jfh@rpp386.Dallas.TX.US (John F. Haugh II) (03/23/89)
In article <2044@viper.Lynx.MN.Org> dave@viper.Lynx.MN.Org (David Messer) writes: >And you will probably get a flood of responses correctly pointing out >that what you say is irrelevent. The original message mentioned that >he "crashed the entire system" by running this program (calling it >a "double fault" rather than "double panic"; which may have misled you). >It doesn't matter that the C program has a bug, it still shouldn't >crash the operating system. Some of us wizards would have shrugged the entire episode off with a ``get a real CPU'' remark. The 80286 does have problems. I doubt that a fully functional and robust operating system for an 80286 can ever be had. The chip is brain dead and a waste of good silicon. Various modes of failure cause the program to be completely aborted, and if that program happens to be your operating system, tough luck. >A true wizard carefully reads the question so that he might answer the >question actually asked, rather than just say the first thing that comes >to mind. Some of that too. Others of us are disgusted with bogus hardware. The 80286 is such an example of a total loser implemented on silicon. Intel created the 286 to keep programmers humble, not to be used for anything productive. -- John F. Haugh II +-Quote of the Week:------------------- VoiceNet: (214) 250-3311 Data: -6272 | "Do not drink and bake" InterNet: jfh@rpp386.Dallas.TX.US | -- Arnold Swartzenegger UucpNet : <backbone>!killer!rpp386!jfh +--------------------------------------
david@daisy.UUCP (David Schachter) (03/24/89)
In article <13866@rpp386.Dallas.TX.US> jfh@rpp386.Dallas.TX.US (John F. Haugh II) writes: >The 80286 does have problems. I doubt that a fully functional and >robust operating system for an 80286 can ever be had. The chip >is brain dead and a waste of good silicon. Various modes of failure >cause the program to be completely aborted, and if that program >happens to be your operating system, tough luck. >... >Others of us are disgusted with bogus hardware. >The 80286 is such an example of a total loser implemented on silicon. >... >Intel created the 286 to keep programmers humble, not to be used >for anything productive. Um, you are incorrect. Daisy Systems has been shipping DNIX, a somewhat-Unix-like operating system since 1985. DNIX is robust, very fast, and has had support since Day 1 for multiple windows, networking a la NFS with automount, loadable/sharable libraries and device drivers, and bitmap graphics. It runs on PC-AT class machines quite nicely. Oh, and it supports virtual memory with demand swapping. User program faults cause the program to be killed and an error message printed in the window where the program was started. Your claim that the 286 isn't to be used for productive work will come as a surprise to the 10,000 or so users of DNIX. The 80286 doesn't implemenent as nice an architecture as a 68000 or NS32016. But is is a lot better than the PDP-11, which is what the chip was modelled after (as well as upwardly compatible with the 8086/8.) The chip was NOT designed with UNIX in mind. At the time it was designed, UNIX was not a clear winner. Nor was PC-DOS. The design was well underway in 1982. What CPU were you using in 1982? How much did it cost? I am willing to bet a very small amount of money your CPU was either a lot more expensive than an 80286-based computer or a lot less power- full. I should know better than to reply to articles when I'm furious. -- David Schachter Disclaimer: I have no financial interest in Intel or its competitors. The success or failure of the 286 is not related to my success or failure. Un- warranted attacks on computer architectures offends me.
ken@gatech.edu (Ken Seefried iii) (03/24/89)
In article <2863@daisy.UUCP> david@daisy.UUCP (David Schachter) writes: >In article <13866@rpp386.Dallas.TX.US> jfh@rpp386.Dallas.TX.US (John F. Haugh II) writes: >>The 80286 does have problems. I doubt that a fully functional and >>robust operating system for an 80286 can ever be had. I beg to differ...there is a group at Rice that has ported BSD 4.3 to the 286. While some might not consider 4.3 functional, they have all been locked up somewhere safe...;') (and, no, you can't get 4.3 unless you have a source license, among other things...) > >The 80286 doesn't implemenent as nice an architecture as a 68000 or NS32016. This is, of course, an understatement... >But >is is a lot better than the PDP-11, which is what the chip was modelled after (as >well as upwardly compatible with the 8086/8.) Um...this is sure news to me. The PDP-11 has a nice, clean instruction set, lots (for the time) general purpose registers, and a different MMU scheme. I LIKE to programme the PDP-11. If I'm wrong, do correct me...could you point out what features the PDP and the 286 have in common? The 286 was based on extentions to the 8086...which was based on the 8080...which was based on the 8008...you get the idea... >The chip was NOT designed with UNIX in mind. True...;') I often wonder WHAT it was designed for. This is a serious question! What was Intel shooting for? iRMX? MS-DOS? CP/M? I find it interesting that noone (that I know of) has REALLY written an operating system based on the 286 features (as apposed to adapting an old arch (MS-DOS) or making the arch look like something else (Unix)) except Microsoft (OS/2) and Intel itself (iRMX/286). >What CPU were you using in 1982? As a matter of fact...a PDP-11/2 (and a Z-80)... >Unwarranted attacks on computer architectures offends me. Like putting the 286 and the PDP-11 in the same group? ;') ;') ...ken seefried iii ken@gatech.edu
peter@ficc.uu.net (Peter da Silva) (03/24/89)
In article <2863@daisy.UUCP>, david@daisy.UUCP (David Schachter) writes: > [the 80286] is a lot better than the PDP-11, which is what the chip was > modelled after... Right. A chip with a random instruction set, weird adressing modes and restrictions, and no general-purpose registers was modelled on the PDP-11, with an orthogonal instruction set and addressing modes, and 6 general purpose registers. > Disclaimer: I have no financial interest in Intel or its competitors. The > success or failure of the 286 is not related to my success or failure. Un- > warranted attacks on computer architectures offends me. So why make one? -- Peter da Silva, Xenix Support, Ferranti International Controls Corporation. Business: uunet.uu.net!ficc!peter, peter@ficc.uu.net, +1 713 274 5180. Personal: ...!texbell!sugar!peter, peter@sugar.hackercorp.com.
david@daisy.UUCP (David Schachter) (03/25/89)
ken@gatech.edu takes issue with my claim that the 80286 was modelled after the PDP-11. My sources are one of the microcoders for the chip, one of the layout engineers for the chip, the manager of the 80186 design team, and a general purpose Intel guru, who worked on various 80x86 family chips and is now a production manager on an 80386 fab line in a city whose name I can't possibly spell, in the South West of the U.S.A. So, Ken, maybe they didn't do such a good job of imitation, but the PDP-11 was the source. The 80286 segmentation scheme is certainly better than the PDP-11. But the -11 clearly wins in register availability. I suspect further discussion should be moved to comp.arch or e-mail. I'll attempt to change the followup line accordingly.
donn@titan.rice.edu (Donn Baumgartner) (03/25/89)
In article <18227@gatech.edu> ken@gatech.UUCP (Ken Seefried iii) writes: >In article <2863@daisy.UUCP> david@daisy.UUCP (David Schachter) writes: >>In article <13866@rpp386.Dallas.TX.US> jfh@rpp386.Dallas.TX.US (John F. Haugh II) writes: >>>The 80286 does have problems. I doubt that a fully functional and >>>robust operating system for an 80286 can ever be had. > >I beg to differ...there is a group at Rice that has ported BSD 4.3 to >the 286. While some might not consider 4.3 functional, they have all >been locked up somewhere safe...;') (and, no, you can't get 4.3 unless >you have a source license, among other things...) In all fairness, I must point out that the 4.3bsd port to the 286 was started at the University of Illinois by Doug Gilmore (now with Multiflow). And it is not a "Rice project", in as much as the people working on it are generally spred-out all over the world. And frankly, the project is frowned on by the Rice CS department because they don't want Rice to get a reputation as an OS porting group (and since I'm the only one in the department that has any thing to do with the project, I think they're safe). The Rice CS department is more oriented towards parallel processing (mild understatement); maybe I should have tried porting MACH to the iPSC Hypercube... >>But >>is is a lot better than the PDP-11, which is what the chip was modelled after (as >>well as upwardly compatible with the 8086/8.) > >Um...this is sure news to me. The PDP-11 has a nice, clean >instruction set, lots (for the time) general purpose registers, and a >different MMU scheme. I LIKE to programme the PDP-11. If I'm wrong, >do correct me...could you point out what features the PDP and the 286 >have in common? Actually, the reference probably refers to the segmented nature of the old PDPs as used in early releases of BSD unix. Anyway... this topic is getting beat into the ground. - Donn Baumgartner ATbsd Project Coordinator donn@rice.edu
vandys@hpcupt1.HP.COM (Andrew Valencia(Seattle)) (03/28/89)
/ hpcupt1:comp.unix.microport / ken@gatech.edu (Ken Seefried iii) / 9:15 pm Mar 23, 1989 / >I find it interesting that noone (that I know of) has REALLY written >an operating system based on the 286 features (as apposed to adapting >an old arch (MS-DOS) or making the arch look like something else >(Unix)) except Microsoft (OS/2) and Intel itself (iRMX/286). Gemini (sp?) Computers offered a B-level secure operating system as an important part of the Trusted Computer Base software line. Fully semaphored MP-capable system... all on the '286. Mr. Tao, the president, claimed it was the only OS at that time which used all the protection features the '286 offered. I wonder how they're doing.... Andy
gregg@ihlpb.ATT.COM (Wonderly) (03/28/89)
From article <18227@gatech.edu>, by ken@gatech.edu (Ken Seefried iii): > In article <2863@daisy.UUCP> david@daisy.UUCP (David Schachter) writes: >>The chip was NOT designed with UNIX in mind. > > True...;') I often wonder WHAT it was designed for. This is a > serious question! What was Intel shooting for? iRMX? MS-DOS? > CP/M? Multics comes to mind... 64K segments, 4 access levels, transfer gates... Naw, couldn't be... -- Gregg Wonderly DOMAIN: gregg@ihlpb.att.com AT&T Bell Laboratories UUCP: att!ihlpb!gregg
ken@gatech.edu (Ken Seefried iii) (03/28/89)
In article <9989@ihlpb.ATT.COM> gregg@ihlpb.ATT.COM (Wonderly) writes: >From article <18227@gatech.edu>, by ken@gatech.edu (Ken Seefried iii): >> In article <2863@daisy.UUCP> david@daisy.UUCP (David Schachter) writes: [me wondering out loud about what the 80286 was designed for] > >Multics comes to mind... 64K segments, 4 access levels, transfer >gates... > I thought about that, but it is my understanding (which surely could be wrong) that multics is based on dynamicly allocated, variable size segments of potentially large size. Certainly, the 80286 doesn't fit this criteria with its fixed size, 64K segments. Also, doesn't Multics use more than 4 rings of protection? Gates are a different story...a really good feature of the 80286. The 80386 *IS*, however, Multics-on-a-chip... ...ken seefried iii ken@gatech.edu
campbell@redsox.UUCP (Larry Campbell) (03/28/89)
In article <18250@gatech.edu> ken@gatech.UUCP (Ken Seefried iii) writes: }In article <9989@ihlpb.ATT.COM> gregg@ihlpb.ATT.COM (Wonderly) writes: }>From article <18227@gatech.edu>, by ken@gatech.edu (Ken Seefried iii): }>> In article <2863@daisy.UUCP> david@daisy.UUCP (David Schachter) writes: } }[me wondering out loud about what the 80286 was designed for] } }>Multics comes to mind... 64K segments, 4 access levels, transfer }>gates... } }I thought about that, but it is my understanding (which surely could }be wrong) that multics is based on dynamicly allocated, variable size }segments of potentially large size. Certainly, the 80286 doesn't fit }this criteria with its fixed size, 64K segments. Also, doesn't }Multics use more than 4 rings of protection? Multics segments were limited in size to 256K 36-bit words. In the 1960s, that seemed like a lot. These days it seems pretty puny. Of course, it's still a heck of a lot bigger than 64K bytes. -- Larry Campbell The Boston Software Works, Inc. campbell@bsw.com 120 Fulton Street wjh12!redsox!campbell Boston, MA 02146
mike@cimcorMN.ORG (Michael Grenier) (03/28/89)
From article <18250@gatech.edu>, by ken@gatech.edu (Ken Seefried iii): > In article <9989@ihlpb.ATT.COM> gregg@ihlpb.ATT.COM (Wonderly) writes: >>Multics comes to mind... 64K segments, 4 access levels, transfer >>gates... > I thought about that, but it is my understanding (which surely could > be wrong) that multics is based on dynamicly allocated, variable size > segments of potentially large size. Certainly, the 80286 doesn't fit > this criteria with its fixed size, 64K segments. Also, doesn't Huh, The 80286 doesn't have fixed 64K segments. You set the length of the segment in the segment descriptor for that segment. From a user level on this Microport box, you can allocate a segment of any size <= 64K using the brk() system call. Of course, its the 64K max segment size that is the real pain. -Mike Grenier mike@cimcor.mn.org