--- Day changed

--- Log opened Thu Jul 03 00:00:01 2003
00:01 < david> Pahan: so fix it, make a patch and sent it to the maintainer
00:02 < Pahan> Right.
00:02 < Pahan> Took me the better part of the day to find it.
00:25 -!- ticallion [~ticallion@213.175.160.230] has joined #uml
00:27 < ticallion> greets folks
00:29 < david> hi
00:29 < ticallion> the david?!
00:31 < david> which david?
00:31 < ticallion> coulson, lxf articles and what nots?!
00:31 < david> yeah
00:32 < ticallion> very cool articles dude... I scour them insanely... congrats :)
00:32 < david> thanks :-)
00:32 < Dave\\> wow it's you!
00:35 < david> apparently ;-)
00:36 < ticallion> I'm looking for work abroad... lxf made me contemplate england, but the problem is u people don't have ski resorts :P and I can't live without my yearly dose of snowboarding mayhem
00:37 < ticallion> I might settle for those indoor thingies in something-keynnes and stuff :)
00:37 < ticallion> it's amazing how lxf and linu-mag uk outdo anything in their american couterparts
00:39 < david> :-)
00:39 < david> ticallion: scotland has ski resorts, as does the lake district
00:40 < ticallion> david: how far by train is scotland from downtown london?! I mean could u go there for the day to ski/snowboard?!
00:41  * ticallion wisps over to mozilla and types snowboarding.org.uk *cross fingers*
00:42 < david> ticallion: er, it's like 6hrs by train
00:42 < david> ticallion: you can fly it in maybe 1:30hrs
00:42 < david> btw, no one says 'downtown' in ENgland :-)
00:42 < david> center of london is 'central london'
00:43 < ticallion> I expected that, felt uneasy typing that :)
00:45 < david> ticallion: :-)
00:47 < ticallion> well, england has its attractions, first off, sexy lin mags, unrivaled worldwide, second, the english r the most fanatical about tuning they subaru imprezas which is my fav. car of all time, third, umm?! year-round indoor snowboarding which is a slight turn on, what else?! I really dunno how the people r, I'm travelling in 2 months to check out france, england, canada, I'll make my mind up by then 
00:48 < david> ticallion: You can get Linux magazines anywhere :-)
00:48 < ticallion> david: I know dude, but that says that linux has market penetration and a respectable geek following too 
00:49 < david> ticallion: in parts - it's not as good as the US
00:49 < ticallion> but I'm sorry to say, Colin McRae gets on my nerves, I absolutely despise that creature
00:49 < david> who the hell is that? ;-)
00:49 < ticallion> david: I beg to differ... their mags suck, why would that be if the states had "better" geeks
00:49 < ticallion> ?!
00:50 < ticallion> I'm so serious, I'm not joking btw
00:50 < ticallion> u get the feeling, that the admins over there just wanna make things work, no passion
00:51 < caker> bahh -- users
00:51 < ticallion> I can't wait till I test my extremely sexy and very much tree-like iptables setup on a uml distro
00:51 < david> ticallion: it depends - There is quite  Linux following in the US, or at least that is what I experienced
00:51 < ticallion> btw, david
00:51 < ticallion> I was about to take the vserver aproach until I read ur article 
00:52 < david> the one in LXF, or the one on my site?
00:52 < david> vserver sucks ;-)
00:52 < ticallion> david: lxf dude... haven't read the one on ur site yet
00:52 < david> caker: hey, did Greg mail you again today?
00:52 < david> ticallion: oh, ok - That's about a year old ;-)
00:53 < ticallion> david: indeed... numbers, yes... but quality *shrug*, I can't say for sure, I'm only going by the quality of the magazines, the lin websites and the general feel of webhost pages in the us :P
00:53 < david> ticallion: Sysadmin magazine is good - Can't say I'm a fan of Linux Journal
00:53 < caker> david: he mailed me yesterday, yes
00:53 < ticallion> david: well, I only contemplated such a setup recently... was working mainly on my gateway "product" for quite a while now
00:54 < david> caker: ok, just wondering
00:54 < caker> david: did he mention it?
00:54 < david> caker: yeah, we talked about a few things eariler today
00:54 < david> ticallion: my gateway runs entirely under UML
00:54 < david> ticallion: works nicely
00:54 < ticallion> david: and I remembered seeing an article on uml, I found it yesterday, read it, did some research today, and decided on uml, then reread ur article
00:54 < ticallion> david: wow, that's cool
00:55 < david> UML is very nice - Needs some extra features to make it *really* useful, but on the whole it's very productive
00:56  * ticallion hastles the chan dwellers with his vision of a perfect life for the couple of years to come: "civilized country, own place, rack/s, broadband, snowboarding, impreza wrx (tuned of course)... in order of importance...
00:56 < ticallion> david: I was surprised to see the timestamp on the 2.4.19 patch
00:57 < david> ticallion: which 2.4.19 patch?
00:57 < caker> david: get your ebtables up and running?
00:57 < ticallion> david: btw, I've looked at virt hosts, the majority seem to be using intermezzo's "true vps"
00:57 < david> caker: yep, works nice
00:57 < ticallion> david: umm, uml patch
00:57 < david> ticallion: oh
00:57 < david> ticallion: I think that is kernel level virtualization - Lower overhead
00:57 < caker> "true vps" = contexts
00:58 < caker> and a shared kernel :(
00:58 < ticallion> caker: yeah, right...
00:58 < caker> There are a lot more limitations to the shared kernel/context approaches then appear at first
00:59 < ticallion> btw, I'm confused about a particular aspect of the whole contexts/uml thing, in contexts it's sposed to be easier to bork the "master" kernel (forgot what it's called), I think david in his article (and definitely in other resources) it's said that uml is safer in that aspect?! I don't get why?! since ALL calls r relayed to the host kernel?!
01:00 < david> ticallion: the UML kernel virtualises the system calls and passes them to the host kernel
01:01 < caker> uml runs like any other process on the host -- it's just another program running
01:01 < ticallion> david: could u elaborate a bit?! why is it safer?!
01:02 < ticallion> caker: fine... I get it, I think
01:02 < david> ticallion: because it's an isolated process
01:02 < ticallion> david: and in the contexts case, there no translation layer in between the vserver and the kernel?!
01:03 < david> ticallion: I've not looked at vserver in much depth - I thought it was just a crazy chroot with some IP binding stuff
01:03 < caker> ticallion: just the context is the only thing -- each user IS uid 0 (root) inside each context
01:04 < ticallion> caker: ouch, how did I miss that?! haven't figured that out while reasing about vserver
01:04 < caker> ticallion: yup - that's vserver's method .. it's pretty cool and works I suppose, but uml is better imho
01:05 < ticallion> anyway, let's take a purist aproach to the whole thing, vserver is just a bsd rip off, grrrr, *bsd gets my blood to a boil
01:05 < david> bbl
01:05  * ticallion waves
01:17 -!- ticallion [~ticallion@213.175.160.230] has quit [Ping timeout: 488 seconds]
01:54 -!- ticallion [~ticallion@213.175.160.112] has joined #uml
02:29 < Pahan> Hrm, a paused UML still answers ICMP echo-replies?
02:31 < green> SMP?
02:32 < Pahan> No/.
02:32 < green> AH, perhaps even paused UML still serves interrupts then
02:32 < Pahan> I suppose the meaning of "stop" is "no user processes are scheduled"
02:33 < green> I do not know ;)
02:36 < Pahan> Hrm, so I got the basic switch daemon functionality working.
02:36 < Pahan> It does everything that uml_switch does, and more.
02:37 < Pahan> Looks like it's not too slow, either.
02:37  * Pahan kills it with macof.
02:38 < ticallion> is ashamed of using debian in a uml
02:38  * ticallion 's a RHL fanatic
02:38 < Pahan> Yours is a misguided loyalty.
02:43 < ticallion> Pahan: I would definitely expect that from a debian follower
02:43 < ticallion> s/follower/peasant/ if i knew u better
02:43 < ticallion> :P
02:44 < ticallion> I like RHL cos debian will never be a corporate linux solution
02:44 < ticallion> that's all
02:44 < Pahan> ticallion: How so?
02:45 < Pahan> And why are "corporate" solutions worth using?
02:46 < ticallion> Pahan: cos I'm a sysadmin, and I'm compulsive to boot, so I'd like to invest every second of my geeky existence in the pursuit, ah crap, forget that, in short, I use RHL as my workstation to stay on top of RHL advances (u would say, lack thereof)
02:46 < ticallion> :P
02:58 < green> ticallion: why don't you use RHL inside f UML jut like all the sane people do? ;)
02:59 < ticallion> green: cos I don't feel like rebooting, installing a minimal thing, rebooting into my workstation install then uml'ing
02:59  * green likes RH too, except their marketing practices :\
02:59 < ticallion> I'm too tired
02:59 < ticallion> :(
02:59 < green> ticallion: install it directly into UML ;)
02:59  * ticallion wants candy
02:59 < ticallion> green: I shouldn't be able to?!
02:59 < ticallion> green: I can?!
03:00 < ticallion> green: that would be uml'ing the install kernel?! I'm guessing
03:00 < green> http://linuxhacker.ru/uml
03:01 < ticallion> green: funny thing is, I loaded that doc, but didn't bother to read it, I assumed debian could do it out-of-the-box and RHL couldn't, cos I'm too tired to read before I go offline and am too tired to research such a trivial step
03:01 < ticallion> in short
03:01 < ticallion> I'm going crazy out of lack of sleep
03:02 < green> hehe, get a sleep and then read the doc ;)
03:03 < ticallion> I should :)
03:26 -!- ticallion [~ticallion@213.175.160.112] has quit [Ping timeout: 490 seconds]
04:17 -!- snide [~snide@81.53.158.107] has joined #uml
04:20 -!- ticallion [~ticallion@213.175.160.31] has joined #uml
04:23 -!- ticallion[afk [~ticallion@213.175.160.31] has joined #uml
04:24 -!- ticallion [~ticallion@213.175.160.31] has quit [Client Quit]
04:28 -!- ticallion[afk is now known as ticallion
04:30 -!- mistral [mistral@212.159.71.212] has quit [Ping timeout: 488 seconds]
04:34 -!- lah [~z@AVATAR.internet.ufg.ac.at] has quit [jupiter.oftc.net arion.oftc.net]
04:49 -!- revenger2 [revenger@80.139.29.134] has joined #uml
04:51 -!- revenger_ [revenger@p508B170B.dip.t-dialin.net] has quit [Ping timeout: 490 seconds]
05:04 -!- mistral [mistral@jstevenson.plus.com] has joined #uml
05:11 -!- DnsInfector [~DnsInfect@80.15.51.253] has joined #uml
05:28 -!- ticallion [~ticallion@213.175.160.31] has quit [Quit: Client exiting]
05:49 -!- lah [~z@AVATAR.internet.ufg.ac.at] has joined #uml
07:12 -!- DnsInfector [~DnsInfect@80.15.51.253] has quit [Ping timeout: 488 seconds]
07:57 -!- mistral [mistral@jstevenson.plus.com] has quit [charon.oftc.net uranium.oftc.net]
07:57 -!- Pahan [pahan@syr-24-59-98-110.twcny.rr.com] has quit [charon.oftc.net uranium.oftc.net]
07:57 -!- ido [~Ido@212.150.75.108] has quit [charon.oftc.net uranium.oftc.net]
07:57 -!- lah [~z@AVATAR.internet.ufg.ac.at] has quit [charon.oftc.net uranium.oftc.net]
07:57 -!- ljlane [ljlane@aphrael.net] has quit [charon.oftc.net uranium.oftc.net]
07:57 -!- caker [~null@pcp507591pcs.nash01.tn.comcast.net] has quit [charon.oftc.net uranium.oftc.net]
07:57 -!- albino [~albino@neuf.dorms.usu.edu] has quit [charon.oftc.net uranium.oftc.net]
07:57 -!- Lathiat [lathiat@seven.sixlabs.org] has quit [charon.oftc.net uranium.oftc.net]
07:57 -!- Z-Wing [zedders@zanzar.net] has quit [charon.oftc.net uranium.oftc.net]
07:57 -!- pirlouit [~peter@64.162.195.202] has quit [charon.oftc.net uranium.oftc.net]
07:57 -!- QSECOFR [~rocky@lab.atipa.com] has quit [charon.oftc.net uranium.oftc.net]
07:57 -!- ichilton [~ian@pc3-stoc3-4-cust203.midd.cable.ntl.com] has quit [charon.oftc.net uranium.oftc.net]
07:57 -!- BB1 [~chris@62.3.208.26] has quit [charon.oftc.net uranium.oftc.net]
07:57 -!- kara [~kara@varstation.lanscape.net] has quit [charon.oftc.net uranium.oftc.net]
07:57 -!- silug [~steve@osiris.silug.org] has quit [charon.oftc.net uranium.oftc.net]
07:57 -!- green [green@linuxhacker.ru] has quit [charon.oftc.net uranium.oftc.net]
07:57 -!- BB [~chris@ns1.8086.net] has quit [charon.oftc.net uranium.oftc.net]
07:57 -!- cameron [~cameron@crosstalk.otago.ac.nz] has quit [charon.oftc.net uranium.oftc.net]
07:57 -!- dilinger [irc@sloth.voxel.net] has quit [charon.oftc.net uranium.oftc.net]
07:58 -!- ido [~Ido@212.150.75.108] has joined #uml
07:58 -!- Pahan [pahan@syr-24-59-98-110.twcny.rr.com] has joined #uml
07:58 -!- mistral [mistral@jstevenson.plus.com] has joined #uml
07:58 -!- lah [~z@AVATAR.internet.ufg.ac.at] has joined #uml
07:58 -!- ljlane [ljlane@aphrael.net] has joined #uml
07:58 -!- caker [~null@pcp507591pcs.nash01.tn.comcast.net] has joined #uml
07:58 -!- albino [~albino@neuf.dorms.usu.edu] has joined #uml
07:58 -!- Lathiat [lathiat@seven.sixlabs.org] has joined #uml
07:58 -!- Z-Wing [zedders@zanzar.net] has joined #uml
07:58 -!- pirlouit [~peter@64.162.195.202] has joined #uml
07:58 -!- QSECOFR [~rocky@lab.atipa.com] has joined #uml
07:58 -!- ichilton [~ian@pc3-stoc3-4-cust203.midd.cable.ntl.com] has joined #uml
07:58 -!- BB1 [~chris@62.3.208.26] has joined #uml
07:58 -!- kara [~kara@varstation.lanscape.net] has joined #uml
07:58 -!- silug [~steve@osiris.silug.org] has joined #uml
07:58 -!- green [green@linuxhacker.ru] has joined #uml
08:00 -!- cameron [~cameron@crosstalk.otago.ac.nz] has joined #uml
08:01 -!- ChanServ changed the topic of #uml to: User Mode Linux | 2.4.20-6um, 2.5.69-1um, 2.4.21-rc7-djc1-5um | http://user-mode-linux.sf.net | http://usermodelinux.org | http://usermodelinux.co.uk | http://uml.openconsultancy.com | http://www.stearns.org/slartibartfast | #uml Archives: http://www.theshore.net/uml/irc/
08:01 -!- dilinger [irc@sloth.voxel.net] has joined #uml
08:01 -!- charon.oftc.net changed the topic of #uml to: User Mode Linux | 2.4.20-6um, 2.5.69-1um, 2.4.21-rc7-djc1-5um | http://user-mode-linux.sf.net | http://usermodelinux.org | http://usermodelinux.co.uk | http://uml.openconsultancy.com | http://www.stearns.org/slartibartfast | #uml Archives: http://www.theshore.net/uml/irc/
08:01 -!- ChanServ changed the topic of #uml to: User Mode Linux | 2.4.20-6um, 2.5.69-1um, 2.4.21-rc7-djc1-5um | http://user-mode-linux.sf.net | http://usermodelinux.org | http://usermodelinux.co.uk | http://uml.openconsultancy.com | http://www.stearns.org/slartibartfast | #uml Archives: http://www.theshore.net/uml/irc/
08:40 -!- Getty [~Getty@217.69.76.51] has quit [Quit: Serverwechsel]
08:46 -!- Getty [~Getty@217.69.76.51] has joined #uml
09:19 -!- jdike [~jdike@129.170.249.137] has joined #uml
09:19 < jdike> hi guys
09:19 < Dave\\> hi jdike
09:22 < ido> hey guys :)
09:34 -!- ido [~Ido@212.150.75.108] has quit [Ping timeout: 488 seconds]
09:38 -!- ido [~Ido@212.150.75.108] has joined #uml
09:54 -!- BB [~chris@ns1.8086.net] has joined #uml
10:12 < green> Hi Jeff!
10:12 < green> jdike: Have you seen the solution to gdb vs gcc3 problems?
10:13 < jdike> Hey Oleg!
10:14 < jdike> green: did you see the post last week from someone who tracked it down to gdb being confused by symbols in UML?
10:14 < jdike> green: that's what you're talking about?
10:14 < green> yes, except that it was confused by incorrect context of some elf data
10:14 < green> s/context/content/
10:14 < jdike> green: was there a solution?
10:15 < jdike> green: that just looked like a diagnosis of the problem
10:15 < green> Yes. There was a tiny patch for gdb to not read that ELF section. And it works
10:15 < jdike> green: A patch to gdb?
10:15 < green> s/tiny patch/description of what to change/
10:15 < green> yes, to gdb.
10:15 < jdike> green: is the bug in gdb?
10:15 < green> I also contacted SuSE and asked them to fix the real bug, we'll what will come out of it
10:15 < green> no, the bug is in toolchain
10:16 < jdike> green: OK, but not in UML
10:16 < green> yes
10:16 < jdike> green: i.e. it's not using symbols or section names which have special meaning to the tools/gdb?
10:17 < green> hm. no, I think it is not. Anyway you cannot write to ELF section directly, I think
10:18 < jdike> green: OK, cool
10:18 < jdike> green: be better if it was something I could fix and make everyone happy, though
10:19 < green> At least we have some solution that works, and this is great.
10:19 < green> and may be SuSe guys will find the real problem
10:20 < jdike> green: yeah, it will be nice to have real stacks again :-)
10:20 < green> I enjoy reals stacks for 2 days already ;)
10:20 < jdike> green: is only UML having this problem, or do other apps see it?
10:20 < jdike> green: hehe
10:22 < green> I do not know, I have not done lots of userspace debugging lately.
10:36 -!- AquaJo [~torbofh@pD9559D3F.dip.t-dialin.net] has joined #uml
10:36 -!- ido [~Ido@212.150.75.108] has quit [Ping timeout: 488 seconds]
10:41 -!- ido [~Ido@212.150.75.108] has joined #uml
10:43 < snide> green: UML debugging *is* userspace debugging <grins>
10:44 < jdike> snide: shhhh
10:44 -!- Maniac [~daManiac@209-5-247-105.mb.skyweb.ca] has joined #uml
10:44 < jdike> snide: green thinks he's an 31337 kernel whacker
10:44 < jdike> snide: don't upset him :-)
10:44 < Maniac> can anyone answer a quick question?
10:45 < snide> jdike: oops...
10:46  * snide fiddles in drivers/block/loop.c
10:47  * jdike was just copying code from loop.c
10:47 < snide> jdike: the funny thing is that i crashed my linux box twice when i realised that i could also do it in UML ;-)
10:47 < Maniac> if i start like so: linux eth0=tuntap,,,192.168.10.51 etc.; all connection to my UML come from the 192.168.10.51 address
10:48 < snide> jdike: i don't really understand how loop.c works actually... 
10:48 < jdike> snide: hehe, crashing UML is a lot more fun
10:48 < Maniac> if I use UML for web serving how do i get external IP to show up?
10:48 < snide> jdike: ... and *much* less painful ;-)
10:49 < green> yeah.
10:49  * green just recovered from writing down a sysrq-T trace on a piece of paper
10:50 < green> if I'd make more such excercises, I'd be able to handwrite again ;)
10:50 < snide> green: why ? it was in dmesg anyway ;-)
10:50 < jdike> snide: not necessarily
10:51 < jdike> snide: not if you can't run dmesg
10:51 < snide> jdike: i know. usually when u do a sysrq-T, dmesg is already far far away
10:51 < snide> jdike: yup, that's what i meant
10:51 < green> snide: Well, it hung immediately after starting init, what dmesg are you apeaking of? ;)
10:52 < snide> green: i told u not to hardlink /sbin/init to /dev/null.....
10:52  * green needs to get longer serial cable, so that it can really reach some nearby box
10:53 < jdike> Maniac: what external IP?
10:54 < snide> btw, does someone know what a TLB is ? 
10:55 < jdike> snide: translation lookaside buffer
10:55 < snide> jdike: oh. thx
10:55 < snide> jdike: and, what does it mean ? ;-)))))))))))))
10:55 < green> hehe
10:55 < Maniac> jdike: well if someone outside accesses the web page, the ip address in access_log shows the 192.168.10.51 addy
10:55 < Maniac> instead of the ip of the box connecting
10:56 < green> snide: this is the secret knowledge, not for mere mortals like me and you ;)
10:56 < snide> Maniac: do u use a reverse proxy ? 
10:56 < snide> green: oh... that explains many things ;-)))
10:57 < jdike> snide: it's a little cache on the processor which holds recent virt to phys mappings
10:58 < snide> jdike: a sort of MMU cache ?
10:58 < jdike> snide: maybe, it might depend on the mmu
11:00 < Maniac> snide: heh, i'm stupid. I thought it was doing it for all services but of course when i ssh'd in it wasn't
11:00 < Maniac> so i spose it's an apache question now 
11:00  * Maniac slaps himself
11:03 < snide> jdike: do u know a good doc on Physical and Virtual Memory mapping ? 
11:04 < jdike> snide: not really
11:04 < jdike> snide: I've just sort of picked it up by osmosis
11:04 < snide> jdike: k
11:06  * snide is going back hom
11:06 -!- snide [~snide@81.53.158.107] has quit [Quit: [BX] Mr. T uses BitchX. Shouldn't you, SUCKAH?]
11:07 < david> Maniac: how are you forwarding the TCP connection to Apache?
11:08 < Maniac> ProxyPass
11:08 < david> ok
11:08 < david> so that'll screw it up
11:09 < Maniac> ja
11:09 < david> you need to use DNAT or something
11:09 < david> jdike: hello
11:09 < jdike> hi david
11:09 < jdike> david: how are things?
11:10 < david> jdike: good
11:10 < Maniac> i'm servering content on the host though, and just want to forward connection to certain sites to the uml box
11:10 < david> jdike: UML seems nice and stable at the moment
11:10 < david> jdike: needs more features to solve that problem ;-)
11:11 < david> Maniac: then you're stuffed then :-)
11:11 < jdike> david: that's just what I like to hear :-)
11:11 < Maniac> btw: all UML people i wuv UML. Had a UML instance running for 40+ days 
11:11 < Maniac> stable as heck
11:11 < david> Maniac: pass the REMOTE_ADDR from the host to the UML as an env-var then modify the access logs to use that rather than the REMOTE_ADDR from the actual proxied connection
11:12 < david>  16:09:52 up 99 days,  8:06,  2 users,  load average: 0.00, 0.00, 0.00
11:12 < Maniac> david: any clues to how I would do that? like where i would look or just google?
11:12 < Maniac> david: you win :P
11:12 < david> Maniac: check the mod_rewrite documentation
11:12 < green> jdike: yeah, we need more features. I have long list ;)
11:12 < jdike> green: hehe
11:12 < david> UML isn't fun anymore
11:12 < Maniac> david: ok cool
11:13 < david> like back in the day when it would be luck if it got to init or not :-)
11:13 < Maniac> i run gentoo in uml
11:13 < mistik1> why do all that
11:13 < jdike> it used to be that UML uptime contests were kind of pointless
11:13 < jdike> by the time you got a good uptime, the latest UML would have so many more bugs fixed
11:13 < david> jdike: indeed
11:14 < mistik1> Maniac: connect to my IP on port 780
11:14 < mistik1> via http
11:15 < green> david: bugs are necessary evil anyway
11:15 < green> s/necessary/unavoidable/
11:15 < jdike> green: bah
11:16 < david> green: I know but it gets boring when the bugs are just little things that are down to bad coding on Jeff's part
11:16 < green> hehe
11:16 < mistik1> someone connect to http://music.geeksinthehood.net:780
11:17  * jdike don't get no respect
11:17 < david> mistik1: worked
11:17 < mistik1> thanks
11:17 < green> 403 forbidden
11:17  * jdike goes to sulk in the corner
11:17 < david> jdike: get swsusp in UML and all will be forgiven ;-)
11:17 < jdike> after which I will add special bugs for david and green
11:17 < david> bbl
11:18 < mistik1> I did that for the benifit of Maniac 
11:18 < mistik1> 64.233.221.3 - - [03/Jul/2003:11:14:46 -0400] "GET /icons/apache_pb.gif HTTP/1.1" 200 2326 "http://music.geeksinthehood.net:780/" "Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.4a) Gecko/20030114"
11:18 < green> david: I'd vote for removing 2 more context switches and zerocopy UBD instead
11:18 < mistik1> Maniac: note the access log
11:18 < Maniac> mistik1: how?
11:18 < mistik1> that is apache running in uml
11:18 < mistik1> I dont use uml_net
11:19 < mistik1> I use a bridge and add my taps to the bridge
11:19 < mistik1> all my taps have no IP as a result
11:19 < Maniac> ok, my problem is i already have apache on the host serving content. I want to proxy additional content the uml
11:20 < mistik1> so I start uml like this 'linux eth0=tuntap,tapname
11:20 < mistik1> It should not matter
11:20 < green> jdike: BTW, I have a proof that skas mode is significantly slower under debugger at least for some tasks
11:21 < green> jdike: I can pass you a program that makes gdb to consume 80+% of cpu in skas mode, while tracing thread consumes much less CPU in tt mode. And tt mode works faster
11:21 < jdike> green: what does it do?
11:21 < green> it mmaps big file and then dirties all the pages sequentally
11:22 < jdike> green: interesting
11:22 < jdike> green: not obvious why there would be a speed difference
11:23 < green> seems that gdb gets called a lot more often that tt thread,I'd say. or may be switching to gdb is more expensive
11:39 < mistik1> 4 days and still going strong
11:39 < mistik1> thanks for a wonderfull project guys
11:40 < mistik1> jdike: did I tell you how i'm using UML right now?
11:40 -!- smcavoy [~smcavoy@207.61.160.163] has joined #uml
11:40 < jdike> mistik1: nope
11:40 < QSECOFR> x86-64 guest work on x86-64 host?
11:41 < mistik1> jdike: Well right now I have it acting as a terminal server, and have 2 diskless machines booted off it
11:41 < jdike> QSECOFR: there is no x86-64 guest
11:41 < jdike> QSECOFR: but the i386 UML works on x86-64
11:42 < mistik1> I was a but shocked at the performance of the thin-client running off a UML server
11:42 < AquaJo> hi
11:43 < jdike> mistik1: good or bad?
11:43 < mistik1> good
11:43 < mistik1> I've had people using the clients off an on over the last 4 days
11:43 < mistik1> no complaints so far
11:45 < AquaJo> jdike: ich shall greet you from Peter, one of the vd-server guys :-)
11:46 < QSECOFR> jdike: ok. thanks...wont help me then..:(
11:46 < jdike> AquaJo: cool
11:46 < jdike> AquaJo: BTW, tell him that the vd-server UML mirror is active, forgot to send him mail about it...
11:46 < AquaJo> he has allready noticed it
11:47 < QSECOFR> if i had de time, i'd take a stab at following the port-howto thangy...:(
11:48 < jdike> AquaJo: how's vd-server doing with UML so far?
11:48 < AquaJo> It's running very well, my only problem ist, with the latest patches I can't compile cipe and (not so important) dazuko
11:50 < david> green: 
11:50 < david> green: uh
11:50 < david> green: I think the two context switch removal may be waiting for the purging of /proc/mm :-)
11:51 < AquaJo> jdike: so we are still using the 'old' 2.4.19-50um
11:58 < david> AquaJo: ah, a nice rootable kernel - Might want to at least apply the ptrace patch to it ;-)
11:59 < AquaJo> no kernel module loader *g*
12:00 < AquaJo> and in the moment I get cipe running with an uml-patch, we will update all vd-Server (the host are running .21)
12:02 < jdike> AquaJo: is there arch-dependent stuff in cipe? (and what is it anyway?)
12:05 < david> jdike: thought it was some sort of network tunnel thing
12:05 < AquaJo> cipe ist an udp-based free s/wan, encryptet iptunnels over udp-base
12:05 < david> http://sites.inka.de/sites/bigred/devel/cipe.html
12:07 < AquaJo> I don't think it uses ARCh-dependant stuff, but it needs the includes, since it isn't a kernelpatch, an the includes sam seems to bee brocken or so
12:08 < jdike> AquaJo: that should be easy to fix
12:13 < jdike> AquaJo: change the CFLAGS definition to - CFLAGS=`cd path/to/uml/tree ; make script 'SCRIPT=echo $(CFLAGS)' ARCH=um`
12:13 < jdike> AquaJo: and that will pick up the correct flags and includes
12:13 < jdike> AquaJo: err
12:13 < jdike> AquaJo: CFLAGS=`cd path/to/uml/tree ; make script 'SCRIPT=@echo $(CFLAGS)' ARCH=um`
12:14 < AquaJo> I'll try it
12:26 -!- dilinger [irc@sloth.voxel.net] has quit [uranium.oftc.net infrared.oftc.net]
12:27 -!- dilinger [irc@sloth.voxel.net] has joined #uml
12:43 < AquaJo> so, now I'll go testing it
13:22 < jdike> gcc: Internal compiler error: program cc1 got fatal signal 11
13:22 < jdike> that gives me the warm fuzzies about gcc
13:31 -!- mdz__ [~mdz@h00034795fb2d.ne.client2.attbi.com] has joined #uml
13:31 -!- mdz__ [~mdz@h00034795fb2d.ne.client2.attbi.com] has left #uml []
13:32 -!- mdz_ [~mdz@24.91.103.187] has joined #uml
13:37 < AquaJo> thaks jeff, it seems to compile
13:37 < AquaJo> bugs.c: In function `disable_lcall':
13:37 < AquaJo> bugs.c:117: warning: implicit declaration of function `modify_ldt'
13:37 < AquaJo> the only messaages
13:39 < jdike> AquaJo: that's harmless
13:40 < AquaJo> <- has forgotten to run the configure file => no module ...
13:45 < AquaJo> with your parameters make throw no errors, but it doesnt compile
13:46 < AquaJo> s simpel make after the configure: ./configure --with-linux=/usr/src/linux-2.4.21 throws out tonns of errors
13:47 < AquaJo> lots of file not found
13:47 < AquaJo> I'll get something to eat
13:47 < AquaJo> cul8ter
13:49 -!- snide [~snide@81.53.158.107] has joined #uml
13:50  * snide is home...
13:51 < BB> put kettle on ;)
13:55 -!- snide [~snide@81.53.158.107] has quit [Quit: [BX] With a BitchX here and a BitchX there, here a BitchX there a BitchX everywhere a BitchX]
14:19 < Pahan> jdike: So! Describe the management interface for the vlan thingie.
14:19 < Pahan> jdike: It's a telnet session.
14:19 < jdike> Pahan: moi?
14:19 < jdike> Pahan: I'm not a vlan guy, how should I know?
14:20 < Pahan> jdike: No, I mean my switch daemon.
14:20 < Pahan> jdike: It's at the point where I can make it suppot vlans.
14:20 < Pahan> support*
14:20  * Pahan winks.
14:21 < jdike> Pahan: don't you also have to fiddle the driver?
14:21 < Pahan> jdike: What driver?
14:21 < jdike> Pahan: the UML network driver
14:21 < jdike> Pahan: to put vlan tags on the frames?
14:21 < Pahan> Nope. I tried, and two UMLs can communicate via vlan over uml_switch
14:21 < Pahan> You use the 8021q module.
14:21 < Pahan> kernel module
14:22 < jdike> Pahan: interesting, so my driver handle the larger frames transparently?
14:22 < jdike> Pahan: OK, you're basing this on uml_switch, correct?
14:22 < Pahan> jdike: No, I rewrote it in Python.
14:23 < jdike> Pahan: hmmm
14:23 < Pahan> There is a bunch of NIC drivers that choke on larger frames. Your doesn't ;-)
14:23 < jdike> Pahan: if I'm going to support it, it needs to be added to uml_switch
14:23 < Pahan> jdike: You don't need to do anything to support it.
14:23 < BB> i think 90% of nic drivers dont support vlan properly
14:24 < Pahan> BB: The 8021q driver homepage lists a bunch of pactches to make at least some drivers work properly.
14:24 < david> why do you need VLAN tagging at all?
14:24 < david> if you control the switching device
14:24 < Pahan> david: jdike said some people are interested in heaving that.
14:24 < BB> Pahan I know, I've used a few already
14:24 < Pahan> having*
14:24 < david> then you can create VLANS within the switch and add specific ports to it
14:25 < jdike> Pahan: my thinking is to add a command line to the switch
14:25 < Pahan> david: But you can't! The virtual switch does not support that.
14:25 < david> Pahan: right, I'm just saying implimenting it with a kernel module is backwards when you can do it on the switch without any compatability issues
14:25 < jdike> Pahan: no telnet interface
14:25 < david> Pahan: you said you rewrote it - So add VLAN support in
14:25 < Pahan> david: I am not implementing a kernel driver.
14:25 < david> <Pahan> You use the 8021q module.
14:25 < david> <Pahan> kernel module
14:25 < Pahan> david: That's on the end nodes.
14:25 < david> Pahan: why do you need it?
14:25 < david> you only need that if you're doing tagged VLANs
14:25 < david> which is pointless
14:26 < david> just make the switch process filter frames for specific vlans based on port
14:26 < BB> emulating a real VLAN switch would be nice
14:26 < david> BB: exactly
14:26 < david> BB: using untagged frames and creating a nice membership array for each port for a specific vlan
14:26 < david> bbl
14:27 < BB> david yes, tagged/untagged with multiple members
14:27 < jdike> david: people (Henrik being one) were asking for real vlan tags though
14:27 < jdike> david: they might not be using the switch, maybe bridging out to the physical ethernet
14:28 -!- Pahan- [Blah@ca-manbch-cuda3-c2a-a-52.stmnca.adelphia.net] has joined #uml
14:28 < Pahan-> Urf, internet hickups.
14:29 < jdike> Pahan: ask on uml-devel for what people want wrt vlan
14:29 < jdike> Pahan: this was a while ago and I don't remember exactly what they wanted
14:29 < Pahan-> david: I wanted something that can daemonize easily and do multiple segments in one process.
14:30 < Pahan-> jdike: Does UML data socket address have anything to do with UML identity?
14:30 < Pahan-> jdike: Or is it assigned semi-randomly?
14:30 < jdike> Pahan-: no
14:30 < jdike> Pahan-: the control address sets the switch identity
14:31 < Pahan-> jdike: No, no. The address that UMLs report to the switch.
14:31 < Pahan-> Is that pre-determined?
14:32 < jdike> Pahan-: if it were they wouldn't have to report it :-)
14:32 < Pahan-> Hrm true :P
14:32 < jdike> Pahan-: in effect, that's the identity of one interface of that UML
14:33 < jdike> Pahan-: but it has no real significance
14:34  * Pahan- nods.
14:37 -!- mdz_ [~mdz@24.91.103.187] has quit [Quit: Client exiting]
14:41 < AquaJo> re
14:45 -!- Pahan- [Blah@ca-manbch-cuda3-c2a-a-52.stmnca.adelphia.net] has quit []
14:45 -!- mdz_ [~mdz@h00034795fb2d.ne.client2.attbi.com] has joined #uml
14:46 < jdike> Hi Matt
14:48 < AquaJo> jeff: the error I got when trying to compile:
14:48 < AquaJo> 80-190-52-9:/usr/src/cipe-1.5.4# make ARCH=um
14:48 < AquaJo> make -C lib all
14:48 < AquaJo> make[1]: Entering directory `/usr/src/cipe-1.5.4/lib'
14:48 < AquaJo> make[1]: Nothing to be done for `all'.
14:48 < AquaJo> make[1]: Leaving directory `/usr/src/cipe-1.5.4/lib'
14:48 < AquaJo> make -C 2.4.21-6um-i386-cb all
14:48 < AquaJo> make[1]: Entering directory `/usr/src/cipe-1.5.4/2.4.21-6um-i386-cb'
14:48 < AquaJo> gcc -MD -DKCC_BUILD -I. -include ./config.h  -D__KERNEL__ -I/usr/src/linux-2.4.21/include -Wall -Wstrict-prototypes -Wno-trigraphs -O2 -fno-strict-aliasing -fno-common -fomit-frame-pointer -pipe -mpreferred-stack-boundary=2 -DMODULE -o bf.o -c ../cipe/bf.c
14:48 < AquaJo> In file included from /usr/src/linux-2.4.21/include/linux/ptrace.h:24,
14:48 < AquaJo> ....
14:48 < AquaJo>                  from ../cipe/bf.c:14:
14:48 < AquaJo> In file included from /usr/src/linux-2.4.21/include/asm/ptrace.h:10,
14:48 < AquaJo> ...
14:48 < AquaJo>                  from ../cipe/bf.c:14:
14:49 < AquaJo> In file included from /usr/src/linux-2.4.21/include/asm/ptrace.h:10,
14:49 < AquaJo> ....
14:49 < AquaJo>                  from ../cipe/bf.c:14:
14:49 < AquaJo> make[1]: *** [bf.o] Error 1
14:49 < AquaJo> make[1]: Leaving directory `/usr/src/cipe-1.5.4/2.4.21-6um-i386-cb'
14:49 < AquaJo> make: *** [all] Error 2
14:49 < AquaJo> ups, beginning /es
14:50  * lah lah
14:50 < AquaJo>  /usr/src/linux-2.4.21/include/asm/ptrace.h:9: sysdep/ptrace.h: No such file or directory
14:50 < AquaJo>  /usr/src/linux-2.4.21/include/asm/ptrace-generic.h:25: sysdep/ptrace.h: No such file or directory
14:50 < AquaJo>  /usr/src/linux-2.4.21/include/asm/ptrace-generic.h:26: skas_ptrace.h: No such file or directory
14:50 < AquaJo>  /usr/src/linux-2.4.21/include/asm/ptrace-generic.h:29: field `regs' has incomplete type
14:50 < jdike> AquaJo: is this a fully built UML tree?
14:50 < lah> anyone played with grsecurity + skas yet? I noticed some guy with nick "pflanze" was looking for it, too, I wonder if he played around with it
14:51 < AquaJo> jdike: yes, make linux ARCH=um has worked in the tree
14:51 < AquaJo> but cipe doesn't want to
14:51  * AquaJo doesn't understand why they dont supply it as an kernel patch
14:51 < jdike> AquaJo: i.e. the tree is built right now, full of object files and crap?
14:52 < AquaJo> that would be easyer I think
14:52 -!- snide [~snide@AMontsouris-108-1-31-107.w81-53.abo.wanadoo.fr] has joined #uml
14:52 < AquaJo> jdike: in this case yes
14:52 < AquaJo> mom, I try it clean
14:52 < jdike> AquaJo: if you look at the cc line, you'll see that the UML options totally aren't there
14:53 < jdike> AquaJo: no, it has to be built
14:53 < snide> jdike: btw, u do prefer full UML patch (ie based on vanilla kernel) instead of a UML patch, right ?
14:53 < AquaJo> jdike: ist only the cipe-module, it has it's own build-skript
14:54 < AquaJo> the uml-kernel itself builds fine
14:54 < jdike> snide: rephrase the question?
14:54 < jdike> AquaJo: you need to fiddle the cipe build script so that it uses the UML CFLAGS instead of its own hardcoded x86 CFLAGS
14:54 < snide> jdike: to send u a patch for UML, what format do u prefer ?
14:55 < AquaJo> ok, ill try
14:55 < jdike> snide: you think I want my whole patch back?
14:56 < snide> jdike: i don't know... that's why i do ask ;-)
14:56 < AquaJo> I'll try tomorrow
14:56 < AquaJo> thanks for your help
14:56 < snide> jdike: once u said that u don't like patch based on ur UML patch ;-)
14:57 -!- Maniac [~daManiac@209-5-247-105.mb.skyweb.ca] has left #uml [Client exiting]
14:57 < jdike> snide: when?
14:57 < jdike> snide: for a patch to UML, I want a patch against a UML tree
14:58 < snide> jdike: oh, ok ;-)
14:58 < snide> jdike: good i asked ;-)
15:34 < mdz_> jdike: hey
15:34 < mdz_> I didn't know this channel existed
15:55 -!- jdike [~jdike@129.170.249.137] has quit [Quit: Leaving]
15:57 < caker> hello
15:58 < BB> hi caker 
15:58 < caker> BB: hey
16:13 -!- snide [~snide@AMontsouris-108-1-31-107.w81-53.abo.wanadoo.fr] has quit [Quit: [BX] I wonder what this button marked "EOF" does...]
16:15 -!- smcavoy [~smcavoy@207.61.160.163] has quit [Quit: ACK!]
16:22 -!- snide [~snide@81.53.158.107] has joined #uml
16:44 < AquaJo> good night
16:48 < mistik1> .
16:49 < mistik1> .
16:49 < Dave\\> .
16:49 < Dave\\> .
16:49 < Dave\\> .
16:49 -!- AquaJo [~torbofh@pD9559D3F.dip.t-dialin.net] has quit [Quit: Die gesamte Gegenwart - ein Punkt in der Ewigkeit.]
17:05 -!- snide [~snide@81.53.158.107] has quit [Quit: [BX] Eat, drink and be merry...for tomorrow we die]
17:27 -!- shak [~shak@80.4.84.78] has quit [Quit: Client exiting]
17:32 -!- mistral [mistral@jstevenson.plus.com] has quit [Remote host closed the connection]
17:33 -!- mistral [mistral@jstevenson.plus.com] has joined #uml
21:54 < caker> is there a method to determine if UML has a filesystem file on the host open, besides grep'ing lsof?
21:54 < caker> (from the host)
23:09 -!- ticallion [~ticallion@213.175.160.178] has joined #uml
23:11 -!- ticallion [~ticallion@213.175.160.178] has quit [Client Quit]
--- Log closed Fri Jul 04 00:00:00 2003