Back to Home / #uml / 2007 / 10 / Prev Day | Next Day
#uml IRC Logs for 2007-10-19

---Logopened Fri Oct 19 00:00:53 2007
02:10|-|aroscha [] has quit [Quit: aroscha]
03:13|-|pieter [] has joined #uml
03:13<pieter>hi guys
03:20<Magotari>Hello, pieter.
03:21<pieter>with uml, can i export the real serial ports / network cards to the vm ?
03:22<Magotari>I don't think you can. Maybe serial ports could be done, but not real network cards, I don't think.
03:23<Magotari>You can put UML onto the network, but I don't think you can have it operate as a driver of a real network card.
03:23<Magotari>In fact, if you look inside UML's config, any real hardware you see is a bug.
03:23<Magotari>Now, serial ports are a different story. You could use some redirection to get there. Close enough, I hope.
03:24<pieter>Thanks, that should be fine. :]
03:25<Magotari>Glad to be of help. If you have more questions about UML, feel free to ask.
03:25<Magotari>I am here for half an hour more before I have to go.
03:38<pieter>How do I make my own filesystem / linux layout? instead of the Fedora core one provided on the website?
03:38<pieter>(a link would be fine if you know of any)
03:38<Magotari>pieter: I have an idea how to do it which no one knows about. :)
03:38<Magotari>Use Qemu to put a distro onto a disk image.
03:39<Magotari>Then use UML to boot it. Use just one partition for the whole disk, it will do just fine. To boot it you need root=/dev/ubda1 in your uml commandline.
03:39<Magotari>There is a different method too, the proper one.
03:39<Magotari>But I could never be bothered with it. Qemu is just faster.
03:40<pieter>Cool, do you have a link to the proper one? Just for interest sake
03:40<Magotari>Sure. Let me find it.
03:42<pieter>Thanks! :)
03:42<Magotari>In the end the proper method is harder to get started on, needs more work to boot the installer.
03:42<Magotari>However, it gives you a chance to edit fstab and other files.
03:43<Magotari>Also if you do it right you might not need the root=/dev/ubda1 line.
03:43<pieter>I'll read through it and then decide yeah. :]
03:43<Magotari>Of course with the quick and dirty method you can extract the partition with dd, if you know a bit about harddisk layouts. (I don't.)
03:46<Magotari>Ah, one more idea would be to use a loopback mount and copy the files you need. Easy to miss an important file that way though.
03:58<Magotari>pieter: I am going off now. I will be back in about 4-5 hours for a short while and then later in the day. jdike comes here nearly every day also in about 5-6 hours from now.
03:58<pieter>Thanks man, enjoy
03:58<Magotari>He knows a lot more than I do, he might be of help as well.
03:58<Magotari>See you later.
04:23|-|Ancalagon [] has joined #uml
04:24|-|Baltam8 [] has quit [Remote host closed the connection]
04:26|-|tyler29 [] has joined #uml
04:27|-|Baltam8 [] has joined #uml
04:29|-|Ancalagon [] has left #uml []
04:40|-|tyler29 [] has quit [Ping timeout: 480 seconds]
04:50|-|tyler29 [] has joined #uml
04:58|-|tyler29 [] has quit [Ping timeout: 480 seconds]
05:08|-|tyler29 [] has joined #uml
05:11|-|ftumch [] has quit [Quit: Goodbye.]
05:19|-|ftumch [~James@] has joined #uml
05:37|-|tyler29 [] has quit [Ping timeout: 480 seconds]
05:54|-|tyler29 [] has joined #uml
06:03|-|tyler29 [] has quit [Ping timeout: 480 seconds]
06:19|-|aroscha [~aroscha@] has joined #uml
06:24|-|tyler29 [] has joined #uml
06:28|-|aroscha [~aroscha@] has quit [Ping timeout: 480 seconds]
06:28|-|Urgleflogue [~plamen@] has quit [Ping timeout: 480 seconds]
08:04|-|tyler29 [] has quit [Ping timeout: 480 seconds]
08:11|-|pieter [] has quit [Read error: Connection reset by peer]
08:15|-|pieter [] has joined #uml
08:21|-|tyler29 [] has joined #uml
08:32|-|dang [] has joined #uml
08:46|-|aroscha [] has joined #uml
09:16|-|pieter [] has quit []
09:40|-|baroni [] has joined #uml
10:01|-|jdike [] has joined #uml
10:01<jdike>Hi guys
10:06|-|tyler29 [] has quit [Ping timeout: 480 seconds]
10:21<kokoko1>Hello jeff
10:22|-|arun [] has joined #uml
10:29|-|tyler29 [] has joined #uml
10:29|-|Magotari [] has quit [Quit: Lost terminal]
10:55|-|linbot [] has quit [Ping timeout: 480 seconds]
10:56|-|linbot [] has joined #uml
12:00|-|kos_tom [] has joined #uml
12:45|-|tyler29 [] has quit [Ping timeout: 480 seconds]
12:49|-|karol [] has joined #uml
12:49|-|karol changed nick to magotari
12:49|-|magotari changed nick to Magotari
12:52|-|duns_s [] has joined #uml
12:53<duns_s>how can i work around the 2 gb max file size in hostfs (without network filesystems)?
12:56<Magotari>duns_s: Splitting the big file?
12:57<duns_s>okay here is what i am doing/want to do:
12:58<duns_s>i want to run rtorrent (and most likely amule), a webserver and ssh for some file sharing with friends in an uml on my server
12:58<duns_s>i have added iptables rules on the host so that uml cannot reach the internal network except the dns server
12:59<duns_s>when files are completed in rtorrent, i want rtorrent to put them on my external usb drive with all the media files (via hostfs, since i would like to avoid putting the uml machine in my network)
13:00<Magotari>How about using the usb drive as ubdb?
13:00<duns_s>that would be an option
13:01<duns_s>then i would have to export the external drive to all the network as a nfs/samba share
13:01<duns_s>the only problem i see here is: a) if the machine gets hacked all my media may be lost (very bad)
13:01<duns_s>b) the performance of uml is not the greatest
13:02|-|tyler29 [] has joined #uml
13:02<duns_s>the killer feature imho is hostfs
13:02<Magotari>Did you try Skas3? I have not tried it yet, but I hear it is a bit faster.
13:02<Magotari>Just a bit, but it does add up.
13:03<Magotari>The data loss is a problem.
13:03<duns_s>it is not a problem at the moment, but i think it could be heavy with a 460 GiB file share on a via c3/512 MB ram
13:04<duns_s>I just don't get why hostfs restricts me to 2 GiB, this is stupid.
13:04<duns_s>It stops you from accessing files on host's fs errors anyway, doesn't itß
13:04<Magotari>I don't think it would be that bad. If you do get problems switch to Xen or OpenVZ. But UML should handle most loads just fine. Not stellar performance, but at least stable.
13:04<Magotari>No idea about the 2G limit.
13:05<Magotari>I could take a look at the source code and see what is going on.
13:05<kokoko1>duns_s, wait for jdike
13:05<duns_s>Magotari: that would be great!
13:05<duns_s>kokoko1: okay thanks
13:06<Magotari>Well, no. Beause I am just learning C. :/ So I might not figure it out.
13:06<Magotari>I do have time to burn though.
13:09<duns_s>I'll have a look, too. I'll be back after dinner.
13:09<duns_s>Cu later...
13:17|-|mgross [] has joined #uml
13:20<Magotari>I cannot figure it out. :/ So much for that. Still lots to learn. Damn.
13:53|-|tyler29 [] has quit [Ping timeout: 480 seconds]
13:55<duns_s>jdike: ping
14:01|-|baroni [] has quit [Read error: Connection reset by peer]
14:06|-|tyler29 [] has joined #uml
14:08|-|aroscha [] has quit [Quit: aroscha]
14:25<duns_s>Magotari: I've found out that it cannot write files larger than 2 GB -1 Byte if open64 is not used
14:26<duns_s>but as far as I can see it is used correctly.
14:26<duns_s>I have never looked at the kernel code before though.
14:27<duns_s>SSIZE_MAX could be a problem, too, I don't know how this is set in arch=UML
14:28|-|remus [~remus@] has quit [Remote host closed the connection]
14:30<duns_s>Okay, leave me a message, I'll be back later.
14:32|-|cl4sh [] has joined #uml
14:33|-|cl4sh [] has left #uml []
14:34|-|cl4sh [] has joined #uml
14:34<cl4sh>hi all
14:35<cl4sh>ive got little problem with compilation of vnuml under gentoo
14:36<cl4sh>the error message is
14:36<cl4sh>i don't know where i can find XML::Checker
14:38<Magotari>Maybe in the perl cheese store?
14:38<Magotari>Erm no.
14:38<Magotari>Python has the cheese store. Perl has cpan.
14:40<Magotari>duns_s: I don't know C well enough, so I have no idea.
14:41<cl4sh>Magotari: yes i was saw that
14:42<cl4sh>Magotari: bu i wander how to emerge it in gentoo
14:42<Magotari>Hey, a gentoo user. Just like me. :) Let me see...
14:42<cl4sh>Magotari: i was try parser and DOM
14:42<cl4sh>Magotari: :)
14:46<Magotari>Does not seem to be in Portage. But maybe this...
14:46<Magotari>I have never used it, so caveat. Might work though.
14:47<cl4sh>Magotari: that looks quite interesting
14:53<cl4sh>Magotari: OMG it works BTW very useful tool thx a lot
14:56|-|remus [~remus@] has joined #uml
15:01<jdike>there shouldn't be a 2G limit
15:04<jdike>UML is O_LARGEFILE everywhere
15:05<jdike>might be userspace within UML
15:18<duns_s>jdike: how can i check that
15:19<duns_s>i've downloaded an opensuse image with 4.2 gb and rtorrent crashed when writing to it.
15:20<duns_s>jdike: Has this been fixed?
15:20<jdike>it's from 2003, I'd imagine so
15:21<jdike>see if you can dd a chunk from > 2G out of the file
15:21<duns_s>this one, too, but in german:
15:26<duns_s>jdike: can i do something like skip=1950 bs=1M ?
15:26<jdike>except something bigger than 1950 would be good
15:27<duns_s>jdike: how do you mean that?
15:27<jdike>would be 3G
15:28<duns_s>jdike: yes, okay, i've thought about skip=1950 bs=1M count=200, because it actually reads the data, but your approach is even faster
15:30|-|dang [] has quit [Quit: Leaving.]
15:34[~]jdike gotta go
15:34|-|jdike [] has quit [Quit: Leaving]
15:59|-|remus [~remus@] has quit [Remote host closed the connection]
16:18|-|remus [~remus@] has joined #uml
16:24<duns_s>i can read files greater than 2 gb, but i cannot write them
16:25|-|fo0bar_ changed nick to fo0bar
16:33|-|mgross [] has quit [Ping timeout: 480 seconds]
16:48<remus>I am running uml under gdb and it crashes when inserting a module
16:48<remus>anyone else has seen this problem ?
16:48<remus>Program received signal SIGSEGV, Segmentation fault.
16:48<remus>0x0000000060012d56 in copy_chunk_from_user (from=1723686928, len=<value optimized out>, arg=0x67fafdd8)
16:48<remus> at arch/um/kernel/skas/uaccess.c:137
16:48<remus>137 arch/um/kernel/skas/uaccess.c: No such file or directory.
16:48<remus> in arch/um/kernel/skas/uaccess.c
16:49<remus>I can post the backtrace if someone is interested
16:51|-|ram [] has quit [Ping timeout: 480 seconds]
16:53<remus>Magotari, caker ?
16:54<Magotari>Hmm. No idea.
17:02|-|ram [] has joined #uml
17:03<ds2>what does x/1b 1723686928
17:16|-|krau [~cktakahas@] has quit [Quit: Varei!!!]
17:16|-|tyler29 [] has quit [Ping timeout: 480 seconds]
17:28<duns_s>if i can read files larger than 2 gb, but cannot write them on hostfs, what could be the problem?
17:50|-|kos_tom [] has quit [Quit: I like core dumps]
17:53|-|aroscha [] has joined #uml
18:02|-|duns_s [] has quit [Remote host closed the connection]
18:48<newbie>Should I worry about the following messages at UML startup:
18:48<newbie>Checking syscall emulation patch for ptrace...missing
18:48<newbie>Checking for tmpfs mount on /dev/shm...OK
18:48<newbie>Checking PROT_EXEC mmap in /dev/shm/...OK
18:48<newbie>Checking for the skas3 patch in the host:
18:48<newbie> - /proc/mm...not found: No such file or directory
18:48<newbie>the missing lline and the /proc/mm concern me - AND the uml hangs after mounting filesystems - exactly when it should be running init
18:49<Magotari>newbie: What is your host kernel?
18:51<newbie>one sec
18:53<newbie>the guest kernel is
18:53<newbie>the rootfs is copied from an fc3 machine running 2.6.12-6
18:54<newbie>(actually the image is from the same machine I'm using as the host system)
18:54<Magotari>Erm, ok. That is an old kernel.
18:54<newbie>yes, i know - it's what I need it to run on though
18:55<newbie>what kernel feature is /proc/mm involved with?
18:55<Magotari>Upgrade to something newer to get the syscall emulation patch, which will boost speed. THe mm is fine, not a reason to worry.
18:55<newbie>oh ok
18:55<Magotari>I don't remember, some memory thingy I suppose.
18:55<newbie>how new do I need to be?
18:55<Magotari>Well, you don't have to.
18:55<Magotari>Unless you are going in for speed or bugfixes.
18:56<Magotari>Try to download a kernel and rootfs from the uml website.
18:56<Magotari>See if those boot.
18:56<Magotari>If yes, something is wrong with your setup.
18:56<Magotari>If they don't at least ponder a kernel upgrade.
18:56<Magotari>I cannot say for sure.
18:56<Magotari>The guy who does know will be here in about 15 hours.
18:56<newbie>those boot fine - we have a standard environment that I need to virtualize so I would prefer to keep as close to it as possible
18:57<Magotari>Now try to boot your rootfs with the kernel from the website.
18:57<newbie>in fact the fc3 pristine from nagafix seems to boot fine
18:58<newbie>that's the kernel I'm using - I saw no need to build one from scratch unless necessary
18:59<Magotari>Diagnosis: Your rootfs is broken somehow. Maybe try root=ubda1 ?
18:59<Magotari>Erm no.
18:59<newbie>hmm ok
18:59<Magotari>Sorry, 2 in the night here.
18:59<newbie>llet's see...
18:59|-|albertito [] has left #uml []
19:00<newbie>I'm using /dev/ubda on the command line and in /etc/fstab for the guest, I also created the dev structure in the root fs using MAKEDEV ubd
19:00|-|aroscha [] has quit [Quit: aroscha]
19:00<newbie>hmm so I should refer to it as ubda1? why?
19:01<newbie>I don't think I have any partitions...
19:01<Magotari>Oh, no need for the 1 then.
19:01<newbie>hmm...I have a thought - could i have broke the cow/root combbo by mucking with the root without using the cow
19:01<newbie>easy enough to test that
19:01<Magotari>Yes, could be.
19:01<newbie>I will toss the cow file and have it create a new one...
19:01<newbie>back in a bit
19:03<Magotari>rootfs and cow must have same mtime
19:03<newbie>Any idea why uml creates so many processes??? Seems to create 30 that I need to kill EVERY TIME I start it
19:03<newbie>ahh ok - I hoped it would complain loudly in such a case but I guess not - let me go test that
19:03<Magotari>and changing a rootfs makes all cows invalid.
19:04<Magotari>on the phone
19:05<newbie>ok, it says it created a new cow for me...still hangs while booting though
19:05<newbie><no problem - let me know when you can talk>
19:05<newbie>back shortly - more research
19:07<Magotari>is your guest uml running in skas0 mode?
19:08<Magotari>Only 4 processes per UML in skas0.
19:14<Magotari>Off the phone.
19:15<Magotari>To correct myself, jdike never comes here on the weekend, so he won't be here till monday.
19:15<Magotari>He would know what to do.
19:16<newbie>maybe I should leave it till monday then
19:16<newbie>obviously I'm not in skas0 mode
19:16<Magotari>Yeah, I am new here too, I don't know the answers to many things.
19:16<newbie>also just noticed I have a messy mount of loopback devices at the moment which may have broken it
19:17<newbie>in fact it looks like I have mounted the same image twice on /mnt and even umount -f won't dump it
19:17|-|ram [] has quit [Read error: Operation timed out]
19:17<newbie>not my machine so I can't just reboot it either
19:17<newbie>back in a minute
19:17<Magotari>Yeah, weird.
19:18|-|krau [~cktakahas@] has joined #uml
19:19<newbie>i'm back - the person who's machine it is not around right I guess I just need too leave it - the image that is mounted is NOT the one that I'm actively running though
19:20<newbie>but I had the active one mounted at /mnt/img1 and the other at /mnt/img2 and apparently somehow also had the other one at /mnt
19:20<newbie>really bad -
19:20<newbie>things should be sorted welll enough now though - let me try the one I want to get working again and see what happens
19:22<newbie>no luck. I did notice that it claims to be running in SKAS0 mode though
19:22<Magotari>That is good.
19:22<Magotari>TT mode was insecure and slow.
19:22<newbie>yes good..still puzzling though -
19:22<newbie>why so many procs
19:22<Magotari>I get four.
19:22<Magotari>Per UML.
19:23<Magotari>No idea why so many indeed.
19:24<Magotari>An idea, if it is not critical, maybe you could umount /mnt/
19:25<newbie>I think they are threads. pstree shows one parent process
19:25<newbie>I tried umount -f /mnt
19:25<newbie>it failed
19:27<Magotari>"Device busy" ?
19:29<Magotari>I'm sorry, I am all out of ideas for now. No idea what is going on either.
19:29<Magotari>I gotta get to sleep.
19:29<Magotari>I will be here tomorrow, all day, so if you have any other questions...
19:30<newbie>Ok, thanks for your help!
19:30<newbie>Get some sleep now.
19:42<newbie>gott it (the unmount prob at least)
19:42<newbie>umount -fl /mnt
19:42<newbie>is your friend when it seems that you somehow mounted a loopback image twice on /mnt
19:43<newbie>ok that's done....will sort the rest out later (ie Monday)
20:38|-|ElectricElf [~dbharris@] has quit [Ping timeout: 480 seconds]
20:46|-|ElectricElf [~dbharris@] has joined #uml
21:00|-|remus [~remus@] has quit [Remote host closed the connection]
21:02|-|Nem^ [] has quit [Read error: Operation timed out]
21:16|-|cl4sh [] has quit [Quit: Lost terminal]
21:17|-|Nem^ [] has joined #uml
22:58|-|VS_ChanLog [] has left #uml [Rotating Logs]
22:58|-|VS_ChanLog [] has joined #uml
---Logclosed Sat Oct 20 00:00:26 2007