Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!uwm.edu!rpi!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!emory!mephisto!udel!mmdf
From: HBO043%DJUKFA11.BITNET@cunyvm.cuny.edu (Christoph van Wuellen)
Newsgroups: comp.os.minix
Subject: Re: minix 1.5 question
Message-ID: <25466@nigel.udel.EDU>
Date: 24 Jul 90 10:19:58 GMT
Sender: mmdf@ee.udel.EDU
Lines: 13

My personal opinion about the progress made from 1.1st --> 1.5 st is:

The disk-I/O is faster because the administration of the buffer pool is
much better.
When dumping dirty buffers to disk, they are sorted by block numbers and
FS talks to FLOPPY via (long) I/O-vectors.

P.S. This is an exellent hook for incorporating track buffering:
switch track buffering on when FLOPPY gets a SCATTERED_IO request,
dump dirty trackbuffers and invalidate all of them when the vector is processed
.

C.v.W.