NetNews Usenet Archive 1992 #18

home *** CD-ROM | disk | FTP | other *** search

/ NetNews Usenet Archive 1992 #18 / NN_1992_18.iso / spool / comp / sys / ibm / pc / digest / 155 < prev next >

Wrap

Internet Message Format | 1992-08-16 | 43.7 KB

Path: sparky!uunet!cis.ohio-state.edu!ucbvax!BRL.MIL!Info-IBMPC From: Info-IBMPC@BRL.MIL (Info-IBMPC Digest) Newsgroups: comp.sys.ibm.pc.digest Subject: Info-IBMPC Digest V92 #133 Message-ID: <920810195223.V92N133@brl.mil> Date: 10 Aug 92 18:52:23 GMT Sender: daemon@ucbvax.BERKELEY.EDU Reply-To: Info-IBMPC@BRL.MIL Distribution: world Organization: The Internet Lines: 1168 Approved: info-ibmpc@wsmr-simtel20.army.mil Info-IBMPC Digest Mon, 10 Aug 92 Volume 92 : Issue 133 Today's Editor: Gregory Hicks - Rota Spain <GHICKS@brl.mil> Today's Topics: AT&T vs. BSDI --> 4.3BSD-NET2 distribution requires AT&T license Borland mktime() bugs Help: What is the difference between 386 assembly and the old 286 Hmm. Random numbers again (2 msgs) HOW CAN I SET PARENT ENV VARS WITH C ? How easy is file I/O from device drivers in CONFIG.SYS at boot time ? moved my ibmdos.com (6 msgs) Polygon Question (2 msgs) problems with new hard disk Reboot from protected mode REQUEST FOR INFO: DISABLING ^C (3 msgs) Send Replies or notes for publication to: <INFO-IBMPC@brl.mil> Send requests of an administrative nature (addition to, deletion from the distribution list, et al) to: <INFO-IBMPC-REQUEST@brl.mil> Addition and Deletion requests for UK readers should be sent to: <INFO-IBMPC-REQUEST@DARESBURY.AC.UK> Archives of past issues of the Info-IBMPC Digest are available by FTP ONLY from WSMR-SIMTEL20.ARMY.MIL in directory PD2:<ARCHIVES.IBMPC>. ---------------------------------------------------------------------- Date: 21 Jul 92 19:31:48 GMT From: merlin <merlin@neuro.usc.edu> Subject: AT&T vs. BSDI --> 4.3BSD-NET2 distribution requires AT&T license Keywords: AT&T 'Death Star' rises over BSDI's horizon [Tel. 1-800-800-4BSD [For people in comp.unix.sysv386 -- two products were recently released in field test versions with a production version intended in the near future. Berkeley Software Design, Inc [Tel. 1-800-800-4BSD] has offered full source code for a product called 'BSD/386' based on the publically available code made available via numerous archive sites from the UC Regents 4.3BSD-NET2 UNIX software distribution. Willian and Lynne Jolitz have offered the full sources for an alternative product called '386BSD' which is also based on the publically available UC Regents 4.3BSD-NET2 UNIX software distribution. 'BSD/386' sells for $1,000 for the full source and $200 for a binary right to copy. '386BSD' is available from numerous public archive sites without any charge of any kind.] [The UC Regents 4.3BSD-NET2 software was claimed not to contain any AT&T derived source code -- as a consequence it was believed by many people to be an appropriate base for development of extremely inexpensive versions of 'Berkeley UNIX' compatible operating systems. This claim is disputed by AT&T as described below.] [This note is not an advertisement of any kind. I am not connected with AT&T, ATTIS, USG, USDL, USL (or whatever AT&T would like to be called in the near future), Bell Laboratories, UC Regents, CSRG, BSDI, or the Jolitz Development Team. This is simply an expression of concern about litigation which will have a dramatic effect on whether or not AT&T is allowed to have monopoly interest in operating systems derived from publically available source codes. Frankly, I am curious about the ultimate judge's decision.] Two days ago I told a colleague in Psychology that the world had changed -- we spent several hours talking about the possibilities for good which arose from the free or at least very low cost release of 386BSD & BSD/386. However, AT&T [and it's massive army of high paid staff attorneys] have fired the first shot in a legal action which may doom such possibilities. Tomorrow I'm going to have to tell him the forces of darkness and greed have decended to crush hope of using UNIX in extremely low cost systems. The bottom line is that AT&T claims NET2 is contaminated with intellectual property misappropriated from AT&T -- perhaps not direct copies of source code -- but at least ways of doing things (trade secrets) -- and therefore any system derived from NET2 requires an AT&T source code license [it used to be about $10,000 to get an AT&T source license]. The threat is twofold o AT&T can sue anyone who has any assets or any prospect of assets for each and every copy of an operating system maintained or allowed to be copied by any individual without payment of the AT&T license fee o AT&T can withdraw it's contribution from any organization which would permit the maintenance or copying of systems derived from 4.3BSD-NET2 AT&T's complaint 92-1667 filed in US District Court--New Jersey claims: o AT&T authorized the UC Regents to distribute certain works derived from their UNIX software to third parties ... subject to restrictions o Those restrictions include a requirement limiting such distribution to persons who have also acquired licenses from AT&T or USL o 'Networking Release 2' contains software code that was copied from, based upon, or derived from, code licensed to the Regents by AT&T o Any operating system derived from 'Networking Release 2' requires a license from AT&T or its successor, USL AT&T's entire complaint is contained in a false advertising and unfair competition claim based on BSDI's brochure which states BSDI sources are not derived from AT&T code --and-- do not require an AT&T source license. AT&T's 1-MAY-92 interrogatory (a series of questions to an adversary) asks: o Whether anyone related to BSDI has ever had access to AT&T UNIX sources o How much employee time was spent to develop BSDI's source code product BSDI's motion to dismiss and subsequent press releases argue that AT&T has not made out an adequate case of copyright infringement -- and -- therefore cannot maintain their claim of false advertising or unfair competition until they prove what BSDI would like to say is a simple copyright claim. However, the AT&T claim is not a copyright claim -- it is a claim that BSDI incorporated intellectual property belonging to AT&T into the BSDI product. The intellectual property may be in the form of copyright, patent, or trade secret protected material. While NET2 may not literally contain any direct copies of AT&T source code -- it is very possible it contains a translation or adaptation of copyrighted material -- or it may contain a patented means of performing some task -- or it may be based on knowledge of the original techniques [trade secrets] embodied in the AT&T source code. Hence, AT&T only has to prove that someone involved in CSRG's NET2 release or in BSDI's BSD/386 development had access to AT&T licensed materials at some time in his/her lifetime to trigger the spectre of contamination of BSDI's product. BSDI's position is in stark contrast the Phoenix BIOS project where two teams of engineers worked in parallel -- one team developing a functional specification by studying the original IBM BIOS ROM chip codes -- and a second completely independent and compartmentalized team developing code. There is no suggestion either CSRG or BSDI made any effort to institute similar means to prevent the incorporation of AT&T technology in BSD/386. Further, AT&T's question about the time investment of BSDI in bringing out their product [compared with their own cost over many years] will likely go a long way toward supporting their unfair competition claim. If it took a small company like BSDI only a couple of years with a small team of people to produce BSD/386 vs the multi year investment of a corporate giant - then it is very possible AT&T may prevail on the unfair competition claim. On a final note, BSDI's own press release states that: Although USL has not sued the University of California, we expect that USL (or its parent, ATT) will threaten to review or withdraw research grants made to any university or research institution using or distributing software based on NET2 ... All in all, this filing by a corporate giant with virtually unlimited funds for legal expenses would seem to spell the doom of 'free' UNIX projects as they are presently conceived. There is simply too much risk someone with a prior exposure to AT&T source codes could manage to contaminate the product. Sigh, AJ Alexander-James Annala Principal Investigator Neuroscience Image Analysis Network HEDCO Neuroscience Building, Fifth Floor University of Southern California University Park Los Angeles, CA 90089-2520 [mod- as discussed here previously, UNIX does not have any patent protection. The only patent relating to it that I am aware of was the SUID patent which was donated to the public some time ago] ------------------------------ Date: 19 Jul 92 02:08:00 GMT From: Vanzandt <jrv@truth.uucp> Subject: Borland mktime() bugs Borland seems to have bugs in their implementations of mktime(), which converts a broken down time to a time_t (seconds since 0000 UCT 1 Jan 1900). It is supposed to adjust invalid parameters (like 93 seconds or the 43rd day of the month) to be within their proper range. However, Borland's mktime() sometimes increments or decrements the hour even if it's valid. (It's not a one time adjustment, either -- if called again with the same structure, it keeps changing the hour.) The version in the Turbo C++ 1.0 library shows the error for several days near each daylight savings time transition, for example: mktime error 11 -> 10 on Apr 03 10:15:54 1989 mktime error 11 -> 10 on Apr 04 10:15:54 1989 mktime error 11 -> 10 on Apr 05 10:15:54 1989 mktime error 11 -> 10 on Oct 26 10:15:54 1989 mktime error 11 -> 10 on Oct 27 10:15:54 1989 mktime error 11 -> 10 on Oct 28 10:15:54 1989 mktime error 11 -> 10 on Oct 29 10:15:54 1989 mktime error 11 -> 10 on Apr 02 10:15:54 1990 mktime error 11 -> 10 on Apr 03 10:15:54 1990 mktime error 11 -> 10 on Apr 04 10:15:54 1990 mktime error 11 -> 12 on Oct 29 12:15:54 1990 mktime error 11 -> 12 on Oct 30 12:15:54 1990 mktime error 11 -> 12 on Oct 31 12:15:54 1990 The version with Borland C++ 3.00 shows it only exactly on the day of the transition, always incrementing in the spring and decrementing in the fall: mktime error 10 -> 11 on Sun Apr 02 11:24 1989 mktime error 10 -> 9 on Sun Oct 29 09:24 1989 mktime error 10 -> 11 on Sun Apr 01 11:24 1990 mktime error 10 -> 9 on Sun Oct 28 09:24 1990 mktime error 10 -> 11 on Sun Apr 07 11:24 1991 mktime error 10 -> 9 on Sun Oct 27 09:24 1991 mktime error 10 -> 11 on Sun Apr 05 11:24 1992 mktime error 10 -> 9 on Sun Oct 25 09:24 1992 mktime error 10 -> 11 on Sun Apr 04 11:24 1993 mktime error 10 -> 9 on Sun Oct 31 09:24 1993 I believe the hour should be incremented only between 0200 and 0259 on the spring Sunday when daylight savings time starts, and that it should never be decremented. A demo program follows. Does the new Borland C++ 3.1 have bug? - Jim Van Zandt <jrv@mbunix.mitre.org> /* demonstrate a bug in the Borland implementation of mktime() */ #include <stdio.h> #include <time.h> /* relevant declarations */ main() { char buf[200]; int yr, mo, da, hr = 10; /* NOT near 2:00 AM */ time_t now = time(0); struct tm bdt = *localtime(&now); /* fill structure with legal values */ for (yr = 89; yr < 94; yr++) /* test all days in a five year span */ for (mo = 0; mo < 12; mo++) for (da = 1; da < 32; da++) { bdt.tm_year = yr; bdt.tm_mon = mo; bdt.tm_mday = da; bdt.tm_hour = hr; mktime(&bdt); if(bdt.tm_hour != hr) { strftime(buf, 199, "%a %b %d %H:%M %Y", &bdt); printf(" mktime error %2d -> %2d on %s\n", hr, bdt.tm_hour, buf); } } } ------------------------------ Date: 18 Jul 92 04:37:33 GMT From: Mark William Hopkins <markh@csd4.csd.uwm.edu> Subject: Help : What is the difference between 386 assembly and the old 286 882028s@dragon.acadiau.ca (Ghazali Shafie) writes: > >I'm wondering what kind of differences between 386 assembly llang and 286 >( I know there should be a difference since 386 is 32bit ) but what are >the differences and are any new instruction is used or introduced into >a 386 . > >Thanks in advance >. 80386 EXTENSIONS Segment registers FS, GS Stack operations: pop FS, pop GS, push FS, push GS 16-bit Registers now contained inside 32-bit registers: EAX, EBX, ECX, EDX, ESI, EBI, ESP, EBP Extra addressing modes for 32 bit operands. Indirect addressing: [I] Scaled-index addressing: [I + S*B]. I: Any 32-bit register. B: Any 32 bit register, except ESP. S: The SCALE: 1, 2, 4, or 8. Size Altering Prefixes to toggle between 16/32 bit addressing mode and 16/32 bit operand size. opsiz: adrsiz: So most 16-bit operations can now be used on 32-bit objects, and operations involving indirect addressing can now used the extended 32-bit addressing modes. Test and debug registers (CR0-CR7, DR0-DR7, TR0-TR7, not all defined). Test register operationd (Rd = 32-bit register). mov CRn, Rd mov DRn, Rd mov Rd, CRn mov Rd, DRn mov TRn, Rd mov Rd, TRn 16/32 bit offset jumps (Dv = 16/32 bit offset) j* Dv Set byte on condition (byte = condition? 1: 0), (Eb = byte operand). set* Eb Bit test operations (Ev = 16/32 bit effective address, Rv = 16/32 bit register, b = 8-bit constant). First operand is the address, second operand is the bit index. CF = bit(Dest, Index); bt Ev, Rv bt Ev, b CF = bit(Dest, Index); bit(Dest, Index) = 1; bts Ev, Rv bts Ev, b CF = bit(Dest, Index); bit(Dest, Index) = 0; btr Ev, Rv btr Ev, b CF = bit(Dest, Index); bit(Dest, Index) ^= 1; btc Ev, Rv btc Ev, b Double shift left: shld D, S, C ... D = D:S << max(C, 31); shld Ev, Rv, b shld Ev, Rv, CL Double shift right: shrd D, S, C ... D = D:S >> max(C, 31); shrd Ev, Rv, b shrd Ev, Rv, CL Byte/Word to Word/Long conversions (Eb = 8 bit operand, Ew = 16 bit operand). Move with Zero Extension movzx Rv, Eb movzx Rv, Ew Move with Sign Extension movsx Rv, Eb movsx Rv, Ew Convert Double to Quad (EDX:EAX = sign-extend(EAX)) cdq Convert Word to Double Extended (EAX = sign-extend(AX)) cwde Bit scan operations Bit Scan Forward bsf Rv, Ev Bit Scan Reverse bsr Rv, Ev More stack operations Push 32-bit General Registers pushad Pop All General Registers popad Pop Stack into EFLAGS popfd Push EFLAGS Register pushfd ------------------------------ Date: 18 Jul 92 21:19:48 GMT From: Philip Smolen <phil@cs.utexas.edu> Subject: Hmm. Random numbers again. Keywords: random knuth Summary: Everything you ever wanted to know about random numbers tck@netlink.cts.com (Kevin Marcus) writes: >Basically, what I need to do is get random numbers, without using the system >clock, or, if it does use the clock, if I were to take two random numbers >rigth next to each other (this is all in ASM!!!) that I would not get the >same number. You need to read _The Art of Computer Programming, volume 2: Semi-numerical Algorithms_ by Donald E. Knuth. Half of this book discusses good algorithms for generating random numbers. Knuth uses assembly to express his algorithms. You'll probably also want volume one of this series, since it explains Knuth's unique way of expressing algorithms. ------------------------------ Date: 20 Jul 92 14:26:46 GMT From: Alan Modra <amodra@ucs.adelaide.edu.au> Subject: Hmm. Random numbers again. >tck@netlink.cts.com (Kevin Marcus): > Basically, what I need to do is get random numbers, without using the system > clock, or, if it does use the clock, if I were to take two random numbers > rigth next to each other (this is all in ASM!!!) that I would not get the > same number. Get the "improved random number generator" random.c from gatekeeper.dec.com in /.0/bsd-sources/src/lib/libc/gen (or from many other places in netland) This files has LOTS of comments, and is fairly easy to understand, even if you're not a C programmer. It will give you an idea how to create the most efficient assembly routines I know that return reasonably random (in a statistical sense) numbers. For instance, a 32 bit RNG in TASM ideal mode for 386 proc Rand uses si public Rand ; Nifty shift register feedback style of random number generator ; Think of it as 32 shift registers in parallel, with polymonial x^63 + x + 1 ; There's also carries from one register to the next ; Period is approximately 63*(2^63-1) cld mov si,[rptr] lodsd rol eax,1 cmp si,offset EndRandDat jb short Ok1 mov si,offset StartRandDat Ok1: add [si],eax mov [rptr],si ret endp rptr is an index into an array of 63 dwords - the shift register. See random.c for initial values to plug into the array. The line 'rol eax,1' is a slight modification I made to the algorithm, an attempt to give bit 0 the same propagated carries that the other bits have. It seems to work OK but I haven't done exhaustive tests, nor done any analysis of the resultant period (any pure mathematicians out there like a problem?). ------------------------------ Date: 20 Jul 92 12:30:06 GMT From: hdtodd@eagle.wesleyan.edu Subject: HOW CAN I SET PARENT ENV VARS WITH C ? rlmeyering@beckman.com writes: > I am writing an application in BORLAND C within which I need to be able > to SET environment variables that will remain once the program completes. I wrote two C programs that manipulate the parent (in fact, root) environment. One trims environment variables (we use it on our Novell servers) and the other manipulates paths. Either one would show how to manipulate the environment to add a new variable, for example. Here is the code for mp.c, which manipulates the path. /* mp.c Program to modify the path environment for MS-DOS. Written by H. D. Todd, Wesleyan University, 12/26/90 */ #include <stdio.h> #include <stdlib.h> #include <string.h> #include <dos.h> #define MAXSTRS 100 /* increase this if you have more env strings */ #define MAXPATH 500 /* max size of path string */ #define PATHSEP ";" /* separates node names in path */ void main(int argc, char *argv[]) { int i, num, envsize, len, nstrs=0, nextra, pathindex; char dowhat, varstg[MAXPATH], tmpstr[MAXPATH], *strptr[MAXSTRS], far *envp, far *envptr, *subloc; unsigned far *p_psp, far *p_env; struct { char link; unsigned int owner_psp; unsigned int blk_len; } far *envmcb; if (argc < 2) { fprintf(stderr, "Usage: mp [ -{p|a} newnodename | -r oldnodename newnodename]*\n"); exit(1); } /* Construct the PSP pointers */ p_psp = MK_FP(_psp,0x16); /*pointer to parent's PSP*/ p_env = MK_FP(*p_psp,0x2c); /*pointer to parent's environment ptr*/ envptr = MK_FP(*p_env,0x00); /*pointer to parent's environment */ envmcb = MK_FP(*p_env-1,0x00); /* pointer to parent env MCB */ envsize = (*envmcb).blk_len<<4; /* copy strings from parent's environment (FAR) to local strings */ envp = envptr; while (*envp) { /* env strings terminated by second null */ int len; char far *fp; char *lp; for (len=0, fp=envp; *fp; fp++) len++; lp = strptr[nstrs++] = malloc(len+1); for ( ; (*lp++=*envp++); ); } /* find the PATH= string among the env vars */ for (pathindex=0; pathindex<nstrs; pathindex++) if (strnicmp(strptr[pathindex],"PATH=",5) ==0 ) break; if (pathindex>=nstrs) { fprintf(stderr, "MP: Can't find path var!\n"); exit(1); } strcpy(varstg,strptr[pathindex]); /* now do what the switches tell us to do */ while (--argc > 0 ) { if ( ( (dowhat=(*++argv)[0])) != '-') { fprintf(stderr, "MP: begin switches with '-', as in '-%c'\n", dowhat); exit(1); } dowhat = tolower(*++argv[0]); if (dowhat!='p' && dowhat!='a' && dowhat!='r') { fprintf(stderr, "MP: invalid switch -%c\n", dowhat); exit(1); }; if (--argc <=0) { fprintf(stderr, "MP: switch -%c missing node name\n", dowhat); exit(1); } switch (dowhat) { case 'a': strcat(varstg,PATHSEP); strcat(varstg,*++argv); break; case 'p': strcpy(tmpstr,"PATH="); strcat(tmpstr,*++argv); strcat(tmpstr,PATHSEP); strcat(tmpstr,strpbrk(varstg,"=")+1); strcpy(varstg,tmpstr); break; case 'r': if ( (subloc=strstr(varstg,*++argv)) == NULL) { fprintf(stderr,"MP: pathnode %s not found in current path\n", *argv); fprintf(stderr," replacement ignored"); while ( *(argv+1)[0] != '-') { ++argv; argc--; } break; }; strcpy(tmpstr,subloc+strlen(*argv)); if (*(argv+1)[0] != '-') { strcpy(subloc,*++argv); --argc; } else subloc = '\0'; strcat(varstg,tmpstr); break; }; } /* make sure we can copy it all back into the parent's env block */ strptr[pathindex] = &varstg; for (i=0, len=0; i<nstrs; i++) len += strlen(strptr[i]) + 1; if (len+1>envsize) { fprintf(stderr,"MP: Env block too small to store updated PATH string\n"); exit(1); } else { /* OK, do it */ for (i=0; i<nstrs; i++) { for ( ; (*envptr++ = *strptr[i]++) ; ) ; *envptr = '\0'; } } } ------------------------------ Date: 22 Jul 92 08:22:08 GMT From: Len Chisholm <L.Chisholm@massey.ac.nz> Subject: How easy is file I/O from device drivers in CONFIG.SYS at boot time ? Keywords: CONFIG.SYS device drivers file I/O Summary: I want to do file I/O at boot time from a device driver in CONFIG.SYS Hi there, I am interested in doing file operations from a CONFIG.SYS loaded device driver at BOOT time to load configuration information. It is obviously possible as QEMM'S LOADHI.SYS must read other device drivers and the OPTIMISE command causes LOADHI to write statistics to a file. What contortions are required to do this safely , and what limitations are involved ( if any ) ? Thanks in advance, ------------------------------------------------------------------------- | LEN CHISHOLM , Systems programmer | EMail: L.Chisholm@massey.ac.nz | | Department of Production Technology | Phone: +64 6 3505262 | | Massey University | FAX: +64 6 3505604 | | Palmerston North, New Zealand | | ------------------------------------------------------------------------- ------------------------------ Date: 20 Jul 92 09:29:08 GMT From: Volkmar Grote <vg@sup.stollmann.de> Subject: moved my ibmdos.com .... jcmorris@mwunix.mitre.org (Joe Morris) writes: >Several other posters have recommended the Norton Disk Doctor's "Make disk >bootable" function. They're right, and I don't know of any other common >program to do this (other than the FORMAT command ... ah, THAT'S A JOKE!) The other common program is DR DOS' "sys", which is much more sensible than the MS variant; when updating, the latter first destroys your system and then tells you, that there is not enough space ... Volkmar ------------------------------ Date: 19 Jul 92 03:32:40 GMT From: Jen Kilmer <jenk@microsoft.com> Subject: moved my ibmdos.com .... Ralf.Brown@B.GP.CS.CMU.EDU writes: >parry@yoyo.cc.monash.edu.au (Tom J Parry) wrote: >}As of DOS 5.00, this problem no longer exists. Well it does but DOS 5.00 >}can handle it. The SYS command of DOS 5.00 is able to install DOS on a >For DOS 4.0+, only the first three sectors of IO.SYS have to be >contiguous. actually, for 3.30+, only the first three sectors of IO.SYS have to be contiguous. but 4.0+, SYS is able to move any file occupying those first three sectors elsewhere. SYS is also able to move any file(s) in the first two directory entries elsewhere. >5.0, even the requirement that those three sectors be the first three on >the disk has been lifted by storing the actual starting location in the >boot sector. also 5.0's SYS copies COMMAND.COM over as a matter of course...which for the majority of DOS users, makes a normal boot floppy. >Internet: RALF+@CS.CMU.EDU |The University would disclaim this if it knew... >FIDO: Ralf Brown 1:129/26.1 |"Wisdom is the quality that keeps you from >BIT: RALF%CS.CMU.EDU@CARNEGIE| getting into situations where you need it." >AT&Tnet: (412)268-3053 school| -- Doug Larson (gee, I'm correcting Ralf Brown...maybe one of these days I'll get to meet BillG. :-) -jen ------------------------------ Date: 19 Jul 92 16:20:10 GMT From: Leonard Erickson <leonard@qiclab.scn.rain.com> Subject: moved my ibmdos.com .... jenk@microsoft.com (Jen Kilmer) writes: >also 5.0's SYS copies COMMAND.COM over as a matter of course...which for >the majority of DOS users, makes a normal boot floppy. No it does *not*! As many users of 4dos (and other shels that take the place of COMMAND.COM could tell you, DOS 5.0's SYS takes *whatever* program is pointed to by the COMSPEC environment variable and copies it to the disk under the *name* COMMAND.COM! Arguably, the copying could be considered the "correct" behavior, and in that case, to boot without a CONFIG.SYS the renaming is necessary. But as some of these shells require overlay files and the like, this is *not* a good idea. I'd suggest that in the next version or revision, SYS should check and if the command processor *isn't* COMMAND.COM, it should prompt the user and ask if he wants <filename> copied to the disk. If not, then have it ask him for the path to COMMAND.COM and copy that. Trust me, this *has* resulted in very puzzled users with non-bootable disks! Leonard Erickson leonard@qiclab.scn.rain.com CIS: [70465,203] 70465.203@compuserve.com FIDO: 1:105/56 Leonard.Erickson@f56.n105.z1.fidonet.org (The CIS address is checked daily. The others infrequently) ------------------------------ Date: 19 Jul 92 18:17:42 GMT From: Ralf.Brown@b.gp.cs.cmu.edu Subject: moved my ibmdos.com .... jenk@microsoft.com (Jen Kilmer) wrote: }Ralf.Brown@B.GP.CS.CMU.EDU writes: }>For DOS 4.0+, only the first three sectors of IO.SYS have to be }>contiguous. } }actually, for 3.30+, only the first three sectors of IO.SYS have to be }contiguous. but 4.0+, SYS is able to move any file occupying those first }three sectors elsewhere. SYS is also able to move any file(s) in the first }two directory entries elsewhere. I thought it was 3.3+, but was only sure for 4.0+. I remember 3.3 being able to deal with fragmented system files, but I wasn't sure whether that only applied to MSDOS.SYS or both IO.SYS and MSDOS.SYS. }(gee, I'm correcting Ralf Brown...maybe one of these days I'll get to }meet BillG. :-) I've never claimed to be perfect, only that I make fewer than average mistakes.... Internet: RALF+@CS.CMU.EDU |The University would disclaim this if it knew... FIDO: Ralf Brown 1:129/26.1 |"Wisdom is the quality that keeps you from BIT: RALF%CS.CMU.EDU@CARNEGIE| getting into situations where you need it." AT&Tnet: (412)268-3053 school| -- Doug Larson ------------------------------ Date: 21 Jul 92 00:14:08 GMT From: Tom J Parry <parry@yoyo.cc.monash.edu.au> Subject: moved my ibmdos.com .... vg@sup.stollmann.de (Volkmar Grote) writes: > jcmorris@mwunix.mitre.org (Joe Morris) writes: > > >Several other posters have recommended the Norton Disk Doctor's "Make disk > >bootable" function. They're right, and I don't know of any other common > >program to do this (other than the FORMAT command ... ah, THAT'S A JOKE!) > > The other common program is DR DOS' "sys", which is much more sensible > than the MS variant; when updating, the latter first destroys your system and > then tells you, that there is not enough space ... That's right! MSDOS sys just destroys your system and then gives an error message. Done that to me at least 57 times this morning already! Come on, let's not drop down to this level. Sure you prefer the DRDOS version of the program, sure you probably don't know that the DOS 5 version will install the system on any disk where there is enough free space by moving things around if necessary. Sure MSDOS sys might have harmed you once (although I have never ever seen it do any harm myself) but this is no reason to announce that the program destorys every disk it is run on. IT DOES NOT, well not for me anyway. BTW Somebody told me on the weekend that there is an undocumented argument to FDISK which will fix the boot record of a disk that is faulty. I have never tried it nor do I know what kind of disks it works on (presumably only hard disks) but the option is FDISK /MBR (for master boot record). Anyone come across this one? Tom J Parry. Your reality is a figment of my imagination. ------------------------------ Date: 26 Jul 92 03:21:34 GMT From: Jon Freivald <jaf@jaflrn.uucp> Subject: moved my ibmdos.com .... parry@yoyo.cc.monash.edu.au (Tom J Parry) writes: > BTW Somebody told me on the weekend that there is an undocumented > argument to FDISK which will fix the boot record of a disk that is > faulty. I have never tried it nor do I know what kind of disks it works > on (presumably only hard disks) but the option is FDISK /MBR > (for master boot record). Anyone come across this one? Yes indeed - in fact, have used it many times... It's new with DOS 5.0, but you can use it on any disk (assuming you booted 5.0 from a floppy) with any DOS version on it. Why do I use it? - Instant erradication of BSI viruses! It's been quite a handy tool for me - re-writing a clean Master Boot Record on an infected hard disk. Of course, you have to be booting from a clean (preferably write protected - helps prevent accidents) floppy - also do a hard-boot, as you are then *SURE* that memory is virus free when you boot. Jon Jon Freivald ( jaflrn!jaf@uunet.UU.NET ) ------------------------------ Date: 21 Jul 92 22:10:38 GMT From: Glenn Clapp <gclapp@javelin.sim.es.com> Subject: Polygon Question... morley@camins.camosun.bc.ca writes: > >Ok, this should be an easy one... Given 3 X,Y points that form a triangle, >what is the simplest (fastest) way to determine if they are clock-wise or >counter-clock-wise? I'm not interested in general polygons of N points, >just plain old triangles. Yes, this is for use in a graphics app. > Traverse the polygon, calculating 3 cross products (v1->v2 X v2->v3, v2->v3 X v3->v1), and (v3->v1 X v1->v2), and if the cross products are positive, the polygon is counterclockwise, otherwise they will be negetive. Also, if you have a sign flip between the different cross products, then the polygon is not convex. ------------------------------ Date: 25 Jul 92 22:37:19 GMT From: Larry Jones <scjones@thor.sdrc.com> Subject: Polygon Question... gclapp@javelin.sim.es.com (Glenn Clapp) writes: > Traverse the polygon, calculating 3 cross products (v1->v2 X v2->v3, > v2->v3 X v3->v1), and (v3->v1 X v1->v2), and if the cross products are > positive, the polygon is counterclockwise, otherwise they will be negetive. > Also, if you have a sign flip between the different cross products, then > the polygon is not convex. One is sufficient - a triangle is always convex. Larry Jones, SDRC, 2000 Eastman Dr., Milford, OH 45150-2789 513-576-2070 Domain: lawrence.jones@sdrc.com IBMMAIL: USSDR7DR at IBMMAIL I thought my life would seem more interesting with a musical score and a laugh track. -- Calvin ------------------------------ Date: 18 Jul 92 07:06:54 GMT From: "Ira H. Nydick" <hinydick@curtiss.seas.ucla.edu> Subject: problems with new hard disk Hi, I am new to this newsgroup and do not know if it is the appropriate one to post this question to. I apologize in advance if it is not. I recently acquired a Hyundai 386sx (it was free) which I am giving to my father. I have my own computer, a 25 mhz 386 clone. I took this as an oportunity to upgrade my hard drive to a larger unit, and purchased a Maxtor 7120 120 MB IDE drive and cheapo IDE card. I installed my old drive, a 66 MB Toshiba w/RLL card in the Hyundai and the new Maxtor and IDE card in my computer. The Hyundai works fine. My computer, unfortunately, is not cooperating as nicely as its neighbor. I partitioned the disk with MSDOS 4.01 FDISK to one partition, installed the MSDOS 4.01 os that came with the Hyundai (recall, the Hyundai has my old Toshiba drive which contains MSDOS 3.3 so no need to use the included DOS 4.01 on it, so why not on my new drive?), copied all directories and files from the old Toshiba with FASTBACK and restored them on the Maxtor, and, oh yeah, changed the CMOS settings in my AMI BIOS to reflect the new drive. Everything seemed fine for a bit, until various problems startedoccuring. These are the ones I have noticed so far: Computer hangs when 1. I try to use FASTBACK to back up files on it 2. I try to save a worksheet in Lotus 123 3. I simply try to run Wordperfect 5.1 from the C prompt. This is weird because it initially ran WP fine, but now it locks up if I try. ??? 4. I try to save setup changes in my faxboard software. This is the first problem I noticed. 5. Xtree gives a weird 'unknown disk error' message when first run but otherwise seems to run ok. Could it be any of the following reasons? a. My new IDE card is bad b. my new Maxtor drive is bad c. I hooked something up incorrectly (can't imagine what, installation was easy)d. drive is not configured properly in CMOS (store I bought the drive at said to use user-defined drive type, 1024 cyl.,heads:14, sector: 17 and not worry about Write Pcomp and Landing zone. I put zero for them;don't know what they are) e. operating system I installed is one that came with Hyundai (MSDOS 4.01). Could it be a version specifically for Hyundai computers? not likely, rt? f. some sort of interrupt conflict (this area is fuzzy to me) Note: the DOS 4.01 install program put some strange stuff in the config.sys and autoexec.bat files. I tried rebooting with this stuff deleted and Xtree error message dissapeared but rest of problems still present. I plan to upgrade to DOS 5.0 but want to figure out the reason for these problems first. If anyone out there thinks he/she know what is wrong, I would REALLY REALLY appreciate hearing from you. My computer ignorance is surely shining through but I am trying to learn as fast as I can. If it is not a problem pleas email me; if it is, then post and I will try to check the group reasonably often. Thanks, Thanks, Thanks in advance Ira Nydick UCLA MANE Department P.S. sorry for the length of the post, but I don't know what is important to include and what isn't so I may have gone overboard. ------------------------------ Date: 30 Jul 92 03:13:22 GMT From: bill parquet <billp@jupiter.cse.utoledo.edu> Subject: Reboot from protected mode sankey@unixg.ubc.ca (Todd Sankey) writes: : Does anyone have a description or code snippet or anything to : describe how to reboot the PC from protected mode? Rebooting and kicking the computer out of protected mode are pretty similar. To reboot just do: mov al,0feh out 64h, al (It seems to be 100% effective on my 286, but it sometimes causes my 486 to crash). To get out of protected mode, there's a few outs to the CMOS before the big fe->64 to kind of give the bios a resume command. I don't know what they are off the top of my head. (BTW You might want to HLT, jump back to the halt after that... It seems like overkill, but the BIOS does it!) ------------------------------ Date: 21 Jul 92 09:54:50 GMT From: Lee Hollingworth <s1110238@giaeb.cc.monash.edu.au> Subject: REQUEST FOR INFO: DISABLING ^C fredc@ibmpcug.co.uk (Fred Curtis) writes: >I want to disable the ^C interrupt. I can trap it and ignore it >by reseting the DOS interrupt vector, but it still prints "^C" on >the screen. Use the bios functions that use Interrupt 16h function 00h, if you are using MSC use _bios_keybrd(K_READ), if Borland TC++ or C++ use bioskey(0) to wait and read user input. Or below is a simple asm listing which can be compiled and included in a C program... Lee Hollingworth s1110238@giaeb.cc.monash.edu.au # This is a shell archive. Remove anything before this line, # then unpack it by saving it in a file and typing "sh file". # # Wrapped by Lee Hollingworth <s1110238@giaeb> on Tue Jul 21 20:07:03 1992 # # This archive contains: # biosibm.h biosibm.asm # LANG=""; export LANG PATH=/bin:/usr/bin:$PATH; export PATH echo x - biosibm.h cat >biosibm.h <<'@EOF' /**** * file: biosibm.h * purpose: defines for biosibm.asm ****/ #ifndef BIOSKEYS #define BIOSKEYS /* bioskeybrd services */ #define K_READ 0x10 #define K_READY 0x11 #define K_STATUS 0x12 /* enhanced keyboard status bits */ #define K_RSHIFT 0x0000 /* right shift is down */ #define K_LSHIFT 0x0002 /* left shift is down */ #define K_CTRL 0x0004 /* both Ctrl keys down */ #define K_ALT 0x0008 /* both Alt keys down */ #define K_SCROLL 0x0010 /* scroll lock toggle on */ #define K_NUM 0x0020 /* num lock toggle on */ #define K_CAP 0x0040 /* cap lock toggle on */ #define K_INS 0x0080 /* insert toggle is on */ #define K_LCTRL 0x0100 /* left Ctrl key down */ #define K_LALT 0x0200 /* left Alt key down */ #define K_RCTRL 0x0400 /* right Ctrl key down */ #define K_RALT 0x0800 /* right Alt key down */ #define K_PSCROLL 0x1000 /* scroll key is down */ #define K_PNUM 0x2000 /* num key is down */ #define K_PCAP 0x4000 /* cap key is down */ #define K_SYSRQ 0x8000 /* sys rq key is down */ #endif extern unsigned bioskeybrd(unsigned int service); @EOF chmod 600 biosibm.h echo x - biosibm.asm cat >biosibm.asm <<'@EOF' ; ; file: biosibm.asm ; purpose: int 16h keyboard status functions ; author: Lee Hollingworth ; header: biosibm.h ; key_ready equ 01h ; function K_READY .model small .code public _bioskeybrd ;----------------------- bioskeybrd ; call unsigned int bioskeybrd(service); ; unsigned int service; ; services ; K_READ read character and scan code, wait if none ready ; K_READY check if a key is waiting to be read ; K_STATUS get the current keyboard status ; refer to biosibm.h for defines ; _bioskeybrd proc push bp mov bp, sp ; mov stack pointer to base pointer push si push di mov ah, [bp+4] ; get service argument cmp ah, key_ready ; special case for key_ready service jne key_read_status ; jump if not key_ready int 16h jnz alldone ; jmp if character waiting to be read xor ax, ax ; otherwise return 0 jmp alldone key_read_status: int 16h ; simple call for read or status alldone: pop di pop si pop bp ret _bioskeybrd endp end @EOF chmod 600 biosibm.asm exit 0 ------------------------------ Date: 23 Jul 92 13:13:44 GMT From: "Scott C. Sadow" <scott@mycro.uucp> Subject: REQUEST FOR INFO: DISABLING ^C Summary: How do you disable the ^C handling? fredc@ibmpcug.co.uk (Fred Curtis) writes: >I want to disable the ^C interrupt. I can trap it and ignore it >by reseting the DOS interrupt vector, but it still prints "^C" on >the screen. The best way I have found of doing this is to create a small interrupt handler that just sets a flag. Then set interrupt vector 1BH to this routine. (1BH is the BIOS ctrl-break handler) I also set interrupt vector 23H to this same routine, or slightly different. (23H is the DOS ctrl-c/ctrl-break handler) Oh yes - make sure you save both vectors before changing them and restore them before your programs ends. (By the way, I have done this also for interrupt vector 5, which is print screen. Then I can check the global variable and, if set, print the graphics screen.) Scott C. Sadow scott@mycro.UUCP ...gatech!nanovx!mycro!scott ------------------------------ Date: 26 Jul 92 05:55:41 GMT From: mark jose <speednut@csource.oz.au> Subject: REQUEST FOR INFO: DISABLING ^C s110238@giaeb.cc.monash.edu.au (Lee Hollingworth) > fredc@ibmpcug.co.uk (Fred Curtis) writes: > > >I want to disable the ^C interrupt. I can trap it and ignore it > >by reseting the DOS interrupt vector, but it still prints "^C" on > >the screen. Sorry for this late post, but I just read this group. Below is the code I use to catch the ^C interrupt (directly). The code below actually sets a flag "abortflag" to trigger other parts of the program that it's time to quite. If you want to disable the Ctrl-Break altogether then just make "FatalAbort()" do nothing. Hope this helps. /*---------------------------------------------------------------------------*/ /* Handles the capturing of Control-C and Control-Break to allow for normal */ /* exitting during program execution. */ /* */ /* Donated to the Public Domain. USE IT AT ***YOUR*** PERIL!!!!!!!!!!!!!!!! */ /*---------------------------------------------------------------------------*/ #include <dos.h> /* Required for getvect/setvect - Turbo C */ #ifdefined MSOFT /* Using a Microsoft C compiler (QUICK-C?) */ #define enable _enable #define disable _disable #define setvect _dos_setvect #define getvect _dos_getvect #endif /*---------------------------------------------------------------------------*/ /* Local function prototypes and variables. */ /*---------------------------------------------------------------------------*/ static void interrupt far FatalAbort(void); /* The abort interrupt routine */ static void (interrupt far *OldBreak)(void); /* The old abort interrupt */ /*---------------------------------------------------------------------------*/ /* Global variables. */ /*---------------------------------------------------------------------------*/ int localabort; /* The abort flag. */ /*---------------------------------------------------------------------------*/ /* Set up the new interrupt routines for Ctrl-C and Ctrl-Break. */ /*---------------------------------------------------------------------------*/ void cdecl SetBreakOut(void) { disable(); /* Turn off interrupts (for now) */ localabort = 0; /* Set the abort flag. */ OldBreak = getvect(0x1B); /* Save the old interrupt control. */ setvect(0x1B, FatalAbort); /* Set fatal abort vector (1) */ setvect(0x23, FatalAbort); /* Set fatal abort vector (2) */ enable(); /* Turn interrupts back on - flush!*/ } /*---------------------------------------------------------------------------*/ /* Restore the old interrupt back in its old vector. */ /*---------------------------------------------------------------------------*/ void cdecl RestoreBreak(void) { setvect(0x1B, OldBreak); /* Restore old interrupt. */ } /*---------------------------------------------------------------------------*/ /* Interrupt will trap Ctrl-C and Ctrl-Break. Just change abortflag to 1 & */ /* the program will pick it up and deal with it. */ /*---------------------------------------------------------------------------*/ static void interrupt far FatalAbort(void) { localabort = 1; /* Just set the flag and that's it! */ } Above done in Turbo C 2.0 mark jose speednut@csource.oz.au Unique Computing Pty Ltd, Melbourne, Australia The opinions expressed above are that of the author only. ------------------------------ End of Info-IBMPC Digest V92 #133 ********************************* -------