Back to Home / #uml / 2007 / 02 / Prev Day | Next Day
#uml IRC Logs for 2007-02-12

---Logopened Mon Feb 12 00:01:00 2007
00:26|-|infowolfe [] has quit [Ping timeout: 480 seconds]
00:26|-|infowolfe [] has joined #uml
00:34|-|infowolfe_ [] has joined #uml
00:40|-|infowolfe [] has quit [Read error: Operation timed out]
02:10|-|SNy [] has quit [Remote host closed the connection]
02:15|-|SNy [] has joined #uml
03:03|-|tyler [~tyler@] has joined #uml
03:09|-|alb [] has quit [Read error: Connection reset by peer]
03:11|-|albertito [] has joined #uml
03:24|-|da-x_ [] has quit [Ping timeout: 480 seconds]
03:39|-|da-x_ [] has joined #uml
03:40|-|kokoko1 [~Slacker@] has joined #uml
04:26|-|motp [~motp@] has joined #uml
04:50|-|motp [~motp@] has quit [Quit: Leaving]
05:32|-|albertito [] has quit [Read error: Connection reset by peer]
05:37|-|albertito [] has joined #uml
05:46|-|tchan [] has quit [Server closed connection]
05:47|-|tchan [] has joined #uml
06:44|-|jjkola [] has joined #uml
07:24|-|AllenJB [] has quit [Server closed connection]
07:25|-|AllenJB [] has joined #uml
07:38|-|fbenites [~chatzilla@] has joined #uml
08:10|-|Hunger [] has quit [Server closed connection]
08:12|-|Hunger [] has joined #uml
08:39|-|Netsplit <-> quits: fbenites
08:39|-|Netsplit over, joins: fbenites
09:10|-|da-x [] has quit [Server closed connection]
09:11|-|da-x [] has joined #uml
09:53|-|jdike [] has joined #uml
09:53<jdike>Hi guys
09:53<infowolfe_>mornin jdike
09:53<infowolfe_>how was the weekend?
09:53|-|infowolfe_ changed nick to infowolfe
09:53<peterz>hey Jeff
09:53<peterz>jdike: you're aware that 3 level pagetables on um-i386 give scary warnings?
09:54<jdike>runtime warnings, I take it?
09:54<peterz>it gives warnings about long >> 32
09:54<peterz>compile time
09:54<jdike>Oh, OK
09:55<peterz>it ought to just work out runtime, but seeing those warnings in that code made me shiver
09:55<jdike>I haven't played with that in a while
09:55<peterz>right, the config option just happened on somehow, and scared me a little until I figured out what was going on
09:55<jdike>that was done as part of the x86_64 port - peel that out into 32-bit UML to separate the debugging
09:56<jdike>it's only useful if you want highmem
09:56<jdike>and most people are happy with the 2.xG you can get without it
09:57<peterz>um on i386 can't really access more than 4gb physical memory, can it?
09:57<peterz>hmm, actually, with all different processes it might
09:58<jdike>no, it can't
09:58<jdike>and the limit right now is 2.somethingG
09:58<jdike>PAE is a hardware idea
09:58<peterz>yeah, real sick one
09:59<jdike>UML exists in software-land, where you get a 4G address space
09:59<infowolfe>sorry, didn't see the 'um' portion
10:22|-|mode2_ [] has joined #uml
10:25|-|mode2 [] has quit [Read error: Operation timed out]
10:34|-|hfb [] has joined #uml
10:34|-|hfb [] has left #uml []
10:35|-|weasel [] has quit [Quit: Changing server]
10:36|-|weasel [weasel@2001:858:10f:6::1] has joined #uml
11:02<kokoko1>hi jdike
11:02<kokoko1>I am wondering do we have any other option to increase UML ram?
11:03<kokoko1>one method i know that is stop/start uml :S
11:03|-|writerz_ [] has joined #uml
11:04<jdike>kokoko1, UML supports hotplug memory
11:04<kokoko1>jdike, actaully i have to assign ram to uml from host RAM
11:05<jdike>you mean you want it guaranteed in RAM, not swapped?
11:06<kokoko1>sorry, imean i want to increase running UML RAM by +128M
11:06<kokoko1>Without stop/starting UML ( just wondering if it is possible)
11:07<jdike>hotplug memory
11:07<jdike>the catch is you can only increase up to what the UML started with
11:07<jdike>So, if you start it with a lot, then pull a bunch out immediately, then you have room to increase it later
11:08<kokoko1>so if uml started with 256M i can make it 256+128 ?
11:09<jdike><jdike> the catch is you can only increase up to what the UML started with
11:09|-|mode2_ [] has quit [Ping timeout: 480 seconds]
11:09|-|Netsplit <-> quits: fbenites
11:09|-|Netsplit over, joins: fbenites
11:12<kokoko1>up 25 days
11:12<kokoko1>its the same uml that was crashing after day or two
11:13<kokoko1>Apparently we didn't do anything, one stopped on spam monitoring script which was running from cron every 5 minutes and taking lot of cpu
11:13<kokoko1>after stopping that script mail server UML working fine from last 25 days
11:14<jdike>I was hoping to see a core dump from it
11:14<kokoko1>Duno it never dumpped it.
11:16<jjkola>jdike: if you need something to do, try running detachtty + sbcl in a uml so that you connect with slime to that, it makes system crawling
11:16<jdike>UML system?
11:17<jjkola>what do you mean?
11:17<jdike>i.e. the UML crawls, not the host?
11:18<jjkola>it affects whole system
11:18<jjkola>as it's taking all hard drive resources
11:20<jjkola>I tried also using caker's limiter patch but then it started to use cpu time and I had to renice the uml process
11:21<kokoko1>Nice article about UML
11:22<kokoko1>who's the author of this article ? :)
11:23<jdike>By: Marc Abramowitz
11:23<kokoko1>jdike, you read the comments?
11:23<kokoko1>I trried to follow along after downloading the kernel and FS but it seems the FedoraCore5-x86-root_fs file system has been built with the wrong /lib/modules/*:
11:24<kokoko1>I wonder where one would report this on the UML site
11:24<kokoko1>I thinks its time you could help to that comment :)
11:25<kokoko1>heheh yeah i missed the author name at top of the page :S
11:26|-|mode2_ [] has joined #uml
11:26<jdike>it doesn't really matter usually
11:26|-|nessie [] has quit [Remote host closed the connection]
11:26|-|nessie [] has joined #uml
11:27<jdike>plus including modules ties the filesysem to the kernel version
11:29<jdike>but probably when you get a kernel and a filesystem from the same place, the modules should match
11:31|-|writerz_ [] has quit [Ping timeout: 480 seconds]
11:42<kokoko1>jdike, it a nice and informative article ins't ?
11:43<jdike>yeah, I skimmed it, but it looked decent
11:43<kokoko1> /. it :D
11:45<kokoko1>I just submitted it to let see if they pass it
11:45|-|writerz_ [] has joined #uml
11:46<kokoko1>jdike, i am wondering why author pointed "User Mode Linux" to
11:46<kokoko1>User Mode Linux (UML) allows you to run ....
11:46<kokoko1>It pointed to wrong site
11:47<kokoko1>while submitting to root.prom.... i corrected that link :S
11:50|-|mode2_ [] has quit [Ping timeout: 480 seconds]
12:07|-|mode2_ [] has joined #uml
12:12|-|writerz_ [] has quit [Ping timeout: 480 seconds]
12:17|-|flatronf700B [~flatronf7@] has quit [Read error: Operation timed out]
12:27<kokoko1>Dike has mentioned the possibility of a SKAS4 patch, so depending on when you're reading this, it might already be available.
12:31|-|kos_tom [] has joined #uml
12:36|-|writerz_ [] has joined #uml
12:39|-|mode2_ [] has quit [Ping timeout: 480 seconds]
12:45|-|Netsplit <-> quits: fbenites
12:45|-|Netsplit over, joins: fbenites
12:47<dgraves>kokoko1: yeah, its been mentioned for a while now.
12:47<dgraves>but not finished.
12:48|-|ram [] has joined #uml
12:55|-|Netsplit over, joins: fbenites
12:56<kokoko1>dgraves, hiya
12:56<kokoko1>any idea squid can proxy ftp ?
13:09<kokoko1>Yep squid do it via CONNECT method
13:10|-|fbenites_ [~chatzilla@] has joined #uml
13:10|-|Netsplit <-> quits: fbenites
13:11|-|fbenites_ changed nick to fbenites
13:33|-|da-x_ [] has quit [Ping timeout: 480 seconds]
13:50|-|da-x_ [] has joined #uml
14:20|-|baroni [] has joined #uml
14:22|-|ram [] has quit [Ping timeout: 480 seconds]
14:39<dgraves>kokoko1: hiya yourself. :)
14:45|-|ram [] has joined #uml
14:45|-|infowolfe_ [] has joined #uml
14:49|-|Cowboy_ [~Cowboy@] has joined #uml
14:52|-|infowolfe [] has quit [Ping timeout: 480 seconds]
14:55|-|Cowboy_ [~Cowboy@] has left #uml [Client exiting]
15:09|-|albertito [] has quit [Read error: Connection reset by peer]
15:14|-|albertito [] has joined #uml
15:20|-|tyler [~tyler@] has quit [Read error: Operation timed out]
15:21|-|tyler [~tyler@] has joined #uml
15:24|-|cmantito [] has quit [Read error: Operation timed out]
16:19<uroboros_>I am running 2.6.20 and ARCH=um 2.6.20 with ext3 ubd (and yes, I removed /lib/tls) - the UML boots up cleanly, everything is OK. Then I try to connect to /dev/pts/X with screen like this: ``screen /dev/pts/X'' and I get login. I type root as an username (there is no password yet) and I get /etc/motd etc. Then the logging in process freezes. :(
16:19<uroboros_>Can anyone help me with this?
16:24<jdike>2.6.20 has good TLS support, BTW
16:25<jdike>what was the command line?
16:27<uroboros_>cmdline for the vmlinux?
16:29<uroboros_>give me few secs to login to my notebook I am experimenting with uml on
16:29<uroboros_>(to copy/paste) ;)
16:30<uroboros_>I will put it in a pastebin, one moment please...
16:31<jdike>it can't be that long
16:31<uroboros_>to rewrite it to pasteble form in here will be longer than to paste it all in pastebin ;)
16:31<uroboros_>It is just a very simple script.
16:32<jdike>you can log in on the main console?
16:32<uroboros_>ubd1 is swap
16:32<uroboros_>I can log in, I type username and get a banner and then ... freezes :(
16:32<jdike>on the main console?
16:33<jdike>before, you were talking about screen
16:33<uroboros_>Obviously some problem running the shell. If /lib/tls is not removed then I get errors from init.
16:33<uroboros_>jdike: yes, screen, the first console /dev/pts/FIRST
16:33<jdike>well, how about the main console
16:33<uroboros_>jdike: the others does not work at all, dunno why yet.
16:33<uroboros_>jdike: ??
16:34<jdike>the one in your xterm window?
16:34<uroboros_>I do not use xterm.
16:34<jdike>wherever the boot output is
16:34<uroboros_>No, no login is offered (and never was if I remember cleanly)
16:34<uroboros_>If you mean the same console Irun uml from.
16:35<jdike>well, adding a :0 to your inittab will help that
16:35<jdike>what's the init error?
16:35<uroboros_>I will have to rename the /lib/tls dir. back, wait please...
16:36<uroboros_>Wov! I started OK.
16:36<uroboros_>W/ it did not. :) Great work! :)
16:36<uroboros_>Now I try to login...
16:37<jdike>2.6.19 has the same support
16:37<uroboros_>Freeze :(
16:37<uroboros_>I will show you.
16:39|-|orionrobots [~danny@] has joined #uml
16:40<uroboros_>Here it is. <- freezes in this state :(
16:41<uroboros_>It seems as if it can not run shell for some reason or something like that...
16:42<uroboros_>I have no network setup in my UML image yet, so I can not test it w/ SSH for the very moment.
16:42<jdike>can you get a shell with init=/bin/bash?
16:43<jdike>what's the UML doing on the host? sleeping? running?
16:43<uroboros_>I will try the init=/bin/sh , great idea! ;)
16:43<uroboros_>simple and effective :)
16:45<uroboros_>I am in,
16:45<uroboros_>init=/bin/sh worked
16:46|-|orionrobots [~danny@] has left #uml []
16:46<jdike>can you cut down the .login, if there is one?
16:46<uroboros_>Ok, when in, I changed the shell for root to /bin/sh from /bin/bash and tried normal bootup. The same result as before, the banner and then freeze :(
16:47<uroboros_>jdike: .login?
16:47<uroboros_>in /root/.login ?
16:47<jdike>just so there is nothing but a shell running when you log in?
16:47<jdike>.profile too
16:48<uroboros_>there's .bashrc and .profile in /root, I will rename both of them, ok?
16:49<uroboros_>Same result :(
16:49|-|kos_tom [] has quit [Quit: I like core dumps]
16:49<uroboros_>I can as well rename /etc/bash_bashrc etc. files.
16:50<uroboros_>But I think there is no mistake in them at all.
16:50<uroboros_>I can strace the vmlinux, perhaps I will see something strange in the strace's output.
16:50<jdike>add a :0 entry to inittab which starts a shell instead of getty
16:51<jdike>that will give you a way to look at what's going on
16:51<uroboros_>ah, it failed with strace ;)
16:51<uroboros_>OK, I will add will check the inittab
16:52<caker>Dike says that Xen "doesn't fit in the Linux world," and called it "a technological dead end." He predicts a "family of virtualization stuff under KVM." Whatever makes it into the mainline kernel, says Dike, is what the distros will follow.
16:52<caker>jdike: you go! :)
16:53<jdike>I was actually a little worried about the reaction inside Intel to that
16:53<jdike>given the amount of effort they're putting in Xen
16:54<jdike>there was another article about my talk titled "Intel boosts Linux virtualization" that got some (positive) attention
16:54<jdike>Computer Weekly or some such
16:54<caker>yeah, read that one too
16:55<jdike>people liked the title as much as the actual content
16:55<caker>So -- the work the Moscow guys did, to support VT -- there was a host component to it essentially does what KVM exposes?
16:55<caker>and that will be going away in favor of KVM?
16:55<jdike>I'm actually mourning the fact that we could have done KVM 6 months - 1 year earlier
16:55<caker>does that fall under your responsibilities, or will the Moscow guys take KVM and run with it?
16:55<jdike>we just missed the boat
16:55<uroboros_>jdike: what do you call :0 entry in inittab?
16:55|-|tyler [~tyler@] has quit [Remote host closed the connection]
16:56<caker>bummer (re 1 year earlier)
16:56<uroboros_>jdike: sorry, but I did not hear (and see) anything similar in the /etc/inittab
16:56<jdike>uroboros_, what do you main?
16:57<jdike>never mind, hold on
16:57<uroboros_>jdike: you mean to replace the "l2:2:wait:/etc/init.d/rc 1" with "l2:2:wait:/bin/sh" ?
16:57<uroboros_>jdike: for instance...
16:57<jdike>0:2345:respawn:/sbin/mingetty tty0
16:58<jdike>except /bin/bash in place of /sbin/mingetty tty0
16:59<uroboros_>INIT: can not execute "/sbin/mingetty" perhaps there no this file, I will check it...
16:59<caker>does that work without fancy /dev/tty0 redirects?
16:59<caker>like: 0:2345:respawn:/bin/bash -login >/dev/tty0 2>&1 </dev/tty0
17:00<jdike>well, whatever the other entries say
17:00<jdike>yeah, maybe
17:00<jdike>I was hoping something simple and obvious would work
17:00<jdike>we just need a no-login shell prompt someplace
17:01<uroboros_>so, if not mingetty, what about getty?
17:01<uroboros_>can Iuse getty for it?
17:01<caker>ok .. the above code works for me under x86 linux (using ttyS0)
17:01<caker>uroboros_: try my line above
17:04<uroboros_>jdike: OK, system booted with caker's inittab line, what next?
17:04<jdike>you have a shell now?
17:05<uroboros_>in the console I ran it from? no
17:05<uroboros_>no, no shell
17:05<uroboros_>Last lines are: Virtual console 1 assigned ... etc. 6 times.
17:05<jdike>well, there should be
17:05<jdike>the whole point of this was to get a shell without having to log in
17:06<uroboros_>The distro inside is Debian/testing (clean debootstrap)
17:06<caker>uroboros_: did you boot with con=null con0=fd:0,fd:1 (is that the default? I always specify on the command line anyway)
17:06<uroboros_>I used it several times on 2.6.something_very_old and worked (but without the /lib/tls stuff only)
17:07<uroboros_>I booted with the script you saw using it as "uml 1"
17:08<uroboros_>I thought the parameters are OK. Perhaps they are not...
17:08<uroboros_>I can get rid of the COW file.
17:08<caker>that looks ok to me
17:08<uroboros_>for me to
17:09<uroboros_>I get one idea. I did no post-install actions. I just debootstrapped in the UBD root partition (Ext3), and set a hostname. Nothing else.
17:10<uroboros_>Perhaps that is wrong and I missed some important step or something.
17:10<uroboros_>But I simply can not imagine such step to disfunt the login. :(
17:10<uroboros_>Forget about what I have just said. I simply should work
17:11|-|jjkola [] has quit [Quit: *pop*]
17:12<uroboros_>I have the strange feeling that it must be something very obvious. You know the feeling you know you have it just before your eyes and you can not see it... um. :(
17:14<uroboros_>I can also compress the whole image and put it somewhere to download.
17:15<uroboros_>the FS looks clean (I tune2fs -c1 on it so fsck runs every bootup)
17:15<uroboros_>I am out of ideas. :(
17:16<uroboros_>caker: btw: when make ARCH=defconfig and then make ARCH=menuconfig -> where the hell is SELinux? :-D
17:19<uroboros_>caker: "make ARCH=um menuconfig" in ^^ of course, sorry. I started to be a little tired (sitting all the day at computer is no healthy)
17:23<uroboros_>Perhaps I added something I shouldn't into the kernel. Perhaps I forgot something? Do you wanna see my .config for the vmlinux?
17:26<uroboros_>Hey, that is really strange!
17:27<uroboros_>When I type fast, the keys are stored in buffer in opposite dirrection inside the uml! wtf?! ;)
17:28<uroboros_>In vi, <esc>:wq lasts in :qw, if I type it very slowly it is OK. LOL
17:28<uroboros_>And the same is for the login screen. "root" became "orot" ;)
17:30<uroboros_>OK, I go sleep. I am really terribly tired today. Tomorrow I will try to make new images w/ rather Debian/sarge. We will see. I will let you know. Good night.
17:43|-|jdike [] has quit [Quit: Leaving]
17:44|-|cmantito [] has joined #uml
17:59<uroboros_>Last idea, I will check tomorrow - for some *unknown* reason waiting for DNS?
18:01<uroboros_>Or, perhaps - too less space left on device?
18:02<uroboros_>I will let you know.
20:42|-|flatronf700B [~flatronf7@] has joined #uml
21:06|-|albertito [] has quit [Read error: Connection reset by peer]
21:08|-|albertito [] has joined #uml
21:22|-|baroni [] has quit [Ping timeout: 480 seconds]
21:23|-|baroni [] has joined #uml
22:07|-|kokoko1 [~Slacker@] has quit [Read error: Operation timed out]
22:32|-|baroni [] has quit [Remote host closed the connection]
22:46|-|Nem^ [] has quit [Ping timeout: 480 seconds]
22:57|-|Nem^ [] has joined #uml
22:59|-|VS_ChanLog [] has left #uml [Rotating Logs]
22:59|-|VS_ChanLog [] has joined #uml
23:37|-|infowolfe_ [] has quit [Quit: Leaving]
23:46|-|ram [] has quit [Ping timeout: 480 seconds]
23:49|-|infowolfe [] has joined #uml
---Logclosed Tue Feb 13 00:00:32 2007