Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!ames!sun-barr!texsun!pollux!killer!elg
From: elg@killer.Dallas.TX.US (Eric Green)
Newsgroups: comp.sys.amiga
Subject: Re: OS/2 vs AmigaDOS
Message-ID: <8078@killer.Dallas.TX.US>
Date: 11 May 89 06:53:23 GMT
References: <11609@well.UUCP>
Organization: The Unix(R) Connection, Dallas, Texas
Lines: 49

in article <11609@well.UUCP>, ewhac@well.UUCP (Leo 'Bols Ewhac' Schwab) says:
> In article <6796@cbmvax.UUCP> jesup@cbmvax.UUCP (Randell Jesup) writes:
>>	One problem with VM: Forbid()/Permit() and Disable()/Enable().
>>If you allow another task to run when a Forbidden task takes a page fault,
>>then the Forbid() is broken.  This is compounded by the fact that

There are those of us who would say "So what?". Forbid/Permit, in my
experience, are most often used when grabbing a system resource.
Breaking it only hurts if you access one of those resources. Since the
only program accessing the SCSI resources is the HD driver, breaking
it thus isn't much of a problem.

On the other hand, there are some programs which use Forbid/Permit
around timing-critical code, e.g. a program which twiddles the printer
port to do weird and wonderful things such as a 1541-style bus. I can
see where there'd be a problem there, alas...

>> Disable is, of course, worse.

Understatement of the year ;-).

> 	Then establish the following rule by fiat:
> 
> 	"If you take a page fault while Forbid()den or Disable()d, you
> crash."

Sorry, Leo, that breaks the First Commandment: "Thou Shalt Not Break
Existing Programs." 

I sort of like the solution that Dave Haynie suggested -- adding a
flag to the task structure that says, "You can put me into virtual
memory." This info would have to be put where it could be accessed by
e.g. AllocMem.  If this flag is not set, the program's pages are
locked into "real" memory. This would require MAJOR hacking to
AllocMem, basically adding the resource tracking that everybody so
clamors for (if you get eight AllocMem calls for 8-byte structures for
task X, you don't want to allocate 8 pages!), plus some hacking of
hacking of the task switcher (boy, I'm glad y'all said "Don't mess
with the Process structure!" because adding another flag to the Task
structure will make the offsets wrong for existing programs accessing
the Process structure). By the way, that's not a violation of the
First Commandment, since CATS has been saying for the past couple of
years that the Process structure will probably change....

--
|    // Eric Lee Green              P.O. Box 92191, Lafayette, LA 70509     |
|   //  ..!{ames,decwrl,mit-eddie,osu-cis}!killer!elg     (318)989-9849     |
|  //    Join the Church of HAL, and worship at the altar of all computers  |
|\X/   with three-letter names (e.g. IBM and DEC). White lab coats optional.|