On Mon, Jun 14, 1999 at 11:10:04AM +0200, Ignatios Souvatzis wrote:
> On Sun, Jun 13, 1999 at 04:07:43PM -0700, Erik E. Fair wrote:
> > At 14:40 -0700 6/13/99, Christoph Badura wrote:
> > >chuq@chuq.com (Chuck Silvers) writes:
> > >
> > >>on a related note, it does seem odd that the current buf interface has
> > >>the offset of the i/o (b_blkno) in sectors, but the size of the i/o
> > >>(b_bcount) in bytes.  it would be more consistent for both offset and size
> > >>to have the same units.  or are there devices that can read partial
> > >>sectors, but only at the beginning of a sector?
> > >
> > >Tape drives do that.
> > 
> > But no one uses tape drives through the block device; everyone uses the
> > character ("raw") device, which does not go through the buffer cache.
> 
> Two things I want to mention:
> 
> 1.) while raw tape drives do not use the buffer cache, raw ("character")
>     accesses to block devices _do_ use the buffer header method to 
>     describe to the device driver what to read/write.
> 
> 2.) somebody might wish to use the block device to get read-ahead and better
>     streaming of tape reads/writes...

I forgot:
  2b) Think of what will happen when we implement something like splice(2)... 

	-is