1166401254 Q * dna Quit: Verlassend 1166401272 J * matti matti@linux.gentoo.pl 1166401280 M * Bertl wb matti! 1166402283 Q * matti Remote host closed the connection 1166404821 M * Bertl okay, off to bed now .. have a good one everyone! 1166404826 N * Bertl Bertl_zZ 1166407119 N * click_ click 1166414248 Q * kugg Remote host closed the connection 1166414658 M * mugwump bah, somebody wrote something to me, but I was in the bush :) 1166419799 Q * hardwire` Quit: Coyote finally caught me 1166420215 J * hardwire ~hardwire@rdbck-4891.wasilla.mtaonline.net 1166424899 J * FireEgl ~FireEgl@adsl-61-147-76.bhm.bellsouth.net 1166425189 J * yarihm ~yarihm@84-75-123-221.dclient.hispeed.ch 1166426216 J * leoleon ~leo@60-242-43-100.static.tpgi.com.au 1166426374 J * Aiken_ ~james@tooax8-118.dialup.optusnet.com.au 1166426692 Q * Aiken Ping timeout: 480 seconds 1166426694 M * leoleon hello 1166427103 J * meandtheshell ~markus@85-124-233-137.work.xdsl-line.inode.at 1166429174 Q * Johnnie Remote host closed the connection 1166429251 J * Johnnie ~jdlewis@jdlewis.org 1166429727 J * matti matti@linux.gentoo.pl 1166430581 Q * Johnnie Remote host closed the connection 1166430962 J * Johnnie ~jdlewis@jdlewis.org 1166432204 M * leoleon hello? 1166432781 Q * leoleon Quit: Leaving 1166433460 Q * transacid Ping timeout: 480 seconds 1166433548 J * dna ~naucki@176-236-dsl.kielnet.net 1166433729 J * transacid ~transacid@transacid.de 1166438620 Q * Aiken_ Quit: Leaving 1166441021 M * daniel_hozac Bertl_zZ: they're in FE, so they're available for yum'ing right after install. 1166442703 J * shedi ~siggi@inferno.lhi.is 1166443280 Q * ensc Killed (NickServ (GHOST command used by ensc_)) 1166443281 Q * FireEgl Quit: ... 1166443291 J * ensc ~irc-ensc@p54B4FFE3.dip.t-dialin.net 1166447190 J * FireEgl ~FireEgl@adsl-61-147-76.bhm.bellsouth.net 1166448117 J * Piet hiddenserv@tor.noreply.org 1166448303 Q * Loki|muh Remote host closed the connection 1166448304 J * Loki|muh loki@satanix.de 1166448764 M * daniel_hozac Hollow: i threw something together to block edits with libx11-data 1166455675 M * doener I wonder why that needs /dev/mem now... 1166455699 M * Piet php4-gd depends on libxpm which depends on x11-common 1166455727 M * doener well, that's the boring dependency ;) 1166455773 M * Piet if you looks at the pastebin entry from above you'll see it seems like the x11-common preconfigure scripts seems to access /dev/mem 1166455774 J * newz2000 ~matt@12-216-147-124.client.mchsi.com 1166455795 M * doener yeah, but which part? and why? :) 1166455809 M * Piet i neither know nor want to find out :) 1166455818 M * doener I had a look at the .preinst script... way too much unrelated stuff 1166457276 Q * matti Remote host closed the connection 1166458138 Q * Johnnie Remote host closed the connection 1166458177 J * Johnnie ~jdlewis@jdlewis.org 1166458373 J * matti ~matti@linux.gentoo.pl 1166460055 Q * matti Remote host closed the connection 1166460385 J * stefani ~stefani@tsipoor.banerian.org 1166461427 M * nebuchadnezzar Bertl_zZ: with the main util-vserver, I still have operation not supported :-/ 1166461440 Q * lilalinux Ping timeout: 480 seconds 1166461444 M * daniel_hozac nebuchadnezzar: what filesystem do your guests live on? 1166461451 J * lilalinux ~plasma@dslb-084-058-216-120.pools.arcor-ip.net 1166461479 M * nebuchadnezzar ext3 1166461486 M * nebuchadnezzar over LVM 1166461497 M * daniel_hozac what does strace say? 1166461658 J * bonbons ~bonbons@83.222.37.103 1166461663 M * nebuchadnezzar sorry, with my compiled util-vserver it says that it can not find vcontext, but it exists 1166461698 M * daniel_hozac what says it can't find vcontext? 1166461709 M * daniel_hozac vhashify shouldn't be using that at all. 1166461753 M * nebuchadnezzar http://paste.linux-vserver.org/791 1166461775 M * daniel_hozac ah. 1166461810 M * daniel_hozac is the guest running? 1166461830 M * nebuchadnezzar yes 1166461840 M * daniel_hozac have you tried stopping it? 1166461849 M * nebuchadnezzar the same 1166461859 M * nebuchadnezzar vnamespace: execvp("/usr/local/sbin/vserver"): No such file or directory 1166461869 M * daniel_hozac even after stopping it? 1166461885 M * nebuchadnezzar this is the result of vserver betch stop 1166461902 M * daniel_hozac oh. 1166461910 M * nebuchadnezzar strange 1166461916 M * daniel_hozac what util-vserver version started it? 1166461921 M * nebuchadnezzar 0.30 1166461931 M * nebuchadnezzar from debian 1166461937 M * daniel_hozac which 0.30? 1166461942 M * nebuchadnezzar vps afx works 1166461955 M * daniel_hozac is /usr/local on a separate filesystem? 1166461962 M * nebuchadnezzar 0.30.212 1166461970 M * daniel_hozac 0.30.212-1 from Debian? 1166461975 M * nebuchadnezzar yes, it's a separate fs 1166461978 M * nebuchadnezzar yes from debian sid 1166461987 M * daniel_hozac so that's why then. 1166461997 M * daniel_hozac /usr/local has been unmounted in the guest's namespace. 1166462338 P * independence 1166462364 M * nebuchadnezzar ok, so I install the debian util-vserver and stop the vserver 1166462372 M * daniel_hozac yep. 1166462414 M * nebuchadnezzar now if I start it with my compiled vserver it should be ok ? 1166462431 M * daniel_hozac yep. 1166462478 J * lilalinux_ ~plasma@80.69.41.2 1166462585 Q * lilalinux Ping timeout: 480 seconds 1166462685 M * nebuchadnezzar daniel_hozac Bertl_zZ ok, so, with a mainline vserver + dietlibc the vhashify works 1166462699 M * nebuchadnezzar I'll try with debian dietlibc 1166462708 M * nebuchadnezzar to localize the problem explicitely 1166462818 M * nebuchadnezzar thanks to daniel_hozac and Bertl_zZ for your disponibility :-) 1166463016 M * daniel_hozac np, will be interesting to see what is causing the problem... 1166463045 M * nebuchadnezzar I'll try a diff between mainline and debian util-vserver 1166463076 M * daniel_hozac for 0.30.212? should just be a weird error message "fix". 1166463102 M * nebuchadnezzar I'll give the result of the tests to Ola... 1166463216 M * daniel_hozac sure, i guess both micah and i get copies anyway :) 1166463240 M * nebuchadnezzar ok, with debian dietlibc and debian libbeecrypt + mainline util-vserver it works 1166463251 M * daniel_hozac really? 1166463251 J * the_hydra ~a_mulyadi@202.59.168.29 1166463269 M * daniel_hozac could you diff the vserver-info outputs? 1166463288 M * nebuchadnezzar sure 1166463438 M * nebuchadnezzar http://paste.linux-vserver.org/792 1166463525 M * daniel_hozac hmm, looks like Debian builds it against sparc32. 1166463536 M * daniel_hozac have you tried doing that? 1166463542 M * nebuchadnezzar no 1166463579 M * nebuchadnezzar I'll try with a mcpu=v8 1166463615 M * nebuchadnezzar the debian one have more API 1166463623 M * daniel_hozac isn't -march=v8 what's required? 1166463653 M * nebuchadnezzar I'll look at the debian source 1166463653 M * daniel_hozac (i don't really know, it seems like the package was actually built on a 32-bit sparc (chroot?)) 1166463687 M * nebuchadnezzar well, 0.30.210 was running on my sparc64 machine 1166463704 M * daniel_hozac what kernel? 1166463718 M * nebuchadnezzar 2.6.16 IIRC 1166463723 M * daniel_hozac or hmm, it was only vhashify that wasn't working with the debian utils, right? 1166463730 M * nebuchadnezzar right 1166463747 M * daniel_hozac ok, that rules out my initial theory. 1166463758 M * nebuchadnezzar which is ? 1166463829 M * daniel_hozac 32 vs. 64 syscalls. 1166463833 M * daniel_hozac +-bit 1166463876 M * the_hydra daniel_hozac: sorry, interrupting...herbert is still zzzz? 1166463883 M * daniel_hozac the_hydra: yes, it appears so. 1166463889 M * the_hydra daniel_hozac: ok thanks 1166463902 M * daniel_hozac nebuchadnezzar: looks like you have a more recent compiler too. 1166463964 M * daniel_hozac the_hydra: anything we (the others :)) can do for you? 1166464011 M * the_hydra daniel_hozac: wanna discuss about the surfacing issues related with virtualization, that is hardware assisted VS software assisted one :) 1166464078 M * the_hydra daniel_hozac: if you don't mind, please kindly share your opinion about it... 1166464083 M * daniel_hozac well, regardless of how assisted the hardware virtualization is, it'll always be slower than OS-level as you can't share caches, buffers, files, etc. 1166464115 M * the_hydra because it operates on different namespace/address space when switching ringlevel? 1166464120 M * nebuchadnezzar daniel_hozac: there is some changes in the lib/vserver-syscall-def.h in the debian source 1166464135 M * daniel_hozac nebuchadnezzar: hmm? why isn't that in the patch? 1166464157 M * daniel_hozac the_hydra: well that, and that you have multiple kernels keeping individual track of it. 1166464180 M * nebuchadnezzar I don't know 1166464182 M * daniel_hozac i.e. every in-kernel structure will be duplicated for each running guest. 1166464199 M * the_hydra daniel_hozac: i see...i think what you mean by "multiple kernels" are host, hypervisor and guest itself...CMIIW 1166464232 M * daniel_hozac the_hydra: well, i was more referring to each of the guests. 1166464241 M * nebuchadnezzar it don't seems to impact sparc http://paste.linux-vserver.org/793 1166464244 M * daniel_hozac but yes, the hypervisor would need it too. 1166464262 M * daniel_hozac nebuchadnezzar: hmm, i don't even have that file. 1166464272 M * the_hydra daniel_hozac: hm ok... 1166464324 M * daniel_hozac ah, i see it in the diff now. 1166464345 M * daniel_hozac i thought micah kept all of the changes as dpatches, but it seems not. 1166464671 M * the_hydra daniel_hozac: thanks a lot... 1166464780 Q * sladen Ping timeout: 480 seconds 1166464824 N * Bertl_zZ Bertl 1166464839 M * Bertl morning folks! 1166464848 M * daniel_hozac morning Bertl! 1166464853 M * the_hydra right on time 1166464857 M * the_hydra hi bert 1166464873 M * Bertl hey the_hydra! how's going? 1166464901 M * the_hydra Bertl: me? quite fine..still far from mud's lake 1166464923 M * the_hydra Bertl: i asked this to daniel_hozac before, hope you don't mind 1166464948 M * the_hydra Bertl : wanna discuss about the surfacing issues related with virtualization, that is hardware assisted VS software assisted one 1166464975 M * the_hydra bertl: perhaps you can share your opinion about it...? 1166464988 M * Bertl hmm, you mean like in Xen vs PC/VP? 1166465056 M * the_hydra Bertl: yes 1166465075 J * sladen paul@starsky.19inch.net 1166465095 M * Bertl well, not quite Linux-VServer related, but what do you wanna discuss there? 1166465148 M * the_hydra in perfomance aspect, as far as you know, do you think hardware assisted one is really faster than software assisted one? 1166465206 M * daniel_hozac nebuchadnezzar: AFAICT, that's not used anywhere at all, so it shouldn't be used anywhere. 1166465228 M * daniel_hozac ... so rewriting messages in the middle isn't a good idea :) 1166465252 M * Bertl the_hydra: well, that's a though question ... is an FPU in hardware faster than in software? probably depends on the commands the CPU has, no? 1166465274 Q * meandtheshell Quit: Leaving. 1166465297 M * the_hydra Bertl: somehow my lousy brain agree with that :) 1166465311 M * Bertl the_hydra: fact is, that both intel and amd (out of heritage) _had_ very broken support for hypervisor like virtualization 1166465334 M * Bertl that was mostly fixed with the VT part of VP/PC 1166465369 M * Bertl the hardware 'virtualization' is just the icing on the cake 1166465378 M * the_hydra Bertl: i got same impression...so actually the problem is, both VT/Pacifica hasn't implemented a really effective hardware based virtualization? 1166465403 J * meandtheshell ~markus@85-124-233-137.work.xdsl-line.inode.at 1166465404 M * the_hydra Bertl: and the approach should always be a hybrid one.. ? 1166465427 M * Bertl nope, good hardware virtualization can be found on s3x0, really big powerpc and sparcs 1166465437 M * Bertl aka as partitioning 1166465457 M * the_hydra oh forgot about that ones...thanks for the reminder 1166465467 M * Bertl but that is not what Xen and friends are heading to 1166465475 M * the_hydra how can I forgot that old irons...grrmffff 1166465494 M * the_hydra no? hm...so, what are they heading to? 1166465513 M * Bertl they want to do soft partitioning at a finer grained level than CPU boards 1166465571 M * the_hydra finer? hm...finer than the current state? 1166465584 M * the_hydra wonder how does that looks like... 1166465779 M * Bertl well, Xen already does 'virtual cpu' and such 1166465787 Q * meandtheshell Quit: Leaving. 1166465807 M * Bertl it's not very effective, because the CPUs do not really support fast context switching across kernels# 1166465814 M * Bertl but it is definitely getting better 1166465859 M * the_hydra so with this virtual cpu, each guest could own a different processing state? 1166465882 M * the_hydra and possibly different pipeline state, cache state and so on? 1166466166 J * meandtheshell ~markus@85-124-37-195.dynamic.xdsl-line.inode.at 1166466997 Q * the_hydra Ping timeout: 480 seconds 1166467003 M * Bertl well, something like that, mostly tlb switches 1166467126 J * m`m`h ~simba@deb30.mgts.by 1166467292 J * djrise ~djrise@lns-bzn-48f-62-147-149-121.adsl.proxad.net 1166467347 M * Bertl welcome djrise! 1166467393 M * djrise hi everyone 1166467628 M * djrise i have a question about hearbeat 1166467635 M * Bertl +t 1166467637 M * djrise does anyone can help me? 1166467646 M * Bertl maybe, depends on the question :) 1166467739 M * djrise i want to do failover service between vserver 1166467752 M * djrise i have a machine with a vserver 1166467766 M * djrise and second with another vserver for backup 1166467790 M * djrise and have also a NAS to share mail 1166467804 M * djrise it's a mail server but i want to make for all services 1166467831 M * djrise does i really need drbd device to make raid1 networking 1166467840 M * djrise between vserver 1166467843 M * djrise ? 1166467923 M * djrise because with the NAS i'm sure mail data are mounted in the two vserver and are the same 1166467934 M * djrise but i don't understant very well 1166467938 M * djrise how does it work 1166467986 M * Bertl hmm, how is the NAS attached? 1166467997 M * Bertl NFS or iSCSI or AOE? 1166468169 M * djrise nfs 1166468190 M * Bertl and only te mail dir is there? 1166468211 M * djrise yes 1166468246 M * Bertl well, if other data changes over time, you probably want to sync the guests somehow 1166468260 M * Bertl that could be via the NAS or via rsync or via drbd 1166468397 M * djrise so 1166468453 M * djrise so i can use only hearbeat 1166468472 M * djrise and rsync to synchronise guest 1166468501 M * djrise maildir is synchronised by the NAS 1166468509 M * Bertl yes, it doesn't matter _how_ you keep the guest in sync with the 'copy' 1166468546 M * Bertl for switchover it would be sufficient to 'copy' the guest, for failover you probably want a periodic synchronization 1166468555 M * hardwire moo 1166468574 M * hardwire or DRBD/GFS 1166468628 M * djrise it's for failover, only one service (postfix) run 1166468645 M * djrise i must synchronise with a cron ? 1166468650 M * hardwire you must! 1166468672 M * Bertl no, if all the relevant data is on the NAS 1166468679 M * hardwire unless you don't want your mailboxes to stay somewhat persistent after failover 1166468683 M * hardwire oh 1166468683 M * hardwire heh 1166468690 M * Bertl you just have to sync when you update postfix 1166468702 M * hardwire cron it so your foot stays firmly hole free 1166468773 M * djrise does i must rsyncro with distribution update or kernel-update 1166468777 M * djrise or it can work 1166468778 M * djrise ? 1166468828 M * Bertl kernel does not really matter, but when you update the relevant service, it would be good to keep the copies in sync 1166468894 M * djrise ok , it's what i think too , the essential it's same postfix version and utilities 1166468988 M * djrise and if the primary guest failover and his mailq is full 1166469036 M * djrise i loose my mailq i presume 1166469055 M * Bertl well, I'd suggest to put the /var/spool/{mail,postfix} on the NAS 1166469066 M * Bertl in this case, you should not lose anything 1166469120 M * djrise logical ;-) 1166469214 M * djrise but i think about another solution what do you think about that 1166469250 M * djrise so, my first machine is running with vserver mail, dns, web 1166469269 M * djrise all works and i don't really want to 1166469334 M * djrise format. I want to mount /vserver dir via NFS on the fist and the backup machine 1166469385 M * djrise and share vserver with hearbeat between vserver 1166469404 M * djrise off course, synchronised it's ok! 1166469450 M * djrise because all vserver data are on the nas, but i don't if this solution are optimal and if it's possible 1166469461 M * djrise know sory 1166469463 M * Bertl sure, that works 1166469484 M * Bertl you just should make damn sure that the guests do not run at the same time (sharing the same storage) 1166469633 M * djrise ok that's hearbeat job, and with this solution i can give the same name between vserver in different machine? 1166469676 M * Bertl sure, but you probably want to use identical ips too 1166469687 M * Bertl and have some arp announcement handle that 1166469745 M * djrise it's vrrp or ucarp but it doesn't work good with vserver 1166469802 M * djrise hearbeat give virtual ip and two real ip for guest's 1166469845 M * djrise what do you think about this solution 1166469859 M * djrise i think it's better than the fist solution 1166469963 M * Bertl if you are comfortable with it, it should be fine 1166469989 M * djrise as you can see i'm never do that ... 1166470087 M * Bertl well, there is always a first time :) 1166470105 M * djrise :-) 1166470203 M * djrise thanks for all your answer 1166470221 M * Bertl you're welcome! 1166470249 M * newz2000 I'm finally getting a chance to upgrade my ctx17 vservers to the latest and greatest. 1166470270 M * newz2000 I'm a little afraid, because so much has changed its almost like starting from scratch learning new tools 1166470308 M * daniel_hozac how old is ctx17? 4 years? 1166470314 M * daniel_hozac 5? 1166470316 M * newz2000 Has it been that long? 1166470351 M * daniel_hozac wow, it's just 3 and a half. 1166470452 M * newz2000 I can't wait to figure things out again 1166470525 M * Bertl let us know how it goes 1166470892 J * Roey ~katz@h-69-3-4-130.mclnva23.covad.net 1166471129 M * Bertl wb Roey! 1166471134 M * Roey hello bertl!!! 1166471164 M * Roey I just wanted to let you know that the VServer-based solution we are using to host www/dns/email is still up and kicking :) 1166471169 M * Roey so that you again : 1166471170 M * Roey :) 1166471225 M * Bertl great to hear! 1166471359 M * Roey Bertl: now, I was wondering what the plans are for integration into the mainline kernel 1166471375 M * Roey now that it's been decided that KVM (and not Xen) will be the standard VM 1166471393 M * Roey do you seee Linux-VServer getting in? 1166471398 M * daniel_hozac though neither would've influenced Linux-VServer. 1166471403 M * Roey I feel bad about this OpenVZ stuff 1166471408 M * daniel_hozac the spaces are more interesting. 1166471418 M * Roey daniel_hozac: what do you mean, ths spaces? 1166471433 M * daniel_hozac (i.e. we're using the uts and IPC spaces introduced in 2.6.19, and will likely use the pid spaces in 2.6.20) 1166471490 Q * bronson Ping timeout: 480 seconds 1166471619 M * daniel_hozac (hmm, looks like pid spaces won't be in 2.6.20. i just assumed) 1166471728 M * bonbons daniel_hozac/Bertl: can 2.1.1.7 work without a vshelper? REBOOT_KILL does not work anymore for me... just get a message in kernel log 1166471970 M * daniel_hozac bonbons: hmm, it should work fine. 1166471973 M * bonbons looks like an empty value in /proc/sys/kernel/vshelper is needed for it to work as before ... 1166472017 M * daniel_hozac really? 1166472022 M * daniel_hozac what does the kernel log say? 1166472087 M * bonbons it says: /sbin/vshelper: (restart 11) returned sync with -2 1166472125 M * bonbons and there is no /sbin/vshelper on that system... so I would have expected the REBOOT_KILL to operate instead of kernel doing a NOOP... 1166472508 M * daniel_hozac hmm, you're right. 1166472525 M * daniel_hozac i guess we should add a && ret != -ENOENT to do_vshelper. 1166472535 M * Bertl legacy vs non legacy? 1166472537 M * daniel_hozac or alternatively vs_reboot. 1166472607 J * johnny ~johnny@ip70-160-117-211.hr.hr.cox.net 1166472649 M * Bertl bonbons: what if you configure /bin/true for the helper? 1166472657 M * johnny hi folks 1166472667 M * johnny i just upgraded to a 2.2rc5 1166472675 M * Bertl good 1166472703 J * Loki|muh_ loki@satanix.de 1166472705 M * bonbons Bertl: will try, but on -ENOENT it should just work as if nothing was set... 1166472724 M * Bertl I disagree, it's like the hotplug helper 1166472767 M * Bertl i.e. it is a misconfiguration to have nothing there 1166472813 M * Bertl we can set a different default (i.e. /bin/true) if the older tools handle that gracefully 1166472827 M * bonbons with /bin/true it works... checking if the helper gets ever set... 1166472834 M * daniel_hozac i don't think there are any tools that set the helper. 1166472851 M * daniel_hozac otherwise we'd just set it to the path directly, i guess. 1166472858 M * johnny fun.. 1166472860 M * johnny chbind: vc_net_create(): Invalid argument 1166472862 M * daniel_hozac i.e. no need for the /sbin/vshelper symlink. 1166472870 M * johnny bet you guys have seen that one before 1166472873 M * bonbons johnny: dynamic contexts? 1166472879 M * daniel_hozac johnny: dynamic contexts, but no support in the kernel? 1166472894 M * johnny well it said it was deprecated :... so i removed it 1166472895 M * Bertl see ML announcement 1166472909 M * johnny ok.. cool that must be it exactly 1166472911 M * Bertl it's okay to disable dynamic contexts 1166472915 M * bonbons but you didn't define the context id in your guest config ;) 1166472918 M * johnny yeah 1166472925 M * Bertl but you should not use them afterwards :) 1166472927 M * johnny what is the actual site 1166472935 M * johnny err address of the great flower page 1166472946 M * johnny it's mentioned in the faq.. but it doesn't link to it 1166472965 M * Bertl echo 42 >/etc/vservers//context 1166472979 M * daniel_hozac http://www.nongnu.org/util-vserver/doc/conf/configuration.html 1166472980 M * johnny hmm.. 42? :) 1166472983 M * Bertl (range is 2-49151 atm, 65535 soon) 1166472993 M * daniel_hozac (also available in /doc/configuration.html) 1166472995 M * johnny i'll go with 2 i guess 1166473012 Q * Loki|muh Ping timeout: 480 seconds 1166473012 N * Loki|muh_ Loki|muh 1166473056 M * johnny so what was the problem with dynamic contexts? 1166473061 M * daniel_hozac everything :) 1166473075 M * daniel_hozac think of it like having a dynamic uid. 1166473108 M * johnny does vserver-new automatically create one? 1166473124 A * johnny wonders how good of shape his /etc/vservers is in 1166473233 M * Bertl make sure each guest has a separate one 1166473238 M * Bertl s/separate/distinct/ 1166473246 M * johnny yeah i guessed that :) 1166473286 M * daniel_hozac i was under the impression vserver-new had been removed. 1166473294 M * daniel_hozac has it just been deprecated? 1166473430 M * johnny oh 1166473460 M * johnny is there a specific version of util-vserver i should be using for 2.2 1166473491 M * Bertl 0.30.212 1166473506 M * johnny hmm.. i'm using 202 1166473517 M * johnny err 210 1166473519 M * johnny oops 1166473559 M * johnny so i should upgrade? 1166473564 M * daniel_hozac yes. 1166473705 M * johnny lol.. i was just wishing for bash completion support in util-vserver and the 212 ebuild has it :) 1166473838 M * micah daniel_hozac: i am pretty sure I've only kept changes in patches, although something could have snuck in 1166473875 M * bonbons Bertl: so what do we agree on, treat -ENOENT in do_vshelper() as if there was no helper set, or switch default to something else (assuming util-vserver sets helper...) 1166473894 M * daniel_hozac micah: there are quite a few new files and such that aren't patches. 1166473907 M * daniel_hozac bonbons: util-vserver doesn't set the helper. 1166473942 M * daniel_hozac micah: http://paste.linux-vserver.org/794 1166473951 M * micah daniel_hozac: hmm, I wonder if those are left over changes that ola made to the upstream source 1166473958 M * bonbons so we treating -ENOENT is more or less the only option... 1166474002 M * micah daniel_hozac: the Makefile.svn is the one I added just recently to be able to build the -rc 1166474034 M * Bertl bonbons: in which cases does a system run without util-vserver? 1166474048 M * Bertl (i.e. without the helper assigned) 1166474064 M * micah the man ones are likely ola, they would be good to move into patches or determine if they should be pushed upstream, but aren't really relevant to this 1166474109 M * Bertl micah: btw, can be something done about Ola no-handling bug reports properly? 1166474112 M * daniel_hozac Bertl: vserver-utils? :) 1166474120 M * Bertl *not-handling 1166474133 M * daniel_hozac Bertl: how do you mean? 1166474136 M * Bertl daniel_hozac: well, then my answer would be this: 1166474139 M * micah Bertl: I wish... 1166474168 M * daniel_hozac Bertl: i guess micah or me will respond to bug reports if we don't agree with him. 1166474175 M * micah definately 1166474190 M * micah i'm actually surprised he has been responding to them recently quite fast 1166474193 M * Bertl daniel_hozac, bonbons: from now on, please configure the helper properly, util-vserver can set /sbin/vshelper, (and/or remove the links) and vserver-utils can set /bin/true 1166474193 M * daniel_hozac yeah. 1166474197 M * micah if only to say, I he doesn't know 1166474238 M * daniel_hozac Bertl: so the default will be changed to /bin/true? 1166474251 M * Bertl micah: well, I'm a little annoyed by wrong conclusions I do not even see (like it must be a broken kernel) and such, which get reported here only by accident 1166474266 M * Bertl s/see/hear/read about/ 1166474283 M * Bertl daniel_hozac: I see no reason for doing that right now 1166474293 M * daniel_hozac well, i was trying to debug that with nebuchadnezzar here. 1166474295 M * Bertl daniel_hozac: but it would not hurt to add setting it properly 1166474321 M * daniel_hozac i prefer debugging on IRC over email :) 1166474333 M * bonbons will do so 1166474336 M * micah daniel_hozac: I'll look into getting these differences from upstream back in sync with your source 1166474357 M * daniel_hozac Bertl: right, well, modifying /etc/sysctl.conf seems rather ugly. 1166474369 J * Aiken ~james@tooax8-009.dialup.optusnet.com.au 1166474378 M * Bertl daniel_hozac: there is proc and sysctl -w :) 1166474384 M * nebuchadnezzar daniel_hozac: it seems that the enable-apis=NOLEGACY cause the problem 1166474424 M * daniel_hozac Bertl: but when would it run? you only want to set it on boot. 1166474449 M * daniel_hozac nebuchadnezzar: so you can reproduce it with --enable-apis=NOLEGACY? 1166474457 M * Bertl daniel_hozac: with the runlevel script which also kills the remaining contexts on shutdown :) 1166474470 M * nebuchadnezzar I'm eating for now, I'll look at it in couple of minutes 1166474495 M * daniel_hozac Bertl: hehe, good point. 1166474512 M * Bertl daniel_hozac: obviously the helper is checked right now on every guest start/stop, it wouldn't hurt to check the proc/sysctl entry there too and correct it if necessary 1166474574 Q * lilalinux_ Remote host closed the connection 1166474597 M * daniel_hozac Bertl: well, it's already checking it. 1166474749 M * johnny hmmm.. ok now time to figure out this sysinit thing 1166475226 M * nebuchadnezzar daniel_hozac: removing the --enable-apis=NOLEGACY from the debian build and vhashify works 1166475260 M * daniel_hozac nebuchadnezzar: so if you add --enable-apis=NOLEGACY, it breaks? 1166475288 M * micah daniel_hozac: I've sorted out all those differences, removing stuff that shouldn't be there, and getting the rest sync'd back to what you have 1166475303 M * micah daniel_hozac: I dont think any of those differences have to do with nebuchadnezzar's problem though 1166475305 M * nebuchadnezzar daniel_hozac: yes, it breaks the mainline util-vserver 1166475309 M * Bertl nebuchadnezzar: did you enable the vserver debugging in the kernel config? 1166475315 M * daniel_hozac micah: i agree. 1166475317 M * nebuchadnezzar Bertl: yes 1166475327 M * Bertl excellent, please do the following: 1166475351 M * nebuchadnezzar ok 1166475361 M * Bertl sysctl -w vserver.debug_switch=255 1166475379 M * nebuchadnezzar done 1166475384 M * daniel_hozac retry vhashify now. 1166475385 M * Bertl then run a trivial vhashify (only a few files in two dirs) 1166475404 M * nebuchadnezzar Bertl: with the working one or the not working one ? 1166475423 M * Bertl with both, but check/save the dmesg output between runs 1166475428 M * nebuchadnezzar ok 1166475441 M * Bertl so that we get the debug messages for both of them 1166475561 M * johnny should i have a vservers.conf ? 1166475569 M * johnny /etc/vservers.conf 1166475623 M * daniel_hozac no. 1166475627 M * Bertl well, mine says BACKGROUND=no 1166475629 M * johnny yeah 1166475633 M * johnny that's what mine says 1166475640 M * johnny just wondering if it was useful at all 1166475650 M * nebuchadnezzar I haven't one 1166475651 M * johnny since i don't see it on the great flower page 1166475661 M * daniel_hozac it's a legacy file. 1166475668 M * nebuchadnezzar it was for the old configuration ? 1166475671 M * nebuchadnezzar arf 1166475685 M * nebuchadnezzar in english we say legacy 1166475702 M * nebuchadnezzar in english _tehey_ say legacy 1166475709 M * nebuchadnezzar f** keyboard 1166475713 M * daniel_hozac hehe. 1166475716 M * nebuchadnezzar stupid user 1166475824 M * nebuchadnezzar with debug it became really slow 1166475833 M * daniel_hozac yeah, it's logging a lot. 1166475853 M * Bertl that's why I sugegsted to use only a few files 1166475885 M * nebuchadnezzar but it checks all the other files 1166475992 M * johnny any of the gentoo peeps here now? i'd like to know about the vserver profile and baselayout-1.13 1166476027 M * nebuchadnezzar Bertl: ok, do you want the whole files or just a diff ? 1166476028 M * daniel_hozac there's no need for a vserver profile for baselayout-1.13. 1166476040 M * daniel_hozac nebuchadnezzar: a diff would suffice. 1166476077 M * johnny that's what i heard 1166476085 M * johnny but.. i seem to be missing an important script 1166476098 M * daniel_hozac which is? 1166476105 M * nebuchadnezzar daniel_hozac: a diff only show the output of the non working call to vhashify 1166476144 M * daniel_hozac well, did you put the output of the first run in one file, and the output of the second run in another? 1166476198 M * johnny init-vservers.sh 1166476214 M * johnny mail twain.localmomentum.net # vserver twain.localmomentum.net start 1166476214 M * johnny vcontext: execvp("/lib/rcscripts/sh/init-vserver.sh"): No such file or directory 1166476214 M * nebuchadnezzar well, I just cp /var/log/messages two times, 1166476224 M * nebuchadnezzar I had the same :-) 1166476227 M * daniel_hozac init-vserver.sh is part of util-vserver. 1166476238 M * nebuchadnezzar near the same :-/ 1166476239 M * daniel_hozac if you install your guest with -d gentoo, it should be copied. 1166476252 M * johnny uhmm.. this guest is far too old for that 1166476253 M * johnny ok 1166476266 M * daniel_hozac johnny: well, do you have baselayout-1.13? 1166476270 M * johnny on the guest 1166476297 M * daniel_hozac so copy it manually. 1166476305 M * johnny yeah 1166476306 M * johnny going to 1166476307 M * daniel_hozac (or run the initpost script) 1166476313 M * johnny i'll copy it 1166476441 M * johnny yay.. now the gentoo init style works again 1166476480 M * johnny so uhmm.. what should the fstab be in a guest.. 1166476546 M * Bertl nebuchadnezzar: bzip2 the file and upload it somewhere 1166476557 M * nebuchadnezzar i'll do that 1166476603 M * daniel_hozac johnny: empty. 1166476734 M * johnny yay.. it almost all works now 1166476819 M * johnny so.. i guess i only have one question left.. do i need a mark file? 1166476861 M * johnny switching to 1.13 got rid of the sysinit problem i was having 1166476989 M * daniel_hozac you need the mark file if you want the guest to start on boot. 1166477145 M * bonbons weird... with a bash script as vshelper I get ... returned sync with 512! 1166477159 M * Bertl executable? 1166477226 M * bonbons ah, seems to be typo in the script... but running it manually returns 2 1166477236 J * bronson ~bronson@adsl-75-36-149-233.dsl.pltn13.sbcglobal.net 1166477293 M * daniel_hozac 512 means WEXITSTATUS(ret) == 0, no? 1166477294 M * bonbons not nice... but probably caused by the exit code stored between other data 1166477309 M * daniel_hozac == 2, of course. 1166477314 M * bonbons hehe 1166477423 M * nebuchadnezzar Bertl: http://www.asgardr.info/~nebu/vserver/messages.fail.bz2 and http://www.asgardr.info/~nebu/vserver/messages.working.bz2 1166477445 M * nebuchadnezzar both just have 6 empty files to hashify 1166477540 M * daniel_hozac 403. 1166477545 M * Bertl hmm, 403 here too :) 1166477550 M * nebuchadnezzar arf 1166477563 M * nebuchadnezzar here too 1166477579 M * nebuchadnezzar ok now 1166477585 M * nebuchadnezzar need to reload page maybe 1166477709 M * daniel_hozac Bertl: do we have a script to convert the switch debugging numbers into the appropriate VCMD_? 1166477712 M * bonbons Bertl: is there a way to kill a "dead" guest? (except rebooting the host) 1166477746 M * daniel_hozac bonbons: how would you kill it if it's dead already? 1166477753 M * daniel_hozac i.e. what is still existing? 1166477772 M * Bertl daniel_hozac: vcmd can do that, but no script yet 1166477775 M * bonbons dead in the sense that it still exists, but the init in there is in it's post-reboot/halt loop 1166477797 M * daniel_hozac bonbons: vkill -9? 1166477940 M * bonbons hm, I tried with signal name instead of number... and that one did not get recognized 1166478032 M * Bertl daniel_hozac: I don't see any difference in the logs, except for the fact that the working version continues with VCMD_38_1[1] 1166478050 M * Bertl while the other does not, so my best guess would be that the tools use the ioctl interface 1166478085 M * daniel_hozac nebuchadnezzar: hmm, did you post your entire vserver-info somewhere? 1166478114 M * nebuchadnezzar which one ? 1166478121 M * nebuchadnezzar working/not working ? 1166478127 M * daniel_hozac doesn't really matter. 1166478138 M * nebuchadnezzar ok 1166478140 M * daniel_hozac i'm mostly interested in the VS-API version. 1166478178 J * Piet_ hiddenserv@tor.noreply.org 1166478251 M * nebuchadnezzar every thing is on http://www.asgardr.info/~nebu/vserver/ now 1166478280 M * nebuchadnezzar the debian one, the mainline which works and the mainline which does not work 1166478304 M * nebuchadnezzar VS-API: 0x00020102 1166478407 M * Bertl nebuchadnezzar: what filesystem? 1166478433 M * nebuchadnezzar ext3 1166478487 M * Bertl userspace is 64 or 32bit? 1166478545 M * nebuchadnezzar quite everything is 32bits, but in debian, some key packages are 64 bits like data bases (IIRC) 1166478554 M * nebuchadnezzar so it can be mixed 1166478580 Q * Piet Ping timeout: 480 seconds 1166478580 M * Bertl okay, I guess that explains it, the ioctl interface uses an int, and there is no compatibility configured 1166478596 M * nebuchadnezzar Bertl: the debug message where produces by 32 bits applications 1166478609 M * nebuchadnezzar -mcpu=v8 1166478635 M * daniel_hozac are you sure? that's not visible at all in the vserver-info output. 1166478655 M * nebuchadnezzar daniel_hozac: check the build/host 1166478667 M * nebuchadnezzar the debian one is sparc-unknown-linux-gnu 1166478678 M * daniel_hozac right, i know that. 1166478685 M * daniel_hozac but both of your mainline ones are 64 bit. 1166478691 M * nebuchadnezzar yes 1166478706 M * nebuchadnezzar but I don't use the mainline to make the test 1166478712 M * nebuchadnezzar I just recompile the debian package 1166478722 M * nebuchadnezzar without the enable-apis=NOLEGACY 1166478775 M * nebuchadnezzar so on debian from debian with the NOLEGACY which fails to run and on debian rebuilded without the NOLEGACY which works 1166478781 A * nebuchadnezzar hopes he is clear 1166478801 M * nebuchadnezzar both debian packages are 32bits 1166478826 M * Bertl it would be interesting to compile a 64bit legacy version 1166478836 M * Bertl (i.e. with NOLEGACY :) 1166478837 M * nebuchadnezzar but I test the 64bits mainline with and without NOLEGACY and I have the same result than with the debian 32 version 1166478853 M * nebuchadnezzar Bertl: I make the test and I have the same result 1166478862 M * nebuchadnezzar but maybe you want the debug output ? 1166478884 M * Bertl hmm, I would have assumed that the 64bit userspace works fine, but let me draft up a test program 1166478891 M * nebuchadnezzar ok 1166478898 M * daniel_hozac oh, you've found the problem already? 1166478960 M * Bertl nebuchadnezzar: http://vserver.13thfloor.at/Experimental/TOOLS/vxid-0.03.tar.bz2 1166478966 M * Bertl (that was fast, no? :) 1166478974 M * nebuchadnezzar piouff 1166478988 M * daniel_hozac Bertl: hmm, so how do i get vcmd to convert it? 1166479023 M * nebuchadnezzar Bertl: I just run make ? 1166479074 M * Bertl daniel_hozac: ./vcmd -nld -C VCMD_38_1[1] 1166479097 M * nebuchadnezzar Bertl: compiled 1166479104 M * daniel_hozac ah, thanks! -d... 1166479122 M * Bertl nebuchadnezzar: okay, now get a file somewhere on that filesystem with 'touch testÄ 1166479127 M * Bertl s/Ä/' 1166479149 M * nebuchadnezzar ok 1166479186 M * Bertl ./vxid -i test 1166479249 M * nebuchadnezzar http://paste.linux-vserver.org/795 1166479256 M * Bertl and try the same without the -i 1166479271 M * nebuchadnezzar xid= 0, flags=0x01000000, mask=0x00060000, /etc/vservers/.defaults/vdirbase/betch/usr/src/truc 1166479319 M * Bertl okay, now pick some other file (one which won't hurt to modify) 1166479336 M * Bertl and do: ./vxid -x 666 /path/to/file 1166479342 M * nebuchadnezzar under the vserver or from the host is ok ? 1166479351 M * Bertl (all from the host) after that, rerun buth commands 1166479354 M * Bertl *both 1166479361 M * nebuchadnezzar ok 1166479365 M * Bertl i.e. the -i and non -i 1166479426 M * nebuchadnezzar I add a comment on the past 795 1166479434 M * nebuchadnezzar for the vxid -x 1166479443 M * nebuchadnezzar there is an invalid argument 1166479527 M * Bertl hmm, the EINVAL is interesting ... 1166479531 M * nebuchadnezzar arf, http://paste.linux-vserver.org/797 1166479575 M * Bertl please try: 1166479585 M * Bertl ./vxid -i -x 666 /path/to/file 1166479619 M * nebuchadnezzar arf, http://paste.linux-vserver.org/798 1166479687 M * Bertl strange, I don't see any EINVAL near the ioctl code ... double checking now 1166479693 M * nebuchadnezzar that's a strange pastbin service, a new ID for each submition on the same paste 1166479741 M * Bertl but it seems that the GETTAG was lost for ext3 1166479753 M * daniel_hozac that's legacy, no? 1166479766 M * Bertl yes, and legacy seems to be enabled, no? 1166479776 M * daniel_hozac but isn't legacy supposed to be removed? 1166479787 M * nebuchadnezzar Bertl: not in the kernel 1166479794 M * nebuchadnezzar I disable the legacy and legacy net 1166479811 M * Bertl sec 1166479889 M * nebuchadnezzar I upload my kernel config if you want http://www.asgardr.info/~nebu/vserver/config-2.6.19.1-vs2.3.0.4+ancalagon.1 1166479895 M * Bertl nebuchadnezzar: please get this: http://vserver.13thfloor.at/Stuff/SCRIPT/testme.sh-0.17 1166479900 M * Bertl and run it like this: 1166479925 M * Bertl ./testme.sh-0.17 -dq 1166479963 M * nebuchadnezzar http://paste.linux-vserver.org/799 1166479968 M * Bertl tx 1166480028 M * Bertl okay, then the kernel side is correct to not-handle the legacy ioctl :) 1166480039 M * Bertl was in the impression that ioctl was used ... 1166480061 M * nebuchadnezzar I see ngen networking is disabled, isn't it automatically enabled when legacy is disabled ? 1166480062 M * Bertl OTOH, if you strace -fF the failing vhashify, what command does it fail on? 1166480085 M * daniel_hozac there is no ngnet right now :) 1166480092 M * Bertl nebuchadnezzar: taht flag will be on with 2.3.something 1166480116 M * nebuchadnezzar arf, ok 1166480150 Q * bonbons Quit: Leaving 1166480151 M * Bertl daniel_hozac: could it be that the tools fall back to ioctl somehow? 1166480167 M * daniel_hozac that shouldn't happen since the legacy version isn't enabled. 1166480190 M * nebuchadnezzar in my dmesg I get this: ioctl32(vxid:2712): Unknown cmd fd(3) cmd(40047805){00} arg(ff98182c) on /var/lib/vservers/betch/usr/src/plouf 1166480207 M * nebuchadnezzar thats the last command I did on the plouf file 1166480208 M * Bertl hmm :) 1166480216 M * nebuchadnezzar I'm waiting the strace 1166480232 M * daniel_hozac did you get anything for vhashify? no, right? 1166480255 M * Bertl guess he is running that now 1166480262 M * nebuchadnezzar arf, my vhashify is T 1166480282 M * daniel_hozac that's expected when you're running strace. 1166480320 M * nebuchadnezzar the strace output is blocked at 11k 1166480423 M * nebuchadnezzar I wonder if my strace is not waiting forever :-/ 1166480451 M * Bertl try strace -fF -o vhasify.trace vhashify ... 1166480464 M * nebuchadnezzar it's what I've done 1166480468 M * daniel_hozac oh, that might not work out. 1166480475 M * daniel_hozac since it enter the guest. 1166480478 M * johnny so, does one have to set a mark in a vserver or will it default to default? 1166480478 M * daniel_hozac +s 1166480489 M * daniel_hozac johnny: you have to set it. 1166480496 M * johnny ok 1166480498 M * johnny cool 1166480513 M * johnny now evertyhing will work i think 1166480515 M * johnny finally.. 1166480526 M * johnny from a restart that is 1166480526 M * Bertl daniel_hozac: hmm, okay, why does it enter the guest again? 1166480540 M * nebuchadnezzar the last things I see in my strace ouput is 1166480542 M * nebuchadnezzar open("/tmp/vunify.YkSX70", O_RDWR|O_CREAT|O_EXCL|O_LARGEFILE, 0600) = 3 1166480545 M * daniel_hozac Bertl: to get a list of filenames. 1166480553 M * nebuchadnezzar unlink("/tmp/vunify.YkSX70") = 0 1166480560 M * nebuchadnezzar fork() = 2828 1166480566 M * nebuchadnezzar wait4(2828, 1166480570 M * nebuchadnezzar and that's all 1166480593 M * Bertl daniel_hozac: is there a way to stop it from doing so? 1166480635 M * Bertl i.e. what if I want to 'hashify' files without a guest config? 1166480652 M * daniel_hozac you can do manual hashifications. 1166480670 M * daniel_hozac that's what --manually is for :) 1166480696 M * nebuchadnezzar is it usefull to enter the vserver to list files ? 1166480711 M * nebuchadnezzar since we know the place where they are ? 1166480723 M * daniel_hozac i guess chroot-sh would work for Debian. 1166480773 M * Bertl okay, do you have a simple test line for nebuchadnezzar with the --manual? 1166480913 M * daniel_hozac vhashify --manually --destination /vservers/.hash -vvvvvvv /vservers/ , i guess. 1166480917 M * daniel_hozac i've never used it myself. 1166480995 M * nebuchadnezzar answer is here http://oldwiki.linux-vserver.org/alpha+util-vserver 1166481090 M * nebuchadnezzar ok, it's running 1166481107 M * nebuchadnezzar I'll upload the strace output 1166481206 M * nebuchadnezzar Bertl, daniel_hozac http://www.asgardr.info/~nebu/vserver/hashify_fF.strace 1166481247 M * nebuchadnezzar arf, wrong file 1166481248 M * nebuchadnezzar sorry 1166481260 M * nebuchadnezzar hmm, I'll bzip2 it 1166481331 M * nebuchadnezzar Bertl, daniel_hozac http://www.asgardr.info/~nebu/vserver/hashify.fF.strace.bz2 1166481398 M * Bertl that looks really funny :) 1166481414 M * nebuchadnezzar ho yes 1166481445 M * nebuchadnezzar I just don't know what you can find in it :-) 1166481465 M * nebuchadnezzar happylly I have a good DSL connection :-) 1166481525 M * nebuchadnezzar But I have too much beer for me alone... it becomes difficult ;-) 1166481536 M * nebuchadnezzar near midnight here 1166481538 Q * johnny Quit: Ex-Chat 1166481552 M * Bertl same here (well, regarding the time :) 1166481612 M * Bertl daniel_hozac: to me it seems like something is going wrong in userspace, there is an lstat which seems to return just fine, but as response the tools start bitching 1166481640 M * daniel_hozac what's 3331 SYS_267(0x10000, 0, 0, 0xf7f661b4, 0) = 51447793 1166481717 M * Bertl no idea ... yet 1166481718 M * daniel_hozac oh, VCI. 1166481745 M * Bertl hum, seems my vcmd doesn't handle that then ... 1166481817 M * daniel_hozac well, it does look like userspace is the one coming up with the ENOSYS. 1166481893 M * daniel_hozac nebuchadnezzar: did you get any warnings when compiling it? 1166481900 M * nebuchadnezzar somes 1166481910 M * nebuchadnezzar you cant the compilation output ? 1166481920 M * daniel_hozac yeah. 1166481950 M * nebuchadnezzar with NOLEGACY 1166481962 M * daniel_hozac right. 1166482097 M * nebuchadnezzar it's coming 1166482223 M * Bertl daniel_hozac: okay, the vcmd-0.07-pre2 also understands # and VCMD_# like used in the strace 1166482248 M * daniel_hozac cool, thanks. 1166482254 M * nebuchadnezzar ok, configure.w.NOLEGACY and make.w.NOLEGACY is here http://www.asgardr.info/~nebu/vserver/ 1166482595 Q * dna Quit: Verlassend 1166482902 M * daniel_hozac i don't see anything that stands out. 1166482989 M * Bertl nebuchadnezzar: which gcc/binutils do you use? 1166483007 M * Bertl nebuchadnezzar, daniel_hozac: was it tried without dietlibc yet? 1166483019 M * nebuchadnezzar no 1166483024 M * daniel_hozac well, vhashify is usually built against glibc anyway. 1166483028 M * daniel_hozac (due to beecrypt) 1166483031 M * nebuchadnezzar 4.1.1-20 from debian sid 1166483056 M * Bertl daniel_hozac: gdb? 1166483057 M * nebuchadnezzar binutils 2.17-3 1166483072 M * daniel_hozac good idea. 1166483079 M * nebuchadnezzar erf 1166483166 M * nebuchadnezzar so, what's the programm ? 1166483182 M * daniel_hozac gdb /usr/lib*/util-vserver/vhashify 1166483190 M * daniel_hozac set args --manually... 1166483210 M * daniel_hozac bt vc_get_iattr 1166483218 M * daniel_hozac s/bt/b/ 1166483230 M * daniel_hozac run 1166483272 M * nebuchadnezzar I'm installing gdb 1166483342 M * nebuchadnezzar ok 1166483365 M * nebuchadnezzar 42 if ( (mask==0) || 1166483426 M * Bertl you can use 'r 1166483438 M * nebuchadnezzar for ? 1166483442 M * Bertl (instead of the set args, run) 1166483449 M * nebuchadnezzar too late :-) 1166483460 M * Bertl well, next time :) 1166483515 M * daniel_hozac type s 1166483529 M * nebuchadnezzar 48 if ( flags ) *flags &= ~*mask; 1166483553 M * daniel_hozac again. 1166483562 M * nebuchadnezzar 50 CALL_VC(CALL_VC_V13 (vc_get_iattr, filename, xid, flags, mask), 1166483584 M * Bertl I would be interested in a p *mask :) 1166483584 M * daniel_hozac again. 1166483605 M * nebuchadnezzar too late :-) 1166483617 M * daniel_hozac too late? 1166483634 M * nebuchadnezzar well, p *mask give me 1166483644 M * nebuchadnezzar Cannot access memory at address 0x0 1166483652 M * daniel_hozac mask is NULL? 1166483660 M * daniel_hozac what does bt show? 1166483667 M * nebuchadnezzar bt 1166483689 M * Bertl probably it was really a little too late 1166483704 M * nebuchadnezzar http://paste.linux-vserver.org/800 1166483711 M * nebuchadnezzar I can run it again 1166483715 M * Bertl np 1166483721 M * Bertl you'll get there again 1166483765 M * daniel_hozac yeah. 1166483790 M * nebuchadnezzar mask is null 1166483804 M * nebuchadnezzar 48 if ( flags ) *flags &= ~*mask; 1166483806 M * daniel_hozac mask shouldn't be available in the current scope. 1166483813 M * nebuchadnezzar (gdb) p *mask 1166483813 M * nebuchadnezzar Cannot access memory at address 0x0 1166483896 M * nebuchadnezzar no 1166483903 M * nebuchadnezzar when the breakpoint appear 1166483913 M * nebuchadnezzar Breakpoint 1, vc_get_iattr (filename=0xf7f229c4 "\235?030\200??, xid=0x11208, flags=0x1483c, mask=0x0) at lib/syscall_getiattr.c:42 1166483917 M * nebuchadnezzar mask is null 1166483984 M * daniel_hozac i don't think so. 1166484037 M * daniel_hozac then it would fail immediately. 1166484057 M * nebuchadnezzar the Breakpoint line is not the vc_get_iattr call ? the mask parameter is null isn't it ? 1166484079 M * daniel_hozac yep. 1166484091 M * daniel_hozac but i'm inclined to think your debugging is screwed up somehow. 1166484114 M * Bertl maybe -g -O2 would be good? 1166484122 M * Bertl (compile time) 1166484126 M * daniel_hozac that filename doesn't look valid, and if mask really was NULL, if ( flags )... wouldn't be executed. 1166484158 M * daniel_hozac -g -O2 is already used, according to the vserver-info outputs. 1166484217 M * daniel_hozac nebuchadnezzar: could you upload the asm disassembly of vc_get_iattr? 1166484219 M * nebuchadnezzar mask is an unint32_t 1166484241 M * nebuchadnezzar daniel_hozac: if you tell me how 1166484303 A * nebuchadnezzar is trying to understand 1166484317 M * Bertl objdump -d 1166484329 M * daniel_hozac objdump -d /usr/lib*/libvserver.so.0 | awk '// { do_print=1 } /^$/ { do_print=0 } do_print==1 { print }' 1166484335 M * Bertl or inside gdb, with disassm 1166484434 M * nebuchadnezzar done, 1166484463 M * nebuchadnezzar http://www.asgardr.info/~nebu/vserver/vc_get_iattr.asm 1166484529 M * Bertl hmm .. 503 :) 1166484539 M * nebuchadnezzar ? 1166484553 M * Bertl now it works 1166484566 M * nebuchadnezzar arf 1166484710 M * nebuchadnezzar maybe the NULL mask come from 32/64 bits ? 1166484747 M * Bertl no, userspace should be consistent 1166484758 M * nebuchadnezzar ok 1166484768 M * Bertl i.e. the toolchain should bail out when you try to link 64bit with 32bit code 1166484805 M * nebuchadnezzar heuuu 1166484806 M * nebuchadnezzar ok 1166484819 M * nebuchadnezzar but gdb ? 1166484924 M * Bertl should know what elf header it loaded 1166485011 M * nebuchadnezzar well, the first call to vc_get_iattr has an xid=0x11208 but if I continue the program, all other call to vc_get_iattr has xid=0 1166485022 M * nebuchadnezzar same for flags 1166485029 M * nebuchadnezzar and filename 1166485036 M * nebuchadnezzar arf 1166485040 M * nebuchadnezzar erverything is 0 1166485138 M * Bertl do you have a different gcc/binutils to try with? 1166485158 M * nebuchadnezzar no 1166485173 M * nebuchadnezzar I can try with debian util-vserver ? 1166485190 M * Bertl nah, probably won't change anything 1166485197 M * nebuchadnezzar arf, I can install gcc-4.0 1166485211 M * nebuchadnezzar or 3.4 1166485238 M * nebuchadnezzar but not for binutils 1166485294 M * nebuchadnezzar ok, 5h before work 1166485297 M * nebuchadnezzar need to sleep now 1166485305 M * nebuchadnezzar do you need something ? 1166485309 M * nebuchadnezzar a coffe ? 1166485309 M * Bertl np, we'll figure it sooner or later ... 1166485313 M * nebuchadnezzar a cigarette ? 1166485338 M * nebuchadnezzar :-) 1166485349 M * nebuchadnezzar thanks a lot for your time 1166485364 M * Bertl np, thank you too 1166485728 J * _dmax ~semaj@bl4-59-176.dsl.telepac.pt 1166485744 J * s0undt3ch_ ~s0undt3ch@bl4-59-176.dsl.telepac.pt 1166486076 Q * dmax Ping timeout: 480 seconds 1166486076 N * _dmax dmax 1166486112 P * stefani I'm Parting (the water) 1166486167 Q * s0undt3ch Ping timeout: 480 seconds 1166486167 N * s0undt3ch_ s0undt3ch