From: Digestifier To: Linux-Activists@news-digests.mit.edu Reply-To: Linux-Activists@news-digests.mit.edu Date: Tue, 21 Apr 92 06:15:09 EDT Subject: Linux-Activists Digest #89 Linux-Activists Digest #89, Volume #2 Tue, 21 Apr 92 06:15:09 EDT Contents: Re: gdb(1) and ptrace(2) (Eric C. Newton x2092) Re: Drive performance in general (William Michael Lye) re: Kernel panic (John B. Jr. Rogerson) Help, I can't backup my hard disk (Michael Pereckas) Oddities with MCC Release (Ben Combee) Re: How to change the path : Followup (Hongjiu Lu -- Graduate Student) 0.95c+ console and gcc 2.1 problem. (Erik Fichtner) Re: How to change the path : Followup (Demian A. Johnston) Re: Yet another X question (Ken Hughes) If not SCSI, why not floppy? (Richard Burgess) Re: Kernel panic (SAB139@psuvm.psu.edu) Re: 0.95c+ console and gcc 2.1 problem. (Hongjiu Lu -- Graduate Student) generic scsi support in OS/2 (Tim Tsai) cpio doesn't accept multi-volume archives (Andrew Moore) fmtflop in 0.95x...??? (BATES, ROBERT PATRICK) Re: If not SCSI, why not floppy? (Drew Eckhardt) ---------------------------------------------------------------------------- From: enewton@uucp (Eric C. Newton x2092) Subject: Re: gdb(1) and ptrace(2) Reply-To: uunet.uu.net!srg!enewton Date: 17 Apr 92 08:10:10 Path: srg!uunet!uunet!boulder!ophelia.cs.colorado.edu!drew From: drew@ophelia.cs.colorado.edu (Drew Eckhardt) Summary: gdb and ptrace Organization: University of Colorado at Boulder Date: Thu, 16 Apr 1992 11:19:55 GMT 1. gdb does not work with the new .95c+ kernel. Breakpoints end up in wierd places, etc. My guess is that internal structures changed enough from .95a to .95c to break things (ie task_info, etc).... I don't still have my .95a sources to verify this, but if it isn't just me someone should recompile gdb with the new kernel header files. (I have 8M free on my disk, and I don't think it's enough.) I have the whole beastie compiled on my machine at home. The problem I have is that the ptrace() function is not in libc.a. I wrote a dummy ptrace() function to link the program. 3. GDB 4.5 has been released. Is anyone building this?> Yes. I will after 4.4 works. Anyone care to volunteer the sources for making a ptrace system call? -Eric Remember to mail to: uunet.uu.net!srg!enewton "The more you complain, enewton@oswego.oswego.edu the longer God lets you newton@cme.nist.gov live." -- Remember to mail to: uunet.uu.net!srg!enewton "The more you complain, enewton@oswego.oswego.edu the longer God lets you newton@cme.nist.gov live." ------------------------------ From: lye@fraser.sfu.ca (William Michael Lye) Subject: Re: Drive performance in general Date: Tue, 21 Apr 1992 02:42:24 GMT liljeber@kruuna.Helsinki.FI (Mika Pekka Liljeberg) writes: >In article <92109.63088.UPP200@ibm.rhrz.uni-bonn.de> UPP200@ibm.rhrz.uni-bonn.de (Guido Kueppers) wrote: >This bears some thinking about. Obviously the standard Linux HD driver >could use the attentions of a determined optimizing programmer. ;) >I understand that Linus has a lot of things with higher priorities to do, >but perhaps some of those guys, who liked to write turbo loaders for the >Commodore 64 are still around and would like another challenging project. ;-) >Just thought I'd add my two, uhum, pennies worth. > Mika >-- >Mika Liljeberg Email: liljeber@kruuna.Helsinki.FI >Helsinki University Mika.Liljeberg@Helsinki.FI >Dept. of Computer Science Oh, you mean, someone like lil' ol' me? I'm only *just* starting to get into low-level programming on my clone. Its a real bugger to get hold of the info I need. (I mean, what's this DMA stuff? :-) Seriously, if someone can suggest a GOOD source of information on the low-level accessing of the drives on a clone, I'm game to start bashing away, that is if someone hasn't already done something similar... -- Bill Lye, lye@sfu.ca SFU claims these views....NOT! ------------------------------ From: jbrogers@phoenix.Princeton.EDU (John B. Jr. Rogerson) Subject: re: Kernel panic Date: Tue, 21 Apr 1992 02:06:07 GMT With regard to a recent note of mine with the subject "Kernel panic", I received a welcome instructional email note from ted@nmsu.edu that I should have included the level of the system that I am using and the equipment that it is running on. Herewith, I offer the missing information. Linux system version 0.95c+, running on a 25Mhz 386 machine by Gateway 2000 with 4Mb memory. Regards, Jack Rogerson jbrogers@phoenix.princeton.edu ------------------------------ From: mper@uipsuxb.ps.uiuc.edu (Michael Pereckas) Subject: Help, I can't backup my hard disk Date: 21 Apr 92 04:24:46 GMT I can't back up my hard disk. Gnu tar 1.10 can't read files in multivolume archives that are split across disks, cpio and pax can't create multivolume archives at all, and diskbackup/diskrestore fails randomly. Does anyone have any ideas? I'm running 0.95c+ on a 2Mbyte system with an MFM harddrive. I'm using floppy drive /dev/PS1, which works fine with mounted filesystems and single-volume tars. /dev/at0 causes the same problems with tar (and db/dr modified for the smaller disks). Michael Pereckas ------------------------------ From: gt0501c@prism.gatech.EDU (Ben Combee) Subject: Oddities with MCC Release Date: 21 Apr 92 04:45:29 GMT I installed the MCC interim version of Linux, 0.95c+, tonight. It was very well done (the easiest way I know of to get all the shared libraries going), but when I installed the second compimag disk (with g++ and bash) and rebooted, sh quit working. It would not let commands complete output and gave the error message after any external like sh [8 1]: EINVAL. I rebooted using the original root and copied the sh from there, and it worked again. Any ideas -- running a 386/33 (no cache) with 150M ESDI drive. Linux 0.95 worked fine. -- Benjamin L. Combee | -===- Internet: gt0501c@prism.gatech.edu | = ___ "Home of the Big G Burger" 1991-1992 joint enrollee | - = -- MST3K, Comedy Central C, Modula-2, C++ resource | -===- ------------------------------ From: hlu@yoda.eecs.wsu.edu (Hongjiu Lu -- Graduate Student) Crossposted-To: alt.os.linux Subject: Re: How to change the path : Followup Date: 21 Apr 92 03:13:03 GMT In article <1992Apr21.104213.1867@csdvax.csd.unsw.edu.au> s9100786@csdvax.csd.unsw.edu.au (Matthew Jackson) writes: >A short while ago I posted a query on how to change the path. Attached are the >responses I received. I have not tried them as my Linux machine is at home and >my net access is at work. > >It seems those who have had the same problem replied after they worked it out >and also some more familiar with Unix than myself replied too, for which I am >grateful. > >Anyway, I thought it would be of some use to others so I post a summary. > >Summary: For historical reasons, the root user does not have the > current directory in the path. WRONG, WRONG, WRONG. That is not for historical reasons. Do your homework about Unix. That is for security reason. > > Under ash and bash, PATH=dir1:dir2:... > Then export PATH > > Under csh or tcsh, PATH=(dir1 dir2 ...) Please DO NOT DO THAT. KEEP THE WAY IT WAS. H.J. ------------------------------ From: techs@triton.unm.edu (Erik Fichtner) Subject: 0.95c+ console and gcc 2.1 problem. Date: Tue, 21 Apr 92 04:36:35 GMT Hey everyone... I installed 0.95c and compiled the 0.95c+ patches last weekend and now there's a major problem with the terminal emulation routines. whenever I use less or elvis (these are the worst offenders, other things do it too) data falls thru to the 25th line and then all my data comes out on one line on the bottom of the screen. stty rows 24 seems to fix less, but elvis is still screwed up. this never was a problem under 0.95 (no letter) kernel, so I've gotta assume that something got messed up along the way from 0.95 to 0.95c+. Can someone point out what's different here? I'd really like to fix this back to the way 0.95 handled it since 0.95 worked! ;-) the systems kinda useless if I can't read and edit files. Also having a problem with gcc 2.1. One of the things I'm intrested in is ray tracing... I used to run DKBtrace under MS-DOS, and then i ran it under AT&T SVR4 (my last enviornment. yuuuck). So, i went and got the source again and tried to compile it under Linux. *KABOOM!* gcc got very angry and spit this error message at me: gcc: Internal compiler error: program cc1 got fatal signal 6 make: *** [parse.o] Error 1 Correct me if i'm wrong, but that's not supposed to happen, is it? :-) -- Techs techs@triton.unm.edu Disclaimer: any inaccuracies are caused by the phase of the moon "...whaddya mean I ain't kind... Just not *your* kind!" - Megadeth ------------------------------ From: dj1l+@andrew.cmu.edu (Demian A. Johnston) Crossposted-To: alt.os.linux Subject: Re: How to change the path : Followup Date: 21 Apr 92 05:10:30 GMT If you are root or even and "regular" joe user and you decide to put '.' in your path at least put it at the end of the path list to prevent strange problems. Demian ---> daj+@cmu.edu ---> dj1l@andrew.cmu.edu ---> demianjn@ids.jvnc.net ------------------------------ From: hughes@succubus.csee.usf.edu (Ken Hughes) Subject: Re: Yet another X question Date: 20 Apr 92 16:42:07 GMT Reply-To: hughes@succubus.csee.usf.edu (Ken Hughes) In article <1992Apr15.145806.16235@ccu.umanitoba.ca>, umthom61@ccu.umanitoba.ca (Adam Thompson) writes: |> In <1992Apr15.053227.24310@serval.net.wsu.edu> hlu@yoda.eecs.wsu.edu |> (Hongjiu Lu -- Graduate Student) writes: |> |> >From my own experience, a real X11R4 can take as little as 10MB to |> run. |> >on a 386. I only keep the minimum stufffs I need. Beside MIT |> stuffs, |> >I added some other programs, like xgif and xdvi. I can put off X |> source |> >from the net and compile it without much problem. I believe, for |> X11R5, |> >20MB should be a safe bet. |> >H.J. |> |> Perhaps you're assuming he doesn't want to compile it :-) |> |> According to the docs in the R5 distribution, the source tree untars |> to |> roughly 150-200 Megs. To compile, expect to need at least another |> 100Meg. |> |> As far as compiled binaries go, using shared libraries, on a Sun |> i386, |> the full MIT core plus about 50% again in added-on goodies takes |> somewhere |> in the range of 60-70 Megs. Of course, to simply run R5 |> 'functional', then, |> yes, expect to use ~20 megs. Perhaps even a bit less if you only |> have one |> WM on call... I've got X11R5 running under Mach2.6 on my 386/40, and it's quite functional and used less than 30M. The major problem here is that the average X client using libX11.a is at least 400K (stripped) because everything must be statically linked. The corresponsing clients under SunOS are around 50-60K. If the X11R5 can be made to use dynamic libraries then the installed size should be reasonable. -- ======================================================================= Ken Hughes | "Which button do I press to turn (hughes@sol.csee.usf.edu) | it off?" "Try the red one, FT-Ph D candidate, PT-ex-sysadm | allright?" "Which red button?" Dept of Comp Sci and Eng | University of South Florida | .... _Heathers_ ======================================================================= ------------------------------ From: burgess@eniac.seas.upenn.edu (Richard Burgess) Subject: If not SCSI, why not floppy? Date: 21 Apr 92 06:33:29 GMT As new user, I have dutifully sifted through the mass of help files and assorted FAQs. The beginner's guide (and other guides) were enormously helpful in obtaining a floppy bootable system. Alas my system is st02 SCSI based. Any attempts to boot a floppy containing the proper SCSI kernal causes a kernal panic. The kernal correctly identifies my drive and finds SCSI ID 0 as my one (now MSDOS) partition. It then dies. Without floppy-booting a SCSI kernal, I am not sure how to get a SCSI-aware-Fdisk to repartition. Do I have to repartition with DOS Fdisk first? Am I missing something in all that documentation? Not to be defeated by a lack of hard-drive, I figured I would use my second floppy to access uemacs, mtools or anything - just as a test. This also failed. Floppy resets are plentiful. Given that my system is a clone with A:1.2M and B:1.44M shouldn't mount /dev/at1 /mnt or mount /dev/PS0 /mnt work to mount my B: drive? Does this have to do with the SCSI card-I doubt it. The drive light goes on but the mounting errors-out, leading me to believe that the os knows where to look (ie /dev/** works) but cannot figure out what it is it finds their. Thanks for any help Richard Burgess ========================================================= DOS 5, OS/2 2.0, WINDOWS 3.1, WIN32NT (soon!), and LINUX: life was much easier when DOS 3.3 was the only OS! ------------------------------ From: SAB139@psuvm.psu.edu Subject: Re: Kernel panic Date: 21 Apr 92 03:38:31 GMT Well, couldn't actually tell you what they mean. But, I would love to know as I got one when I firsted tried to use my hard drive as root under 0.95c+. Of course it was a different error and different numbers. It wasn't causing a kernel panic just not giving me a shell. If anyone is interested, I still have the EIP, EFLAGS, fs, etc... and the info from scroll-lock and control-shift-scroll-lock. -- Steven A. Bairstow * "There ain't no rules around here! InternNet : sab139@psuvm.psu.edu * We're trying to accomplish something!" or : bairstow@vivaldi.psu.edu * - Thomas Edison WWIV-Net : 216 @8450 * ------------------------------ From: hlu@yoda.eecs.wsu.edu (Hongjiu Lu -- Graduate Student) Subject: Re: 0.95c+ console and gcc 2.1 problem. Date: Tue, 21 Apr 92 06:23:50 GMT In article techs@triton.unm.edu (Erik Fichtner) writes: >Also having a problem with gcc 2.1. One of the things I'm intrested in is ray >tracing... I used to run DKBtrace under MS-DOS, and then i ran it under AT&T >SVR4 (my last enviornment. yuuuck). So, i went and got the source again and >tried to compile it under Linux. *KABOOM!* gcc got very angry and spit this >error message at me: > >gcc: Internal compiler error: program cc1 got fatal signal 6 >make: *** [parse.o] Error 1 > >Correct me if i'm wrong, but that's not supposed to happen, is it? :-) > > That may be bug in gcc 2.1 or the source code is not very portable. First thing I would do is add -Wall to CFLAGS in Makefile and try to fix ALL the error/warning messages by including the appropriate header files. If parse.c is generated from parse.y, I would use bison 1.16 or 1.15 (I heard something about 1.16 bugs.) Bison in 2.1shared-A.tar.Z is 1.16 and untested. You may want to generate parse.c from the other machines to see if they are the same. H.J. ------------------------------ From: it1@ra.msstate.edu (Tim Tsai) Crossposted-To: comp.os.os2.programmer Subject: generic scsi support in OS/2 Date: 21 Apr 92 07:44:53 GMT I asked this before in comp.os.os2.misc but got no answers, so I'm going to give it another shot. On OS/2 2.0 GA, IBM has built-in support for scsi controllers that intercepts INT13. Does anybody have comments on how this works? My guess is that OS/2 uses the virtual-8086 mode to call the controller's BIOS routines and pass data back to protected mode through some sort of share memory. While there probably is a degradation in performance, a much broader range of devices are now available. Thoughts? BTW, this is crossposted to comp.os.linux because I think support like this would get a lot more people to use Linux. We can still have controller-specific drivers that emphasis on performance, but those that have exotic controllers can at least get in the door. Tim ------------------------------ From: alm@netcom.com (Andrew Moore) Subject: cpio doesn't accept multi-volume archives Date: 21 Apr 92 04:16:05 GMT cpio will prompt for "next tape" but immediately dies upon pressing ENTER. I am using the command: cpio -itv ------------------------------ From: rpb0804@zeus.tamu.edu (BATES, ROBERT PATRICK) Subject: fmtflop in 0.95x...??? Date: 21 Apr 92 08:39:00 GMT I was just tooling through my S5R4 reference and was trying to figure out how to format a Unix floppy. I found a few pages concerning fmtflop, and tried it in Linux to format some disks to use on a Sun for FTP and the like. However, to my dismay, Linux spits it back up and says that there's no such critter. If it's there, where is it? If it's not, why not? Also, is the gcc2.ps (I hope I got that right :->) a PostScript version of the softdocs for gcc2? Just wondering so's I make sure I'm on a PS printer when I dump it... Thanks, Rob. ============================================================================== Robert Bates Disclaimer: rpb0804@zeus.tamu.edu "I make no claims..." Texas A & M University ============================================================================== ------------------------------ From: drew@kinglear.cs.colorado.edu (Drew Eckhardt) Subject: Re: If not SCSI, why not floppy? Date: Tue, 21 Apr 1992 09:43:17 GMT In article <75082@netnews.upenn.edu> burgess@eniac.seas.upenn.edu (Richard Burgess) writes: >As new user, I have dutifully sifted through the >mass of help files and assorted FAQs. The beginner's >guide (and other guides) were enormously helpful in >obtaining a floppy bootable system. BUT NOT THE SCSI FAQ. The guy running the SCSI project (me) gets lazy, and when rushed just takes a snapshot of his /vmunix file. I mount root from a NORMAL hard disk, major 3 minor 1. Obviously, if you don't have a minix filesystem at that location, you will get a kernel panic about being unable to read root. >Alas my system is st02 SCSI based. Any attempts to >boot a floppy containing the proper SCSI kernal >causes a kernal panic. The kernal correctly >identifies my drive and finds SCSI ID 0 as my one >(now MSDOS) partition. It then dies. Without >floppy-booting a SCSI kernal, I am not sure how to >get a SCSI-aware-Fdisk to repartition. Do I >have to repartition with DOS Fdisk first? Am I missing >something in all that documentation? > See above. The solution is to change the bytes at offset 0508 and 0509 in the boot image to 0's. This will rectify the problem. >work to mount my B: drive? >Does this have to do with the SCSI card-I doubt it. No. ------------------------------ ** FOR YOUR REFERENCE ** The service address, to which questions about the list itself and requests to be added to or deleted from it should be directed, is: Internet: Linux-Activists-Request@NEWS-DIGESTS.MIT.EDU You can send mail to the entire list (and comp.os.linux) via: Internet: Linux-Activists@NEWS-DIGESTS.MIT.EDU Linux may be obtained via one of these FTP sites: nic.funet.fi pub/OS/Linux tsx-11.mit.edu pub/linux tupac-amaru.informatik.rwth-aachen.de pub/msdos/replace The current version of Linux is 0.95a released on March 17, 1992 End of Linux-Activists Digest ******************************