From: Digestifier To: Linux-Development@senator-bedfellow.mit.edu Reply-To: Linux-Development@senator-bedfellow.mit.edu Date: Sun, 9 Oct 94 15:13:06 EDT Subject: Linux-Development Digest #283 Linux-Development Digest #283, Volume #2 Sun, 9 Oct 94 15:13:06 EDT Contents: Re: Telnet & ftp freeze! (Ralph Sims) Re: Linux For Mac (Peter N Lewis) 1.1.50 SIGNAL BUG! EMERGENCY (Davor Jadrijevic) Re: What GUI to write for? (Karl Keyte) Re: Improving SLIP latency under Linux (Frank Lofaro) Re: linux-activists@Niksula.hut.fi (Karl Keyte) Re: Multiprocessing Pentium Systems (Andrew Anderson) Re: Improving SLIP latency under Linux (Christopher Michael Joslyn) Re: Korn Shell '93 Now Available from AT&T (Chris Bitmead) Re: Flame on the attitude of Linux towards GCC development (Richard L. Goerwitz) Re: DOSEMU question (Rob Janssen) 1.1.52 bug? (Riku Saikkonen) Re: Linux killed my floppy drive! (Jeff Kesselman) ---------------------------------------------------------------------------- From: ralphs@halcyon.halcyon.com (Ralph Sims) Crossposted-To: comp.os.linux.help,comp.os.linux.misc,comp.os.linux.admin Subject: Re: Telnet & ftp freeze! Date: 7 Oct 1994 01:36:59 GMT coling@ivory.torolab.ibm.com (Colin Beckmann) writes: >: And what about those of us that DON'T see it? Basic setup is a >: dedicated PPP link on a 14.4 dialup, NET-3 stuff, ppd 2.1.2a, >: etc., with an InfoMagic/TransAmeritech CD-ROM combined install. >I am NOT seeing th problem, Have a 14.4 modem using NET-3 pppd 2.2.2a with >slackware 1.2 , and kernel 1.1.30. I have downloaded 20 and 30 megs in a >single session via ftp and never had a problem. I regularly rlogin to >other sites, once again without problem Well, the only thing I didn't add was 1.1.51 for the kernel. While nothing is locking up, there is a lag while telnetting with an FTP session going on that is annoying. I've received a patch that I hope to be able to roll in this weekend and do some serious testing. Other things that run are Sendmail+IDA as a daemon, xntpd, and CERN's web server. The ftp session definitely takes over the system. ------------------------------ From: peter.lewis@info.curtin.edu.au (Peter N Lewis) Subject: Re: Linux For Mac Date: Sun, 09 Oct 1994 11:56:11 +0800 >Henry Ware (hware@bronze.coil.com) wrote: >> Whats to tell? The Linux FAQ lists no 68k mac ports, GNU doesn't support >> Apple (because of Apple's "look and feel" lawsuits), and I haven't heard >> of any (except for the PowerMac port). What do the FSF's political goals have to do with it? GNU software is (by definiteion) free, and people can and do port it to any platform they feel like. Even by discouraging this the FSF flies in the face of it's own goals. In any event, I've certainly heard of both a Mac 68k and Mac PPC Linux port. I've got no idea on the status of either. The 68k one seems to be doing quite well (it was discussed in a recent Lunux Journal article), and the PPC one will almost certainly happen since future Power Mac's will most probably be designed with the new PPC hardware standards so that porting to a Power Mac will be equivalent to porting to most any PowerPC machine. I agree with Andi though, it'd be nice to have more frequent updates on the states of all the Linux ports. Perhaps a monthly posting listing the current status and contact points for all Linux ports? Peter. -- Peter N Lewis - Macintosh TCP fingerpainter FTP my programs from redback.cs.uwa.edu.au:Others/PeterLewis/ or amug.org:pub/peterlewis/ or nic.switch.ch:software/mac/peterlewis/ ------------------------------ From: davj@ds5000.irb.hr (Davor Jadrijevic) Subject: 1.1.50 SIGNAL BUG! EMERGENCY Date: 4 Oct 1994 23:10:50 GMT /* take linux-1.1.50, gcc 2.5.8 compile this program with $ cc -O6 -s -N bug.c bug.c: In function 'sig_hup': bug.c:26: warning: passing arg 2 of 'sigaction' from incompatible pointer type bug.c: In function 'main': bug.c:39: warning: passing arg 2 of 'sigaction' from incompatible pointer type $ get it running $ a.out and kill it: $ killall -HUP a.out 23:59:51 emard:~/src/so $ Unable to handle kernel paging request at virtual address 6ae58955 current->tss.cr3 = 003d5000, %cr3 = 003d5000 *pde = 00000000 Oops: 0000 EIP: 0010:0011773c EFLAGS: 00010206 eax: 0000002b ebx: 00000001 ecx: 00000000 edx: 6ae58955 esi: 0053304c edi: 00000001 ebp: 0045bfbc esp: 0045bf98 ds: 0018 es: 0018 fs: 002b gs: 002b ss: 0018 Process a.out (pid: 510, process nr: 18, stackpage=0045b000) Stack: 00000000 0003b328 bffffb10 bffffb10 00000001 Code: 64 f6 02 00 66 c7 45 34 23 00 66 c7 45 40 2b 00 66 c7 45 1c 0011703c T _sys_sigaction 0011715c T _sys_sigreturn 001172fc t _setup_frame 0011743c T _do_signal <----- this?? 001177b0 t ___gnu_compiled_c 001177b0 t _get_long 001177b0 t gcc2_compiled. Best regards, Davor */ #include #include /* FD_SET, timer stuff */ #include struct sigaction sa; void sig_hup(int sig) { sigaction(SIGHUP, sig_hup, NULL); } int main(int argc, char **argv) { fd_set rmask; struct timeval timeout, *event; sa.sa_handler = sig_hup; sa.sa_mask = 0*( (sigset_t)1 << (SIGHUP-1) ); sa.sa_flags = (0*SA_ONESHOT) | (0*SA_NOMASK) | (0*SA_INTERRUPT) | (0*SA_NOCLDSTOP); sa.sa_restorer = NULL; sigaction(SIGHUP, sig_hup, NULL); FD_ZERO(&rmask); FD_SET(0, &rmask); FD_SET(1, &rmask); select(2, &rmask, NULL, NULL, NULL); } -- , ================ Davor Jadrijevic ==================== ------------------------------ Date: Tue, 4 Oct 1994 15:40:23 +0100 From: kkeyte@esoc.bitnet (Karl Keyte) Reply-To: kkeyte@esoc.bitnet Subject: Re: What GUI to write for? Check out also the OI kit from ParcPlace - builds for OpenLook or Motif. You can select which toolkit look & feel you want at RUN-TIME! It's freely available for Linux and a new version is expected sometime following the stable releases of Linux (1.2), libraries (4.6) and gcc (2.6). XFree 3.1 is already out. Karl ========================================================================= Vitrociset S.p.A. Tel : +(49) 6151 902041 European Space Agency Fax : +(49) 6151 904041 64293 Darmstadt, Germany e-Mail: KKEYTE@ESOC.BITNET ------------------------------ From: ftlofaro@unlv.edu (Frank Lofaro) Subject: Re: Improving SLIP latency under Linux Date: Fri, 7 Oct 94 16:10:24 GMT In article <36uvav$1qo@news1.shell> dwm@shell.portal.com (David - Morris) writes: >callahan@maths.ox.ac.uk (Michael Callahan) writes: > > >>I personally don't see how this can be better than TOS queuing, >>without changing the SLIP or PPP protocols. With TOS, an interactive >>packet will always go out before any FTP packets that are waiting. > >Well, it is not always possible (and historically rarely possible) >to tell if a connection really deserved TOS queuing at high >priorities. A couple expamples ... a terminal session being used >to 'download' a large file; a terminal emulator with a programable >interface (TN3270 w/Chameleon for NETMANAG or PC/3270 from IBM) >performing batch bulk transactions, file downloads, etc. > >Dave Morris > How about giving packets below a certain size higher priority? That would help interactive traffic, and handle cases such as downloads though a terminal session properly. ------------------------------ Date: Tue, 4 Oct 1994 16:12:35 +0100 From: kkeyte@esoc.bitnet (Karl Keyte) Reply-To: kkeyte@esoc.bitnet Subject: Re: linux-activists@Niksula.hut.fi In article 58K@PE1CHL.AMPR.ORG, rob@pe1chl.ampr.org (Rob Janssen) writes: >But unfortunately you can't specify the address yourself. It blindly >takes the address it finds in the From: line. This loses badly when your >address has changed, or the translation of the From: address which happens >in the mail routers has changed. > >I have experienced already two times that the only way to unsubscribe was >to write a message to the operator :-( The change in mail routing is exactly the problem I have. I cannot unsubscribe. I have also tried mailing root, postmaster and operator, but to no avail. I think the system needs to be made a little more inteligent in its handling of addresses. How do I even SEE the addresses it THINKS it has for me? Karl ========================================================================= Vitrociset S.p.A. Tel : +(49) 6151 902041 European Space Agency Fax : +(49) 6151 904041 64293 Darmstadt, Germany e-Mail: KKEYTE@ESOC.BITNET ------------------------------ From: andrew@amelia.db.erau.edu (Andrew Anderson) Subject: Re: Multiprocessing Pentium Systems Date: 4 Oct 1994 23:21:36 GMT Just curious: Does anyone have an MP system that they can try running Linux on? Does it croak, or just not take advantage of the 2nd CPU? Also, I've been looking at the MP Pentiums, and I was wondering if I purchased a system with just 1 CPU and added a second later when Linux gets some kind of MP support if that would work. -- |===========================================================================| | Andrew Anderson andrew@db.erau.edu | | Novell Network System Administrator "Making the impossible | | Linux System Administrator possible -- daily!" | | | | I don't speak for ERAU, and God knows I don't want them to speak for me! | |===========================================================================| ------------------------------ From: chrisj@pvi.com (Christopher Michael Joslyn) Subject: Re: Improving SLIP latency under Linux Date: Tue, 4 Oct 1994 23:18:38 GMT In article <1994Oct4.185217.27344@inca.comlab.ox.ac.uk>, Michael Callahan wrote: >In article <1994Oct3.191439.12908@pvi.com>, >Christopher Michael Joslyn wrote: >>In article <36m9lc$6q0@agate.berkeley.edu>, >>Nick Kralevich wrote: >>> >>>Is there any way to improve/derease the latency associated with >>>SLIP under linux? Specifically, when I am ftping a large file, >>>I frequently get ping times of 6+ seconds. This murders interactive >>>traffic. I've tried setting my MTU to 256, but it doesn't make >>>any difference. >> >>The main problem is that an ftp packet is much larger than, say, a telnet or >>ping packet. Because the ftp packet is large, the packets must be broken up >>into sizes that can be sent over the line (this is your MTU at work), thus >>the ftp packet takes longer to send. Additionaly, an ftp connection typically >>is doing things in a batch where it is constantly pushing these large packets >>onto the queue. Factor all things in and what you get is a long queue filled >>mostly with ftp packets and a few, notably smaller, interactive packets. > >If you use recent client programs and a recent kernel, Linux PPP and >SLIP will send interactive packets before FTP ones. Recent clients >will set the IP TOS (type-of-service) field to favor interactive >traffic, and recent kernels know to send packets with "minimize >latency" TOS before anything else. They even know how to do this >for packets they are routing. IMHO, this sounds like a very simple way of doing things, however, I have the following problems with TOS: (a) Any user can build a copy of ftp that resets the packet TOS back to interactive, putting the network right back where to it started. (b) TOS only has two levels--interactive, and non-interactive. However, most network applications really should be considered at more levels than two. For example, I would order them as: telnet/r* should be first, NFS should be second, HTTP should be third, ..., and ftp should be last as long as it doesn't cause a timeout. (c) Users must upgrade all their, possibly customized, client applications for this to work. (d) Extensive kernel modification was needed. (e) The prioritization is static. I believe that different networks have different needs and may want to control their network traffic more than others. In all actuality, I wish I would have know about the TOS modifications before I started this patch Lars and I've been working on. I must have the older clients which don't support TOS. I'd really like to take a look at the performance differences. Has a package been created where I could grab one file and have all the latest client applications (e.g., which Slackware (version ?) file(s) do I need)? >So, if you set up a link with a small MTU, you should get OK performance. What do you consider small? I have a MTU of <=1006 which isn't too bad considering I can usually get 1.6 kb/s using ftp. >>>Theres was a thread a couple of months ago that said the problem >>>was in the kernel. However, there was never a solution posted. > >I don't remember the thread, but it is possible that it was before >the kernel did TOS queuing properly. I think that would have been >many months ago, though. Yeah, I missed it too. What version of the kernel (and client apps) started using TOS properly? >>Not really. It is a network problem. Lars Fenneberg (lf@gimli.comlink.de) >>and I are currently working on a package that would balance the load on a >>SLIP or PPP dial-up network connection. Unfortunantly, we are just getting >>started and will be a little bit before it is complete. The implementation >>at this time simple slices the bandwidth into equal sections so each >>connection has an equal bandwidth. What we are planning on adding is >>unequal sharing. Interactive connections would probably have more of the >>bandwidth since they usually transmit in bursts. The ftp connection would >>then be slowed during these bursts and then be back up to speed when nothing >>else is communicating. Of course, this kind of unequal sharing will be >>tunable at runtime. > >I personally don't see how this can be better than TOS queuing, >without changing the SLIP or PPP protocols. With TOS, an interactive >packet will always go out before any FTP packets that are waiting. >If the interactive traffic wants more bandwidth than is available, >there will be user-visible delays, no matter what your priority >scheme. If it doesn't want more bandwidth than there is, it will >get as much as it asks for, in preference to the bulk traffic. From a simple, single user, standard installation stand-point, I agree, there wouldn't be much difference. However, I'd like to have a way of dynamically changing the packet weights depending on the network traffic. >Actually, I guess there is one thing you could do. You could set >things up so that if an interactive packet gets queued while a bulk >packet is in the middle of transmission, you immediately interrupt >the bulk packet (by sending an end-of-frame character and relying >on the remote end to discard the incomplete frame) and start the >interactive one instead. Ugly, and I don't recommend it for SLIP >(which has no link error detection). It would improve latency >somewhat. Sounds like a nightmare of an operation. I have a hard time thinking of what could generate a packet so big that you would need to actually cut if off mid-transfer. >On the other hand, it's fine for people to implement whatever they >want. Whatever floats your boat. No, I really appreciate the comments. I helps us all, especially me. Thanks. - Chris -- ---- Purgamentum Init, Exit Purgamentum ------------------------------------ Christopher M. Joslyn | Visual Numerics, Inc. | "A la fin de l'envoi, +1 (303) 581-3269 | 6230 Lookout Road | je touche!" chrisj@boulder.vni.com | Boulder, CO 80301 | My opinions are mine! ------------------------------ From: chrisb@wombat.cssc-syd.tansu.com.au (Chris Bitmead) Subject: Re: Korn Shell '93 Now Available from AT&T Date: 4 Oct 94 15:24:37 In article <36higp$f69@superb.csc.ti.com> andy@tidmmpl.csc.ti.com (Andrew J. Piziali) writes: >In article <36f6eb$eov@psu_075.sb2.pdx.edu>, James E. McNalley > asked: > > What does ksh do that BASH can't do? > >How about coprocess communication (print/read -p ...), the "select" statement, Recent versions of bash have "select" >builtin arithmetic, bash has builtin arithmetic >a command line editor which will drop you into vi/emacs, bash has this. >and parameter attributes (typeset -LZ ...), to name a few? Don't know about this. ------------------------------ Crossposted-To: comp.os.linux.help,comp.os.linux.misc From: goer@ellis.uchicago.edu (Richard L. Goerwitz) Subject: Re: Flame on the attitude of Linux towards GCC development Reply-To: goer@midway.uchicago.edu Date: Fri, 7 Oct 1994 20:24:30 GMT xjzhu@math.uwaterloo.ca (Xiaojun Zhu) writes: > >Think about the scenario, when linus announced his first release of >linux and no body wanted to try that out? You know the answer, >don't you? I think if everyone beta tested every piece of software to come out, no one would every get any work done. We all select the stuff that is most important to us, and try to help out as best we can. I'm currently working hard to help get kinks out of DOSEMU, which I think is a worthwhile project. It takes a lot of time, and I have to weed through a lot of debugging output and ftp patches every couple of days (plus the ones the authors want me to try out). I simply could not do this for every piece of software. -- -Richard L. Goerwitz goer%midway@uchicago.bitnet goer@midway.uchicago.edu rutgers!oddjob!ellis!goer ------------------------------ From: rob@pe1chl.ampr.org (Rob Janssen) Subject: Re: DOSEMU question Reply-To: pe1chl@rabo.nl Date: Tue, 4 Oct 1994 22:30:32 GMT In <1994Oct4.152636.21456@alw.nih.gov> drankin@myhost.subdomain.domain (Douglas Rankin) writes: > I have a question. I have been able to get dosemu0.53 running with a >diskiamge for c: I was wondering how I configure it in order for it >to boot my dos drives which is /dev/sda. My linux is on /dev/sdb1. >Any help would be appreciated. You can reply here or email me at the >following: > drankin@mason1.gmu.edu or > drankin@dgs.dgsys.com >thanks for any help!!!! Your best bet is to move the diskimage to a:, boot from that, and then lredir the /dev/sda disk to c: Booting from diskimage in a: is simple: bootA # Startup drive valid values: bootA bootC # Use the following option to boot from the specified file, and then # once booted, have bootoff execute in autoexec.bat. Thanks Ted :-). # Notice it follows a typical floppy spec. To create this file use # dd if=/dev/fd0 of=/var/lib/dosemu/bdisk bs=16k # bootdisk { heads 2 sectors 9 tracks 80 threeinch file /var/lib/dosemu/bdisk } You can still have both floppies available: floppy { device /dev/fd0 threeinch } floppy { device /dev/fd1 fiveinch } Then, in the config.sys on the image, put something like: files=30 buffers=5 lastdrive=h stacks=0,0 dos=high,umb device=a:\setver.exe devicehigh=a:\emufs.sys /dos devicehigh=c:\dosemu\ems.sys shell=c:\command.com /e:512 /p In the autoexec.bat on the image, put something like: @echo off prompt $p$g c: path c:\dosemu;c:\bin;c:\dos set comspec=c:\command.com set dircmd=/o/l set lib=c:\lib set tz=MET0 lh c:\dos\doskey.com >NUL lredir d: LINUX\FS\cdrom R >NUL lredir e: LINUX\FS\ R >NUL lredir c:\dosemu\bootoff.bat Finally, you have the file \dosemu\bootoff.bat on the DOS disk: bootoff This does the trick. C: is now a redirected drive, and A: and B: are the floppies again. Rob -- ========================================================================= | Rob Janssen | AMPRnet: rob@pe1chl.ampr.org | | e-mail: pe1chl@rabo.nl | AX.25 BBS: PE1CHL@PI8UTR.#UTR.NLD.EU | ========================================================================= ------------------------------ Subject: 1.1.52 bug? From: riku.saikkonen@compart.fi (Riku Saikkonen) Date: Sun, 9 Oct 94 17:16:00 +0200 Hmm... Linux crashes on me, seemingly randomly. Nothing works after the crash, not virtual console switching, not Ctrl-Alt-Del. 'Unable to handle kernel paging request' and the normal register dump appears in the log, for two or three processes. The EIP is in a totally different location each time... So, it seems that it's the 1.1.52 kernel which is having some problems. I noticed some modification of memory.c in the .52 patch; perhaps that's the culprit? Anyone else seen the same problems? I can't be really sure if this is a kernel bug, because I just did a quite heavy upgrade of the machine (switched to a VLB motherboard + display adapter + IDE host adapter). That's why I'm asking... Hmm, now I compiled 1.1.50, and it seems to work. Except for one crash in X, but that is probably something else (no kernel panic or anything in the log). So I think it is a bug in 1.1.52... -=- Rjs -=- riku.saikkonen@compart.fi - IRC: Rjs "From cavern pale the moist moon eyes / the white mists that from earth arise / to hide the morrow's sun and drip / all the grey day from each twig's tip." - J. R. R. Tolkien ------------------------------ Crossposted-To: comp.os.linux.help From: jeffpk@netcom.com (Jeff Kesselman) Subject: Re: Linux killed my floppy drive! Date: Tue, 4 Oct 1994 05:33:47 GMT In article , Ahmed Naas wrote: > > >So, did Linux kill my drive or is this one of those rare coincidences? > >-- Well, I've nevre heard of software killing a floppy drive, and I boot from my floppy and have created multiple boot floppies without trouble. Sounds like coincidence to me, but I could be wrong. Jeff Kesselman ------------------------------ ** 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-Request@NEWS-DIGESTS.MIT.EDU You can send mail to the entire list (and comp.os.linux.development) via: Internet: Linux-Development@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 Digest ****************************** .