Received: from PACIFIC-CARRIER-ANNEX.MIT.EDU by po7.MIT.EDU (5.61/4.7) id AA23612; Sat, 9 Dec 95 15:31:12 EST Received: from SENATOR-BEDFELLOW.MIT.EDU by MIT.EDU with SMTP id AA29219; Sat, 9 Dec 95 15:31:17 EST Received: (from root@localhost) by senator-bedfellow.MIT.EDU (8.6.12/2.3JIK) id PAA25454 for Linux-Development-System-Dist@senator-bedfellow.mit.edu; Sat, 9 Dec 1995 15:14:14 -0500 Message-Id: <199512092014.PAA25454@senator-bedfellow.MIT.EDU> From: Digestifier To: Linux-Development-System@senator-bedfellow.MIT.EDU Reply-To: Linux-Development-System@senator-bedfellow.MIT.EDU Date: Sat, 9 Dec 95 15:14:10 EST Subject: Linux-Development-System Digest #99 Linux-Development-System Digest #99, Volume #2 Sat, 9 Dec 95 15:14:10 EST Contents: Re: Status of kernel list (Uwe Bonnes) Re: Help for DEC Tulip w/1.3.XX (John Burton) Re: Priorities in 1.3.45? (Alex Butcher) Re: Swap-space on diskless client (Stephane Billiart) Re: rlogin lookalike (Miquel van Smoorenburg) Re: Priorities in 1.3.45? (Larry Daffner) Re: ELF: Can Only Build Static Binaries! (Steven M. Harrington) Re: Limiting diskcache/buffers? (Perry F Nguyen) Re: Will Linux work with the P6 ??? (Matthias Urlichs) Re: Latest Stable Kernel ??? (Matthias Urlichs) Re: reading lp port.. (Matthias Urlichs) Errors with 1.3.45 (John Meacham) device driver mail groups (Tom Bjorkholm) Re: Library version number bugs (A. Rohde) Re: ELF: Can Only Build Static Binaries! (Markus Schoder) ---------------------------------------------------------------------------- From: bon@elektron.ikp.physik.th-darmstadt.de (Uwe Bonnes) Subject: Re: Status of kernel list Date: 7 Dec 1995 12:23:38 GMT Frank Racis (FWR100@psuvm.psu.edu) wrote: : I haven't gotten any mail from the kernel list in about a week. : Is it down, has everyone been quiet, or have I been randomly : unsubscribed? : Check with a mail to Majordome@vger.rutgers.edu containing which end if you are still subscribed. (The list was scrapped about a week ago). If not, resubscribe, possibly to "linux-kernel-digest". With the digest you get only one letter per day and so releaf vger from sendings zillions of single mails. -- Uwe Bonnes bon@elektron.ikp.physik.th-darmstadt.de Institut fuer Kernphysik Schlossgartenstrasse 9 64289 Darmstadt ========= Tel. 06151 162516 ======== Fax. 06151 164321 ========== ------------------------------ From: john@ns.gats.hampton.va.us (John Burton) Subject: Re: Help for DEC Tulip w/1.3.XX Date: 08 Dec 1995 13:43:03 GMT In article <4a5v8r$i2o@cocoa.brown.edu> peterv@foo.net (Peter J. Vessenes) writes: I just compiled the tulip.c v0.10 with Linux 1.3.45 kernel distribution (under elf), but it took a small kludge -- the tulip.c file includes the etherdevice.h file, but the etherdevice.h file has a different definition for init_etherdev than tulip.c wants. I just copied etherdevice.h in /usr/include/linux to /usr/include/linux/\ etherdevice-2.h and commented out the init_etherdev line in etherdevice-2.h. I then update the include to point to etherdevice-2.h, and it compiled. Ummm...I had thought of this approach, but discarded it due to one small problem...The prototype for init_etherdev in etherdevice.h has two parameters, the routine init_etherdev as defined in net_init.c has two parameters that match the prototype. The prototype for init_etherdev in tulip.c v0.10 has 3 parameters. The calls to init_etherdev in tulip.c v0.10 use three parameters. Being unfamiliar with any of these routines, I could not say which parameter was unnecessary, and also from a program correctness stand point I did not like the idea of passing 3 parameters to a routine that only expects 2. On the other hand, how does it perform ? John -- John Burton GATS, Inc. j.c.burton@gats.hampton.va.us 28 Research Drive j.c.burton@larc.nasa.gov Hampton, VA 23666 (804) 865-7491 (voice) (804) 865-1021 (fax) ------------------------------ From: Alex.Butcher@bris.ac.uk (Alex Butcher) Subject: Re: Priorities in 1.3.45? Date: Fri, 8 Dec 1995 12:52:27 GMT Hiya ldaffner@convex.com (Larry Daffner) wrote: >In <4a5hsa$j2m@questor.org> sp@questor.org writes: >>In article , >>Christopher Neufeld wrote: >Getting >the procps-0.69-1.3.39.tgz package solves the problem (If I remember >right it's under /pub/Linux/system/mic/ps or something like that on >sunsite) /pub/Linux/system/Status/ps/procps-0.97-1.3.39.tar.gz I also read something about having to recompile libc too :C, but I'll give procps a try first. :) >HTH >-Larry Regards, Alex. -- Alex Butcher - Micro Support Technician Tel +44 (0)117 928 9000 x3038 Computing Service, University of Bristol, Tyndall Ave. Bristol BS8 1UD http://www.cse.bris.ac.uk/~ccajb/Welcome.html #include ------------------------------ From: billiart@irisa.fr (Stephane Billiart) Subject: Re: Swap-space on diskless client Date: 8 Dec 1995 14:16:11 GMT Reply-To: billiart@irisa.fr In article <4a4cq1$73b@bmtlh10.bnr.ca>, sjuneau@bnr.ca (Steve Juneau) writes: >Rob Janssen (rob@pe1chl.ampr.org) wrote: >: In <4a21pn$1jd@fbi-news.Informatik.Uni-Dortmund.DE> loebbing@turing.informatik.uni-dortmund.de (Martin Loebbing) writes: > >: Linux can't swap over the network, or at least I don't think it can. > >: You will need to use a local diskdrive with a swap partition, or install >: enough memory to run without swapping if you don't have a local disk. > A few weeks ago, I saw an article in comp.os.research related to a rswap patch for linux to provide a remote swap facility. It has been developped in a spanish (or was it portugese) university, and is designed for swapping over the network between two linux boxes running 1.1.0 kernel. Unfortunately, I deleted this article and forgot the URL for downloading... Stephane -- "We can all survive and get along | Stephane Billiart if we can add one and one and get two" | IRISA - projet SOLIDOR David Goodis | 35042 RENNES CEDEX - FRANCE ------------------------------ From: miquels@drinkel.ow.org (Miquel van Smoorenburg) Crossposted-To: comp.os.linux.networking,comp.mail.uucp Subject: Re: rlogin lookalike Date: Sat, 9 Dec 1995 16:31:05 +0100 (MET) In article <4a6du0$ajs@work.smurf.noris.de>, Matthias Urlichs wrote: >In comp.os.linux.development.system, article <49vre4$2qh@ntserver.prolution.iaf.nl>, > renee@ntserver.prolution.iaf.nl (Renee Teunissen) writes: >> >> Data transfers from the uucp to the client are just fine we get about >> 7400cps over a 64kbit connection) but tranfers from the client to >> the uucp machine goes very and very slow (about 500cps!!). it seems >> that rlogin sends every byte it recieves over the serialline in a single >> ethernet frame. >You might want to verify that with tcpdump. It is true, we had the same problem. rlogin does read()s from its input with size 1, to reckognize the escape character (usually a `~'). Since for purposes like these you usually use "rlogin -E -8" (8 bit clean, no escape character) anyway, that isn't nessecary at all and rlogin can just do a read(0, buf, 4096) if it wants to. The fixed (BSD 4.4) version we use at Cistron is available on ftp://ftp.cistron.nl/pub/People/miquels/ It has only been tested under Linux. I have also rewritten the terminal handling code to use Posix termios instead of the BSD sgtty nightmare. Anybody knows how to contribute these enhancements back to BSD ? Mike. -- + Miquel van Smoorenburg + Cistron Internet Services + Living is a | | miquels@cistron.nl (SP5) | Independent Dutch ISP | horizontal | + miquels@drinkel.ow.org + http://www.cistron.nl/ + fall + ------------------------------ From: ldaffner@convex.com (Larry Daffner) Subject: Re: Priorities in 1.3.45? Date: 7 Dec 1995 11:11:17 -0600 In <4a5hsa$j2m@questor.org> sp@questor.org writes: >In article , >Christopher Neufeld wrote: >> I just realized that the upgrade to v1.3.45 seems to have had an >>interesting side-effect on my pentium box. All processes are now running >>at nice level 15, and I can't renice them to anything else. I rebooted >>with the old 1.3.34 kernel, and the problem went away. Am I the only one >>who had this happen? >Same here on an 80486, but if I tell "top" to "renice" a process showing >nice level 15, to nice level 20, I show it running at level -4. The values for priorities were messed around with in 1.3.39. Getting the procps-0.69-1.3.39.tgz package solves the problem (If I remember right it's under /pub/Linux/system/mic/ps or something like that on sunsite) HTH -Larry -- Larry Daffner - Software Engineer | email: ldaffner@convex.com | Convex Computer Corporation | tel: (214)497-4274 / home: (214)380-4382 | Ray's Rule of Precision: Measure with a micrometer. Mark with chalk. Cut with an axe. ------------------------------ From: harringt@Rainier.eskimo.com (Steven M. Harrington) Subject: Re: ELF: Can Only Build Static Binaries! Reply-To: harringt@eskimo.com Date: Sat, 9 Dec 1995 17:37:58 GMT In article <4a54hq$7c7@holly.aa.net>, Thomas Boutell wrote: >Hello, > >When I compile new programs, the linker seems to totally ignore >shared libraries and build static ELF binaries only. "Hello World" >is 80K. ldd confirms that the binaries I make are all statically linked >ELF binaries. This is Not A Good Thing. Please help. > >Existing ELF binaries find their shared libraries just fine. > The first reasonable guess is: check to make sure that all elf libs have links like: libc.so -> libc.so.5.x.x.x and not like: libc.so.5 -> libc.so.5.x.x.x (Yes, its in HJ's documentation and no, I didn't set it up right the first time). -- Steve harringt@eskimo.com ------------------------------ Crossposted-To: comp.os.linux.misc From: pfnguyen@Viet.Viet.Com (Perry F Nguyen) Subject: Re: Limiting diskcache/buffers? Date: Sat, 9 Dec 1995 16:03:57 GMT In article , >Well... Yes, that is probably a good idea, but I still dont want it to >have more than 5 megs of disk buffers. In my opinion that is never >needed, and there is no reason for it to grow to more than that. Is there >any chance of this being an option in a future kernel? Disk buffers are dynamically allocated as your run the system, it can be almost all of your ram, to 0 of your ram, depending on when your system needs it. So, just let your kernel handle the buffer size, it knows what the best size to use is. -- -- pub 1024/0D97E00D 1995/01/01 Perry "Huy" Francis Nguyen Key fingerprint = CE 62 F2 01 33 87 9D 89 BC 53 8D 11 F9 A0 DE 8F - FTP ftp://ftp.netcom.com/pub/pf/pfn FTP or finger pfnguyen@netcom.com for PGP Public Key. ------------------------------ From: smurf@smurf.noris.de (Matthias Urlichs) Subject: Re: Will Linux work with the P6 ??? Date: 7 Dec 1995 10:51:23 +0100 In comp.os.linux.development.system, article <30C373B0.58357802@mmm.com>, jdsundberg@mmm.com (John D. Sundberg) writes: > Yep > > p6:~> cat /proc/cpuinfo > cpu : 686 > model : PP"PJPPzPrP#PPHT$LD$ > tV t/ > t+ > t6DRQYRQRQRQUWVS\$0|$< > mask : Unknown Bah. Somebody forgot a range check here. :-( > BogoMips : 132.88 > Neat. Which CPU speed is that? 133 MHz? 200? -- Matthias Urlichs \ XLink-POP N|rnberg | EMail: urlichs@smurf.noris.de Schleiermacherstra_e 12 \ Unix+Linux+Mac | Phone: ...please use email. 90491 N|rnberg (Germany) \ Consulting+Networking+Programming+etc'ing 42 PGP: 1B 89 E2 1C 43 EA 80 44 15 D2 29 CF C6 C7 E0 DE Click here. ------------------------------ From: smurf@smurf.noris.de (Matthias Urlichs) Subject: Re: Latest Stable Kernel ??? Date: 7 Dec 1995 10:54:02 +0100 In comp.os.linux.development.system, article <49vh2d$6v5@kelly.teleport.com>, seano@teleport.com (Sean O'Halloran) writes: > The same sort of > consolidations appears to occur as 1.3.45 is approached - so I > would ??guess?? that 1.3.45 should be a pretty stable release as well .. > Linus noted that he wanted to do a 1.3.46 before leaving for Japan, but no such luck. He'll be back on the 10th or so, so expect a new kernel then. ;-) -- Matthias Urlichs \ XLink-POP N|rnberg | EMail: urlichs@smurf.noris.de Schleiermacherstra_e 12 \ Unix+Linux+Mac | Phone: ...please use email. 90491 N|rnberg (Germany) \ Consulting+Networking+Programming+etc'ing 42 PGP: 1B 89 E2 1C 43 EA 80 44 15 D2 29 CF C6 C7 E0 DE Click here. ------------------------------ From: smurf@work.smurf.noris.de (Matthias Urlichs) Subject: Re: reading lp port.. Date: 7 Dec 1995 11:09:16 +0100 In comp.os.linux.development.system, article , pe1chl@wab-tis.rabobank.nl writes: > > The LP driver on Linux does not support reading the printer port. > How exactly does DOS support reading them? I don't think I have ever > seen that. It's possible, though there seem to be two or three incompatible ways of doing that. I'd love that capability too -- cu -l /dev/lpt1 and interactively talk to our Postscript printer... right now the beast prints an error page when it sees a problem; I'd much rather get mail with the error message from the printer daemon. (Yes, that'd mean hacking lpd. So be it. I'll do it if/when somebody else enhances the kernel driver. :-) -- Matthias Urlichs \ XLink-POP N|rnberg | EMail: urlichs@smurf.noris.de Schleiermacherstra_e 12 \ Unix+Linux+Mac | Phone: ...please use email. 90491 N|rnberg (Germany) \ Consulting+Networking+Programming+etc'ing 42 PGP: 1B 89 E2 1C 43 EA 80 44 15 D2 29 CF C6 C7 E0 DE Click here. ------------------------------ From: John Meacham Subject: Errors with 1.3.45 Date: Thu, 07 Dec 1995 23:24:16 -0800 I have been running 1.3.45 for a while now and am very satisfied with it except for one major problem, every now again (about one every two days non-stop running, my ethernet connection dies and i get this message repeated over and over... eth0: Infinite loop in interrupt, status 2013. eth0: Infinite loop in interrupt, status 2013. eth0: Infinite loop in interrupt, status 2013. eth0: Infinite loop in interrupt, status 2013. eth0: Infinite loop in interrupt, status 2013. eth0: Infinite loop in interrupt, status 2013. eth0: Infinite loop in interrupt, status 2013. eth0: Infinite loop in interrupt, status 2013. eth0: Infinite loop in interrupt, status 2013. eth0: Infinite loop in interrupt, status 2013. I have a 3Com 3X9 ethernet card and am running a basic slack 3.0 system (ELF) except for the newer kernel. It is a 486 dx2 66 VLB motherboard IDE, SB16 and GUS sound cards..... any help would be apreciated... John ------------------------------ From: tomb@tomb.mydata.se (Tom Bjorkholm) Subject: device driver mail groups Date: 9 Dec 1995 17:20:17 GMT My idea is this: Let's form a few groups of Linux device driver writers. These groups should be fairly small 8 - 20 persons in a group. Each group will have its own mailing list. In this way you can easily ask people involved in similar activities for advice, without flooding the comp.os.linux.* news groups. Background: I have just started to port a set of device drivers from SVR3 to Linux. I am sure that I am not the only one that is doing this kind of driver porting, and I also know that there is a lot of people out there writing new device drivers for Linux. There is some documentation about how to write Linux device drivers (Kernel Hacker's Guide, etc) but not very much. You have got to read the kernel source to find out what is there for you to use. I think that everyone that is writing device drivers for Linux (except Linus :-) has got a few questions they would like to ask. (Is there a kernel function to perform this task? Has anyone got a smart solution for foo?) Action: Mail me! If there is enough interest the groups are formed, and I can arrange the mailing lists. Best, Tom -- =============================================================== Tom Bjorkholm MYDATA automation AB tel: +46 8 629 09 00 Karlsbodavagen 39 fax: +46 8 629 09 09 S-161 70 Bromma, Sweden ------------------------------ From: exp109@modcomp.physik.uni-kiel.de (A. Rohde) Subject: Re: Library version number bugs Date: 8 Dec 1995 12:19:58 GMT Reply-To: rohde@physik.uni-kiel.de Patrick hit the head of the nail..... I think, that there should be a possiblity to tell the programs, which library it has to load. For example: if I compile the jpeg library version 6 and call it libjpeg.so.6.0.0 and somebody else compiles absolutely the same library but calls it ...so.2.0.0, the binary will refuse to load it (major version number to small). What we need to get around this trouble, is a small program to patch the binary to load libjpeg.so.6.0 instead. If the libraries are equal, it will work, if not, it *perhaps* won't (who cares or wonders in this case?) A similar utility exist for go32 under dos. Axel P.S.: I use Patricks Slackware 3.0 (thank you VERY much!)..... Bug reports (hopefully constructive): -Elf has perhaps broken gdb, it is unable to read core dumps. I saw it yesterday, no idea. -xxgdb prints each character twice. 1.12 compiles fine and runs. -Bash 1.14.5 does not like to work with make and f2c's f77 command. 1.14.2 does. the 'exit 4' bug was introduced in 1.14.3 I remember. -Patrick, you please write *explicitely*, that ash is not able to replace bash as /bin/sh, since something is going wrong during the boot process with the rc.* files. The disks are not mounted rw. rc.*-scripts compatible with ash or execurted by #/bin/bash instead of #/bin/sh solve this problem. A *small* and fast /bin/sh is a good idea (IMO). -six or seven times the mtools died, and locked the floppy drive. It never happended before. I don't know what happens. The mtools processes can not be killed. No idea. I recompiled the with less optimization, it still happens sometimes, but seldomly. -Please install a resource file for xterm, termcap and terminfo entries/files for 'linux' and 'xterm', that are compatible with each other. I recommend to add this to /usr/lib/X11/app-defaults/XTerm: *vt100.translations: #override\n\ NoneHome: string("\033[1~")\n\ NoneEnd: string("\033[4~")\n\ NoneDelete: string("\033[3~")\n\ NoneBackSpace: string("\177")\n (I'm sitting in front of a Sun, therefore I can't post my modified termcap and terminfo, but if you like to have them, next week.....) I have patched rxvt 2.10 to be able to read a keymapping. Do you like to have it? -As we can see in this file, Patrick is going to fix the # of lines bug in ncurses. In article <4a84au$ac4@nkosi.well.com>, gonzo@magnet.mednet.net (Patrick J. Volkerding) writes: |> I've noticed what I think are serious problems in the selection of version |> numbers for shared libraries under Linux. Allow me to provide a couple of |> examples: |> |> 1. Linux libc-5.2.16 has been made public. The GNU gdbm and db libraries |> are no longer part of this release, and we've been pointed to what are |> supposed to be the new versions. However, both of these libraries have |> lower version numbers (*.so.1 instead of *.so.2) than the ones in |> libc-5.0.9. This effectively breaks all existing apps linked with the |> versions from the last public release. |> |> What happened? ELF was supposed to save us from these kinds of |> problems, and would have here if only a library version number had been |> correctly chosen. Note that the libraries now follow the version number |> of the package e.g. gdbm-1.7.3 = libgdbm.so.1.7.3. Previously 2.0.0 |> had been used as the so version number... probably a bad idea, but |> IMHO it represented some kind of commitment to stick with the numbering |> that had been decided upon. |> |> 2. An even crazier example: The first public release of shared ncurses |> libraries was just made on tsx-11.mit.edu in /pub/linux/packages/GCC. |> The links for the shared libraries have a totally different version |> number than the libraries, e.g.: |> |> -rw-r--r-- 226650 /usr/lib/libncurses.so.1.9.7a |> lrwxrwxrwx 0 /usr/lib/libncurses.so.2.1 -> libncurses.so.1.9.7a |> |> This is the expected name. If the link is changed to libncurses.so.1.9 |> or libncurses.so.1 (the expected name, IMO) it won't work. Resulting |> binaries will be expecting to find libncurses.so.2.1 and therefore won't |> start. |> |> In addition, if these links don't exist they are not correctly made by |> ldconfig. |> |> Comments anyone? |> |> Pat |> ------------------------------ From: schoder@rbg.informatik.th-darmstadt.de (Markus Schoder) Subject: Re: ELF: Can Only Build Static Binaries! Date: 8 Dec 1995 11:04:07 GMT Thomas Boutell (@boutell.com) wrote: : When I compile new programs, the linker seems to totally ignore : shared libraries and build static ELF binaries only. "Hello World" : is 80K. ldd confirms that the binaries I make are all statically linked : ELF binaries. This is Not A Good Thing. Please help. : Existing ELF binaries find their shared libraries just fine. : Here's what my system looks like: : libc 5.2.16 : gcc 2.7.0 : ld.so 1.7.9 : binutils: : This is what ld -v says: : ld version cygnus/linux-2.5.2l.17 (with BFD cygnus/linux-2.5.2l.11) : X11R6 (ELF) : Help! Any reasonable guesses as to what may be wrong with my system? Here is what the libc-5.2.16 release file states as requirements: SYSTEM REQUIREMENTS: * kernel 1.1.92 or above. It may work with an older kernel if the QMAGIC format is supported. If you use the kernel 1.3.x, you should upgrade to 1.3.40 or above. Otherwise readv/writev system calls won't work right. * gcc-2.7.2 or above and binutils 2.6.0.2 or above. * ld.so-1.7.10 or above. * libg++ 2.7.1.3 or above. This is only necessary for development using c++. Plus libc-5.2.16 breaks gnu make (but it can be fixed). I think you should do some serious upgrading and of course read the release file. Its on sunsite. Hope this helps Markus ------------------------------ ** 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-Development-System-Request@NEWS-DIGESTS.MIT.EDU You can send mail to the entire list (and comp.os.linux.development.system) via: Internet: Linux-Development-System@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 sunsite.unc.edu pub/Linux End of Linux-Development-System Digest ******************************