1153009445 Q * coocoon Quit: KVIrc 3.2.0 'Realia' 1153010192 Q * Radiance Remote host closed the connection 1153010446 J * mkhl ~mkhl@200-148-40-202.dsl.telesp.net.br 1153010596 J * Radiance 5a0ff45aff@halt.1984world.eu 1153010865 Q * bonbons Quit: Leaving 1153016697 Q * yarihm Quit: Leaving 1153019130 Q * mkhl Quit: 1153022894 Q * micah Ping timeout: 480 seconds 1153023966 M * Skram "some CPU instructions require additional privileges and may not be executed in user space" is that talking about cpu-sys calls? Im confused 1153027646 J * s0undt3c1 ~s0undt3ch@bl7-242-186.dsl.telepac.pt 1153027805 Q * meandtheshell Quit: bye bye ... 1153028104 Q * s0undt3ch Ping timeout: 480 seconds 1153028115 N * s0undt3c1 s0undt3ch 1153031060 Q * gerrit Quit: Client exiting 1153031079 Q * FaUl_ Ping timeout: 480 seconds 1153034645 J * dna ~naucki@dialer-167-86.kielnet.net 1153036979 M * matled Skram: I'd guess things like IO etc 1153037869 J * Viper0482 ~Viper0482@p5497604C.dip.t-dialin.net 1153039867 M * jake- morning 1153040025 J * schimmi ~sts@port-212-202-73-176.dynamic.qsc.de 1153041258 J * bonbons ~bonbons@83.222.39.166 1153041865 J * FaUl immo@shell.chaostreff-dortmund.de 1153043751 J * Smutje_ ~Smutje@xdsl-87-78-57-42.netcologne.de 1153043859 Q * Smutje Ping timeout: 480 seconds 1153043859 N * Smutje_ Smutje 1153047029 J * lilalinux ~plasma@dslb-084-058-195-249.pools.arcor-ip.net 1153048356 J * win[X]amp ~sapujagad@202.152.172.1 1153048511 J * Aiken ~james@tooax6-107.dialup.optusnet.com.au 1153049080 N * Bertl_oO Bertl 1153049085 M * Bertl morning folks! 1153049573 M * Bertl Skram, matled: there are several categories of ring0 ops, like modifying TLB tables (IO/MMU) changing system state (halt and similar) and generating traps/irqs, you might want to visit the intel/amd site and download an instruction manual, it will show you what operations are priviledged ... 1153049610 Q * s0undt3ch Read error: Connection reset by peer 1153049695 J * insomnia1 ~insomniac@slackware.it 1153049695 Q * insomniac Read error: Connection reset by peer 1153049700 M * Bertl okay, back later ... cya :) 1153049704 N * Bertl Bertl_oO 1153049706 Q * win[X]amp Quit: 1153049765 J * win[X]amp ~sapujagad@202.152.172.1 1153052018 J * Aiken_ ~james@tooax8-034.dialup.optusnet.com.au 1153052314 Q * Aiken Ping timeout: 480 seconds 1153052644 Q * shedi Quit: Leaving 1153053066 J * shedi ~siggi@inferno.lhi.is 1153053723 J * meandtheshell ~markus@85-124-207-97.dynamic.xdsl-line.inode.at 1153054647 J * Pazzo ~thomas@host130-250-static.72-81-b.business.telecomitalia.it 1153055164 Q * win[X]amp Quit: 1153055250 Q * h01ger Quit: h01ger 1153055257 J * h01ger ~holger@socket.layer-acht.org 1153056964 Q * Aiken_ Quit: Leaving 1153057618 J * s0undt3ch ~s0undt3ch@bl7-252-90.dsl.telepac.pt 1153057774 Q * matti Quit: 8-X 1153058055 J * yarihm ~yarihm@84-74-17-70.dclient.hispeed.ch 1153060744 Q * lilalinux Remote host closed the connection 1153061730 J * DavidS ~david@chello062178045213.16.11.tuwien.teleweb.at 1153061956 M * DavidS uhm, i have renamed interfaces with udev rules and now i can't use them as vserver interfaces since /proc/net/dev doesn't carry that change... 1153061967 M * DavidS see chbind.c:existsDevice 1153062009 M * DavidS additionally, the tools fail to report _anything_ and just start the vserver without the interfaces 1153062038 M * daniel_hozac so ifconfig and ip show the new names, but /proc/net/dev remains the same? 1153062054 M * DavidS yes 1153062065 M * daniel_hozac that sounds like a bug. 1153062071 M * DavidS kernel? 1153062096 M * doener hm, IIRC there was something about that on lkml lately, lemme check that... 1153062107 A * DavidS sobs 1153062275 M * DavidS ah! on my 1.6.17 devel host /proc/net/dev is already right 1153062281 M * DavidS 2.6.17 1153062293 M * DavidS my 2.6.16 production host is f*cked ... 1153062296 M * DavidS not my week ... 1153062445 M * daniel_hozac hmm, i don't get how chbind is supposed to work in this instance. (as usual...) 1153062515 M * daniel_hozac it seems to treat the argument to --ip as an interface name, but it's an IP address. 1153062572 M * DavidS daniel_hozac: as i said: 2.6.17 has this already fixed ... 1153062588 M * daniel_hozac right, but chbind is still horribly broken. :) 1153062615 M * DavidS daniel_hozac: it works for me .. on .17 at least .. ;) 1153062679 M * daniel_hozac oh i see, it lets you pass a device name to --ip. 1153062688 M * daniel_hozac rather than specifying the address. 1153062710 M * daniel_hozac so really, unless you're doing that, the /proc/net/dev usage in chbind won't affect you at all. 1153062823 M * DavidS daniel_hozac: i have ip="some ip not yet configured" dev="uplink" and nodev; with the intent of configuring the ip on my uplink interface without nuking it ... 1153062856 M * daniel_hozac so you setup the IP addresses on the host. 1153062907 M * daniel_hozac because that's what nodev means. the utils won't do any of the ip addr commands. 1153062918 M * DavidS daniel_hozac: of course! it's been a long weekend, pardon me ... 1153063255 M * DavidS ok, i have configured the ip now on an alias device and have sucessfully bound a vserver to it 1153063266 M * DavidS thanks for the help ... 1153063290 M * daniel_hozac you're welcome. 1153064265 M * DavidS so, back to my ruby segfault :-/ 1153064671 Q * weasel Quit: The system is going DOWN for reboot in 2 minutes! 1153065020 J * weasel ~weasel@asteria.debian.or.at 1153065058 J * coocoon ~coocoon@p54A061BF.dip.t-dialin.net 1153065505 J * lilo2 hiddenserv@tor.noreply.org 1153067527 Q * coocoon Ping timeout: 480 seconds 1153068035 Q * romke Server closed connection 1153068042 J * romke ~romke@procyon.romke.net 1153068097 J * coocoon ~coocoon@p54A07536.dip.t-dialin.net 1153069351 Q * Smutje Quit: leaving 1153069396 Q * yarihm Write error: connection closed 1153069621 J * ekc ekc@netblock-66-245-252-180.dslextreme.com 1153069881 Q * coocoon Quit: KVIrc 3.2.0 'Realia' 1153070783 J * mef ~mef@c-68-39-177-97.hsd1.nj.comcast.net 1153070789 Q * mef Remote host closed the connection 1153075543 Q * click Ping timeout: 480 seconds 1153076513 J * click click@ti511110a080-3396.bb.online.no 1153076561 J * nammie ~nam@S0106001195551ff0.va.shawcable.net 1153076849 Q * Nam Read error: Operation timed out 1153077551 Q * anonc Server closed connection 1153078108 N * click [STAR]Ravenedge 1153079259 N * Ben_zZz Ben_ 1153080285 Q * Viper0482 Remote host closed the connection 1153080289 Q * DavidS Ping timeout: 480 seconds 1153081011 T * * http://linux-vserver.org/ | latest stable 2.01, 1.2.10, 1.2.11-rc1, devel 2.1.0, exp 2.{0.2,1.1}-rc26 | util-vserver-0.30.210 | libvserver-1.0.2 & vserver-utils-1.0.3 | He who asks a question is a fool for a minute; he who doesn't ask is a fool for a lifetime -- share the gained knowledge on the wiki, and we'll forget about the minute ;) 1153081011 T * Bertl - 1153081032 J * coocoon ~coocoon@p54A06D08.dip.t-dialin.net 1153081035 Q * morrigan Remote host closed the connection 1153081064 J * morrigan morrigan@212.16.62.52 1153082514 M * Skram are there any diagrams of how VServer works? I could make my own but am still trying to find software for OSX to do so 1153082650 M * coocoon Skram: http://images.google.de/images?q=vserver&hl=de&btnG=Bilder-Suche 1153082720 M * Skram ok 1153082735 M * Skram i should make my own 1153082746 M * Skram http://www.vsen.dk/gfx/vserver-pres-sslug_html_m18982bb4.gif is such low quality 1153082798 M * coocoon Skram: maybe u will finde there some interesting 1153082840 M * Skram a picture of Virtuozzo would work too 1153083537 Q * bonbons Quit: Leaving 1153083630 N * Bertl_oO Bertl 1153083634 M * Bertl evening folks! 1153083682 Q * dna Quit: Verlassend 1153083687 M * daniel_hozac evening! 1153083764 M * coocoon Skram: maybe this will inspire u it is in french but there are also a lot of images http://www.unilim.fr/dokuwiki/lib/exe/fetch.php?id=systeme%3Avserver%3Aaccueil&cache=cache&media=systeme:vserver:presentation-vserver.ppt 1153083828 M * Skram blah 1153083832 M * Skram i need a graphics designer 1153083833 M * Skram heh 1153084130 M * doener Bertl: did you get my mail? (shiny) 1153084220 M * Bertl yes, but I didn't get around to check it, although I know what you do 1153084280 M * doener ok, just wondered if you got it. it's buggy anyway, one of the cases where we have ebp is broken 1153084292 M * Bertl problem here AFAICR was that older compilers did not get the assignment right and the result was a register error, plus moves like %eax,%eax 1153084344 M * Skram whats the official plural of VPS? 1153084346 N * Ben_ Ben_zZz 1153084347 M * Skram VPSes? 1153084359 M * Skram I guess it should be VPSs since Server => ServerS 1153084359 M * Bertl Guests? 1153084385 M * doener hm, that happened with the constraints all being "g", gcc then tries to "preload" some stuff into registers and that might conflict with the inline assembly, in this case probably because there was no clobber for eax 1153084454 M * Bertl doener: IMHO the problem is not really solveable with 7 registers and 7 arguments, where at least 3 registers need special handling and the esp is broken :) 1153084466 M * Skram Linux-VServer provides virtualization for GNU/Linux systems. 1153084467 M * Skram Accomplished by kernel level isolation 1153084467 M * Skram Allows the execution of multiple virtual units (Virtual Private Servers / Guests) at once. 1153084470 M * Skram Said Guests are sufficiently isolated to... 1153084472 M * Skram Guarantee the requested security 1153084475 M * Skram Utilize available resources efficiently 1153084477 M * Skram All Guests in a Linux-VServer environment run on the same kernel. 1153084480 M * Skram does that make sense 1153084547 M * Bertl doener: I would tend to hard code i386 ignoring the gcc register assignment and such, basically putting all 'sensible' arguments on the stack and fetching them from there 1153084587 M * Bertl doener: that is damn ugly but the gcc handling of inline asm is severly broken for x86 1153084651 M * Bertl Skram: are you rewriting the introduction pages or so? 1153084668 M * Skram Bertl: just putting them into bullet format 1153084682 M * Skram it says it well, and I am not taking credit for writing it.. it is for a presentation 1153084685 M * doener Bertl: hm, I don't see the "broken" part 1153084716 M * Bertl doener: the following behaviours are IMHO 'broken' 1153084737 M * Skram the second and last point are kind of redundant 1153084739 M * Skram ill combine them 1153084742 M * Bertl - a register which is marked as clobbered is not stored by gcc (ebx for fix) 1153084750 M * Bertl s/fix/pic/ 1153084792 M * daniel_hozac Bertl, doener: any reason we can't use mov %ebx, -4(%esp) rather than the pushl? 1153084865 M * Bertl - a modification of the stack pointer (esp) is not recognized and register arguments rom the stack are not corrected 1153084876 M * Bertl s/rom/from/ 1153084886 M * Skram brb 1153084906 M * Bertl - finally, register assignments are not solved/honore with not at least one register left empty 1153084906 Q * [STAR]Ravenedge Read error: Connection reset by peer 1153084999 M * Bertl daniel_hozac: we can save %ebx in many different variations 1153085019 M * Bertl the rpoblem here is, that gcc decides about the left side (arguments) 1153085036 M * Bertl it can be a memory location, a stack offset or a register 1153085083 M * daniel_hozac ah, right. 1153085085 M * Bertl and while that works pretty fine for tiny inlines, it fails horribly for more complex stuff (because of the issues mentioned above) 1153085128 M * Bertl my original idea was to make that syscall inlineable, and it works pretty fine except for x86 :) 1153085154 M * Bertl maybe we should just accept that this arch is borked and 'require' a stack passing convention 1153085166 M * Bertl (which would immediately solve the issues) 1153085466 M * doener Bertl: hm, what about a local var to save ebx/ebp? 1153085494 Q * lilo2 Remote host closed the connection 1153085546 J * lilo2 hiddenserv@tor.noreply.org 1153085548 M * doener if we force a memory constraint for them for the inline assembly, that should get us some space on the stack that we can savely address 1153085677 M * Bertl yes, that's what I thought about, passing another 'm' constraint arg 1153085701 M * Bertl and storing the ebx/ebp there (or in two of them) 1153085732 Q * coocoon Quit: KVIrc 3.2.0 'Realia' 1153085738 M * Bertl well, ebp could be pushed in any case I guess 1153085776 M * Bertl but needs some testing, especially with older compilers 1153086397 Q * mire Quit: Leaving 1153087293 M * doener Bertl: hm, I'm not convinced about the brokeness of gcc not recognizing the push... Sure, it could try to parse the inline assembly and be smart, but as soon as there's some instruction that it doesn't know about you're screwed anyway 1153087453 M * Bertl well, as it currently is, you are screwed when you use _any_ op which modifies the stack, no? :) 1153087491 M * Bertl (without -fomit-frame-pointer of course :) 1153087494 M * doener with -fomit-frame-pointer 1153087512 M * Bertl yep, I meant with :) 1153087516 M * doener without it gcc uses ebp and all is fine 1153087529 M * Bertl yes, as long as _you_ do not use ebp :) 1153087553 M * Bertl at which point, gcc starts to misbehave again ... 1153087626 M * Bertl personally, I would we fine if the interface was dumb, but would allow to say, put this argument in that register and don't be too smart about it ... 1153087651 M * Bertl but that's something the inline asm does not allow on x86 1153087718 M * Bertl but hey, even if they change it tomorrow, that does not help us a bit :) 1153089013 Q * DrEeevil Quit: Leaving 1153091764 Q * michal` Ping timeout: 480 seconds 1153092086 J * michal` ~michal@www.rsbac.org 1153092772 J * Aiken ~james@tooax6-127.dialup.optusnet.com.au 1153093514 Q * michal` Ping timeout: 480 seconds 1153093532 J * matti matti@linux.gentoo.pl 1153093701 J * coocoon ~coocoon@p54A05263.dip.t-dialin.net 1153093855 J * michal` ~michal@www.rsbac.org