Re: xen, guest won't start, Hotplug scripts not working
On Sun, Apr 26, 2009 at 10:54:10AM +0200, Javier Barroso wrote: On Sun, Apr 26, 2009 at 10:18 AM, gcr...@vcn.bc.ca wrote: # mount [irrelevant stuff deleted] /dev/mapper/Bun_vg-mabelleSys_lv on /mnt type ext3 (rw) Having partition mounted is not a good idea for xen machines, umount it ant try again. Latter you can see at /var/log/xend.log , it could give you any hints The way I presented that wasn't clear. Actually I only mounted the device afterwards so I could show the contents of fstab on the guest. When I tried to start the guest, the partition wasn't mounted. this is trying to start the guest machine dumps to /var/log/xen/xend.log begin log exterpt [2009-04-26 02:22:25 2162] DEBUG (XendDomainInfo:84) XendDomainInfo.create(['vm', ['name', 'mabelle_vm'], ['memory', 128], ['vcpus', 1], ['on_xend_start','ignore'], ['on_xend_stop', 'ignore'], ['image', ['linux', ['kernel','/boot/vmlinuz-2.6.26-2-xen-686'], ['ramdisk', '/boot/initrd.img-2.6.26-2-xen-686'],['root', '/dev/hda1 ro'], ['args', '2']]], ['device', ['vbd', ['uname', 'phy:/dev/Bun_vg/mabelleSys_lv'], ['dev', 'hda1'], ['mode', 'w']]], ['device', ['vbd',['uname', 'phy:/dev/Bun_vg/mabelleSwap_lv'], ['dev', 'hda2'], ['mode', 'w') [2009-04-26 02:22:25 2162] DEBUG (XendDomainInfo:1618)XendDomainInfo.constructDomain [2009-04-26 02:22:25 2162] DEBUG (balloon:132) Balloon: 132032 KiB free; need 2048; done. [2009-04-26 02:22:25 2162] DEBUG (XendDomain:443) Adding Domain: 13 [2009-04-26 02:22:25 2162] DEBUG (XendDomainInfo:1703) XendDomainInfo.initDomain: 13 256 [2009-04-26 02:22:25 2162] DEBUG (XendDomainInfo:1738) _initDomain:shadow_memory=0x0, memory_static_max=0x800, memory_static_min=0x0. [2009-04-26 02:22:25 2162] DEBUG (balloon:132) Balloon: 132032 KiB free;need 131072; done. [2009-04-26 02:22:25 2162] INFO (image:139) buildDomain os=linux dom=13 vcpus=1 [2009-04-26 02:22:25 2162] DEBUG (image:363) domid = 13 [2009-04-26 02:22:25 2162] DEBUG (image:364) memsize= 128 [2009-04-26 02:22:25 2162] DEBUG (image:365) image = /boot/vmlinuz-2.6.26-2-xen-686 [2009-04-26 02:22:25 2162] DEBUG (image:366) store_evtchn = 1 [2009-04-26 02:22:25 2162] DEBUG (image:367) console_evtchn = 2 [2009-04-26 02:22:25 2162] DEBUG (image:368) cmdline= root=/dev/hda1 ro 2 [2009-04-26 02:22:25 2162] DEBUG (image:369) ramdisk= /boot/initrd.img-2.6.26-2-xen-686 [2009-04-26 02:22:25 2162] DEBUG (image:370) vcpus = 1 [2009-04-26 02:22:25 2162] DEBUG (image:371) features = [2009-04-26 02:22:26 2162] INFO (XendDomainInfo:1514) createDevice: vbd :{'uuid': '5881dca0-bef6-edc9-71a0-1a0c23a0a003', 'bootable': 1, 'driver':'paravirtualised', 'dev': 'hda1', 'uname': ' phy:/dev/Bun_vg/mabelleSys_lv', 'mode': 'w'} [2009-04-26 02:22:26 2162] DEBUG (DevController:118) DevController: writing {'virtual-device': '769', 'device-type': 'disk', 'protocol': 'x86_32-abi', 'backend-id': '0', 'state': '1', 'backend': '/local/domain/0/backend/vbd/13/769'} to /local/domain/13/device/vbd/769. [2009-04-26 02:22:26 2162] DEBUG (DevController:120) DevController: writing {'domain': 'mabelle_vm', 'frontend': '/local/domain/13/device/vbd/769', 'uuid': '5881dca0-bef6-edc9-71a0-1a0c23a0a003', 'dev': 'hda1', 'state': '1', 'params':'/dev/Bun_vg/mabelleSys_lv', 'mode': 'w', 'online': '1', 'frontend-id': '13','type': ' phy'} to /local/domain/0/backend/vbd/13/769. [2009-04-26 02:22:26 2162] INFO (XendDomainInfo:1514) createDevice: vbd : {'uuid': '35406b91-040e-68f4-5112-0719676b9cb9', 'bootable': 0, 'driver': 'paravirtualised', 'dev': 'hda2', 'uname': 'phy:/dev/Bun_vg/mabelleSwap_lv', 'mode': 'w'} [2009-04-26 02:22:26 2162] DEBUG (DevController:118) DevController: writing {'virtual-device': '770', 'device-type': 'disk', 'protocol': 'x86_32-abi','backend-id': '0', 'state': '1', 'backend': '/local/domain/0/backend/vbd/13/770'} to /local/domain/13/device/vbd/770. [2009-04-26 02:22:26 2162] DEBUG (DevController:120) DevController: writing {'domain': 'mabelle_vm', 'frontend': '/local/domain/13/device/vbd/770', 'uuid':'35406b91-040e-68f4-5112-0719676b9cb9', 'dev': 'hda2', 'state': '1', 'params': '/dev/Bun_vg/mabelleSwap_lv', 'mode': 'w', 'online': '1', 'frontend-id': '13','type': 'phy'} to /local/domain/0/backend/vbd/13/770. [2009-04-26 02:22:26 2162] DEBUG (XendDomainInfo:2195) Storing VM details: {'on_xend_stop': 'ignore', 'shadow_memory': '0', 'uuid':'17b7bcec-beb5-f2df-2752-ca9137654204', 'on_reboot': 'restart', 'start_time': '1240737746.81','on_poweroff': 'destroy', 'on_xend_start': 'ignore', 'on_crash': 'restart','xend/restart_count': '0', 'vcpus': '1', 'vcpu_avail': '1', 'image': (linux (kernel/boot/vmlinuz-2.6.26-2-xen-686) (ramdisk /boot/initrd.img-2.6.26-2-xen-686) (args/'root=/dev/hda1 ro 2') (notes (HV_START_LOW 4118806528) (FEATURES/'writable_page_tables|writable_descriptor_tables|auto_translated_physmap|pae_pgdir_above_4gb|supervisor_mode_kernel')
Re: Screen fonts in X
On Thu, 29 Jan 2004, Nano Nano wrote: On Thu, Jan 29, 2004 at 04:27:07PM -0800, G. Crimp wrote: Hi, Just installed Debian 3.0 r.0 from CD (No internet connection, I know 3.0 is at r2). Fonts are very fuzzy. Title bars are unreadable, text in a terminal window is, but not for very long, menus can be used too, but very hard on the eyes. I've looked in the archives. I'm forced to do my searches on time limited public terminals at the moment, so what I've found so far comes from threads in May 2003 principally: http://egads.ertius.org/~rob/font_guide.txt http://lists.debian.org/debian-user/2003/debian-user-200310/thrd5.html#03827 Haven't read your specific questions, but just do what the guide says. Read the font guide. Keep in mind that my internet connectivity is limited. If it ain't on my CD, it is a time consuming to acquire a package. I have, however, improved my CD situation. I now have the first three CD's of 3.0 r2. Didn't help me with the font guide you referred me to. I don't really need to have TrueType fonts on my system. X has already worked marvellously for me without them. Thought I'd try anyway thought, except that I can only find Japanese and Korean TrueType fonts packages on my CDs, (specifically, I don't have the bitstream package mentioned in the guide) so I am no farther ahead. Plus, fontconfig is not part of stable. Can't I just get some readable X fonts using Deb 3.0 without TrueType fonts and all the font config stuff. I can't imagine that the X redesign in v4 was so drastic as to remove this basic ability. If I absolutely must get the bitstream fonts and the fontconfig package from testing to make my screen readable, so be it. But given my Inet access woes, and the fact I have already spent a week getting connected whereever to try and find docs and packages to work this out, I'm still hoping someone can tell me how I can go home and get some easily readable menus, widget labels, and window content with what I have already. Is this maybe and anti-aliasing issue ? Can I shut that off somewhere (anything I've read so far about turning off AA was specific to software that doesn't concern me) Thanks again. G. Crimp -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Screen fonts in X
On Sat, 31 Jan 2004, Adam Aube wrote: Correct. Anyone know why x-window-system seems to depend on xfs, in that case ? I tried to uninstall xfs in my attempts to solve my font pb. I don't know why the dependency still exists, but I just used update-rc.d to remove all the symlinks so it won't start. Remember to comment out the line referring to xfs (local font server) in /etc/X11/XF86Config-4 - otherwise X will take a while to start (tries to connect and waits until it times out). Adam Haven't gone that far yet, but I have '/etc/init.d/xfs stop' d the font server for some tests, but it didn't improve anything. :( G. Crimp -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Screen fonts in X
On Fri, 30 Jan 2004, Joris Huizer wrote: G. Crimp wrote: - dpkg-reconfigure fontconfig in order to allow bitmapped fonts, but the command complained it didn't know what fontconfig was (neither do I, but that was the suggestion in one of the threads; Hello, Just a small note - I think this `dpkg-reconfigure fontconfig` failed because you don't have fontconfig installed; If a site told you you should (re)configure fontconfig, try I don't have the package on my CD. Might download it. But first maybe I should explain that I am not looking for the latest greatest whizbang gui. I run neither kde nor gnome. I'm using wmaker. I just want to be able easily to read menus, title bars, screen output.Until this release I have never had trouble doing so. Anyone know how I can get back to some simple readable fonts (don't need TrueType, etc). Anyone know if installing xfonts-pex will help (that will have to a download to floppy and carry home). Secondary question. In more than one place I have read that in v4 of X, font serving is integrated, so a font server is not necessary unless one wishes to make fonts available off thte workstation. Anyone know why x-window-system seems to depend on xfs, in that case ? I tried to uninstall xfs in my attempts to solve my font pb. Thanks, G, Crimp -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Screen fonts in X
Hi, Just installed Debian 3.0 r.0 from CD (No internet connection, I know 3.0 is at r2). Fonts are very fuzzy. Title bars are unreadable, text in a terminal window is, but not for very long, menus can be used too, but very hard on the eyes. I've looked in the archives. I'm forced to do my searches on time limited public terminals at the moment, so what I've found so far comes from threads in May 2003 principally: - I've tried reordering the font path in both XF86Config-4 and /etc/X11/fs/config (or something like that) to no avail (stopping and restarting xfs or xdm as the case warranted; - dpkg-reconfigure fontconfig in order to allow bitmapped fonts, but the command complained it didn't know what fontconfig was (neither do I, but that was the suggestion in one of the threads; - made sure I had freetype installed and that the module was being called for loading in XF86Config-4 - Checked the /var/log/XFree86.+.log logs finding nothing alarming. - I tried a couple of other dpkg-reconfigure suggestions, but the notes I've brought with me for this email aren't as good as I thought, so I can't remember what they were [he added lamely]. Any guesses ? Hmmm, as I write, I realize that I didn't change font orders for X and for xfs simultaneously (the font orders would be different between files in my tests) would this matter? I must admit that I don't understand very well how fonts work in X, especially since font serving was introduced, so I don't even know which file is being used if only one is, though I did try shutting down the font server totally for a couple of tests. I even attempted to remove the xfs package since one of the threads I found wondered out loud why people even ran them. Since X had this functionality built in now, an external server was really needed except for special fonts. But I was warned that X depended on xfs so I backed out of making that change. I realize that I haven't done a very complete archive search yet, but that is a real pain in the butt using heavily used public machines. ONe thing I did not in the X output was that PEX fonts couldn't be found. Sure enough xfonts-pex is not on the CD I have. Could this be the problem. I'd love to send the X output logs but can't use floppies from these terminals. If there is anything that needs to be checked, though, I will find a way to post it. There were a few glitches with some modules (DRI, hal, mga). Any help welcome. Id really like to get X usable as soon as possible. I've been switching to a VT to scoure the docs available with the install because its just too painful to read in X at the moment. Thanks, G. Crimp -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Don't understand Unix timestamps
On Thu, Sep 13, 2001 at 02:43:02AM +0100, Chris Kenrick wrote: Hi all, I'm a bit confused about Unix timestamps on files. In particular, I want to know what the timestamp on an 'ls -l' or a 'find . -ls' means. You can usually get this kind of information from man pages or info docs. For ls the info system gives the most complete information. See the following excerpts. $ info ls -l --format=verbose In addition to the name of each file, print the file type, permissions, ..., and timestamp (by default, the modification time). For files with a time more than six months old or more than one hour into the future, the timestamp contains the year instead of the time of day... You can also list the access or status change times. From the same source: -c --time=ctime --time=status --time=use If the long listing format (e.g., `-l', `-o') is being used, print the status change time (the `ctime' in the inode) instead of the modification time... -u --time=atime --time=access if the long listing format (e.g., `--format=long') is being used, print the last access time (the `atime' in the inode)... On a different but related note, what is the easiest combination of commands to find A) A list of files in a given directory that have been accessed in the last 24 hours Try 'man find' or 'info find'. $ find given directory -atime -24 B) The total disk usage by the given file list A) The command you want is 'du' (disk usage, see 'man du' or 'info du'). $ du -sh `find given directory -atime -24` Notice the back ticks (left single quotation mark) that surround the second command. This sends the output from the back ticked command line to the preceding command, from the `find' command to the `du' command in this instance. 'find given directory -atime -24 | xargs du -sh' should give the same result. HTH, Gerald
DHCP over cable
Hi, I need help setting up a dhcp client. I have recently been told by [EMAIL PROTECTED], my cable ISP, that they will be moving to an entirely dynamic setup from a hybrid dynamic/static one. I tried setting up dhcpcd once several months ago when I suddenly found myself with no DNS. They had changed the IPs of the servers, given to me as part of my static setup, without giving any notice. I figured I would move to a dynamic configuration to avoid this in the future, but I couldn't get it to work. Now I have no choice. If anyone is connected via cable, especially if you are a Shaw customer (Vancouver, BC), I'd love to hear how you are setup. Which dhcp client do you use (pump, dhcpcd, etc) ? What does your configuration file look like ? Thanks, G
Thanks, Re: cron: how ? WAS: cron: why not?
On Sat, Sep 08, 2001 at 06:29:39AM +0200, Martin F Krafft wrote: also sprach G. Crimp (on Fri, 07 Sep 2001 07:34:41PM -0700): Suppose one wants a command to run on the third of every month at 8:30 am, and only then. Neither 30 8 3 * * command nor 30 8 3 * 0 command will work according to the crontab(5) exerpt above. The first will cause it to run on the third, and every day of the week of every month. The second will cause the command to run on the third and on every Sunday. that's not true. the first part isn't. 30 8 3 * * will run only on the third of the month. either/or only applies when both dom and dow are non-*'s. Thanks.
Re: cron: how ? WAS: cron: why not?
On Fri, Sep 07, 2001 at 01:58:10AM +0200, Martin F Krafft wrote: crontab(5): Note: The day of a command's execution can be specified by two fields -- day of month, and day of week. If both fields are restricted (ie, aren't *), the command will be run when either field matches the current time. For exam ple, ``30 4 1,15 * 5'' would cause a command to be run at 4:30 am on the 1st and 15th of each month, plus every Friday. I was about to reply and point out the above characteristic of crontab behaviour. I'd still like to know if anyone knows how to get the behaviour that Martin wants. I'm in the midst of setting up a crontab entry myself and can't see a way around crontab using whatever matches from either the dom and dow fields. Suppose one wants a command to run on the third of every month at 8:30 am, and only then. Neither 30 8 3 * * command nor 30 8 3 * 0 command will work according to the crontab(5) exerpt above. The first will cause it to run on the third, and every day of the week of every month. The second will cause the command to run on the third and on every Sunday. What value does one put in the dow field so that it is null ? The same is true if one wants a command run on a given day of the week, and only on that day. Suppose the command is to run on Monday only. 30 8 * * 1 command will cause the command to run on Monday, but also every other day of the month. 30 8 1-31 * 1 command will cause the command to run every Monday, but also on whatever date one used in the dom field. Anyone know if there is a way around this ? Thanks, G.
fetchmail, 2 problems
Hi all, I recently updated fetchmail to 5.3.3-1.2. I have noticed two unusual conditions. First, I run fetchmail in daemon mode (fetchmail -d interval) and it occasionally appears to get lazy. If I haven't noticed any new mail in my box for awhile, I 'kill -QUIT' it and run it in the foreground. There will often be a ton of messages just sitting waiting to be POPped. Once I have this ton of messages POPped, I can put fetchmail back into daemon mode and all is well for some time. I have never had this problem on my home box before the upgrade, though I did have it on a work machine at a previous job. The second problem is that, when POPping a large number of messages, fetchmail appears to hang at the end of the run. For example, I just had the above described problem (ps shows fetchmail apparently running, but I was not getting any mail). I killed fetchmail and ran it manually. There were 238 messages on one server. After fetching all the messages, the command prompt doesn't come back. If I run ps axf in a separate window, I see that fetchmail has spawned a 'uname -a'. That's it, nothing more happens until I Ctrl-C. I don't know exactly how many messages are required before this happens, but I have had it not hang on POP run of 8 messages, and had it hang on a run of 35. This never happened before the upgrade. I suppose this could be what is making the background process choke, too. As a note, the fetchmail update was part of a dselect update/install, so fetchmail was not the only thing that was upgraded. Might be something else causing this I suppose. Any thoughts ? Thanks, G.
Re: Moving Directories
On Mon, Aug 06, 2001 at 03:30:06PM +1000, Steven Farrier wrote: On Sun, 5 Aug 2001 18:49:28 -0400 dman [EMAIL PROTECTED] wrote: Boot without mounting /usr. Mount /usr somewhere else and 'mkdir /usr'. Then 'mv /other_mount/* /usr'. how do I boot without mounting /usr? how do I mount /usr somewhere else. Why do you want to do this ? What is your goal in swapping the two partitions ? I would guess, from your two secondary questions above, that you don't have a lot of experience with Linux. I would suggest that the task you have set out might be a little ambitious for a beginner, though a good learning process.
Re: keyboard problem in console
On Wed, Aug 01, 2001 at 11:20:40PM +0300, Petteri Heinonen wrote: Hello all. Problems arise when I changed my hard drive. I copied (with tar) everything from my old drive to the new one, and made it bootable. It works all fine, except that all chracters are not correct, i.e. can't produce some characters by corresponding keys when I'm in console. In X I don't have any problems. I'm using scandinavian keyboard, so I think for some reason system thinks that I have US keyboard and produces wrong characters. What are most common reasons for keyboard problems and where could I find good information about how keyboards are handled? Can't answer your question directly. I use a non-US kbd, too, though, and have encountered problems from time to time. So, here is what I think I know. The console kbd is loaded from the /etc/init.d/keymap.sh script, which is actually referenced from the symbolic link /etc/rcS.d/S05keymaps.sh. On my system, that script loads the keyboard definition file /etc/kbd/default.kbd.gz. In my case, that file is a Canadian French keyboard layout. You can find other kbd maps in /usr/share/keymaps/arch/kbd-family/. If you are not already familiar with that shorthand, azerty, qwerty, and qwertz refer to the first six keys, left to right on the top row of letter keys on the keyboard. First, have a look at the /etc/init.d/keymaps.sh script to see which keymap it is loading at boot. Once you know what file it is loading, have a look at the file to see whether it defines a Scandinavian or US keyboard. It might be handy to visually compare it to the US and Scandinavian keymaps in /usr/share/keymaps/etc/ if you aren't familiar with how to read them. If it is the US keymap, that is your problem. You might have had a symbolic link from /etc/kbd/default.kbd.gz that didn't survive the tar from the old to new drive, or something like that. Once you have figured out which keymap you want from /usr/share/keymaps/blah/, you can copy it to /etc/kbd/default.kmap.gz (might want to mv the target to a backup filename first). I don't know if this is the official way, or the best way, nor that it won't blow up your computer or cause you to lose data. Just my thoughts. HTH.
compiling dxpc
Hi, I'm trying to compile the newest version of dxpc on two different machines (one is running Deb 2.0, the other Deb 2.1). When I get to the `make' step, neither machine will compile. I appear to be missing some header files (I think, I know little of C). I don't know what packages I need to install to get the missing files. Here's the make output from both machines (slightly different on each). - start errors on Deb 2.0 box --- # make gcc -c -g -O2 -DX_DISPLAY_MISSING=1 -DACCEPT_SOCKLEN_T=int main.C main.C:1: iostream.h: No such file or directory main.C:26: X11/Xproto.h: No such file or directory In file included from main.C:36: util.H:4: fstream.h: No such file or directory make: *** [main.o] Error 1 # - end Deb 2.0 box, start Deb 2.1 box - # make gcc -c -g -O2 -DX_DISPLAY_MISSING=1 -DACCEPT_SOCKLEN_T=int main.C main.C:26: X11/Xproto.h: No such file or directory make: *** [main.o] Error 1 # -- end make errors for both boxes Any solutions ? TIA Gerald Crimp
Re: compiling dxpc
On Sun, Jan 16, 2000 at 07:47:45PM -0800, Eric G . Miller wrote: On Sun, Jan 16, 2000 at 07:19:54PM -0800, G. Crimp wrote: [snip] Yes, you need to use g++ to compile these programs. They are C++, not C. So you need to have the relevant C++ development debs. Looks like you also need the X development debs. -- Do you have any package names ? Thx, Gerald ++ | Eric G. Milleregm2@jps.net | | GnuPG public key: http://www.jps.net/egm2/gpg.asc | ++ -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null
Re: TeX, THANKS
On Mon, Jan 03, 2000 at 04:07:16PM -0800, Brian Lavender wrote: Go to an ftp archive and go into the proposed-updates for slink. There was some sort of y2k problem with tex where the make format or something to that effect failed. Anyway, the updated package will work. In fact here is the url for the cdrom debian archive. Thanks. That did the trick. The packages on the main Debian ftp site must have changed between the first install that worked and the second that didn't. How weird that the y2k problem was even a problem before 2000. Anyway, looks like I have a working system again. -- Gerald Crimp
TeX
Hi, I've been using TeX at work for a few weeks now. Last week I got a new computer and installed 2.1 on it (same as was on the old box). As far as I can tell, I have all the appropriate tetex packages in place. TeX no longer works, though. When I try to run tex on a .tex file, I get the following error: This is TeX, Version 3.14159 (Web2C 7.2) I can't find the format file `tex.fmt'! The tetex docs say to run the command `texconfig init' in such a case. When I run it I get tons of output, including the following excerpts: [...] Beginning to dump on file tex.fmt (format=tex 1999.12.30) [...] No pages of output. Transcript written on tex.log [...] The command continues to run producing similar output concerning latex and metafont. However, the file tex.fmt is nowhere to be found on my file system. Maybe the line No pages of output. explains that. I thought I could get some answers from the transcript in tex.log, but that file appears not to have been written anywhere either. Anybody know what I have done wrong on this new computer, or more positively, how I can back to writing documentation with TeX ? Thanks, Gerald Crimp
Re: KDE: can't connect to X server
On Sun, Jul 18, 1999 at 12:54:22PM -0700, Nate wrote: On Sun, Jul 18, 1999 at 12:00:29PM -0700, G. Crimp wrote: Is there an ftp site where I can get the debs for kde and qt (with matching version :)). You can get qt1g (which is qt-1.4) from the following... http://http.us.debian.org/debian/dists/slink/non-free/binary-i386/ libs/qt1g-1.40-1.1.deb ^^ I know about this version of qt, but kde 1.1 wants at least 1.42 and the deb you point out is only 1.40. I tried it anyway, forcing dpkg to overlook the 'needs qt1g (= 1.42) whereas installed qt1g is only 1.40'. Not surprisingly, since the kde site explicitly states that 1.42 is the minimum for kde 1.1, once everything was installed, it wouldn't fly. potato has a qt version 1.44. As far as kde 1.1 is concerned, that is OK. But the deb package of qt 1.44 depends on higher version of some of the slink libraries, plus a library or two that don't exist in slink. I'm afraid of upgrading the existing libraries and adding new ones from potato for fear of breaking my slink setup. I'm not having any trouble getting ahold of the kde debs. The problme is finding a deb of a version of qt that is suitable for the version of kde available in deb package form. Does anyone know of a qt*1.42*.deb anywhere ? Thanks again, Gerald
KDE: can't connect to X server
Hi, I'm trying to get KDE running, just to play a little bit. When I run kde from the command line, I am told that various kde components can't connect the X server, and it craps out, even as root. I have a fairly messy setup, and I am not sure what to try next. I've already spent several hours downloading various versions of components and trying them out. I'm using kde 1.1 (kde doesn't seem to have debianized 1.1.1, yet, at least not at the ftp sites I tried). It needs at qt of at least version 1.42. The only debs I could find were 1.40 and 1.44. When I tried to install 1.44, it wanted more recent version of a couple of libraries. Looked like stuff from potato which scared me off, so I am using the qt tar ball which I compiled myself. Does anyone know why kde would be unable to connect to the X server ? I don't think it is because it can't find the qt libraries. I set up all the environment variables to make qt visible even though it's in /usr/local/lib (I did have a problem but ldd told what was invisible and I fixed that). Failing answers to the above, does anyone know where I can get a deb of qt 1.42 ? Thanks, Gerald
Re: KDE: can't connect to X server
On Sat, Jul 17, 1999 at 10:28:05PM -0700, Nate wrote: On Sat, Jul 17, 1999 at 10:06:29PM -0700, G. Crimp wrote: Hi, I'm trying to get KDE running, just to play a little bit. When I run kde from the command line, I am told that various kde components can't connect the X server, and it craps out, even as root. Add the following line to /etc/apt/sources.list deb http://kde.tdyc.com slink kde contrib rkrusty Run apt-get update When that is done apt-get install kdebase kdeutils kdenetwork kdemultimedia, etc, etc, This will install both qt1g and qt1g-dev and all other dependencies. Thanks. I'll have a couple problems with this though. First, the machine I want on which I am trying to install KDE isn't connected to the internet, and I don't really want to install everything anyway, at least, not at first. I was hoping to get away with base, libs and support. I am not that comfortable with all the details of apt yet, but it seems to me that the above precludes the use of apt. Plus, I really DON'T want to install KDE to the box that is hooked up to the net. Is there an ftp site where I can get the debs for kde and qt (with matching version :)). Thanks, Gerald
Re: .bash_history
On Wed, 14 Jul 1999, Patrick Kirk wrote: Hi all, I love the ability to get at previous commands just by flicking the up arrow. But if the last 15 commands were mutt, its a little less conenient. Is there a way that if the last 15 commands were ls preceded by 10 mutts amd preceded by three tops, then the uparrow would first show ls, then mutt and then top? -- Patrick Kirk If I understand you correctly, you only want to hit the up arrow three times to get to top rather than 25 times. One way is to tell bash not to store seqential duplicates of a command. Put this in your bash environment: export HISTCONTROL=ignoredups With that, if you do 15 'ls -l', 10 'mutt' and 3 'top', all your history file will retain is ... whatever whatever whatever whatever whatever whatever ls -l mutt top whatever whatever whatever ... However, this works on the whole command line, not just the command name. Thus, if you do 7 'ls -l', a 'ls -sAF', 7 more 'ls -l', 9 'mutt', a 'mutt -R', and 3 'top', your history file will retain ... ls -l ls -sAF ls -l mutt mutt -R top ... HTH, G.
make menuconfig
Hi, I tried doing a make menuconfig instead of make config to configure the kernel. Very near the beginning it craps out because of a missing curses.h file. (This is a Deb 2.0 box by the way, kernel 2.0.34) The make script cd's to /usr/src/linux/scripts/lxdialog. curses.h is an include in lxdialog.c: In file included from lxdialog.c:22: dialog.h:29: curses.h: No such file or directory make[1]: *** [lxdialog.o] Error 1 make[1]: Leaving directory /usr/src/kernel-source-2.0.34/scripts/lxdialog' make: *** [menuconfig] Error 2 Did a find on the entire file system and it didn't turn up anywhere. Any idea why I wouldn't have it on my system ? Thanks, Gerald
Re: DOS MBR
On Tue, Jul 06, 1999 at 10:55:01PM -0400, William T Wilson wrote: On Tue, 6 Jul 1999, Felipe Alvarez Harnecker wrote: Can someone send me a DOS Master Boot record so a can dd it to /dev/hda? No, but if you boot a DOS floppy then you can reconstitute it by running (DOS) fdisk /mbr. Linux fdisk will not do this. This procedure will not alter your partition table, but it will, of course, make Linux unbootable from the HD. Unbootable, that is, if LILO was installed to /dev/hda rather than to one of the partitions.
Re: make menuconfig
On Tue, Jul 06, 1999 at 04:26:18PM -0700, G. Crimp wrote: Hi, I tried doing a make menuconfig instead of make config to configure the kernel. Very near the beginning it craps out because of a missing curses.h file. (This is a Deb 2.0 box by the way, kernel 2.0.34) The make script cd's to /usr/src/linux/scripts/lxdialog. curses.h is an include in lxdialog.c: In file included from lxdialog.c:22: dialog.h:29: curses.h: No such file or directory make[1]: *** [lxdialog.o] Error 1 make[1]: Leaving directory /usr/src/kernel-source-2.0.34/scripts/lxdialog' make: *** [menuconfig] Error 2 Did a search on ncurses in dselect and saw that I had 4 ncurses packages installed. Didn't even occur to me to look if there were any others. I'll install the -dev as well. Gerald
Re: I/O error on /dev/dsp and /dev/audio
On Fri, Jul 02, 1999 at 08:40:39AM +0200, Jens Ritter wrote: On Thu, 1 Jul 1999, G. Crimp wrote: Ah ha. The kernel log reports the following: kernel: Sound: DMA (output) timed out - IRQ/DRQ config error? give me any more help with this. I have read the sound HOWTO and the isapnp docs, but when it comes to reading the pnpdump file I find it pretty bewildering. As I mentioned in my initial post, the fact that I had this working on a Deb 1,3 system was more good luck than anything else. The pnpdump output often says to choose only one of a selection, but I am never sure which one to choose, nor in fact which of several lines constitute a single choice. Here is the output from cat /dev/sndstat: [...] Card config: Sound Blaster at 0x220 irq 5 drq 0,1 Roland MPU-401 at 0x330 irq 12 drq 0 6860 UART Midi irq 11 drq 0 (SB MPU-401 at 0x800 irq 5 drq 0) OPL-2/OPL-3 FM at 0x388 drq 0 [...] Make this match with you kernel config and the isapnp config. Have a look at /proc/dma (when all modules are loaded). I am not sure what the () around SB MPU have to say. Check in the sound howto or in the kernel sources. HTH, Jens I'm still stuck with this. I went over the Sound-HOWTO and the isapnp docs again and then tried to reconfigure the sound card and recompile the kernel sound support. I thought I had this licked, but I still get the I/O errors on /dev/audio and /dev/dsp. If anybody understands sound and hardware registers, etc, better than I do, I would sure love a hand with this. I have included the pnpdump file which I edited to use as isapnp.conf, and the relevant bits from the kernel .config file in case anyone can spot what i am doing wrong. I've tried a number of different combinations and nothing seems to work, which is frustrating since I had this working first go round on a bo (1.3) box. Here are the files, Thanks in advance. -- isapnp.conf # $Id: pnpdump.c,v 1.1.1.2 1998/01/07 05:17:47 fred Exp $ # This is free software, see the sources for details. # This software has NO WARRANTY, use at your OWN RISK # # For details of this file format, see isapnp.conf(5) # # For latest information on isapnp and pnpdump see: # http://www.roestock.demon.co.uk/isapnptools/ # # Compiler flags: -DREALTIME -DNEEDSETSCHEDULER -DNEEDNANOSLEEP # # Trying port address 0203 # Board 1 has serial identifier 27 ff ff ff ff 68 18 73 16 # (DEBUG) (READPORT 0x0203) (ISOLATE) (IDENTIFY *) # Card 1: (serial identifier 27 ff ff ff ff 68 18 73 16) # Vendor Id ESS1868, No Serial Number (-1), checksum 0x27. # Version 1.0, Vendor version 1.0 # ANSI string --ESS ES1868 Plug and Play AudioDrive-- # # Logical device id ESS # # Edit the entries below to uncomment out the configuration required. # Note that only the first value of any range is given, this may be changed if required # Don't forget to uncomment the activate (ACT Y) when happy (CONFIGURE ESS1868/-1 (LD 0 # Logical device decodes 16 bit IO address lines # Minimum IO base address 0x0800 # Maximum IO base address 0x0ff8 # IO base alignment 8 bytes # Number of IO addresses required: 8 (IO 0 (BASE 0x0800)) (ACT Y) )) # # Logical device id ESS1868 # # Edit the entries below to uncomment out the configuration required. # Note that only the first value of any range is given, this may be changed if required # Don't forget to uncomment the activate (ACT Y) when happy (CONFIGURE ESS1868/-1 (LD 1 # Multiple choice time, choose one only ! # Start dependent functions: priority preferred # First DMA channel 1. # 8 bit DMA only # Logical device is not a bus master # DMA may execute in count by byte mode # DMA may not execute in count by word mode # DMA channel speed in compatible mode (DMA 0 (CHANNEL 1)) # Next DMA channel 0 or 3. # 8 bit DMA only # Logical device is not a bus master # DMA may execute in count by byte mode # DMA may not execute in count by word mode # DMA channel speed in compatible mode (DMA 1 (CHANNEL 0)) # IRQ 5. # High true, edge sensitive interrupt (by default) (INT 0 (IRQ 5 (MODE +E))) # Fixed IO base address 0x0220 # Number of IO addresses required: 16 (IO 0 (BASE 0x0220)) # Fixed IO base address 0x0388 # Number of IO addresses required: 4 (IO 1 (BASE 0x0388)) # Fixed IO base address 0x0330 # Number of IO addresses required: 2 (IO 2 (BASE 0x0330)) # Start dependent functions: priority acceptable # First DMA channel 1. # 8 bit DMA only # Logical device is not a bus master # DMA may execute in count by byte mode # DMA may not execute in count by word mode # DMA channel speed in compatible mode
xmcd sometimes fouls up the cdrom drive
Hi, I have a NEC 4 disk cdrom drive. Sometimes after using xmcd, the drive seems to get all balled up. It seems to get confused about whether there are any CD's in any of the 4 slots and just keeps cycling through them looking. When it gets in this state, I can't play any more CD's If I quit xmcd, remove the audio cd and replace with a iso9660 cd, I can't mount the CD. I get the bad super-block, device busy or wrong fs type message. If I switch to single user mode the error message is more verbose when trying to mount the cdrom drive: --- error from single user mode # mount /cd hdb : tray open or drive not ready hdb: code: 0x70 key: 0x02 asc: 0x06 ascq: 0x00 hdb: packet command error: status=0x51 hdb: packet command error: error=0x54 hdb: code: 0x70 key: 0x05 asc: 0x24 ascq: 0x00 hdb : tray open or drive not ready hdb: code: 0x70 key: 0x02 asc: 0x06 ascq: 0x00 hdb : tray open or drive not ready hdb: code: 0x70 key: 0x02 asc: 0x06 ascq: 0x00 hdb : tray open or drive not ready hdb: code: 0x70 key: 0x02 asc: 0x06 ascq: 0x00 hdb : tray open end_request: I/O error, dev 03:40, sector 64 isofs_read_super: bread failed, dev 03:40 iso_blknum 16 block 32 hdb: code: 0x70 key: 0x02 asc: 0x06 ascq: 0x00 mount: wrong fs type, bad option, bad superblock on /dev/hdb, or too many mounted file systems hdb : tray open or drive not ready hdb: code: 0x70 key: 0x02 asc: 0x06 ascq: 0x00 hdb: packet command error: status=0x51 hdb: packet command error: error=0x54 hdb: code: 0x70 key: 0x05 asc: 0x24 ascq: 0x00 hdb : tray open or drive not ready hdb: code: 0x70 key: 0x02 asc: 0x06 ascq: 0x00 end single user mode error message-- I'm wondering if anyone knows what gives, and better yet, how I can reset the device to regain control. I've tried rm /dev/hdb followed by mknod -m 771 /dev/hdb b 3 64 but that does't help. The only solution I've found thus far is to reboot ( :¬( ) Any ideas welcome. Gerald
Re: User menu problem
On Mon, Jul 05, 1999 at 10:04:28AM +1000, Tadeusz Bak wrote: Hi all, I have just installed StarOffice 5.1 in my Debian 2.1 system. I tried to put an entry into the menu system (I use fvwm2 as a window manager). So I created a file .menu/soffice in my home directory with the contents: ?package(local.soffice):needs=x11 section=Apps/Editors title=StarOffice \ command=~/Office51/bin/soffice and then I run update-menus command. Unfortunately there is no new entry in the menu. What did I wrong? Thank you for any suggestions. Assuming Deb 2.1 is like Deb 2.0 If you ran update-menus as root user, it won't look in ~/.menu for packages to add to the system. You have to run it as the user in question. However, there is some problem running it as an ordinary user. Can't remember what it is. I think you'll find the explanation in either man update-menus or somewhere under /usr/doc/menu. Why not put the entry in /etc/menu and run as root ? Of course, after running update-menus, you have restart fvwm2 to see the changes in the menu system. Gerald
Re: User menu problem
On Mon, Jul 05, 1999 at 01:52:05PM -0500, ktb wrote: You could edit /etc/X11/fvwm2/menudefs.hook directly. kent Bad idea. /etc/X11/fvwm2/menudefs.hook is an automatically generated file. Next time the file is regenerated, any changes you made to it will be lost. The Debian menu system has supplied specific mechanisms for menu customization. Why side step what was specifically provided for this purpose ? G. Tadeusz Bak wrote: Hi all, I have just installed StarOffice 5.1 in my Debian 2.1 system. I tried to put an entry into the menu system (I use fvwm2 as a window manager). So I created a file .menu/soffice in my home directory with the contents: ?package(local.soffice):needs=x11 section=Apps/Editors title=StarOffice \ command=~/Office51/bin/soffice and then I run update-menus command. Unfortunately there is no new entry in the menu. What did I wrong? Thank you for any suggestions. -- Tad -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null
Re: X won't start
On Thu, Jul 01, 1999 at 09:14:14PM -0700, Bob Nielsen wrote: I recompiled my kernel today (minor changes) and after installing the new version, when I try to start X, I get: ... [snip] (**) FontPath set to /usr/X11R6/lib/X11/fonts/misc/,/usr/X11R6/lib/X11/fonts/75dpi/:unscaled,/usr/X11R6/lib/X11/fonts/100dpi/:unscaled,/usr/X11R6/lib/X11/fonts/75dpi/,/usr/X11R6/lib/X11/fonts/100dpi/ (--) SVGA: PCI: Cirrus Logic GD5430 rev 45, Memory @ 0xe500 (--) SVGA: chipset: clgd5430 (--) SVGA: videoram: 512k (**) SVGA: Using 16 bpp, Depth 16, Color weight: 565 (--) SVGA: Maximum allowed dot-clock: 22.778 MHz (--) SVGA: There is no defined dot-clock matching mode 1024x768 ^ Fatal server error: No valid modes found. I have no idea what caused this. /etc/X11/XF86Config hasn't changed and ran fine for a couple of years. I reinstalled the old kernel and it made no difference. I'm running potato, kernel 2.2.10 on a K6-2/350 with 64 MB ram. It's a generic video card with a Cirrus 5430 chip. I have no idea where the 22.778 MHz maximum dot-clock line came from. Any ideas? Looks like something must have changed somehow. If this was the entirey of the output from X, it looks like you only had one mode defined in XF86Config: 1024x768. The dot clock this mode asks (the first number in the Modeline definition of the Monitor section) for is not is not available on your videocard. Maybe you deleted the DotClocks definitions in the Device section ? The 22.778 MHz maximum dot-clock line came from the X server probing your card to see what clock frequencies it was capable of generating. So there is a mismatch between what you ask for in the 1024x768 Mode definition and what the video card is capable of generating. When the X server determines that the video card cannot provide the dot clock requested in a given mode definition, it deletes that mode definition from the list of available resolutions. If you had other (valid) mode definitions in XF86Config, the X server would continue with the next mode in the Display section. It looks like you don't have any other modes defined. Have you examined your XF86Config to see if it got hosed somehow ? You might have to reconfigure. Hope this helps. Gerald
Re: Boot Floppy
On Wed, Jun 30, 1999 at 07:07:32PM -0300, The FreeStuff Web Ring wrote: Hello, A friend of mine is trying to install Debian 2.1 from the 4 CD Set onto a computer that doesn't enable booting from a CD Rom. Can you tell me how to make a boot floppy please? Much appreciated! Hello Nova Scotia ! The following assumes that your friend is running DOS or Windows of some flavour and can access his CD drive. I've outlined the procedure for making rescue floppy below. The Debian rescue floppy is what boots the computer into the install routine. Its also nice to have around if you hose your system so badly that you can't reboot. However, if your friend already has an OS on the computer and can access the CD, there may be am easier way to boot. There is a DOS batch script on the CD that you can run from a DOS prompt that reboots the computer with Linux. The batch script is called boot.bat and it in the install directory of CD. Let's say that his CD is D: under DOS/Win: C\: d: D\: cd \install D\install\: boot.bat and you should be off to the races. To make a rescue floppy from DOS, you use the rawrite2.exe programme. Simply copying the images won't work -- you must use rawrite. Also, you may have to be in true DOS mode rather than just a DOS prompt in a window. I gather some people have trouble doing this from a window prompt. Before you use rawrite2.exe, you have to choose which image file you are going to use. If you are installing to a desktop computer with a 3.5 high density floppy, you probably want to use resc1440.bin. rawrite2.exe and the floppy images are in CD-drive-letter\dists\stable\main\disks-i386\current. See section 5.3 of the document install.txt (or install.html with a web browser) to decide if resc1440.bin is appropriate for you. To use rawrite: C\: d: D\: cd \dists\stable\main\disks-i386\current D\dists\stable\main\disks-i386\current: rawrite2 -f image_u_want -d a: or b: -- drive your blank floppy is This information is free and without warranty. Your friend really should read the README.* files in the root directory of the CD as well as the install.txt file noted above. Good luck.
-xrm option to X apps
Just wondering if anyone knows what the -xrm option does for apps that run in X. man X gives a little blurb, but it is not very clear. I've tried a couple of experiments like: xterm -xrm title=foobar but it doesn't seem to work. The is no discussion in man X about the syntax of expression in -xrm expression so I don't know if the above is correct. Anyone have knowledge of this curious little command attribute ?
Re: I/O error on /dev/dsp and /dev/audio
On Thu, Jul 01, 1999 at 08:20:01AM +0200, Jens Ritter wrote: Does cat /dev/sndstat work? Check if sound is compiled in or loaded when a sound request happens (check logs). Jens Ah ha. The kernel log reports the following: kernel: Sound: DMA (output) timed out - IRQ/DRQ config error? I don't doubt that when I was configuring isapnp I made a mistake. Can you give me any more help with this. I have read the sound HOWTO and the isapnp docs, but when it comes to reading the pnpdump file I find it pretty bewildering. As I mentioned in my initial post, the fact that I had this working on a Deb 1,3 system was more good luck than anything else. The pnpdump output often says to choose only one of a selection, but I am never sure which one to choose, nor in fact which of several lines constitute a single choice. Here is the output from cat /dev/sndstat: --- cat /dev/sndstat - Sound Driver:3.5.4-960630 (Tue Mar 23 11:47:08 PST 1999 root, Linux humble 2.0.34 #1 Sun Mar 21 10:18:45 PST 1999 i586 unknown) Kernel: Linux humble 2.0.34 #1 Tue Mar 23 11:48:33 PST 1999 i586 Config options: 0 Installed drivers: Type 1: OPL-2/OPL-3 FM Type 5: Roland MPU-401 Type 8: 6860 UART Midi Type 2: Sound Blaster Type 7: SB MPU-401 Card config: Sound Blaster at 0x220 irq 5 drq 0,1 Roland MPU-401 at 0x330 irq 12 drq 0 6860 UART Midi irq 11 drq 0 (SB MPU-401 at 0x800 irq 5 drq 0) OPL-2/OPL-3 FM at 0x388 drq 0 Audio devices: 0: ESS ES1688 AudioDrive (rev 11) Synth devices: 0: Yamaha OPL-3 Midi devices: 0: MPU-401 0.0 Midi interface #1 1: 6850 UART Timers: 0: System clock Mixers: 0: Sound Blaster -- end cat /dev/sndstat -- Thanks, Gerald G. Crimp [EMAIL PROTECTED] writes: Hi, I can't write to either /dev/audio or /dev/dsp, even as root. The permissions on both are crw-rw. I discovered when trying to set up RealAudio for a Linux broadcast. When I tried to cat a file to either of these devices I get cat: write error: Input/output error.
Re: Problems changing colour depth.
On Thu, Jul 01, 1999 at 07:50:58PM +1000, Revenant wrote: I dlded and installed the newer version of the i740 drivers, and I've now got the screen running in 1024 x 768 mode okay (yay!) but it won't use a colour depth of greater than 8. I've tried running startx with the -bpp option and it still came up in 8-bit mode. Any thoughts? Thanx. What command line did you use ?
Re: TAR.GZ
On Thu, Jul 01, 1999 at 10:56:04PM +0200, Cuno Sonnemans wrote: Hi, I've downloaded GUILGNL0.GZ (WP8 language module). Now I want to try to extract it. I've tried, tar -xzvf .., and gunzip . In both cases I got the message: not a gzip format. How is this possible and what is the way to extract GUILGNL0.GZ !!! HTH Cuno Sonnemans Hi, tar zxvvf worked for me on WP 8 language modules. Maybe the file was corrupted on download. I've heard that some browser will do tranlations on a file when downloading. If you downloaded use and ftp client, were you in binary mode ? Try doing: $ file GUILGNL0.GZ and see if it says that it is gzip compressed data. Gerald
X -- switching colour depths on the fly
Hi all, I know about changing resolutions in X by using the Ctrl and Alt keys along with the keypad + or -. I am wondering now if anyone knows if it is possible to switch colour depths without restarting X and demanding a specific depth. Thanks, Gerald
X -- switching colour depths on the fly
Hi all, I know about changing resolutions in X by using the Ctrl and Alt keys along with the keypad + or -. I am wondering now if anyone knows if it is possible to switch colour depths without restarting X and demanding a specific depth. Thanks, Gerald
I/O error on /dev/dsp and /dev/audio
Hi, I can't write to either /dev/audio or /dev/dsp, even as root. The permissions on both are crw-rw. I discovered when trying to set up RealAudio for a Linux broadcast. When I tried to cat a file to either of these devices I get cat: write error: Input/output error. I am wondering if I have maybe misconfigured my sound card. I know I could cat sound files through to the device when I was still running Deb 1.3. I don't know much about sound though. When I was compiling the kernel for sound, is was more by guess and by golly than thorough understanding that I got it to work in 1.3. I thought I had compiled things the same way when I upgraded. Anyone know what I should be looking at ? I know at least part of the kernel support is OK because I can play a music CD throught the sound card. Thanks, Gerald
booting with NT bootloader fails
Hi, I'm trying to make Linux bootable from a scsi harddrive using NT bootloader. I'm reasonably used to lilo, though far from understanding all the intricacies of the process. I've also read the Linux+NT HOWTO and think I have followed it appropriately. After choosing Linux from the NT boot menu, the kernel is found and hardware detection begins, but when it comes time to mount the / partition, I get the following panic message: --- Kernel panic: VFS: unable to mount the root from 08:06 --- 08:06 is /dev/sda6 which is the root partition. I think it is not finding the drive, because during scsi detection, just before attempting to mount /, is the message SCSI: 0 hosts detected total. I know the kernel is aware of the the drive though, because I can boot from a floppy no problem and mount all the necessary partitions from sda. Any ideas what might be the problem ? Thanks, Gerald
Re: LILO
On Fri, May 07, 1999 at 10:00:11AM -0400, Wayne Topa wrote: [snip] # Generated by liloconfig # Specifies the boot device boot=/dev/hda1 boot should be the drive not that partition, ie boot = /dev/hda What do you mean by should ? I have several computers booting from a partition boot record, such as hda1 rather than from the MBR such as hda. G.
Cobalt Qube
Anyone know anything about the Cobalt Qube. I have someone asking me questions about setting up a Linux environment for himself (diskless, fanless box, remember ?). He specifically asked me about the Qube. There are a couple of things that I can't answer on my own. The specs on the web say it is tuned as a web server. I wonder a) if that would have serious consequences if it's prime use was general purpose, rather than web serving; b) how easy it would be to put other apps and servers on the box for general purpose use. I'm also wondering about mixing architechtures. This guy wants to sit in front of a quiet diskless box at his desk (not entirely solved yet) that runs all apps across the net from another box (the Qube, maybe) sitting in another room. If you have a ix86 diskless box on the desk, is it going to be able to run apps served from a RISC port of Linux ? Could one put x86 binaries onto the RISC harddrive and have them served to the diskless box ? Thanks alot everyone. Gerald
Re: Cobalt Qube
On Tue, May 04, 1999 at 10:22:21PM -0400, Sergey V Kovalyov wrote: On Tue, 4 May 1999, G. Crimp wrote: I'm also wondering about mixing architechtures. This guy wants to sit in front of a quiet diskless box at his desk (not entirely solved yet) Why not just keep the existing Sun box and just use it as X-terminal. You can either keep Solaris or install Linux on it. that runs all apps across the net from another box (the Qube, maybe) sitting in another room. You'd probably want something more powerful here. If you have a ix86 diskless box on the desk, is it going to be able to run apps served from a RISC port of Linux ? What you mean served ? You can run applications on RISC and display them on your local worstation. Any architecture. I guess I am not too sure of what the various networking possibilities are, nor how they work. He currently runs a diskless box. I suggested he could just buy a used x-terminal and put the Linux box in another room. That's when I found out he was running a diskless sparc booting off the network. He pointed out it had it's own RAM and cpu. From that and my own brief reading of diskless boots over a network, that the local work station does more than just act as a display. Rather than having many users sharing cpu cycles on the central box, everyone, including the diskless ones just got the apps remotely and did the actual processing on their own cpu using their own memory. So by served, I guess I meant getting the code off the remote disk, but running it locally. Gee, I guess that means that swap must me done over the net as well. Doesn't sound good. Maybe I need some straightening out on the basic concepts involved here. Gerald Could one put x86 binaries onto the RISC harddrive and have them served to the diskless box ? Sure. Put them in a separate directory , export it via NFS. Sergey.
Re: diskless box: fanless too ?
On Mon, May 03, 1999 at 04:08:47PM -0700, George Bonser wrote: On Mon, 3 May 1999, G. Crimp wrote: Does anyone know if the same is possible in the ix86 architechture ? I know I could set him up with a diskless box booting off a server in another room, but could that diskless box also have a fanless powersupply ? I figure the cpu would still need its fan. Anyone think that the absence of a fan in a diskless box would cause heating problems ?? Not sure about ix86 but a netwinder might work ... they draw only 15 watts. I would think that even if they have a fan, it could be disconnected. Would be interesting getting to to boot over the net but should't be impossible. This raises two questions in my mind, if you have the patience to deal with them. 1) What about application/utility compatibility ? I often heard a few years back, when the alphs port was the only other Linux, that it worked fine, but there were not as many goodies to run on it. If I suggest a strongarm cpu to this person, how limited is he going to be ? I don't know why stuff can't simply be recompiled for a new architecture, I jus tknow it can't. 2) In the I don't know why, it just is department, why would this be any more difficult to boot over the network than an ix86 ? Simply because no one has taken the time to write the code ? Or, as your statement above leads me to suspect, because something in the Arm architecture makes this an exercise in binary gymnastics ? Thanks for the info, Gerald
Re: diskless box: fanless too ?
On Mon, May 03, 1999 at 06:21:53PM -0500, Jens B. Jorgensen wrote: I would think you could do it if you used once of those heat sinks which have a peltier junction on the bottom. I don't know if anyone's tried this of course but I do know such heat sinks exist (I saw and felt one a Comdex last year and boy was it cold!) Isn't this risky ? I don't know exactly what you are talking about, but it sounds very similar to what I saw a wholesalesman bring into a computer store and demo: small cpu fan the contact side of which got VERY cold as soon as it was given power. A few seconds after that, it also became thouroughly covered in condensation. I'm not sure I would feel comfortable having an active rain cloud inside my box next to all those sensitive electronic components. Are we talkinga aboput the same thing ? Gerald
Re: system shutdown from xdm
You could also use sudo to let certain people have the privilege of running shutdown. They won't need to know the root password, only their own, and you could even set it up so they don't need their own password. Here's an example sudoers entry: username machinename= /sbin/shutdown -[hr] now would allow the user _username_ to run either the shutdown or reboot command on the host _machinename_. A simple modification to that would allow _username_ to run the command without her own password. See man 8 sudo, man 5 sudoers and man 8 visudo. Gerald On Tue, May 04, 1999 at 09:45:03AM +0200, Nils Rennebarth wrote: On Mon, May 03, 1999 at 07:32:38PM +0200, Helge Hafting wrote: If you want anybody to shutdown without password, make /sbin/shutdown a suid executable. Make shutdown a menu choice in your window manager for user friendliness. Don't do the suid thing unless *anybody* logging into your machine should be able to stop it! This includes logging in via network or possibly the internet. It is ok for an unconnected home machine though. You might create a group of trusted people (local perhaps?), change the group of /sbin/shutdown to this group and remove the x bit for 'other': chgrp local /sbin/shutdown chmod o-x /sbin/shutdown Now only user in group local may shutdown the machine. Nils -- Plug-and-Play is really nice, unfortunately it only works 50% of the time. To be specific the Plug almost always works.--unknown source
Re: finding and using applications
A person has to start somewhere. If your novice user knows enough to get by, they already have a tremendous advance over the absolute green horn. apropos only scans man pages, but that is a good place to begin. Of course, if the person doesn't know what they are looking for, they won't be able to find it. They could always try asking someone who know more than thmeselves. There are also any number of good un*x books that can help a person learn more about what might be available. Do you know of any OS where looking for a tool to do a specific job is easy if you don't already know about ? G. On Tue, May 04, 1999 at 07:14:53AM -0400, Tommy Malloy wrote: Suppose you have a Debian Gnu/Linux system set up and fully loaded with applications. A new user appears who is going to use the system. The new user is a unix novice. He/she knows enough basic commands to get by. Is there a simple way for that user to find every available application on the system, what the application does, and how to use it? I really don't think so. Remember apropos only scans man pages. Looking in /usr/bin isn't much help for finding a tool to do a specific job unless you already know about it. I really believe that any user should be able to step up to the machine and quickly and easily find if an application to do what they want is available. Yes this is available for many applications, but not for all. I believe that this serious problem, which is an impediment to Linuxes mass acceptance could easily be fixed. Debian should not include application that are not fully documented ie have manpages, info pages etc Also some frontend appliction for finding applicatons would be helpful Somethnig based on he code for dselect would probably work fine. Please don't suggest that I write it. I can't. I am only commenting on a feature I would like to see. Please don't ask what type of applicaton I am looking for so you can help me find it. I am not looking for an application. But I would like to be able, and have any users be able, to know what applications are available on my system and how to use them. I would like to be able to get that informantion exclusivly from my computer and not depend on this list, irc, usenet, my big pile of tech books, or any external source. I am root for heavens sake. Strictly from a System administration perspective, There should be a simple way for users to know what apps are available to them. If you know one please let me know it -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null
Re: mutt+exim not working
On Mon, May 03, 1999 at 08:46:04PM +0200, J.H.M. Dassen wrote: On Mon, May 03, 1999 at 10:32:36 -0700, Paul Nathan Puri wrote: my_hdr From: [EMAIL PROTECTED] The user you run mutt under must be listed in exim's trusted user list. Say your local username is paul, you'd have to have something like trusted_users = mail:paul in /etc/exim.conf . I don't think this is correct. I am using mutt and exim and have not changed trusted_users to include the username I work under when mailing with mutt. Ger
diskless box: fanless too ?
I have been asked to help someone learn about Linux. So far, I have installed Linux on an existing box. I must also give advice on new hardware. One of the concerns this person has, is not having anything spinning in his office. He currently uses a diskless sparc station, which is apparently also fanless since it is very very quiet. Does anyone know if the same is possible in the ix86 architechture ? I know I could set him up with a diskless box booting off a server in another room, but could that diskless box also have a fanless powersupply ? I figure the cpu would still need its fan. Anyone think that the absence of a fan in a diskless box would cause heating problems ?? Thanks, Gerald
Re: What is /etc/.pwd.lock for ?
On Sat, Apr 24, 1999 at 10:23:53AM +0200, Sami Dalouche wrote: Is it not a lockfile for vipw or other similar utils which are used to change the /etc/passwd cleanly ? I don't think so. If it were, I would expect that vipw or other utils (eg., adduser) would delete it upon finishing alterations to the password file. This file seems to be constantly present. It is there on a Red Hat 5.2 system, on another debian system, and was on the system which prompted my initial message. Try 'ls -l /etc/.pwd.lock' and see if you don't have it on your system.
What is /etc/.pwd.lock for ?
Hi, I deleted /etc/.pwd.lock on a test box I play with. It was dated 15 apr. I thought it was just a stale lock file from the last time I created a user. Rebooting the same machine to an install of Red Hat, I discovered that the same file existed in /etc of the RH system. The time stamps on the files correspond to the dates I installed the respective systems. I think now that it was not a stale file, but can't find what purpose it serves. Its name would suggest it has something to do with passwords, but nonw of the man pages associated with passwd or shadow make mention of it. The shadow man pages speak of file locking without giving details (ie, not specifically naming the log file). Does anyone know what the file is for, or failing that, where I might look, aside from the above man pages, to learn more ? Thanks, Gerald
Re: new mail not visible in mail readers
On Fri, Apr 09, 1999 at 07:21:14PM -0400, Ben Collins wrote: On Sun, Apr 11, 1999 at 02:26:01PM +1000, Alan Eugene Davis wrote: I have been having a problem that when I fetch my mail using fetchmail, the new messages do not always appear, at least immediately. Sometimes, I have to run fetchmail a second time, then a few messages appear, and as I delete messages, new messages begin showing at the tail end of the buffer. I use mutt, but with af the mail isn't visible either. Any ideas? Could a lagging system clock cause such a problem? No, the mail does not immediately go into the mail folders. Incoming I think this depends on the MTA. When I used smail, the first message fetched didn't get delivered until all messages were fetched. The more messages there are on the remote server you're popping, the longer you wait for the messages to be delivered locally. However, delivery started as soon as the fetch was terminated. I recently switched to exim. exim starts local delivery right away. The first messages start appearing in mailboxes while more messages are still being popped. I am not a mail expert by any stretch of the imagination. Perhaps the above behaviour is configurable, but that is the way they behaved as installed out of the box from Deb 2.0. HTH, Gerald
Re: Script to check whether modem telephone line in use --- how?
On Thu, Apr 01, 1999 at 06:42:40PM +0930, Mark Phillips wrote: [major snip] a while. But from what Bud Rogers has said, it would seem the abort should happen straight away.) Yes, it should. I don't have problems with interfering with other modems here, but there are other people using the phone. If someone else is on the line, my modem will hang up immediately without making any noises aside from the the clicks of going off hook and then back on hook. Maybe your modem doesn't understand this. I can also confirm what others have already said, that not dialing is not gaurantee that you will not crash your fathers connection. If I am on line and someone picks up the phone, the noise is often, though not always, sufficient to cause my connection to drop. HTH, Gerald
Re: MTA and SMTP ident
On Sun, Mar 28, 1999 at 07:32:52PM +0200, Martin Bialasinski wrote: GC == G Crimp [EMAIL PROTECTED] writes: GC I think you are right, though, that it is the list software and GC not the MTA that is silently discarding my mail. Also note that the list software doesn't see the SMTP envelope (it is not passed on final delivery). GC My From: field is not a problem (thanks mutt !) but I'll give the GC Sender header a crack and see if that makes a difference. Good Luck. It is not failure to have the Sender different from the From:-header. Actually the Sender should be the correct mailbox of the sender, even if he changes the From-header. Maybe you don't have a registred hostname for your box, and the list software tries to resolve it and fails (or such some). No I don't have a registered name for my box, but my box's name doesn't appear anywhere, I have visible_name set to the name of my ISP. It is just the local user name that gets sent out (ie, my user name on my local box) and appended to my ISP domain. My local user name is not the same as my user name on any of my exteranl acounts, so the resulting combination is illegal. BTW, you have spoken a couple of times of the acceptable behaviour of mail agents in handling of From: and Sender: header etc. Where are you getting this information ? I have looked at 4 or 5 RFCs and have seen nothing of the sort. I have discovered a couple of sources of RFCs but none so far that does nice indexing according to topic. Any help in trying to find info ? Thanks again, Gerald
Re: MTA and SMTP ident
On Sun, Mar 28, 1999 at 07:32:52PM +0200, Martin Bialasinski wrote: GC == G Crimp [EMAIL PROTECTED] writes: GC I think you are right, though, that it is the list software and GC not the MTA that is silently discarding my mail. Also note that the list software doesn't see the SMTP envelope (it is not passed on final delivery). I am not sure about this, I can see the envelope From header on messages sent to the lists. GC My From: field is not a problem (thanks mutt !) but I'll give the GC Sender header a crack and see if that makes a difference. Good Luck. It is not failure to have the Sender different from the From:-header. Actually the Sender should be the correct mailbox of the sender, even if he changes the From-header. Maybe you don't have a registred hostname for your box, and the list software tries to resolve it and fails (or such some). I tried both telling mutt to change Sender: and Return-Path:. The mail is still not getting through. Can you tell me how the receiving end is getting the user name off my local box (a dial in ppp link, not part of any registered domain) [Reminder: on this box I am [EMAIL PROTECTED], at one of my pop servers I am [EMAIL PROTECTED] The Sender: Return-Path: etc headers are getting written [EMAIL PROTECTED] At first I thought is was the ident protocol (RFC1413), because I would see that in the Received: headers sometimes. However, I have told my local identd not to identify me, and my mail is still ending up with my local ID attached to the domainname of my ISP, so I think ident can't be the only culprit. Is is possible that the MAIL FROM: command in the SMTP protocol is responsible, and if so, is there any way to circumvent ? TNA, Gerald
Re: MTA and SMTP ident
On Sat, Mar 27, 1999 at 02:20:48PM +0200, Martin Bialasinski wrote: GC == G Crimp [EMAIL PROTECTED] writes: GC The From: header is okay, but the envelope From header gets GC written as [EMAIL PROTECTED] The my.isp.com is getting through GC because I could tell smail that my visible_name was the domain of GC my ISP and not the local machine name. GC Some mailing lists don't like the fact the From: field and the GC envelope don't agree and reject my mail. I believe you are wrong here. This would be damn stupid (BTW, the mailinglist might reject the message based on this, the receiving MTA however must not do this). I have no problems with something like that. Did you choose to send _all_ outgoing mail via your isp's smtp server (smarthost)? I have tried using both smart_host and routing mail myself. In both cases, by mail seems to get through OK except to certain mailing lists. I never get any bounces. My mail just quietly disappears. I have done a test where I telnet to the mail daemon at the destination, and I am able to go through the SMTP protocol manually and get my message through. I think you are right, though, that it is the list software and not the MTA that is silently discarding my mail. When flame wars erupt in one particular list, we go through periods of moderation. The moderator receives my mail OK and can forward it no problem, but on auto-pilot, my mail is dropped without exception. For me, I just set the visible_name and use the method described in http://www.debian.org/fom/137.html to change the From: and Sender: header (so I don't have to do it im every programm sending mails). My From: field is not a problem (thanks mutt !) but I'll give the Sender header a crack and see if that makes a difference.
MTA and SMTP ident
What do people on this list do when there local name is not the same as there login at their ISP ? My mail doesn't make it to some mailing lists because of the SMTP ident thingy. I'm talking about dial up users like myself, not those who are running servers and have their own domains, etc. For example, suppose my ID on my local Linux box is me_here but my ID at my ISP is me_there. I can set the From: header correctly, eg. [EMAIL PROTECTED] But, when an e-mail message gets sent, the receiving end writes an envelope From header using information my local MTA has sent along with the message. The From: header is okay, but the envelope From header gets written as [EMAIL PROTECTED] The my.isp.com is getting through because I could tell smail that my visible_name was the domain of my ISP and not the local machine name. It seems that MTA's look up the loginname of the user sending the mail, and pass that along with the message. The receiving end then uses that to write the envelope. Some mailing lists don't like the fact the From: field and the envelope don't agree and reject my mail. I can get others' posts, but can not post myself. Anyone know how I can work around this ? I am currently using smail, and had a look at exim, but couldn't find any feature that would allow this. Thanks, Gerald
Re: Problems installing Debian on a 486
On Tue, Mar 23, 1999 at 03:03:28PM -0800, Alan Bailward wrote: other machine, it brings up the boot: prompt, and then starts to load the root FS from root.bin.. After loading for a while it fails with the message A20 gate not responding!... [snip] The problem machine is a 486DX33 with 8MB of RAM, if you need more info about the hardware, please let me know. Could it be that the kernel on the rescue disk was compiled for 586+? Could that be causing it? I'm grasping I know, but... :) alan, out on a limb. The A20 line has something to do with working around a bug in memory addressing that first showed up in 286's I think. I don't really know much about it, but on my 486, their is an option in the BIOS to set the line. You might try changing this option in the Bios at boot and see if it helps. Gerald
Re: iso9660 module in kernel 2.0.34
On Tue, Mar 16, 1999 at 07:12:34PM -0600, Paul Miller wrote: G. Crimp wrote: Hi, Anybody know what an unresolved symbol is ? I've just compiled a This is a function call or variable that the modules wishes to use but cannot find. kernel making iso9660 support a module. I've done this in bo with no problems. In hamm, however, when I try to mount a cd, I get an error saying that the kernel does not support iso9660 filesystems. kerneld is running. If I do insmod isofs I get the following errors: - # insmod -p isofs /lib/modules/2.0.34/fs/isofs.o: unresolved symbol load_nls_default_R86e263f4 /lib/modules/2.0.34/fs/isofs.o: unresolved symbol unload_nls_R3f17924a /lib/modules/2.0.34/fs/isofs.o: unresolved symbol load_nls_Rbc00b63b /lib/modules/2.0.34/fs/isofs.o: unresolved symbol utf8_wctomb_Rf531b5d3 -- Try using modprobe instead of insmod. It will load other necessary modules for the requested module to run. modprobe isofs For modprobe to work, you need to build the dependancy list first. This happens every time the machine boots, but you can also do this by typing depmod -a something must be amiss somewhere, because I depmod is obviously not doing its job at boot. I even tried running depmod -a manually. anyway, I am running kerneld, so all this should be happening without my intervention should it not ? Gerald
iso9660 module in kernel 2.0.34
Hi, Anybody know what an unresolved symbol is ? I've just compiled a kernel making iso9660 support a module. I've done this in bo with no problems. In hamm, however, when I try to mount a cd, I get an error saying that the kernel does not support iso9660 filesystems. kerneld is running. If I do insmod isofs I get the following errors: - # insmod -p isofs /lib/modules/2.0.34/fs/isofs.o: unresolved symbol load_nls_default_R86e263f4 /lib/modules/2.0.34/fs/isofs.o: unresolved symbol unload_nls_R3f17924a /lib/modules/2.0.34/fs/isofs.o: unresolved symbol load_nls_Rbc00b63b /lib/modules/2.0.34/fs/isofs.o: unresolved symbol utf8_wctomb_Rf531b5d3 -- Seems to me I've done pretty much the same thing as in previous compiles, but it ain't workin' this time. Any answers ? Thanks, Gerald
kernel module dependencies
I compiled a kernel with NLS and iso9660 as modules. When I tried to mount a cd I got a message saying that iso9660 was not supported by the kernel. Tried installing iso9660 with insmod and got a bunch of unresolved symbol errors all relating to nls. If I run 'insmod nls' first and then try to mount the cd, it works great. I figure that kerneld should see that when mounting a cd, isofs is needed which in turn needs nls and deal with the dependencies. Am I right about this ? I didn't have this problem with kernel 2.0.30 although I couldn't swear that I made nls a module when I was compiling 2.0.30. If I am right about this, should I be reporting this as a bug to someone,and if so, to whom ? Thanks, Gerald
Re: gcc vs egcs
On Tue, Jan 19, 1999 at 04:54:19PM -0500, fantumn Steven Baker wrote: Hi Steve, Glad to see you are still amongst the ranks of Debian users. I am not the most knowledgable person on this subject, but when I say the name fantumn, I couldn't resist answering. You might get a better answer from someone else, but I will give it a try. Apparently, gcc and g++ development became fragmented as developers split into a number of camps. GNU gcc and g++ development (Deb gcc and g++ were the GNU versions) has halted, or at least slowed to a trickle. egcs is a movement to bring back together the various gcc/g++ development streams. The egcs versions of gcc and g++ are not both yet ready for prime time. I may get corrected on this, but I think it is gcc that still has some bugs, so Deb (2.0 at least) is still using GNU gcc. g++ on Deb is now the egcs version. So I don't think it is a question of one sucks and the other doesn't. It is more a case of c and c++ compiler development moving to egcs 'cause that is where development is happening. You might want to check out the Deb mailing list archives to get more informative detail than I can give. Yours from Victoria, Gerald Okay, I don't want to start a holy war or anything here, but I have some questions about egcs and gcc. First, was wondering _what_ the differences between gcc and egcs were. (Note: when I say gcc, I mean gcc _and_ g++) Compatibility: First, can egcs compile everything that gcc can? IE: kernel? Also, is g++ incapable of compiling programs written for compilation with egcs? (ie: Berlin)... Please include your own opinions, but please don't just say (egcs|gcc) sucks, and not provide reason. Sure, (egcs|gcc) sucks because... is acceptable. Thanks in advance... -fantumn -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null
booting from 2nd IDE interface
Hi all, I've a LILO question. The lilo docs say that MAYBE ones bios will allow booting from a drive on the second IDE interface (hdc of hdd). Apparently, I can as I have been booting Deb 1.3 from hdc. I have since installed Deb 2.0 to hdd, but when I make an appropriate entry in /etc/lilo.conf and run lilo, I get the following error: Warning: BIOS drive 0x82 may not be accessible And, in fact, I cannot boot the kernel on hdd. My question is therefore, if I can boot from hdc, should I not be able to boot from hdd ? If I remember correctly, 0x82 means major number 8, minor #2. This is sda2 on my box. I have no scsi devices. Anyone know what is going on ? Here is an excerpt from lilo.conf: begin excerpt -- boot=/dev/hda3 compact install=/boot/boot.b map=/boot/map vga=extended delay=30 image=/vmlinuz label=Linux root=/dev/hdc1 read-only image=/mnt/boot/vmlinuz-2.0.34 label=deb2 root=/dev/hdd1 vga=ask read-only -- end excerpt Thanks, Gerald
smail -oMP
I am having trouble with my smail configuration. One of the things I am looking at is how smail is being invoked. In /var/spool/smail/input/stupid_message_that_for_some_reason_refuses_\ to_leave_my_box I see that three options are used (if I am interpreting this correctly) -t -oem and -oMP. I can find -t and -oem in man(8) smail, but not -oMP. Does anyone know what this switch means ? Thanx, Gerald
no sound until I run doom
I need help with sound. If I want to run a music CD (internal CD drive) or a cassette (from a walkman plugged into the input jack on the back of the sound card) I have first to run doom. If I don't, I can't hear a thing. This is on Deb 1.3. I assume the same thing is happening on Deb 2.0. I don't have doom installed on Deb 2.0, so I can't test it, but I currently can't get any sound out in 2.0. Anyone know what I have to do to remedy this ? As a tack on sound question, when I try to run Xmixer or aumix from the fvwm2 menus (Apps - Sound), at best I see a window flash briefly to life on the desk top, only to disappear, at worst I see nothing. The same thing happens on Deb2.0. I have more sound utilities, like CD player consoles, on 2.0 that do the same thing. XPlaycd works but the others don't (can't tell you which ones, 'cause I am writing this from Deb1.3. Don't have mail configured on 2.0, yet.) Any suggestions ? thanks for any help, Gerald Crimp
Re: [Off Topic] An EXCELLENT Microsoft Confidential document on
On Thu, Nov 05, 1998 at 09:28:45AM -0200, Vera Lucia Mazzocchi wrote: Hi, I was thinking if this document wasn't released purposely, that is, let them know only what we want they know and let's see what they think about. Seems to me that M$ might be playing with us, releasing such document, or allowing somebody to have access and releasing it. Doesn't this sound to ease? To have access to a confidential document from M$ exactly about OSS? I have some doubts ... just thinking... I don't think I doubt its authenticity or sincerity, but my guess is the author of the memo is not a major M$ player. His personnal home page is linked off the Linux Weekly News site (http://lwn.net). This leak comes curious close to the time where BG is trying to convince the court in the anti-trust case against him, that M$ is not the only player, that there alternatives out there. He needs Linux to be present and to appear to the court to be a major threat to 'Doze. Could work in their favour. On the other hand, if it creates negative publicity and heat in the court, it would be easy to distance themselves from it as being the mere enthusiastic meanderings of a junior underling and not official corporate policy nor practice. Have a look at his site. He doesn't come across as a big corporate savvy suit. Ta, Gerald Crimp
Re: [Off Topic] An EXCELLENT Microsoft Confidential document on
On Thu, Nov 05, 1998 at 05:02:43PM +, Thomas Lakofski wrote: On Thu, 5 Nov 1998 [EMAIL PROTECTED] wrote: Me too! I'm usually a very cautious person...How do we *know* that this has even originated from MicroSoft? So there is the issue of whether or not it's from MS, and if it is it it truly confidential? It was confirmed by MS. See slashdot.org somewhere... And the Wall Street Journal of Tues., page B4. Gerald
Re: [Off Topic] An EXCELLENT Microsoft Confidential document on
On Fri, Nov 06, 1998 at 12:25:27PM +1300, [EMAIL PROTECTED] wrote: Also, it's unlikely that this was deliberatly leaked by microsoft (IMHO), because it's not exactly good publicity for them. Far more PR damage is possible because of (admittedly somewhat backhanded) admissions of how well the Free Software community is doing and the subtefuge they are planning to combat it than any gain they could get from seeing our relations to it. Rememeber it's not only Free Software enthusiasts and microsoft haters who will be reading it. Ah, but they were SO quick to confirm that it was an official memo with SO little effort put into trying to downplay its significance and distance themselves from what seems to be damaging. I think they could be very well playing this. Gerald Crimp
Slightly off topic: FUD ??
I've just been reading an interesting article by Eric S. Raymond (_The Cathedral and the Bazaar_, http://www.tuxedo.org/~esr/writings/cathedral-bazaar). In it he is commenting on what he believes to be an authentic internal M$ document talking about the challenge presented by the open source movement. Anyway, my question for the list is, does anyone know what FUD is. I have recently seen this acronym floating around in relation to how big companies like micromush are going to combat against the growing threat posed by open source. These references are always saying stuff like normal FUD tactics won't work. I feel like I'm missing just a little bit of the flavour of these articles, so if anyone can enlighten me I'd be really FUDing happy :). BTW, the article I said I was reading is at http://www.tuxedo.org/~esr/halloween.html TIA, Gerald
Re: apropos in man corrupted ! - howto fix ?
On Wed, Oct 28, 1998 at 10:59:50AM -0800, Geoffrey L. Brimhall wrote: At some point, my apropos has gotten corrupted so that it never returns anything. I've tried un-installing man-db, and re-installing it, but it apropos still does not work. Though I'm not sure if this has anything to do with it, when I issue 'catman' as super-user, I get the following error: I don't know if this is related to a man problem I had or not. Trying to use man with any command caused a segmentation fault. I was told on this list to do `mandb --create'. This process generated many error messages, but man started working again. HTH, Gerald
Re: Linux on 4Mb
On Tue, Oct 27, 1998 at 11:49:08AM +0100, Bostjan JERKO wrote: Is it possible to install Linux on machine with 4 Mb and of course it should work. Bostjan Don't know about Debian 2.0, but I installed Debian 1.3 on a 486sx 25MHz with 5 MB of RAM and a ~110MB harddrive. I had about 16MB of memory after using about 12MB of the harddrive for a swap partition. And swap it did, especially when running X. With a fairly minimal install that included X-Window I used about 60MB of the harddrive. When I added the kernel sources and a graphical card game, about another 15 MB of disk space were eaten up. It worked, but couldn't do much with it except play cards :) Ta, Gerald
Re: lilo - booting from a device
Thanks ! This even worked for a bootable partition on hdd =:¬) (except that my bios was not then able to load from hdd |:¬( ). On Wed, Oct 14, 1998 at 12:16:00AM -0400, Richardson,Anthony wrote: I don't know the answer to your question, but I've got an alternative approach. Mount the boot partition of the new distribution: mount /dev/hda1 /mnt Then add the following stanza (it's based on your entry below) to lilo.conf: image=/mnt/vmlinuz label=newdistro root=/dev/hda12 Then lilo will figure out the sector to file mapping for you. Tony On Tuesday, October 13, 1998 1:32 AM, G. Crimp [SMTP:[EMAIL PROTECTED] wrote: Does anyone know how to determine what sectors on a partition a given file is occupying ? The lilo docs state that when specifying a device name, one also has to say what sectors to map (either by giving a beginning and an end, or a beginning and an offset). Here is what I am trying to do. I want to play with different distributions just to see what they are like, but boot them from the harddrive. I currently have Deb installed to hda3. hda1 and hda2 are 5MB partitions, on which I install /boot from another distro, so that the kernel images within fall well below the 1024 cyl. limit. I am not at the machine I am doing this on, nor are they networked, so the following details may not be 100% accurate (syntax wise). This box has no DOS. Lilo is installed directly to the MBR. So, the lilo.conf looks something like --- boot=/dev/hda [prompt, delay, timeout, etc.] image=/vmlinuz label=olddistro root=/dev/hda3 -- to which I want to add something like - image=/dev/hda1 # this is the partition /w kernel of new distro label=newdistro root=/dev/hda12 #this is / of new distro sectors[or whatever the variable name is]= For floppies I've seen that it is something like 1+512, so I guess the image must start on the second sector of the floppy. there must be someway to determine the sector info for any file. Any ideas ? TIA. Gerald -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null
lilo - booting from a device
Does anyone know how to determine what sectors on a partition a given file is occupying ? The lilo docs state that when specifying a device name, one also has to say what sectors to map (either by giving a beginning and an end, or a beginning and an offset). Here is what I am trying to do. I want to play with different distributions just to see what they are like, but boot them from the harddrive. I currently have Deb installed to hda3. hda1 and hda2 are 5MB partitions, on which I install /boot from another distro, so that the kernel images within fall well below the 1024 cyl. limit. I am not at the machine I am doing this on, nor are they networked, so the following details may not be 100% accurate (syntax wise). This box has no DOS. Lilo is installed directly to the MBR. So, the lilo.conf looks something like --- boot=/dev/hda [prompt, delay, timeout, etc.] image=/vmlinuz label=olddistro root=/dev/hda3 -- to which I want to add something like - image=/dev/hda1 # this is the partition /w kernel of new distro label=newdistro root=/dev/hda12 #this is / of new distro sectors[or whatever the variable name is]= For floppies I've seen that it is something like 1+512, so I guess the image must start on the second sector of the floppy. there must be someway to determine the sector info for any file. Any ideas ? TIA. Gerald
Re: Login from a Terminal
Hi, I've a friend who just installed Debian 2.0 on a 486. He reads through a talking terminal. Under a Slackware (don't know what version) he used to type agetty 9600 ttyS0 vt100. He says this doesn't work with the new installation. I don't know anything about getty. A look at the man page suggests that this is an appropriate command. Does anyone know what he needs to do to get this to work ? Thanks, Gerald
re-rebooting during installation
Hi, Question first, details follow for those that want them. Question: How can I get Deb to present me with the profile selection screen that one gets during installation of Deb 2.0, after one has partitioned, initialized, configured the net and the base system and rebooted ? You know, where one decides if one wants to create an Admin box, stand alone box, home box, development box, etc. Details: I'm just having a go at installing 2.0 from an official CD image. Got through all the initial installation routine stuff, rebooted, root password etc, etc. Then I got to the part where you pick a profile. I went and had a look at custom, then wanted to go back and have a look at some of the other profiles. I hit cancel at the custom selection screen. It didn't take me back to where I could select another profile. I didn't want to go straight to dselect, so I rebooted. We are having a nation wide installfest in Canada next weekend. I'm doing my part and I want to know more about the various profiles so I can better help installees go home with an appropriate configuration. When I rebooted, I was sent directly to a login as if installation was complete. I guess it was complete, but I want to play more with the installation procedure and profile selection. If anyone can tell me what the mechanism is to restart the root password selection, user account creation, profile selection routine, it would be greatly appreciated. I don't really want to have to go back and redo the whole install thing. Thanks, Gerald
diald and kernel problems
Hi, I trying to set up diald. I had to recompile my kernel to enable slip support. I made it a module. I did make mrproper, make config, make dep, make clean, make zImage, make modules, make modules-install. It looked to me as though they all terminated normally. I can boot from the new kernel no problems. The boot up messages tell me that diald is being started. However, `ps -ax | grep diald' reveals nothing. /var/log/daemon.log tells me: Sep 13 02:55:46 modprobe: no dependency information for module: /lib/modules/2.0.30/net/slip.o In /var/log/messages: Sep 13 02:55:47 diald[112]: SLIP not supported by kernel, can't build proxy. diald[112]: Diald is dieing with code 1 I don't know if this is pertinent, but in /var/log/degub... Sep 13 02:42:10 kernel: Max size:324376 Log zone size:2048 kernel: First datazone:28 Root inode number 57344 /var/log/diald is empty. Can anyone tell me what I did wrong with SLIP ? Here's the configure details from /usr/src/linux/.config: # Network device support # CONFIG_NETDEVICES=y CONFIG_DUMMY=m # CONFIG_EQUALIZER is not set CONFIG_PLIP=m CONFIG_PPP=m # # CCP compressors for PPP are only built as modules. # CONFIG_SLIP=m CONFIG_SLIP_COMPRESSED=y # CONFIG_SLIP_SMART is not set # CONFIG_SLIP_MODE_SLIP6 is not set Thanks, Gerald
Re: man: Segmentation fault
On Sat, Aug 22, 1998 at 10:37:29AM +1200, Michael Beattie wrote: Could be as simple as a corrupt database. try `mandb --create` Wooow ! Tried it. Did get use of man back (Thanks !), but got many warnings (see below). And anyway, what would cause an otherwise sweet and innocent DB to become corrupt in the first place ? Are there some underlying problems that I should be worried about ? Thanks, Gerald mandb: warning: /usr/man/man8/in.smtpd.8: whatis parse for in.smtpd(8) failed whatis parse for mailq(8) failed whatis parse for sendmail(8) failed whatis parse for runq(8) failed whatis parse for rmail(8) failed whatis parse for rsmtp(8) failed whatis parse for qcrack(1) failed whatis parse for endian(1) failed whatis parse for qext(1) failed whatis parse for qinit(1) failed whatis parse for uupath(1) failed whatis parse for ps2frag(1) failed mandb: warning: /usr/man/man3/man.tmp: ignoring bogus filename ignoring bogus filename whatis parse for paperdone(3paper) failed whatis parse for paperfirst(3paper) failed whatis parse for paperwithsize(3paper) failed whatis parse for paperlast(3paper) failed whatis parse for defaultpapername(3paper) failed whatis parse for papernext(3paper) failed whatis parse for paperprev(3paper) failed whatis parse for systempapersizefile(3paper)failed whatis parse for ioctl_list(2) failed /usr/man/man7/undocumented.7*: competing extensions whatis parse for groff_mmse(7) failed Checking for stray cats under /usr/man... Checking for stray cats under /var/catman... Processing manual pages under /usr/local/man... Checking for stray cats under /usr/local/man... Checking for stray cats under /var/catman/local... Processing manual pages under /usr/X11R6/man... Checking for stray cats under /usr/X11R6/man... Checking for stray cats under /var/catman/X11R6... 27 man subdirectories contained newer manual pages. 1880 manual pages and 0 stray cats were added. # mandb: warning: /usr/man/man
man: Segmentation fault
Hi, I can't use 'man' anymore. It doesn't dump core, but I do get a Segmentation fault in X as on the console, when I am root as when I am joeuser. I can't recall having made any changes to the system except to have used dselect to install the dxpc package. Oh yeah, and a while ago I set up a small script to set up a ppp connection with another pc through the serial ports using a null-modem cable. It isn't impossible that I have made another change, but it will be minor like having added or removed some package with dselect. With only Segmentation fault as a message, I don't have the slightest little crumb of a notion what to look at. Can anyone guide me through some detective work ? Thanks, Gerald
THANKS ! RE: minimal files essential for booting ?
Thanks to all who helped on this one!! Indeed, it was /lib that was missing during boot. I repartitioned the disk putting /etc, /bin and /lib back on the / partition, and everything is hunky dorry. I am now set to enjoy the freedom acquired by moving from a 115 MB drive to a 6.1 GB drive. Ta, Gerald On Thu, Aug 06, 1998 at 01:02:55AM -0700, G. Crimp wrote: Sorry to be a pain with this, but I want to try one more time for help. I recently copied my file system over to a new bigger disk, but I can't boot to the new disk. Question and hypothesis first, explanation following. I went nuts partitioning the new disk. I was mostly just experimenting. Many will think I have gone needlessly overboard. I won't disagree. The old disk has two partitions, one being swap. The new disk has a partition for just about everything. These directories all live on their own partitions: /usr /usr/local /var /home /etc /bin /tmp /lib I'm thinking now that the boot sequence needs something from /etc, /bin or /lib that it can't find 'cause it hasn't been mounted yet (see below). That's why I am wondering if anyone can tell me what files, configuration, executable or otherwise are essential for booting the system. I'd be happy to RTFM if someone can tell me what manual I should be reading. I had a look at the howto index. The section on the boot sequence in the boot disk howto gave some insights but not enough. Can't think where else to look for info. Steps taken, explanation and things tried follows. Before anyone asks, the copy was not the problem. I used a tar mostly and cpio for the device files once I discovered that tar was not going to work. I copied directory by directory and for each directory under /, I did 'ls -lAFR file' where file was either oldfs or newfs. I then did a diff on the two files. The only differences were things like number of used blocks, date stamps and a ton of pairs of directory names because one of each pair was prefixed with /mnt. There were no differences (after a few tries :-)) in the mode bits nor the ownership. Other steps taken: 1) copied kernel image to floppy using instructions for compiling kernel found in the source tree; 2) edited fstab and lilo.conf(even though I wasn't counting on booting from the hard drive right off the bat); 3) made the new drive master; 4) changed the cmos configuration (not really necessary); When let 'er rip, I only get as far as this message in the boot sequence: VFS: root partition (ext2 filesystem) mounted read-only. If I boot from the old drive, the next message is : INIT: something else that flashes by too quickly to be read. I tried copying the contents of /bin and /etc to root partition instead of having them on their own partitions, but that didn't help. Anyway, init is in /sbin which is on the root partition anyway. Should I have /lib on the root partition too ? Thanks for any more help. Perplexed, Gerald -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null
minimal files essential for booting ?
Sorry to be a pain with this, but I want to try one more time for help. I recently copied my file system over to a new bigger disk, but I can't boot to the new disk. Question and hypothesis first, explanation following. I went nuts partitioning the new disk. I was mostly just experimenting. Many will think I have gone needlessly overboard. I won't disagree. The old disk has two partitions, one being swap. The new disk has a partition for just about everything. These directories all live on their own partitions: /usr /usr/local /var /home /etc /bin /tmp /lib I'm thinking now that the boot sequence needs something from /etc, /bin or /lib that it can't find 'cause it hasn't been mounted yet (see below). That's why I am wondering if anyone can tell me what files, configuration, executable or otherwise are essential for booting the system. I'd be happy to RTFM if someone can tell me what manual I should be reading. I had a look at the howto index. The section on the boot sequence in the boot disk howto gave some insights but not enough. Can't think where else to look for info. Steps taken, explanation and things tried follows. Before anyone asks, the copy was not the problem. I used a tar mostly and cpio for the device files once I discovered that tar was not going to work. I copied directory by directory and for each directory under /, I did 'ls -lAFR file' where file was either oldfs or newfs. I then did a diff on the two files. The only differences were things like number of used blocks, date stamps and a ton of pairs of directory names because one of each pair was prefixed with /mnt. There were no differences (after a few tries :-)) in the mode bits nor the ownership. Other steps taken: 1) copied kernel image to floppy using instructions for compiling kernel found in the source tree; 2) edited fstab and lilo.conf(even though I wasn't counting on booting from the hard drive right off the bat); 3) made the new drive master; 4) changed the cmos configuration (not really necessary); When let 'er rip, I only get as far as this message in the boot sequence: VFS: root partition (ext2 filesystem) mounted read-only. If I boot from the old drive, the next message is : INIT: something else that flashes by too quickly to be read. I tried copying the contents of /bin and /etc to root partition instead of having them on their own partitions, but that didn't help. Anyway, init is in /sbin which is on the root partition anyway. Should I have /lib on the root partition too ? Thanks for any more help. Perplexed, Gerald -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null
boot: unable to open initial console
I've moved my old system over to a new hard drive. I thought I was fairly meticulous in copying things over (compared all permissions, ownerships, links and other special files between the two file systems). I edited lilo.conf and fstab, then shutdown, changed the drives master/slave jumpers and rebooted from the rescue floppy. I used the install program to mount all the partitions and to mkswap. Next I switched to the console and copied my kernel to a floppy. Evidently, I have overlooked something. I only get through the first little bit of the boot sequence (I can write out the whole thing , one screen full, if anyone wants it) up to where all the partitions are named and VHS: Mounted root (ext2 filesystem) readonly. thenUnable to open an initial console Does anyone know what I need to do to fix this ? Thanks, Gerald -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null
SOLVED: Re: fdisk and fujitsu drives
Got this figured out, sort of. I don't know why, but fdisk seems to act differently for the fujitsu drives as compared to Western Digital, Samsung, Seagate and O'Connor drives I have or have had. I always just accepted defaults for the first cylinder when creating new partitions. Had to explicitly type them out when partitioning the fujitsu. Don't yet have the final word on it working but I am optimistic as there were no warning about bad data starts and partitions not ending on boundaries. Ta. On Thu, Jul 30, 1998 at 09:27:44PM -0700, G. Crimp wrote: I've got a 6.4 GB fujitsu drive that I am trying partition. Actually, I had it partitioned, formatted (no reported errors), and file system from an old disk transferred to it. However, I made it the master after the fs transfer, and was using a rescue floppy to boot, mount all the partitions and set up LILO, but LILO setup failed with: Device 0X0300: Invalid partition table, 3rd entry 3D address: 1/0/78 (73710) linear address: 55/11/21 (20592) fdisk reported the first four partitions not ending on cylinder boundaries. Also, the beginning, starting and ending cylinders are all wonky. Eg., begin start end typesize part 11 1 11 native 5M part 240 11 22 native 5M part 379 22 239 native 100M part 4855 239 2859extendd rest of disk part 5855 239 889 native 300M part 61024889 1539native 300M part 7102415392189native 300M part 8202421892623native 200M So, v (verify) gives all kinds of errors (end of cylinders not on boundaries, bad start of data, partitions overlapping). I deleted all the partitions and started over which fixed up the beginning cylinders on parts 1-4 but not the rest. ( I didn't write the partition table so I still have the old setup) QUESTIONS: 1) Does this matter ? (apparently it does to LILO) I was able to mke2fs -c all partitions no problem, and copy over the old file system. and if it does matter, 2a) Anyone know what is causing this ? 2b) I can do to fix it ? I suspect this might be a logical/physical geometry thing. Linux required me to explicitly state the physical geometry of a 1.2 Gig Seagate (now a paper weight) before it would install on it. However, I have a Western Digital 3.1 Gig in another box that Linux happily installed on without the physical parameters. All help appreciated. Thanks, Gerald -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null
Re: boot: unable to open initial console
On Mon, Aug 03, 1998 at 02:58:58PM +0200, Dirk Bonne wrote: G. Crimp wrote: I've moved my old system over to a new hard drive. I thought I was fairly meticulous in copying things over (compared all permissions, ownerships, links and other special files between the two file systems). I edited lilo.conf and fstab, then shutdown, changed the drives master/slave jumpers and rebooted from the rescue floppy. I used the install program to mount all the partitions and to mkswap. Next I switched to the console and copied my kernel to a floppy. Evidently, I have overlooked something. I only get through the first little bit of the boot sequence (I can write out the whole thing , one screen full, if anyone wants it) up to where all the partitions are named and VHS: Mounted root (ext2 filesystem) readonly. thenUnable to open an initial console Looks like /dev/tty0 isn't there. Had a look. :( It's there. Symlink to console. Console is there, too. Permissions, ownership and major/minor numbers are exactly the same between the original and cloned tty0 and console. The only that is different is the time stamp. Thanks, Gerald -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null
Re: boot: unable to open initial console
On Mon, Aug 03, 1998 at 05:52:36PM +0200, Dirk Bonne wrote: Taren wrote: [snip] Or the permissions aren't set right. I've found that when copying files from /dev, the permissions rarely stay the way they were originally They should with tar. e.g.: tar cf - . | (cd somewhere; tar xvf -) Of course, you must set umask to 000 beforehand Actually, I tried pretty much the same command as you suggest: tar -cf - $dir | (cd /mnt; tar -xpsf -) This worked well for ordinary files and some special files, but any sockets in /dev or /var were turned into pipes. At least, the first mode bit changed from an 's' to a 'p'. I don't think the device numbers changed, but the mode bit change was enough to scare me. I used cpio in /dev and /var and it seemed to preserve everything. But I still have my Can't open initial console problem. Maybe all the device files got moved fine as far as ls -l is concerned but not on a more fundamental and functional level. Perhaps I should just cd /mnt/dev; rm * and then remake the device files. I've read about a command to make the devices. Can't remember where nor what the command was. Anyone know where to go to get info on recreating the device files ? Thanks, Gerald -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null
Re: boot: unable to open initial console
On Mon, Aug 03, 1998 at 05:58:46AM -0700, Taren wrote: VHS: Mounted root (ext2 filesystem) readonly. thenUnable to open an initial console Looks like /dev/tty0 isn't there. Are use sure you did copy the devices correctly (cp does not copy devices correctly, use tar or cpio). Dirk Or the permissions aren't set right. I've found that when copying files from /dev, the permissions rarely stay the way they were originally I'd checked that already. I used cpio to copy the dev and var directories. All mode bits and ownership are intact. Only the time stamp is different between the original and copied tty0 and console. I think more than likely the problem is somewhere else. I've copied everything over, but the partition structure is not the same between the two drives. In fact, on the old drive, there was only one partition (except for the swap of course). I've broken this down on the new disk. Also, root is partition 3 whereas it was 1 on the old disk (I'm not using the first two partitions for Debian). I'm thinking that there is some boot script that got blindly copied over to the new disk, and is still looking on hda1 from something that is now on a different partition. Kind of like if I'd forgotten to change the boot= and root= in lilo.conf I don't understand the Linux boot procedure very well, so I don't know where to start to look. Any more ideas out there? Thanks, Gerald -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null
fdisk and fujitsu drives
I've got a 6.4 GB fujitsu drive that I am trying partition. Actually, I had it partitioned, formatted (no reported errors), and file system from an old disk transferred to it. However, I made it the master after the fs transfer, and was using a rescue floppy to boot, mount all the partitions and set up LILO, but LILO setup failed with: Device 0X0300: Invalid partition table, 3rd entry 3D address: 1/0/78 (73710) linear address: 55/11/21 (20592) fdisk reported the first four partitions not ending on cylinder boundaries. Also, the beginning, starting and ending cylinders are all wonky. Eg., begin start end typesize part 1 1 1 11 native 5M part 2 40 11 22 native 5M part 3 79 22 239 native 100M part 4 855 239 2859extendd rest of disk part 5 855 239 889 native 300M part 6 1024889 1539native 300M part 7 102415392189native 300M part 8 202421892623native 200M So, v (verify) gives all kinds of errors (end of cylinders not on boundaries, bad start of data, partitions overlapping). I deleted all the partitions and started over which fixed up the beginning cylinders on parts 1-4 but not the rest. ( I didn't write the partition table so I still have the old setup) QUESTIONS: 1) Does this matter ? (apparently it does to LILO) I was able to mke2fs -c all partitions no problem, and copy over the old file system. and if it does matter, 2a) Anyone know what is causing this ? 2b) I can do to fix it ? I suspect this might be a logical/physical geometry thing. Linux required me to explicitly state the physical geometry of a 1.2 Gig Seagate (now a paper weight) before it would install on it. However, I have a Western Digital 3.1 Gig in another box that Linux happily installed on without the physical parameters. All help appreciated. Thanks, Gerald -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null
srwxrwxrwx prwxrwxrwx
I'm transferring an existing Debian installation from an small disk to a new big one. I used tar to move all the files over. For the most part all the permissions and other attributes were preserved, but a few were not. Most notably all sockets were made into pipes in /dev and in /var/run. Does this matter ? If so, any ideas how to change them back to srwxrwxrwx ? Thanks, Gerald Crimp -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null
lexmark printers
There are some pretty good deals on Lexmark printers at my University. They use PCL5 and PS emulation. Anybody have any experience with Lexmark's emulation and Linux ? I've heard that the emulation isn't perfect. Since they don't support Linux, they are not likely to offer any solutions. Any one come across any notable problems with the printer output ? Thanks, Gerald Crimp -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null
mke2fs - bad blocks
Hi, I got a small used disk given to me that I am trying to put into a small system I have. When I tried to run mke2fs on any of the partitions I had created I get Checking fro bad blocks (read-only test): Bad block 0 out of range;ignored. done Block 1 in primary superblock/group descriptor area bad. Blocks 1 through 3 must be good in order to build a filesystem. Aborting Similarly, mkswap gives: --- 4120 bad pages mkswap: fatal: first page unreadable --- I suspect the drive is toast, but thought I would check first. Does anyone know if a low level format or something else can save this, or is it just garbage ? Thanks, Gerald Crimp -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null
removing kernel sources
Hi, I have the kernel source tree on a machine with very limited space. I have compiled from it but it is just plain occupying too much space. I want to get rid of it now. Just about anything that could be done as a module was. Following the instructions in the README file in the root of the src tree I have made /usr/include/asm /usr/include/linux and /usr/include/scsi symlinks to the respective include directories of the src tree (oh, BTW this is 2.0.30). Q: Can I 1) restore these three directories by deleting the symlinks doing mkdir /usr/include/x and then copying from the pointed to src directories to the newly recreated /usr/include/ directories and then 2) delete the source tree and still have a bootable runnable system ? I guess it is the module stuff that worries me the most. Is there anything in the source tree, now that I have compiled from it a couple of times, that is necessary to the continued health of the box ? Thanks, Gerald Crimp -- Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null
More help, pls Re: simple interface/low-power users
On Tue, May 26, 1998 at 08:16:52PM -0700, Darren Benham wrote: On 27-May-98 G. Crimp wrote: I thought it would be great if I could set up their accounts so that when they logged in a button bar greated them. They could invoke xpat2, logout, shutdown, or call up a graphic mail user agent (maybe xfmail or netscape) just by clicking the appropriate button. Here is where I need ideas and guidance. If you use X (xdm) and one of the flavors of fvwm*, you can use FvwmButtons to give them a button bar on the screen when X starts up. Thanks to all who sent suggestions. I haven't had much time to work on this so its a little slow. The route I have decided to try is as above. I am using fvwm2 and xdm, so it is just a question of getting FvwmButtons going ... That's why I am writing again. I've been through man fvwm2, man FvwmButtons, /etc/X11/fvwm2/system.fvwm2rc, /usr/doc/fvwm2/README.sysrc, etc, etc, and I can't get a button box to come up. I have created a ~/.fvwm2 directory in which I have two files: init-restart.hook and post.hook. Their contents follow after which I'll decipher a little. init-restart.hook: ### ## ## ~/.fvwm2/init-restart.hook ## ### ## Load any modules which should be started during fvwm ## initialization. #ModulePath /usr/lib/X11/fvwm:/usr/bin/X11:/usr/X11R6/lib/X11/fvwm2/ ## Make sure FvwmButtons and FvwmPager are always there. #AddToFunc InitFunction I Module FvwmButtons #AddToFunc RestartFunction I Module FvwmButtons #Module FvwmButtons #+ I Module /usr/X11R6/lib/X11/fvwm2/FvwmButtons + I Exec FvwmButtons + I Module FvwmButtons ## Make it titlebar-less, sticky, and give it an icon Style FvwmButtons Icon toolbox.xpm, NoTitle, Sticky ## Alright, let's configure this puppy. *FvwmButtonsFore White *FvwmButtonsBack DarkSlateGrey *FvwmButtonsColumns 1 *FvwmButtonsGeometry -0-0 *FvwmButtonsFont -adobe-helvetica-bold-r-*-*-14-* *FvwmButtons(Title Shutdown, Back LightRed, Fore Black, Action 'sudo shutdown -h now') *FvwmButtons(Title Solitaire, Fore Black, Action 'xpat2 -geometry 700x450-0+0 ') #Module FvwmButtons end of init-restart.hook post.hook: ## Make it titlebar-less, sticky, and give it an icon Style FvwmButtons Icon toolbox.xpm, NoTitle, Sticky ## Alright, let's configure this puppy. *FvwmButtonsFore White *FvwmButtonsBack DarkSlateGrey *FvwmButtonsColumns 1 *FvwmButtonsGeometry -0-0 *FvwmButtonsFont -adobe-helvetica-bold-r-*-*-14-* *FvwmButtons(Title Shutdown, Back LightRed, Fore Black, Action 'sudo shutdown -h now') *FvwmButtons(Title Solitaire, Fore Black, Action 'xpat2 -geometry 700x450-0+0 ') #Module FvwmButtons --- end of post.hook - You can see a bit what I have tried from what is commented out. My first attempt was to put everything in init-restart.hook. I tried a couple of ways of adding the buttons to the init and restart functions. I have tried adding adding Module FvwmButtons on a separate line and in the post.hook file. After many fruitless experiments at little variations, I tried putting the configuration stuff in the post.hook file as well. Nothing I have tried has given the slightest hint of success. All that starts is the pager and an xterm window. ANYone know what the heck I am doing wrong (ie., how to make this right ) Thanks a million, Gerald Crimp -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
creating simple interface for low-power users
Hi, I'm trolling for ideas and guidance. I'm trying to help my parents enter the electronic age. They basically know nothing about computers. They're 80 and I'm stubborn. This means that things have to be super easy for them, and I'm hoping to avoid windoze. So far I have set up a 486 with Deb 1.3 for them. They are starting with xpat2 to learn how to use a mouse. The mouse part is going OK. Typing commands is proving to be quite a hurdle. For example, I've set up a couple of aliases so they can invoke xpat2 (they'll never remember that) and shutdown the computer (using sudoers). The notion of a command line and shell are proving troubling, but it is not just the command line where keyboard commands are difficult. One of my ultimate goals is to get them using e-mail. They've seen pine and mutt. Control key command sequences are discouraging them. I think my only hope of keeping them interested is by pointing and clicking. I thought it would be great if I could set up their accounts so that when they logged in a button bar greated them. They could invoke xpat2, logout, shutdown, or call up a graphic mail user agent (maybe xfmail or netscape) just by clicking the appropriate button. Here is where I need ideas and guidance. I think this should be feasible. I would make this button bar their login shell. All that would remain would be to create the button bar. I don't know where to go from here. I have taken some programming classes and diddled with various languages, but have never done anything graphical. I have heard of various widget sets like tcl/tk, motif, lesstif, etc., but don't know which might be the most appropriate. I would love to delve into this but I need a starting point. Any feedback ? Is there a better way to go ? TIA, Gerald Crimp -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Trantor 128F SCSI - FDD, HDD controller failure
Hi, I just acquired a used Trantor 128F SCSI adapter that I am trying to install in a 486sx-33 to run a Maxtor lxt-340sy harddrive. The card has onboard bios that can be enabled/disabled. The computer bios is AMI (06/06/92). I have set the jumpers and dipps on the drive and card to what I think should be reasonable values according to what I have so far read about SCSI. After the memory check and a couple of seconds of who knows what, and after the SCSI bios message if I have it enabled, I get FDD controller failure HDD controller failure press F1 to resume If I press F1 I get that nice little table that gives a run down of what is in the box, three lines about memory, shadow memory and something else, and them I am dead in the water because there is neither a floppy nor a hd from which to boot. I'm not really sure what to look for or try. I did try changing the IRQ of the board ( I picked 10 because I don't think anything uses it) but I got the same message I got about a thousand times before. I have made some progress in ajusting the dipp switches (I didn't event get the option of F1 to continue at first). I don't know enought about the start up of the box to know what is causing this. Can anyone give me an idea of what to look at ? Thanks, Gerald Crimp -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Help making a 486 into an X Terminal
On Wed, May 20, 1998 at 12:32:10AM -0400, Kiyan Azarbar wrote: Is this possible? If so, what is the bare minimum I need. How do I get started? Is there a bootdisk package? I can copy any debs I need from my computer over to the share drive on the win95 machine, but I'm wondering how I should go about creating a partition for ext2fs, and how to install the kernel, etc. Should I just compile a tiny minimalist kernel there? And most importantly, where do I get fips? I can't help you with the network aspects of your desired setup, but I thought I would share my playing around with a small set up to give you an idea of space requirements. I set up Deb with X on a 486sx-33, 5 Mb RAM, and a HD with something like 115 or 120 Mb capacity. I set aside 12 Mb for a swap partition which left about 100 or so Mb for a Linux partition. From dselect, I then picked out what I thought would be the bare minimum to get a running system. This was a trial and error process, at least for me as I sometimes think I won't need a package but some other package does need it. I ended up with a running box on which I could start X with fvwm2. df reported 60% disk usage. Of course about the only thing I had on the machine that one could pretend to do anything with was a text editor and minicom. With 5 megs of RAM I couldn't complain that it was too fast for me to keep up with, but the machine was only slightly slower with X than it in its previous life running M$ Windows 3.1 with 4 megs of RAM. The machine is fast enough that I decided I could use it to try and teach my parents how to use a computer. The current setup is a reincarnation as the old Deb (1.1) was wiped to remake it a DOS box for awhile. I can't say how exactly the chosen packages match the new 1.3 setup, but I again aimed for a minimal setup. joe and minicom are there again, but I added the full source tree to make a custom kernel, and xpat2 so mater and pater could have some fun while learning to use a mouse. df reports 89% disk usage. Same partitions as before (they weren't changed). Go for it man; it's possible ! As for fips, I don't know if all distro put it on the CD, but that's where I got it (infomagic LDR). I'd be interested to hear how you make out. Ciao, Gerald Crimp -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: French-Canadian console keymap?
On Thu, May 14, 1998 at 12:23:36AM -0400, David Huggins-Daines wrote: Hi! Does anyone know where to find a console keymap for a 102-key Canadian multilingual keyboard? There doesn't seem to be an option for it in the menu kbdconfig gives, even though X has a Canadian layout that matches my keyboard exactly... Hmm, I do have an option in kbdconfig. kbdconfig, however, at least as far as I understand it, no longer affects the keyboard under X. At least, the default XF86Config on Deb 1.3.1 didn't for me. There is a new scheme for handling the keyboard with the newer versions of XFree86, apparently. IIRC, it is possible to tell X to use the old method which uses the kbd for text console. What version of Deb are you using ? What version or revision or release of X ? As for this new setup under X, I went in and hacked my own keyboard. I can't remember what was wrong, but something about the one supplied was unsatisfactory. So, let us know whether we're talkin' console, or X, and what version, and maybe I can give some more help. Ta, Gérald Crimp -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: dselect oddities
On Fri, May 15, 1998 at 08:45:45AM -0800, Steve Lamb wrote: Am I the only one who feels that dselect should not update packages unless explicitly told to? No you are not alone. As someone suggested, this has shades of M$ taking control of one's machine. Someone else mentioned that having to tell it to go ahead and update when there are approx 300 packages, is onerous. I agree. And it is just as onerous for those who don't want to update automatically. Perhaps the solution would be a command line option: give one option and dselect will automatically mark new stuff for an update (the user would have to specifically mark those things they don't want to have updated), give another option, and all updated packages will be marked to hold leaving the admin to then select those packages which they would like to have updated. Ta, Gerald -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Sound - permissions on /dev devices
Hi, I finally got my sound enabled in the kernel. On to the next problem. Whenever I try to use something like workbone, or anything that needs to use a device such as /dev/audio, or /dev/dsp, or even /dev/hdb (my cdrom) as an ordinary user I get permission denied. The permissions on these are ?rw-rw. For the time being I have just done a chmod to give rw permission to the world. From a security point of view, is this an acceptable solution ? Would there be a better way to allow myself to use these devices without being root ? I thought of making myself a member of the groups audio and disk (for hdb), but this seemed like a messy solution, and I wasn't sure of the wisdom of being a member of the group disk. Anyone know what the right way to do this is ? TIA, Gerald Crimp -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: IPX and Soundcard - loadable modules?
On Sun, May 10, 1998 at 11:23:44PM +0100, M.C. Vernon wrote: Dear Debian, I need support for a soundcard and IPX. I have been informed by some helpful people that I need to recompile and include support for these. I would suggest that you start by reading some of the docs available on the subject of compiling a kernel. A good place to look for up to date docs is at http://sunsite.unc.edu/mdw/linux.html#ldp which is the Linux Documentation Project. I think there is a HOWTO or mini-HOWTO about compiling a kernel. I also constantly referred to the README in the kernel source code top level directory, and in the sub-directory Documentation of the source code, the file modules.txt. Can I do this via loadable modules, or do I need to recompile (which is scarey for a newbie - how do I do it?) I am not a technical wizard, but I have compiled kernels a couple of times myself. If I remember correctly, you do have to compile even to use things as loadable modules. I think the standard kernel that you get when you install comes as one monolithic block. You have to compile to enable loadable modules and to select which things you want to use as modules. I don't guarantee the accuracy of that information though. I too found compiling a kernel to be a bit intimidating, but I can report that having thrice compiled a kernel, I have not yet met with disaster. I did meet with numerous hick-ups that slowed me down while I tried to figure them out, but nothing directly related to the compile itself caused me grief. I did, however, proceed with extreme caution _after_ having read the above docs, and following them closely all the while. I think if you go slowly, and make sure you understand what the docs tell you to do, you shouldn't have much trouble. Just be prepared to spend a little time at it. 'Course, I can't guarantee anything, just speaking from my experience. Who knows, if you have a more technically inclined noodle than myself, you may zip through the whole thing. Good luck ! Gerald Crimp -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
logical devices on sound card
Anyone understand logical devices as they relate to kernel configuration ? I recently compiled sound support into the kernel. I just kind of guessed at what to do in putting together isapnptools and make config, and guessing at the capabilities of this poorly documented card. Dumb luck got me fairly good results. I didn't know what mpu-401 is (still don't for that matter), so I didn't choose it during make config. Following the Sound HOWTO I ran `cat /dev/sndstat' to see how things looked. Under the section card config I got something along the lines of Card config: Sound Blaster at 0x220 irq 5 drq 1,1 ( MPU-401 irq 1 drq 0) midi ... OPL-2/OPL-3 FM at 0x388 drq 0 The sound HOWTO said that anything in parentheses was configured but not detected. I'm not sure I understand because I didn't enable this in the kernel, yet it was reported. I decided I would enable it in the kernel so I recompiled. Now cat /dev/sndstat reports Card config: Sound Blaster at 0x220 irq 5 drq 1,1 (Roland MPU-401 at 0x168 irq 12 drq 0) (SB MPU-401 irq 1 drq 0) OPL-2/OPL-3 FM at 0x388 drq 0 So, my dumb luck has run out. I might just go back to the old configure and hopefully get back the midi device, but I'd like know if I can also get this mpu device enabled. Anybody know about this, specifically, why did mpu appear in /dev/stat after the first compile even though I hadn't enabled it during make config ? isapnp did find 4 logical devices on the sound card. I still only partially understand what a logical device is so I'm not sure where to go from here. Any comments or ideas appreciated. TIA, Gerald Crimp -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
sound utilities, /usr/X11R6/bin/au
Sorry to keep pestering the list with my sound questions. I thought it would be better to send each one as separate message rather than dump only loosely related questions under one subject header. After having enabled sound in the kernel, I went into dselect and installed a number of sound utilities in order to play around and see what the various packages did and which ones might be of interest to me. Now when I boot, I get a process named /usr/X11R6/bin/au running all the time. This is a slink to auvoxware. This ties up the sound devices in /dev so that if as a user I want to do something that uses sound, I can't because the device is busy. Does anyone know what auvoxware does and to which package it belongs ? I couldn't find a man page for either au or auvoxware. I went back to dselect, but couldn't find any mention of au or auvoxware in the sound package descriptions so I don't know which one to potentially remove from my system. Thanks once again, Gerald -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
SOLVED: Re: recompiling kernel
On Tue, May 05, 1998 at 12:02:30PM -0400, Bill Leach wrote: Hi Gerald; Though I noticed your original posting I did not then comment for it being too much of a 'blind leading the blind' situation. More like the blind leading the stupid... Read on. What I did notice is that the lines you quoted: [snip] should work. I am most assuredly not a shell guru but the command 'hash' is a bash internal command. _I_ would expect that shells other than bash should work correctly because the 'hash' command itself would not exist and that would be an error. Your talk of shells other than bash made something click that should have clicked before. Bash is the shell I use. However, not so long ago, I changed the /bin/sh symlink from bash to ash so that netscape helper apps would not get screwed up by the double parantheses problem. I keep a log of changes to and maintainance on my system. When I saw this symlink change, I reasoned that root's shell is bash, and since I never saw any #!/bin/sh as the first line of any of the files I looked through, bash would be used. I never gave it a second thought. Thanks for coming forward. And thanks to all those who offered help. I'll likely be back as I have compiled but things are not working as smoothly as they might. But I am going to go back to the docs first. Thanks again to all. Gerald Crimp -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: recompiling kernel
On Sun, Apr 19, 1998 at 11:39:52PM +0200, Marcus Brinkmann wrote: Try kernel-package. make-kpkg --revision custom.1.0 kernel-image will creat you a deb file with your kernel. Then install the deb file in /usr/src with dpkg -i filename and answer yes when it is offering to make a boot disk. This will install the kernel on the root partition, too. You may not want this... but your old image is still in vmlinuz.old (and it is a good idea to have a lilo entry for it). Marcus Hi, I finally installed the kernel-package as you suggested. I was hoping first to find out where my problem was coming from with the non-Debian specific kernel compile procedures. I am back to square one after trying `make-kpkg --revision custom.1.0 kernel-image'. It poops out at exactly the same place. I am at my wits end. This is basically a paint by numbers procedure for me. I follow the instructions and do what I'm told. I don't have a deep understanding of all the pieces that come together to produce a kernel. I am especially frustrated that I compiled a kernel a few months ago with no problems. I resubmit the error output in the hopes that someone can help me get to the bottom of this, or point me to other resources to which I might turn (I have already posted to comp.kernel.sources without result). I would really like to get sound support into the kernel. Please note the following highlights from the following output. I do not have encaps on my system (`find / -iname encaps', `type encaps', 'file encaps' all negative). To my mind, this means that the `else' clause should be selected rather than the `then' clause of the `if' statement (thus sidestepping the curious illegal `-k' option). It remains a mystery to me why I get this error even if I use the old .config file (no sound support) which was used to compile the kernel I currently run. What could I have possibly changed on my system that is causing me this grief ? Any help would be greatly appreciated. If less abbreviated output, or other info (config files, directory listings, etc) is required, please let me know. Here is the standard output: [...] make[4]: Entering directory /usr/src/kernel-source-2.0.30/arch/i386/boot/compressed' tmppiggy=/tmp/$$piggy; \ rm -f $tmppiggy $tmppiggy.gz $tmppiggy.lnk; \ if hash encaps 2 /dev/null; then \ objdump -k -q -o 0x10 /usr/src/kernel-source-2.0.30/vmlinux $tmppiggy; \ else \ objcopy -O binary -R .note -R .comment -R .stab -R .stabstr /usr/src/kernel-source-2.0.30/vmlinux $tmppiggy; \ fi; \ [...] objdump: illegal option -- k Usage: objdump [-ahifdDprRtTxsSlw] [-b bfdname] [-m machine] [-j section-name] [--archive-headers] [--target=bfdname] [--debugging] [--disassemble] [--disassemble-all] [--disassemble-zeroes] [--file-headers] [--section-headers] [--headers] [--info] [--section=section-name] [--line-numbers] [--source] [--architecture=machine] [--reloc] [--full-contents] [--stabs] [--syms] [--all-headers] [--dynamic-syms] [--dynamic-reloc] [--wide] [--version] [--help] [--private-headers] [--start-address=addr] [--stop-address=addr] [--prefix-addresses] [--show-raw-insn] [-EB|-EL] [--endian={big|little}] objfile... at least one option besides -l (--line-numbers) must be given objdump: supported targets: elf32-i386 a.out-i386-linux srec symbolsrec tekhex binary ihex trad-core encaps: not found [...] make: *** [stamp-image] Error 2 root:/usr/src/linux # Ta, Gerald Crimp -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: recompiling kernel
On Wed, Apr 29, 1998 at 11:32:27AM -0400, Scott Ellis wrote: [snip] Somewhere in usr/src/linux/arch/i386/boot/compressed there is a file that is invoking objdump (whatever that might be) with an option `-k'. make zdisk and make zImage say this is an illegal option. [snip] The objdump man page doesn't mention the `-k' option. So why is some script somewhere issuing this command ? This is more puzzling since I compiled a kernel when I first installed without any problem. Same sources. Now I try to add sound and I get this. Even if I take out the sound support and try to compile the kernel exactly like the one I am currently using (from a backed up .config) I get this error. [snip] Argh! This keeps popping up. The proper fix for this is rm `which encaps` Thanks for the answer. Sorry if I'm dredging up old muck. Before I just blindly apply your solution, though, I want to make sure I know at least at little what's going on. I've been trying to decipher the stdout from make zImage (or zdisk), and I am not sure the above solution will solve the problem. Once again, here is where the compile craps out: (...) make[2]: Entering directory /usr/src/kernel-source-2.0.30/arch/i386/boot/compressed' (...) if hash encaps 2 /dev/null; then \ objdump -k -q -o 0x10 /usr/src/kernel-source-2.0.30/vmlinux $tmppiggy; \ else \ objcopy -O binary -R .note -R .comment -R .stab -R .stabstr /usr/src/kernel-source-2.0.30/vmlinux $tmppiggy; \ fi; (...) objdump: illegal option -- k Usage: objdump [-ahifdDprRtTx (...) (...) make: *** [zdisk] Error 2 root:/usr/src/linux # Let me explain why I am unsure that the above solution is going to solve my problem. If I understand the above excerpt correctly, `objdump -k -q ...' will be executed if `hash encaps' returns an exit status of true. It shouldn't. I don't have encaps on my system. I have checked using `find / -iname encaps', `type encaps', and `file encaps'. Therefore, `objcopy -O binary ...' should be executed instead. I have tested this with the following little script : if hash encaps 2 /dev/null then echo Hello ! else echo Loser ! fi which effectively returns `Loser !' rather than `Hello !'. So I don't understand why the compile script trys to run `objdump -k ...' instead of `objcopy ...' Can anyone explain ? My understanding of all this is pretty scant. Maybe there is something behind the scenes that I can't see with my little analysis. If someone could confirm that rm 'which encaps' will indeed solve the problem, it would be greatly appreciated. (If that someone had the time and inclination, I'd love to know the how and why of the solution !) Thanks all for your help, Gerald Crimp -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]