"Stephen C. Tweedie" wrote:
>
> Hi,
>
> On Fri, Apr 07, 2000 at 02:05:51PM -0600, Jeff V. Merkey wrote:
> > > Shouldn't this be handled by the partition-detection code anyhow
> > > (/proc/partitions et al)?
> >
> > Probably, but this would mean putting more code into the 2.4 release. I
> > looked at the partition stuff in genhd.c - it should go here -- except
> > for one point. NWFS does not share the Unix limitation of "one
> > partition - one file system tree".
>
> There is nothing in Linux which enforces this. You do need to have
> a block device to identify a fs superblock, but there is no reason why
> the filesystem cannot access other block devices too. Ext3 will be
> using this to place the journal on a separate disk from the main
> filesystem, and we have had patches against ext2 in the past to allow
> the filesystem to span multiple block devices.
>
> As far as the block device layer is concerned, all you do is submit
> block IO requests to devices. It doesn't care which devices, and it
> doesn't care who is doing the IO.
Completely agree here.
>
> So, scan the gendisk list for the available partitions, and do what
> you like with them. Your filesystem should not need to make any
> other assumptions about what is going on in the block device layer.
> The partition list is just a list of available devices. It doesn't
> represent filesystems in ANY way, so you can bind multiple such
> partitions together in NWFS if you want to.
Cool.
Some performance issues I am seeing.
>
> --Stephen
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.rutgers.edu
> Please read the FAQ at http://www.tux.org/lkml/
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sat Apr 15 2000 - 21:00:14 EST