jfb@vi.ri.cmu.edu (John Brennen) (08/12/87)
First, someone posted saying that a user-level task can bring down a 386 machine. How? Assuming the user has one readable code segment and one readable data segment, neither of which contains system data; further, the GDT, IDT, and paging directories are protected...now, what's the trick that subverts the protection (also, assume that the O/S attempts no error recovery, but terminates the executing task on any violation). Second, I have heard that Sun is planning a 386-based machine. If this is true, it says something interesting: that someone with no financial interest in Intel and a considerable investment in Motorola has chosen an Intel part. Sun certainly isn't turning its back on Motorola, but broadening one's horizons never hurt anyone. I've written multitaskers for both the 80xxx family and the 68xxx family, and I'll admit that the 68xxx program was cleaner and easier to write. The problem is that a UN*X primitive like fork() is difficult to implement because pointers in the parent and child point to the same object. In my 80xxx version, **because of segmentation**, the pointers do what they are supposed to. Using register-based addressing is the only consistent solution for the 68000, and is (in my opinion) unacceptable. The other "solution" (a 68020 with MMU) was unacceptable at the time, and restricts the use of the program, while the 80xxx version will run identically on anything from an 8088 to an 80386 (32-bit registers and all). Sorry to ramble on without any serious flaming (:-), but to tell the truth, I've been programming these two processor families for over two years, and still haven't made up my mind. Flames to /dev/null, constructive comments are welcome. John Brennen jfb@vi.ri.cmu.edu Visual Inspection Lab Carnegie Mellon U.
jru@etn-rad.UUCP (John Unekis) (08/13/87)
In article <1014@vi.ri.cmu.edu> jfb@vi.ri.cmu.edu (John Brennen) writes: > >Second, I have heard that Sun is planning a 386-based machine. If this >is true, it says something interesting: that someone with no financial >interest in Intel and a considerable investment in Motorola has chosen >an Intel part. Sun certainly isn't turning its back on Motorola, but >broadening one's horizons never hurt anyone. > As far as I know, SUN has no plans to implement an intel based machine. This sounds like a self-serving rumor started by Intel. The next SUN machine out will be the SUN 4, which will be based on a very high speed RISC processor(10+MIPS). The only rumor that I have ever heard about SUN and Intel is that SUN was able to emulate the instruction set of the 8088 in a software routine on the SUN3, and still beat out the PC in performance. They intended to offer this software emulation as their answer to IBM PC compatibility since, as far as I know, SUN would rather die than dirty their hands with an intel chip. >I've written multitaskers for both the 80xxx family and the 68xxx family, >and I'll admit that the 68xxx program was cleaner and easier to write. >The problem is that a UN*X primitive like fork() is difficult to implement .... >"solution" (a 68020 with MMU) was unacceptable at the time, and restricts .... This is a little like saying 'I bought a ford without wheels and I couldn't drive it, so now I only buy chevrolets.'
jfb@vi.ri.cmu.edu.UUCP (08/14/87)
In article <249@etn-rad.UUCP>, jru@etn-rad.UUCP (John Unekis) writes: > In article <1014@vi.ri.cmu.edu> jfb@vi.ri.cmu.edu (John Brennen) writes: > > > >Second, I have heard that Sun is planning a 386-based machine. If this > > As far as I know, SUN has no plans to implement an intel based machine. > This sounds like a self-serving rumor started by Intel. The next SUN ......... > PC compatibility since, as far as I know, SUN would rather die than dirty > their hands with an intel chip. > > >The problem is that a UN*X primitive like fork() is difficult to implement > .... > >"solution" (a 68020 with MMU) was unacceptable at the time, and restricts > .... > This is a little like saying 'I bought a ford without wheels and I > couldn't drive it, so now I only buy chevrolets.' First, who are you to say whether Intel is spreading rumors or whether SUN would dirty its hands with an Intel chip? Seriously, if you know something I don't, tell me. Is there an official SUN policy regarding Intel? As far as I can tell, you aren't affiliated with SUN... (And I heard about the SUN/4 weeks ago; something like 18000 Dhrystones -- wow!) As for your other comment... totally inept analogy. I must admit that I invited it, mentioning the 8086 and the 68020 in the same paragraph. Sure, the 68020 is (in general) a better processor than the 8086. But let's compare the current technology -- 68020 vs. 80386. I use 68020-based Apollo systems regularly, and a Compaq 386 system daily. I'll say that I get much more productive work out of the 386. Note that the 80386 can do on-chip memory management akin to the 68020-family's MMU. As one final, very opinionated comment, what the hell are people doing bashing Intel on comp.sys.ibm.pc? What the hell are people who despise Intel doing reading this newsgroup? I honestly don't understand why you don't just leave us alone -- those of us who happen to use PCs for productive work and who want a forum for discussing aspects of MSDOS machines and their uses. Go away. Thanks. John Brennen CMU visual inspection lab These are my opinions, and I'll be damned if anyone else tries to claim them.
boykin@custom.UUCP (08/14/87)
> As far as I know, SUN has no plans to implement an intel based machine. > This sounds like a self-serving rumor started by Intel. The next SUN > machine out will be the SUN 4, which will be based on a very high speed > RISC processor(10+MIPS). The only rumor that I have ever heard about SUN > and Intel is that SUN was able to emulate the instruction set of the 8088 > in a software routine on the SUN3, and still beat out the PC in performance. > They intended to offer this software emulation as their answer to IBM > PC compatibility since, as far as I know, SUN would rather die than dirty > their hands with an intel chip. Looks like SUN is dead than... They've had a '286 add-on board for their systems for at least a year now, maybe more. By adding this board you have a fully compatable DOS system. DOS comes up as a window on your system that you can toggle between. I know the guy who designed the board and got an early demo, it was pretty nice. Joe Boykin Custom Software Systems ...necntc!custom!boykin
dbercel@toto.uucp (Danielle Bercel, MIS Systems Programming) (08/14/87)
In article <249@etn-rad.UUCP> jru@etn-rad.UUCP (0000-John Unekis) writes: >In article <1014@vi.ri.cmu.edu> jfb@vi.ri.cmu.edu (John Brennen) writes: >> >>Second, I have heard that Sun is planning a 386-based machine. If this >>is true, it says something interesting: that someone with no financial >>interest in Intel and a considerable investment in Motorola has chosen >>an Intel part. Sun certainly isn't turning its back on Motorola, but >>broadening one's horizons never hurt anyone. >> > As far as I know, SUN has no plans to implement an intel based machine. > This sounds like a self-serving rumor started by Intel. The next SUN > machine out will be the SUN 4, which will be based on a very high speed > RISC processor(10+MIPS). The only rumor that I have ever heard about SUN > and Intel is that SUN was able to emulate the instruction set of the 8088 > in a software routine on the SUN3, and still beat out the PC in performance. > They intended to offer this software emulation as their answer to IBM > PC compatibility since, as far as I know, SUN would rather die than dirty > their hands with an intel chip. > First, let me say that I know nothing of Sun's future design and marketing plans. They usually don't consult with me on these decisions :-). We have a product called Sun IPC that is a AT on a board. We don't emulate an 8088, we actually have an Intel 80286 that interfaces with a Sun 3 via the VME bus. The IPC board runs in a window and, for all intents and purposes, is an AT. The clocl speed is 10Mhz and includes one parallel port and two emulated serial ports. In addition to the physical parallel port there are two emulated parallel ports. We also can emulate extended memory, up to 4MB. Also, as part of the stanard software, there is an EPSON to Postscript filter that works perfectly. This includes EPSON bit-mapped graphics. In additon to all this, the board has some further advantages over an AT though. By using NFS we can link our Unix file system to the MSDOS file system. And files can be moved back and forth trannparently. For example, I am using a Sun 3/160 and I have an IPC board in this system. If I were to open the IPC window I would find that I have Drive A: and B: (hardware attatched) and Drive C:, D:, and E:. Drive C: is am emulated hard disk. In reality it is a single unix file. As a consequence, whenever I back up my 3/160 Drive C: gets backed up along with everything else. Drive D: is a RAM disk and Drive E: is actually a directory on my 3/160. This is different from the emulated C: drive. If I were to do a DIR of drive E: I would find that I have 240MB of available disk space. I can access disc space this way (via NFS) on many systems throughout Sun. I can store data and/or program files on an NFSed file system and execute/access them just as I would a file/program on Drive C:. The Sun IPC board is very impressive and I love it. Not as much as I love my Sun 3/160, but it's close. The IPC board allows me to take advantage of all the MSDOS software without having to leave my Unix environment. So, we are in fact using an Intel chip and as far as I know, nobody has died yet :-). danielle --------- -- ---\ UUCP: {hplabs,decvax,}!sun!toto!{danielle,dbercel} ----> Toto, I don't this this is Kansas ---/ ARPA: dbercel@sun.arpa or COM: dbercel%toto@sun
CHRIS@pucc.Princeton.EDU (Christopher Dietrich) (08/14/87)
In article <1016@vi.ri.cmu.edu>, jfb@vi.ri.cmu.edu (John Brennen) writes: >As one final, very opinionated comment, what the hell are people doing >bashing Intel on comp.sys.ibm.pc? What the hell are people who despise >Intel doing reading this newsgroup? I honestly don't understand why you >don't just leave us alone -- those of us who happen to use PCs for >productive work and who want a forum for discussing aspects of MSDOS >machines and their uses. Go away. Thanks. I agree. And I wish the people who have nothing better to do than bash IBM would follow. Chris Dietrich, Systems Programmer Center for Information Technology Princeton University CHRIS@PUCC.PRINCETON.EDU <BITNET>
jjboritz@watdragon.UUCP (08/16/87)
In article <3245@pucc.Princeton.EDU> CHRIS@pucc.Princeton.EDU writes: >In article <1016@vi.ri.cmu.edu>, jfb@vi.ri.cmu.edu (John Brennen) writes: > >>As one final, very opinionated comment, what the hell are people doing >>bashing Intel on comp.sys.ibm.pc? What the hell are people who despise >>Intel doing reading this newsgroup? I honestly don't understand why you >>don't just leave us alone -- those of us who happen to use PCs for >>productive work and who want a forum for discussing aspects of MSDOS >>machines and their uses. Go away. Thanks. > > I agree. And I wish the people who have nothing better to do than >bash IBM would follow. > Well excuse us all for voicing an opinion, but if no one ever complained and we all relied upon good ol' IBM to come up with improvements on their own, where would the PC be now. Jim Boritz <backbone>!watmath!watdragon!jjboritz
roger@celtics.UUCP (Roger B.A. Klorese) (08/16/87)
In article <774@custom.UUCP> boykin@custom.UUCP (Joseph Boykin) writes: >(some other uninformed but talkative soul said:) >> As far as I know, SUN has no plans to implement an intel based machine. Glad you started with the "As far as I know" part... Sun announced several months ago their intention to bring out 80x86 based systems. The obvious business reason is to provide "one-stop shopping" for the people who need MS-DOS applications for some users, and don't need UNIX, or the price tags of even the 3/50 and 3/60 systems. Nobody says they're choosing this as a _technically desirable_ solution, merely a practical one. But they _are_ doing it. -- ///==\\ (No disclaimer - nobody's listening anyway.) /// Roger B.A. Klorese, CELERITY (Northeast Area) \\\ 40 Speen St., Framingham, MA 01701 +1 617 872-1552 \\\==// celtics!roger@seismo.CSS.GOV - seismo!celtics!roger
dleigh@hplabsz.HPL.HP.COM (Darren Leigh) (08/18/87)
In article <1016@vi.ri.cmu.edu> jfb@vi.ri.cmu.edu (John Brennen) writes: >As one final, very opinionated comment, what the hell are people doing >bashing Intel on comp.sys.ibm.pc? What the hell are people who despise >Intel doing reading this newsgroup? I honestly don't understand why you >don't just leave us alone -- those of us who happen to use PCs for >productive work and who want a forum for discussing aspects of MSDOS >machines and their uses. Go away. Thanks. > > John Brennen > CMU visual inspection lab So, everyone who uses an IBM compatible machine has to love Intel? Did it ever cross your mind that there are those of us out here who are spending lots of time doing PC development and still cursing IBM and Intel because of the junk we have to put up with? Do you realize how much development time is wasted because software is so hard to write-for/port-to the PC environment. Intel's segmented architecture and IBM's decision to use the 80x86 has probably set the industry back ten years. In my office I have a 68020 machine running Unix and an AT compatible running MS-DOS. Guess which one I prefer to develop on. In the next cubicle my coworker has an PS/2 Model 80. I've played with it. I've also seen and laughed at OS/2. Guess what I'd still rather work with. The 68020 box wins every time. My 68020 machine does multitasking and has virtual memory. Can any machine with an Intel processor do that? The Amiga does a nice job of multitasking (although no virtual memory) and all it has is a lowly 68000 with no memory management chip. I've seen multitasking kludges for IBM/Intel machines, but none that work as well as the Amiga. Sure there is a huge market for machines that use Intel processors, but can any informed person out there honestly sat that they are "better"? Darren Leigh dleigh@hplabs.hp.com DISCLAIMER: The preceding opinions are mine and may or may not be shared by my employers.
ballou@wheatena (Kenneth R. Ballou) (08/27/87)
In article <1014@vi.ri.cmu.edu> jfb@vi.ri.cmu.edu (John Brennen) writes: >First, someone posted saying that a user-level task can bring down a 386 >machine. How? I was surprised while reading the 80386 Programmer's Reference Manual to see that in Virtual 8086 mode, the instructions IN, INS, OUT, and OUTS are *not* IOPL (I/O Privilege Level) sensitive. So, for example, a user task could reset the CPU in the same way as one resets the 80286 in the AT by sending an appropriate command to the 8042.
rich@devvax.JPL.NASA.GOV (Richard Pettit) (09/11/87)
In article <1197@cartan.Berkeley.EDU> ballou@wheatena.UUCP (Kenneth R. Ballou) writes: >In article <1014@vi.ri.cmu.edu> jfb@vi.ri.cmu.edu (John Brennen) writes: >... >Manual to see that in Virtual 8086 mode, the instructions IN, INS, OUT, >and OUTS are *not* IOPL (I/O Privilege Level) sensitive. So, for example, >a user task could reset the CPU in the same way as one resets the 80286 in >the AT by sending an appropriate command to the 8042. However, those instructions ARE sensitive to the I/O permission bit map. Therefore, if you want to allow your virtual PC to reboot the ENTIRE machine, help yourself. What do you think a V86 monitor is for ? So that you can model the limitations of the virtual PC. rich@devvax.jpl.nasa.gov
alexande@drivax.UUCP (Mark Alexander) (09/11/87)
In article <1197@cartan.Berkeley.EDU> ballou@wheatena.UUCP (Kenneth R. Ballou) writes: > I was surprised while reading the 80386 Programmer's Reference >Manual to see that in Virtual 8086 mode, the instructions IN, INS, OUT, >and OUTS are *not* IOPL (I/O Privilege Level) sensitive. So, for example, >a user task could reset the CPU in the same way as one resets the 80286 in >the AT by sending an appropriate command to the 8042. However, there is an I/O permission bitmap in the TSS that protects against unwanted I/O instructions. So if the OS sets up the TSS correctly, the system will be protected. [stuff to keep news happy] -- Mark Alexander ...{hplabs,seismo,sun,ihnp4}!amdahl!drivax!alexande "Bob-ism: the Faith that changes to meet YOUR needs." -- Bob
stever@tekgen.TEK.COM (Imants Golts) (09/12/87)
In regards to Sun's plans for an intel based machine... In Software Business, an insert in Computer Systems News, on page 3 is a report that Sun is working out details of incorporating Phoenix's MS-DOS compatibility technology in Sun's Unix workstations. In addition, a systems analyst at an investment firm suggests that one of Sun's options is a 386-based workstation running Unix on MS-DOS via Phoenix BIOS technology. (Sept 7, l987 issue) I hope they do it as I am not impressed with any of the 386's out there now especially the ones using the Intel mother board however cheap they are or become. ---Steve Rogers