that I'd like to
> >    mount on-the-fly (via a pcmia slot or a USB reader). Pretend its from
> >    a digital camera, if it helps.
> >
> > * Mounting other Uni* filesystems. I've seen a number of NetBSD developers
> >   using multiboot systems with both NetBSD and other OSes. Some are folks
> >   like darrenr, who target FreeBSD in addition to NetBSD; I've also seen
> >   comments from people with employer-provided laptops running FreeBSD or
> >   Linux, with NetBSD in a `spare' partition.
>
>
> Could you clarify just how your proposal addreses these cases?  If
> static (boot-time) case is different from hot-swap media, how does
> each work?

At boot, we itterate over all disks in the system. We read them for
different disklabel types, and find partitions. We add them to a struct
diskpart (a new disklabel-replacing structure). When we have found
everything (or found 255 user ones), we shove the diskpart into the
kernel.

We then listen for disk-insert events. When a disk is inserted, we read it
for partitions as above.

> Or if some or all of these scenarios are all ``future work'',
> to be addressed when we have kernel-to-user/user-to-kernel event
> notification in place, could that be stated upfront?

Said event framework is in, AFAICT.

Take care,

Bill