Re: [gentoo-user] reiserfs undelete ?
Andrew Gaffney wrote: Matthew Baxa wrote: On Mon, 2003-11-24 at 17:11, Oliver Lange wrote: Phil Sexton wrote: Someone with more technical backround (less lazyness ;-P) would provide you with a more exaustive answer, but in short you cannot undelete files in a ReiserFS partition because of its very internal structure. Does this count? http://recover.sourceforge.net/unix/ Oh boy, that's a clear minus point for reiserfs.. Um, all the Unix filesystems (ext2, ext3, ReiserFS, XFS, JFS, UFS) have no undelete stuff. There was a program I used to use with ext2 before ext3 was big. I can't remember what it was called. You would add a 'LD_PRELINK' environment variable pointing at its dynamic lib that would override the unlink() function and move files into a protected environment. You could then run the associated utility to recover them. Also, it came with an *actual* ext2 undelete program that worked pretty well. I believe you're referring to libtrash, http://m-arriaga.net/software/libtrash/ . Not for me, but interesting idea. -Michael -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] Re: Gentoo internal structure
Ciaran McCreesh wrote: On Thu, 20 Nov 2003 23:37:08 +0200 Wayne Oliver [EMAIL PROTECTED] | You can take an approach similar to debian simply put | them in a non-free folder in portage, possibly put a banner on | the ebuild informing the user tha it's non-free ??? Hmm, maybe we should implement a keyword in the ebuild that tells people under what licence a package is distributed. That way, it would be easy for the user to check and avoid non-Free-as-in-Stallman software if they were that way inclined. I propose that we call this keyword 'LICENSE' (US spelling seems to be policy...), add it to every ebuild and make repoman check that it's there. Oh, wait... If only the world were that simple. But in reality developers never, ever anticipate user's needs that way :-/ I mean, if they did there would be an ACCEPT_LICENSE setting scheduled to be implemented in make.conf to do exactly what this guy wants. But no...that would be too easy... *heavy sigh* Ah well. We can always dream, eh? -Heschi -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] after kernel upgrade get / mount error
Basically an initrd is a completely different filesystem. This step in the boot process is trying to turn the filesystem from this: / = / on initrd into this: / = / on root volume /initrd = / on initrd by doing mount /dev/whatever /new_root. then pivot_root /new_root /initrd It's like a chroot, but not quite. So your problem is that it can't remount / on the initrd to /initrd, probably because you haven't created the mount point. So it fails by not remounting the initrdin short, it unmounts it. So yes, it's not a problem, and yes, you can ignore it, and no, it has nothing to do with reiserfs. Read linux/Documentation/initrd.txt for more details. -Heschi - Original Message - From: Joshua Banks [EMAIL PROTECTED] To: [EMAIL PROTECTED]; [EMAIL PROTECTED] Sent: Wednesday, October 01, 2003 12:42 AM Subject: RE: [gentoo-user] after kernel upgrade get / mount error --- Corey Larsen [EMAIL PROTECTED] wrote: Make sure that ReiserFs support is compiled into the kernel monolithically, instead of being included as a module. e.g. you should have: File systems --- * Reiserfs support instead of: File systems --- M Reiserfs support also make sure that you have the correct options compiled in/excluded per the docs at http://www.gentoo.org/doc/en/gentoo-x86-install.xml under Code listing 16.6: Ok so I have it compiled into the kernel from what I can see. grep -I REI .config CONFIG_REISERFS_FS=y # CONFIG_REISERFS_CHECK is not set # CONFIG_REISERFS_PROC_INFO is not set So I take it that the bootup messages that are posted below can safely be ignored? ReiserFS version 3.6.25 Sep 29 17:57:18 deadmeat kernel: VFS: Mounted root (reiserfs filesystem) readonly. Sep 29 17:57:18 deadmeat kernel: Trying to move old root to /initrd ... failed this message Sep 29 17:57:18 deadmeat kernel: Unmounting old root Sep 29 17:57:18 deadmeat kernel: Trying to free ramdisk memory ... okay Sep 29 17:57:18 deadmeat kernel: Freeing unused kernel memory: 112k freed Sep 29 17:57:18 deadmeat kernel: Adding Swap: 506036k swap-space (priority -1) Thanks, JBanks __ Do you Yahoo!? The New Yahoo! Shopping - with improved product search http://shopping.yahoo.com -- [EMAIL PROTECTED] mailing list -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] alsa-driver
If Alsa is truly 'part of the kernel' then my question was, and still is, where is the source code in the kernel tree? Out of morbid curiousity-- What's wrong with the drivers in linux-2.6.0-test*/sound? -Heschi -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] Cleaning up ../distfiles from make.conf
Have I missed something again? Is there really no option or flag or feature or something to set in /etc/make.conf so that emerge will remove the /usr/portage/distfiles/* that were downloaded for the installation? Yes. You've missed http://article.gmane.org/gmane.linux.gentoo.user/46323/match=distfiles http://forums.gentoo.org/viewtopic.php?t=3011highlight=distfiles+clean http://forums.gentoo.org/viewtopic.php?t=70275highlight=distfiles+clean http://forums.gentoo.org/viewtopic.php?t=69654highlight=distfiles+clean http://forums.gentoo.org/viewtopic.php?t=67849 Honestly, people... -Heschi -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] cpuinfo documentation
- The cpuinfo is incompletely described at http://search.cpan.org/author/JSTOWE/Linux-Cpuinfo-1.5/Cpuinfo.pm , but even the author of this Perl module admits that he doesn't know anything about 'wp', which is the cpuinfo entity that I'm currently trying to understand. Following the wp bit through arch/i386/kernel/setup.c, then arch/i386/mm/init.c, then googling, it looks like wp is write protect. Take a look at http://testing.lkml.org/slashdot.php?mid=249185 and you'll probably get the gist of it. Fun research project. HAND. -Heschi -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] /mnt/cdrom problems
I try not to add junk posts to the mailing list, but... *jawdrop* I salute you, sir! -Heschi I don't know if this will hel,p but here in South Florida, we have these tiny sugar ants. They will eat nearly anything. Cleaning up the yard after my 2 dogs isn't a problem as long as we have ants. If you have them in your neck of the woods you might try slopping a trail of Hershey's Syrup from the floor up to the CD-ROM drive. Once the ants eat up all of the chocalate, they'll probably go for the toast. One black jelly bean smashed down to about 1/8 high and stuck with some spit to the cd tray should coax them inside.This will probably take 2 to 3 weeks, unless you have particularly large and hungry ants. Mmmm, I COULD be wrong here but I hope this helps. The other alternative is to break down and shell out 30 or 40 dollars for a new CD-ROM drive -- Regards, Ernie 100% Microsoft and Intel free -- [EMAIL PROTECTED] mailing list -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] qpkg -q - am I interpreting it correctly?
Unfortunately, kdebase is a glob of a lot of closely-related packages. It's a kludge, but about a year ago (wild guess) the maintainer discussed splitting kde into a bunch of different packages and the eventual decision was to leave it globbed. So...try ldd `which rosegarden` to display the libraries RoseGarden links to. Then you can qpkg -f the libraries. BTW; I think that qpkg only traces one level of dependencies...kdelibs was probably an indirect dep. emerge -ep rosegarden might give you some insight. Or not. Dunno. -Heschi I don't suppose you know how to go any further in terms of determining what portion of kdemultimedia Rosegarden is actually using do you? None of the options in the man page seem to suggest this tool can go that far. -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] Booting kernel on pld pentium
This is exactly what happens when you forget to compile in console support...you should double-check that. -Heschi Hello, i installed gentoo on my old pentium computer to function as a gateway. But there is a problem when i am booting the kernel. After some moments it stops booting. The last output is: Freeing unued kernel memory 124k freed. And after that it stops and doesnt go on. What can I do? Why is that? Bye Jan -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] emerge mask issue: I checked emerge info
The people telling you that ACCEPT_KEYWORDS should be just ~x86 are wrong. Had they bothered to look on their own systems they'd see that they have both also (assuming they run unstable). If you had only ~x86, you wouldn't be able to install stable packages. Settings in make.conf are cumulative to settings in make.globals and the make.profile files. Oh, and wrt your original question, it's hardmasked in package.mask just like people have been telling you. If you emerge the ebuild manually it'll at least try to install, but you should probably find out why it was hardmasked first. -Heschi but . . but from my make.conf i have .. . . # '~ppc', '~sparc' are the unstable KEYWORDS for their respective platforms. # DO NOT PUT ANYTHING BUT YOUR SPECIFIC ~ARCHITECTURE IN THE LIST. # IF YOU ARE UNSURE OF YOUR ARCH, OR THE IMPLICATIONS, DO NOT MODIFY THIS. # #ACCEPT_KEYWORDS=~arch ACCEPT_KEYWORDS=~x86 # Portage Directories # === # -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] Automatic perl packages, emerged from CPAN?
Take a look at g-cpan.pl...I've never used it myself, but it seems to be kinda what you're looking for. Not much information on usage; seems to be g-cpan.pl Category::Module but I'm not positive. Try searching the forums. -Heschi I've been wondering about this, is there some way for me to build the various perl modules using emerge, but getting them directly from CPAN? I'm thinking about something analagous to Debian's dh-make-perl command, where you can manage your perl modules with the debian package management system, but you also have the flexibility of CPAN. Perhaps something like: 'emerge --cpan Module::Name'? Just a suggestion, thanks for reading! Adam, half Gentoo half Debian user (the server will be on Gentoo RSN) -- [EMAIL PROTECTED] mailing list -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] auto Emerge world
I've been following this thread and it occurs to me that doing an auto emerge world might be better handled if there was some way to --pretend this a print a log of what would be updated. It could even be a menu-driven process allowing you to choose what to update. Does this sound feasible? As I said, I am not a programmer so this might be totally useless thinking. Just trying to be helpful. http://forums.gentoo.org/viewtopic.php?t=67849 Take a look at bemerge and femerge. -Heschi -- [EMAIL PROTECTED] mailing list
Re: Re: [gentoo-user] how to trigger ._cfg* to be installed?]
I'm no developer, but try emerge --noconfmem :-/ -Heschi - Original Message - From: Lloyd D Budd [EMAIL PROTECTED] To: [EMAIL PROTECTED] Cc: [EMAIL PROTECTED] Sent: Thursday, August 21, 2003 12:46 PM Subject: [Fwd: Re: [gentoo-user] how to trigger ._cfg* to be installed?] Hi, before I further boffo my system, any devel that is online know the solution? Cheers, Lloyd -Forwarded Message- From: Lloyd D Budd [EMAIL PROTECTED] To: [EMAIL PROTECTED] Subject: Re: [gentoo-user] how to trigger ._cfg* to be installed? Date: Thu, 21 Aug 2003 10:03:50 -0400 On Thu, 2003-08-21 at 10:25, Andrew Gaffney wrote: Lloyd D Budd wrote: On Thu, 2003-08-21 at 10:10, Andrew Gaffney wrote: Lloyd D Budd wrote: I did an 'emerge -e world' in the hopes that I would have all of the cfg change files. This did not occur. Any one know how to accomplish this? Are you trying to overwrite all your config files with the defaults? If so: Nope, just re-inspect the changes. Like someone else already said, I think your only option is to re-emerge. *cringe* Which I did with 'emerge -e world' :-( -- [EMAIL PROTECTED] mailing list -- Lloyd D Budd [EMAIL PROTECTED] -- [EMAIL PROTECTED] mailing list -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] grub messed up XP partition
I'm assuming that Disk error is coming from Grub, since I don't remember seeing that from Windows' boot loader. I guess it could be your motherboard. Knowing which would be a big help. I've done what you did before, and I was able to fix it with a combination of fixboot and fixmbr (despite its name, it seems to do partition boot records as well in Rescue mode. fdisk /mbr probably couldn't hurt either. You said you reinstalled Windows? Wow. Drastic. That *should* have fixed *everything* :-p Since it doesn't seem to have, either I'm missing something, I misunderstood what you said, or you broke something completely different. In any rate, if you're sure that the problem is that the Windows partition boot record is clobbered, you can make a backup pretty easily. I was going to write out the commands to do all this stuff but then I was googling and I found this page, http://astron.berkeley.edu/~jhall/grub_install_hda1.html , which, heh, seems pretty well targeted to you :-p The only problem is that this is designed to restore a 9x windows, not XP. XP doesn't have any equivalent command to sys, I don't think, so you'll need to copy a good bootsector, like this: dd if=/dev/hda1 of=boot.bin bs=512 count=1 then get it to your Linux install on the broken machine, and do: dd if=boot.bin of=/dev/hda1 bs=512 count=1 I've never personally done this myself, but there ought not to be any important information in that area anyway (now that you've clobbered it) but you still might want make a back up. Which I shouldn't need to explain. If that doesn't do it for you, you might have broken something else. Alternatively, if you don't have a functioning Windows machine, I think I can muster up the 512 bytes of bandwidth to email you a good boot sector off one of my machines. Just ask me off-list. -Heschi In a moment of lessened attention, I executed 'setup (hd0,0)' in grub, rather than the 'setup (hd0)' that I intended to use. The problem now is that I can't boot XP that's installed on the (hd0,0) partition. It says Disk error\nPress any key to retry. I have tried the XP fixboot utility to rewrite the bootsector on that drive, but to no avail. Is there a way I can undo the changes that grub made to this partition? -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] Portage: Updating all installed packages
You're acting like this is a big deal. Unless I'm missing something, it's really not. I don't know how much people know about how Portage calculates its dep tree, so I'll start from scratch. When you emerge a package, two important (for our purposes) things happen. 1) a directory in /var/db/pkg is created containing the current state of a number of Portage variables, including the USE variables, and the ebuild. 2) If the emerge was done without -u and --oneshot the package is added to /var/cache/edb/world. How are dependancy packages handled? As if they had --oneshot. They are not added to world. (NB: I'm ignoring the system profile) At this point, emerge -puD world updates all packages, because they are all in the dependancy graph--that is, every package is installed for a reason. Either it was manually emerged, or it is a dependancy or of one that was. What happens when you unmerge this package? Portage uses the /var/db/pkg ebuild and CONTENTS file to un-install the package. Any dependencies that you have are *left alone*. This is a much commented-on problem when removing, for example, kde. Now you have packages on your system that you don't need, and aren't in the depgraph. So when you do emerge -puD world they will not be found, because emerge -puD world is really emerge -puD `cat /var/cache/edb/world`. (again, ignoring the system profile) But why should they be? Unless you've lied to portage (removed something from the world file) or there's a portage bug (it removed it for you when it shouldn't have) nobody really uses any of these--why update them? But, you ask, how do I get rid of them? That's what dep-clean is for. It's in gentoolkit. Run it. I clean my systems out like this: emerge -C `dep-clean -U -C|grep /` , but you should check which packages you're about to destroy first. Then run revdep-rebuild, because sometimes there have been dependency changes without revision bumps (which is, IMO, a bug). Once you do this, if you still see a deviation, THEN you have a bug. If not, as usual, I think this is a documentation bug, not a portage bug. Would you really prefer that Portage waste its time and yours on packages nobody uses? So, now to address your bug entry. The problem of course with executing [2] using qpkg is that all packages get added to the world file so portage needs to handle this differently. emerge -uD --oneshot `qpkg -I -nc` Deps file - Another will be a 'deps' file in the same location and this will contain all packages pulled in as a result of a manually installed package. This 'dependencies' file will also contain all system packages which form part of the initial base system. I'm not sure what you mean by this--this is either the difference between world+system and qpkg -I -nc (i.e., dep-clean -U) , or the /var/db/pkg/category/package/{,R,P}DEPEND files. Either way, unless I've misunderstood, it's already available in one way or another. Finally, there are plenty of things in Gentoo that de facto require a tool to do them right. There's ufed, mirrorselect, and etc-update. I don't see why qpkg shouldn't be part of that list if you really, really want to update all these packages. -Heschi http://bugs.gentoo.org/show_bug.cgi?id=26540 Please read it and provide feedback either on the bug or on this list. Basically, the issue at hand is the difference between [1] and [2] and how portage should handle this difference. [1] emerge -Duvp world [2] emerge -Duvp `qpkg -I -nc` Look forward to receiving feedback. -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] Portage: Updating all installed packages
Your initial discussion did not relate to my bug at all. Yes, it did. It explained, in fact, exactly why this happens. It may have been too verbose or basic for you, but again, I have no idea how much you know about how Portage works, and you don't seem to have read it at all. It's possible that I was wrong (it always is) but if I am I would appreciate you telling me so. But, you ask, how do I get rid of them? That's what dep-clean is for. It's I did not ask this. It's a rhetorical technique. You don't need to take it literally. Of course not. I'm suggesting adding a switch to portage which gives you the option to use it or not use it. I was attempting to explain why, by default, Portage doesn't try to update these packages. So, now to address your bug entry. /var/db/pkg/category/package/{,R,P}DEPEND files. Either way, unless I've misunderstood, it's already available in one way or another. There is no method available currently in portage to achieve the same effect as [2] in the bug. I gave you one in the line you snipped. Again, if it doesn't work, I'd like it if you told me why instead of ignoring me. And you haven't at all explained what this extra deps file is intended to do. Why can't I just do find in /var/db/pkg to get this information? Finally, there are plenty of things in Gentoo that de facto require a tool to do them right. There's ufed, mirrorselect, and etc-update. I don't see why qpkg shouldn't be part of that list if you really, really want to update all these packages. Well, the reason is simple. This is, the way I see it, base functionality. It is extending --deep to include the full map of packages. And this is why it requires due attention from the parent package. No, it is forcing Portage to include packages that have *no reason* to be on your system. I would prefer it got rid of them altogether, rather than updating them. But that's hard. Thanks for feedback. I would appreciate it if you went back and reread what I wrote, because I thought I addressed the problem fairly completely. -Heschi -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] Re: Portage: Updating all installed packages
In fact some of them are in the world file so I have no clue why dep-clean thinks they need cleaning, but it's also strange that some I emerged manually are not in world. If you're not completely up to date, old versions of packages that you need will show up as unneeded (after all, they aren't...the latest versions are, you just don't have them installed). A handful of unstable packages I emerged after moving the ebuild to PORTAGE_OVERLAY_DIR and changing ~x86 to x86. These show up in world, but are on the list from dep-clean -U. Dep-clean probably doesn't take into account package specific ~x86 settings. But if you manually marked them stable, I'm not sure why it would want to get rid of them, unless they're old versions. -Heschi -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] Re: etc-update and fstab...
How do I recognize trivial changes? Only upt to 3 lines affected? Then there are no trivial changes;-) There aren't any trivial changes! If a 1-line change can bring your machine down, then EVERY single line must be chacked with the greatest of care. Spoken like someone who recently installed gentoo, and never had to look at a bunch of files where the only thing that had changed was the CVS header. I'm quite pleased with the automerging support. Quite often I get confused which side is old (left?) and new, respectively. My problem too! Then use a different diff command. You can change it in etc-update.conf. I don't have any suggestions but there have to be more out there. To the person who said -5 is useless, I disagree that. Every time I do an upgrade of XFree there's a buch of X config files modified that I don't care about. I merge the files I've modified, then -5 the rest of them. To the person (people) who think /etc/fstab never changes, older versions of baselayout required tmpfs mounted at /mnt/.init.d/ . New versions (maybe not in stable yet) don't. How do you suggest those changes get pointed out to the user? You're complaining that the automated tools don't do what you want them to--and now people are suggesting that fstab get run through *sed*?? Sounds like a recipe for disaster to me. If you don't like etc-update, edit the files manually. If you have a concrete suggestion for improving etc-update, feel free to say something. Etc-update is by no means perfect, but I don't see an obvious way to improve it. You might try the menu-based mode, which has been in development for quite some time. That will, at least, fix the too many files to fit on the screen problem. -Heschi -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] Re: etc-update and fstab...
I agree completely that fstab needs at times, like recently, to be updated. However, for all the smart tools around here, I think it amazingly dense that etc-update -5 will replace a working partition number like /dev/hda6 with something like /dev/boot! It certainly should be able to find out which partitions I'm using for which purpose: /dev/hda6 /boot ext3 noauto,noatime 1 1 It requires me to remember which partition is which. Possibly fine for programmers and hardware techs, but not so nice for users. This is the core of the problem. You're asking for specialized treatment for fstab. It might be doable. OK, so that goes in, and the next question is, what about rc.conf, and modules.d/alsa, and, and, and... Someone else pointed out that there is no standard format to config files in Unix. That's why this question is not easily, maybe at all, solvable. Maybe with some sort of plugin architecture, where each package supplies its own config updater for etc-update to run. But inevitably those updaters are going to have bugs. So people will have to check their configs after they're updated, and really, how much more time does it take to merge them by hand than just checking whether the merge was done correctly? that the automated tools don't do what you want them to--and now people are suggesting that fstab get run through *sed*?? I don't know 'sed' and didn't suggest anything about it, even though I know you're just making an example. Sorry, this was my poor attempt to address half a dozen posts without responding to each of them--it wasn't really directed at you. I do think that some sort of editor that would show the changes side by side would be an improvement, but I don't know what tools would do that today. This is exactly what merge interactively does. I think it's option 3 once you've selected a file to update. Someone else posted a more detailed explanation. I think the real problem that this has pointed out is that people expect etc-update to update config files for you. It doesn't. It doesn't even *try*. Used to be that Portage just printed something like: There are config files to be updated! use find -name .__cfg* to find them. Well, people weren't too happy with that. Surprise. So someone wrote etc-update, which basically did the find for you and gave you a couple options on how to handle the new one--delete the update, blindly accept the changes, or merge them with a diff command. But etc-update was in gentoolkit, and newbies never found it. And then they posted annoying messages to lists and groups about how stupid having to find config files manually was. So etc-update was moved into the portage package, and the help message was updated to mention it instead. So that brings us to today, where we have messages (these are the most recent in a long series of how -5 clobbered my system) about how stupid etc-update is. Well, yeah. It's not supposed to be smart. You're supposed to be. But OTOH, I'd say that it's sort of a documentation bug that this isn't explained very clearly anywhere, and I guess auto-merge is not obviously synonymous with DESTROY YOUR CONFIG FILES!! BAHAHAHAHAH!!'. Maybe someone should write a config file manual for the user docs section. But there's not anything wrong with etc-update, just with people's understanding of how it should be used. Hope that clears things up. -Heschi (PS: Mark: I may write angry-sounding emails, but generally that's just my style. If I'm really frustrated I don't write anything at all. No hard feelings on my side--I wrote because I thought your points, and other people's, were worth addressing.) -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] etc-update and fstab...
forget the space between noauto, noatime and all will be well, do this on every line. [snip] none/dev/sh tmpfs defaults0 0 Also, this should be /dev/shm, not /dev/sh, I believe. Not sure if this will be a problem, but things could get wierd if you don't fix it. -Heschi -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] gentoo-sources
Does anyone know when the gento-sources 2.4.21 will come out? When it's ready. In the meantime you can use pfeifer-sources, which is the development branch for gentoo-sources. -Heschi -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] No usb mouse support, after 12 re-compiles. Missing modules? USB support proper also fails...
Did you mount /boot before you copied the kernel? It's hard to debug this because you haven't given us much information. What do you mean, all show up in dmesg? The exact output would be useful. What do you mean, the modules are not available? There's no reason that modules should not be found, unless you're running a different version of the kernel than you installed the modules for. What does uname -a give you? Is it the right compile time? What kernel source are you using now? -Heschi Anubis still can't get usb mouse support working Anubis It seems like I get a different system with each new install using the same configs Anubis wtf? Anubis if I compile as modular, all the modules are not availible Anubis if I compile proper, like when it worked before, I get nothing as well. Anubis When I ask for help, I get ignored mostly. What am I to do? Anubis I could put the ps2 adapter on but why? Anubis I had this very same mouse on this very same system working with gentoo and this gaming-sources Anubis now I get nada Anubis very frustrating Anubis no consistency Anubis uhci.c,usb.c hid-core.c Anubis all show up in dmesg -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] alsa-driver 0.9.5 - any ebuild yet?
~x86 -Heschi I think a ACCEPT_KEYWORDS=~86 emerge -p alsa-driver would do the trick. On my test machine at work the proposed command: ACCEPT_KEYWORDS=~86 emerge -p alsa-driver -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] alsa-driver 0.9.5 - any ebuild yet?
Probably make.conf is the only place that really even mentions them. But basically, in an ebuild is a list of KEYWORDS. Currently I believe the only use of KEYWORDS is to define what architectures (sparc, x86, ppc, ...) a package works on. In your make.profile, make.defaults sets ACCEPT_KEYWORDS to x86, because you're almost certainly using default-x86-1.4. You're using default-x86-1.4 because that's what your stage tarball came with; you used to have to set the link manually but that was back in the 1.0 days. Anyway, for every arch keyword there's a corresponding ~arch keyword, which indicates that the package is works-for-me stable (me being the developer) and ready for testing by general users. Portage ties all of this together by comparing your ACCEPT_KEYWORDS to the ebuild's KEYWORDS, and if there aren't any shared, it considers the package masked. So an ebuild is first put into the portage tree with KEYWORDS=~x86 and only people with ~x86 see it unmasked; later, the ebuild is marked stable with KEYWORDS=x86. Note that like many other variables in make.{globals,defaults,conf}, ACCEPT_KEYWORDS is cumulative, so even if you set ACCEPT_KEYWORDS=~x86 in make.conf, you still have x86 too. Or maybe you're asking why it's x86, in which case: 8086, 286, 386, 486, 586 (pentium), 686 (pentium 2+)...x86...get it? -Heschi What document explains the use of these key words? I really need to go read it and understand. -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] Can't boot after install L 02, and no GRUB either!
For future reference, one of the options in the advanced menu of fdisk is to renumber partitions in disk order. But I had them out of order on my laptop for ages and no problem. Shouldn't really matter to anything that doesn't suck. :-p What hardware are you running (ide card, motherboard, drive)? Also, for precision's sake, what does `fdisk -l /dev/hda` give you? Given what you've said it sounds like there's something unusual about your setup. You seem to have done everything right, but that's hardly unusual for wierd stuff like this. You might want to try attacking it from the floppy first--boot from the liveCD, back up your .config, copy /proc/.config over, and use that kernel to boot off the floppy. If that doesn't work, something's *really* wierd. Hmm. Maybe the disk has a bad sector in the first track? Good luck...you'll need it, I think. At some point you're probably going to want to go to the LILO or GRUB lists. -Heschi Suggestions? Cheers, Robin -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] hostname with dynamic IP?
Take a look at the -H flag for dhcpcd, maybe -D too. Add them to dhcpcd_eth0 in /etc/conf.d/net. -Heschi At booting, I see the message you need to set /etc/hostname to a valid hostname. How can I do that? The install manual (chap. 19) seems to address people with a fixed IP. Am I wrong? I have cable modem connection with dynamic IP. With RH8, the command hostname shows me the real hostname (i.e. the hostname from the point of view of the ISP, if I understood correctly); I know I can use that name to receive e-mail (I did it for testing purposes). In gentoo, the same command only outputs localhost; I suppose some script could get the name at boot time and cat it to /etc/hostname. The same question relating /etc/dnsdomainname... -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] Problem with scp
Someone gave you what I think is the right answer but I don't believe you ever responded, so I'll reiterate. I saw this line in your example login: Good day to avoid cops. Crawl to work. which looks to me like you have `fortune` in your .bashrc or .bash_profile. scp will not work if commands you have in either generate output for non-interactive shells. Apparently you can test for $- containing i, but that's just ripped off a Google search, so... [snip?] -Heschi -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] Problems installing Gentoo on Toshiba TE2100 Laptop
There was some bug with the spinner from something or other never going away. I gather you're new to unix in general--cdimage root# is the prompt; it's done, but whatever made that spinner won't give up. One message from a month back says you should try `killall hwsetup`. Anyway, even if that doesn't work, if you can deal with the silly thing filling up your screen all the time while you do the install, you're OK. -Heschi - Original Message - From: Jamie Dobbs [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Friday, June 27, 2003 6:12 AM Subject: [gentoo-user] Problems installing Gentoo on Toshiba TE2100 Laptop The install starts off well but sits at the 'CDIMAGE' bar for over 15 minutes before I give up hope and reboot the laptop. Is it worth me waiting longer or is there some trick to getting Gentoo 1.4rc4 to install on a laptop reliably? I should note that the install find the laptops NIC fine and assigns an IP from the DHCP server, it finds the USB bus but just never makes it past the 'cdimage root #' bar that makes it all the way across the screen and just sits there appears to be doing things (the 'spinner' at the end of the bar goes round and round) but nothing else ever happens. ...or am I just too impatient? I want to have Linux on this laptop but don't want ot have to go back to Mandrake after running Gentoo on my dekstop PC and seeing just how fast it is! -- [EMAIL PROTECTED] mailing list -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] Still Failing Library
This is a problem with the ABI break between 2.95 and 3.x. You're going to have to compile with 2.95. You should be able to emerge sys-devel/gcc-2.95.2 and use gcc-config to switch. I assume that all this lib stuff is you munging the proprietary stuff. The include path probably comes from their system, not yours, (since that's a linker message, it's possible it's pulling that error out of the object file, I suppose) but it's hard to impossible to tell since we can't even see any of the code. If you need help with an NDA'd library, you should talk to people you can talk to under the NDA. -Heschi Alright, this is still not working. I've played with the -lstdc++ but it didn't help (and it was already there anyhoo). Now it looks to me like it was compiled with gcc 2.95.2. Would that make a difference? Then in the line: -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] Why keeping old kernel sources?
Works better if you use the right names. emerge -C gentoo-sources-2.4.20-r3 or whatever. -Heschi - Original Message - From: Frank Hellmuth [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Monday, June 23, 2003 12:27 PM Subject: Re: [gentoo-user] Why keeping old kernel sources? On Monday 23 June 2003 15:14, MAL wrote: Frank Hellmuth wrote: Is it safe to do a #emerge -C linux-2.4.20-gentoo-r3 linux-2.4.20-gentoo-r4 They aren't cleaned, as often previous sources are needed still, such as when you want to compile a module for a previous version. Even when you run the above command, the directories will remain in /usr/src, because when you compiled your kernel, you generate additional files that weren't installed by the emerge. So _after_ unmerging, you can rm -rf /usr/src/linux-2.4.20-gentoo-r{3,4} Thanks for the information, now that I've tried it, I found that there are no ebuilds for r3 and r4 in the portage tree anymore. So is there any way to unmerge them using portage-tools? Or do I just have to remove the directories? Hmmm --- esp. for something important like the kernel sources it leaves a bad taste: install via emerge, uninstall via adding ebuilds manually to PORTDIR_OVERLAY or removing directories... Frank -- [EMAIL PROTECTED] mailing list -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] Why keeping old kernel sources?
It's out of the portage tree, but for many reasons, which you can probably guess a few of, it's a good idea to keep around information about the package, and your system, as they were when you installed it. If you look in /var/db/pkg, you'll see that every package you saves a whole bunch of information--including the ebuild. -Heschi ChangeLog gentoo-sources-2.4.19-r10.ebuild Manifest files gentoo-sources-2.4.20-r5.ebuild So I thought it's missing ebuild. Frank -- [EMAIL PROTECTED] mailing list -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] Date for Final Release?
I think the single largest problem here with Gentoo and release management is that it has NEVER been said what exactly is going to be in a release and what exactly needs to be done to have a release. Version 1.4 includes what? Does that answer apply to RC[1,2,3,4,*]? If not, then the definition is incorrect. So ... anybody have an idea when we can have some bullet points to determine what is going to be in 1.4 and what is left to get it to 1.4? Is a month and a half ago soon enough for you? http://www.gentoo.org/news/en/gwn/20030505-newsletter.xml -Heschi PS: Norberto, isn't 1 incoherent thread a little slim to justify blocking me? :-p -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] [gentoo-dev] broken portage ;P and yes .. it's my fault .. sortoff
What file system are you using? I'm guessing XFS? From the install guide: XFS is a filesystem with metadata journaling that is fully supported under Gentoo Linux's xfs-sources kernel. It comes with a robust feature-set and is optimized for scalability. We only recommend using this filesystem on Linux systems with high-end SCSI and/or fibre channel storage and a uninterruptible power supply. Because XFS aggressively caches in-transit data in RAM, improperly designed programs (those that don't take proper precautions when writing files to disk, and there are quite a few of them) can lose a good deal of data if the system goes down unexpectedly. IIRC there was a long discussion about this a while back when gentoo.org stopped using XFS. I think the problem was that any file that is being lengthened when you crash gets destroyed. If you were doing emerges at the same time that would have a tendency to trash portage, or maybe the portage database. Your error appears similar to the one other people on the list are getting...but I think that's only costmetic. You might want to try reading /usr/portage/sys-apps/portage/files/README.RESCUE. And if trying the nvidia drivers is screwing your machine this badly, maybe you should *stop*. -Heschi IS there some way to fix this ? -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] pkg's safe for removing ?
Any new revision (-r1,-r2...) will almost certainly use the old source tarball. Of course Portage will happily redownload any file it needs, but here's a more-or-less bulletproof way to clean out your distfiles: mv /usr/portage/distfiles /tmp mkdir /usr/portage/distfiles FETCHCOMMAND='mv /tmp/distfiles/${FILE} ${DISTDIR}' emerge -fe world rm -rf /tmp/distfiles This should be about as perfect as you're likely to see. Unfortunately it does take a while to run. If you have leaf packages that aren't in your world file I guess you could add qpkg -I -nc|xargs. -Heschi This comes up every few weeks. What am I missing here? I've never found anything except the kernel sources an the xfree sources that are ever referenced again. 99% of the packages are discrete installs rather than a base packages with patches. I regularly wipe/usr/portage/distfiles and /var/tmp/portage without any dire consequences. I almost never see anything reloaded that I wiped out. Since I have cable that's not a problem, but how often does one update the kernel or xfree sources anyway? It would seem that the developers should put a proper rationale for what to do in the gentoo documentation, and then we could close this out permanently. -- Collins Richey - Denver Area gentoo stable - ext3 -- [EMAIL PROTECTED] mailing list -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] Create a new grub only partition? Looking for ideas.
You won't be able to chroot from tomsrtbt. When I needed to do a floppy-based install, I had to make my own 2.4.20 kernel and rebuild the disk image, which isn't terribly hard, but requires a functional installed Linux. See: http://forums.gentoo.org/viewtopic.php?t=27951 for other alternatives. -Heschi - Original Message - From: Mark Knecht [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Monday, June 02, 2003 10:53 AM Subject: RE: [gentoo-user] Create a new grub only partition? Looking for ideas. MOST IMPORTANT! Can you explain just a bit about why I'm chroot-ing when I do this? I don't understand chroot at all. If it's not too much typing, give me a few commands to sort of start the flow. Tom's is a very thin Linux (one floppy, after all!). By chrooting to your own root disk, you're putting yourself into your installed Gentoo environment - albeit running Tom's kernel instead of the one you built. Let's say your root disk is /dev/hda3 and you're /boot disk is /dev/hda4. I think that Tom's provides a /mnt directory... so do this: mount /dev/hda3 /mnt mount /dev/hda4 /mnt/boot (and any other things you need to mount). Then: chroot /mnt and you'll be in the environment you built - with all the commands (like grub) that aren't in Tom's environment. From there, you can start repairs. Thanks Nathan. Sounds cool. Have Tomsrtbt downloaded. Trying to figure out how to make the floppy on an Win XP Pro machine. Looks like I need to go find a DOS machine. Thanks for your help, Mark -- [EMAIL PROTECTED] mailing list -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] Create a new grub only partition? Looking for ideas.
Guess I should have reread the thread--I assumed you were going for tomsrtbt for a reason. Yeah, the liveCD will work fine for what you want. I'm on my way out right now, so I can't go into too much detail, but my suggestion would be to create a GRUB boot floppy (instructions on this are available http://www.gentoo.org/doc/en/altinstall.xml#doc_chap2 but that's for making a netboot image; ignore the part about tulip and just do listing 2.4). Make your best guesses for a grub.conf, write it to the floppy, then boot off it. One of the nice things about GRUB is you can edit its config on the fly (use the e key), so if you didn't get your config file right you can play around with it. (It doesn't actually save to the disk, though..) Once you have a working grub.conf that can boot all your OS's, you should be able to just choose a GRUB root partition, write it to the MBR, and go. In the meantime, you aren't messing with your HD, and so you won't be breaking anything. -Heschi Why not use a Gentoo LiveCD and boot from it. It's got everything you need to run a system. Brett, Is the LiveCD the same things as the Stage123 install CD? I suggested this possibility earlier but one responder said it wasn't built to do that. I booted from the Stage 123 install CD last night to play around with this idea. My thought was I installed grub from that CD in the first place. It must have enough on it to make this work, if I could find the right set of command to give it. Maybe we could come up with a document about how to use one of these CD's to repair a system? I'd be up for helping on that. Thanks, Mark -- [EMAIL PROTECTED] mailing list