1166746603 M * Wonka re 1166746614 M * Wonka now running 2.2.0-rc6 1166746625 M * Wonka and "vserver $name enter" works again :) 1166746777 M * daniel_hozac chcontext --hostname won't though :) (i.e. you really should update your utils ;)) 1166746805 Q * mire Ping timeout: 480 seconds 1166746821 M * Wonka debian has 0.30.212 in unstable, am just preparing to install that 1166746830 M * Wonka "aptitude -u" is already running 1166746986 M * Guy- are you guys familiar with runit or daemontools? I'd like to run some services in their own vserver without going through the fuss of starting and entire linux boot process there 1166747051 M * Guy- what I'd need is to do "vfoo vservername service", where vfoo is a vserver utility, service is a program to run relative to the root of vservername, and vfoo shouldn't exit until service exits 1166747073 M * Guy- also, vfoo would inherit an stdout descriptor that it should pass on to service 1166747117 M * Guy- is this possible? 1166747126 Q * dna Quit: Verlassend 1166747131 M * daniel_hozac vserver exec? 1166747154 M * Guy- how would the parent process see this? 1166747159 M * daniel_hozac see what? 1166747168 M * Guy- wait, I'm trying to explain :) 1166747198 M * Guy- so, assuming I'm a runsv process (from runit, doesn't really matter); I fork(), and the child execs 'vserver guest exec' 1166747208 Q * FireEgl Quit: ... 1166747220 M * Guy- is the resulting process, that runs inside the vserver, still my child? 1166747253 M * Bertl yes, unless the parent exist 1166747268 M * Guy- ok, can the parent send signals to the child? 1166747271 M * daniel_hozac it's a distant grand child. 1166747273 M * Bertl in which case the child is reparented to the guest init 1166747281 M * daniel_hozac no. 1166747287 M * Bertl no, signalling across contexts is blocked 1166747302 M * Guy- will the parent get a sigchld when the child exits? 1166747308 M * Bertl daniel_hozac: btw, already managed to unlock the blocks and erase a page :) 1166747359 M * daniel_hozac haha, awesome! is there a need to reflash it, or is it just for kicks? :) 1166747374 M * daniel_hozac Guy-: yes. 1166747395 M * daniel_hozac at least, i think so... 1166747404 M * Guy- so, in all respects, the child running in a different contexts acts like a 'normal' child, except I can't send signals to it? 1166747425 M * Bertl mobo doesn't boot after an interrupted bios flash, no recovery procedure for this bios, but replacing the mobo would have been the cheaper solution 1166747427 M * daniel_hozac that's the idea. 1166747446 M * Bertl so, I'd say it is for the fun of doing it :) 1166747453 M * daniel_hozac Bertl: ah, interrupted by what? 1166747464 M * Bertl power surge 1166747466 J * mire ~mire@185-166-222-85.adsl.verat.net 1166747481 M * daniel_hozac that sucks. 1166747498 M * Guy- does process state inheritance work across virtual contexts? 1166747510 M * Bertl like? 1166747513 M * daniel_hozac though flashing it the "hard way" is definitely cooler than using those boring utilities :) 1166747516 M * Guy- obviously cwd can't be inherited 1166747527 M * Guy- but what about resource limits, umask, filedescriptors, ...? 1166747542 M * Bertl daniel_hozac: indeed, took me some time and patience to solder the tiny wires to the chip :) 1166747544 M * Guy- oh, and in some cases, cwd could be inherited too 1166747562 M * Bertl yes, those things are all inherited 1166747564 M * daniel_hozac cwd is never inherited. 1166747571 M * daniel_hozac (for vserver ... exec) 1166747585 M * Bertl correct, was talking about a context switch 1166747594 M * daniel_hozac yeah. 1166747608 M * Guy- so the new process starts with cwd = / (the root of the virtual context), I assume 1166747616 M * daniel_hozac yes. 1166747634 M * Guy- I guess only root can switch contexts like this? 1166747638 M * daniel_hozac yes. 1166747647 M * daniel_hozac or well, anyone with CAP_CONTEXT. 1166747656 M * Bertl yeah, was going to add that 1166747672 M * Guy- can I have a pipe between the parent (running in context 0) and the child (running in a different context)? 1166747684 M * Bertl yes 1166747687 M * Guy- cool 1166747700 M * Bertl at least I cannot think of anything blocking that, atm 1166747712 M * daniel_hozac well, that definitely works. 1166747727 M * daniel_hozac otherwise sysv initstyle wouldn't be able to output anything either. 1166747753 M * Bertl hmm, that uses pipes? 1166747763 M * daniel_hozac well, i assume pipe means fd. 1166747788 M * Bertl was thinking of named pipes ... 1166747815 M * Guy- no, I meant inheriting a filedescriptor that's not associated with a real file 1166747827 M * Guy- or any filesystem object 1166747844 M * Bertl works, as long as the file is not tagged with a different context 1166747844 M * Guy- what one process puts in at one end the other receives at the other 1166747868 M * Guy- like grep | vserver ... exec sed 1166747877 M * Guy- (stupid example, just to show what I mean) 1166747905 M * Bertl should work, daniel_hozac? 1166747906 M * daniel_hozac should work fine. 1166747910 M * Guy- and actually it'd be the other way round, vserver ... exec grep | sed 1166747925 M * Guy- OK, that takes care of logging then 1166747984 M * Guy- and I assume environment variables are also inherited across a vserver exec 1166748012 M * Guy- what about filedescriptors that refer to sockets that use IP addresses isolated from the virtual context? 1166748035 M * daniel_hozac sockets are tagged with the nid, so that won't work. 1166748042 M * Bertl entering a process context should not affect that with recent versions 1166748059 M * Bertl entering a network context might give interesting results 1166748093 M * Guy- assuming I run inetd in context 0 and telnetd is started in context 42 (by inetd) 1166748099 M * Guy- inetd listens on 0.0.0.0:23 1166748113 M * Guy- context 42 can only use 1.2.3.4 1166748118 M * Guy- a connection comes in on 2.3.4.5 1166748125 M * Guy- what happens? 1166748235 M * daniel_hozac Bertl: 2.2.0-rc6/net/ipv4/inet_diag.c:inet_diag_dump, should the time wait sockets really use vx_check? 1166748257 M * Bertl probably not 1166748263 M * Bertl checking 1166748328 Q * yarihm Quit: Leaving 1166748328 M * Bertl yep, is a bug, good catch, tx 1166748333 M * daniel_hozac Guy-: i think it might work... you'd have to try. 1166748347 M * Guy- daniel_hozac: isn't this a bug if it works? 1166748391 M * daniel_hozac is it? 1166748424 M * Guy- I'm not sure 1166748437 M * Guy- after all, you have a process in context 42 sending packets from 2.3.4.5 1166748447 M * Bertl Guy-: the checks happen at bind time 1166748450 M * Guy- even though context 42 shouldn't be able to do that 1166748469 M * daniel_hozac but if we check on creation, shouldn't that keep us safe? 1166748487 M * daniel_hozac i mean, how often do you enter a guest from the host with lots of fds that you don't want the guest to use? 1166748498 M * daniel_hozac and if that's the case, close the fds before execing. 1166748504 M * Bertl i.e. if the inetd is already bound (host) then it will hand over that socket to the guest, which will be allowed to do certain things with it 1166748543 M * Bertl such setups might not be secure per default, they need to have some policy to become secure 1166748565 M * Wonka »Why does gbd do know about "break" but not about "fix"?« 1166748593 M * Bertl because the gdb folks are lazy guys :) 1166748607 M * Guy- but hey, it's supposed to be a DEbugger :) 1166748639 M * daniel_hozac if you write a bug fixing gdb, i'm sure you will become the king of something :) 1166748685 M * Bertl guess we'd instantly add you to the Hall'o'Fame *G* 1166748694 M * Guy- daniel_hozac: I'm actually pleased that inherited sockets work, and even more pleased because you apparently don't think this is a bug that needs fixing :) 1166751375 Q * Piet_ Ping timeout: 480 seconds 1166754109 M * Medivh hi 1166754135 M * Medivh i'm wondering, if i use disk limits for a vserver... what is the best way to count current usage on start of the vserver so it will have correct values? 1166754474 M * Bertl hmm, tools should already do that for you 1166754495 M * Medivh hmm, doesn't look like they do it here tho :( 1166754532 M * Bertl ask daniel_hozac, IIRC some 'minor' magic was requires, like touching a file in the config tree 1166754554 M * Medivh i got /etc/vservers/v10/dlimits/0/{directory|inodes_total|reserved|space_total}, maybe i need smth else? 1166754557 M * Medivh ah... ok 1166754600 M * Medivh lemme check the GFP, maybe it's on there ;) 1166754612 M * Bertl yep, it should be 1166754841 M * Medivh hm, either it's not, or i'm blind 1166754921 M * Bertl okay, what did you put in dlimits? 1166754942 M * Medivh file 'directory', content '/vservers/v10' 1166754956 M * Medivh file 'inodes_total', content '1000000' 1166754963 M * Medivh file 'reserved', content '5' 1166754976 M * Medivh file 'space_total', content '10485760' 1166755080 M * Bertl and when you start the guest, it doesn't fill in the current values? 1166755105 M * Medivh at least when i do 'df' in the guest, i get used = 0 1166755119 M * Bertl hmm, what tool version is that? 1166755125 M * daniel_hozac does /vservers/vl0 have files tagged with the xid? 1166755131 M * daniel_hozac is it mounted with tagxid? 1166755138 M * Bertl ah, good point 1166755146 M * Medivh guys... shoot me ;) 1166755161 M * Medivh tagxid is a good point indeed heh 1166755178 M * Medivh can i -o remount, or does it really need umount first? 1166755189 M * daniel_hozac umount; mount. 1166755200 M * daniel_hozac -o remount,tagxid is explicitly forbidden. 1166755243 M * Bertl the probelm with a remount is, that inodes already in the cache would get a wrong tagging 1166755272 M * Medivh i see... ok, gotta wait a bit then, someone on that machine who would shoot me for restarting her vserver now ;) 1166755297 M * Medivh thanks :) 1166761162 Q * kaner Ping timeout: 480 seconds 1166762044 J * kaner kaner@strace.org 1166763891 M * Bertl okay, I'm off to bed now ... have a good one everyone ... cya tomorrow! 1166763897 N * Bertl Bertl_zZ 1166767381 Q * DreamerC Quit: leaving 1166767398 J * DreamerC ~dreamerc@125-225-96-17.dynamic.hinet.net 1166769052 Q * mcp Read error: Connection reset by peer 1166769067 J * mcp ~hightower@wolk-project.de 1166769564 J * Aiken_ ~james@tooax8-038.dialup.optusnet.com.au 1166769867 Q * anonc Remote host closed the connection 1166769892 Q * Aiken Ping timeout: 480 seconds 1166771387 Q * eGnarF Quit: leaving 1166772963 Q * m`m`h Ping timeout: 480 seconds 1166774553 J * m`m`h ~simba@deb30.mgts.by 1166775727 Q * m`m`h Ping timeout: 480 seconds 1166777125 J * lylix ~eric@dynamic-acs-24-154-34-43.zoominternet.net 1166777472 J * m`m`h ~simba@deb30.mgts.by 1166778367 Q * m`m`h Ping timeout: 480 seconds 1166779446 J * m`m`h ~simba@deb30.mgts.by 1166779524 J * anonc ~anonc@staffnet.internode.com.au 1166779915 J * pisco ~pisco@p5087B77D.dip0.t-ipconnect.de 1166780588 Q * pisco Quit: leaving 1166780619 J * pisco ~pisco@tor-irc.dnsbl.oftc.net 1166780671 Q * pisco 1166780677 Q * m`m`h Ping timeout: 480 seconds 1166780912 J * MrX ~chaos@219.95.234.244 1166781200 J * dna ~naucki@p54BCDE23.dip.t-dialin.net 1166781906 J * meandtheshell ~markus@85-124-174-163.dynamic.xdsl-line.inode.at 1166782875 Q * rob-84x^ Quit: That's it for today 1166782878 J * m`m`h ~simba@deb30.mgts.by 1166784949 J * MooN ~MooN@kazha.id.lv 1166786164 J * lilalinux ~plasma@80.69.41.2 1166786542 Q * Aiken_ Ping timeout: 480 seconds 1166789037 Q * ensc Ping timeout: 480 seconds 1166789651 A * MooN -=> 666 - D.E.V.I.L. (What the Hell Mix) 1166790140 Q * lylix Read error: Connection reset by peer 1166790341 Q * meandtheshell Quit: Leaving. 1166793462 Q * m`m`h Ping timeout: 480 seconds 1166793811 Q * sid3windr Ping timeout: 480 seconds 1166793937 J * orzelf ~orzel@217.167.255.5 1166794012 M * orzelf hello! Can you confirm that on debian stable, there's no other way than to compile your kernel yourself to get vserver ? 1166794036 M * daniel_hozac on sarge, you mean? 1166794085 M * daniel_hozac backports.org should have some precompiled kernels, i think. 1166794163 J * sid3windr luser@bastard-operator.from-hell.be 1166794250 M * orzelf i'm using backports 1166794253 M * orzelf but can't find packages 1166794259 M * orzelf i'm going to check again 1166794335 M * orzelf or better : there are some that refuse to install 1166794338 M * orzelf i remember now ;) 1166794373 M * daniel_hozac what do you mean? 1166794460 A * MooN -=> Смысловые галлюцинации - Доживем до солнца 1166794461 M * orzelf i mean that : http://pastebin.com/843168 1166794553 M * daniel_hozac initramfs-tools is at 0.85c in backports. 1166794596 M * orzelf initramfs-tools: Depends: udev (>= 0.086-1) but 0.056-3 is to be installed 1166794611 M * orzelf if i install udev, then only 0.056-3 is installed 1166794640 M * daniel_hozac udev is at 0.103 in backports. 1166794675 M * orzelf ah.. so i have something wrong with my setup. i just added the backports line in sources.list. 1166794698 M * phedny did you also do "apt-get update" ? 1166794725 M * orzelf yes, sure 1166794762 M * orzelf i had this problem one month ago. I retry today, and first thing i did was this update 1166795068 M * orzelf the main 'howto' link on mainpage of backports is broken. I wonder how alive this is.. Has been broken for long. 1166795095 M * orzelf at least i seem to use the right line in sources.list 1166795722 J * m`m`h ~simba@deb30.mgts.by 1166796508 Q * MooN Quit: n-o-n-e 1166797477 Q * orzelf Ping timeout: 480 seconds 1166797854 J * Piet hiddenserv@tor.noreply.org 1166798189 J * orzelf ~orzel@217.167.255.5 1166798351 Q * orzelf Remote host closed the connection 1166798436 J * marmotas ~whatareul@mail.medlife.ro 1166798455 M * marmotas hi.can anyone help me with a printer???plsssss 1166798567 M * daniel_hozac inside a guest, or what? how does it relate to Linux-VServer? 1166798592 M * marmotas it doesn't 1166798594 M * marmotas srry 1166798596 P * marmotas 1166798765 J * click click@ti511110a080-0625.bb.online.no 1166799305 N * Bertl_zZ Bertl 1166799316 M * Bertl morning folks! 1166799464 M * daniel_hozac morning Bertl! 1166799717 J * gcj ~chris@cpc1-cmbg7-0-0-cust497.cmbg.cable.ntl.com 1166799727 M * Bertl welcome gcj! 1166799740 M * gcj hi bertl! hi all, i'm having some problems with a gentoo vserver not starting. i hope someone can help me 1166799760 M * Bertl I'm sure we will be able to help you ... 1166799780 M * gcj i think it fails to mount /proc, and then drops to sulogin 1166799802 M * gcj i don't see any output when i run vserver x start from an ssh session, so i'm not sure exactly where it fails 1166799817 M * Bertl that could be, why does it try to mount proc? 1166799828 M * gcj it's in /sbin/rc 1166799841 M * gcj i tried to comment out that part, but it still behaves the same 1166799850 M * gcj perhaps something fails afterwards 1166799864 M * gcj i'm using plain initstyle, tried gentoo but it wasn't recognised 1166799880 M * Bertl hmm, what tools do you use? what kernel patch? 1166799892 M * Bertl (and probably what is the host distro :) 1166799923 M * gcj host is gentoo as well 1166799939 M * gcj util-vserver-0.30.210-r13 (gentoo package) on the host 1166799960 M * gcj patch-2.6.15.6-vs2.0.2-rc12.diff 1166799972 M * daniel_hozac that's really old. 1166799974 M * Bertl okay, I'd suggest to update to newer versions there 1166799979 M * gcj i tried patch-2.6.17.13-vs2.0.2.1.diff.bz2 on a 2.6.17.13 kernel, still happens 1166799992 M * Bertl gentoo fixed a bunch of things regarding modified guest layout 1166800003 M * Bertl and the tools now support the gentoo style again, IIRC 1166800012 M * gcj i saw benedikt's page, but could quite follow his instructions 1166800020 M * gcj *couldn't 1166800040 M * daniel_hozac i think the gentoo initstyle requires baselayout 1.13. 1166800068 M * gcj this vserver is quite old and wasn't build using vserver build, i don't know how to change the baselayout of an existing vserver 1166800088 M * daniel_hozac you are using baselayout-vserver, right? 1166800113 M * phreak`` daniel_hozac: it does .. 1166800113 M * gcj daniel_hozac, probably not, this used to be a UML virtual machine and i didn't change much when porting to vserver 1166800145 M * gcj can i just do emerge -av baselayout? 1166800145 M * daniel_hozac gcj: i'm amazed it ever worked at all then. 1166800184 M * gcj perhaps it didn't :-) i don't remember when i last restarted it, or if i had to bring up the services by hand that time 1166800208 M * phreak`` daniel_hozac: plain is what should be used with baselayout-vserver-1.12, iirc we even had a patch to deprecate gentoo (in util-vserver-0.30.211) 1166800230 M * daniel_hozac yes, gentoo was removed completely in 0.30.211. 1166800240 M * daniel_hozac (and probably earlier through patches) 1166800273 M * gcj what should i upgrade first, the host util-vserver? 1166800289 M * phreak`` gcj: yep, then the inside of the vServer 1166800304 M * gcj phreak``, baselayout inside the vserver? 1166800338 M * phreak`` gcj: not only, but make sure you either use baselayout-1.13_alpha* *or* baselayout-vserver-1.12* 1166800392 M * gcj do i need to add an overlay to get these? 1166800425 M * phreak`` gcj: nope, its in the main tree 1166800472 M * daniel_hozac the first is ~x86 though, right? 1166800480 M * phreak`` daniel_hozac: yeah, and masked iirc 1166800490 M * daniel_hozac okay. 1166800698 M * gcj sorry, this may take a while, i'm having to emerge sync first 1166800920 Q * michal` Ping timeout: 480 seconds 1166801039 M * Bertl np, take your time ... 1166801077 M * Bertl okay, off shopping ... back later ... 1166801082 N * Bertl Bertl_oO 1166801161 J * rob-84x^ ~rob@submarine.ath.cx 1166801215 J * michal` ~michal@www.rsbac.org 1166801299 M * gcj i'm using an old standard baselayout, is it safe to emerge baselayout-vserver? 1166801442 M * phreak`` gcj: needs some tricks .. 1166801502 M * gcj phreak``, can you help me? i'd hate to break this vserver, it would be a disaster (it's our mail server) 1166801548 M * phreak`` gcj: well I'd build a binary package for the new baselayout-vserver in another vserver 1166801597 M * gcj would it be easier to emerge a new alpha baselayout? (emerge -s doesn't show the alpha version you mentioned, so i'm not sure where to get it from) 1166801664 M * phreak`` gcj: its package.masked (or in ~arch), so thats why you don't see it. and no the new baselayout-1.13 isn't that easy, since you need to change something on the vserver config itself 1166801703 M * gcj ok, i've never built a gentoo binary package before, where should i start reading? 1166801738 M * phreak`` gcj: emerge -B helps you (or read man emerge :) 1166801748 M * phreak`` emerge -B baselayout that is 1166801775 M * phreak`` bah, baselayout-vserver even 1166801815 M * gcj thanks, will try that :-) 1166803423 M * gcj phreak``, i built a binary package for baselayout-vserver on another box and i'm trying to emerge it, but I get the error "You need to adjust PKGDIR to emerge this package". any ideas 1166803424 M * gcj ? 1166803539 M * gcj ok i fixed it, moved the file to /usr/portage/packages/x86/All 1166803555 M * phreak`` heh, thats what i was going to tell you :) 1166804030 M * gcj what's the best initstyle to use with baselayout-vserver? plain, gentoo or default? 1166804086 M * daniel_hozac plain is the only one that works. 1166804092 M * daniel_hozac so that might be a good idea. 1166804097 M * gcj ok thanks :-) 1166804334 M * gcj i'm being warned to run "setattr --barrier /vservers". do i need to do that just once, or on every boot? 1166804352 J * s[e]bastian ~info@pD957E770.dip.t-dialin.net 1166804674 M * cehteh just once 1166804707 M * cehteh (afaik with recent kernels it isnt even nesseary anymore because vserver uses namespaces) 1166804766 M * gcj do you know how recent? i have 2.6.15 1166805305 M * gcj by the way, i like the new website, much cleaner and easier to find things 1166805322 Q * s[e]bastian 1166805433 M * daniel_hozac you need the barrier even though it's using namespaces. 1166805773 J * s[e]bastian ~info@pD957E770.dip.t-dialin.net 1166805846 M * cehteh ah ok 1166806428 Q * s[e]bastian 1166807035 J * s[e]bastian ~info@pD957E770.dip.t-dialin.net 1166807958 Q * tso Ping timeout: 480 seconds 1166808010 Q * s[e]bastian Ping timeout: 480 seconds 1166808048 J * FireEgl ~FireEgl@adsl-147-90-220.bhm.bellsouth.net 1166808240 J * s[e]bastian ~info@pD957D595.dip.t-dialin.net 1166808426 Q * renihs Quit: Leaving 1166808830 J * bonbons ~bonbons@83.222.37.103 1166808968 M * gcj phreak``, i updated host and guest and now the gentoo vserver starts nicely, but still i don't get any console output when starting it. Is that expected? 1166809282 Q * s[e]bastian 1166810301 M * daniel_hozac gcj: yes. with the plain initstyle, you don't get any output. 1166810400 M * gcj ok thanks, is that likely to be fixed? 1166810446 M * daniel_hozac i don't know of anyone working on it. 1166810464 M * gcj but gentoo initstyle will work again one day? 1166810557 M * daniel_hozac should already work if you use baselayout 1.13. 1166810598 M * gcj maybe when it comes out of alpha and is unmasked :-) 1166810673 M * gcj thank you all for your help 1166810676 M * gcj it's working well now 1166810686 M * gcj happy christmas :-) 1166810695 M * daniel_hozac you too! 1166810704 Q * gcj Quit: Leaving 1166811624 M * bXi w00t 1166811640 M * bXi just did a new install of the gentoo vserver kernel and now it works correctly with genkernel 1166812298 M * hardwire hehe.. should I run badblocks on an AoE partition :) 1166812564 N * Bertl_oO Bertl 1166812567 M * Bertl back now 1166812587 M * daniel_hozac last minute christmas shopping? 1166812683 M * Hollow ha! i still have to.. 1166812701 M * Bertl kind of, well, I needed a bidirectional bus transceiver, don't know if that counts :) 1166812707 M * Hollow heh 1166812746 M * daniel_hozac hehe 1166812834 M * Bertl figured yesterday, after I managed to read and even erase certain parts of the flash chip, that the parallel port cannot supply enough power when all lines are high 1166812857 M * Bertl gives funny and unexpected results ... 1166813315 M * bXi Bertl: remember that issue i had with getuid16 ? 1166813325 Q * ruskie Quit: Disconnecting from stoned server. 1166813329 M * bXi its solved now 1166813348 M * Bertl how_ 1166813353 M * Bertl ? 1166813357 M * bXi dunno 1166813364 M * bXi i just did a new system at work 1166813368 M * bXi same steps 1166813372 M * daniel_hozac i guess that kernel had it enabled. 1166813383 M * Bertl different kernel config, I guess 1166813398 M * bXi that wouldnt make sense 1166813405 M * bXi i had the same kernel and such 1166813415 M * bXi and i used genkernel in both situations 1166813429 M * Bertl hmm, maybe you could double check that in the config 1166813460 M * bXi the config came from the livecd i installed from 1166813463 M * bXi which was the same cd 1166813478 M * bXi my guess is that they did a bugfix in genkernel or so 1166813507 M * bXi but i'll compare them later 1166813515 M * Bertl okay, TIA! 1166813580 J * ruskie ~ruskie@ruskie.user.oftc.net 1166813725 J * prae ~benjamin@foxhound.sherpadown.net 1166814098 Q * prae Quit: Pwet 1166815556 Q * kevinp Quit: Leaving 1166815560 M * bXi oh man 1166815573 M * bXi now i need to fix apache in a vserver 1166815578 M * bXi but it needs net.lo 1166815612 M * nebuchadnezzar hi 1166815619 M * Bertl bXi: hmm? 1166815625 M * Bertl hey nebuchadnezzar! 1166815738 J * Kerberos ~satan@85.138.138.8 1166815844 M * bXi Bertl: i dont have a localhost interface in my vserver 1166815854 M * bXi and apache somehow requires one 1166815862 M * Bertl that would be new 1166815877 M * Bertl at least all my apaches are fine without 1166815889 M * bXi http://paste.linux-vserver.org/816 1166815921 M * Bertl gentoo? 1166815929 M * bXi yeah 1166815949 M * Bertl I guess you have an old baselayout then 1166815965 M * Bertl networking should be off inside a guest, IIRC 1166816098 M * Kerberos dumb question: by adding capabilities.. like net_admin... doesnt affect the host? its all about the guest? 1166816143 M * Bertl if you give net_admin, then the guest will be able to mess with _all_ network stuff 1166816154 M * Bertl i.e. it will be able to do the same as the host 1166816195 M * Kerberos but messing in the guest doenst alter anything in the host right? 1166816220 M * Bertl sure it does 1166816230 M * Bertl if you remove eth0, it will be gone 1166816258 M * Bertl Linux-VServer doesn't do full system virtualization, it is all based on _isolation_ 1166816289 M * Bertl by giving CAP_NET_ADMIN, you basically allow the guest to mess with the entire network stack 1166816465 M * matti Bertl: :)) 1166817114 J * jayCampbell ~jay@spcxn.got.net 1166817176 Q * jayCampbell Quit: using sirc version 2.211+ssfe 1166817238 J * Aiken ~james@tooax8-218.dialup.optusnet.com.au 1166818150 M * Bertl hey matti! 1166818181 J * tso ~tso@254-049.adsl.pool.ew.hu 1166818184 Q * tso 1166818271 Q * matti Quit: 8-X 1166818408 Q * virtuoso Read error: Connection reset by peer 1166818563 J * tso ~tso@254-049.adsl.pool.ew.hu 1166818762 J * s[e]bastian ~info@pD957D595.dip.t-dialin.net 1166818916 M * bXi how does one set up a debian guest 1166818943 M * daniel_hozac vserver debian build -m debootstrap --context ... ... -- -d etch 1166819067 M * hardwire sexy 1166819078 M * hardwire bXi: check the debian grimoire page (via google) 1166820285 M * bXi hmmm i seem to recall a page with lots of downloadable hosts 1166820388 M * bXi s/hosts/guests/ 1166820416 Q * FireEgl Quit: ... 1166820854 M * daniel_hozac tarballs? yes. 1166820877 M * daniel_hozac but Gentoo is pretty much the only distribution (that matters :)) that can't be built on your own. 1166820901 M * daniel_hozac Debian guests are especially easy as the utils will even download debootstrap for you, so you need not install it before-hand. 1166821073 M * bXi i wanna check that list 1166821108 Q * michal` Ping timeout: 480 seconds 1166821434 J * michal` ~michal@www.rsbac.org 1166821822 Q * Piet Quit: Piet 1166823546 J * s0undt3ch_ ~s0undt3ch@bl4-60-11.dsl.telepac.pt 1166823605 J * _dmax ~semaj@81.193.60.11 1166823951 Q * dmax Ping timeout: 480 seconds 1166823951 N * _dmax dmax 1166823969 Q * s[e]bastian 1166823988 Q * s0undt3ch Ping timeout: 480 seconds 1166823988 N * s0undt3ch_ s0undt3ch 1166824113 Q * bonbons Quit: Leaving 1166824416 J * yarihm ~yarihm@84-75-123-221.dclient.hispeed.ch 1166824818 J * C20XE ~moonlight@62.68.61.4 1166824969 N * C20XE Anark1 1166825053 P * Anark1 1166825376 M * daniel_hozac Bertl: hmm. I think the syscall for sparc is broken when using glibc (like vhashify does). 1166825435 M * Bertl ah, details? 1166825470 M * daniel_hozac http://www.asgardr.info/~nebu/vserver/vc_get_iattr_debian_sid.asm 1166825487 M * daniel_hozac (i was writing my reply for the Debian bug, but i can't seem to find anything else wrong with it now...) 1166825500 M * daniel_hozac 6bf0 1166825513 M * Bertl hmm, does that code use my syscall wrappers? 1166825550 M * daniel_hozac it should... yes. 1166825555 M * daniel_hozac according to http://www.asgardr.info/~nebu/vserver/vserver_info.debian 1166825584 M * Bertl hmm, but with dietlibc it works? 1166825594 M * daniel_hozac yes, errno is just an int there, no? 1166825601 M * daniel_hozac i wonder how this is handled on x86, actually. 1166825609 M * Bertl okay, please try to explain what you figured is wrong 1166825625 M * daniel_hozac st %o0, [ %o0 ] 1166825647 M * daniel_hozac (that's trying to store the return value from the syscall in errno) 1166825654 M * daniel_hozac 6bf8 1166825824 M * Bertl okay, that is wrong 1166825845 M * Bertl it should not allocate the o0 for the pointer 1166825874 M * Bertl could you try to disable my wrappers for vhashify? nebuchadnezzar, can you test that? 1166825922 M * daniel_hozac %o0 is the return value from the call. 1166825931 M * Bertl yes 1166825932 M * daniel_hozac from what i've gathered. 1166825945 M * Bertl %i is input, %o is output 1166825955 M * daniel_hozac right. 1166825959 M * Bertl (funtion perspective) 1166825999 M * daniel_hozac yeah, the thing i can't wrap my head around is that it's %o0 in the caller as well as the callee :) 1166826023 M * Bertl nah 1166826050 M * daniel_hozac ? 1166826052 M * Bertl the register mappings are switched IIRC 1166826059 M * Bertl sec, have to get my sparc manual 1166826098 M * Bertl http://www.sics.se/~psm/sparcstack.html 1166826105 M * Bertl google was faster :) 1166826139 M * Bertl but that's not really important for syscalls 1166826171 M * Bertl only relevant thing is that you cannot change %i in syscalls, only return %o 1166826174 M * daniel_hozac hmm, that suggests %i0 is the return value from this function. 1166826266 M * daniel_hozac that makes more sense and is what i've been thinking it should be all along. 1166826276 M * daniel_hozac the asm dumps seem to use %o0 though. 1166826285 M * daniel_hozac (6ad8, 6aa8, etc.) 1166826320 M * daniel_hozac and that would explain why it's failing. 1166826346 M * daniel_hozac hmm, no, it's checking for -1 explicitly. 1166826399 M * Bertl yep, but %o0, [%o0] looks strange 1166826406 M * daniel_hozac yeah, that's definitely wrong. 1166826423 M * Bertl let me check my notes about syscall convention 1166826441 M * daniel_hozac IMHO the alternative syscalls look right. 1166826455 M * daniel_hozac i.e. i'd blame gcc optimizing it a bit too much. 1166826519 M * Bertl so %o0 - %o5 are the arguments, no branch slots are used 1166826539 M * Bertl %o0 is the return code 1166826554 M * Bertl error is in the carry 1166826558 M * Bertl (flag) 1166826653 M * Bertl we are 32bit in userspace, right? 1166826662 M * daniel_hozac yeah. 1166826702 M * Bertl %g1 is syscall number, syscall is ta 10 1166826840 M * Bertl okay, can you describe once again what you see from debugging? 1166826910 M * daniel_hozac the debugging was somewhat inconclusive. 1166826950 M * daniel_hozac i thought the asm showed some serious problems, but as i was researching my reply i found that my assumptions were incorrect. 1166827008 M * Bertl okay, what was the 'observed' issue (at the call/function level) 1166827028 M * Bertl syscall returns wrong values? 1166827031 M * daniel_hozac well, gdb showed that the arguments were messed up. 1166827043 M * daniel_hozac i don't think the syscall was the problem. 1166827068 M * Bertl okay, so shortly before the syscall (ta 0x10) do we have a register dump? 1166827093 M * daniel_hozac no... 1166827113 M * daniel_hozac i guess we should've run gdb in asm mode. 1166827137 M * Bertl okay, does it happen on all systems? 1166827143 M * daniel_hozac all? 1166827158 M * daniel_hozac nebuchadnezzar is the only one who has reported problems. 1166827161 M * Bertl I mean, would it be possible to trigger it on my sparc without jumping through hoops? 1166827174 M * daniel_hozac i guess so. 1166827193 M * daniel_hozac --enable-apis=NOLEGACY and -O2 seemed to be required though. 1166827213 M * daniel_hozac (so that is pointing to a miscompilation, but i can't see it) 1166827231 M * Bertl 0.30.212, yes? 1166827233 M * daniel_hozac yep. 1166827259 M * Bertl do we have the compiled binaries and libraries? 1166827284 M * daniel_hozac they're the ones from Debian. 1166827291 M * daniel_hozac so just getting the .deb should suffice. 1166827343 M * Bertl ah, good 1166827518 M * daniel_hozac the delayed branching really threw me off... 1166827757 M * Bertl yeah, the branch slots are confusing if you are not used to them 1166827799 Q * dna Quit: Verlassend 1166827993 Q * Hollow Quit: Disconnecting from stoned server. 1166828243 J * phreak``_ ~phreak``@140.211.166.183 1166828257 Q * meebey Remote host closed the connection 1166828259 J * meebey meebey@booster.qnetp.net 1166828265 Q * phreak`` Read error: Connection reset by peer 1166828282 J * Hollow ~hollow@styx.xnull.de 1166828289 Q * phreak``_ 1166828315 Q * m4z Read error: Connection reset by peer 1166828319 J * phreak`` ~phreak``@140.211.166.183 1166828376 Q * nox Remote host closed the connection 1166828389 J * nox ~nox@static.88-198-17-175.clients.your-server.de 1166828394 Q * Loki|muh Ping timeout: 480 seconds 1166828413 J * m4z m4z@bastard-operator.from-hell.net 1166828416 J * Loki|muh loki@satanix.de 1166828437 Q * mire Ping timeout: 480 seconds 1166828447 Q * phreak`` 1166828964 J * phreak`` ~phreak``@styx.xnull.de 1166829174 J * mire ~mire@150-167-222-85.adsl.verat.net 1166829689 M * Bertl hmm, okay, will have to wait, sparc is not coming up and I have no serial cable left 1166829992 M * daniel_hozac okay. 1166830024 M * Bertl but if nebuchadnezzar finds the time, a register dump before the syscall would be interesting 1166830048 M * Bertl i.e. show registers and for structures also some inspection 1166830070 M * daniel_hozac i don't think we get to the syscall, at least not with the debug compiles. 1166830087 M * daniel_hozac because the printf i added after it didn't output anything. 1166830099 M * bXi hmmm 1166830110 M * bXi i still cant find the downloadable guest tarball list 1166830157 M * daniel_hozac http://linux-vserver.org/Downloads 1166830859 M * bXi hmmm 1166830864 M * bXi i just got an ubuntu vserver 1166830875 M * bXi and basicly it gives this at random 1166830876 M * bXi root@ubuntu:/# 1166830876 M * bXi [1]+ Stopped vserver svn enter 1166830880 M * bXi and i'm dropped to my host 1166830903 M * daniel_hozac what's stopping it? 1166830921 M * Bertl sounds interesting, maybe you are generating CTRL-Z somehow? 1166830950 M * bXi http://mirrors.sandino.net/vserver/images/ubuntu-6.06-2006-08-23.tar.bz2 1166830952 M * bXi i got that one 1166830966 M * bXi vserver-new svn --context 1337 --hostname svn template ubuntu-6.06-2006-08-23.tar.bz2 1166830969 M * bXi and used that command 1166830975 M * bXi i guess some error in the guest 1166831378 M * daniel_hozac i don't see where vserver-new would give you a shell in the newly created guest. 1166831524 M * bXi oh ya i started and entered it 1166831566 M * Bertl and what happened right before the shell got suspended? 1166831590 M * daniel_hozac you could just type fg to get back to it.