
       OS/2 Upload Information Template for ftp-os2.nmsu.edu

Archive Name: pentium.dsk, fix21_01.ram, loadram.exe
		(First two compressed, third uncompressor)
Program Description: IBM fixes for Pentium and 486/quick
Program Description: loader for IBM fixes 
Operating System: OS/2 2.1

Program Source: (IBM BBS)
		(if individual, please give email address of author)
Replaces: 

Your name: Ilya Zakharevich
Your email address: ilya@math.mit.edu

Proposed directory for placement: 2_1/patches

See also loadram.exe, fix2_01.ram. Message from IBM follows:

Subject: Re: Where is Pentium fix? (The real answer)
Sender: news@sernews.raleigh.ibm.com (news)
Message-ID: <CpCA8K.1ttp@sernews.raleigh.ibm.com>
Date: Thu, 5 May 1994 17:07:32 GMT
Lines: 33
Reply-To: seg@vnet.com.com
References: <21214@blue.cis.pitt.edu> <Cp93uM.Hpn@visix.com> <21784@blue.cis.pitt.edu>
Nntp-Posting-Host: seg.bocaraton.ibm.com
Organization: IBM OS/2 Support Center
X-Newsreader: IBM NewsReader/2 v1.00

OK, I did a little of the debugging on this and can at least set the record
straight. The Pentium, and subsequently various 486 chips (probably all of
them by now) included a new feature called "virtual mode extensions (VMEs)."
Even though the Pentium was supposed to be 100% backward-compatible, it wasn't.
The new VMEs (which, note, were not out until after 2.1 was finalized in 93)
cause a new way to generate a trap 000d (access violation) (for the more 
technically inclined, this is when both the virtual interrupt flag and the
virtual interrupt pending flag are set in the extended flags register).

Anyway, two changes (called "defects" for internal tracking purposes) in the
os2krnl file were made to fix the problem. The os2krnl with the 6.585 internal
revision included the fixes to this problem (as well as many others). This
interim fix (like all other interim fixes) has been superceded by the ServicePak
Xr06200 (aka OS/2 2.11 or "MR1"). Therefore, if you run DOS or Windows apps and
have a Pentium, a 486 DX/2 or possibly other very newly-made chips, you will
probably end up running into this bug. the bug tends to croup up particularly
often if you run interupt-driven vdm stuff like comm or networking software.

In addition to the PJ09844 (and a few other) fix packages that had the 6.585
fix, two other "special" interim fixes were also build with the requisite
fixes.  One "pentium.dsk" is a 6.582 build + these 2 defects. The other,
the fix21_01.ram on the BBS, is a 2.1GA kernel with only these two mods.
Personally, I guess I would rather have pentium.dsk; however, I would MUCH
rather have the service pack on my system (and that's what I use at home).
Yes, I know about the problems and, yes, I know that the ServicePak for OS/2
for Windows isn't out, yet.

I have no idea where you get any of these interim fixes by ftp at this time.

I hope this helps sort out some confusion.

Scott E. Garfinkle
OS/2 Tech Support




