Path: utzoo!attcan!uunet!lll-winken!ames!pasteur!ucbvax!bloom-beacon!think!ephraim
From: ephraim@think.COM (Ephraim Vishniac)
Newsgroups: comp.sys.mac
Subject: Re: Virtual Memory for Mac II -- report from a beta-test site
Keywords: virtual
Message-ID: <35421@think.UUCP>
Date: 16 Jan 89 14:20:40 GMT
References: <1250@cascade.Stanford.EDU>
Sender: news@think.UUCP
Reply-To: ephraim@think.com (Ephraim Vishniac)
Organization: Thinking Machines Corporation, Cambridge MA, USA
Lines: 31

In article <1250@cascade.Stanford.EDU> trewitt@cascade.ARPA (Glenn Trewitt) writes:

>Several people asked about the new product "Virtual" (TM) from
>Connectix that provides virtual memory for Macintosh II and other
>Mac's with PMMU's.  In this message, I hope to answer some of those
>questions.

>There are some problems that Connectix freely admits:

>2) Virtual does not currently work with software that does direct
>SCSI I/O, bypassing the OS file manager (?) calls.  Connectix is
>working on this.  Most scanner software does this when actually
>scanning something (but not when doing post-processing).

I think this answers an earlier article's question about, "What if a
SCSI transfer page-faults?"  (Or words to that effect.)  It sounds
like Virtual hooks the I/O calls and examines the length and address
for transfers to be sure that they're paged in.

Besides the problem mentioned with scanners, this could also be a
problem for SCSI disk formatters, which might issue SCSI commands
directly in the course of installing the driver.  (SCSI Evaluator is
another problem if you select certain options.)  I hope Virtual's
instructions are quite clear about possible gotcha's and how to debug
them.

Ephraim Vishniac					  ephraim@think.com
Thinking Machines Corporation / 245 First Street / Cambridge, MA 02142-1214

	"Arlo Guthrie, it seems, has found what he was looking for:
		God, and the Macintosh." (Boston Globe)