--- Day changed --- Log opened Thu Jun 26 00:00:02 2003 00:28 < jp7x7> ok, I'm running Debian Sarge, i've install packages user-mode-linux, uml-utlities, umlrun,umlrun-uml, etc. (all packages relating to uml). The packaged uml kernel is 2.4.20-6um. Using a downloaded root_fs (Debian_3.0r0 i.e. Sarge) I could only boot single user. I found the devfs related rtfms and changed my devices accordingly. That worked, however, I still cannot create/mount new filesystems... 00:30 < jp7x7> I use "dd if=/dev/zero of=new_fs bs=1M count=500" then boot with "linux ubd1=new_fs" but when I try and mkfs or fdisk from inside uml it tells me apparently device does not exist or as 0 size. 00:30 < mistik1> do the mkfs outside uml 00:31 < jp7x7> Also networking does not work. I have tried Ethertap and TUN/TAP methods. in both I get errors. Execat error in a sec... 00:31 < jp7x7> I tried makefs on the file outside uml. Still can't mount it for some reason 00:32 < jp7x7> mkfs rather 00:35 < jp7x7> john@xian:~/UML/woody-crypto$ sudo mkfs -t ext3 new_fs 00:35 < jp7x7> Password: 00:35 < jp7x7> mke2fs 1.34-WIP (21-May-2003) 00:35 < jp7x7> new_fs is not a block special device. 00:35 < jp7x7> Proceed anyway? (y,n) y 00:35 < jp7x7> Filesystem label= 00:35 < jp7x7> OS type: Linux 00:35 < jp7x7> Block size=1024 (log=0) 00:35 < jp7x7> Fragment size=1024 (log=0) 00:35 < jp7x7> 25896 inodes, 103424 blocks 00:35 < jp7x7> 5171 blocks (5.00%) reserved for the super user 00:35 < jp7x7> First data block=1 00:35 < jp7x7> 13 block groups 00:35 < jp7x7> 8192 blocks per group, 8192 fragments per group 00:36 < jp7x7> 1992 inodes per group 00:36 < jp7x7> Superblock backups stored on blocks: 00:36 < jp7x7> 8193, 24577, 40961, 57345, 73729 00:36 < jp7x7> 00:36 < jp7x7> Writing inode tables: done 00:36 < jp7x7> Creating journal (4096 blocks): done 00:36 < jp7x7> Writing superblocks and filesystem accounting information: done 00:36 < jp7x7> 00:36 < jp7x7> This filesystem will be automatically checked every 36 mounts or 00:36 < jp7x7> 180 days, whichever comes first. Use tune2fs -c or -i to override. 00:36 < jp7x7> did I mkfs the filesystem file correctly 00:36 < jp7x7> ? 00:36 < jp7x7> ** Sorry about the accidental paste 00:36 < jp7x7> I then booted with "linux ubd1=new_fs" 00:37 < jp7x7> from uml, I used "mount -t ext3 /dev/ubd1 /mnt" 00:37 < jp7x7> I still get "Wrong fs type, bad superblock, blah blah' 00:39 < mistik1> the debian image support ext3 yes? 00:39 < jp7x7> It says so in /proc/filesystems, but... 00:39 < mistik1> nm 00:39 < jp7x7> Let me try an ext2 00:40 < mistik1> you should certainly be able to mount to image 00:41 < mistik1> unless the perms on the file are not in favor of the user running uml 00:41 < jp7x7> They are owned by me and uml is run by me 00:41 < mistik1> just for kicks 00:42 < mistik1> can you mount the image outside of uml? 00:42 < jp7x7> yes 00:42 < jp7x7> I can loop it. copy fiels, etc. 00:42 < mistik1> I've never had this problem so i'm just chasing down the obvious 00:43 < jp7x7> still appreciated 00:43 < jp7x7> even weirder... I just tried with new_fs as ext2. When I issued mnt command in UML it mounted my root_fs again on /mnt 00:44 < jp7x7> I know I didn't accidentally use the wrong command 00:44 < jp7x7> on boot: ubd1=new_fs 00:44 < jp7x7> in UML: mount -t ext2 /dev/ubd1 /mnt 00:44 < jp7x7> (none):~# df -h 00:44 < jp7x7> Filesystem Size Used Avail Use% Mounted on 00:44 < jp7x7> /dev/ubd0 58M 49M 6.3M 89% / 00:44 < jp7x7> /dev/ubd1 58M 49M 6.3M 89% /mnt 00:44 < jp7x7> (none):~# 00:44 < mistik1> lol 00:44 < jp7x7> craziness 00:45 < mistik1> that's bloody odd 00:45 < mistik1> but i'm sure the kernel supports ext3 at this point 00:45 < caker> jp7x7: print the output of this for me: ls -d /dev/ubd* 00:45 < caker> er make that a long list -l too 00:46 < jp7x7> I've been fighting this thing for three days and I'm not *quite* a newbie, but I figure there must be something wrong with the packaged uml kernel 00:46 < jp7x7> do you want me to do it from UML or host? 00:46 < caker> uml has ubd devices, not host :-) 00:46 < mistik1> my mind was going in a curve 00:46 < jp7x7> oops 00:46 < caker> I want to see the major/minor numbers 00:47 < jp7x7> brw-rw---- 1 root disk 98, 0 Jul 28 2002 /dev/ubd0 00:47 < jp7x7> brw-rw---- 1 root disk 98, 1 Jul 28 2002 /dev/ubd1 00:47 < jp7x7> brw-rw---- 1 root disk 98, 10 Jul 28 2002 /dev/ubd10 00:47 < jp7x7> brw-rw---- 1 root disk 98, 100 Jul 28 2002 /dev/ubd100 00:47 < jp7x7> brw-rw---- 1 root disk 98, 101 Jul 28 2002 /dev/ubd101 00:47 < jp7x7> brw-rw---- 1 root disk 98, 102 Jul 28 2002 /dev/ubd102 00:47 < jp7x7> brw-rw---- 1 root disk 98, 103 Jul 28 2002 /dev/ubd103 00:47 < jp7x7> brw-rw---- 1 root disk 98, 104 Jul 28 2002 /dev/ubd104 00:47 < jp7x7> I just did the fist few... 00:47 < caker> wow 00:47 < caker> I'm thinking that's incorrect :-) 00:47 < jp7x7> Yes, there are that many 00:48 < caker> difference between each minor should be 16 00:49 < caker> It's the /dev/ubd entries that are wrong, not your kernels or host... 00:49 < jp7x7> If I understand you comment... looking at all of them (there are 255) the major number is 98, the minor number counts up by one for each device 00:50 < caker> Someone just set it up strange ... there's a few ways to do it 00:50 < jp7x7> So... e.g. your saying that ubd0 should be 98, 0 ; and ubd1 should be 98,16 ? 00:50 < caker> but if you're going for ubd1 on the command line = /dev/ubd1 inside uml, you've got to run this: 00:50 < caker> (inside uml) - 00:50 < caker> cd /dev 00:50 < caker> rm -rf ubd* 00:51 < caker> err, 1 sec 00:51 < jp7x7> wont that kill my running system? 00:52 < caker> jp7x7: yes 00:52 < caker> jp7x7: it won't kill your system 00:52 < jp7x7> I mean my uml system 00:52 < caker> try this instead: mount /dev/ubd16 /mnt 00:52 < caker> that's ubd1 on the command line 00:53 < jp7x7> That worked! 00:54 < jp7x7> but wait... 00:55 < jp7x7> on the host system, the file new_fs shows up as 8.4 MB, in UML it shows up as 98M ??? 00:57 < mistik1> its a sparse file 00:57 < jp7x7> is that because I used the "seek=" option to dd? 01:05 -!- fiddo [~fiddo@dialup-195-242.conceptual.net.au] has joined #uml 01:09 -!- fiddo [~fiddo@dialup-195-242.conceptual.net.au] has quit [] 01:09 -!- fiddo [~fiddo@dialup-195-242.conceptual.net.au] has joined #uml 01:10 < fiddo> hello all 01:10 < fiddo> i have a question regarding representing UML packages 01:12 < david> yes 01:13 < fiddo> when drawing the overall project using packages, I design it in such a way as the packages 'mimic' the way the classes reside in their respective namespaces... 01:13 < fiddo> so basically the 'package' representa a 'namespace'. But... 01:14 < david> woh 01:14 < david> I think you have the wrong UML 01:15 < fiddo> this logical structure of the code is different to the way the files are physically stored (obviously). So my question is do I design ANOTHER package diagram for the pysical storage of the files? 01:16 < jp7x7> this UML = User Mode Linux rather than Uniform Modeling Language (or whatever) 01:16 -!- _fiddo_ [~fiddo@dialup-195-242.conceptual.net.au] has joined #uml 01:16 < fiddo> omg! 01:16 < fiddo> dont i feel silly 01:17 < _fiddo_> f 01:17 -!- _fiddo_ [~fiddo@dialup-195-242.conceptual.net.au] has left #uml [] 01:19 < fiddo> ok :) 01:19 -!- fiddo [~fiddo@dialup-195-242.conceptual.net.au] has left #uml [] 01:32 < jp7x7> Thank you caker! My UML system is now working! Networking as well. You tipped me off when you asked me about permissions on the filesystem files. Thos were ok, but I didn;t have permission for the /dev/net/tun file on the host. 01:32 < caker> jp7x7: no problem - glad it worked 02:01 < BB> morn 02:01 -!- jp7x7 [~jp7x7@216.187.157.238] has quit [Ping timeout: 490 seconds] 02:03 < caker> unified modeling language .. anyone use that, or enjoy using it rather? 02:03 < caker> Maybe I've never built anything big enough, but I think its a pain in the ass 02:11 < green> it is ;) 02:11 < green> esp. if your projects cannot be clearly mapped to OO semantic ;) 02:14 -!- jp7x7 [~jp7x7@216.187.157.238] has joined #uml 02:16 -!- jp7x7 [~jp7x7@216.187.157.238] has quit [Client Quit] 02:40 -!- mistral [mistral@212.159.71.212] has joined #uml 03:47 < BB> i had fun at uni, spent 2yr learning their modeling methods, then came to do my final yr project and neither me or the lecturers could figure out how to represent an X11 app in it 04:33 < green> hehe 04:53 -!- ichilton [~ian@80.6.255.203] has quit [Ping timeout: 488 seconds] 04:57 -!- ichilton [~ian@pc3-stoc3-4-cust203.midd.cable.ntl.com] has joined #uml 08:19 -!- litost [~mondain@phynp6.phy-astr.gsu.edu] has joined #uml 09:10 -!- litost [~mondain@phynp6.phy-astr.gsu.edu] has quit [Quit: Run Away!! Run Awaayyyy...] 09:21 -!- DnsInfector [~DnsInfect@193.252.189.93] has joined #uml 09:25 -!- Electric1lf [david@CPE0040f6b429d8-CM400049226055.cpe.net.cable.rogers.com] has joined #uml 09:25 -!- ElectricElf [david@elf.noc.oftc.net] has quit [Killed (NickServ (GHOST command used by Electric1lf))] 09:25 -!- Electric1lf is now known as ElectricElf 09:36 -!- ElectricElf [david@elf.noc.oftc.net] has quit [Quit: Reboot] 09:56 -!- ElectricElf [david@elf.noc.oftc.net] has joined #uml 10:05 -!- jdike [~jdike@dhcp37.ISTS.dartmouth.edu] has joined #uml 10:05 < jdike> hi guys 10:11 < green> Hi Jeff 10:13 < jdike> Hi Oleg 10:16 < mistik1> hey jdike, green 10:17 < DnsInfector> i have a comment on the download page of http://user-mode-linux.sourceforge.net/ 10:17 < DnsInfector> s/of/at/ 10:18 < DnsInfector> when i try to download something the download doesn't start automatically 10:18 < DnsInfector> when i download things from other projects at sourceforge everything is fine 10:20 < jdike> DnsInfector: Well, I don't have any control over that at all 10:21 < jdike> DnsInfector: that is totally run by sf 10:21 < DnsInfector> yes but 10:21 < DnsInfector> i remember that downloads were working fine some time ago 10:21 < DnsInfector> perhaps something has changed at sourceforge in the download policy 10:22 * DnsInfector is comparing downloads that don't work with one that do 10:22 < jdike> DnsInfector: maybe 10:22 < jdike> DnsInfector: I have lots of other mirrors, so if you don't like SF, just use one of the others 10:22 < DnsInfector> sorry for all 10:22 < DnsInfector> it looks like it works now 10:23 * DnsInfector slaps himself 10:26 < DnsInfector> this is quite off topic 10:27 < DnsInfector> but after hearing the gentoo story 10:27 < DnsInfector> i wanted to know if there's some support from 10:27 < DnsInfector> enterprises behind UML ? 10:28 < jdike> DnsInfector: not really 10:28 < DnsInfector> i've seen that many university seems to use UML 10:29 < DnsInfector> is it a university project ? 10:29 < jdike> DnsInfector: there are small tidbits here and there, but nothing major 10:29 < jdike> DnsInfector: No 10:30 < DnsInfector> this is still Open Source spirit ;) 10:30 < jdike> DnsInfector: it's my spare time project 10:30 < DnsInfector> great job 10:31 < DnsInfector> your spare time is quite creative 10:32 * jdike is borderline insane 10:32 < mistik1> heh 10:33 < mistik1> jdike: is that what you call it? 10:35 < Lathiat> jdike: i think you crossed the border when u started kernel hacking :P 10:35 < jdike> Lathiat: hehe 10:47 < ido> assuming i'd like to install a distro off a CD, could i run a temp UML, mount the CD, chroot it, and install it like i would if it weren't a UML ? 10:49 < DnsInfector> except the hardware detection it should work 10:50 < ido> but everything else would, right ? 10:51 < jdike> ido: you generally have to tweak things in order to fake out the installer 10:51 < jdike> ido: device names in particular 10:52 < ido> mknod the hd devices ? 10:53 < jdike> ido: with major 98 10:53 < DnsInfector> it may be the case where devfs is not recommanded 10:53 < jdike> devfs is definitely not recommended 10:53 < ido> gonna try installing slackware 10:54 < jdike> ido: slackware is good, you can tell it where to install from, where to install to, and it just does it 10:54 < jdike> ido: no faking needed 10:54 < ido> jep. 10:54 < ido> eh. 10:54 < ido> so i just need to chroot the CDROM mount, right ? 10:54 < ido> any other guidelines ? 10:54 < jdike> ido: the other distros are "smarter" 10:55 < jdike> ido: you don't need to chroot anything 10:55 < ido> so what do i do ? 10:57 < ido> oh 10:57 < ido> eh 10:57 < ido> i see 10:57 < ido> that incase i have it already mounted 11:01 < ido> hmm 11:02 < ido> on the slackware UML howto on SF, they say host% linux ubd0=inst_fs_slackware_7_0 ubd1=root_fs_slackware ubd2r=/dev/cdrom devfs=nomount rw , what does ubd0=inst_fs_slackware_7_0 refer to ? 11:02 < Lathiat> the install cdrom? 11:02 < ido> isn't that /dev/cdrom ? 11:02 < Lathiat> uh hrm, thats a good question 11:02 < jdike> ido: you have to pick apart the boot image 11:03 < ido> what do you mean by that? 11:03 < jdike> ido: inst_fs_blahblah is the little root filesystem inside it 11:03 < ido> uh ? 11:04 < ido> and if i've got the slackware ISO image, how would i install it from there other then first burning a CD ? 11:05 < jdike> ido: run file on the boot image, that will tell you what you need to do to start picking it apart 11:05 < ido> oh i see, inst_fs_slackware... is a temp fs that it boots into - some starting point that we could run the installation from, right ? 11:06 < jdike> ido: yeah 11:06 < ido> and what did you mean by your last sentence ? 11:09 < ido> on how to install it from the CDROM image 11:41 < david> hello 12:13 < BB> why do i have to move flat the day my new server arrives 12:14 < ido> eh 12:14 < ido> bummer 12:15 < ido> jdike: the link on SF on the slackware howto, to Martin Volf's hints doesn't work anymore 12:48 < jdike> green: there? 12:50 < green> kind of 12:51 < jdike> green: I want to submit some vmalloc memory for IO 12:51 < jdike> green: I'm thinking virt_to_phys, then get a struct page *, then create_empty_buffers 12:52 < jdike> green: the IO is sub-page in size, so it won't span pages 12:52 < jdike> green: is this the right way? 12:53 < green> I do not know, never tried that. But that might work 12:54 < green> this virt_to_phys -> ? -> struct page * thing looks tricky and fragile to me. Also may be vmallocked pages are specially marked. But you will never know until you try ;) 12:55 < jdike> green: I have to make sure that vmalloc mappings don't change until the IO is done 12:56 < jdike> green: maybe copying into a kmalloc buffer and submitting that would be better 12:56 < green> why do you want to kmalloc? just do getblk() and then submit the buffer yuo got 12:57 * jdike reads more code 12:57 < jdike> green: OK, getblk, copy data into bh->b_data, submit to ll_rw_blk? 12:58 < green> yes. 12:58 < jdike> green: sounds sane, thanks 13:12 -!- yeti [~yeti@p3EE22973.dip.t-dialin.net] has quit [Ping timeout: 492 seconds] 13:24 -!- nick [anandpanta@PPP-219-65-132-185.bng.vsnl.net.in] has joined #uml 14:07 < david> jdike: what do I have to do to get SMP support into skas? 14:07 < david> jdike: er, rather, what do I have to do to encourage you to put SMP support into skas? 14:08 < jdike> david: hmmm, that's a slightly different question :-) 14:08 < jdike> david: Add a few extra hours to my days? 14:09 < david> jdike: sleep less then 14:09 < david> :-) 14:09 < jdike> david: hehe 14:10 < jdike> david: You want it to be able for differential hosting? 14:10 < jdike> err 14:10 < jdike> You want it to for differential hosting? 14:10 < jdike> err 14:10 < jdike> You want it for differential hosting? 14:16 < david> jdike: mostly so Apache doesn't suck on my UMLs 14:16 < jdike> david: why would smp help? 14:17 < david> jdike: because two Apache processes could do something at the same time, rather than one 14:17 < david> jdike: it made an improvement when I was running tt 15:05 -!- nick [anandpanta@PPP-219-65-132-185.bng.vsnl.net.in] has quit [] 15:29 -!- revenger_ [revenger@p508B2ACC.dip.t-dialin.net] has joined #uml 15:31 -!- revenger2 [revenger@p508B36F0.dip.t-dialin.net] has quit [Ping timeout: 490 seconds] 16:48 -!- QSECOFR [~rocky@lab.atipa.com] has joined #uml 16:50 -!- sssssssh______ [~chatzilla@adsl-19-239-102.jax.bellsouth.net] has joined #uml 16:52 < QSECOFR> ac patches no longer contain uml? 16:53 < jdike> QSECOFR: they haven't for a long time 16:53 < DnsInfector> in which tree ? 16:54 < DnsInfector> 2.4 or 2.5 ? 16:54 * jdike assumed 2.4 16:54 < jdike> It should be in 2.5-ac 16:54 < DnsInfector> skas patch hasn't been proposed for integration to marcelo ? 16:54 < QSECOFR> 2.4, sorry. 16:54 < jdike> DnsInfector: not yet 16:55 < jdike> DnsInfector: it doesn't exist in a form that could be acceptable 16:55 < DnsInfector> what acceptable means ? 16:55 < jdike> DnsInfector: acceptable interface 16:55 < DnsInfector> ok 16:55 < jdike> DnsInfector: i.e. /proc/mm has to go 16:55 < QSECOFR> jdike: ok, was just following the howto. 16:56 < jdike> QSECOFR: some versions mentioned are horribly old 16:56 < DnsInfector> jdike: the acceptable patch should add a syscall ? 16:56 < jdike> QSECOFR: I've been pondering automating that so the docs always refer to the most recent UML 16:57 < jdike> DnsInfector: two new syscalls 16:57 < jdike> DnsInfector: new_mm and mm_indirect 16:59 < DnsInfector> there's something in the pipe or this is planned for 2.5 16:59 < DnsInfector> ? 16:59 < jdike> DnsInfector: probably not 2.5, too late 16:59 < jdike> DnsInfector: maybe 2.6 17:00 < DnsInfector> you mean the version after 2.6 ? 17:00 < jdike> DnsInfector: no 17:01 < jdike> DnsInfector: it's non-intrusive enough that it could get into 2.6 17:01 < DnsInfector> ok 17:26 < QSECOFR> dam..thought i read there was a amd64 port. 17:27 < DnsInfector> i also heard about a freeBSD and win32 port 17:29 < jdike> QSECOFR: standard UML runs on x86-64 17:30 < jdike> DnsInfector: win32 == umlwin32.sf.net 17:30 < jdike> DnsInfector: there's someone working on a FreeBSD port, but nothing public yet 17:31 < DnsInfector> that's what i had heard 17:31 < DnsInfector> the BSD port should be as hard as the win32 one 17:32 < jdike> DnsInfector: no, FreeBSD is pretty close to Linux 17:32 < jdike> DnsInfector: that would probably be the easiest OS port 17:32 < DnsInfector> the win32 port already has releases, hum impressive 17:33 -!- jdike [~jdike@dhcp37.ISTS.dartmouth.edu] has quit [Quit: Leaving] 17:42 -!- DnsInfector [~DnsInfect@193.252.189.93] has quit [Remote host closed the connection] 17:44 < Pahan> omfg eek 17:44 < Pahan> umlwin32 is active and releasing? 17:44 * Pahan bounces happily. 17:45 < Pahan> ugh wtf 17:45 < Pahan> Lies, that thing is dead. 17:45 < Pahan> It's been dead since2002. 17:50 -!- DnsInfector [~DnsInfect@ARennes-204-1-3-93.w193-252.abo.wanadoo.fr] has joined #uml 17:51 < QSECOFR> dam.. where'd he go? it does run on x86-64? there's no stuff in arch/um for it? 17:51 < Pahan> DnsInfector: You liar, umlwin32 is dead!1 17:52 < DnsInfector> the web site ? 17:52 < DnsInfector> it was slow but not dead 17:52 < Pahan> The project. 17:53 < Pahan> Last release is from 1st of July 2002. 17:53 < Pahan> Did it move off SF? 17:53 < DnsInfector> i just had a quick look 17:53 < Pahan> TOO QUICK 17:53 < DnsInfector> U're right 18:32 < david> hello 18:41 -!- mistral [mistral@212.159.71.212] has quit [Ping timeout: 488 seconds] 20:43 -!- DnsInfector_ [~DnsInfect@81.53.250.50] has joined #uml 20:43 * DnsInfector_ is going to sleep 20:43 -!- DnsInfector_ [~DnsInfect@81.53.250.50] has left #uml [] 20:51 -!- DnsInfector [~DnsInfect@ARennes-204-1-3-93.w193-252.abo.wanadoo.fr] has quit [Ping timeout: 496 seconds] 22:27 -!- gump [gump@81.5.136.90] has joined #uml --- Log closed Fri Jun 27 00:00:00 2003