Skip site navigation (1)Skip section navigation (2)
Date:      19 Nov 1996 07:18 EST 
From:      "barry (b.a.) scott" <tsbarry@nortel.ca>
To:        joerg_wunsch@uriah.heep.sax.de
Cc:        freebsd-scsi@freebsd.org, peter@taronga.com
Subject:   Re: Drive with 1024 byte logical blocks 
Message-ID:  <199611191226.EAA01372@freefall.freebsd.org>

next in thread | raw e-mail | index | archive | help
> 
> Our UFS is not yet up to the task.  I've seen references to 1024-byte
> sectoring inside #ifdef PC98 though i'm afraid that they've simply
> special-cased this one as opposed to walking through every function
> and removing the implicit assumption that all the world's a DEV_BSIZE
> world.
> 
> MO drives (which are hard-sectored) also suffer from this.
> 

	John Gumb and My self are about to submit a patch against
	2.2 that supports 512, 1024 and 2048 byte sectors for MO.
	(The algorithms scale able 2048 trivially)

	The changes for sd.c to support 1024 byte sectors
	would be simple to create as well, modelled on the od.c
	changes we made.

	Specifically we have msdosfs and ufs fully working on
	2048 byte MO media. We installed the 1006 snap from 2048 byte
	MO media. The code changes are small.

	There are small changes to fdisk to read and write the mbr
	in real section size chunks.

	#ifdef'ed as tahoe in newfs.c and mkfs.c is the place to apply
	sector size scaleing. The kernal ufs code is untouched.

	In the driver we did not fight the DEV_BSIZE assumption. All
	we do is scale block number by the ratio of the real sector size
	over DEV_BSIZE.

	We are still testing the changes to make sure that we have not
	missed something.


			BArry




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199611191226.EAA01372>