kurt@tc.fluke.COM (Kurt Guntheroth) (09/30/88)
Here's a question I want answered...Do PS/2 (OS/2) machines have memory protection? I don't mean software segments. I mean hardware protection. Otherwise, you know what will happen. The first OS/2 applications may be quite buggy and may have a tendency to wipe out memory, causing system alerts (we call them GURUs.) What would Jerry think about GURUs from the Gods In Blue? In a year or so, Amiga may have serious revenge on Mr Pournelle.
jimm@amiga.UUCP (Jim Mackraz) (09/30/88)
In article <5384@fluke.COM> kurt@tc.fluke.COM (Kurt Guntheroth) writes:
)Here's a question I want answered...Do PS/2 (OS/2) machines have memory
)protection? I don't mean software segments. I mean hardware protection.
)Otherwise, you know what will happen. The first OS/2 applications may be
)quite buggy and may have a tendency to wipe out memory, causing system
)alerts (we call them GURUs.)
As I understand it OS/2 uses the hardware memory protection afforded
by the 80286. This affords task protection against accidental and
(as I understand it), intentional violations. They had to add special
stuff to support things like keyboard enhancers (input handler hacks).
)What would Jerry think about GURUs from the
)Gods In Blue?
Sometimes I think if we went into a dead-end loop and just hung the
machine, people would have thought we had a more reliable operating
system. They'd say: "this program crashed." As it is, the OS catches
the exceptions, and people say "the system crashed."
Maybe we'll change the Guru message to say:
"Operating system has intercepted fatal application error.
Data has been protected, reboot system."
(By the way, we have more serious plans for the guru, let's not rehash.")
I still think the first products to make money on OS/2 will be debuggers,
but yep, it is a protected multi-tasking OS on a personal computer, stressing
message-based inter-process communication.
)In a year or so, Amiga may have serious revenge on Mr Pournelle.
Why bother?
jimm
--
Jim Mackraz, I and I Computing
amiga!jimm BIX:jmackraz
Opinions are my own. Comments regarding the Amiga operating system, and
all others, are not to be taken as Commodore official policy.
rsilvers@hawk.ulowell.edu (Robert Silvers) (09/30/88)
in article <5384@fluke.COM> kurt@tc.fluke.COM (Kurt Guntheroth) writes: >Here's a question I want answered...Do PS/2 (OS/2) machines have memory >protection? I don't mean software segments. I mean hardware protection. >Otherwise, you know what will happen. The first OS/2 applications may be >quite buggy and may have a tendency to wipe out memory, causing system >alerts (we call them GURUs.) What would Jerry think about GURUs from the >Gods In Blue? > >In a year or so, Amiga may have serious revenge on Mr Pournelle. I believe all 80(2|3)86 processors have hardware memory protection on the chip. This is why they can run Xenix and Unix. --Rob. Robert Silvers. Box #1003 University of Lowell. Lowell Ma, 01854 (617) 452-8823 Rm. 322 "Live free or live in Massachusetts."
cmcmanis%pepper@Sun.COM (Chuck McManis) (09/30/88)
In article <5384@fluke.COM> kurt@tc.fluke.COM (Kurt Guntheroth) writes: >Here's a question I want answered...Do PS/2 (OS/2) machines have memory >protection? I don't mean software segments. I mean hardware protection. >Otherwise, you know what will happen. The first OS/2 applications may be >quite buggy and may have a tendency to wipe out memory, causing system >alerts (we call them GURUs.) What would Jerry think about GURUs from the >Gods In Blue? > >In a year or so, Amiga may have serious revenge on Mr Pournelle. The 80286 and 80386 both have hardware memory protection (with 4 levels of security) and OS/2 uses them to it's advantage. The only differences between OS/2 and AmigaDOS spring from this very fact. I hope C/A comes out with a 68020 version of AmigaDOS that does support the 68851 PMMU in a real way. Yes, it should be able to run in "compatibility" mode and not do rigid protection, but it should also be able to run a process in "enforcement" mode where all those silly crashes result in a *recoverable* alert. Click left button to wipe the process, click the right button for a postmortem dump, press the Middle button for wack (:-)). Sorry, but just because the AT/286 and AT/386 boxes act like glorified MS-DOS machines now, they are capable of much, much more. --Chuck McManis uucp: {anywhere}!sun!cmcmanis BIX: cmcmanis ARPAnet: cmcmanis@sun.com These opinions are my own and no one elses, but you knew that didn't you.
dleigh@hplabsz.HPL.HP.COM (Darren Leigh) (09/30/88)
In article <5384@fluke.COM> kurt@tc.fluke.COM (Kurt Guntheroth) writes: >Here's a question I want answered...Do PS/2 (OS/2) machines have memory >protection? I don't mean software segments. I mean hardware protection. >Otherwise, you know what will happen. The first OS/2 applications may be >quite buggy and may have a tendency to wipe out memory, causing system >alerts (we call them GURUs.) What would Jerry think about GURUs from the >Gods In Blue? > >In a year or so, Amiga may have serious revenge on Mr Pournelle. OS/2 will run on any AT class machine (ATs, PS/2s and 386 boxes running in emulation mode) that has enough memory (about 2 MBytes just for the system if I recall -- I wouldn't try running with less than 4). The 80286 operating in protected mode does have hardware protected memory (thus the name) on a segment by segment basis. Segments may be up to 64K bytes long (ha ha ha ha ha ha . . . ) but can be smaller. This allows you to have somewhat finely grained protection until you run out of segment descriptors. The protection works OK for runaway programs, but not for security as I understand it is easy to get around. Oh, and Jerry's gonna love all those protection violation errors he gets when he trys to write a simple program in C, especially when the the OS spits them out for many reasons. By swapping segments to disk, OS/2 has "virtual memory" (derisive laughter fills the background). There are some problems though. Since segments are normally 64K bytes in length (way too large) the OS thrashes like mad, and since segments can be less than 64K bytes long, the memory gets fragmented and the OS wastes time copying big chunks of memory around. The Presentation Manager is just a glorified version of MS Windows and you have to be a lawyer to program it. The OS/2 Software Development Kit manual set is the most amazing thing I have seen: the weight to usefullness ratio approaches infinity. Look, what can we expect from lousy companies developing for a lousy machine. When all is said and done, OS/2 certainly does beat MS-DOS. But I keep saying to myself: if only the Amiga had memory protection, if only the Amiga had memory protection, if only the Amiga had memory protection . . . ======== Darren Leigh -- I speak for myself, not the company. Internet: dleigh@hplabs.hp.com UUCP: hplabs!dleigh
scott@applix.UUCP (Scott Evernden) (09/30/88)
In article <5384@fluke.COM> kurt@tc.fluke.COM (Kurt Guntheroth) writes: >Here's a question I want answered...Do PS/2 (OS/2) machines have memory >protection? I don't mean software segments. I mean hardware protection. Absolutely. The 80286 maintains chip-level hardware protection and memory management. >Otherwise, you know what will happen. The first OS/2 applications may be >quite buggy and may have a tendency to wipe out memory, causing system >alerts (we call them GURUs.) No, no system alerts. Programs can still crash (obviously), but they are unable to affect other processes or memory. -scott
rminnich@super.ORG (Ronald G Minnich) (09/30/88)
in article <5384@fluke.COM> kurt@tc.fluke.COM (Kurt Guntheroth) writes: >Here's a question I want answered...Do PS/2 (OS/2) machines have memory >protection? I don't mean software segments. I mean hardware protection. AN even more interesting question. (Yes, OS/2 has protection btw. But that is only one of the bomb-producing issues. Unix V6 had protection and it crashed *often*- that's where i cut my kernel teeth). But how does the message passing work? Is it 'copy the message' or 'map it into your space'? If it is the latter, where do they get mapped in? To the same virtual address per process? Or to a different virtual address? If to a different virtual address, then all the pointers in the message are invalid. If to the same, how do they guarantee that there is not something already there? On the amiga if it ever does got memory management it is done via the equivalent of 'map it in'. Since all processes live in a large linear address space it will work right even with memory management- it will still be VERY efficient. Is OS/2 copy or map in? Anybody know? ron
limonce@pilot.njin.net (Tom Limoncelli) (10/02/88)
In article <800@super.ORG> rminnich@super.ORG (Ronald G Minnich) writes: > [dicussing OS/2 & J.P. and life in general with kurt@tc.fluke.COM ] > [ (Kurt Guntheroth) ... much deleted ] > But how does the message passing work? Is it 'copy the message' or > 'map it into your space'? If it is the latter, where do they get mapped > in? To the same virtual address per process? Or to a different virtual > address? If to a different virtual address, then all the pointers > in the message are invalid. If to the same, how do they guarantee > that there is not something already there? > [much deleted] > ron The book "Inside OS/2" really explains things well. Basically, at every corner and every decision the designers made sure that not only could resources be tracked/recovered but there'd be a way to kill the process without ending up in a deadlock. (The book in general is good too.) Basically, when you allocate shared memory it gives you a virtual address that is unused on other processes unique. In other words, it sets aside a range of virtual addresses that will not be allocated with the same virtual address to any other process. This means that shared memory must be explictly allocated. The one thing about OS/2 that kills me is that all the theory looks like it's going to work. Every design choice is the one that I would have picked (except one small one) if I had to implement an OS on a segmented CPU. Almost makes me want to but a machine with an Intel chip... naaaah. I'll just wait for a MMU and a complete redesign of the AmigaOS to handle resource tracking. :-) Tom -- Tom Limoncelli -- Drew University, Box 1060, Madison, NJ 07940 TLimonce@Drew.Bitnet -- limonce@pilot.njin.net -- VoiceMail (201)408-5389 Drew College of Liberal Arts: male/female ratio: 2:3 student/pc ratio: 1:1 "The opinions expressed are mine... just mine."
peter@sugar.uu.net (Peter da Silva) (10/02/88)
In article <Oct.1.21.29.37.1988.8668@pilot.njin.net>, limonce@pilot.njin.net (Tom Limoncelli) writes: > The one thing about OS/2 that kills me is that all the theory looks > like it's going to work. Every design choice is the one that I would > have picked (except one small one) if I had to implement an OS on a > segmented CPU. Almost makes me want to but a machine with an Intel > chip... naaaah. Me to, except in my case that one design change would be: Start with UNIX. Rip the kernel to shreds, if you like, but start with the UNIX programmer interface. Interim versions would be smaller and available a lot sooner. UNIX is overdue for an internal redesign anyway, but the interface is still the best I've ever seen. There are just a couple of system calls that need to be added, and Mach has them. -- Peter da Silva `-_-' peter@sugar.uu.net Have you hugged U your wolf today?
limonce@pilot.njin.net (Tom Limoncelli) (10/03/88)
In article <2731@sugar.uu.net> peter@sugar.uu.net (Peter da Silva) writes: > In article <Oct.1.21.29.37.1988.8668@pilot.njin.net>, limonce@pilot.njin.net (Tom Limoncelli) writes: > > The one thing about OS/2 that kills me is that all the theory looks > > like it's going to work. Every design choice is the one that I would > > have picked (except one small one) if I had to implement an OS on a > > segmented CPU. Almost makes me want to but a machine with an Intel > > chip... naaaah. > > Me to, except in my case that one design change would be: > > Start with UNIX. No, if I wanted Unix, I'd get Unix. Remember, this is a fresh-start OS. I really love Unix but I dread seeing it in homes. Then again, I'm sure someone has come up with a version of Unix that doesn't require hiring a $18,500 sysadmin to keep it running. (Is anyone still selling "I [heart] my sysadmin" bumper stickers?) > > Rip the kernel to shreds, if you like, but start with the UNIX programmer > interface. Interim versions would be smaller and available a lot sooner. > UNIX is overdue for an internal redesign anyway, but the interface is still > the best I've ever seen. There are just a couple of system calls that need > to be added, and Mach has them. I'd like to see Mach. I'd also like to see OS/2 (besides BSing with programmers at PC-Expo in NYC over the summer and reading the book). Of course, I'd like to have enough money to be able to quit school and re-write Unix. Ha ha ha, yeah right. Have you read the book yet? OS/2 has a real nice name space and the interface to the OS is nice too. Everything is called by putting things on the stack like PASCAL does it. All names begin with the name of library. It's very consistant. > -- > Peter da Silva `-_-' peter@sugar.uu.net Tom "I like AmigaDOS... really!" Limoncelli -- Tom Limoncelli -- Drew University, Box 1060, Madison, NJ 07940 TLimonce@Drew.Bitnet -- limonce@pilot.njin.net -- VoiceMail (201)408-5389 Drew College of Liberal Arts: male/female ratio: 2:3 student/pc ratio: 1:1 "The opinions expressed are mine... just mine."
peter@sugar.uu.net (Peter da Silva) (10/03/88)
In article <Oct.2.19.24.59.1988.19140@pilot.njin.net>, limonce@pilot.njin.net (Tom Limoncelli) writes: > In article <2731@sugar.uu.net> peter@sugar.uu.net (Peter da Silva) writes: > > (for OS/2) > > Start with UNIX. > No, if I wanted Unix, I'd get Unix. Remember, this is a fresh-start > OS. I really love Unix but I dread seeing it in homes. Then again, > I'm sure someone has come up with a version of Unix that doesn't > require hiring a $18,500 sysadmin to keep it running. (Is anyone > still selling "I [heart] my sysadmin" bumper stickers?) HP sold a machine called the Integral (oh no, here he goes again) that came with UNIX in ROM. You turned it on and there you were in a friendly windowing shell. No system administrator. When you stuck a disk in its name showed up in a menu. You *couldn't* pull a disk out until it was unmounted. Everything OS/2 is promising, in 1 meg and NO HARD DISK! > Have you read the book yet? OS/2 has a real nice name space and the > interface to the OS is nice too. It really doesn't matter what the innards of the system look like (like I said, the kernel needs replacing anyway), or what the user sees. The really nice thing about UNIX is that there are all these truly clever ideas (setuid bits, fork, /dev/tty, ...) that make it possible to run it effectively on systems that won't even *fit* OS/2. The UNIX paradigm can be cleanly expanded to allow for windowing and real-time (look at Mach). Yes, both AT&T and Berkeley fubbed this one up, but there are clean IPC and realtime systems out there. If they started there they would have a system out, today, that would let people with stock ATs actually start using it. They might even be able to license HPUX/ROM. The reason for OS/2 has nothing to do with technical superiority, a fresh start, or anything like that. It's a blatant attempt to sell IBM hardware, and to lock out the competition. Just business as usual for IBM. -- Peter da Silva `-_-' peter@sugar.uu.net Have you hugged U your wolf today?
scott%applix.uucp@UDEL.EDU (10/04/88)
Received: from CUNYVM by CUNYVM.BITNET (Mailer X2.00) with BSMTP id 7624; Sat, 01 Oct 88 00:21:12 EDT Received: from UDEL.EDU by CUNYVM.CUNY.EDU (IBM VM SMTP R1.1) with TCP; Sat, 01 Oct 88 00:21:08 EDT Received: from Louie.UDEL.EDU by Louie.UDEL.EDU id ab17280; 30 Sep 88 20:37 EDT Received: by Louie.UDEL.EDU id ab17131; 30 Sep 88 20:25 EDT Received: from USENET by Louie.UDEL.EDU id aa16707; 30 Sep 88 20:10 EDT From: Scott Evernden <scott@applix.uucp> Subject: Re: OS/2 vs. Pournelle Message-ID: <814@applix.UUCP> Date: 30 Sep 88 04:00:09 GMT Organization: APPLiX Inc., Westboro MA To: amiga-relay@UDEL.EDU Sender: amiga-relay-request@UDEL.EDU In article <5384@fluke.COM> kurt@tc.fluke.COM (Kurt Guntheroth) writes: >Here's a question I want answered...Do PS/2 (OS/2) machines have memory >protection? I don't mean software segments. I mean hardware protection. Absolutely. The 80286 maintains chip-level hardware protection and memory management. >Otherwise, you know what will happen. The first OS/2 applications may be >quite buggy and may have a tendency to wipe out memory, causing system >alerts (we call them GURUs.) No, no system alerts. Programs can still crash (obviously), but they are unable to affect other processes or memory. -scott
rminnich%super.org@UDEL.EDU (10/04/88)
Received: from CUNYVM by CUNYVM.BITNET (Mailer X2.00) with BSMTP id 5349; Sun, 02 Oct 88 01:01:21 EDT Received: from UDEL.EDU by CUNYVM.CUNY.EDU (IBM VM SMTP R1.1) with TCP; Sun, 02 Oct 88 01:01:15 EDT Received: from Louie.UDEL.EDU by Louie.UDEL.EDU id aa20178; 1 Oct 88 0:15 EDT Received: from USENET by Louie.UDEL.EDU id aa20176; 1 Oct 88 0:14 EDT From: Ronald G Minnich <rminnich@super.org> Subject: Re: OS/2 vs. Pournelle Message-ID: <800@super.ORG> Date: 30 Sep 88 13:37:59 GMT Organization: Supercomputing Research Center, Lanham, MD To: amiga-relay@UDEL.EDU Sender: amiga-relay-request@UDEL.EDU in article <5384@fluke.COM> kurt@tc.fluke.COM (Kurt Guntheroth) writes: >Here's a question I want answered...Do PS/2 (OS/2) machines have memory >protection? I don't mean software segments. I mean hardware protection. AN even more interesting question. (Yes, OS/2 has protection btw. But that is only one of the bomb-producing issues. Unix V6 had protection and it crashed *often*- that's where i cut my kernel teeth). But how does the message passing work? Is it 'copy the message' or 'map it into your space'? If it is the latter, where do they get mapped in? To the same virtual address per process? Or to a different virtual address? If to a different virtual address, then all the pointers in the message are invalid. If to the same, how do they guarantee that there is not something already there? On the amiga if it ever does got memory management it is done via the equivalent of 'map it in'. Since all processes live in a large linear address space it will work right even with memory management- it will still be VERY efficient. Is OS/2 copy or map in? Anybody know? ron
limonce%pilot.njin.net@UDEL.EDU (10/04/88)
Received: from CUNYVM by CUNYVM.BITNET (Mailer X2.00) with BSMTP id 4482; Sun, 02 Oct 88 00:00:55 EDT Received: from UDEL.EDU by CUNYVM.CUNY.EDU (IBM VM SMTP R1.1) with TCP; Sun, 02 Oct 88 00:00:52 EDT Received: from Louie.UDEL.EDU by Louie.udel.EDU id ad03247; 1 Oct 88 22:09 EDT Received: from USENET by Louie.UDEL.EDU id aa03242; 1 Oct 88 22:07 EDT From: Tom Limoncelli <limonce@pilot.njin.net> Subject: Re: OS/2 vs. Pournelle Message-ID: <Oct.1.21.29.37.1988.8668@pilot.njin.net> Date: 2 Oct 88 01:29:37 GMT Organization: NJ InterCampus Network, New Brunswick, N.J. To: amiga-relay@UDEL.EDU Sender: amiga-relay-request@UDEL.EDU In article <800@super.ORG> rminnich@super.ORG (Ronald G Minnich) writes: > [dicussing OS/2 & J.P. and life in general with kurt@tc.fluke.COM ] > [ (Kurt Guntheroth) ... much deleted ] > But how does the message passing work? Is it 'copy the message' or > 'map it into your space'? If it is the latter, where do they get mapped > in? To the same virtual address per process? Or to a different virtual > address? If to a different virtual address, then all the pointers > in the message are invalid. If to the same, how do they guarantee > that there is not something already there? > [much deleted] > ron The book "Inside OS/2" really explains things well. Basically, at every corner and every decision the designers made sure that not only could resources be tracked/recovered but there'd be a way to kill the process without ending up in a deadlock. (The book in general is good too.) Basically, when you allocate shared memory it gives you a virtual address that is unused on other processes unique. In other words, it sets aside a range of virtual addresses that will not be allocated with the same virtual address to any other process. This means that shared memory must be explictly allocated. The one thing about OS/2 that kills me is that all the theory looks like it's going to work. Every design choice is the one that I would have picked (except one small one) if I had to implement an OS on a segmented CPU. Almost makes me want to but a machine with an Intel chip... naaaah. I'll just wait for a MMU and a complete redesign of the AmigaOS to handle resource tracking. :-) Tom -- Tom Limoncelli -- Drew University, Box 1060, Madison, NJ 07940 TLimonce@Drew.Bitnet -- limonce@pilot.njin.net -- VoiceMail (201)408-5389 Drew College of Liberal Arts: male/female ratio: 2:3 student/pc ratio: 1:1 "The opinions expressed are mine... just mine."
peter%sugar.uu.net@cunyvm.cuny.edu (10/04/88)
Received: from CUNYVM by CUNYVM.BITNET (Mailer X2.00) with BSMTP id 1031; Mon, 03 Oct 88 14:28:48 EDT Received: from UDEL.EDU by CUNYVM.CUNY.EDU (IBM VM SMTP R1.1) with TCP; Mon, 03 Oct 88 14:28:42 EDT Received: by Louie.UDEL.EDU id ad05553; 3 Oct 88 13:36 EDT Received: from Louie.UDEL.EDU by Louie.UDEL.EDU id aa04723; 3 Oct 88 12:54 EDT Received: by Louie.UDEL.EDU id ag04658; 3 Oct 88 12:49 EDT Received: from USENET by Louie.UDEL.EDU id aa04128; 3 Oct 88 12:11 EDT From: Peter da Silva <peter@sugar.uu.net> Subject: Re: OS/2 vs. Pournelle Message-ID: <2742@sugar.uu.net> Date: 3 Oct 88 11:33:02 GMT Organization: Sugar Land Unix - Houston, TX To: amiga-relay@UDEL.EDU Sender: amiga-relay-request@UDEL.EDU In article <Oct.2.19.24.59.1988.19140@pilot.njin.net>, limonce@pilot.njin.net (Tom Limoncelli) writes: > In article <2731@sugar.uu.net> peter@sugar.uu.net (Peter da Silva) writes: > > (for OS/2) > > Start with UNIX. > No, if I wanted Unix, I'd get Unix. Remember, this is a fresh-start > OS. I really love Unix but I dread seeing it in homes. Then again, > I'm sure someone has come up with a version of Unix that doesn't > require hiring a $18,500 sysadmin to keep it running. (Is anyone > still selling "I [heart] my sysadmin" bumper stickers?) HP sold a machine called the Integral (oh no, here he goes again) that came with UNIX in ROM. You turned it on and there you were in a friendly windowing shell. No system administrator. When you stuck a disk in its name showed up in a menu. You *couldn't* pull a disk out until it was unmounted. Everything OS/2 is promising, in 1 meg and NO HARD DISK! > Have you read the book yet? OS/2 has a real nice name space and the > interface to the OS is nice too. It really doesn't matter what the innards of the system look like (like I said, the kernel needs replacing anyway), or what the user sees. The really nice thing about UNIX is that there are all these truly clever ideas (setuid bits, fork, /dev/tty, ...) that make it possible to run it effectively on systems that won't even *fit* OS/2. The UNIX paradigm can be cleanly expanded to allow for windowing and real-time (look at Mach). Yes, both AT&T and Berkeley fubbed this one up, but there are clean IPC and realtime systems out there. If they started there they would have a system out, today, that would let people with stock ATs actually start using it. They might even be able to license HPUX/ROM. The reason for OS/2 has nothing to do with technical superiority, a fresh start, or anything like that. It's a blatant attempt to sell IBM hardware, and to lock out the competition. Just business as usual for IBM. -- Peter da Silva `-_-' peter@sugar.uu.net Have you hugged U your wolf today?
scott%applix.uucp%UDEL.EDU@cunyvm.cuny.edu (10/04/88)
Received: from CUNYVM by CUNYVM.BITNET (Mailer X2.00) with BSMTP id 7125; Tue, 04 Oct 88 01:07:47 EDT Received: from UDEL.EDU by CUNYVM.CUNY.EDU (IBM VM SMTP R1.1) with TCP; Tue, 04 Oct 88 01:07:45 EDT Received: from Louie.UDEL.EDU by Louie.UDEL.EDU id aq10567; 3 Oct 88 18:16 EDT Received: from USENET by Louie.UDEL.EDU id aa10444; 3 Oct 88 18:01 EDT From: scott%applix.uucp@UDEL.EDU Subject: Re: OS/2 vs. Pournelle Message-ID: <4395@louie.udel.EDU> Date: 3 Oct 88 22:00:43 GMT To: amiga-relay@UDEL.EDU Sender: amiga-relay-request@UDEL.EDU Received: from CUNYVM by CUNYVM.BITNET (Mailer X2.00) with BSMTP id 7624; Sat, 01 Oct 88 00:21:12 EDT Received: from UDEL.EDU by CUNYVM.CUNY.EDU (IBM VM SMTP R1.1) with TCP; Sat, 01 Oct 88 00:21:08 EDT Received: from Louie.UDEL.EDU by Louie.UDEL.EDU id ab17280; 30 Sep 88 20:37 EDT Received: by Louie.UDEL.EDU id ab17131; 30 Sep 88 20:25 EDT Received: from USENET by Louie.UDEL.EDU id aa16707; 30 Sep 88 20:10 EDT From: Scott Evernden <scott@applix.uucp> Subject: Re: OS/2 vs. Pournelle Message-ID: <814@applix.UUCP> Date: 30 Sep 88 04:00:09 GMT Organization: APPLiX Inc., Westboro MA To: amiga-relay@UDEL.EDU Sender: amiga-relay-request@UDEL.EDU In article <5384@fluke.COM> kurt@tc.fluke.COM (Kurt Guntheroth) writes: >Here's a question I want answered...Do PS/2 (OS/2) machines have memory >protection? I don't mean software segments. I mean hardware protection. Absolutely. The 80286 maintains chip-level hardware protection and memory management. >Otherwise, you know what will happen. The first OS/2 applications may be >quite buggy and may have a tendency to wipe out memory, causing system >alerts (we call them GURUs.) No, no system alerts. Programs can still crash (obviously), but they are unable to affect other processes or memory. -scott
wtm@neoucom.UUCP (Bill Mayhew) (10/04/88)
Back in 1983 or so we dabbled in thinking about making flavors of Unix the operating system on all in house computers. That idea, fortunately, was fairly short lived. At that time, SCO sold a version of Xenix which could be shoe-horned into an IBM XT with 512K of memory and a 10 meg 85 mS hard disk. The performance was pretty laughable, however. At the same time, IBM was also selling it's Unix-like flavor called PC/IX. I think that one required 640K. PC/IX was pretty ridiculous too. For both, there was no hardware protection. It worked, but it wasn't much to look at. --Bill
dan-hankins@cup.portal.com (10/05/88)
In article <2742@sugar.uu.net> peter@sugar.uu.net (Peter da Silva) writes: >The reason for OS/2 has nothing to do with technical superiority, a fresh >start, or anything like that. It's a blatant attempt to sell IBM hardware, >and to lock out the competition. Just business as usual for IBM. Whoa there! OS/2 is as much a product of Microsoft as it is IBM. Plus, it *will* run on any AT compatible, so I don't see how it is an attempt to sell IBM hardware and lock out competition. I do think they could have done a better job, although I'm not sure considering all the hacks that must be done to get by Intel's kludgy architectures. Dan Hankins My opinions and IBM's are mutually orthogonal.
peter@sugar.uu.net (Peter da Silva) (10/06/88)
In article <9747@cup.portal.com>, dan-hankins@cup.portal.com writes: > In article <2742@sugar.uu.net> peter@sugar.uu.net (Peter da Silva) writes: > >The reason for OS/2 has nothing to do with technical superiority, a fresh > >start, or anything like that. It's a blatant attempt to sell IBM hardware, > >and to lock out the competition. Just business as usual for IBM. > Whoa there! OS/2 is as much a product of Microsoft as it is IBM. Plus, it > *will* run on any AT compatible, so I don't see how it is an attempt to > sell IBM hardware and lock out competition. Well, it's not a portable operating system (like, say, Mach), so it will at least lock people out of the Mac. Also, it was envisioned as running ONLY on the PS/2 line at one point. There was some discussion (ahem) about this, so they backed down. > I do think they could have done a better job, although I'm not sure > considering all the hacks that must be done to get by Intel's kludgy > architectures. UNIX runs on intel's kludgy architectures. And runs better than OS/2. So I doubt these hacks are that necessary. I still say they should have started with UNIX and built a real-time kernel to support it... or licensed one (say, Mach from Carnegie Mellon). Have I mentioned Mach yet? I just got the Mach docs and I'm going gaga over them. What will I do NeXT? -- Peter da Silva `-_-' peter@sugar.uu.net Have you hugged U your wolf today?
wtm@neoucom.UUCP (Bill Mayhew) (10/12/88)
Peter was wrting about the hacks of OS/2 fitting to the Intel arcitecture versus the relatively high portability of Unix derivative operating systems. Versus Unix derivatives, the apparent hackishness (is that a word?) of OS/2 seems peculiar at first. Part of the problem is that OS/2 has to go to lengths to be backwards compatible with regular DOS hardware that wants to scribble randomly onto the video RAM and things like that. EGA and CGA video display adapters are pretty stupid, and have fixed addresses for the display buffer. This is a pretty nasty thing to have to deal with when you are trying to multitask two original flavor non-windowed DOS applications that both are trying to scribble into the video RAM. I don't know what OS/2 does, but what Windows 386 does is apparently generate a page fault when an application scribbles into the video RAM, and then switches to single tasking mode until the miscreant task exits. Yuck. At least with Unix, good behavior is enforced from the ground up, so writing a portable O/S is more straight forward, if not easier. I got a chance to mess arond with the beta test version of AT&T's DOS merge product running on a 6386. DOS merge had to go to extensive lengths similar to OS/2ishness to run DOS applications on top of Unix. As you might guess, graphic applications were quite often able to trash the DOS emulator, but at least the Unix underneath kept going. I suppose that I should mention the Amiga since this is its group. One nice thing that the Amiga has going for it is a very sensible video architecture. Now all we need to have is an mmu, so that we can map the video display into virtual memory, and then use the blitter to our advantage for managing what's on the screen. Not that we really are all that excited about multitasking misbehaving DOS applications on our Amigas :-). It really is too bad that the poor IBM doesn't have nice video hardware. That's the problem with an evolutionary product like the PC. You've got to assume least common denominator hardware for anything you write. Jerry would probably even agree. --Bill ...!lll-winken!neoucom!impulse!wtm ps: if you want to flame, please mail to ...neoucom!impulse!wtm for the moment, we're having a bug with incoming mail to ...!neoucom!user. Neoucom's mail server is having a minor brain siezure, which I hope to have fixed soon. My apologies to anybody that's emailed me recently, and didn't get a reply back; your letter probably accidentally disappeared into a black hole. :-)