Subject: Re: Non-512-byte sectors
To: Chuck Silvers <>
From: Bill Studenmund <>
List: tech-kern
Date: 09/14/1999 12:13:48
On Tue, 14 Sep 1999, Chuck Silvers wrote:

> (I'm assuming Bill was talking about filesystem access to non-512-byte-sector
> devices... raw access is entirely different, and I haven't looked at that yet.
> it should just be a matter of setting b_blkno differently somewhere in the
> physio() path.)
> er, also someone needs to go thru all the device strategy routines and make
> sure they interpret b_blkno the new way instead of the old way.  ok, so
> there's probably a bunch of work left to be done now that I think about it.
> :-)

My main thought was what did we want to do in terms of Koji Imada's three
proposals (which have been languishing in PR's 3790, 3791, and 3792).

To summarize, they are:

3790:	File system sector size == block io unit == physical sector size

3791:	block io unit == DEV_BSIZE, fs's can do what they want, and
	drivers adjust i/o for different size devices

3792:	block io unit == physical sector size and file system sector size
	can be != block io unit

I gather then either 3790 or 3792 is the consensus?

Does anyone have recent diffs?

Take care,